aboutsummaryrefslogtreecommitdiffstats
path: root/arch/s390
diff options
context:
space:
mode:
Diffstat (limited to 'arch/s390')
-rw-r--r--arch/s390/Kbuild6
-rw-r--r--arch/s390/Kconfig230
-rw-r--r--arch/s390/Kconfig.debug21
-rw-r--r--arch/s390/Makefile8
-rw-r--r--arch/s390/appldata/appldata_base.c2
-rw-r--r--arch/s390/appldata/appldata_mem.c2
-rw-r--r--arch/s390/boot/Makefile2
-rw-r--r--arch/s390/boot/compressed/Makefile6
-rw-r--r--arch/s390/boot/compressed/misc.c10
-rw-r--r--arch/s390/crypto/Makefile1
-rw-r--r--arch/s390/crypto/aes_s390.c383
-rw-r--r--arch/s390/crypto/crypt_s390.h112
-rw-r--r--arch/s390/crypto/des_check_key.c132
-rw-r--r--arch/s390/crypto/des_s390.c370
-rw-r--r--arch/s390/crypto/ghash_s390.c162
-rw-r--r--arch/s390/crypto/prng.c5
-rw-r--r--arch/s390/crypto/sha1_s390.c2
-rw-r--r--arch/s390/crypto/sha256_s390.c2
-rw-r--r--arch/s390/crypto/sha512_s390.c2
-rw-r--r--arch/s390/crypto/sha_common.c1
-rw-r--r--arch/s390/defconfig160
-rw-r--r--arch/s390/hypfs/Makefile2
-rw-r--r--arch/s390/hypfs/hypfs.h33
-rw-r--r--arch/s390/hypfs/hypfs_dbfs.c116
-rw-r--r--arch/s390/hypfs/hypfs_diag.c100
-rw-r--r--arch/s390/hypfs/hypfs_vm.c61
-rw-r--r--arch/s390/hypfs/inode.c27
-rw-r--r--arch/s390/include/asm/Kbuild1
-rw-r--r--arch/s390/include/asm/atomic.h28
-rw-r--r--arch/s390/include/asm/bitops.h40
-rw-r--r--arch/s390/include/asm/cache.h1
-rw-r--r--arch/s390/include/asm/cacheflush.h28
-rw-r--r--arch/s390/include/asm/ccwdev.h18
-rw-r--r--arch/s390/include/asm/ccwgroup.h4
-rw-r--r--arch/s390/include/asm/cio.h2
-rw-r--r--arch/s390/include/asm/cmpxchg.h224
-rw-r--r--arch/s390/include/asm/compat.h2
-rw-r--r--arch/s390/include/asm/cpu.h2
-rw-r--r--arch/s390/include/asm/cputime.h12
-rw-r--r--arch/s390/include/asm/dasd.h43
-rw-r--r--arch/s390/include/asm/delay.h8
-rw-r--r--arch/s390/include/asm/diag.h17
-rw-r--r--arch/s390/include/asm/elf.h19
-rw-r--r--arch/s390/include/asm/ftrace.h11
-rw-r--r--arch/s390/include/asm/futex.h12
-rw-r--r--arch/s390/include/asm/hardirq.h20
-rw-r--r--arch/s390/include/asm/hugetlb.h43
-rw-r--r--arch/s390/include/asm/ioctls.h88
-rw-r--r--arch/s390/include/asm/irq.h43
-rw-r--r--arch/s390/include/asm/irqflags.h51
-rw-r--r--arch/s390/include/asm/jump_label.h37
-rw-r--r--arch/s390/include/asm/kprobes.h20
-rw-r--r--arch/s390/include/asm/kvm_virtio.h7
-rw-r--r--arch/s390/include/asm/lowcore.h15
-rw-r--r--arch/s390/include/asm/mmu.h9
-rw-r--r--arch/s390/include/asm/mmu_context.h8
-rw-r--r--arch/s390/include/asm/mutex.h2
-rw-r--r--arch/s390/include/asm/page.h71
-rw-r--r--arch/s390/include/asm/percpu.h68
-rw-r--r--arch/s390/include/asm/perf_event.h3
-rw-r--r--arch/s390/include/asm/pgalloc.h65
-rw-r--r--arch/s390/include/asm/pgtable.h678
-rw-r--r--arch/s390/include/asm/processor.h12
-rw-r--r--arch/s390/include/asm/ptrace.h55
-rw-r--r--arch/s390/include/asm/qdio.h133
-rw-r--r--arch/s390/include/asm/qeth.h51
-rw-r--r--arch/s390/include/asm/rwsem.h63
-rw-r--r--arch/s390/include/asm/s390_ext.h32
-rw-r--r--arch/s390/include/asm/scatterlist.h2
-rw-r--r--arch/s390/include/asm/setup.h3
-rw-r--r--arch/s390/include/asm/smp.h3
-rw-r--r--arch/s390/include/asm/suspend.h10
-rw-r--r--arch/s390/include/asm/syscall.h4
-rw-r--r--arch/s390/include/asm/sysinfo.h40
-rw-r--r--arch/s390/include/asm/system.h256
-rw-r--r--arch/s390/include/asm/thread_info.h16
-rw-r--r--arch/s390/include/asm/timex.h20
-rw-r--r--arch/s390/include/asm/tlb.h108
-rw-r--r--arch/s390/include/asm/tlbflush.h13
-rw-r--r--arch/s390/include/asm/topology.h29
-rw-r--r--arch/s390/include/asm/types.h8
-rw-r--r--arch/s390/include/asm/uaccess.h15
-rw-r--r--arch/s390/include/asm/unistd.h8
-rw-r--r--arch/s390/kernel/Makefile8
-rw-r--r--arch/s390/kernel/asm-offsets.c27
-rw-r--r--arch/s390/kernel/compat_linux.c1
-rw-r--r--arch/s390/kernel/compat_ptrace.h56
-rw-r--r--arch/s390/kernel/compat_wrapper.S33
-rw-r--r--arch/s390/kernel/debug.c1
-rw-r--r--arch/s390/kernel/diag.c21
-rw-r--r--arch/s390/kernel/dis.c148
-rw-r--r--arch/s390/kernel/early.c69
-rw-r--r--arch/s390/kernel/entry.S322
-rw-r--r--arch/s390/kernel/entry.h6
-rw-r--r--arch/s390/kernel/entry64.S122
-rw-r--r--arch/s390/kernel/ftrace.c238
-rw-r--r--arch/s390/kernel/head.S10
-rw-r--r--arch/s390/kernel/irq.c177
-rw-r--r--arch/s390/kernel/jump_label.c59
-rw-r--r--arch/s390/kernel/kprobes.c523
-rw-r--r--arch/s390/kernel/machine_kexec.c2
-rw-r--r--arch/s390/kernel/mcount.S32
-rw-r--r--arch/s390/kernel/mcount64.S29
-rw-r--r--arch/s390/kernel/mem_detect.c4
-rw-r--r--arch/s390/kernel/nmi.c13
-rw-r--r--arch/s390/kernel/process.c80
-rw-r--r--arch/s390/kernel/processor.c20
-rw-r--r--arch/s390/kernel/ptrace.c309
-rw-r--r--arch/s390/kernel/reipl64.S2
-rw-r--r--arch/s390/kernel/s390_ext.c142
-rw-r--r--arch/s390/kernel/setup.c146
-rw-r--r--arch/s390/kernel/signal.c2
-rw-r--r--arch/s390/kernel/smp.c93
-rw-r--r--arch/s390/kernel/switch_cpu.S4
-rw-r--r--arch/s390/kernel/switch_cpu64.S4
-rw-r--r--arch/s390/kernel/syscalls.S5
-rw-r--r--arch/s390/kernel/sysinfo.c43
-rw-r--r--arch/s390/kernel/time.c34
-rw-r--r--arch/s390/kernel/topology.c250
-rw-r--r--arch/s390/kernel/traps.c222
-rw-r--r--arch/s390/kernel/vdso.c16
-rw-r--r--arch/s390/kernel/vdso32/Makefile3
-rw-r--r--arch/s390/kernel/vdso32/clock_getres.S6
-rw-r--r--arch/s390/kernel/vdso32/clock_gettime.S4
-rw-r--r--arch/s390/kernel/vdso64/Makefile3
-rw-r--r--arch/s390/kernel/vdso64/clock_getres.S6
-rw-r--r--arch/s390/kernel/vdso64/clock_gettime.S4
-rw-r--r--arch/s390/kernel/vmlinux.lds.S2
-rw-r--r--arch/s390/kernel/vtime.c32
-rw-r--r--arch/s390/kvm/Kconfig7
-rw-r--r--arch/s390/kvm/Makefile2
-rw-r--r--arch/s390/kvm/kvm-s390.c30
-rw-r--r--arch/s390/kvm/priv.c6
-rw-r--r--arch/s390/kvm/sie64a.S6
-rw-r--r--arch/s390/lib/delay.c31
-rw-r--r--arch/s390/lib/uaccess.h8
-rw-r--r--arch/s390/lib/uaccess_pt.c17
-rw-r--r--arch/s390/lib/uaccess_std.c18
-rw-r--r--arch/s390/math-emu/Makefile2
-rw-r--r--arch/s390/mm/Makefile3
-rw-r--r--arch/s390/mm/cmm.c9
-rw-r--r--arch/s390/mm/extmem.c6
-rw-r--r--arch/s390/mm/fault.c321
-rw-r--r--arch/s390/mm/gup.c224
-rw-r--r--arch/s390/mm/hugetlbpage.c12
-rw-r--r--arch/s390/mm/init.c60
-rw-r--r--arch/s390/mm/maccess.c8
-rw-r--r--arch/s390/mm/mmap.c49
-rw-r--r--arch/s390/mm/pageattr.c61
-rw-r--r--arch/s390/mm/pgtable.c238
-rw-r--r--arch/s390/mm/vmem.c14
-rw-r--r--arch/s390/oprofile/Makefile3
-rw-r--r--arch/s390/oprofile/hwsampler.c1241
-rw-r--r--arch/s390/oprofile/hwsampler.h113
-rw-r--r--arch/s390/oprofile/init.c176
155 files changed, 6807 insertions, 3730 deletions
diff --git a/arch/s390/Kbuild b/arch/s390/Kbuild
new file mode 100644
index 000000000000..ae4b01060edd
--- /dev/null
+++ b/arch/s390/Kbuild
@@ -0,0 +1,6 @@
1obj-y += kernel/
2obj-y += mm/
3obj-y += crypto/
4obj-y += appldata/
5obj-y += hypfs/
6obj-y += kvm/
diff --git a/arch/s390/Kconfig b/arch/s390/Kconfig
index f0777a47e3a5..c03fef7a9c22 100644
--- a/arch/s390/Kconfig
+++ b/arch/s390/Kconfig
@@ -1,18 +1,8 @@
1#
2# For a description of the syntax of this configuration file,
3# see Documentation/kbuild/kconfig-language.txt.
4#
5
6config SCHED_MC
7 def_bool y
8 depends on SMP
9
10config MMU 1config MMU
11 def_bool y 2 def_bool y
12 3
13config ZONE_DMA 4config ZONE_DMA
14 def_bool y 5 def_bool y
15 depends on 64BIT
16 6
17config LOCKDEP_SUPPORT 7config LOCKDEP_SUPPORT
18 def_bool y 8 def_bool y
@@ -30,12 +20,10 @@ config RWSEM_XCHGADD_ALGORITHM
30 def_bool y 20 def_bool y
31 21
32config ARCH_HAS_ILOG2_U32 22config ARCH_HAS_ILOG2_U32
33 bool 23 def_bool n
34 default n
35 24
36config ARCH_HAS_ILOG2_U64 25config ARCH_HAS_ILOG2_U64
37 bool 26 def_bool n
38 default n
39 27
40config GENERIC_HWEIGHT 28config GENERIC_HWEIGHT
41 def_bool y 29 def_bool y
@@ -47,9 +35,7 @@ config GENERIC_CLOCKEVENTS
47 def_bool y 35 def_bool y
48 36
49config GENERIC_BUG 37config GENERIC_BUG
50 bool 38 def_bool y if BUG
51 depends on BUG
52 default y
53 39
54config GENERIC_BUG_RELATIVE_POINTERS 40config GENERIC_BUG_RELATIVE_POINTERS
55 def_bool y 41 def_bool y
@@ -60,14 +46,14 @@ config NO_IOMEM
60config NO_DMA 46config NO_DMA
61 def_bool y 47 def_bool y
62 48
49config ARCH_DMA_ADDR_T_64BIT
50 def_bool 64BIT
51
63config GENERIC_LOCKBREAK 52config GENERIC_LOCKBREAK
64 bool 53 def_bool y if SMP && PREEMPT
65 default y
66 depends on SMP && PREEMPT
67 54
68config PGSTE 55config PGSTE
69 bool 56 def_bool y if KVM
70 default y if KVM
71 57
72config VIRT_CPU_ACCOUNTING 58config VIRT_CPU_ACCOUNTING
73 def_bool y 59 def_bool y
@@ -75,8 +61,6 @@ config VIRT_CPU_ACCOUNTING
75config ARCH_SUPPORTS_DEBUG_PAGEALLOC 61config ARCH_SUPPORTS_DEBUG_PAGEALLOC
76 def_bool y 62 def_bool y
77 63
78mainmenu "Linux Kernel Configuration"
79
80config S390 64config S390
81 def_bool y 65 def_bool y
82 select USE_GENERIC_SMP_HELPERS if SMP 66 select USE_GENERIC_SMP_HELPERS if SMP
@@ -84,22 +68,28 @@ config S390
84 select HAVE_FUNCTION_TRACER 68 select HAVE_FUNCTION_TRACER
85 select HAVE_FUNCTION_TRACE_MCOUNT_TEST 69 select HAVE_FUNCTION_TRACE_MCOUNT_TEST
86 select HAVE_FTRACE_MCOUNT_RECORD 70 select HAVE_FTRACE_MCOUNT_RECORD
71 select HAVE_C_RECORDMCOUNT
87 select HAVE_SYSCALL_TRACEPOINTS 72 select HAVE_SYSCALL_TRACEPOINTS
88 select HAVE_DYNAMIC_FTRACE 73 select HAVE_DYNAMIC_FTRACE
89 select HAVE_FUNCTION_GRAPH_TRACER 74 select HAVE_FUNCTION_GRAPH_TRACER
90 select HAVE_REGS_AND_STACK_ACCESS_API 75 select HAVE_REGS_AND_STACK_ACCESS_API
91 select HAVE_DEFAULT_NO_SPIN_MUTEXES
92 select HAVE_OPROFILE 76 select HAVE_OPROFILE
93 select HAVE_KPROBES 77 select HAVE_KPROBES
94 select HAVE_KRETPROBES 78 select HAVE_KRETPROBES
95 select HAVE_KVM if 64BIT 79 select HAVE_KVM if 64BIT
96 select HAVE_ARCH_TRACEHOOK 80 select HAVE_ARCH_TRACEHOOK
97 select INIT_ALL_POSSIBLE 81 select INIT_ALL_POSSIBLE
82 select HAVE_IRQ_WORK
98 select HAVE_PERF_EVENTS 83 select HAVE_PERF_EVENTS
99 select HAVE_KERNEL_GZIP 84 select HAVE_KERNEL_GZIP
100 select HAVE_KERNEL_BZIP2 85 select HAVE_KERNEL_BZIP2
101 select HAVE_KERNEL_LZMA 86 select HAVE_KERNEL_LZMA
102 select HAVE_KERNEL_LZO 87 select HAVE_KERNEL_LZO
88 select HAVE_KERNEL_XZ
89 select HAVE_GET_USER_PAGES_FAST
90 select HAVE_ARCH_MUTEX_CPU_RELAX
91 select HAVE_ARCH_JUMP_LABEL if !MARCH_G5
92 select HAVE_RCU_TABLE_FREE if SMP
103 select ARCH_INLINE_SPIN_TRYLOCK 93 select ARCH_INLINE_SPIN_TRYLOCK
104 select ARCH_INLINE_SPIN_TRYLOCK_BH 94 select ARCH_INLINE_SPIN_TRYLOCK_BH
105 select ARCH_INLINE_SPIN_LOCK 95 select ARCH_INLINE_SPIN_LOCK
@@ -130,8 +120,7 @@ config S390
130 select ARCH_INLINE_WRITE_UNLOCK_IRQRESTORE 120 select ARCH_INLINE_WRITE_UNLOCK_IRQRESTORE
131 121
132config SCHED_OMIT_FRAME_POINTER 122config SCHED_OMIT_FRAME_POINTER
133 bool 123 def_bool y
134 default y
135 124
136source "init/Kconfig" 125source "init/Kconfig"
137 126
@@ -144,20 +133,21 @@ comment "Processor type and features"
144source "kernel/time/Kconfig" 133source "kernel/time/Kconfig"
145 134
146config 64BIT 135config 64BIT
147 bool "64 bit kernel" 136 def_bool y
137 prompt "64 bit kernel"
148 help 138 help
149 Select this option if you have a 64 bit IBM zSeries machine 139 Select this option if you have an IBM z/Architecture machine
150 and want to use the 64 bit addressing mode. 140 and want to use the 64 bit addressing mode.
151 141
152config 32BIT 142config 32BIT
153 bool 143 def_bool y if !64BIT
154 default y if !64BIT
155 144
156config KTIME_SCALAR 145config KTIME_SCALAR
157 def_bool 32BIT 146 def_bool 32BIT
158 147
159config SMP 148config SMP
160 bool "Symmetric multi-processing support" 149 def_bool y
150 prompt "Symmetric multi-processing support"
161 ---help--- 151 ---help---
162 This enables support for systems with more than one CPU. If you have 152 This enables support for systems with more than one CPU. If you have
163 a system with only one CPU, like most personal computers, say N. If 153 a system with only one CPU, like most personal computers, say N. If
@@ -189,25 +179,44 @@ config NR_CPUS
189 approximately sixteen kilobytes to the kernel image. 179 approximately sixteen kilobytes to the kernel image.
190 180
191config HOTPLUG_CPU 181config HOTPLUG_CPU
192 bool "Support for hot-pluggable CPUs" 182 def_bool y
183 prompt "Support for hot-pluggable CPUs"
193 depends on SMP 184 depends on SMP
194 select HOTPLUG 185 select HOTPLUG
195 default n
196 help 186 help
197 Say Y here to be able to turn CPUs off and on. CPUs 187 Say Y here to be able to turn CPUs off and on. CPUs
198 can be controlled through /sys/devices/system/cpu/cpu#. 188 can be controlled through /sys/devices/system/cpu/cpu#.
199 Say N if you want to disable CPU hotplug. 189 Say N if you want to disable CPU hotplug.
200 190
191config SCHED_MC
192 def_bool y
193 prompt "Multi-core scheduler support"
194 depends on SMP
195 help
196 Multi-core scheduler support improves the CPU scheduler's decision
197 making when dealing with multi-core CPU chips at a cost of slightly
198 increased overhead in some places.
199
200config SCHED_BOOK
201 def_bool y
202 prompt "Book scheduler support"
203 depends on SMP && SCHED_MC
204 help
205 Book scheduler support improves the CPU scheduler's decision making
206 when dealing with machines that have several books.
207
201config MATHEMU 208config MATHEMU
202 bool "IEEE FPU emulation" 209 def_bool y
210 prompt "IEEE FPU emulation"
203 depends on MARCH_G5 211 depends on MARCH_G5
204 help 212 help
205 This option is required for IEEE compliant floating point arithmetic 213 This option is required for IEEE compliant floating point arithmetic
206 on older S/390 machines. Say Y unless you know your machine doesn't 214 on older ESA/390 machines. Say Y unless you know your machine doesn't
207 need this. 215 need this.
208 216
209config COMPAT 217config COMPAT
210 bool "Kernel support for 31 bit emulation" 218 def_bool y
219 prompt "Kernel support for 31 bit emulation"
211 depends on 64BIT 220 depends on 64BIT
212 select COMPAT_BINFMT_ELF 221 select COMPAT_BINFMT_ELF
213 help 222 help
@@ -217,23 +226,10 @@ config COMPAT
217 executing 31 bit applications. It is safe to say "Y". 226 executing 31 bit applications. It is safe to say "Y".
218 227
219config SYSVIPC_COMPAT 228config SYSVIPC_COMPAT
220 bool 229 def_bool y if COMPAT && SYSVIPC
221 depends on COMPAT && SYSVIPC
222 default y
223 230
224config AUDIT_ARCH 231config AUDIT_ARCH
225 bool 232 def_bool y
226 default y
227
228config S390_EXEC_PROTECT
229 bool "Data execute protection"
230 help
231 This option allows to enable a buffer overflow protection for user
232 space programs and it also selects the addressing mode option above.
233 The kernel parameter noexec=on will enable this feature and also
234 switch the addressing modes, default is disabled. Enabling this (via
235 kernel parameter) on machines earlier than IBM System z9-109 EC/BC
236 will reduce system performance.
237 233
238comment "Code generation options" 234comment "Code generation options"
239 235
@@ -242,46 +238,52 @@ choice
242 default MARCH_G5 238 default MARCH_G5
243 239
244config MARCH_G5 240config MARCH_G5
245 bool "S/390 model G5 and G6" 241 bool "System/390 model G5 and G6"
246 depends on !64BIT 242 depends on !64BIT
247 help 243 help
248 Select this to build a 31 bit kernel that works 244 Select this to build a 31 bit kernel that works
249 on all S/390 and zSeries machines. 245 on all ESA/390 and z/Architecture machines.
250 246
251config MARCH_Z900 247config MARCH_Z900
252 bool "IBM eServer zSeries model z800 and z900" 248 bool "IBM zSeries model z800 and z900"
253 help 249 help
254 Select this to optimize for zSeries machines. This 250 Select this to enable optimizations for model z800/z900 (2064 and
255 will enable some optimizations that are not available 251 2066 series). This will enable some optimizations that are not
256 on older 31 bit only CPUs. 252 available on older ESA/390 (31 Bit) only CPUs.
257 253
258config MARCH_Z990 254config MARCH_Z990
259 bool "IBM eServer zSeries model z890 and z990" 255 bool "IBM zSeries model z890 and z990"
260 help 256 help
261 Select this enable optimizations for model z890/z990. 257 Select this to enable optimizations for model z890/z990 (2084 and
262 This will be slightly faster but does not work on 258 2086 series). The kernel will be slightly faster but will not work
263 older machines such as the z900. 259 on older machines.
264 260
265config MARCH_Z9_109 261config MARCH_Z9_109
266 bool "IBM System z9" 262 bool "IBM System z9"
267 help 263 help
268 Select this to enable optimizations for IBM System z9-109, IBM 264 Select this to enable optimizations for IBM System z9 (2094 and
269 System z9 Enterprise Class (z9 EC), and IBM System z9 Business 265 2096 series). The kernel will be slightly faster but will not work
270 Class (z9 BC). The kernel will be slightly faster but will not 266 on older machines.
271 work on older machines such as the z990, z890, z900, and z800.
272 267
273config MARCH_Z10 268config MARCH_Z10
274 bool "IBM System z10" 269 bool "IBM System z10"
275 help 270 help
276 Select this to enable optimizations for IBM System z10. The 271 Select this to enable optimizations for IBM System z10 (2097 and
277 kernel will be slightly faster but will not work on older 272 2098 series). The kernel will be slightly faster but will not work
278 machines such as the z990, z890, z900, z800, z9-109, z9-ec 273 on older machines.
279 and z9-bc. 274
275config MARCH_Z196
276 bool "IBM zEnterprise 196"
277 help
278 Select this to enable optimizations for IBM zEnterprise 196
279 (2817 series). The kernel will be slightly faster but will not work
280 on older machines.
280 281
281endchoice 282endchoice
282 283
283config PACK_STACK 284config PACK_STACK
284 bool "Pack kernel stack" 285 def_bool y
286 prompt "Pack kernel stack"
285 help 287 help
286 This option enables the compiler option -mkernel-backchain if it 288 This option enables the compiler option -mkernel-backchain if it
287 is available. If the option is available the compiler supports 289 is available. If the option is available the compiler supports
@@ -294,7 +296,8 @@ config PACK_STACK
294 Say Y if you are unsure. 296 Say Y if you are unsure.
295 297
296config SMALL_STACK 298config SMALL_STACK
297 bool "Use 8kb for kernel stack instead of 16kb" 299 def_bool n
300 prompt "Use 8kb for kernel stack instead of 16kb"
298 depends on PACK_STACK && 64BIT && !LOCKDEP 301 depends on PACK_STACK && 64BIT && !LOCKDEP
299 help 302 help
300 If you say Y here and the compiler supports the -mkernel-backchain 303 If you say Y here and the compiler supports the -mkernel-backchain
@@ -306,7 +309,8 @@ config SMALL_STACK
306 Say N if you are unsure. 309 Say N if you are unsure.
307 310
308config CHECK_STACK 311config CHECK_STACK
309 bool "Detect kernel stack overflow" 312 def_bool y
313 prompt "Detect kernel stack overflow"
310 help 314 help
311 This option enables the compiler option -mstack-guard and 315 This option enables the compiler option -mstack-guard and
312 -mstack-size if they are available. If the compiler supports them 316 -mstack-size if they are available. If the compiler supports them
@@ -329,25 +333,16 @@ config STACK_GUARD
329 The minimum size for the stack guard should be 256 for 31 bit and 333 The minimum size for the stack guard should be 256 for 31 bit and
330 512 for 64 bit. 334 512 for 64 bit.
331 335
332config WARN_STACK 336config WARN_DYNAMIC_STACK
333 bool "Emit compiler warnings for function with broken stack usage" 337 def_bool n
338 prompt "Emit compiler warnings for function with dynamic stack usage"
334 help 339 help
335 This option enables the compiler options -mwarn-framesize and 340 This option enables the compiler option -mwarn-dynamicstack. If the
336 -mwarn-dynamicstack. If the compiler supports these options it 341 compiler supports this options generates warnings for functions
337 will generate warnings for function which either use alloca or 342 that dynamically allocate stack space using alloca.
338 create a stack frame bigger than CONFIG_WARN_STACK_SIZE.
339 343
340 Say N if you are unsure. 344 Say N if you are unsure.
341 345
342config WARN_STACK_SIZE
343 int "Maximum frame size considered safe (128-2048)"
344 range 128 2048
345 depends on WARN_STACK
346 default "2048"
347 help
348 This allows you to specify the maximum frame size a function may
349 have without the compiler complaining about it.
350
351config ARCH_POPULATES_NODE_MAP 346config ARCH_POPULATES_NODE_MAP
352 def_bool y 347 def_bool y
353 348
@@ -365,24 +360,24 @@ config ARCH_SPARSEMEM_DEFAULT
365 def_bool y 360 def_bool y
366 361
367config ARCH_SELECT_MEMORY_MODEL 362config ARCH_SELECT_MEMORY_MODEL
368 def_bool y 363 def_bool y
369 364
370config ARCH_ENABLE_MEMORY_HOTPLUG 365config ARCH_ENABLE_MEMORY_HOTPLUG
371 def_bool y 366 def_bool y if SPARSEMEM
372 depends on SPARSEMEM
373 367
374config ARCH_ENABLE_MEMORY_HOTREMOVE 368config ARCH_ENABLE_MEMORY_HOTREMOVE
375 def_bool y 369 def_bool y
376 370
377config ARCH_HIBERNATION_POSSIBLE 371config ARCH_HIBERNATION_POSSIBLE
378 def_bool y if 64BIT 372 def_bool y if 64BIT
379 373
380source "mm/Kconfig" 374source "mm/Kconfig"
381 375
382comment "I/O subsystem configuration" 376comment "I/O subsystem configuration"
383 377
384config QDIO 378config QDIO
385 tristate "QDIO support" 379 def_tristate y
380 prompt "QDIO support"
386 ---help--- 381 ---help---
387 This driver provides the Queued Direct I/O base support for 382 This driver provides the Queued Direct I/O base support for
388 IBM System z. 383 IBM System z.
@@ -393,7 +388,8 @@ config QDIO
393 If unsure, say Y. 388 If unsure, say Y.
394 389
395config CHSC_SCH 390config CHSC_SCH
396 tristate "Support for CHSC subchannels" 391 def_tristate m
392 prompt "Support for CHSC subchannels"
397 help 393 help
398 This driver allows usage of CHSC subchannels. A CHSC subchannel 394 This driver allows usage of CHSC subchannels. A CHSC subchannel
399 is usually present on LPAR only. 395 is usually present on LPAR only.
@@ -411,7 +407,8 @@ config CHSC_SCH
411comment "Misc" 407comment "Misc"
412 408
413config IPL 409config IPL
414 bool "Builtin IPL record support" 410 def_bool y
411 prompt "Builtin IPL record support"
415 help 412 help
416 If you want to use the produced kernel to IPL directly from a 413 If you want to use the produced kernel to IPL directly from a
417 device, you have to merge a bootsector specific to the device 414 device, you have to merge a bootsector specific to the device
@@ -443,7 +440,8 @@ config FORCE_MAX_ZONEORDER
443 default "9" 440 default "9"
444 441
445config PFAULT 442config PFAULT
446 bool "Pseudo page fault support" 443 def_bool y
444 prompt "Pseudo page fault support"
447 help 445 help
448 Select this option, if you want to use PFAULT pseudo page fault 446 Select this option, if you want to use PFAULT pseudo page fault
449 handling under VM. If running native or in LPAR, this option 447 handling under VM. If running native or in LPAR, this option
@@ -455,7 +453,8 @@ config PFAULT
455 this option. 453 this option.
456 454
457config SHARED_KERNEL 455config SHARED_KERNEL
458 bool "VM shared kernel support" 456 def_bool y
457 prompt "VM shared kernel support"
459 help 458 help
460 Select this option, if you want to share the text segment of the 459 Select this option, if you want to share the text segment of the
461 Linux kernel between different VM guests. This reduces memory 460 Linux kernel between different VM guests. This reduces memory
@@ -466,7 +465,8 @@ config SHARED_KERNEL
466 doing and want to exploit this feature. 465 doing and want to exploit this feature.
467 466
468config CMM 467config CMM
469 tristate "Cooperative memory management" 468 def_tristate n
469 prompt "Cooperative memory management"
470 help 470 help
471 Select this option, if you want to enable the kernel interface 471 Select this option, if you want to enable the kernel interface
472 to reduce the memory size of the system. This is accomplished 472 to reduce the memory size of the system. This is accomplished
@@ -478,14 +478,16 @@ config CMM
478 option. 478 option.
479 479
480config CMM_IUCV 480config CMM_IUCV
481 bool "IUCV special message interface to cooperative memory management" 481 def_bool y
482 prompt "IUCV special message interface to cooperative memory management"
482 depends on CMM && (SMSGIUCV=y || CMM=SMSGIUCV) 483 depends on CMM && (SMSGIUCV=y || CMM=SMSGIUCV)
483 help 484 help
484 Select this option to enable the special message interface to 485 Select this option to enable the special message interface to
485 the cooperative memory management. 486 the cooperative memory management.
486 487
487config APPLDATA_BASE 488config APPLDATA_BASE
488 bool "Linux - VM Monitor Stream, base infrastructure" 489 def_bool n
490 prompt "Linux - VM Monitor Stream, base infrastructure"
489 depends on PROC_FS 491 depends on PROC_FS
490 help 492 help
491 This provides a kernel interface for creating and updating z/VM APPLDATA 493 This provides a kernel interface for creating and updating z/VM APPLDATA
@@ -500,7 +502,8 @@ config APPLDATA_BASE
500 The /proc entries can also be read from, showing the current settings. 502 The /proc entries can also be read from, showing the current settings.
501 503
502config APPLDATA_MEM 504config APPLDATA_MEM
503 tristate "Monitor memory management statistics" 505 def_tristate m
506 prompt "Monitor memory management statistics"
504 depends on APPLDATA_BASE && VM_EVENT_COUNTERS 507 depends on APPLDATA_BASE && VM_EVENT_COUNTERS
505 help 508 help
506 This provides memory management related data to the Linux - VM Monitor 509 This provides memory management related data to the Linux - VM Monitor
@@ -516,7 +519,8 @@ config APPLDATA_MEM
516 appldata_mem.o. 519 appldata_mem.o.
517 520
518config APPLDATA_OS 521config APPLDATA_OS
519 tristate "Monitor OS statistics" 522 def_tristate m
523 prompt "Monitor OS statistics"
520 depends on APPLDATA_BASE 524 depends on APPLDATA_BASE
521 help 525 help
522 This provides OS related data to the Linux - VM Monitor Stream, like 526 This provides OS related data to the Linux - VM Monitor Stream, like
@@ -530,7 +534,8 @@ config APPLDATA_OS
530 appldata_os.o. 534 appldata_os.o.
531 535
532config APPLDATA_NET_SUM 536config APPLDATA_NET_SUM
533 tristate "Monitor overall network statistics" 537 def_tristate m
538 prompt "Monitor overall network statistics"
534 depends on APPLDATA_BASE && NET 539 depends on APPLDATA_BASE && NET
535 help 540 help
536 This provides network related data to the Linux - VM Monitor Stream, 541 This provides network related data to the Linux - VM Monitor Stream,
@@ -547,31 +552,34 @@ config APPLDATA_NET_SUM
547source kernel/Kconfig.hz 552source kernel/Kconfig.hz
548 553
549config S390_HYPFS_FS 554config S390_HYPFS_FS
550 bool "s390 hypervisor file system support" 555 def_bool y
556 prompt "s390 hypervisor file system support"
551 select SYS_HYPERVISOR 557 select SYS_HYPERVISOR
552 default y
553 help 558 help
554 This is a virtual file system intended to provide accounting 559 This is a virtual file system intended to provide accounting
555 information in an s390 hypervisor environment. 560 information in an s390 hypervisor environment.
556 561
557config KEXEC 562config KEXEC
558 bool "kexec system call" 563 def_bool n
564 prompt "kexec system call"
559 help 565 help
560 kexec is a system call that implements the ability to shutdown your 566 kexec is a system call that implements the ability to shutdown your
561 current kernel, and to start another kernel. It is like a reboot 567 current kernel, and to start another kernel. It is like a reboot
562 but is independent of hardware/microcode support. 568 but is independent of hardware/microcode support.
563 569
564config ZFCPDUMP 570config ZFCPDUMP
565 bool "zfcpdump support" 571 def_bool n
572 prompt "zfcpdump support"
566 select SMP 573 select SMP
567 default n
568 help 574 help
569 Select this option if you want to build an zfcpdump enabled kernel. 575 Select this option if you want to build an zfcpdump enabled kernel.
570 Refer to <file:Documentation/s390/zfcpdump.txt> for more details on this. 576 Refer to <file:Documentation/s390/zfcpdump.txt> for more details on this.
571 577
572config S390_GUEST 578config S390_GUEST
573bool "s390 guest support for KVM (EXPERIMENTAL)" 579 def_bool y
580 prompt "s390 guest support for KVM (EXPERIMENTAL)"
574 depends on 64BIT && EXPERIMENTAL 581 depends on 64BIT && EXPERIMENTAL
582 select VIRTUALIZATION
575 select VIRTIO 583 select VIRTIO
576 select VIRTIO_RING 584 select VIRTIO_RING
577 select VIRTIO_CONSOLE 585 select VIRTIO_CONSOLE
@@ -582,9 +590,9 @@ bool "s390 guest support for KVM (EXPERIMENTAL)"
582 the default console. 590 the default console.
583 591
584config SECCOMP 592config SECCOMP
585 bool "Enable seccomp to safely compute untrusted bytecode" 593 def_bool y
594 prompt "Enable seccomp to safely compute untrusted bytecode"
586 depends on PROC_FS 595 depends on PROC_FS
587 default y
588 help 596 help
589 This kernel feature is useful for number crunching applications 597 This kernel feature is useful for number crunching applications
590 that may need to compute untrusted bytecode during their 598 that may need to compute untrusted bytecode during their
diff --git a/arch/s390/Kconfig.debug b/arch/s390/Kconfig.debug
index 45e0c6199f36..d76cef3fef37 100644
--- a/arch/s390/Kconfig.debug
+++ b/arch/s390/Kconfig.debug
@@ -1,13 +1,25 @@
1menu "Kernel hacking" 1menu "Kernel hacking"
2 2
3config TRACE_IRQFLAGS_SUPPORT 3config TRACE_IRQFLAGS_SUPPORT
4 bool 4 def_bool y
5 default y
6 5
7source "lib/Kconfig.debug" 6source "lib/Kconfig.debug"
8 7
8config STRICT_DEVMEM
9 def_bool y
10 prompt "Filter access to /dev/mem"
11 ---help---
12 This option restricts access to /dev/mem. If this option is
13 disabled, you allow userspace access to all memory, including
14 kernel and userspace memory. Accidental memory access is likely
15 to be disastrous.
16 Memory access is required for experts who want to debug the kernel.
17
18 If you are unsure, say Y.
19
9config DEBUG_STRICT_USER_COPY_CHECKS 20config DEBUG_STRICT_USER_COPY_CHECKS
10 bool "Strict user copy size checks" 21 def_bool n
22 prompt "Strict user copy size checks"
11 ---help--- 23 ---help---
12 Enabling this option turns a certain set of sanity checks for user 24 Enabling this option turns a certain set of sanity checks for user
13 copy operations into compile time warnings. 25 copy operations into compile time warnings.
@@ -19,4 +31,7 @@ config DEBUG_STRICT_USER_COPY_CHECKS
19 31
20 If unsure, or if you run an older (pre 4.4) gcc, say N. 32 If unsure, or if you run an older (pre 4.4) gcc, say N.
21 33
34config DEBUG_SET_MODULE_RONX
35 def_bool y
36 depends on MODULES
22endmenu 37endmenu
diff --git a/arch/s390/Makefile b/arch/s390/Makefile
index 0c9e6c6d2a64..27a0b5df5ead 100644
--- a/arch/s390/Makefile
+++ b/arch/s390/Makefile
@@ -40,6 +40,7 @@ cflags-$(CONFIG_MARCH_Z900) += $(call cc-option,-march=z900)
40cflags-$(CONFIG_MARCH_Z990) += $(call cc-option,-march=z990) 40cflags-$(CONFIG_MARCH_Z990) += $(call cc-option,-march=z990)
41cflags-$(CONFIG_MARCH_Z9_109) += $(call cc-option,-march=z9-109) 41cflags-$(CONFIG_MARCH_Z9_109) += $(call cc-option,-march=z9-109)
42cflags-$(CONFIG_MARCH_Z10) += $(call cc-option,-march=z10) 42cflags-$(CONFIG_MARCH_Z10) += $(call cc-option,-march=z10)
43cflags-$(CONFIG_MARCH_Z196) += $(call cc-option,-march=z196)
43 44
44#KBUILD_IMAGE is necessary for make rpm 45#KBUILD_IMAGE is necessary for make rpm
45KBUILD_IMAGE :=arch/s390/boot/image 46KBUILD_IMAGE :=arch/s390/boot/image
@@ -79,8 +80,7 @@ endif
79endif 80endif
80 81
81ifeq ($(call cc-option-yn,-mwarn-dynamicstack),y) 82ifeq ($(call cc-option-yn,-mwarn-dynamicstack),y)
82cflags-$(CONFIG_WARN_STACK) += -mwarn-dynamicstack 83cflags-$(CONFIG_WARN_DYNAMIC_STACK) += -mwarn-dynamicstack
83cflags-$(CONFIG_WARN_STACK) += -mwarn-framesize=$(CONFIG_WARN_STACK_SIZE)
84endif 84endif
85 85
86KBUILD_CFLAGS += -mbackchain -msoft-float $(cflags-y) 86KBUILD_CFLAGS += -mbackchain -msoft-float $(cflags-y)
@@ -94,8 +94,8 @@ head-y := arch/s390/kernel/head.o
94head-y += arch/s390/kernel/$(if $(CONFIG_64BIT),head64.o,head31.o) 94head-y += arch/s390/kernel/$(if $(CONFIG_64BIT),head64.o,head31.o)
95head-y += arch/s390/kernel/init_task.o 95head-y += arch/s390/kernel/init_task.o
96 96
97core-y += arch/s390/mm/ arch/s390/kernel/ arch/s390/crypto/ \ 97# See arch/s390/Kbuild for content of core part of the kernel
98 arch/s390/appldata/ arch/s390/hypfs/ arch/s390/kvm/ 98core-y += arch/s390/
99 99
100libs-y += arch/s390/lib/ 100libs-y += arch/s390/lib/
101drivers-y += drivers/s390/ 101drivers-y += drivers/s390/
diff --git a/arch/s390/appldata/appldata_base.c b/arch/s390/appldata/appldata_base.c
index 5c91995b74e4..24bff4f1cc52 100644
--- a/arch/s390/appldata/appldata_base.c
+++ b/arch/s390/appldata/appldata_base.c
@@ -130,9 +130,7 @@ static void appldata_work_fn(struct work_struct *work)
130{ 130{
131 struct list_head *lh; 131 struct list_head *lh;
132 struct appldata_ops *ops; 132 struct appldata_ops *ops;
133 int i;
134 133
135 i = 0;
136 get_online_cpus(); 134 get_online_cpus();
137 mutex_lock(&appldata_ops_mutex); 135 mutex_lock(&appldata_ops_mutex);
138 list_for_each(lh, &appldata_ops_list) { 136 list_for_each(lh, &appldata_ops_list) {
diff --git a/arch/s390/appldata/appldata_mem.c b/arch/s390/appldata/appldata_mem.c
index e43fe7537031..f7d3dc555bdb 100644
--- a/arch/s390/appldata/appldata_mem.c
+++ b/arch/s390/appldata/appldata_mem.c
@@ -92,9 +92,7 @@ static void appldata_get_mem_data(void *data)
92 mem_data->pswpin = ev[PSWPIN]; 92 mem_data->pswpin = ev[PSWPIN];
93 mem_data->pswpout = ev[PSWPOUT]; 93 mem_data->pswpout = ev[PSWPOUT];
94 mem_data->pgalloc = ev[PGALLOC_NORMAL]; 94 mem_data->pgalloc = ev[PGALLOC_NORMAL];
95#ifdef CONFIG_ZONE_DMA
96 mem_data->pgalloc += ev[PGALLOC_DMA]; 95 mem_data->pgalloc += ev[PGALLOC_DMA];
97#endif
98 mem_data->pgfault = ev[PGFAULT]; 96 mem_data->pgfault = ev[PGFAULT];
99 mem_data->pgmajfault = ev[PGMAJFAULT]; 97 mem_data->pgmajfault = ev[PGMAJFAULT];
100 98
diff --git a/arch/s390/boot/Makefile b/arch/s390/boot/Makefile
index 8800cf090694..635d677d3281 100644
--- a/arch/s390/boot/Makefile
+++ b/arch/s390/boot/Makefile
@@ -6,7 +6,7 @@ COMPILE_VERSION := __linux_compile_version_id__`hostname | \
6 tr -c '[0-9A-Za-z]' '_'`__`date | \ 6 tr -c '[0-9A-Za-z]' '_'`__`date | \
7 tr -c '[0-9A-Za-z]' '_'`_t 7 tr -c '[0-9A-Za-z]' '_'`_t
8 8
9EXTRA_CFLAGS := -DCOMPILE_VERSION=$(COMPILE_VERSION) -gstabs -I. 9ccflags-y := -DCOMPILE_VERSION=$(COMPILE_VERSION) -gstabs -I.
10 10
11targets := image 11targets := image
12targets += bzImage 12targets += bzImage
diff --git a/arch/s390/boot/compressed/Makefile b/arch/s390/boot/compressed/Makefile
index 1c999f726a58..10e22c4ec4a7 100644
--- a/arch/s390/boot/compressed/Makefile
+++ b/arch/s390/boot/compressed/Makefile
@@ -7,7 +7,8 @@
7BITS := $(if $(CONFIG_64BIT),64,31) 7BITS := $(if $(CONFIG_64BIT),64,31)
8 8
9targets := vmlinux.lds vmlinux vmlinux.bin vmlinux.bin.gz vmlinux.bin.bz2 \ 9targets := vmlinux.lds vmlinux vmlinux.bin vmlinux.bin.gz vmlinux.bin.bz2 \
10 vmlinux.bin.lzma vmlinux.bin.lzo misc.o piggy.o sizes.h head$(BITS).o 10 vmlinux.bin.xz vmlinux.bin.lzma vmlinux.bin.lzo misc.o piggy.o \
11 sizes.h head$(BITS).o
11 12
12KBUILD_CFLAGS := -m$(BITS) -D__KERNEL__ $(LINUX_INCLUDE) -O2 13KBUILD_CFLAGS := -m$(BITS) -D__KERNEL__ $(LINUX_INCLUDE) -O2
13KBUILD_CFLAGS += $(cflags-y) 14KBUILD_CFLAGS += $(cflags-y)
@@ -48,6 +49,7 @@ suffix-$(CONFIG_KERNEL_GZIP) := gz
48suffix-$(CONFIG_KERNEL_BZIP2) := bz2 49suffix-$(CONFIG_KERNEL_BZIP2) := bz2
49suffix-$(CONFIG_KERNEL_LZMA) := lzma 50suffix-$(CONFIG_KERNEL_LZMA) := lzma
50suffix-$(CONFIG_KERNEL_LZO) := lzo 51suffix-$(CONFIG_KERNEL_LZO) := lzo
52suffix-$(CONFIG_KERNEL_XZ) := xz
51 53
52$(obj)/vmlinux.bin.gz: $(vmlinux.bin.all-y) 54$(obj)/vmlinux.bin.gz: $(vmlinux.bin.all-y)
53 $(call if_changed,gzip) 55 $(call if_changed,gzip)
@@ -57,6 +59,8 @@ $(obj)/vmlinux.bin.lzma: $(vmlinux.bin.all-y)
57 $(call if_changed,lzma) 59 $(call if_changed,lzma)
58$(obj)/vmlinux.bin.lzo: $(vmlinux.bin.all-y) 60$(obj)/vmlinux.bin.lzo: $(vmlinux.bin.all-y)
59 $(call if_changed,lzo) 61 $(call if_changed,lzo)
62$(obj)/vmlinux.bin.xz: $(vmlinux.bin.all-y)
63 $(call if_changed,xzkern)
60 64
61LDFLAGS_piggy.o := -r --format binary --oformat $(LD_BFD) -T 65LDFLAGS_piggy.o := -r --format binary --oformat $(LD_BFD) -T
62$(obj)/piggy.o: $(obj)/vmlinux.scr $(obj)/vmlinux.bin.$(suffix-y) 66$(obj)/piggy.o: $(obj)/vmlinux.scr $(obj)/vmlinux.bin.$(suffix-y)
diff --git a/arch/s390/boot/compressed/misc.c b/arch/s390/boot/compressed/misc.c
index 0851eb1e919e..028f23ea81d1 100644
--- a/arch/s390/boot/compressed/misc.c
+++ b/arch/s390/boot/compressed/misc.c
@@ -19,6 +19,7 @@
19#undef memset 19#undef memset
20#undef memcpy 20#undef memcpy
21#undef memmove 21#undef memmove
22#define memmove memmove
22#define memzero(s, n) memset((s), 0, (n)) 23#define memzero(s, n) memset((s), 0, (n))
23 24
24/* Symbols defined by linker scripts */ 25/* Symbols defined by linker scripts */
@@ -54,6 +55,10 @@ static unsigned long free_mem_end_ptr;
54#include "../../../../lib/decompress_unlzo.c" 55#include "../../../../lib/decompress_unlzo.c"
55#endif 56#endif
56 57
58#ifdef CONFIG_KERNEL_XZ
59#include "../../../../lib/decompress_unxz.c"
60#endif
61
57extern _sclp_print_early(const char *); 62extern _sclp_print_early(const char *);
58 63
59int puts(const char *s) 64int puts(const char *s)
@@ -133,11 +138,12 @@ unsigned long decompress_kernel(void)
133 unsigned long output_addr; 138 unsigned long output_addr;
134 unsigned char *output; 139 unsigned char *output;
135 140
136 check_ipl_parmblock((void *) 0, (unsigned long) output + SZ__bss_start); 141 output_addr = ((unsigned long) &_end + HEAP_SIZE + 4095UL) & -4096UL;
142 check_ipl_parmblock((void *) 0, output_addr + SZ__bss_start);
137 memset(&_bss, 0, &_ebss - &_bss); 143 memset(&_bss, 0, &_ebss - &_bss);
138 free_mem_ptr = (unsigned long)&_end; 144 free_mem_ptr = (unsigned long)&_end;
139 free_mem_end_ptr = free_mem_ptr + HEAP_SIZE; 145 free_mem_end_ptr = free_mem_ptr + HEAP_SIZE;
140 output = (unsigned char *) ((free_mem_end_ptr + 4095UL) & -4096UL); 146 output = (unsigned char *) output_addr;
141 147
142#ifdef CONFIG_BLK_DEV_INITRD 148#ifdef CONFIG_BLK_DEV_INITRD
143 /* 149 /*
diff --git a/arch/s390/crypto/Makefile b/arch/s390/crypto/Makefile
index 1cf81d77c5a5..7f0b7cda6259 100644
--- a/arch/s390/crypto/Makefile
+++ b/arch/s390/crypto/Makefile
@@ -8,3 +8,4 @@ obj-$(CONFIG_CRYPTO_SHA512_S390) += sha512_s390.o sha_common.o
8obj-$(CONFIG_CRYPTO_DES_S390) += des_s390.o 8obj-$(CONFIG_CRYPTO_DES_S390) += des_s390.o
9obj-$(CONFIG_CRYPTO_AES_S390) += aes_s390.o 9obj-$(CONFIG_CRYPTO_AES_S390) += aes_s390.o
10obj-$(CONFIG_S390_PRNG) += prng.o 10obj-$(CONFIG_S390_PRNG) += prng.o
11obj-$(CONFIG_CRYPTO_GHASH_S390) += ghash_s390.o
diff --git a/arch/s390/crypto/aes_s390.c b/arch/s390/crypto/aes_s390.c
index 58f46734465f..a9ce135893f8 100644
--- a/arch/s390/crypto/aes_s390.c
+++ b/arch/s390/crypto/aes_s390.c
@@ -31,7 +31,8 @@
31#define AES_KEYLEN_192 2 31#define AES_KEYLEN_192 2
32#define AES_KEYLEN_256 4 32#define AES_KEYLEN_256 4
33 33
34static char keylen_flag = 0; 34static u8 *ctrblk;
35static char keylen_flag;
35 36
36struct s390_aes_ctx { 37struct s390_aes_ctx {
37 u8 iv[AES_BLOCK_SIZE]; 38 u8 iv[AES_BLOCK_SIZE];
@@ -45,6 +46,24 @@ struct s390_aes_ctx {
45 } fallback; 46 } fallback;
46}; 47};
47 48
49struct pcc_param {
50 u8 key[32];
51 u8 tweak[16];
52 u8 block[16];
53 u8 bit[16];
54 u8 xts[16];
55};
56
57struct s390_xts_ctx {
58 u8 key[32];
59 u8 xts_param[16];
60 struct pcc_param pcc;
61 long enc;
62 long dec;
63 int key_len;
64 struct crypto_blkcipher *fallback;
65};
66
48/* 67/*
49 * Check if the key_len is supported by the HW. 68 * Check if the key_len is supported by the HW.
50 * Returns 0 if it is, a positive number if it is not and software fallback is 69 * Returns 0 if it is, a positive number if it is not and software fallback is
@@ -504,15 +523,337 @@ static struct crypto_alg cbc_aes_alg = {
504 } 523 }
505}; 524};
506 525
526static int xts_fallback_setkey(struct crypto_tfm *tfm, const u8 *key,
527 unsigned int len)
528{
529 struct s390_xts_ctx *xts_ctx = crypto_tfm_ctx(tfm);
530 unsigned int ret;
531
532 xts_ctx->fallback->base.crt_flags &= ~CRYPTO_TFM_REQ_MASK;
533 xts_ctx->fallback->base.crt_flags |= (tfm->crt_flags &
534 CRYPTO_TFM_REQ_MASK);
535
536 ret = crypto_blkcipher_setkey(xts_ctx->fallback, key, len);
537 if (ret) {
538 tfm->crt_flags &= ~CRYPTO_TFM_RES_MASK;
539 tfm->crt_flags |= (xts_ctx->fallback->base.crt_flags &
540 CRYPTO_TFM_RES_MASK);
541 }
542 return ret;
543}
544
545static int xts_fallback_decrypt(struct blkcipher_desc *desc,
546 struct scatterlist *dst, struct scatterlist *src,
547 unsigned int nbytes)
548{
549 struct s390_xts_ctx *xts_ctx = crypto_blkcipher_ctx(desc->tfm);
550 struct crypto_blkcipher *tfm;
551 unsigned int ret;
552
553 tfm = desc->tfm;
554 desc->tfm = xts_ctx->fallback;
555
556 ret = crypto_blkcipher_decrypt_iv(desc, dst, src, nbytes);
557
558 desc->tfm = tfm;
559 return ret;
560}
561
562static int xts_fallback_encrypt(struct blkcipher_desc *desc,
563 struct scatterlist *dst, struct scatterlist *src,
564 unsigned int nbytes)
565{
566 struct s390_xts_ctx *xts_ctx = crypto_blkcipher_ctx(desc->tfm);
567 struct crypto_blkcipher *tfm;
568 unsigned int ret;
569
570 tfm = desc->tfm;
571 desc->tfm = xts_ctx->fallback;
572
573 ret = crypto_blkcipher_encrypt_iv(desc, dst, src, nbytes);
574
575 desc->tfm = tfm;
576 return ret;
577}
578
579static int xts_aes_set_key(struct crypto_tfm *tfm, const u8 *in_key,
580 unsigned int key_len)
581{
582 struct s390_xts_ctx *xts_ctx = crypto_tfm_ctx(tfm);
583 u32 *flags = &tfm->crt_flags;
584
585 switch (key_len) {
586 case 32:
587 xts_ctx->enc = KM_XTS_128_ENCRYPT;
588 xts_ctx->dec = KM_XTS_128_DECRYPT;
589 memcpy(xts_ctx->key + 16, in_key, 16);
590 memcpy(xts_ctx->pcc.key + 16, in_key + 16, 16);
591 break;
592 case 48:
593 xts_ctx->enc = 0;
594 xts_ctx->dec = 0;
595 xts_fallback_setkey(tfm, in_key, key_len);
596 break;
597 case 64:
598 xts_ctx->enc = KM_XTS_256_ENCRYPT;
599 xts_ctx->dec = KM_XTS_256_DECRYPT;
600 memcpy(xts_ctx->key, in_key, 32);
601 memcpy(xts_ctx->pcc.key, in_key + 32, 32);
602 break;
603 default:
604 *flags |= CRYPTO_TFM_RES_BAD_KEY_LEN;
605 return -EINVAL;
606 }
607 xts_ctx->key_len = key_len;
608 return 0;
609}
610
611static int xts_aes_crypt(struct blkcipher_desc *desc, long func,
612 struct s390_xts_ctx *xts_ctx,
613 struct blkcipher_walk *walk)
614{
615 unsigned int offset = (xts_ctx->key_len >> 1) & 0x10;
616 int ret = blkcipher_walk_virt(desc, walk);
617 unsigned int nbytes = walk->nbytes;
618 unsigned int n;
619 u8 *in, *out;
620 void *param;
621
622 if (!nbytes)
623 goto out;
624
625 memset(xts_ctx->pcc.block, 0, sizeof(xts_ctx->pcc.block));
626 memset(xts_ctx->pcc.bit, 0, sizeof(xts_ctx->pcc.bit));
627 memset(xts_ctx->pcc.xts, 0, sizeof(xts_ctx->pcc.xts));
628 memcpy(xts_ctx->pcc.tweak, walk->iv, sizeof(xts_ctx->pcc.tweak));
629 param = xts_ctx->pcc.key + offset;
630 ret = crypt_s390_pcc(func, param);
631 BUG_ON(ret < 0);
632
633 memcpy(xts_ctx->xts_param, xts_ctx->pcc.xts, 16);
634 param = xts_ctx->key + offset;
635 do {
636 /* only use complete blocks */
637 n = nbytes & ~(AES_BLOCK_SIZE - 1);
638 out = walk->dst.virt.addr;
639 in = walk->src.virt.addr;
640
641 ret = crypt_s390_km(func, param, out, in, n);
642 BUG_ON(ret < 0 || ret != n);
643
644 nbytes &= AES_BLOCK_SIZE - 1;
645 ret = blkcipher_walk_done(desc, walk, nbytes);
646 } while ((nbytes = walk->nbytes));
647out:
648 return ret;
649}
650
651static int xts_aes_encrypt(struct blkcipher_desc *desc,
652 struct scatterlist *dst, struct scatterlist *src,
653 unsigned int nbytes)
654{
655 struct s390_xts_ctx *xts_ctx = crypto_blkcipher_ctx(desc->tfm);
656 struct blkcipher_walk walk;
657
658 if (unlikely(xts_ctx->key_len == 48))
659 return xts_fallback_encrypt(desc, dst, src, nbytes);
660
661 blkcipher_walk_init(&walk, dst, src, nbytes);
662 return xts_aes_crypt(desc, xts_ctx->enc, xts_ctx, &walk);
663}
664
665static int xts_aes_decrypt(struct blkcipher_desc *desc,
666 struct scatterlist *dst, struct scatterlist *src,
667 unsigned int nbytes)
668{
669 struct s390_xts_ctx *xts_ctx = crypto_blkcipher_ctx(desc->tfm);
670 struct blkcipher_walk walk;
671
672 if (unlikely(xts_ctx->key_len == 48))
673 return xts_fallback_decrypt(desc, dst, src, nbytes);
674
675 blkcipher_walk_init(&walk, dst, src, nbytes);
676 return xts_aes_crypt(desc, xts_ctx->dec, xts_ctx, &walk);
677}
678
679static int xts_fallback_init(struct crypto_tfm *tfm)
680{
681 const char *name = tfm->__crt_alg->cra_name;
682 struct s390_xts_ctx *xts_ctx = crypto_tfm_ctx(tfm);
683
684 xts_ctx->fallback = crypto_alloc_blkcipher(name, 0,
685 CRYPTO_ALG_ASYNC | CRYPTO_ALG_NEED_FALLBACK);
686
687 if (IS_ERR(xts_ctx->fallback)) {
688 pr_err("Allocating XTS fallback algorithm %s failed\n",
689 name);
690 return PTR_ERR(xts_ctx->fallback);
691 }
692 return 0;
693}
694
695static void xts_fallback_exit(struct crypto_tfm *tfm)
696{
697 struct s390_xts_ctx *xts_ctx = crypto_tfm_ctx(tfm);
698
699 crypto_free_blkcipher(xts_ctx->fallback);
700 xts_ctx->fallback = NULL;
701}
702
703static struct crypto_alg xts_aes_alg = {
704 .cra_name = "xts(aes)",
705 .cra_driver_name = "xts-aes-s390",
706 .cra_priority = CRYPT_S390_COMPOSITE_PRIORITY,
707 .cra_flags = CRYPTO_ALG_TYPE_BLKCIPHER |
708 CRYPTO_ALG_NEED_FALLBACK,
709 .cra_blocksize = AES_BLOCK_SIZE,
710 .cra_ctxsize = sizeof(struct s390_xts_ctx),
711 .cra_type = &crypto_blkcipher_type,
712 .cra_module = THIS_MODULE,
713 .cra_list = LIST_HEAD_INIT(xts_aes_alg.cra_list),
714 .cra_init = xts_fallback_init,
715 .cra_exit = xts_fallback_exit,
716 .cra_u = {
717 .blkcipher = {
718 .min_keysize = 2 * AES_MIN_KEY_SIZE,
719 .max_keysize = 2 * AES_MAX_KEY_SIZE,
720 .ivsize = AES_BLOCK_SIZE,
721 .setkey = xts_aes_set_key,
722 .encrypt = xts_aes_encrypt,
723 .decrypt = xts_aes_decrypt,
724 }
725 }
726};
727
728static int ctr_aes_set_key(struct crypto_tfm *tfm, const u8 *in_key,
729 unsigned int key_len)
730{
731 struct s390_aes_ctx *sctx = crypto_tfm_ctx(tfm);
732
733 switch (key_len) {
734 case 16:
735 sctx->enc = KMCTR_AES_128_ENCRYPT;
736 sctx->dec = KMCTR_AES_128_DECRYPT;
737 break;
738 case 24:
739 sctx->enc = KMCTR_AES_192_ENCRYPT;
740 sctx->dec = KMCTR_AES_192_DECRYPT;
741 break;
742 case 32:
743 sctx->enc = KMCTR_AES_256_ENCRYPT;
744 sctx->dec = KMCTR_AES_256_DECRYPT;
745 break;
746 }
747
748 return aes_set_key(tfm, in_key, key_len);
749}
750
751static int ctr_aes_crypt(struct blkcipher_desc *desc, long func,
752 struct s390_aes_ctx *sctx, struct blkcipher_walk *walk)
753{
754 int ret = blkcipher_walk_virt_block(desc, walk, AES_BLOCK_SIZE);
755 unsigned int i, n, nbytes;
756 u8 buf[AES_BLOCK_SIZE];
757 u8 *out, *in;
758
759 if (!walk->nbytes)
760 return ret;
761
762 memcpy(ctrblk, walk->iv, AES_BLOCK_SIZE);
763 while ((nbytes = walk->nbytes) >= AES_BLOCK_SIZE) {
764 out = walk->dst.virt.addr;
765 in = walk->src.virt.addr;
766 while (nbytes >= AES_BLOCK_SIZE) {
767 /* only use complete blocks, max. PAGE_SIZE */
768 n = (nbytes > PAGE_SIZE) ? PAGE_SIZE :
769 nbytes & ~(AES_BLOCK_SIZE - 1);
770 for (i = AES_BLOCK_SIZE; i < n; i += AES_BLOCK_SIZE) {
771 memcpy(ctrblk + i, ctrblk + i - AES_BLOCK_SIZE,
772 AES_BLOCK_SIZE);
773 crypto_inc(ctrblk + i, AES_BLOCK_SIZE);
774 }
775 ret = crypt_s390_kmctr(func, sctx->key, out, in, n, ctrblk);
776 BUG_ON(ret < 0 || ret != n);
777 if (n > AES_BLOCK_SIZE)
778 memcpy(ctrblk, ctrblk + n - AES_BLOCK_SIZE,
779 AES_BLOCK_SIZE);
780 crypto_inc(ctrblk, AES_BLOCK_SIZE);
781 out += n;
782 in += n;
783 nbytes -= n;
784 }
785 ret = blkcipher_walk_done(desc, walk, nbytes);
786 }
787 /*
788 * final block may be < AES_BLOCK_SIZE, copy only nbytes
789 */
790 if (nbytes) {
791 out = walk->dst.virt.addr;
792 in = walk->src.virt.addr;
793 ret = crypt_s390_kmctr(func, sctx->key, buf, in,
794 AES_BLOCK_SIZE, ctrblk);
795 BUG_ON(ret < 0 || ret != AES_BLOCK_SIZE);
796 memcpy(out, buf, nbytes);
797 crypto_inc(ctrblk, AES_BLOCK_SIZE);
798 ret = blkcipher_walk_done(desc, walk, 0);
799 }
800 memcpy(walk->iv, ctrblk, AES_BLOCK_SIZE);
801 return ret;
802}
803
804static int ctr_aes_encrypt(struct blkcipher_desc *desc,
805 struct scatterlist *dst, struct scatterlist *src,
806 unsigned int nbytes)
807{
808 struct s390_aes_ctx *sctx = crypto_blkcipher_ctx(desc->tfm);
809 struct blkcipher_walk walk;
810
811 blkcipher_walk_init(&walk, dst, src, nbytes);
812 return ctr_aes_crypt(desc, sctx->enc, sctx, &walk);
813}
814
815static int ctr_aes_decrypt(struct blkcipher_desc *desc,
816 struct scatterlist *dst, struct scatterlist *src,
817 unsigned int nbytes)
818{
819 struct s390_aes_ctx *sctx = crypto_blkcipher_ctx(desc->tfm);
820 struct blkcipher_walk walk;
821
822 blkcipher_walk_init(&walk, dst, src, nbytes);
823 return ctr_aes_crypt(desc, sctx->dec, sctx, &walk);
824}
825
826static struct crypto_alg ctr_aes_alg = {
827 .cra_name = "ctr(aes)",
828 .cra_driver_name = "ctr-aes-s390",
829 .cra_priority = CRYPT_S390_COMPOSITE_PRIORITY,
830 .cra_flags = CRYPTO_ALG_TYPE_BLKCIPHER,
831 .cra_blocksize = 1,
832 .cra_ctxsize = sizeof(struct s390_aes_ctx),
833 .cra_type = &crypto_blkcipher_type,
834 .cra_module = THIS_MODULE,
835 .cra_list = LIST_HEAD_INIT(ctr_aes_alg.cra_list),
836 .cra_u = {
837 .blkcipher = {
838 .min_keysize = AES_MIN_KEY_SIZE,
839 .max_keysize = AES_MAX_KEY_SIZE,
840 .ivsize = AES_BLOCK_SIZE,
841 .setkey = ctr_aes_set_key,
842 .encrypt = ctr_aes_encrypt,
843 .decrypt = ctr_aes_decrypt,
844 }
845 }
846};
847
507static int __init aes_s390_init(void) 848static int __init aes_s390_init(void)
508{ 849{
509 int ret; 850 int ret;
510 851
511 if (crypt_s390_func_available(KM_AES_128_ENCRYPT)) 852 if (crypt_s390_func_available(KM_AES_128_ENCRYPT, CRYPT_S390_MSA))
512 keylen_flag |= AES_KEYLEN_128; 853 keylen_flag |= AES_KEYLEN_128;
513 if (crypt_s390_func_available(KM_AES_192_ENCRYPT)) 854 if (crypt_s390_func_available(KM_AES_192_ENCRYPT, CRYPT_S390_MSA))
514 keylen_flag |= AES_KEYLEN_192; 855 keylen_flag |= AES_KEYLEN_192;
515 if (crypt_s390_func_available(KM_AES_256_ENCRYPT)) 856 if (crypt_s390_func_available(KM_AES_256_ENCRYPT, CRYPT_S390_MSA))
516 keylen_flag |= AES_KEYLEN_256; 857 keylen_flag |= AES_KEYLEN_256;
517 858
518 if (!keylen_flag) 859 if (!keylen_flag)
@@ -535,9 +876,40 @@ static int __init aes_s390_init(void)
535 if (ret) 876 if (ret)
536 goto cbc_aes_err; 877 goto cbc_aes_err;
537 878
879 if (crypt_s390_func_available(KM_XTS_128_ENCRYPT,
880 CRYPT_S390_MSA | CRYPT_S390_MSA4) &&
881 crypt_s390_func_available(KM_XTS_256_ENCRYPT,
882 CRYPT_S390_MSA | CRYPT_S390_MSA4)) {
883 ret = crypto_register_alg(&xts_aes_alg);
884 if (ret)
885 goto xts_aes_err;
886 }
887
888 if (crypt_s390_func_available(KMCTR_AES_128_ENCRYPT,
889 CRYPT_S390_MSA | CRYPT_S390_MSA4) &&
890 crypt_s390_func_available(KMCTR_AES_192_ENCRYPT,
891 CRYPT_S390_MSA | CRYPT_S390_MSA4) &&
892 crypt_s390_func_available(KMCTR_AES_256_ENCRYPT,
893 CRYPT_S390_MSA | CRYPT_S390_MSA4)) {
894 ctrblk = (u8 *) __get_free_page(GFP_KERNEL);
895 if (!ctrblk) {
896 ret = -ENOMEM;
897 goto ctr_aes_err;
898 }
899 ret = crypto_register_alg(&ctr_aes_alg);
900 if (ret) {
901 free_page((unsigned long) ctrblk);
902 goto ctr_aes_err;
903 }
904 }
905
538out: 906out:
539 return ret; 907 return ret;
540 908
909ctr_aes_err:
910 crypto_unregister_alg(&xts_aes_alg);
911xts_aes_err:
912 crypto_unregister_alg(&cbc_aes_alg);
541cbc_aes_err: 913cbc_aes_err:
542 crypto_unregister_alg(&ecb_aes_alg); 914 crypto_unregister_alg(&ecb_aes_alg);
543ecb_aes_err: 915ecb_aes_err:
@@ -548,6 +920,9 @@ aes_err:
548 920
549static void __exit aes_s390_fini(void) 921static void __exit aes_s390_fini(void)
550{ 922{
923 crypto_unregister_alg(&ctr_aes_alg);
924 free_page((unsigned long) ctrblk);
925 crypto_unregister_alg(&xts_aes_alg);
551 crypto_unregister_alg(&cbc_aes_alg); 926 crypto_unregister_alg(&cbc_aes_alg);
552 crypto_unregister_alg(&ecb_aes_alg); 927 crypto_unregister_alg(&ecb_aes_alg);
553 crypto_unregister_alg(&aes_alg); 928 crypto_unregister_alg(&aes_alg);
diff --git a/arch/s390/crypto/crypt_s390.h b/arch/s390/crypto/crypt_s390.h
index 0ef9829f2ad6..49676771bd66 100644
--- a/arch/s390/crypto/crypt_s390.h
+++ b/arch/s390/crypto/crypt_s390.h
@@ -24,13 +24,18 @@
24#define CRYPT_S390_PRIORITY 300 24#define CRYPT_S390_PRIORITY 300
25#define CRYPT_S390_COMPOSITE_PRIORITY 400 25#define CRYPT_S390_COMPOSITE_PRIORITY 400
26 26
27#define CRYPT_S390_MSA 0x1
28#define CRYPT_S390_MSA3 0x2
29#define CRYPT_S390_MSA4 0x4
30
27/* s390 cryptographic operations */ 31/* s390 cryptographic operations */
28enum crypt_s390_operations { 32enum crypt_s390_operations {
29 CRYPT_S390_KM = 0x0100, 33 CRYPT_S390_KM = 0x0100,
30 CRYPT_S390_KMC = 0x0200, 34 CRYPT_S390_KMC = 0x0200,
31 CRYPT_S390_KIMD = 0x0300, 35 CRYPT_S390_KIMD = 0x0300,
32 CRYPT_S390_KLMD = 0x0400, 36 CRYPT_S390_KLMD = 0x0400,
33 CRYPT_S390_KMAC = 0x0500 37 CRYPT_S390_KMAC = 0x0500,
38 CRYPT_S390_KMCTR = 0x0600
34}; 39};
35 40
36/* 41/*
@@ -51,6 +56,10 @@ enum crypt_s390_km_func {
51 KM_AES_192_DECRYPT = CRYPT_S390_KM | 0x13 | 0x80, 56 KM_AES_192_DECRYPT = CRYPT_S390_KM | 0x13 | 0x80,
52 KM_AES_256_ENCRYPT = CRYPT_S390_KM | 0x14, 57 KM_AES_256_ENCRYPT = CRYPT_S390_KM | 0x14,
53 KM_AES_256_DECRYPT = CRYPT_S390_KM | 0x14 | 0x80, 58 KM_AES_256_DECRYPT = CRYPT_S390_KM | 0x14 | 0x80,
59 KM_XTS_128_ENCRYPT = CRYPT_S390_KM | 0x32,
60 KM_XTS_128_DECRYPT = CRYPT_S390_KM | 0x32 | 0x80,
61 KM_XTS_256_ENCRYPT = CRYPT_S390_KM | 0x34,
62 KM_XTS_256_DECRYPT = CRYPT_S390_KM | 0x34 | 0x80,
54}; 63};
55 64
56/* 65/*
@@ -75,6 +84,26 @@ enum crypt_s390_kmc_func {
75}; 84};
76 85
77/* 86/*
87 * function codes for KMCTR (CIPHER MESSAGE WITH COUNTER)
88 * instruction
89 */
90enum crypt_s390_kmctr_func {
91 KMCTR_QUERY = CRYPT_S390_KMCTR | 0x0,
92 KMCTR_DEA_ENCRYPT = CRYPT_S390_KMCTR | 0x1,
93 KMCTR_DEA_DECRYPT = CRYPT_S390_KMCTR | 0x1 | 0x80,
94 KMCTR_TDEA_128_ENCRYPT = CRYPT_S390_KMCTR | 0x2,
95 KMCTR_TDEA_128_DECRYPT = CRYPT_S390_KMCTR | 0x2 | 0x80,
96 KMCTR_TDEA_192_ENCRYPT = CRYPT_S390_KMCTR | 0x3,
97 KMCTR_TDEA_192_DECRYPT = CRYPT_S390_KMCTR | 0x3 | 0x80,
98 KMCTR_AES_128_ENCRYPT = CRYPT_S390_KMCTR | 0x12,
99 KMCTR_AES_128_DECRYPT = CRYPT_S390_KMCTR | 0x12 | 0x80,
100 KMCTR_AES_192_ENCRYPT = CRYPT_S390_KMCTR | 0x13,
101 KMCTR_AES_192_DECRYPT = CRYPT_S390_KMCTR | 0x13 | 0x80,
102 KMCTR_AES_256_ENCRYPT = CRYPT_S390_KMCTR | 0x14,
103 KMCTR_AES_256_DECRYPT = CRYPT_S390_KMCTR | 0x14 | 0x80,
104};
105
106/*
78 * function codes for KIMD (COMPUTE INTERMEDIATE MESSAGE DIGEST) 107 * function codes for KIMD (COMPUTE INTERMEDIATE MESSAGE DIGEST)
79 * instruction 108 * instruction
80 */ 109 */
@@ -83,6 +112,7 @@ enum crypt_s390_kimd_func {
83 KIMD_SHA_1 = CRYPT_S390_KIMD | 1, 112 KIMD_SHA_1 = CRYPT_S390_KIMD | 1,
84 KIMD_SHA_256 = CRYPT_S390_KIMD | 2, 113 KIMD_SHA_256 = CRYPT_S390_KIMD | 2,
85 KIMD_SHA_512 = CRYPT_S390_KIMD | 3, 114 KIMD_SHA_512 = CRYPT_S390_KIMD | 3,
115 KIMD_GHASH = CRYPT_S390_KIMD | 65,
86}; 116};
87 117
88/* 118/*
@@ -284,6 +314,45 @@ static inline int crypt_s390_kmac(long func, void *param,
284} 314}
285 315
286/** 316/**
317 * crypt_s390_kmctr:
318 * @func: the function code passed to KMCTR; see crypt_s390_kmctr_func
319 * @param: address of parameter block; see POP for details on each func
320 * @dest: address of destination memory area
321 * @src: address of source memory area
322 * @src_len: length of src operand in bytes
323 * @counter: address of counter value
324 *
325 * Executes the KMCTR (CIPHER MESSAGE WITH COUNTER) operation of the CPU.
326 *
327 * Returns -1 for failure, 0 for the query func, number of processed
328 * bytes for encryption/decryption funcs
329 */
330static inline int crypt_s390_kmctr(long func, void *param, u8 *dest,
331 const u8 *src, long src_len, u8 *counter)
332{
333 register long __func asm("0") = func & CRYPT_S390_FUNC_MASK;
334 register void *__param asm("1") = param;
335 register const u8 *__src asm("2") = src;
336 register long __src_len asm("3") = src_len;
337 register u8 *__dest asm("4") = dest;
338 register u8 *__ctr asm("6") = counter;
339 int ret = -1;
340
341 asm volatile(
342 "0: .insn rrf,0xb92d0000,%3,%1,%4,0 \n" /* KMCTR opcode */
343 "1: brc 1,0b \n" /* handle partial completion */
344 " la %0,0\n"
345 "2:\n"
346 EX_TABLE(0b,2b) EX_TABLE(1b,2b)
347 : "+d" (ret), "+a" (__src), "+d" (__src_len), "+a" (__dest),
348 "+a" (__ctr)
349 : "d" (__func), "a" (__param) : "cc", "memory");
350 if (ret < 0)
351 return ret;
352 return (func & CRYPT_S390_FUNC_MASK) ? src_len - __src_len : __src_len;
353}
354
355/**
287 * crypt_s390_func_available: 356 * crypt_s390_func_available:
288 * @func: the function code of the specific function; 0 if op in general 357 * @func: the function code of the specific function; 0 if op in general
289 * 358 *
@@ -291,13 +360,17 @@ static inline int crypt_s390_kmac(long func, void *param,
291 * 360 *
292 * Returns 1 if func available; 0 if func or op in general not available 361 * Returns 1 if func available; 0 if func or op in general not available
293 */ 362 */
294static inline int crypt_s390_func_available(int func) 363static inline int crypt_s390_func_available(int func,
364 unsigned int facility_mask)
295{ 365{
296 unsigned char status[16]; 366 unsigned char status[16];
297 int ret; 367 int ret;
298 368
299 /* check if CPACF facility (bit 17) is available */ 369 if (facility_mask & CRYPT_S390_MSA && !test_facility(17))
300 if (!(stfl() & 1ULL << (31 - 17))) 370 return 0;
371 if (facility_mask & CRYPT_S390_MSA3 && !test_facility(76))
372 return 0;
373 if (facility_mask & CRYPT_S390_MSA4 && !test_facility(77))
301 return 0; 374 return 0;
302 375
303 switch (func & CRYPT_S390_OP_MASK) { 376 switch (func & CRYPT_S390_OP_MASK) {
@@ -316,6 +389,10 @@ static inline int crypt_s390_func_available(int func)
316 case CRYPT_S390_KMAC: 389 case CRYPT_S390_KMAC:
317 ret = crypt_s390_kmac(KMAC_QUERY, &status, NULL, 0); 390 ret = crypt_s390_kmac(KMAC_QUERY, &status, NULL, 0);
318 break; 391 break;
392 case CRYPT_S390_KMCTR:
393 ret = crypt_s390_kmctr(KMCTR_QUERY, &status, NULL, NULL, 0,
394 NULL);
395 break;
319 default: 396 default:
320 return 0; 397 return 0;
321 } 398 }
@@ -326,4 +403,31 @@ static inline int crypt_s390_func_available(int func)
326 return (status[func >> 3] & (0x80 >> (func & 7))) != 0; 403 return (status[func >> 3] & (0x80 >> (func & 7))) != 0;
327} 404}
328 405
406/**
407 * crypt_s390_pcc:
408 * @func: the function code passed to KM; see crypt_s390_km_func
409 * @param: address of parameter block; see POP for details on each func
410 *
411 * Executes the PCC (PERFORM CRYPTOGRAPHIC COMPUTATION) operation of the CPU.
412 *
413 * Returns -1 for failure, 0 for success.
414 */
415static inline int crypt_s390_pcc(long func, void *param)
416{
417 register long __func asm("0") = func & 0x7f; /* encrypt or decrypt */
418 register void *__param asm("1") = param;
419 int ret = -1;
420
421 asm volatile(
422 "0: .insn rre,0xb92c0000,0,0 \n" /* PCC opcode */
423 "1: brc 1,0b \n" /* handle partial completion */
424 " la %0,0\n"
425 "2:\n"
426 EX_TABLE(0b,2b) EX_TABLE(1b,2b)
427 : "+d" (ret)
428 : "d" (__func), "a" (__param) : "cc", "memory");
429 return ret;
430}
431
432
329#endif /* _CRYPTO_ARCH_S390_CRYPT_S390_H */ 433#endif /* _CRYPTO_ARCH_S390_CRYPT_S390_H */
diff --git a/arch/s390/crypto/des_check_key.c b/arch/s390/crypto/des_check_key.c
deleted file mode 100644
index 5706af266442..000000000000
--- a/arch/s390/crypto/des_check_key.c
+++ /dev/null
@@ -1,132 +0,0 @@
1/*
2 * Cryptographic API.
3 *
4 * Function for checking keys for the DES and Tripple DES Encryption
5 * algorithms.
6 *
7 * Originally released as descore by Dana L. How <how@isl.stanford.edu>.
8 * Modified by Raimar Falke <rf13@inf.tu-dresden.de> for the Linux-Kernel.
9 * Derived from Cryptoapi and Nettle implementations, adapted for in-place
10 * scatterlist interface. Changed LGPL to GPL per section 3 of the LGPL.
11 *
12 * s390 Version:
13 * Copyright IBM Corp. 2003
14 * Author(s): Thomas Spatzier
15 * Jan Glauber (jan.glauber@de.ibm.com)
16 *
17 * Derived from "crypto/des.c"
18 * Copyright (c) 1992 Dana L. How.
19 * Copyright (c) Raimar Falke <rf13@inf.tu-dresden.de>
20 * Copyright (c) Gisle Sflensminde <gisle@ii.uib.no>
21 * Copyright (C) 2001 Niels Mvller.
22 * Copyright (c) 2002 James Morris <jmorris@intercode.com.au>
23 *
24 * This program is free software; you can redistribute it and/or modify
25 * it under the terms of the GNU General Public License as published by
26 * the Free Software Foundation; either version 2 of the License, or
27 * (at your option) any later version.
28 *
29 */
30#include <linux/init.h>
31#include <linux/module.h>
32#include <linux/errno.h>
33#include <linux/crypto.h>
34#include "crypto_des.h"
35
36#define ROR(d,c,o) ((d) = (d) >> (c) | (d) << (o))
37
38static const u8 parity[] = {
39 8,1,0,8,0,8,8,0,0,8,8,0,8,0,2,8,0,8,8,0,8,0,0,8,8,0,0,8,0,8,8,3,
40 0,8,8,0,8,0,0,8,8,0,0,8,0,8,8,0,8,0,0,8,0,8,8,0,0,8,8,0,8,0,0,8,
41 0,8,8,0,8,0,0,8,8,0,0,8,0,8,8,0,8,0,0,8,0,8,8,0,0,8,8,0,8,0,0,8,
42 8,0,0,8,0,8,8,0,0,8,8,0,8,0,0,8,0,8,8,0,8,0,0,8,8,0,0,8,0,8,8,0,
43 0,8,8,0,8,0,0,8,8,0,0,8,0,8,8,0,8,0,0,8,0,8,8,0,0,8,8,0,8,0,0,8,
44 8,0,0,8,0,8,8,0,0,8,8,0,8,0,0,8,0,8,8,0,8,0,0,8,8,0,0,8,0,8,8,0,
45 8,0,0,8,0,8,8,0,0,8,8,0,8,0,0,8,0,8,8,0,8,0,0,8,8,0,0,8,0,8,8,0,
46 4,8,8,0,8,0,0,8,8,0,0,8,0,8,8,0,8,5,0,8,0,8,8,0,0,8,8,0,8,0,6,8,
47};
48
49/*
50 * RFC2451: Weak key checks SHOULD be performed.
51 */
52int
53crypto_des_check_key(const u8 *key, unsigned int keylen, u32 *flags)
54{
55 u32 n, w;
56
57 n = parity[key[0]]; n <<= 4;
58 n |= parity[key[1]]; n <<= 4;
59 n |= parity[key[2]]; n <<= 4;
60 n |= parity[key[3]]; n <<= 4;
61 n |= parity[key[4]]; n <<= 4;
62 n |= parity[key[5]]; n <<= 4;
63 n |= parity[key[6]]; n <<= 4;
64 n |= parity[key[7]];
65 w = 0x88888888L;
66
67 if ((*flags & CRYPTO_TFM_REQ_WEAK_KEY)
68 && !((n - (w >> 3)) & w)) { /* 1 in 10^10 keys passes this test */
69 if (n < 0x41415151) {
70 if (n < 0x31312121) {
71 if (n < 0x14141515) {
72 /* 01 01 01 01 01 01 01 01 */
73 if (n == 0x11111111) goto weak;
74 /* 01 1F 01 1F 01 0E 01 0E */
75 if (n == 0x13131212) goto weak;
76 } else {
77 /* 01 E0 01 E0 01 F1 01 F1 */
78 if (n == 0x14141515) goto weak;
79 /* 01 FE 01 FE 01 FE 01 FE */
80 if (n == 0x16161616) goto weak;
81 }
82 } else {
83 if (n < 0x34342525) {
84 /* 1F 01 1F 01 0E 01 0E 01 */
85 if (n == 0x31312121) goto weak;
86 /* 1F 1F 1F 1F 0E 0E 0E 0E (?) */
87 if (n == 0x33332222) goto weak;
88 } else {
89 /* 1F E0 1F E0 0E F1 0E F1 */
90 if (n == 0x34342525) goto weak;
91 /* 1F FE 1F FE 0E FE 0E FE */
92 if (n == 0x36362626) goto weak;
93 }
94 }
95 } else {
96 if (n < 0x61616161) {
97 if (n < 0x44445555) {
98 /* E0 01 E0 01 F1 01 F1 01 */
99 if (n == 0x41415151) goto weak;
100 /* E0 1F E0 1F F1 0E F1 0E */
101 if (n == 0x43435252) goto weak;
102 } else {
103 /* E0 E0 E0 E0 F1 F1 F1 F1 (?) */
104 if (n == 0x44445555) goto weak;
105 /* E0 FE E0 FE F1 FE F1 FE */
106 if (n == 0x46465656) goto weak;
107 }
108 } else {
109 if (n < 0x64646565) {
110 /* FE 01 FE 01 FE 01 FE 01 */
111 if (n == 0x61616161) goto weak;
112 /* FE 1F FE 1F FE 0E FE 0E */
113 if (n == 0x63636262) goto weak;
114 } else {
115 /* FE E0 FE E0 FE F1 FE F1 */
116 if (n == 0x64646565) goto weak;
117 /* FE FE FE FE FE FE FE FE */
118 if (n == 0x66666666) goto weak;
119 }
120 }
121 }
122 }
123 return 0;
124weak:
125 *flags |= CRYPTO_TFM_RES_WEAK_KEY;
126 return -EINVAL;
127}
128
129EXPORT_SYMBOL(crypto_des_check_key);
130
131MODULE_LICENSE("GPL");
132MODULE_DESCRIPTION("Key Check function for DES & DES3 Cipher Algorithms");
diff --git a/arch/s390/crypto/des_s390.c b/arch/s390/crypto/des_s390.c
index cc5420118393..a52bfd124d86 100644
--- a/arch/s390/crypto/des_s390.c
+++ b/arch/s390/crypto/des_s390.c
@@ -3,7 +3,7 @@
3 * 3 *
4 * s390 implementation of the DES Cipher Algorithm. 4 * s390 implementation of the DES Cipher Algorithm.
5 * 5 *
6 * Copyright IBM Corp. 2003,2007 6 * Copyright IBM Corp. 2003,2011
7 * Author(s): Thomas Spatzier 7 * Author(s): Thomas Spatzier
8 * Jan Glauber (jan.glauber@de.ibm.com) 8 * Jan Glauber (jan.glauber@de.ibm.com)
9 * 9 *
@@ -22,22 +22,19 @@
22 22
23#include "crypt_s390.h" 23#include "crypt_s390.h"
24 24
25#define DES3_192_KEY_SIZE (3 * DES_KEY_SIZE) 25#define DES3_KEY_SIZE (3 * DES_KEY_SIZE)
26 26
27struct crypt_s390_des_ctx { 27static u8 *ctrblk;
28 u8 iv[DES_BLOCK_SIZE];
29 u8 key[DES_KEY_SIZE];
30};
31 28
32struct crypt_s390_des3_192_ctx { 29struct s390_des_ctx {
33 u8 iv[DES_BLOCK_SIZE]; 30 u8 iv[DES_BLOCK_SIZE];
34 u8 key[DES3_192_KEY_SIZE]; 31 u8 key[DES3_KEY_SIZE];
35}; 32};
36 33
37static int des_setkey(struct crypto_tfm *tfm, const u8 *key, 34static int des_setkey(struct crypto_tfm *tfm, const u8 *key,
38 unsigned int keylen) 35 unsigned int key_len)
39{ 36{
40 struct crypt_s390_des_ctx *dctx = crypto_tfm_ctx(tfm); 37 struct s390_des_ctx *ctx = crypto_tfm_ctx(tfm);
41 u32 *flags = &tfm->crt_flags; 38 u32 *flags = &tfm->crt_flags;
42 u32 tmp[DES_EXPKEY_WORDS]; 39 u32 tmp[DES_EXPKEY_WORDS];
43 40
@@ -47,22 +44,22 @@ static int des_setkey(struct crypto_tfm *tfm, const u8 *key,
47 return -EINVAL; 44 return -EINVAL;
48 } 45 }
49 46
50 memcpy(dctx->key, key, keylen); 47 memcpy(ctx->key, key, key_len);
51 return 0; 48 return 0;
52} 49}
53 50
54static void des_encrypt(struct crypto_tfm *tfm, u8 *out, const u8 *in) 51static void des_encrypt(struct crypto_tfm *tfm, u8 *out, const u8 *in)
55{ 52{
56 struct crypt_s390_des_ctx *dctx = crypto_tfm_ctx(tfm); 53 struct s390_des_ctx *ctx = crypto_tfm_ctx(tfm);
57 54
58 crypt_s390_km(KM_DEA_ENCRYPT, dctx->key, out, in, DES_BLOCK_SIZE); 55 crypt_s390_km(KM_DEA_ENCRYPT, ctx->key, out, in, DES_BLOCK_SIZE);
59} 56}
60 57
61static void des_decrypt(struct crypto_tfm *tfm, u8 *out, const u8 *in) 58static void des_decrypt(struct crypto_tfm *tfm, u8 *out, const u8 *in)
62{ 59{
63 struct crypt_s390_des_ctx *dctx = crypto_tfm_ctx(tfm); 60 struct s390_des_ctx *ctx = crypto_tfm_ctx(tfm);
64 61
65 crypt_s390_km(KM_DEA_DECRYPT, dctx->key, out, in, DES_BLOCK_SIZE); 62 crypt_s390_km(KM_DEA_DECRYPT, ctx->key, out, in, DES_BLOCK_SIZE);
66} 63}
67 64
68static struct crypto_alg des_alg = { 65static struct crypto_alg des_alg = {
@@ -71,7 +68,7 @@ static struct crypto_alg des_alg = {
71 .cra_priority = CRYPT_S390_PRIORITY, 68 .cra_priority = CRYPT_S390_PRIORITY,
72 .cra_flags = CRYPTO_ALG_TYPE_CIPHER, 69 .cra_flags = CRYPTO_ALG_TYPE_CIPHER,
73 .cra_blocksize = DES_BLOCK_SIZE, 70 .cra_blocksize = DES_BLOCK_SIZE,
74 .cra_ctxsize = sizeof(struct crypt_s390_des_ctx), 71 .cra_ctxsize = sizeof(struct s390_des_ctx),
75 .cra_module = THIS_MODULE, 72 .cra_module = THIS_MODULE,
76 .cra_list = LIST_HEAD_INIT(des_alg.cra_list), 73 .cra_list = LIST_HEAD_INIT(des_alg.cra_list),
77 .cra_u = { 74 .cra_u = {
@@ -86,7 +83,7 @@ static struct crypto_alg des_alg = {
86}; 83};
87 84
88static int ecb_desall_crypt(struct blkcipher_desc *desc, long func, 85static int ecb_desall_crypt(struct blkcipher_desc *desc, long func,
89 void *param, struct blkcipher_walk *walk) 86 u8 *key, struct blkcipher_walk *walk)
90{ 87{
91 int ret = blkcipher_walk_virt(desc, walk); 88 int ret = blkcipher_walk_virt(desc, walk);
92 unsigned int nbytes; 89 unsigned int nbytes;
@@ -97,7 +94,7 @@ static int ecb_desall_crypt(struct blkcipher_desc *desc, long func,
97 u8 *out = walk->dst.virt.addr; 94 u8 *out = walk->dst.virt.addr;
98 u8 *in = walk->src.virt.addr; 95 u8 *in = walk->src.virt.addr;
99 96
100 ret = crypt_s390_km(func, param, out, in, n); 97 ret = crypt_s390_km(func, key, out, in, n);
101 BUG_ON((ret < 0) || (ret != n)); 98 BUG_ON((ret < 0) || (ret != n));
102 99
103 nbytes &= DES_BLOCK_SIZE - 1; 100 nbytes &= DES_BLOCK_SIZE - 1;
@@ -108,7 +105,7 @@ static int ecb_desall_crypt(struct blkcipher_desc *desc, long func,
108} 105}
109 106
110static int cbc_desall_crypt(struct blkcipher_desc *desc, long func, 107static int cbc_desall_crypt(struct blkcipher_desc *desc, long func,
111 void *param, struct blkcipher_walk *walk) 108 u8 *iv, struct blkcipher_walk *walk)
112{ 109{
113 int ret = blkcipher_walk_virt(desc, walk); 110 int ret = blkcipher_walk_virt(desc, walk);
114 unsigned int nbytes = walk->nbytes; 111 unsigned int nbytes = walk->nbytes;
@@ -116,20 +113,20 @@ static int cbc_desall_crypt(struct blkcipher_desc *desc, long func,
116 if (!nbytes) 113 if (!nbytes)
117 goto out; 114 goto out;
118 115
119 memcpy(param, walk->iv, DES_BLOCK_SIZE); 116 memcpy(iv, walk->iv, DES_BLOCK_SIZE);
120 do { 117 do {
121 /* only use complete blocks */ 118 /* only use complete blocks */
122 unsigned int n = nbytes & ~(DES_BLOCK_SIZE - 1); 119 unsigned int n = nbytes & ~(DES_BLOCK_SIZE - 1);
123 u8 *out = walk->dst.virt.addr; 120 u8 *out = walk->dst.virt.addr;
124 u8 *in = walk->src.virt.addr; 121 u8 *in = walk->src.virt.addr;
125 122
126 ret = crypt_s390_kmc(func, param, out, in, n); 123 ret = crypt_s390_kmc(func, iv, out, in, n);
127 BUG_ON((ret < 0) || (ret != n)); 124 BUG_ON((ret < 0) || (ret != n));
128 125
129 nbytes &= DES_BLOCK_SIZE - 1; 126 nbytes &= DES_BLOCK_SIZE - 1;
130 ret = blkcipher_walk_done(desc, walk, nbytes); 127 ret = blkcipher_walk_done(desc, walk, nbytes);
131 } while ((nbytes = walk->nbytes)); 128 } while ((nbytes = walk->nbytes));
132 memcpy(walk->iv, param, DES_BLOCK_SIZE); 129 memcpy(walk->iv, iv, DES_BLOCK_SIZE);
133 130
134out: 131out:
135 return ret; 132 return ret;
@@ -139,22 +136,22 @@ static int ecb_des_encrypt(struct blkcipher_desc *desc,
139 struct scatterlist *dst, struct scatterlist *src, 136 struct scatterlist *dst, struct scatterlist *src,
140 unsigned int nbytes) 137 unsigned int nbytes)
141{ 138{
142 struct crypt_s390_des_ctx *sctx = crypto_blkcipher_ctx(desc->tfm); 139 struct s390_des_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
143 struct blkcipher_walk walk; 140 struct blkcipher_walk walk;
144 141
145 blkcipher_walk_init(&walk, dst, src, nbytes); 142 blkcipher_walk_init(&walk, dst, src, nbytes);
146 return ecb_desall_crypt(desc, KM_DEA_ENCRYPT, sctx->key, &walk); 143 return ecb_desall_crypt(desc, KM_DEA_ENCRYPT, ctx->key, &walk);
147} 144}
148 145
149static int ecb_des_decrypt(struct blkcipher_desc *desc, 146static int ecb_des_decrypt(struct blkcipher_desc *desc,
150 struct scatterlist *dst, struct scatterlist *src, 147 struct scatterlist *dst, struct scatterlist *src,
151 unsigned int nbytes) 148 unsigned int nbytes)
152{ 149{
153 struct crypt_s390_des_ctx *sctx = crypto_blkcipher_ctx(desc->tfm); 150 struct s390_des_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
154 struct blkcipher_walk walk; 151 struct blkcipher_walk walk;
155 152
156 blkcipher_walk_init(&walk, dst, src, nbytes); 153 blkcipher_walk_init(&walk, dst, src, nbytes);
157 return ecb_desall_crypt(desc, KM_DEA_DECRYPT, sctx->key, &walk); 154 return ecb_desall_crypt(desc, KM_DEA_DECRYPT, ctx->key, &walk);
158} 155}
159 156
160static struct crypto_alg ecb_des_alg = { 157static struct crypto_alg ecb_des_alg = {
@@ -163,7 +160,7 @@ static struct crypto_alg ecb_des_alg = {
163 .cra_priority = CRYPT_S390_COMPOSITE_PRIORITY, 160 .cra_priority = CRYPT_S390_COMPOSITE_PRIORITY,
164 .cra_flags = CRYPTO_ALG_TYPE_BLKCIPHER, 161 .cra_flags = CRYPTO_ALG_TYPE_BLKCIPHER,
165 .cra_blocksize = DES_BLOCK_SIZE, 162 .cra_blocksize = DES_BLOCK_SIZE,
166 .cra_ctxsize = sizeof(struct crypt_s390_des_ctx), 163 .cra_ctxsize = sizeof(struct s390_des_ctx),
167 .cra_type = &crypto_blkcipher_type, 164 .cra_type = &crypto_blkcipher_type,
168 .cra_module = THIS_MODULE, 165 .cra_module = THIS_MODULE,
169 .cra_list = LIST_HEAD_INIT(ecb_des_alg.cra_list), 166 .cra_list = LIST_HEAD_INIT(ecb_des_alg.cra_list),
@@ -182,22 +179,22 @@ static int cbc_des_encrypt(struct blkcipher_desc *desc,
182 struct scatterlist *dst, struct scatterlist *src, 179 struct scatterlist *dst, struct scatterlist *src,
183 unsigned int nbytes) 180 unsigned int nbytes)
184{ 181{
185 struct crypt_s390_des_ctx *sctx = crypto_blkcipher_ctx(desc->tfm); 182 struct s390_des_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
186 struct blkcipher_walk walk; 183 struct blkcipher_walk walk;
187 184
188 blkcipher_walk_init(&walk, dst, src, nbytes); 185 blkcipher_walk_init(&walk, dst, src, nbytes);
189 return cbc_desall_crypt(desc, KMC_DEA_ENCRYPT, sctx->iv, &walk); 186 return cbc_desall_crypt(desc, KMC_DEA_ENCRYPT, ctx->iv, &walk);
190} 187}
191 188
192static int cbc_des_decrypt(struct blkcipher_desc *desc, 189static int cbc_des_decrypt(struct blkcipher_desc *desc,
193 struct scatterlist *dst, struct scatterlist *src, 190 struct scatterlist *dst, struct scatterlist *src,
194 unsigned int nbytes) 191 unsigned int nbytes)
195{ 192{
196 struct crypt_s390_des_ctx *sctx = crypto_blkcipher_ctx(desc->tfm); 193 struct s390_des_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
197 struct blkcipher_walk walk; 194 struct blkcipher_walk walk;
198 195
199 blkcipher_walk_init(&walk, dst, src, nbytes); 196 blkcipher_walk_init(&walk, dst, src, nbytes);
200 return cbc_desall_crypt(desc, KMC_DEA_DECRYPT, sctx->iv, &walk); 197 return cbc_desall_crypt(desc, KMC_DEA_DECRYPT, ctx->iv, &walk);
201} 198}
202 199
203static struct crypto_alg cbc_des_alg = { 200static struct crypto_alg cbc_des_alg = {
@@ -206,7 +203,7 @@ static struct crypto_alg cbc_des_alg = {
206 .cra_priority = CRYPT_S390_COMPOSITE_PRIORITY, 203 .cra_priority = CRYPT_S390_COMPOSITE_PRIORITY,
207 .cra_flags = CRYPTO_ALG_TYPE_BLKCIPHER, 204 .cra_flags = CRYPTO_ALG_TYPE_BLKCIPHER,
208 .cra_blocksize = DES_BLOCK_SIZE, 205 .cra_blocksize = DES_BLOCK_SIZE,
209 .cra_ctxsize = sizeof(struct crypt_s390_des_ctx), 206 .cra_ctxsize = sizeof(struct s390_des_ctx),
210 .cra_type = &crypto_blkcipher_type, 207 .cra_type = &crypto_blkcipher_type,
211 .cra_module = THIS_MODULE, 208 .cra_module = THIS_MODULE,
212 .cra_list = LIST_HEAD_INIT(cbc_des_alg.cra_list), 209 .cra_list = LIST_HEAD_INIT(cbc_des_alg.cra_list),
@@ -235,10 +232,10 @@ static struct crypto_alg cbc_des_alg = {
235 * property. 232 * property.
236 * 233 *
237 */ 234 */
238static int des3_192_setkey(struct crypto_tfm *tfm, const u8 *key, 235static int des3_setkey(struct crypto_tfm *tfm, const u8 *key,
239 unsigned int keylen) 236 unsigned int key_len)
240{ 237{
241 struct crypt_s390_des3_192_ctx *dctx = crypto_tfm_ctx(tfm); 238 struct s390_des_ctx *ctx = crypto_tfm_ctx(tfm);
242 u32 *flags = &tfm->crt_flags; 239 u32 *flags = &tfm->crt_flags;
243 240
244 if (!(memcmp(key, &key[DES_KEY_SIZE], DES_KEY_SIZE) && 241 if (!(memcmp(key, &key[DES_KEY_SIZE], DES_KEY_SIZE) &&
@@ -248,141 +245,276 @@ static int des3_192_setkey(struct crypto_tfm *tfm, const u8 *key,
248 *flags |= CRYPTO_TFM_RES_WEAK_KEY; 245 *flags |= CRYPTO_TFM_RES_WEAK_KEY;
249 return -EINVAL; 246 return -EINVAL;
250 } 247 }
251 memcpy(dctx->key, key, keylen); 248 memcpy(ctx->key, key, key_len);
252 return 0; 249 return 0;
253} 250}
254 251
255static void des3_192_encrypt(struct crypto_tfm *tfm, u8 *dst, const u8 *src) 252static void des3_encrypt(struct crypto_tfm *tfm, u8 *dst, const u8 *src)
256{ 253{
257 struct crypt_s390_des3_192_ctx *dctx = crypto_tfm_ctx(tfm); 254 struct s390_des_ctx *ctx = crypto_tfm_ctx(tfm);
258 255
259 crypt_s390_km(KM_TDEA_192_ENCRYPT, dctx->key, dst, (void*)src, 256 crypt_s390_km(KM_TDEA_192_ENCRYPT, ctx->key, dst, src, DES_BLOCK_SIZE);
260 DES_BLOCK_SIZE);
261} 257}
262 258
263static void des3_192_decrypt(struct crypto_tfm *tfm, u8 *dst, const u8 *src) 259static void des3_decrypt(struct crypto_tfm *tfm, u8 *dst, const u8 *src)
264{ 260{
265 struct crypt_s390_des3_192_ctx *dctx = crypto_tfm_ctx(tfm); 261 struct s390_des_ctx *ctx = crypto_tfm_ctx(tfm);
266 262
267 crypt_s390_km(KM_TDEA_192_DECRYPT, dctx->key, dst, (void*)src, 263 crypt_s390_km(KM_TDEA_192_DECRYPT, ctx->key, dst, src, DES_BLOCK_SIZE);
268 DES_BLOCK_SIZE);
269} 264}
270 265
271static struct crypto_alg des3_192_alg = { 266static struct crypto_alg des3_alg = {
272 .cra_name = "des3_ede", 267 .cra_name = "des3_ede",
273 .cra_driver_name = "des3_ede-s390", 268 .cra_driver_name = "des3_ede-s390",
274 .cra_priority = CRYPT_S390_PRIORITY, 269 .cra_priority = CRYPT_S390_PRIORITY,
275 .cra_flags = CRYPTO_ALG_TYPE_CIPHER, 270 .cra_flags = CRYPTO_ALG_TYPE_CIPHER,
276 .cra_blocksize = DES_BLOCK_SIZE, 271 .cra_blocksize = DES_BLOCK_SIZE,
277 .cra_ctxsize = sizeof(struct crypt_s390_des3_192_ctx), 272 .cra_ctxsize = sizeof(struct s390_des_ctx),
278 .cra_module = THIS_MODULE, 273 .cra_module = THIS_MODULE,
279 .cra_list = LIST_HEAD_INIT(des3_192_alg.cra_list), 274 .cra_list = LIST_HEAD_INIT(des3_alg.cra_list),
280 .cra_u = { 275 .cra_u = {
281 .cipher = { 276 .cipher = {
282 .cia_min_keysize = DES3_192_KEY_SIZE, 277 .cia_min_keysize = DES3_KEY_SIZE,
283 .cia_max_keysize = DES3_192_KEY_SIZE, 278 .cia_max_keysize = DES3_KEY_SIZE,
284 .cia_setkey = des3_192_setkey, 279 .cia_setkey = des3_setkey,
285 .cia_encrypt = des3_192_encrypt, 280 .cia_encrypt = des3_encrypt,
286 .cia_decrypt = des3_192_decrypt, 281 .cia_decrypt = des3_decrypt,
287 } 282 }
288 } 283 }
289}; 284};
290 285
291static int ecb_des3_192_encrypt(struct blkcipher_desc *desc, 286static int ecb_des3_encrypt(struct blkcipher_desc *desc,
292 struct scatterlist *dst, 287 struct scatterlist *dst, struct scatterlist *src,
293 struct scatterlist *src, unsigned int nbytes) 288 unsigned int nbytes)
294{ 289{
295 struct crypt_s390_des3_192_ctx *sctx = crypto_blkcipher_ctx(desc->tfm); 290 struct s390_des_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
296 struct blkcipher_walk walk; 291 struct blkcipher_walk walk;
297 292
298 blkcipher_walk_init(&walk, dst, src, nbytes); 293 blkcipher_walk_init(&walk, dst, src, nbytes);
299 return ecb_desall_crypt(desc, KM_TDEA_192_ENCRYPT, sctx->key, &walk); 294 return ecb_desall_crypt(desc, KM_TDEA_192_ENCRYPT, ctx->key, &walk);
300} 295}
301 296
302static int ecb_des3_192_decrypt(struct blkcipher_desc *desc, 297static int ecb_des3_decrypt(struct blkcipher_desc *desc,
303 struct scatterlist *dst, 298 struct scatterlist *dst, struct scatterlist *src,
304 struct scatterlist *src, unsigned int nbytes) 299 unsigned int nbytes)
305{ 300{
306 struct crypt_s390_des3_192_ctx *sctx = crypto_blkcipher_ctx(desc->tfm); 301 struct s390_des_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
307 struct blkcipher_walk walk; 302 struct blkcipher_walk walk;
308 303
309 blkcipher_walk_init(&walk, dst, src, nbytes); 304 blkcipher_walk_init(&walk, dst, src, nbytes);
310 return ecb_desall_crypt(desc, KM_TDEA_192_DECRYPT, sctx->key, &walk); 305 return ecb_desall_crypt(desc, KM_TDEA_192_DECRYPT, ctx->key, &walk);
311} 306}
312 307
313static struct crypto_alg ecb_des3_192_alg = { 308static struct crypto_alg ecb_des3_alg = {
314 .cra_name = "ecb(des3_ede)", 309 .cra_name = "ecb(des3_ede)",
315 .cra_driver_name = "ecb-des3_ede-s390", 310 .cra_driver_name = "ecb-des3_ede-s390",
316 .cra_priority = CRYPT_S390_COMPOSITE_PRIORITY, 311 .cra_priority = CRYPT_S390_COMPOSITE_PRIORITY,
317 .cra_flags = CRYPTO_ALG_TYPE_BLKCIPHER, 312 .cra_flags = CRYPTO_ALG_TYPE_BLKCIPHER,
318 .cra_blocksize = DES_BLOCK_SIZE, 313 .cra_blocksize = DES_BLOCK_SIZE,
319 .cra_ctxsize = sizeof(struct crypt_s390_des3_192_ctx), 314 .cra_ctxsize = sizeof(struct s390_des_ctx),
320 .cra_type = &crypto_blkcipher_type, 315 .cra_type = &crypto_blkcipher_type,
321 .cra_module = THIS_MODULE, 316 .cra_module = THIS_MODULE,
322 .cra_list = LIST_HEAD_INIT( 317 .cra_list = LIST_HEAD_INIT(
323 ecb_des3_192_alg.cra_list), 318 ecb_des3_alg.cra_list),
324 .cra_u = { 319 .cra_u = {
325 .blkcipher = { 320 .blkcipher = {
326 .min_keysize = DES3_192_KEY_SIZE, 321 .min_keysize = DES3_KEY_SIZE,
327 .max_keysize = DES3_192_KEY_SIZE, 322 .max_keysize = DES3_KEY_SIZE,
328 .setkey = des3_192_setkey, 323 .setkey = des3_setkey,
329 .encrypt = ecb_des3_192_encrypt, 324 .encrypt = ecb_des3_encrypt,
330 .decrypt = ecb_des3_192_decrypt, 325 .decrypt = ecb_des3_decrypt,
331 } 326 }
332 } 327 }
333}; 328};
334 329
335static int cbc_des3_192_encrypt(struct blkcipher_desc *desc, 330static int cbc_des3_encrypt(struct blkcipher_desc *desc,
336 struct scatterlist *dst, 331 struct scatterlist *dst, struct scatterlist *src,
337 struct scatterlist *src, unsigned int nbytes) 332 unsigned int nbytes)
338{ 333{
339 struct crypt_s390_des3_192_ctx *sctx = crypto_blkcipher_ctx(desc->tfm); 334 struct s390_des_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
340 struct blkcipher_walk walk; 335 struct blkcipher_walk walk;
341 336
342 blkcipher_walk_init(&walk, dst, src, nbytes); 337 blkcipher_walk_init(&walk, dst, src, nbytes);
343 return cbc_desall_crypt(desc, KMC_TDEA_192_ENCRYPT, sctx->iv, &walk); 338 return cbc_desall_crypt(desc, KMC_TDEA_192_ENCRYPT, ctx->iv, &walk);
344} 339}
345 340
346static int cbc_des3_192_decrypt(struct blkcipher_desc *desc, 341static int cbc_des3_decrypt(struct blkcipher_desc *desc,
347 struct scatterlist *dst, 342 struct scatterlist *dst, struct scatterlist *src,
348 struct scatterlist *src, unsigned int nbytes) 343 unsigned int nbytes)
349{ 344{
350 struct crypt_s390_des3_192_ctx *sctx = crypto_blkcipher_ctx(desc->tfm); 345 struct s390_des_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
351 struct blkcipher_walk walk; 346 struct blkcipher_walk walk;
352 347
353 blkcipher_walk_init(&walk, dst, src, nbytes); 348 blkcipher_walk_init(&walk, dst, src, nbytes);
354 return cbc_desall_crypt(desc, KMC_TDEA_192_DECRYPT, sctx->iv, &walk); 349 return cbc_desall_crypt(desc, KMC_TDEA_192_DECRYPT, ctx->iv, &walk);
355} 350}
356 351
357static struct crypto_alg cbc_des3_192_alg = { 352static struct crypto_alg cbc_des3_alg = {
358 .cra_name = "cbc(des3_ede)", 353 .cra_name = "cbc(des3_ede)",
359 .cra_driver_name = "cbc-des3_ede-s390", 354 .cra_driver_name = "cbc-des3_ede-s390",
360 .cra_priority = CRYPT_S390_COMPOSITE_PRIORITY, 355 .cra_priority = CRYPT_S390_COMPOSITE_PRIORITY,
361 .cra_flags = CRYPTO_ALG_TYPE_BLKCIPHER, 356 .cra_flags = CRYPTO_ALG_TYPE_BLKCIPHER,
362 .cra_blocksize = DES_BLOCK_SIZE, 357 .cra_blocksize = DES_BLOCK_SIZE,
363 .cra_ctxsize = sizeof(struct crypt_s390_des3_192_ctx), 358 .cra_ctxsize = sizeof(struct s390_des_ctx),
364 .cra_type = &crypto_blkcipher_type, 359 .cra_type = &crypto_blkcipher_type,
365 .cra_module = THIS_MODULE, 360 .cra_module = THIS_MODULE,
366 .cra_list = LIST_HEAD_INIT( 361 .cra_list = LIST_HEAD_INIT(
367 cbc_des3_192_alg.cra_list), 362 cbc_des3_alg.cra_list),
368 .cra_u = { 363 .cra_u = {
369 .blkcipher = { 364 .blkcipher = {
370 .min_keysize = DES3_192_KEY_SIZE, 365 .min_keysize = DES3_KEY_SIZE,
371 .max_keysize = DES3_192_KEY_SIZE, 366 .max_keysize = DES3_KEY_SIZE,
372 .ivsize = DES_BLOCK_SIZE, 367 .ivsize = DES_BLOCK_SIZE,
373 .setkey = des3_192_setkey, 368 .setkey = des3_setkey,
374 .encrypt = cbc_des3_192_encrypt, 369 .encrypt = cbc_des3_encrypt,
375 .decrypt = cbc_des3_192_decrypt, 370 .decrypt = cbc_des3_decrypt,
376 } 371 }
377 } 372 }
378}; 373};
379 374
380static int des_s390_init(void) 375static int ctr_desall_crypt(struct blkcipher_desc *desc, long func,
376 struct s390_des_ctx *ctx, struct blkcipher_walk *walk)
377{
378 int ret = blkcipher_walk_virt_block(desc, walk, DES_BLOCK_SIZE);
379 unsigned int i, n, nbytes;
380 u8 buf[DES_BLOCK_SIZE];
381 u8 *out, *in;
382
383 memcpy(ctrblk, walk->iv, DES_BLOCK_SIZE);
384 while ((nbytes = walk->nbytes) >= DES_BLOCK_SIZE) {
385 out = walk->dst.virt.addr;
386 in = walk->src.virt.addr;
387 while (nbytes >= DES_BLOCK_SIZE) {
388 /* align to block size, max. PAGE_SIZE */
389 n = (nbytes > PAGE_SIZE) ? PAGE_SIZE :
390 nbytes & ~(DES_BLOCK_SIZE - 1);
391 for (i = DES_BLOCK_SIZE; i < n; i += DES_BLOCK_SIZE) {
392 memcpy(ctrblk + i, ctrblk + i - DES_BLOCK_SIZE,
393 DES_BLOCK_SIZE);
394 crypto_inc(ctrblk + i, DES_BLOCK_SIZE);
395 }
396 ret = crypt_s390_kmctr(func, ctx->key, out, in, n, ctrblk);
397 BUG_ON((ret < 0) || (ret != n));
398 if (n > DES_BLOCK_SIZE)
399 memcpy(ctrblk, ctrblk + n - DES_BLOCK_SIZE,
400 DES_BLOCK_SIZE);
401 crypto_inc(ctrblk, DES_BLOCK_SIZE);
402 out += n;
403 in += n;
404 nbytes -= n;
405 }
406 ret = blkcipher_walk_done(desc, walk, nbytes);
407 }
408
409 /* final block may be < DES_BLOCK_SIZE, copy only nbytes */
410 if (nbytes) {
411 out = walk->dst.virt.addr;
412 in = walk->src.virt.addr;
413 ret = crypt_s390_kmctr(func, ctx->key, buf, in,
414 DES_BLOCK_SIZE, ctrblk);
415 BUG_ON(ret < 0 || ret != DES_BLOCK_SIZE);
416 memcpy(out, buf, nbytes);
417 crypto_inc(ctrblk, DES_BLOCK_SIZE);
418 ret = blkcipher_walk_done(desc, walk, 0);
419 }
420 memcpy(walk->iv, ctrblk, DES_BLOCK_SIZE);
421 return ret;
422}
423
424static int ctr_des_encrypt(struct blkcipher_desc *desc,
425 struct scatterlist *dst, struct scatterlist *src,
426 unsigned int nbytes)
427{
428 struct s390_des_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
429 struct blkcipher_walk walk;
430
431 blkcipher_walk_init(&walk, dst, src, nbytes);
432 return ctr_desall_crypt(desc, KMCTR_DEA_ENCRYPT, ctx, &walk);
433}
434
435static int ctr_des_decrypt(struct blkcipher_desc *desc,
436 struct scatterlist *dst, struct scatterlist *src,
437 unsigned int nbytes)
438{
439 struct s390_des_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
440 struct blkcipher_walk walk;
441
442 blkcipher_walk_init(&walk, dst, src, nbytes);
443 return ctr_desall_crypt(desc, KMCTR_DEA_DECRYPT, ctx, &walk);
444}
445
446static struct crypto_alg ctr_des_alg = {
447 .cra_name = "ctr(des)",
448 .cra_driver_name = "ctr-des-s390",
449 .cra_priority = CRYPT_S390_COMPOSITE_PRIORITY,
450 .cra_flags = CRYPTO_ALG_TYPE_BLKCIPHER,
451 .cra_blocksize = 1,
452 .cra_ctxsize = sizeof(struct s390_des_ctx),
453 .cra_type = &crypto_blkcipher_type,
454 .cra_module = THIS_MODULE,
455 .cra_list = LIST_HEAD_INIT(ctr_des_alg.cra_list),
456 .cra_u = {
457 .blkcipher = {
458 .min_keysize = DES_KEY_SIZE,
459 .max_keysize = DES_KEY_SIZE,
460 .ivsize = DES_BLOCK_SIZE,
461 .setkey = des_setkey,
462 .encrypt = ctr_des_encrypt,
463 .decrypt = ctr_des_decrypt,
464 }
465 }
466};
467
468static int ctr_des3_encrypt(struct blkcipher_desc *desc,
469 struct scatterlist *dst, struct scatterlist *src,
470 unsigned int nbytes)
471{
472 struct s390_des_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
473 struct blkcipher_walk walk;
474
475 blkcipher_walk_init(&walk, dst, src, nbytes);
476 return ctr_desall_crypt(desc, KMCTR_TDEA_192_ENCRYPT, ctx, &walk);
477}
478
479static int ctr_des3_decrypt(struct blkcipher_desc *desc,
480 struct scatterlist *dst, struct scatterlist *src,
481 unsigned int nbytes)
482{
483 struct s390_des_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
484 struct blkcipher_walk walk;
485
486 blkcipher_walk_init(&walk, dst, src, nbytes);
487 return ctr_desall_crypt(desc, KMCTR_TDEA_192_DECRYPT, ctx, &walk);
488}
489
490static struct crypto_alg ctr_des3_alg = {
491 .cra_name = "ctr(des3_ede)",
492 .cra_driver_name = "ctr-des3_ede-s390",
493 .cra_priority = CRYPT_S390_COMPOSITE_PRIORITY,
494 .cra_flags = CRYPTO_ALG_TYPE_BLKCIPHER,
495 .cra_blocksize = 1,
496 .cra_ctxsize = sizeof(struct s390_des_ctx),
497 .cra_type = &crypto_blkcipher_type,
498 .cra_module = THIS_MODULE,
499 .cra_list = LIST_HEAD_INIT(ctr_des3_alg.cra_list),
500 .cra_u = {
501 .blkcipher = {
502 .min_keysize = DES3_KEY_SIZE,
503 .max_keysize = DES3_KEY_SIZE,
504 .ivsize = DES_BLOCK_SIZE,
505 .setkey = des3_setkey,
506 .encrypt = ctr_des3_encrypt,
507 .decrypt = ctr_des3_decrypt,
508 }
509 }
510};
511
512static int __init des_s390_init(void)
381{ 513{
382 int ret; 514 int ret;
383 515
384 if (!crypt_s390_func_available(KM_DEA_ENCRYPT) || 516 if (!crypt_s390_func_available(KM_DEA_ENCRYPT, CRYPT_S390_MSA) ||
385 !crypt_s390_func_available(KM_TDEA_192_ENCRYPT)) 517 !crypt_s390_func_available(KM_TDEA_192_ENCRYPT, CRYPT_S390_MSA))
386 return -EOPNOTSUPP; 518 return -EOPNOTSUPP;
387 519
388 ret = crypto_register_alg(&des_alg); 520 ret = crypto_register_alg(&des_alg);
@@ -394,23 +526,46 @@ static int des_s390_init(void)
394 ret = crypto_register_alg(&cbc_des_alg); 526 ret = crypto_register_alg(&cbc_des_alg);
395 if (ret) 527 if (ret)
396 goto cbc_des_err; 528 goto cbc_des_err;
397 ret = crypto_register_alg(&des3_192_alg); 529 ret = crypto_register_alg(&des3_alg);
398 if (ret) 530 if (ret)
399 goto des3_192_err; 531 goto des3_err;
400 ret = crypto_register_alg(&ecb_des3_192_alg); 532 ret = crypto_register_alg(&ecb_des3_alg);
401 if (ret) 533 if (ret)
402 goto ecb_des3_192_err; 534 goto ecb_des3_err;
403 ret = crypto_register_alg(&cbc_des3_192_alg); 535 ret = crypto_register_alg(&cbc_des3_alg);
404 if (ret) 536 if (ret)
405 goto cbc_des3_192_err; 537 goto cbc_des3_err;
538
539 if (crypt_s390_func_available(KMCTR_DEA_ENCRYPT,
540 CRYPT_S390_MSA | CRYPT_S390_MSA4) &&
541 crypt_s390_func_available(KMCTR_TDEA_192_ENCRYPT,
542 CRYPT_S390_MSA | CRYPT_S390_MSA4)) {
543 ret = crypto_register_alg(&ctr_des_alg);
544 if (ret)
545 goto ctr_des_err;
546 ret = crypto_register_alg(&ctr_des3_alg);
547 if (ret)
548 goto ctr_des3_err;
549 ctrblk = (u8 *) __get_free_page(GFP_KERNEL);
550 if (!ctrblk) {
551 ret = -ENOMEM;
552 goto ctr_mem_err;
553 }
554 }
406out: 555out:
407 return ret; 556 return ret;
408 557
409cbc_des3_192_err: 558ctr_mem_err:
410 crypto_unregister_alg(&ecb_des3_192_alg); 559 crypto_unregister_alg(&ctr_des3_alg);
411ecb_des3_192_err: 560ctr_des3_err:
412 crypto_unregister_alg(&des3_192_alg); 561 crypto_unregister_alg(&ctr_des_alg);
413des3_192_err: 562ctr_des_err:
563 crypto_unregister_alg(&cbc_des3_alg);
564cbc_des3_err:
565 crypto_unregister_alg(&ecb_des3_alg);
566ecb_des3_err:
567 crypto_unregister_alg(&des3_alg);
568des3_err:
414 crypto_unregister_alg(&cbc_des_alg); 569 crypto_unregister_alg(&cbc_des_alg);
415cbc_des_err: 570cbc_des_err:
416 crypto_unregister_alg(&ecb_des_alg); 571 crypto_unregister_alg(&ecb_des_alg);
@@ -422,9 +577,14 @@ des_err:
422 577
423static void __exit des_s390_exit(void) 578static void __exit des_s390_exit(void)
424{ 579{
425 crypto_unregister_alg(&cbc_des3_192_alg); 580 if (ctrblk) {
426 crypto_unregister_alg(&ecb_des3_192_alg); 581 crypto_unregister_alg(&ctr_des_alg);
427 crypto_unregister_alg(&des3_192_alg); 582 crypto_unregister_alg(&ctr_des3_alg);
583 free_page((unsigned long) ctrblk);
584 }
585 crypto_unregister_alg(&cbc_des3_alg);
586 crypto_unregister_alg(&ecb_des3_alg);
587 crypto_unregister_alg(&des3_alg);
428 crypto_unregister_alg(&cbc_des_alg); 588 crypto_unregister_alg(&cbc_des_alg);
429 crypto_unregister_alg(&ecb_des_alg); 589 crypto_unregister_alg(&ecb_des_alg);
430 crypto_unregister_alg(&des_alg); 590 crypto_unregister_alg(&des_alg);
diff --git a/arch/s390/crypto/ghash_s390.c b/arch/s390/crypto/ghash_s390.c
new file mode 100644
index 000000000000..b1bd170f24b1
--- /dev/null
+++ b/arch/s390/crypto/ghash_s390.c
@@ -0,0 +1,162 @@
1/*
2 * Cryptographic API.
3 *
4 * s390 implementation of the GHASH algorithm for GCM (Galois/Counter Mode).
5 *
6 * Copyright IBM Corp. 2011
7 * Author(s): Gerald Schaefer <gerald.schaefer@de.ibm.com>
8 */
9
10#include <crypto/internal/hash.h>
11#include <linux/module.h>
12
13#include "crypt_s390.h"
14
15#define GHASH_BLOCK_SIZE 16
16#define GHASH_DIGEST_SIZE 16
17
18struct ghash_ctx {
19 u8 icv[16];
20 u8 key[16];
21};
22
23struct ghash_desc_ctx {
24 u8 buffer[GHASH_BLOCK_SIZE];
25 u32 bytes;
26};
27
28static int ghash_init(struct shash_desc *desc)
29{
30 struct ghash_desc_ctx *dctx = shash_desc_ctx(desc);
31
32 memset(dctx, 0, sizeof(*dctx));
33
34 return 0;
35}
36
37static int ghash_setkey(struct crypto_shash *tfm,
38 const u8 *key, unsigned int keylen)
39{
40 struct ghash_ctx *ctx = crypto_shash_ctx(tfm);
41
42 if (keylen != GHASH_BLOCK_SIZE) {
43 crypto_shash_set_flags(tfm, CRYPTO_TFM_RES_BAD_KEY_LEN);
44 return -EINVAL;
45 }
46
47 memcpy(ctx->key, key, GHASH_BLOCK_SIZE);
48 memset(ctx->icv, 0, GHASH_BLOCK_SIZE);
49
50 return 0;
51}
52
53static int ghash_update(struct shash_desc *desc,
54 const u8 *src, unsigned int srclen)
55{
56 struct ghash_desc_ctx *dctx = shash_desc_ctx(desc);
57 struct ghash_ctx *ctx = crypto_shash_ctx(desc->tfm);
58 unsigned int n;
59 u8 *buf = dctx->buffer;
60 int ret;
61
62 if (dctx->bytes) {
63 u8 *pos = buf + (GHASH_BLOCK_SIZE - dctx->bytes);
64
65 n = min(srclen, dctx->bytes);
66 dctx->bytes -= n;
67 srclen -= n;
68
69 memcpy(pos, src, n);
70 src += n;
71
72 if (!dctx->bytes) {
73 ret = crypt_s390_kimd(KIMD_GHASH, ctx, buf,
74 GHASH_BLOCK_SIZE);
75 BUG_ON(ret != GHASH_BLOCK_SIZE);
76 }
77 }
78
79 n = srclen & ~(GHASH_BLOCK_SIZE - 1);
80 if (n) {
81 ret = crypt_s390_kimd(KIMD_GHASH, ctx, src, n);
82 BUG_ON(ret != n);
83 src += n;
84 srclen -= n;
85 }
86
87 if (srclen) {
88 dctx->bytes = GHASH_BLOCK_SIZE - srclen;
89 memcpy(buf, src, srclen);
90 }
91
92 return 0;
93}
94
95static void ghash_flush(struct ghash_ctx *ctx, struct ghash_desc_ctx *dctx)
96{
97 u8 *buf = dctx->buffer;
98 int ret;
99
100 if (dctx->bytes) {
101 u8 *pos = buf + (GHASH_BLOCK_SIZE - dctx->bytes);
102
103 memset(pos, 0, dctx->bytes);
104
105 ret = crypt_s390_kimd(KIMD_GHASH, ctx, buf, GHASH_BLOCK_SIZE);
106 BUG_ON(ret != GHASH_BLOCK_SIZE);
107 }
108
109 dctx->bytes = 0;
110}
111
112static int ghash_final(struct shash_desc *desc, u8 *dst)
113{
114 struct ghash_desc_ctx *dctx = shash_desc_ctx(desc);
115 struct ghash_ctx *ctx = crypto_shash_ctx(desc->tfm);
116
117 ghash_flush(ctx, dctx);
118 memcpy(dst, ctx->icv, GHASH_BLOCK_SIZE);
119
120 return 0;
121}
122
123static struct shash_alg ghash_alg = {
124 .digestsize = GHASH_DIGEST_SIZE,
125 .init = ghash_init,
126 .update = ghash_update,
127 .final = ghash_final,
128 .setkey = ghash_setkey,
129 .descsize = sizeof(struct ghash_desc_ctx),
130 .base = {
131 .cra_name = "ghash",
132 .cra_driver_name = "ghash-s390",
133 .cra_priority = CRYPT_S390_PRIORITY,
134 .cra_flags = CRYPTO_ALG_TYPE_SHASH,
135 .cra_blocksize = GHASH_BLOCK_SIZE,
136 .cra_ctxsize = sizeof(struct ghash_ctx),
137 .cra_module = THIS_MODULE,
138 .cra_list = LIST_HEAD_INIT(ghash_alg.base.cra_list),
139 },
140};
141
142static int __init ghash_mod_init(void)
143{
144 if (!crypt_s390_func_available(KIMD_GHASH,
145 CRYPT_S390_MSA | CRYPT_S390_MSA4))
146 return -EOPNOTSUPP;
147
148 return crypto_register_shash(&ghash_alg);
149}
150
151static void __exit ghash_mod_exit(void)
152{
153 crypto_unregister_shash(&ghash_alg);
154}
155
156module_init(ghash_mod_init);
157module_exit(ghash_mod_exit);
158
159MODULE_ALIAS("ghash");
160
161MODULE_LICENSE("GPL");
162MODULE_DESCRIPTION("GHASH Message Digest Algorithm, s390 implementation");
diff --git a/arch/s390/crypto/prng.c b/arch/s390/crypto/prng.c
index aa819dac2360..0808fbf0f7d3 100644
--- a/arch/s390/crypto/prng.c
+++ b/arch/s390/crypto/prng.c
@@ -76,7 +76,7 @@ static void prng_seed(int nbytes)
76 76
77 /* Add the entropy */ 77 /* Add the entropy */
78 while (nbytes >= 8) { 78 while (nbytes >= 8) {
79 *((__u64 *)parm_block) ^= *((__u64 *)buf+i*8); 79 *((__u64 *)parm_block) ^= *((__u64 *)(buf+i));
80 prng_add_entropy(); 80 prng_add_entropy();
81 i += 8; 81 i += 8;
82 nbytes -= 8; 82 nbytes -= 8;
@@ -152,6 +152,7 @@ static const struct file_operations prng_fops = {
152 .open = &prng_open, 152 .open = &prng_open,
153 .release = NULL, 153 .release = NULL,
154 .read = &prng_read, 154 .read = &prng_read,
155 .llseek = noop_llseek,
155}; 156};
156 157
157static struct miscdevice prng_dev = { 158static struct miscdevice prng_dev = {
@@ -165,7 +166,7 @@ static int __init prng_init(void)
165 int ret; 166 int ret;
166 167
167 /* check if the CPU has a PRNG */ 168 /* check if the CPU has a PRNG */
168 if (!crypt_s390_func_available(KMC_PRNG)) 169 if (!crypt_s390_func_available(KMC_PRNG, CRYPT_S390_MSA))
169 return -EOPNOTSUPP; 170 return -EOPNOTSUPP;
170 171
171 if (prng_chunk_size < 8) 172 if (prng_chunk_size < 8)
diff --git a/arch/s390/crypto/sha1_s390.c b/arch/s390/crypto/sha1_s390.c
index f6de7826c979..e9868c6e0a08 100644
--- a/arch/s390/crypto/sha1_s390.c
+++ b/arch/s390/crypto/sha1_s390.c
@@ -90,7 +90,7 @@ static struct shash_alg alg = {
90 90
91static int __init sha1_s390_init(void) 91static int __init sha1_s390_init(void)
92{ 92{
93 if (!crypt_s390_func_available(KIMD_SHA_1)) 93 if (!crypt_s390_func_available(KIMD_SHA_1, CRYPT_S390_MSA))
94 return -EOPNOTSUPP; 94 return -EOPNOTSUPP;
95 return crypto_register_shash(&alg); 95 return crypto_register_shash(&alg);
96} 96}
diff --git a/arch/s390/crypto/sha256_s390.c b/arch/s390/crypto/sha256_s390.c
index 61a7db372121..5ed8d64fc2ed 100644
--- a/arch/s390/crypto/sha256_s390.c
+++ b/arch/s390/crypto/sha256_s390.c
@@ -86,7 +86,7 @@ static struct shash_alg alg = {
86 86
87static int sha256_s390_init(void) 87static int sha256_s390_init(void)
88{ 88{
89 if (!crypt_s390_func_available(KIMD_SHA_256)) 89 if (!crypt_s390_func_available(KIMD_SHA_256, CRYPT_S390_MSA))
90 return -EOPNOTSUPP; 90 return -EOPNOTSUPP;
91 91
92 return crypto_register_shash(&alg); 92 return crypto_register_shash(&alg);
diff --git a/arch/s390/crypto/sha512_s390.c b/arch/s390/crypto/sha512_s390.c
index 4bf73d0dc525..32a81383b69c 100644
--- a/arch/s390/crypto/sha512_s390.c
+++ b/arch/s390/crypto/sha512_s390.c
@@ -132,7 +132,7 @@ static int __init init(void)
132{ 132{
133 int ret; 133 int ret;
134 134
135 if (!crypt_s390_func_available(KIMD_SHA_512)) 135 if (!crypt_s390_func_available(KIMD_SHA_512, CRYPT_S390_MSA))
136 return -EOPNOTSUPP; 136 return -EOPNOTSUPP;
137 if ((ret = crypto_register_shash(&sha512_alg)) < 0) 137 if ((ret = crypto_register_shash(&sha512_alg)) < 0)
138 goto out; 138 goto out;
diff --git a/arch/s390/crypto/sha_common.c b/arch/s390/crypto/sha_common.c
index f42dbabc0d30..48884f89ab92 100644
--- a/arch/s390/crypto/sha_common.c
+++ b/arch/s390/crypto/sha_common.c
@@ -38,6 +38,7 @@ int s390_sha_update(struct shash_desc *desc, const u8 *data, unsigned int len)
38 BUG_ON(ret != bsize); 38 BUG_ON(ret != bsize);
39 data += bsize - index; 39 data += bsize - index;
40 len -= bsize - index; 40 len -= bsize - index;
41 index = 0;
41 } 42 }
42 43
43 /* process as many blocks as possible */ 44 /* process as many blocks as possible */
diff --git a/arch/s390/defconfig b/arch/s390/defconfig
index e40ac6ee6526..29c82c640a88 100644
--- a/arch/s390/defconfig
+++ b/arch/s390/defconfig
@@ -2,17 +2,24 @@ CONFIG_EXPERIMENTAL=y
2CONFIG_SYSVIPC=y 2CONFIG_SYSVIPC=y
3CONFIG_POSIX_MQUEUE=y 3CONFIG_POSIX_MQUEUE=y
4CONFIG_AUDIT=y 4CONFIG_AUDIT=y
5CONFIG_RCU_TRACE=y
5CONFIG_IKCONFIG=y 6CONFIG_IKCONFIG=y
6CONFIG_IKCONFIG_PROC=y 7CONFIG_IKCONFIG_PROC=y
7CONFIG_CGROUPS=y 8CONFIG_CGROUPS=y
8CONFIG_CGROUP_NS=y 9CONFIG_CPUSETS=y
9CONFIG_SYSFS_DEPRECATED_V2=y 10CONFIG_CGROUP_CPUACCT=y
10CONFIG_UTS_NS=y 11CONFIG_RESOURCE_COUNTERS=y
11CONFIG_IPC_NS=y 12CONFIG_CGROUP_MEM_RES_CTLR=y
13CONFIG_CGROUP_MEM_RES_CTLR_SWAP=y
14CONFIG_CGROUP_SCHED=y
15CONFIG_RT_GROUP_SCHED=y
16CONFIG_BLK_CGROUP=y
12CONFIG_BLK_DEV_INITRD=y 17CONFIG_BLK_DEV_INITRD=y
13# CONFIG_CC_OPTIMIZE_FOR_SIZE is not set 18# CONFIG_CC_OPTIMIZE_FOR_SIZE is not set
14# CONFIG_COMPAT_BRK is not set 19# CONFIG_COMPAT_BRK is not set
15CONFIG_SLAB=y 20CONFIG_SLAB=y
21CONFIG_PROFILING=y
22CONFIG_OPROFILE=y
16CONFIG_KPROBES=y 23CONFIG_KPROBES=y
17CONFIG_MODULES=y 24CONFIG_MODULES=y
18CONFIG_MODULE_UNLOAD=y 25CONFIG_MODULE_UNLOAD=y
@@ -20,24 +27,14 @@ CONFIG_MODVERSIONS=y
20CONFIG_DEFAULT_DEADLINE=y 27CONFIG_DEFAULT_DEADLINE=y
21CONFIG_NO_HZ=y 28CONFIG_NO_HZ=y
22CONFIG_HIGH_RES_TIMERS=y 29CONFIG_HIGH_RES_TIMERS=y
23CONFIG_64BIT=y
24CONFIG_SMP=y
25CONFIG_NR_CPUS=32
26CONFIG_COMPAT=y
27CONFIG_S390_EXEC_PROTECT=y
28CONFIG_PACK_STACK=y
29CONFIG_CHECK_STACK=y
30CONFIG_PREEMPT=y 30CONFIG_PREEMPT=y
31CONFIG_MEMORY_HOTPLUG=y 31CONFIG_MEMORY_HOTPLUG=y
32CONFIG_MEMORY_HOTREMOVE=y 32CONFIG_MEMORY_HOTREMOVE=y
33CONFIG_QDIO=y 33CONFIG_KSM=y
34CONFIG_CHSC_SCH=m
35CONFIG_IPL=y
36CONFIG_BINFMT_MISC=m 34CONFIG_BINFMT_MISC=m
37CONFIG_PFAULT=y 35CONFIG_CMM=m
38CONFIG_HZ_100=y 36CONFIG_HZ_100=y
39CONFIG_KEXEC=y 37CONFIG_KEXEC=y
40CONFIG_S390_GUEST=y
41CONFIG_PM=y 38CONFIG_PM=y
42CONFIG_HIBERNATION=y 39CONFIG_HIBERNATION=y
43CONFIG_PACKET=y 40CONFIG_PACKET=y
@@ -46,16 +43,15 @@ CONFIG_NET_KEY=y
46CONFIG_AFIUCV=m 43CONFIG_AFIUCV=m
47CONFIG_INET=y 44CONFIG_INET=y
48CONFIG_IP_MULTICAST=y 45CONFIG_IP_MULTICAST=y
46# CONFIG_INET_LRO is not set
49CONFIG_IPV6=y 47CONFIG_IPV6=y
50CONFIG_NETFILTER=y 48CONFIG_NET_SCTPPROBE=m
51CONFIG_NETFILTER_NETLINK_QUEUE=m 49CONFIG_L2TP=m
52CONFIG_NETFILTER_NETLINK_LOG=m 50CONFIG_L2TP_DEBUGFS=m
53CONFIG_NF_CONNTRACK=m 51CONFIG_VLAN_8021Q=y
54# CONFIG_NF_CT_PROTO_SCTP is not set
55CONFIG_NET_SCHED=y 52CONFIG_NET_SCHED=y
56CONFIG_NET_SCH_CBQ=m 53CONFIG_NET_SCH_CBQ=m
57CONFIG_NET_SCH_PRIO=m 54CONFIG_NET_SCH_PRIO=m
58CONFIG_NET_SCH_MULTIQ=y
59CONFIG_NET_SCH_RED=m 55CONFIG_NET_SCH_RED=m
60CONFIG_NET_SCH_SFQ=m 56CONFIG_NET_SCH_SFQ=m
61CONFIG_NET_SCH_TEQL=m 57CONFIG_NET_SCH_TEQL=m
@@ -69,28 +65,14 @@ CONFIG_NET_CLS_U32=m
69CONFIG_CLS_U32_MARK=y 65CONFIG_CLS_U32_MARK=y
70CONFIG_NET_CLS_RSVP=m 66CONFIG_NET_CLS_RSVP=m
71CONFIG_NET_CLS_RSVP6=m 67CONFIG_NET_CLS_RSVP6=m
72CONFIG_NET_CLS_FLOW=m
73CONFIG_NET_CLS_ACT=y 68CONFIG_NET_CLS_ACT=y
74CONFIG_NET_ACT_POLICE=y 69CONFIG_NET_ACT_POLICE=y
75CONFIG_NET_ACT_NAT=m
76CONFIG_CAN=m
77CONFIG_CAN_RAW=m
78CONFIG_CAN_BCM=m
79CONFIG_CAN_VCAN=m
80CONFIG_UEVENT_HELPER_PATH="/sbin/hotplug" 70CONFIG_UEVENT_HELPER_PATH="/sbin/hotplug"
81# CONFIG_FIRMWARE_IN_KERNEL is not set 71# CONFIG_FIRMWARE_IN_KERNEL is not set
82CONFIG_BLK_DEV_LOOP=m 72CONFIG_BLK_DEV_LOOP=m
83CONFIG_BLK_DEV_NBD=m 73CONFIG_BLK_DEV_NBD=m
84CONFIG_BLK_DEV_RAM=y 74CONFIG_BLK_DEV_RAM=y
85CONFIG_BLK_DEV_XIP=y 75CONFIG_VIRTIO_BLK=y
86CONFIG_BLK_DEV_XPRAM=m
87CONFIG_DASD=y
88CONFIG_DASD_PROFILE=y
89CONFIG_DASD_ECKD=y
90CONFIG_DASD_FBA=y
91CONFIG_DASD_DIAG=y
92CONFIG_DASD_EER=y
93CONFIG_VIRTIO_BLK=m
94CONFIG_SCSI=y 76CONFIG_SCSI=y
95CONFIG_BLK_DEV_SD=y 77CONFIG_BLK_DEV_SD=y
96CONFIG_CHR_DEV_ST=y 78CONFIG_CHR_DEV_ST=y
@@ -102,101 +84,93 @@ CONFIG_SCSI_CONSTANTS=y
102CONFIG_SCSI_LOGGING=y 84CONFIG_SCSI_LOGGING=y
103CONFIG_SCSI_SCAN_ASYNC=y 85CONFIG_SCSI_SCAN_ASYNC=y
104CONFIG_ZFCP=y 86CONFIG_ZFCP=y
105CONFIG_SCSI_DH=m 87CONFIG_ZFCP_DIF=y
106CONFIG_SCSI_DH_RDAC=m
107CONFIG_SCSI_DH_HP_SW=m
108CONFIG_SCSI_DH_EMC=m
109CONFIG_SCSI_DH_ALUA=m
110CONFIG_SCSI_OSD_INITIATOR=m
111CONFIG_SCSI_OSD_ULD=m
112CONFIG_MD=y
113CONFIG_BLK_DEV_MD=y
114CONFIG_MD_LINEAR=m
115CONFIG_MD_RAID0=m
116CONFIG_MD_RAID1=m
117CONFIG_MD_MULTIPATH=m
118CONFIG_BLK_DEV_DM=y
119CONFIG_DM_CRYPT=y
120CONFIG_DM_SNAPSHOT=y
121CONFIG_DM_MIRROR=y
122CONFIG_DM_ZERO=y
123CONFIG_DM_MULTIPATH=m
124CONFIG_NETDEVICES=y 88CONFIG_NETDEVICES=y
125CONFIG_DUMMY=m 89CONFIG_DUMMY=m
126CONFIG_BONDING=m 90CONFIG_BONDING=m
127CONFIG_EQUALIZER=m 91CONFIG_EQUALIZER=m
128CONFIG_TUN=m 92CONFIG_TUN=m
129CONFIG_VETH=m
130CONFIG_NET_ETHERNET=y 93CONFIG_NET_ETHERNET=y
131CONFIG_LCS=m 94CONFIG_VIRTIO_NET=y
132CONFIG_CTCM=m
133CONFIG_QETH=y
134CONFIG_QETH_L2=y
135CONFIG_QETH_L3=y
136CONFIG_VIRTIO_NET=m
137CONFIG_HW_RANDOM_VIRTIO=m
138CONFIG_RAW_DRIVER=m 95CONFIG_RAW_DRIVER=m
139CONFIG_TN3270=y
140CONFIG_TN3270_TTY=y
141CONFIG_TN3270_FS=m
142CONFIG_TN3270_CONSOLE=y
143CONFIG_TN3215=y
144CONFIG_TN3215_CONSOLE=y
145CONFIG_SCLP_TTY=y
146CONFIG_SCLP_CONSOLE=y
147CONFIG_SCLP_VT220_TTY=y
148CONFIG_SCLP_VT220_CONSOLE=y
149CONFIG_SCLP_CPI=m
150CONFIG_SCLP_ASYNC=m
151CONFIG_S390_TAPE=m
152CONFIG_S390_TAPE_BLOCK=y
153CONFIG_S390_TAPE_34XX=m
154CONFIG_ACCESSIBILITY=y
155CONFIG_EXT2_FS=y 96CONFIG_EXT2_FS=y
156CONFIG_EXT3_FS=y 97CONFIG_EXT3_FS=y
157# CONFIG_EXT3_DEFAULTS_TO_ORDERED is not set 98# CONFIG_EXT3_DEFAULTS_TO_ORDERED is not set
99CONFIG_EXT4_FS=y
100CONFIG_EXT4_FS_POSIX_ACL=y
101CONFIG_EXT4_FS_SECURITY=y
158CONFIG_PROC_KCORE=y 102CONFIG_PROC_KCORE=y
159CONFIG_TMPFS=y 103CONFIG_TMPFS=y
160CONFIG_TMPFS_POSIX_ACL=y 104CONFIG_TMPFS_POSIX_ACL=y
161CONFIG_NFS_FS=y 105# CONFIG_NETWORK_FILESYSTEMS is not set
162CONFIG_NFS_V3=y
163CONFIG_NFSD=y
164CONFIG_NFSD_V3=y
165CONFIG_PARTITION_ADVANCED=y 106CONFIG_PARTITION_ADVANCED=y
166CONFIG_IBM_PARTITION=y 107CONFIG_IBM_PARTITION=y
167CONFIG_DLM=m 108CONFIG_DLM=m
168CONFIG_MAGIC_SYSRQ=y 109CONFIG_MAGIC_SYSRQ=y
169CONFIG_DEBUG_KERNEL=y 110CONFIG_DEBUG_KERNEL=y
170# CONFIG_SCHED_DEBUG is not set 111CONFIG_TIMER_STATS=y
171CONFIG_DEBUG_SPINLOCK=y 112CONFIG_PROVE_LOCKING=y
172CONFIG_DEBUG_MUTEXES=y 113CONFIG_PROVE_RCU=y
114CONFIG_LOCK_STAT=y
115CONFIG_DEBUG_LOCKDEP=y
173CONFIG_DEBUG_SPINLOCK_SLEEP=y 116CONFIG_DEBUG_SPINLOCK_SLEEP=y
117CONFIG_DEBUG_LIST=y
118CONFIG_DEBUG_NOTIFIERS=y
174# CONFIG_RCU_CPU_STALL_DETECTOR is not set 119# CONFIG_RCU_CPU_STALL_DETECTOR is not set
120CONFIG_KPROBES_SANITY_TEST=y
175CONFIG_DEBUG_FORCE_WEAK_PER_CPU=y 121CONFIG_DEBUG_FORCE_WEAK_PER_CPU=y
122CONFIG_CPU_NOTIFIER_ERROR_INJECT=m
123CONFIG_LATENCYTOP=y
176CONFIG_SYSCTL_SYSCALL_CHECK=y 124CONFIG_SYSCTL_SYSCALL_CHECK=y
177CONFIG_SAMPLES=y 125CONFIG_DEBUG_PAGEALLOC=y
178CONFIG_CRYPTO_FIPS=y 126# CONFIG_FTRACE is not set
127# CONFIG_STRICT_DEVMEM is not set
128CONFIG_CRYPTO_NULL=m
129CONFIG_CRYPTO_CRYPTD=m
179CONFIG_CRYPTO_AUTHENC=m 130CONFIG_CRYPTO_AUTHENC=m
131CONFIG_CRYPTO_TEST=m
180CONFIG_CRYPTO_CCM=m 132CONFIG_CRYPTO_CCM=m
181CONFIG_CRYPTO_GCM=m 133CONFIG_CRYPTO_GCM=m
134CONFIG_CRYPTO_CBC=y
182CONFIG_CRYPTO_CTS=m 135CONFIG_CRYPTO_CTS=m
183CONFIG_CRYPTO_ECB=m 136CONFIG_CRYPTO_ECB=m
137CONFIG_CRYPTO_LRW=m
184CONFIG_CRYPTO_PCBC=m 138CONFIG_CRYPTO_PCBC=m
139CONFIG_CRYPTO_XTS=m
140CONFIG_CRYPTO_XCBC=m
185CONFIG_CRYPTO_VMAC=m 141CONFIG_CRYPTO_VMAC=m
142CONFIG_CRYPTO_MD4=m
143CONFIG_CRYPTO_MICHAEL_MIC=m
186CONFIG_CRYPTO_RMD128=m 144CONFIG_CRYPTO_RMD128=m
187CONFIG_CRYPTO_RMD160=m 145CONFIG_CRYPTO_RMD160=m
188CONFIG_CRYPTO_RMD256=m 146CONFIG_CRYPTO_RMD256=m
189CONFIG_CRYPTO_RMD320=m 147CONFIG_CRYPTO_RMD320=m
148CONFIG_CRYPTO_SHA256=m
149CONFIG_CRYPTO_SHA512=m
150CONFIG_CRYPTO_TGR192=m
151CONFIG_CRYPTO_WP512=m
152CONFIG_CRYPTO_ANUBIS=m
153CONFIG_CRYPTO_ARC4=m
154CONFIG_CRYPTO_BLOWFISH=m
190CONFIG_CRYPTO_CAMELLIA=m 155CONFIG_CRYPTO_CAMELLIA=m
156CONFIG_CRYPTO_CAST5=m
157CONFIG_CRYPTO_CAST6=m
158CONFIG_CRYPTO_DES=m
191CONFIG_CRYPTO_FCRYPT=m 159CONFIG_CRYPTO_FCRYPT=m
160CONFIG_CRYPTO_KHAZAD=m
192CONFIG_CRYPTO_SALSA20=m 161CONFIG_CRYPTO_SALSA20=m
193CONFIG_CRYPTO_SEED=m 162CONFIG_CRYPTO_SEED=m
163CONFIG_CRYPTO_SERPENT=m
164CONFIG_CRYPTO_TEA=m
165CONFIG_CRYPTO_TWOFISH=m
166CONFIG_CRYPTO_DEFLATE=m
194CONFIG_CRYPTO_ZLIB=m 167CONFIG_CRYPTO_ZLIB=m
195CONFIG_CRYPTO_LZO=m 168CONFIG_CRYPTO_LZO=m
196CONFIG_ZCRYPT=m 169CONFIG_ZCRYPT=m
170CONFIG_CRYPTO_SHA1_S390=m
171CONFIG_CRYPTO_SHA256_S390=m
197CONFIG_CRYPTO_SHA512_S390=m 172CONFIG_CRYPTO_SHA512_S390=m
198CONFIG_CRC_T10DIF=y 173CONFIG_CRYPTO_DES_S390=m
199CONFIG_CRC32=m 174CONFIG_CRYPTO_AES_S390=m
200CONFIG_CRC7=m 175CONFIG_CRC7=m
201CONFIG_KVM=m 176CONFIG_VIRTIO_BALLOON=y
202CONFIG_VIRTIO_BALLOON=m
diff --git a/arch/s390/hypfs/Makefile b/arch/s390/hypfs/Makefile
index b08d2abf6178..2e671d5004ca 100644
--- a/arch/s390/hypfs/Makefile
+++ b/arch/s390/hypfs/Makefile
@@ -4,4 +4,4 @@
4 4
5obj-$(CONFIG_S390_HYPFS_FS) += s390_hypfs.o 5obj-$(CONFIG_S390_HYPFS_FS) += s390_hypfs.o
6 6
7s390_hypfs-objs := inode.o hypfs_diag.o hypfs_vm.o 7s390_hypfs-objs := inode.o hypfs_diag.o hypfs_vm.o hypfs_dbfs.o
diff --git a/arch/s390/hypfs/hypfs.h b/arch/s390/hypfs/hypfs.h
index fa487d4cc08b..d9df5a060a83 100644
--- a/arch/s390/hypfs/hypfs.h
+++ b/arch/s390/hypfs/hypfs.h
@@ -12,6 +12,8 @@
12#include <linux/fs.h> 12#include <linux/fs.h>
13#include <linux/types.h> 13#include <linux/types.h>
14#include <linux/debugfs.h> 14#include <linux/debugfs.h>
15#include <linux/workqueue.h>
16#include <linux/kref.h>
15 17
16#define REG_FILE_MODE 0440 18#define REG_FILE_MODE 0440
17#define UPDATE_FILE_MODE 0220 19#define UPDATE_FILE_MODE 0220
@@ -38,6 +40,33 @@ extern int hypfs_vm_init(void);
38extern void hypfs_vm_exit(void); 40extern void hypfs_vm_exit(void);
39extern int hypfs_vm_create_files(struct super_block *sb, struct dentry *root); 41extern int hypfs_vm_create_files(struct super_block *sb, struct dentry *root);
40 42
41/* Directory for debugfs files */ 43/* debugfs interface */
42extern struct dentry *hypfs_dbfs_dir; 44struct hypfs_dbfs_file;
45
46struct hypfs_dbfs_data {
47 void *buf;
48 void *buf_free_ptr;
49 size_t size;
50 struct hypfs_dbfs_file *dbfs_file;
51 struct kref kref;
52};
53
54struct hypfs_dbfs_file {
55 const char *name;
56 int (*data_create)(void **data, void **data_free_ptr,
57 size_t *size);
58 void (*data_free)(const void *buf_free_ptr);
59
60 /* Private data for hypfs_dbfs.c */
61 struct hypfs_dbfs_data *data;
62 struct delayed_work data_free_work;
63 struct mutex lock;
64 struct dentry *dentry;
65};
66
67extern int hypfs_dbfs_init(void);
68extern void hypfs_dbfs_exit(void);
69extern int hypfs_dbfs_create_file(struct hypfs_dbfs_file *df);
70extern void hypfs_dbfs_remove_file(struct hypfs_dbfs_file *df);
71
43#endif /* _HYPFS_H_ */ 72#endif /* _HYPFS_H_ */
diff --git a/arch/s390/hypfs/hypfs_dbfs.c b/arch/s390/hypfs/hypfs_dbfs.c
new file mode 100644
index 000000000000..b478013b7fec
--- /dev/null
+++ b/arch/s390/hypfs/hypfs_dbfs.c
@@ -0,0 +1,116 @@
1/*
2 * Hypervisor filesystem for Linux on s390 - debugfs interface
3 *
4 * Copyright (C) IBM Corp. 2010
5 * Author(s): Michael Holzheu <holzheu@linux.vnet.ibm.com>
6 */
7
8#include <linux/slab.h>
9#include "hypfs.h"
10
11static struct dentry *dbfs_dir;
12
13static struct hypfs_dbfs_data *hypfs_dbfs_data_alloc(struct hypfs_dbfs_file *f)
14{
15 struct hypfs_dbfs_data *data;
16
17 data = kmalloc(sizeof(*data), GFP_KERNEL);
18 if (!data)
19 return NULL;
20 kref_init(&data->kref);
21 data->dbfs_file = f;
22 return data;
23}
24
25static void hypfs_dbfs_data_free(struct kref *kref)
26{
27 struct hypfs_dbfs_data *data;
28
29 data = container_of(kref, struct hypfs_dbfs_data, kref);
30 data->dbfs_file->data_free(data->buf_free_ptr);
31 kfree(data);
32}
33
34static void data_free_delayed(struct work_struct *work)
35{
36 struct hypfs_dbfs_data *data;
37 struct hypfs_dbfs_file *df;
38
39 df = container_of(work, struct hypfs_dbfs_file, data_free_work.work);
40 mutex_lock(&df->lock);
41 data = df->data;
42 df->data = NULL;
43 mutex_unlock(&df->lock);
44 kref_put(&data->kref, hypfs_dbfs_data_free);
45}
46
47static ssize_t dbfs_read(struct file *file, char __user *buf,
48 size_t size, loff_t *ppos)
49{
50 struct hypfs_dbfs_data *data;
51 struct hypfs_dbfs_file *df;
52 ssize_t rc;
53
54 if (*ppos != 0)
55 return 0;
56
57 df = file->f_path.dentry->d_inode->i_private;
58 mutex_lock(&df->lock);
59 if (!df->data) {
60 data = hypfs_dbfs_data_alloc(df);
61 if (!data) {
62 mutex_unlock(&df->lock);
63 return -ENOMEM;
64 }
65 rc = df->data_create(&data->buf, &data->buf_free_ptr,
66 &data->size);
67 if (rc) {
68 mutex_unlock(&df->lock);
69 kfree(data);
70 return rc;
71 }
72 df->data = data;
73 schedule_delayed_work(&df->data_free_work, HZ);
74 }
75 data = df->data;
76 kref_get(&data->kref);
77 mutex_unlock(&df->lock);
78
79 rc = simple_read_from_buffer(buf, size, ppos, data->buf, data->size);
80 kref_put(&data->kref, hypfs_dbfs_data_free);
81 return rc;
82}
83
84static const struct file_operations dbfs_ops = {
85 .read = dbfs_read,
86 .llseek = no_llseek,
87};
88
89int hypfs_dbfs_create_file(struct hypfs_dbfs_file *df)
90{
91 df->dentry = debugfs_create_file(df->name, 0400, dbfs_dir, df,
92 &dbfs_ops);
93 if (IS_ERR(df->dentry))
94 return PTR_ERR(df->dentry);
95 mutex_init(&df->lock);
96 INIT_DELAYED_WORK(&df->data_free_work, data_free_delayed);
97 return 0;
98}
99
100void hypfs_dbfs_remove_file(struct hypfs_dbfs_file *df)
101{
102 debugfs_remove(df->dentry);
103}
104
105int hypfs_dbfs_init(void)
106{
107 dbfs_dir = debugfs_create_dir("s390_hypfs", NULL);
108 if (IS_ERR(dbfs_dir))
109 return PTR_ERR(dbfs_dir);
110 return 0;
111}
112
113void hypfs_dbfs_exit(void)
114{
115 debugfs_remove(dbfs_dir);
116}
diff --git a/arch/s390/hypfs/hypfs_diag.c b/arch/s390/hypfs/hypfs_diag.c
index 1211bb1d2f24..6023c6dc1fb7 100644
--- a/arch/s390/hypfs/hypfs_diag.c
+++ b/arch/s390/hypfs/hypfs_diag.c
@@ -555,80 +555,38 @@ struct dbfs_d204 {
555 char buf[]; /* d204 buffer */ 555 char buf[]; /* d204 buffer */
556} __attribute__ ((packed)); 556} __attribute__ ((packed));
557 557
558struct dbfs_d204_private { 558static int dbfs_d204_create(void **data, void **data_free_ptr, size_t *size)
559 struct dbfs_d204 *d204; /* Aligned d204 data with header */
560 void *base; /* Base pointer (needed for vfree) */
561};
562
563static int dbfs_d204_open(struct inode *inode, struct file *file)
564{ 559{
565 struct dbfs_d204_private *data;
566 struct dbfs_d204 *d204; 560 struct dbfs_d204 *d204;
567 int rc, buf_size; 561 int rc, buf_size;
562 void *base;
568 563
569 data = kzalloc(sizeof(*data), GFP_KERNEL);
570 if (!data)
571 return -ENOMEM;
572 buf_size = PAGE_SIZE * (diag204_buf_pages + 1) + sizeof(d204->hdr); 564 buf_size = PAGE_SIZE * (diag204_buf_pages + 1) + sizeof(d204->hdr);
573 data->base = vmalloc(buf_size); 565 base = vmalloc(buf_size);
574 if (!data->base) { 566 if (!base)
575 rc = -ENOMEM; 567 return -ENOMEM;
576 goto fail_kfree_data; 568 memset(base, 0, buf_size);
569 d204 = page_align_ptr(base + sizeof(d204->hdr)) - sizeof(d204->hdr);
570 rc = diag204_do_store(d204->buf, diag204_buf_pages);
571 if (rc) {
572 vfree(base);
573 return rc;
577 } 574 }
578 memset(data->base, 0, buf_size);
579 d204 = page_align_ptr(data->base + sizeof(d204->hdr))
580 - sizeof(d204->hdr);
581 rc = diag204_do_store(&d204->buf, diag204_buf_pages);
582 if (rc)
583 goto fail_vfree_base;
584 d204->hdr.version = DBFS_D204_HDR_VERSION; 575 d204->hdr.version = DBFS_D204_HDR_VERSION;
585 d204->hdr.len = PAGE_SIZE * diag204_buf_pages; 576 d204->hdr.len = PAGE_SIZE * diag204_buf_pages;
586 d204->hdr.sc = diag204_store_sc; 577 d204->hdr.sc = diag204_store_sc;
587 data->d204 = d204; 578 *data = d204;
588 file->private_data = data; 579 *data_free_ptr = base;
589 return nonseekable_open(inode, file); 580 *size = d204->hdr.len + sizeof(struct dbfs_d204_hdr);
590
591fail_vfree_base:
592 vfree(data->base);
593fail_kfree_data:
594 kfree(data);
595 return rc;
596}
597
598static int dbfs_d204_release(struct inode *inode, struct file *file)
599{
600 struct dbfs_d204_private *data = file->private_data;
601
602 vfree(data->base);
603 kfree(data);
604 return 0; 581 return 0;
605} 582}
606 583
607static ssize_t dbfs_d204_read(struct file *file, char __user *buf, 584static struct hypfs_dbfs_file dbfs_file_d204 = {
608 size_t size, loff_t *ppos) 585 .name = "diag_204",
609{ 586 .data_create = dbfs_d204_create,
610 struct dbfs_d204_private *data = file->private_data; 587 .data_free = vfree,
611
612 return simple_read_from_buffer(buf, size, ppos, data->d204,
613 data->d204->hdr.len +
614 sizeof(data->d204->hdr));
615}
616
617static const struct file_operations dbfs_d204_ops = {
618 .open = dbfs_d204_open,
619 .read = dbfs_d204_read,
620 .release = dbfs_d204_release,
621}; 588};
622 589
623static int hypfs_dbfs_init(void)
624{
625 dbfs_d204_file = debugfs_create_file("diag_204", 0400, hypfs_dbfs_dir,
626 NULL, &dbfs_d204_ops);
627 if (IS_ERR(dbfs_d204_file))
628 return PTR_ERR(dbfs_d204_file);
629 return 0;
630}
631
632__init int hypfs_diag_init(void) 590__init int hypfs_diag_init(void)
633{ 591{
634 int rc; 592 int rc;
@@ -637,18 +595,21 @@ __init int hypfs_diag_init(void)
637 pr_err("The hardware system does not support hypfs\n"); 595 pr_err("The hardware system does not support hypfs\n");
638 return -ENODATA; 596 return -ENODATA;
639 } 597 }
640 rc = diag224_get_name_table();
641 if (rc) {
642 diag204_free_buffer();
643 pr_err("The hardware system does not provide all "
644 "functions required by hypfs\n");
645 }
646 if (diag204_info_type == INFO_EXT) { 598 if (diag204_info_type == INFO_EXT) {
647 rc = hypfs_dbfs_init(); 599 rc = hypfs_dbfs_create_file(&dbfs_file_d204);
648 if (rc) 600 if (rc)
649 diag204_free_buffer(); 601 return rc;
650 } 602 }
651 return rc; 603 if (MACHINE_IS_LPAR) {
604 rc = diag224_get_name_table();
605 if (rc) {
606 pr_err("The hardware system does not provide all "
607 "functions required by hypfs\n");
608 debugfs_remove(dbfs_d204_file);
609 return rc;
610 }
611 }
612 return 0;
652} 613}
653 614
654void hypfs_diag_exit(void) 615void hypfs_diag_exit(void)
@@ -656,6 +617,7 @@ void hypfs_diag_exit(void)
656 debugfs_remove(dbfs_d204_file); 617 debugfs_remove(dbfs_d204_file);
657 diag224_delete_name_table(); 618 diag224_delete_name_table();
658 diag204_free_buffer(); 619 diag204_free_buffer();
620 hypfs_dbfs_remove_file(&dbfs_file_d204);
659} 621}
660 622
661/* 623/*
diff --git a/arch/s390/hypfs/hypfs_vm.c b/arch/s390/hypfs/hypfs_vm.c
index ee5ab1a578e7..e54796002f61 100644
--- a/arch/s390/hypfs/hypfs_vm.c
+++ b/arch/s390/hypfs/hypfs_vm.c
@@ -20,8 +20,6 @@ static char local_guest[] = " ";
20static char all_guests[] = "* "; 20static char all_guests[] = "* ";
21static char *guest_query; 21static char *guest_query;
22 22
23static struct dentry *dbfs_d2fc_file;
24
25struct diag2fc_data { 23struct diag2fc_data {
26 __u32 version; 24 __u32 version;
27 __u32 flags; 25 __u32 flags;
@@ -104,7 +102,7 @@ static void *diag2fc_store(char *query, unsigned int *count, int offset)
104 return data; 102 return data;
105} 103}
106 104
107static void diag2fc_free(void *data) 105static void diag2fc_free(const void *data)
108{ 106{
109 vfree(data); 107 vfree(data);
110} 108}
@@ -239,42 +237,29 @@ struct dbfs_d2fc {
239 char buf[]; /* d2fc buffer */ 237 char buf[]; /* d2fc buffer */
240} __attribute__ ((packed)); 238} __attribute__ ((packed));
241 239
242static int dbfs_d2fc_open(struct inode *inode, struct file *file) 240static int dbfs_diag2fc_create(void **data, void **data_free_ptr, size_t *size)
243{ 241{
244 struct dbfs_d2fc *data; 242 struct dbfs_d2fc *d2fc;
245 unsigned int count; 243 unsigned int count;
246 244
247 data = diag2fc_store(guest_query, &count, sizeof(data->hdr)); 245 d2fc = diag2fc_store(guest_query, &count, sizeof(d2fc->hdr));
248 if (IS_ERR(data)) 246 if (IS_ERR(d2fc))
249 return PTR_ERR(data); 247 return PTR_ERR(d2fc);
250 get_clock_ext(data->hdr.tod_ext); 248 get_clock_ext(d2fc->hdr.tod_ext);
251 data->hdr.len = count * sizeof(struct diag2fc_data); 249 d2fc->hdr.len = count * sizeof(struct diag2fc_data);
252 data->hdr.version = DBFS_D2FC_HDR_VERSION; 250 d2fc->hdr.version = DBFS_D2FC_HDR_VERSION;
253 data->hdr.count = count; 251 d2fc->hdr.count = count;
254 memset(&data->hdr.reserved, 0, sizeof(data->hdr.reserved)); 252 memset(&d2fc->hdr.reserved, 0, sizeof(d2fc->hdr.reserved));
255 file->private_data = data; 253 *data = d2fc;
256 return nonseekable_open(inode, file); 254 *data_free_ptr = d2fc;
257} 255 *size = d2fc->hdr.len + sizeof(struct dbfs_d2fc_hdr);
258
259static int dbfs_d2fc_release(struct inode *inode, struct file *file)
260{
261 diag2fc_free(file->private_data);
262 return 0; 256 return 0;
263} 257}
264 258
265static ssize_t dbfs_d2fc_read(struct file *file, char __user *buf, 259static struct hypfs_dbfs_file dbfs_file_2fc = {
266 size_t size, loff_t *ppos) 260 .name = "diag_2fc",
267{ 261 .data_create = dbfs_diag2fc_create,
268 struct dbfs_d2fc *data = file->private_data; 262 .data_free = diag2fc_free,
269
270 return simple_read_from_buffer(buf, size, ppos, data, data->hdr.len +
271 sizeof(struct dbfs_d2fc_hdr));
272}
273
274static const struct file_operations dbfs_d2fc_ops = {
275 .open = dbfs_d2fc_open,
276 .read = dbfs_d2fc_read,
277 .release = dbfs_d2fc_release,
278}; 263};
279 264
280int hypfs_vm_init(void) 265int hypfs_vm_init(void)
@@ -287,18 +272,12 @@ int hypfs_vm_init(void)
287 guest_query = local_guest; 272 guest_query = local_guest;
288 else 273 else
289 return -EACCES; 274 return -EACCES;
290 275 return hypfs_dbfs_create_file(&dbfs_file_2fc);
291 dbfs_d2fc_file = debugfs_create_file("diag_2fc", 0400, hypfs_dbfs_dir,
292 NULL, &dbfs_d2fc_ops);
293 if (IS_ERR(dbfs_d2fc_file))
294 return PTR_ERR(dbfs_d2fc_file);
295
296 return 0;
297} 276}
298 277
299void hypfs_vm_exit(void) 278void hypfs_vm_exit(void)
300{ 279{
301 if (!MACHINE_IS_VM) 280 if (!MACHINE_IS_VM)
302 return; 281 return;
303 debugfs_remove(dbfs_d2fc_file); 282 hypfs_dbfs_remove_file(&dbfs_file_2fc);
304} 283}
diff --git a/arch/s390/hypfs/inode.c b/arch/s390/hypfs/inode.c
index 98a4a4c267a7..6fe874fc5f8e 100644
--- a/arch/s390/hypfs/inode.c
+++ b/arch/s390/hypfs/inode.c
@@ -46,8 +46,6 @@ static const struct super_operations hypfs_s_ops;
46/* start of list of all dentries, which have to be deleted on update */ 46/* start of list of all dentries, which have to be deleted on update */
47static struct dentry *hypfs_last_dentry; 47static struct dentry *hypfs_last_dentry;
48 48
49struct dentry *hypfs_dbfs_dir;
50
51static void hypfs_update_update(struct super_block *sb) 49static void hypfs_update_update(struct super_block *sb)
52{ 50{
53 struct hypfs_sb_info *sb_info = sb->s_fs_info; 51 struct hypfs_sb_info *sb_info = sb->s_fs_info;
@@ -316,10 +314,10 @@ static int hypfs_fill_super(struct super_block *sb, void *data, int silent)
316 return 0; 314 return 0;
317} 315}
318 316
319static int hypfs_get_super(struct file_system_type *fst, int flags, 317static struct dentry *hypfs_mount(struct file_system_type *fst, int flags,
320 const char *devname, void *data, struct vfsmount *mnt) 318 const char *devname, void *data)
321{ 319{
322 return get_sb_single(fst, flags, data, hypfs_fill_super, mnt); 320 return mount_single(fst, flags, data, hypfs_fill_super);
323} 321}
324 322
325static void hypfs_kill_super(struct super_block *sb) 323static void hypfs_kill_super(struct super_block *sb)
@@ -449,12 +447,13 @@ static const struct file_operations hypfs_file_ops = {
449 .write = do_sync_write, 447 .write = do_sync_write,
450 .aio_read = hypfs_aio_read, 448 .aio_read = hypfs_aio_read,
451 .aio_write = hypfs_aio_write, 449 .aio_write = hypfs_aio_write,
450 .llseek = no_llseek,
452}; 451};
453 452
454static struct file_system_type hypfs_type = { 453static struct file_system_type hypfs_type = {
455 .owner = THIS_MODULE, 454 .owner = THIS_MODULE,
456 .name = "s390_hypfs", 455 .name = "s390_hypfs",
457 .get_sb = hypfs_get_super, 456 .mount = hypfs_mount,
458 .kill_sb = hypfs_kill_super 457 .kill_sb = hypfs_kill_super
459}; 458};
460 459
@@ -470,13 +469,12 @@ static int __init hypfs_init(void)
470{ 469{
471 int rc; 470 int rc;
472 471
473 hypfs_dbfs_dir = debugfs_create_dir("s390_hypfs", NULL); 472 rc = hypfs_dbfs_init();
474 if (IS_ERR(hypfs_dbfs_dir)) 473 if (rc)
475 return PTR_ERR(hypfs_dbfs_dir); 474 return rc;
476
477 if (hypfs_diag_init()) { 475 if (hypfs_diag_init()) {
478 rc = -ENODATA; 476 rc = -ENODATA;
479 goto fail_debugfs_remove; 477 goto fail_dbfs_exit;
480 } 478 }
481 if (hypfs_vm_init()) { 479 if (hypfs_vm_init()) {
482 rc = -ENODATA; 480 rc = -ENODATA;
@@ -498,9 +496,8 @@ fail_hypfs_vm_exit:
498 hypfs_vm_exit(); 496 hypfs_vm_exit();
499fail_hypfs_diag_exit: 497fail_hypfs_diag_exit:
500 hypfs_diag_exit(); 498 hypfs_diag_exit();
501fail_debugfs_remove: 499fail_dbfs_exit:
502 debugfs_remove(hypfs_dbfs_dir); 500 hypfs_dbfs_exit();
503
504 pr_err("Initialization of hypfs failed with rc=%i\n", rc); 501 pr_err("Initialization of hypfs failed with rc=%i\n", rc);
505 return rc; 502 return rc;
506} 503}
@@ -509,7 +506,7 @@ static void __exit hypfs_exit(void)
509{ 506{
510 hypfs_diag_exit(); 507 hypfs_diag_exit();
511 hypfs_vm_exit(); 508 hypfs_vm_exit();
512 debugfs_remove(hypfs_dbfs_dir); 509 hypfs_dbfs_exit();
513 unregister_filesystem(&hypfs_type); 510 unregister_filesystem(&hypfs_type);
514 kobject_put(s390_kobj); 511 kobject_put(s390_kobj);
515} 512}
diff --git a/arch/s390/include/asm/Kbuild b/arch/s390/include/asm/Kbuild
index 42e512ba8b43..287d7bbb6d36 100644
--- a/arch/s390/include/asm/Kbuild
+++ b/arch/s390/include/asm/Kbuild
@@ -5,6 +5,7 @@ header-y += chsc.h
5header-y += cmb.h 5header-y += cmb.h
6header-y += dasd.h 6header-y += dasd.h
7header-y += debug.h 7header-y += debug.h
8header-y += kvm_virtio.h
8header-y += monwriter.h 9header-y += monwriter.h
9header-y += qeth.h 10header-y += qeth.h
10header-y += schid.h 11header-y += schid.h
diff --git a/arch/s390/include/asm/atomic.h b/arch/s390/include/asm/atomic.h
index 76daea117181..d9db13810d15 100644
--- a/arch/s390/include/asm/atomic.h
+++ b/arch/s390/include/asm/atomic.h
@@ -9,7 +9,7 @@
9 * 9 *
10 * Atomic operations that C can't guarantee us. 10 * Atomic operations that C can't guarantee us.
11 * Useful for resource counting etc. 11 * Useful for resource counting etc.
12 * s390 uses 'Compare And Swap' for atomicity in SMP enviroment. 12 * s390 uses 'Compare And Swap' for atomicity in SMP environment.
13 * 13 *
14 */ 14 */
15 15
@@ -36,14 +36,19 @@
36 36
37static inline int atomic_read(const atomic_t *v) 37static inline int atomic_read(const atomic_t *v)
38{ 38{
39 barrier(); 39 int c;
40 return v->counter; 40
41 asm volatile(
42 " l %0,%1\n"
43 : "=d" (c) : "Q" (v->counter));
44 return c;
41} 45}
42 46
43static inline void atomic_set(atomic_t *v, int i) 47static inline void atomic_set(atomic_t *v, int i)
44{ 48{
45 v->counter = i; 49 asm volatile(
46 barrier(); 50 " st %1,%0\n"
51 : "=Q" (v->counter) : "d" (i));
47} 52}
48 53
49static inline int atomic_add_return(int i, atomic_t *v) 54static inline int atomic_add_return(int i, atomic_t *v)
@@ -128,14 +133,19 @@ static inline int atomic_add_unless(atomic_t *v, int a, int u)
128 133
129static inline long long atomic64_read(const atomic64_t *v) 134static inline long long atomic64_read(const atomic64_t *v)
130{ 135{
131 barrier(); 136 long long c;
132 return v->counter; 137
138 asm volatile(
139 " lg %0,%1\n"
140 : "=d" (c) : "Q" (v->counter));
141 return c;
133} 142}
134 143
135static inline void atomic64_set(atomic64_t *v, long long i) 144static inline void atomic64_set(atomic64_t *v, long long i)
136{ 145{
137 v->counter = i; 146 asm volatile(
138 barrier(); 147 " stg %1,%0\n"
148 : "=Q" (v->counter) : "d" (i));
139} 149}
140 150
141static inline long long atomic64_add_return(long long i, atomic64_t *v) 151static inline long long atomic64_add_return(long long i, atomic64_t *v)
diff --git a/arch/s390/include/asm/bitops.h b/arch/s390/include/asm/bitops.h
index 2e05972c5085..667c6e9f6a34 100644
--- a/arch/s390/include/asm/bitops.h
+++ b/arch/s390/include/asm/bitops.h
@@ -621,6 +621,7 @@ static inline unsigned long find_first_zero_bit(const unsigned long *addr,
621 bits = __ffz_word(bytes*8, __load_ulong_be(addr, bytes)); 621 bits = __ffz_word(bytes*8, __load_ulong_be(addr, bytes));
622 return (bits < size) ? bits : size; 622 return (bits < size) ? bits : size;
623} 623}
624#define find_first_zero_bit find_first_zero_bit
624 625
625/** 626/**
626 * find_first_bit - find the first set bit in a memory region 627 * find_first_bit - find the first set bit in a memory region
@@ -641,6 +642,7 @@ static inline unsigned long find_first_bit(const unsigned long * addr,
641 bits = __ffs_word(bytes*8, __load_ulong_be(addr, bytes)); 642 bits = __ffs_word(bytes*8, __load_ulong_be(addr, bytes));
642 return (bits < size) ? bits : size; 643 return (bits < size) ? bits : size;
643} 644}
645#define find_first_bit find_first_bit
644 646
645/** 647/**
646 * find_next_zero_bit - find the first zero bit in a memory region 648 * find_next_zero_bit - find the first zero bit in a memory region
@@ -677,6 +679,7 @@ static inline int find_next_zero_bit (const unsigned long * addr,
677 } 679 }
678 return offset + find_first_zero_bit(p, size); 680 return offset + find_first_zero_bit(p, size);
679} 681}
682#define find_next_zero_bit find_next_zero_bit
680 683
681/** 684/**
682 * find_next_bit - find the first set bit in a memory region 685 * find_next_bit - find the first set bit in a memory region
@@ -713,6 +716,7 @@ static inline int find_next_bit (const unsigned long * addr,
713 } 716 }
714 return offset + find_first_bit(p, size); 717 return offset + find_first_bit(p, size);
715} 718}
719#define find_next_bit find_next_bit
716 720
717/* 721/*
718 * Every architecture must define this function. It's the fastest 722 * Every architecture must define this function. It's the fastest
@@ -742,18 +746,7 @@ static inline int sched_find_first_bit(unsigned long *b)
742 * 23 22 21 20 19 18 17 16 31 30 29 28 27 26 25 24 746 * 23 22 21 20 19 18 17 16 31 30 29 28 27 26 25 24
743 */ 747 */
744 748
745#define ext2_set_bit(nr, addr) \ 749static inline int find_first_zero_bit_le(void *vaddr, unsigned int size)
746 __test_and_set_bit((nr)^(__BITOPS_WORDSIZE - 8), (unsigned long *)addr)
747#define ext2_set_bit_atomic(lock, nr, addr) \
748 test_and_set_bit((nr)^(__BITOPS_WORDSIZE - 8), (unsigned long *)addr)
749#define ext2_clear_bit(nr, addr) \
750 __test_and_clear_bit((nr)^(__BITOPS_WORDSIZE - 8), (unsigned long *)addr)
751#define ext2_clear_bit_atomic(lock, nr, addr) \
752 test_and_clear_bit((nr)^(__BITOPS_WORDSIZE - 8), (unsigned long *)addr)
753#define ext2_test_bit(nr, addr) \
754 test_bit((nr)^(__BITOPS_WORDSIZE - 8), (unsigned long *)addr)
755
756static inline int ext2_find_first_zero_bit(void *vaddr, unsigned int size)
757{ 750{
758 unsigned long bytes, bits; 751 unsigned long bytes, bits;
759 752
@@ -763,8 +756,9 @@ static inline int ext2_find_first_zero_bit(void *vaddr, unsigned int size)
763 bits = __ffz_word(bytes*8, __load_ulong_le(vaddr, bytes)); 756 bits = __ffz_word(bytes*8, __load_ulong_le(vaddr, bytes));
764 return (bits < size) ? bits : size; 757 return (bits < size) ? bits : size;
765} 758}
759#define find_first_zero_bit_le find_first_zero_bit_le
766 760
767static inline int ext2_find_next_zero_bit(void *vaddr, unsigned long size, 761static inline int find_next_zero_bit_le(void *vaddr, unsigned long size,
768 unsigned long offset) 762 unsigned long offset)
769{ 763{
770 unsigned long *addr = vaddr, *p; 764 unsigned long *addr = vaddr, *p;
@@ -790,11 +784,11 @@ static inline int ext2_find_next_zero_bit(void *vaddr, unsigned long size,
790 size -= __BITOPS_WORDSIZE; 784 size -= __BITOPS_WORDSIZE;
791 p++; 785 p++;
792 } 786 }
793 return offset + ext2_find_first_zero_bit(p, size); 787 return offset + find_first_zero_bit_le(p, size);
794} 788}
789#define find_next_zero_bit_le find_next_zero_bit_le
795 790
796static inline unsigned long ext2_find_first_bit(void *vaddr, 791static inline unsigned long find_first_bit_le(void *vaddr, unsigned long size)
797 unsigned long size)
798{ 792{
799 unsigned long bytes, bits; 793 unsigned long bytes, bits;
800 794
@@ -804,8 +798,9 @@ static inline unsigned long ext2_find_first_bit(void *vaddr,
804 bits = __ffs_word(bytes*8, __load_ulong_le(vaddr, bytes)); 798 bits = __ffs_word(bytes*8, __load_ulong_le(vaddr, bytes));
805 return (bits < size) ? bits : size; 799 return (bits < size) ? bits : size;
806} 800}
801#define find_first_bit_le find_first_bit_le
807 802
808static inline int ext2_find_next_bit(void *vaddr, unsigned long size, 803static inline int find_next_bit_le(void *vaddr, unsigned long size,
809 unsigned long offset) 804 unsigned long offset)
810{ 805{
811 unsigned long *addr = vaddr, *p; 806 unsigned long *addr = vaddr, *p;
@@ -831,10 +826,17 @@ static inline int ext2_find_next_bit(void *vaddr, unsigned long size,
831 size -= __BITOPS_WORDSIZE; 826 size -= __BITOPS_WORDSIZE;
832 p++; 827 p++;
833 } 828 }
834 return offset + ext2_find_first_bit(p, size); 829 return offset + find_first_bit_le(p, size);
835} 830}
831#define find_next_bit_le find_next_bit_le
832
833#include <asm-generic/bitops/le.h>
834
835#define ext2_set_bit_atomic(lock, nr, addr) \
836 test_and_set_bit_le(nr, addr)
837#define ext2_clear_bit_atomic(lock, nr, addr) \
838 test_and_clear_bit_le(nr, addr)
836 839
837#include <asm-generic/bitops/minix.h>
838 840
839#endif /* __KERNEL__ */ 841#endif /* __KERNEL__ */
840 842
diff --git a/arch/s390/include/asm/cache.h b/arch/s390/include/asm/cache.h
index 24aafa68b643..2a30d5ac0667 100644
--- a/arch/s390/include/asm/cache.h
+++ b/arch/s390/include/asm/cache.h
@@ -13,6 +13,7 @@
13 13
14#define L1_CACHE_BYTES 256 14#define L1_CACHE_BYTES 256
15#define L1_CACHE_SHIFT 8 15#define L1_CACHE_SHIFT 8
16#define NET_SKB_PAD 32
16 17
17#define __read_mostly __attribute__((__section__(".data..read_mostly"))) 18#define __read_mostly __attribute__((__section__(".data..read_mostly")))
18 19
diff --git a/arch/s390/include/asm/cacheflush.h b/arch/s390/include/asm/cacheflush.h
index 405cc97c6249..3e20383d0921 100644
--- a/arch/s390/include/asm/cacheflush.h
+++ b/arch/s390/include/asm/cacheflush.h
@@ -1,32 +1,16 @@
1#ifndef _S390_CACHEFLUSH_H 1#ifndef _S390_CACHEFLUSH_H
2#define _S390_CACHEFLUSH_H 2#define _S390_CACHEFLUSH_H
3 3
4/* Keep includes the same across arches. */
5#include <linux/mm.h>
6
7/* Caches aren't brain-dead on the s390. */ 4/* Caches aren't brain-dead on the s390. */
8#define flush_cache_all() do { } while (0) 5#include <asm-generic/cacheflush.h>
9#define flush_cache_mm(mm) do { } while (0)
10#define flush_cache_dup_mm(mm) do { } while (0)
11#define flush_cache_range(vma, start, end) do { } while (0)
12#define flush_cache_page(vma, vmaddr, pfn) do { } while (0)
13#define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE 0
14#define flush_dcache_page(page) do { } while (0)
15#define flush_dcache_mmap_lock(mapping) do { } while (0)
16#define flush_dcache_mmap_unlock(mapping) do { } while (0)
17#define flush_icache_range(start, end) do { } while (0)
18#define flush_icache_page(vma,pg) do { } while (0)
19#define flush_icache_user_range(vma,pg,adr,len) do { } while (0)
20#define flush_cache_vmap(start, end) do { } while (0)
21#define flush_cache_vunmap(start, end) do { } while (0)
22
23#define copy_to_user_page(vma, page, vaddr, dst, src, len) \
24 memcpy(dst, src, len)
25#define copy_from_user_page(vma, page, vaddr, dst, src, len) \
26 memcpy(dst, src, len)
27 6
28#ifdef CONFIG_DEBUG_PAGEALLOC 7#ifdef CONFIG_DEBUG_PAGEALLOC
29void kernel_map_pages(struct page *page, int numpages, int enable); 8void kernel_map_pages(struct page *page, int numpages, int enable);
30#endif 9#endif
31 10
11int set_memory_ro(unsigned long addr, int numpages);
12int set_memory_rw(unsigned long addr, int numpages);
13int set_memory_nx(unsigned long addr, int numpages);
14int set_memory_x(unsigned long addr, int numpages);
15
32#endif /* _S390_CACHEFLUSH_H */ 16#endif /* _S390_CACHEFLUSH_H */
diff --git a/arch/s390/include/asm/ccwdev.h b/arch/s390/include/asm/ccwdev.h
index f3ba0fa98de6..623f2fb71774 100644
--- a/arch/s390/include/asm/ccwdev.h
+++ b/arch/s390/include/asm/ccwdev.h
@@ -92,6 +92,16 @@ struct ccw_device {
92}; 92};
93 93
94/* 94/*
95 * Possible events used by the path_event notifier.
96 */
97#define PE_NONE 0x0
98#define PE_PATH_GONE 0x1 /* A path is no longer available. */
99#define PE_PATH_AVAILABLE 0x2 /* A path has become available and
100 was successfully verified. */
101#define PE_PATHGROUP_ESTABLISHED 0x4 /* A pathgroup was reset and had
102 to be established again. */
103
104/*
95 * Possible CIO actions triggered by the unit check handler. 105 * Possible CIO actions triggered by the unit check handler.
96 */ 106 */
97enum uc_todo { 107enum uc_todo {
@@ -102,13 +112,13 @@ enum uc_todo {
102 112
103/** 113/**
104 * struct ccw driver - device driver for channel attached devices 114 * struct ccw driver - device driver for channel attached devices
105 * @owner: owning module
106 * @ids: ids supported by this driver 115 * @ids: ids supported by this driver
107 * @probe: function called on probe 116 * @probe: function called on probe
108 * @remove: function called on remove 117 * @remove: function called on remove
109 * @set_online: called when setting device online 118 * @set_online: called when setting device online
110 * @set_offline: called when setting device offline 119 * @set_offline: called when setting device offline
111 * @notify: notify driver of device state changes 120 * @notify: notify driver of device state changes
121 * @path_event: notify driver of channel path events
112 * @shutdown: called at device shutdown 122 * @shutdown: called at device shutdown
113 * @prepare: prepare for pm state transition 123 * @prepare: prepare for pm state transition
114 * @complete: undo work done in @prepare 124 * @complete: undo work done in @prepare
@@ -117,16 +127,15 @@ enum uc_todo {
117 * @restore: callback for restoring after hibernation 127 * @restore: callback for restoring after hibernation
118 * @uc_handler: callback for unit check handler 128 * @uc_handler: callback for unit check handler
119 * @driver: embedded device driver structure 129 * @driver: embedded device driver structure
120 * @name: device driver name
121 */ 130 */
122struct ccw_driver { 131struct ccw_driver {
123 struct module *owner;
124 struct ccw_device_id *ids; 132 struct ccw_device_id *ids;
125 int (*probe) (struct ccw_device *); 133 int (*probe) (struct ccw_device *);
126 void (*remove) (struct ccw_device *); 134 void (*remove) (struct ccw_device *);
127 int (*set_online) (struct ccw_device *); 135 int (*set_online) (struct ccw_device *);
128 int (*set_offline) (struct ccw_device *); 136 int (*set_offline) (struct ccw_device *);
129 int (*notify) (struct ccw_device *, int); 137 int (*notify) (struct ccw_device *, int);
138 void (*path_event) (struct ccw_device *, int *);
130 void (*shutdown) (struct ccw_device *); 139 void (*shutdown) (struct ccw_device *);
131 int (*prepare) (struct ccw_device *); 140 int (*prepare) (struct ccw_device *);
132 void (*complete) (struct ccw_device *); 141 void (*complete) (struct ccw_device *);
@@ -135,7 +144,6 @@ struct ccw_driver {
135 int (*restore)(struct ccw_device *); 144 int (*restore)(struct ccw_device *);
136 enum uc_todo (*uc_handler) (struct ccw_device *, struct irb *); 145 enum uc_todo (*uc_handler) (struct ccw_device *, struct irb *);
137 struct device_driver driver; 146 struct device_driver driver;
138 char *name;
139}; 147};
140 148
141extern struct ccw_device *get_ccwdev_by_busid(struct ccw_driver *cdrv, 149extern struct ccw_device *get_ccwdev_by_busid(struct ccw_driver *cdrv,
@@ -192,6 +200,8 @@ int ccw_device_tm_start_timeout(struct ccw_device *, struct tcw *,
192 unsigned long, u8, int); 200 unsigned long, u8, int);
193int ccw_device_tm_intrg(struct ccw_device *cdev); 201int ccw_device_tm_intrg(struct ccw_device *cdev);
194 202
203int ccw_device_get_mdc(struct ccw_device *cdev, u8 mask);
204
195extern int ccw_device_set_online(struct ccw_device *cdev); 205extern int ccw_device_set_online(struct ccw_device *cdev);
196extern int ccw_device_set_offline(struct ccw_device *cdev); 206extern int ccw_device_set_offline(struct ccw_device *cdev);
197 207
diff --git a/arch/s390/include/asm/ccwgroup.h b/arch/s390/include/asm/ccwgroup.h
index c79c1e787b86..f2ea2c56a7e1 100644
--- a/arch/s390/include/asm/ccwgroup.h
+++ b/arch/s390/include/asm/ccwgroup.h
@@ -29,8 +29,6 @@ struct ccwgroup_device {
29 29
30/** 30/**
31 * struct ccwgroup_driver - driver for ccw group devices 31 * struct ccwgroup_driver - driver for ccw group devices
32 * @owner: driver owner
33 * @name: driver name
34 * @max_slaves: maximum number of slave devices 32 * @max_slaves: maximum number of slave devices
35 * @driver_id: unique id 33 * @driver_id: unique id
36 * @probe: function called on probe 34 * @probe: function called on probe
@@ -46,8 +44,6 @@ struct ccwgroup_device {
46 * @driver: embedded driver structure 44 * @driver: embedded driver structure
47 */ 45 */
48struct ccwgroup_driver { 46struct ccwgroup_driver {
49 struct module *owner;
50 char *name;
51 int max_slaves; 47 int max_slaves;
52 unsigned long driver_id; 48 unsigned long driver_id;
53 49
diff --git a/arch/s390/include/asm/cio.h b/arch/s390/include/asm/cio.h
index e34347d567a6..fc50a3342da3 100644
--- a/arch/s390/include/asm/cio.h
+++ b/arch/s390/include/asm/cio.h
@@ -183,7 +183,7 @@ struct esw3 {
183 * The irb that is handed to the device driver when an interrupt occurs. For 183 * The irb that is handed to the device driver when an interrupt occurs. For
184 * solicited interrupts, the common I/O layer already performs checks whether 184 * solicited interrupts, the common I/O layer already performs checks whether
185 * a field is valid; a field not being valid is always passed as %0. 185 * a field is valid; a field not being valid is always passed as %0.
186 * If a unit check occured, @ecw may contain sense data; this is retrieved 186 * If a unit check occurred, @ecw may contain sense data; this is retrieved
187 * by the common I/O layer itself if the device doesn't support concurrent 187 * by the common I/O layer itself if the device doesn't support concurrent
188 * sense (so that the device driver never needs to perform basic sene itself). 188 * sense (so that the device driver never needs to perform basic sene itself).
189 * For unsolicited interrupts, the irb is passed as-is (expect for sense data, 189 * For unsolicited interrupts, the irb is passed as-is (expect for sense data,
diff --git a/arch/s390/include/asm/cmpxchg.h b/arch/s390/include/asm/cmpxchg.h
new file mode 100644
index 000000000000..81d7908416cf
--- /dev/null
+++ b/arch/s390/include/asm/cmpxchg.h
@@ -0,0 +1,224 @@
1/*
2 * Copyright IBM Corp. 1999, 2011
3 *
4 * Author(s): Martin Schwidefsky <schwidefsky@de.ibm.com>,
5 */
6
7#ifndef __ASM_CMPXCHG_H
8#define __ASM_CMPXCHG_H
9
10#include <linux/types.h>
11
12extern void __xchg_called_with_bad_pointer(void);
13
14static inline unsigned long __xchg(unsigned long x, void *ptr, int size)
15{
16 unsigned long addr, old;
17 int shift;
18
19 switch (size) {
20 case 1:
21 addr = (unsigned long) ptr;
22 shift = (3 ^ (addr & 3)) << 3;
23 addr ^= addr & 3;
24 asm volatile(
25 " l %0,%4\n"
26 "0: lr 0,%0\n"
27 " nr 0,%3\n"
28 " or 0,%2\n"
29 " cs %0,0,%4\n"
30 " jl 0b\n"
31 : "=&d" (old), "=Q" (*(int *) addr)
32 : "d" (x << shift), "d" (~(255 << shift)),
33 "Q" (*(int *) addr) : "memory", "cc", "0");
34 return old >> shift;
35 case 2:
36 addr = (unsigned long) ptr;
37 shift = (2 ^ (addr & 2)) << 3;
38 addr ^= addr & 2;
39 asm volatile(
40 " l %0,%4\n"
41 "0: lr 0,%0\n"
42 " nr 0,%3\n"
43 " or 0,%2\n"
44 " cs %0,0,%4\n"
45 " jl 0b\n"
46 : "=&d" (old), "=Q" (*(int *) addr)
47 : "d" (x << shift), "d" (~(65535 << shift)),
48 "Q" (*(int *) addr) : "memory", "cc", "0");
49 return old >> shift;
50 case 4:
51 asm volatile(
52 " l %0,%3\n"
53 "0: cs %0,%2,%3\n"
54 " jl 0b\n"
55 : "=&d" (old), "=Q" (*(int *) ptr)
56 : "d" (x), "Q" (*(int *) ptr)
57 : "memory", "cc");
58 return old;
59#ifdef CONFIG_64BIT
60 case 8:
61 asm volatile(
62 " lg %0,%3\n"
63 "0: csg %0,%2,%3\n"
64 " jl 0b\n"
65 : "=&d" (old), "=m" (*(long *) ptr)
66 : "d" (x), "Q" (*(long *) ptr)
67 : "memory", "cc");
68 return old;
69#endif /* CONFIG_64BIT */
70 }
71 __xchg_called_with_bad_pointer();
72 return x;
73}
74
75#define xchg(ptr, x) \
76({ \
77 __typeof__(*(ptr)) __ret; \
78 __ret = (__typeof__(*(ptr))) \
79 __xchg((unsigned long)(x), (void *)(ptr), sizeof(*(ptr)));\
80 __ret; \
81})
82
83/*
84 * Atomic compare and exchange. Compare OLD with MEM, if identical,
85 * store NEW in MEM. Return the initial value in MEM. Success is
86 * indicated by comparing RETURN with OLD.
87 */
88
89#define __HAVE_ARCH_CMPXCHG
90
91extern void __cmpxchg_called_with_bad_pointer(void);
92
93static inline unsigned long __cmpxchg(void *ptr, unsigned long old,
94 unsigned long new, int size)
95{
96 unsigned long addr, prev, tmp;
97 int shift;
98
99 switch (size) {
100 case 1:
101 addr = (unsigned long) ptr;
102 shift = (3 ^ (addr & 3)) << 3;
103 addr ^= addr & 3;
104 asm volatile(
105 " l %0,%2\n"
106 "0: nr %0,%5\n"
107 " lr %1,%0\n"
108 " or %0,%3\n"
109 " or %1,%4\n"
110 " cs %0,%1,%2\n"
111 " jnl 1f\n"
112 " xr %1,%0\n"
113 " nr %1,%5\n"
114 " jnz 0b\n"
115 "1:"
116 : "=&d" (prev), "=&d" (tmp), "=Q" (*(int *) ptr)
117 : "d" (old << shift), "d" (new << shift),
118 "d" (~(255 << shift)), "Q" (*(int *) ptr)
119 : "memory", "cc");
120 return prev >> shift;
121 case 2:
122 addr = (unsigned long) ptr;
123 shift = (2 ^ (addr & 2)) << 3;
124 addr ^= addr & 2;
125 asm volatile(
126 " l %0,%2\n"
127 "0: nr %0,%5\n"
128 " lr %1,%0\n"
129 " or %0,%3\n"
130 " or %1,%4\n"
131 " cs %0,%1,%2\n"
132 " jnl 1f\n"
133 " xr %1,%0\n"
134 " nr %1,%5\n"
135 " jnz 0b\n"
136 "1:"
137 : "=&d" (prev), "=&d" (tmp), "=Q" (*(int *) ptr)
138 : "d" (old << shift), "d" (new << shift),
139 "d" (~(65535 << shift)), "Q" (*(int *) ptr)
140 : "memory", "cc");
141 return prev >> shift;
142 case 4:
143 asm volatile(
144 " cs %0,%3,%1\n"
145 : "=&d" (prev), "=Q" (*(int *) ptr)
146 : "0" (old), "d" (new), "Q" (*(int *) ptr)
147 : "memory", "cc");
148 return prev;
149#ifdef CONFIG_64BIT
150 case 8:
151 asm volatile(
152 " csg %0,%3,%1\n"
153 : "=&d" (prev), "=Q" (*(long *) ptr)
154 : "0" (old), "d" (new), "Q" (*(long *) ptr)
155 : "memory", "cc");
156 return prev;
157#endif /* CONFIG_64BIT */
158 }
159 __cmpxchg_called_with_bad_pointer();
160 return old;
161}
162
163#define cmpxchg(ptr, o, n) \
164 ((__typeof__(*(ptr)))__cmpxchg((ptr), (unsigned long)(o), \
165 (unsigned long)(n), sizeof(*(ptr))))
166
167#ifdef CONFIG_64BIT
168#define cmpxchg64(ptr, o, n) \
169({ \
170 cmpxchg((ptr), (o), (n)); \
171})
172#else /* CONFIG_64BIT */
173static inline unsigned long long __cmpxchg64(void *ptr,
174 unsigned long long old,
175 unsigned long long new)
176{
177 register_pair rp_old = {.pair = old};
178 register_pair rp_new = {.pair = new};
179
180 asm volatile(
181 " cds %0,%2,%1"
182 : "+&d" (rp_old), "=Q" (ptr)
183 : "d" (rp_new), "Q" (ptr)
184 : "cc");
185 return rp_old.pair;
186}
187#define cmpxchg64(ptr, o, n) \
188 ((__typeof__(*(ptr)))__cmpxchg64((ptr), \
189 (unsigned long long)(o), \
190 (unsigned long long)(n)))
191#endif /* CONFIG_64BIT */
192
193#include <asm-generic/cmpxchg-local.h>
194
195static inline unsigned long __cmpxchg_local(void *ptr,
196 unsigned long old,
197 unsigned long new, int size)
198{
199 switch (size) {
200 case 1:
201 case 2:
202 case 4:
203#ifdef CONFIG_64BIT
204 case 8:
205#endif
206 return __cmpxchg(ptr, old, new, size);
207 default:
208 return __cmpxchg_local_generic(ptr, old, new, size);
209 }
210
211 return old;
212}
213
214/*
215 * cmpxchg_local and cmpxchg64_local are atomic wrt current CPU. Always make
216 * them available.
217 */
218#define cmpxchg_local(ptr, o, n) \
219 ((__typeof__(*(ptr)))__cmpxchg_local((ptr), (unsigned long)(o), \
220 (unsigned long)(n), sizeof(*(ptr))))
221
222#define cmpxchg64_local(ptr, o, n) cmpxchg64((ptr), (o), (n))
223
224#endif /* __ASM_CMPXCHG_H */
diff --git a/arch/s390/include/asm/compat.h b/arch/s390/include/asm/compat.h
index a875c2f542e1..da359ca6fe55 100644
--- a/arch/s390/include/asm/compat.h
+++ b/arch/s390/include/asm/compat.h
@@ -169,7 +169,7 @@ static inline compat_uptr_t ptr_to_compat(void __user *uptr)
169 169
170static inline int is_compat_task(void) 170static inline int is_compat_task(void)
171{ 171{
172 return test_thread_flag(TIF_31BIT); 172 return is_32bit_task();
173} 173}
174 174
175#else 175#else
diff --git a/arch/s390/include/asm/cpu.h b/arch/s390/include/asm/cpu.h
index 471234b90574..e0b69540216f 100644
--- a/arch/s390/include/asm/cpu.h
+++ b/arch/s390/include/asm/cpu.h
@@ -20,7 +20,7 @@ struct cpuid
20 unsigned int ident : 24; 20 unsigned int ident : 24;
21 unsigned int machine : 16; 21 unsigned int machine : 16;
22 unsigned int unused : 16; 22 unsigned int unused : 16;
23} __packed; 23} __attribute__ ((packed, aligned(8)));
24 24
25#endif /* __ASSEMBLY__ */ 25#endif /* __ASSEMBLY__ */
26#endif /* _ASM_S390_CPU_H */ 26#endif /* _ASM_S390_CPU_H */
diff --git a/arch/s390/include/asm/cputime.h b/arch/s390/include/asm/cputime.h
index 8b1a52a137c5..081434878296 100644
--- a/arch/s390/include/asm/cputime.h
+++ b/arch/s390/include/asm/cputime.h
@@ -73,18 +73,18 @@ cputime64_to_jiffies64(cputime64_t cputime)
73} 73}
74 74
75/* 75/*
76 * Convert cputime to milliseconds and back. 76 * Convert cputime to microseconds and back.
77 */ 77 */
78static inline unsigned int 78static inline unsigned int
79cputime_to_msecs(const cputime_t cputime) 79cputime_to_usecs(const cputime_t cputime)
80{ 80{
81 return cputime_div(cputime, 4096000); 81 return cputime_div(cputime, 4096);
82} 82}
83 83
84static inline cputime_t 84static inline cputime_t
85msecs_to_cputime(const unsigned int m) 85usecs_to_cputime(const unsigned int m)
86{ 86{
87 return (cputime_t) m * 4096000; 87 return (cputime_t) m * 4096;
88} 88}
89 89
90/* 90/*
@@ -202,7 +202,7 @@ static inline void s390_idle_check(struct pt_regs *regs, __u64 int_clock,
202 202
203static inline int s390_nohz_delay(int cpu) 203static inline int s390_nohz_delay(int cpu)
204{ 204{
205 return per_cpu(s390_idle, cpu).nohz_delay != 0; 205 return __get_cpu_var(s390_idle).nohz_delay != 0;
206} 206}
207 207
208#define arch_needs_cpu(cpu) s390_nohz_delay(cpu) 208#define arch_needs_cpu(cpu) s390_nohz_delay(cpu)
diff --git a/arch/s390/include/asm/dasd.h b/arch/s390/include/asm/dasd.h
index 218bce81ec70..0be28efe5b66 100644
--- a/arch/s390/include/asm/dasd.h
+++ b/arch/s390/include/asm/dasd.h
@@ -73,6 +73,7 @@ typedef struct dasd_information2_t {
73 * 0x02: use diag discipline (diag) 73 * 0x02: use diag discipline (diag)
74 * 0x04: set the device initially online (internal use only) 74 * 0x04: set the device initially online (internal use only)
75 * 0x08: enable ERP related logging 75 * 0x08: enable ERP related logging
76 * 0x20: give access to raw eckd data
76 */ 77 */
77#define DASD_FEATURE_DEFAULT 0x00 78#define DASD_FEATURE_DEFAULT 0x00
78#define DASD_FEATURE_READONLY 0x01 79#define DASD_FEATURE_READONLY 0x01
@@ -80,6 +81,8 @@ typedef struct dasd_information2_t {
80#define DASD_FEATURE_INITIAL_ONLINE 0x04 81#define DASD_FEATURE_INITIAL_ONLINE 0x04
81#define DASD_FEATURE_ERPLOG 0x08 82#define DASD_FEATURE_ERPLOG 0x08
82#define DASD_FEATURE_FAILFAST 0x10 83#define DASD_FEATURE_FAILFAST 0x10
84#define DASD_FEATURE_FAILONSLCK 0x20
85#define DASD_FEATURE_USERAW 0x40
83 86
84#define DASD_PARTN_BITS 2 87#define DASD_PARTN_BITS 2
85 88
@@ -217,6 +220,25 @@ typedef struct dasd_symmio_parms {
217 int rssd_result_len; 220 int rssd_result_len;
218} __attribute__ ((packed)) dasd_symmio_parms_t; 221} __attribute__ ((packed)) dasd_symmio_parms_t;
219 222
223/*
224 * Data returned by Sense Path Group ID (SNID)
225 */
226struct dasd_snid_data {
227 struct {
228 __u8 group:2;
229 __u8 reserve:2;
230 __u8 mode:1;
231 __u8 res:3;
232 } __attribute__ ((packed)) path_state;
233 __u8 pgid[11];
234} __attribute__ ((packed));
235
236struct dasd_snid_ioctl_data {
237 struct dasd_snid_data data;
238 __u8 path_mask;
239} __attribute__ ((packed));
240
241
220/******************************************************************************** 242/********************************************************************************
221 * SECTION: Definition of IOCTLs 243 * SECTION: Definition of IOCTLs
222 * 244 *
@@ -261,25 +283,10 @@ typedef struct dasd_symmio_parms {
261/* Set Attributes (cache operations) */ 283/* Set Attributes (cache operations) */
262#define BIODASDSATTR _IOW(DASD_IOCTL_LETTER,2,attrib_data_t) 284#define BIODASDSATTR _IOW(DASD_IOCTL_LETTER,2,attrib_data_t)
263 285
286/* Get Sense Path Group ID (SNID) data */
287#define BIODASDSNID _IOWR(DASD_IOCTL_LETTER, 1, struct dasd_snid_ioctl_data)
288
264#define BIODASDSYMMIO _IOWR(DASD_IOCTL_LETTER, 240, dasd_symmio_parms_t) 289#define BIODASDSYMMIO _IOWR(DASD_IOCTL_LETTER, 240, dasd_symmio_parms_t)
265 290
266#endif /* DASD_H */ 291#endif /* DASD_H */
267 292
268/*
269 * Overrides for Emacs so that we follow Linus's tabbing style.
270 * Emacs will notice this stuff at the end of the file and automatically
271 * adjust the settings for this buffer only. This must remain at the end
272 * of the file.
273 * ---------------------------------------------------------------------------
274 * Local variables:
275 * c-indent-level: 4
276 * c-brace-imaginary-offset: 0
277 * c-brace-offset: -4
278 * c-argdecl-indent: 4
279 * c-label-offset: -4
280 * c-continued-statement-offset: 4
281 * c-continued-brace-offset: 0
282 * indent-tabs-mode: nil
283 * tab-width: 8
284 * End:
285 */
diff --git a/arch/s390/include/asm/delay.h b/arch/s390/include/asm/delay.h
index 8a096b83f51f..0e3b35f96be1 100644
--- a/arch/s390/include/asm/delay.h
+++ b/arch/s390/include/asm/delay.h
@@ -14,10 +14,12 @@
14#ifndef _S390_DELAY_H 14#ifndef _S390_DELAY_H
15#define _S390_DELAY_H 15#define _S390_DELAY_H
16 16
17extern void __udelay(unsigned long long usecs); 17void __ndelay(unsigned long long nsecs);
18extern void udelay_simple(unsigned long long usecs); 18void __udelay(unsigned long long usecs);
19extern void __delay(unsigned long loops); 19void udelay_simple(unsigned long long usecs);
20void __delay(unsigned long loops);
20 21
22#define ndelay(n) __ndelay((unsigned long long) (n))
21#define udelay(n) __udelay((unsigned long long) (n)) 23#define udelay(n) __udelay((unsigned long long) (n))
22#define mdelay(n) __udelay((unsigned long long) (n) * 1000) 24#define mdelay(n) __udelay((unsigned long long) (n) * 1000)
23 25
diff --git a/arch/s390/include/asm/diag.h b/arch/s390/include/asm/diag.h
index 72b2e2f2d32d..7e91c58072e2 100644
--- a/arch/s390/include/asm/diag.h
+++ b/arch/s390/include/asm/diag.h
@@ -9,9 +9,22 @@
9#define _ASM_S390_DIAG_H 9#define _ASM_S390_DIAG_H
10 10
11/* 11/*
12 * Diagnose 10: Release pages 12 * Diagnose 10: Release page range
13 */ 13 */
14extern void diag10(unsigned long addr); 14static inline void diag10_range(unsigned long start_pfn, unsigned long num_pfn)
15{
16 unsigned long start_addr, end_addr;
17
18 start_addr = start_pfn << PAGE_SHIFT;
19 end_addr = (start_pfn + num_pfn - 1) << PAGE_SHIFT;
20
21 asm volatile(
22 "0: diag %0,%1,0x10\n"
23 "1:\n"
24 EX_TABLE(0b, 1b)
25 EX_TABLE(1b, 1b)
26 : : "a" (start_addr), "a" (end_addr));
27}
15 28
16/* 29/*
17 * Diagnose 14: Input spool file manipulation 30 * Diagnose 14: Input spool file manipulation
diff --git a/arch/s390/include/asm/elf.h b/arch/s390/include/asm/elf.h
index 354d42616c7e..64b61bf72e93 100644
--- a/arch/s390/include/asm/elf.h
+++ b/arch/s390/include/asm/elf.h
@@ -161,7 +161,9 @@ extern unsigned int vdso_enabled;
161 use of this is to invoke "./ld.so someprog" to test out a new version of 161 use of this is to invoke "./ld.so someprog" to test out a new version of
162 the loader. We need to make sure that it is out of the way of the program 162 the loader. We need to make sure that it is out of the way of the program
163 that it will "exec", and that there is sufficient room for the brk. */ 163 that it will "exec", and that there is sufficient room for the brk. */
164#define ELF_ET_DYN_BASE (STACK_TOP / 3 * 2) 164
165extern unsigned long randomize_et_dyn(unsigned long base);
166#define ELF_ET_DYN_BASE (randomize_et_dyn(STACK_TOP / 3 * 2))
165 167
166/* This yields a mask that user programs can use to figure out what 168/* This yields a mask that user programs can use to figure out what
167 instruction set this CPU supports. */ 169 instruction set this CPU supports. */
@@ -194,17 +196,7 @@ do { \
194} while (0) 196} while (0)
195#endif /* __s390x__ */ 197#endif /* __s390x__ */
196 198
197/* 199#define STACK_RND_MASK 0x7ffUL
198 * An executable for which elf_read_implies_exec() returns TRUE will
199 * have the READ_IMPLIES_EXEC personality flag set automatically.
200 */
201#define elf_read_implies_exec(ex, executable_stack) \
202({ \
203 if (current->mm->context.noexec && \
204 executable_stack != EXSTACK_DISABLE_X) \
205 disable_noexec(current->mm, current); \
206 current->mm->context.noexec == 0; \
207})
208 200
209#define ARCH_DLINFO \ 201#define ARCH_DLINFO \
210do { \ 202do { \
@@ -218,4 +210,7 @@ struct linux_binprm;
218#define ARCH_HAS_SETUP_ADDITIONAL_PAGES 1 210#define ARCH_HAS_SETUP_ADDITIONAL_PAGES 1
219int arch_setup_additional_pages(struct linux_binprm *, int); 211int arch_setup_additional_pages(struct linux_binprm *, int);
220 212
213extern unsigned long arch_randomize_brk(struct mm_struct *mm);
214#define arch_randomize_brk arch_randomize_brk
215
221#endif 216#endif
diff --git a/arch/s390/include/asm/ftrace.h b/arch/s390/include/asm/ftrace.h
index 96c14a9102b8..b7931faaef6d 100644
--- a/arch/s390/include/asm/ftrace.h
+++ b/arch/s390/include/asm/ftrace.h
@@ -4,25 +4,20 @@
4#ifndef __ASSEMBLY__ 4#ifndef __ASSEMBLY__
5 5
6extern void _mcount(void); 6extern void _mcount(void);
7extern unsigned long ftrace_dyn_func;
8 7
9struct dyn_arch_ftrace { }; 8struct dyn_arch_ftrace { };
10 9
11#define MCOUNT_ADDR ((long)_mcount) 10#define MCOUNT_ADDR ((long)_mcount)
12 11
13#ifdef CONFIG_64BIT 12#ifdef CONFIG_64BIT
14#define MCOUNT_OFFSET_RET 18 13#define MCOUNT_INSN_SIZE 12
15#define MCOUNT_INSN_SIZE 24
16#define MCOUNT_OFFSET 14
17#else 14#else
18#define MCOUNT_OFFSET_RET 26 15#define MCOUNT_INSN_SIZE 20
19#define MCOUNT_INSN_SIZE 30
20#define MCOUNT_OFFSET 8
21#endif 16#endif
22 17
23static inline unsigned long ftrace_call_adjust(unsigned long addr) 18static inline unsigned long ftrace_call_adjust(unsigned long addr)
24{ 19{
25 return addr - MCOUNT_OFFSET; 20 return addr;
26} 21}
27 22
28#endif /* __ASSEMBLY__ */ 23#endif /* __ASSEMBLY__ */
diff --git a/arch/s390/include/asm/futex.h b/arch/s390/include/asm/futex.h
index 5c5d02de49e9..81cf36b691f1 100644
--- a/arch/s390/include/asm/futex.h
+++ b/arch/s390/include/asm/futex.h
@@ -7,7 +7,7 @@
7#include <linux/uaccess.h> 7#include <linux/uaccess.h>
8#include <asm/errno.h> 8#include <asm/errno.h>
9 9
10static inline int futex_atomic_op_inuser (int encoded_op, int __user *uaddr) 10static inline int futex_atomic_op_inuser (int encoded_op, u32 __user *uaddr)
11{ 11{
12 int op = (encoded_op >> 28) & 7; 12 int op = (encoded_op >> 28) & 7;
13 int cmp = (encoded_op >> 24) & 15; 13 int cmp = (encoded_op >> 24) & 15;
@@ -18,7 +18,7 @@ static inline int futex_atomic_op_inuser (int encoded_op, int __user *uaddr)
18 if (encoded_op & (FUTEX_OP_OPARG_SHIFT << 28)) 18 if (encoded_op & (FUTEX_OP_OPARG_SHIFT << 28))
19 oparg = 1 << oparg; 19 oparg = 1 << oparg;
20 20
21 if (! access_ok (VERIFY_WRITE, uaddr, sizeof(int))) 21 if (! access_ok (VERIFY_WRITE, uaddr, sizeof(u32)))
22 return -EFAULT; 22 return -EFAULT;
23 23
24 pagefault_disable(); 24 pagefault_disable();
@@ -39,13 +39,13 @@ static inline int futex_atomic_op_inuser (int encoded_op, int __user *uaddr)
39 return ret; 39 return ret;
40} 40}
41 41
42static inline int futex_atomic_cmpxchg_inatomic(int __user *uaddr, 42static inline int futex_atomic_cmpxchg_inatomic(u32 *uval, u32 __user *uaddr,
43 int oldval, int newval) 43 u32 oldval, u32 newval)
44{ 44{
45 if (! access_ok (VERIFY_WRITE, uaddr, sizeof(int))) 45 if (! access_ok (VERIFY_WRITE, uaddr, sizeof(u32)))
46 return -EFAULT; 46 return -EFAULT;
47 47
48 return uaccess.futex_atomic_cmpxchg(uaddr, oldval, newval); 48 return uaccess.futex_atomic_cmpxchg(uval, uaddr, oldval, newval);
49} 49}
50 50
51#endif /* __KERNEL__ */ 51#endif /* __KERNEL__ */
diff --git a/arch/s390/include/asm/hardirq.h b/arch/s390/include/asm/hardirq.h
index 498bc3892385..e4155d3eb2cb 100644
--- a/arch/s390/include/asm/hardirq.h
+++ b/arch/s390/include/asm/hardirq.h
@@ -12,10 +12,6 @@
12#ifndef __ASM_HARDIRQ_H 12#ifndef __ASM_HARDIRQ_H
13#define __ASM_HARDIRQ_H 13#define __ASM_HARDIRQ_H
14 14
15#include <linux/threads.h>
16#include <linux/sched.h>
17#include <linux/cache.h>
18#include <linux/interrupt.h>
19#include <asm/lowcore.h> 15#include <asm/lowcore.h>
20 16
21#define local_softirq_pending() (S390_lowcore.softirq_pending) 17#define local_softirq_pending() (S390_lowcore.softirq_pending)
@@ -25,20 +21,4 @@
25 21
26#define HARDIRQ_BITS 8 22#define HARDIRQ_BITS 8
27 23
28void clock_comparator_work(void);
29
30static inline unsigned long long local_tick_disable(void)
31{
32 unsigned long long old;
33
34 old = S390_lowcore.clock_comparator;
35 S390_lowcore.clock_comparator = -1ULL;
36 return old;
37}
38
39static inline void local_tick_enable(unsigned long long comp)
40{
41 S390_lowcore.clock_comparator = comp;
42}
43
44#endif /* __ASM_HARDIRQ_H */ 24#endif /* __ASM_HARDIRQ_H */
diff --git a/arch/s390/include/asm/hugetlb.h b/arch/s390/include/asm/hugetlb.h
index bb8343d157bc..799ed0f1643d 100644
--- a/arch/s390/include/asm/hugetlb.h
+++ b/arch/s390/include/asm/hugetlb.h
@@ -37,32 +37,6 @@ static inline int prepare_hugepage_range(struct file *file,
37int arch_prepare_hugepage(struct page *page); 37int arch_prepare_hugepage(struct page *page);
38void arch_release_hugepage(struct page *page); 38void arch_release_hugepage(struct page *page);
39 39
40static inline pte_t pte_mkhuge(pte_t pte)
41{
42 /*
43 * PROT_NONE needs to be remapped from the pte type to the ste type.
44 * The HW invalid bit is also different for pte and ste. The pte
45 * invalid bit happens to be the same as the ste _SEGMENT_ENTRY_LARGE
46 * bit, so we don't have to clear it.
47 */
48 if (pte_val(pte) & _PAGE_INVALID) {
49 if (pte_val(pte) & _PAGE_SWT)
50 pte_val(pte) |= _HPAGE_TYPE_NONE;
51 pte_val(pte) |= _SEGMENT_ENTRY_INV;
52 }
53 /*
54 * Clear SW pte bits SWT and SWX, there are no SW bits in a segment
55 * table entry.
56 */
57 pte_val(pte) &= ~(_PAGE_SWT | _PAGE_SWX);
58 /*
59 * Also set the change-override bit because we don't need dirty bit
60 * tracking for hugetlbfs pages.
61 */
62 pte_val(pte) |= (_SEGMENT_ENTRY_LARGE | _SEGMENT_ENTRY_CO);
63 return pte;
64}
65
66static inline pte_t huge_pte_wrprotect(pte_t pte) 40static inline pte_t huge_pte_wrprotect(pte_t pte)
67{ 41{
68 pte_val(pte) |= _PAGE_RO; 42 pte_val(pte) |= _PAGE_RO;
@@ -137,21 +111,10 @@ static inline void huge_ptep_invalidate(struct mm_struct *mm,
137{ 111{
138 pmd_t *pmdp = (pmd_t *) ptep; 112 pmd_t *pmdp = (pmd_t *) ptep;
139 113
140 if (!MACHINE_HAS_IDTE) { 114 if (MACHINE_HAS_IDTE)
141 __pmd_csp(pmdp);
142 if (mm->context.noexec) {
143 pmdp = get_shadow_table(pmdp);
144 __pmd_csp(pmdp);
145 }
146 return;
147 }
148
149 __pmd_idte(address, pmdp);
150 if (mm->context.noexec) {
151 pmdp = get_shadow_table(pmdp);
152 __pmd_idte(address, pmdp); 115 __pmd_idte(address, pmdp);
153 } 116 else
154 return; 117 __pmd_csp(pmdp);
155} 118}
156 119
157#define huge_ptep_set_access_flags(__vma, __addr, __ptep, __entry, __dirty) \ 120#define huge_ptep_set_access_flags(__vma, __addr, __ptep, __entry, __dirty) \
diff --git a/arch/s390/include/asm/ioctls.h b/arch/s390/include/asm/ioctls.h
index 2f3d8736361f..960a4c1ebdf1 100644
--- a/arch/s390/include/asm/ioctls.h
+++ b/arch/s390/include/asm/ioctls.h
@@ -1,94 +1,8 @@
1/*
2 * include/asm-s390/ioctls.h
3 *
4 * S390 version
5 *
6 * Derived from "include/asm-i386/ioctls.h"
7 */
8
9#ifndef __ARCH_S390_IOCTLS_H__ 1#ifndef __ARCH_S390_IOCTLS_H__
10#define __ARCH_S390_IOCTLS_H__ 2#define __ARCH_S390_IOCTLS_H__
11 3
12#include <asm/ioctl.h>
13
14/* 0x54 is just a magic number to make these relatively unique ('T') */
15
16#define TCGETS 0x5401
17#define TCSETS 0x5402
18#define TCSETSW 0x5403
19#define TCSETSF 0x5404
20#define TCGETA 0x5405
21#define TCSETA 0x5406
22#define TCSETAW 0x5407
23#define TCSETAF 0x5408
24#define TCSBRK 0x5409
25#define TCXONC 0x540A
26#define TCFLSH 0x540B
27#define TIOCEXCL 0x540C
28#define TIOCNXCL 0x540D
29#define TIOCSCTTY 0x540E
30#define TIOCGPGRP 0x540F
31#define TIOCSPGRP 0x5410
32#define TIOCOUTQ 0x5411
33#define TIOCSTI 0x5412
34#define TIOCGWINSZ 0x5413
35#define TIOCSWINSZ 0x5414
36#define TIOCMGET 0x5415
37#define TIOCMBIS 0x5416
38#define TIOCMBIC 0x5417
39#define TIOCMSET 0x5418
40#define TIOCGSOFTCAR 0x5419
41#define TIOCSSOFTCAR 0x541A
42#define FIONREAD 0x541B
43#define TIOCINQ FIONREAD
44#define TIOCLINUX 0x541C
45#define TIOCCONS 0x541D
46#define TIOCGSERIAL 0x541E
47#define TIOCSSERIAL 0x541F
48#define TIOCPKT 0x5420
49#define FIONBIO 0x5421
50#define TIOCNOTTY 0x5422
51#define TIOCSETD 0x5423
52#define TIOCGETD 0x5424
53#define TCSBRKP 0x5425 /* Needed for POSIX tcsendbreak() */
54#define TIOCSBRK 0x5427 /* BSD compatibility */
55#define TIOCCBRK 0x5428 /* BSD compatibility */
56#define TIOCGSID 0x5429 /* Return the session ID of FD */
57#define TCGETS2 _IOR('T',0x2A, struct termios2)
58#define TCSETS2 _IOW('T',0x2B, struct termios2)
59#define TCSETSW2 _IOW('T',0x2C, struct termios2)
60#define TCSETSF2 _IOW('T',0x2D, struct termios2)
61#define TIOCGPTN _IOR('T',0x30, unsigned int) /* Get Pty Number (of pty-mux device) */
62#define TIOCSPTLCK _IOW('T',0x31, int) /* Lock/unlock Pty */
63#define TIOCSIG _IOW('T',0x36, int) /* Generate signal on Pty slave */
64
65#define FIONCLEX 0x5450 /* these numbers need to be adjusted. */
66#define FIOCLEX 0x5451
67#define FIOASYNC 0x5452
68#define TIOCSERCONFIG 0x5453
69#define TIOCSERGWILD 0x5454
70#define TIOCSERSWILD 0x5455
71#define TIOCGLCKTRMIOS 0x5456
72#define TIOCSLCKTRMIOS 0x5457
73#define TIOCSERGSTRUCT 0x5458 /* For debugging only */
74#define TIOCSERGETLSR 0x5459 /* Get line status register */
75#define TIOCSERGETMULTI 0x545A /* Get multiport config */
76#define TIOCSERSETMULTI 0x545B /* Set multiport config */
77
78#define TIOCMIWAIT 0x545C /* wait for a change on serial input line(s) */
79#define TIOCGICOUNT 0x545D /* read serial port inline interrupt counts */
80#define FIOQSIZE 0x545E 4#define FIOQSIZE 0x545E
81 5
82/* Used for packet mode */ 6#include <asm-generic/ioctls.h>
83#define TIOCPKT_DATA 0
84#define TIOCPKT_FLUSHREAD 1
85#define TIOCPKT_FLUSHWRITE 2
86#define TIOCPKT_STOP 4
87#define TIOCPKT_START 8
88#define TIOCPKT_NOSTOP 16
89#define TIOCPKT_DOSTOP 32
90#define TIOCPKT_IOCTL 64
91
92#define TIOCSER_TEMT 0x01 /* Transmitter physically empty */
93 7
94#endif 8#endif
diff --git a/arch/s390/include/asm/irq.h b/arch/s390/include/asm/irq.h
index 7da991a858f8..ba7b01c726a3 100644
--- a/arch/s390/include/asm/irq.h
+++ b/arch/s390/include/asm/irq.h
@@ -1,23 +1,42 @@
1#ifndef _ASM_IRQ_H 1#ifndef _ASM_IRQ_H
2#define _ASM_IRQ_H 2#define _ASM_IRQ_H
3 3
4#ifdef __KERNEL__
5#include <linux/hardirq.h> 4#include <linux/hardirq.h>
6 5#include <linux/types.h>
7/*
8 * the definition of irqs has changed in 2.5.46:
9 * NR_IRQS is no longer the number of i/o
10 * interrupts (65536), but rather the number
11 * of interrupt classes (2).
12 * Only external and i/o interrupts make much sense here (CH).
13 */
14 6
15enum interruption_class { 7enum interruption_class {
16 EXTERNAL_INTERRUPT, 8 EXTERNAL_INTERRUPT,
17 IO_INTERRUPT, 9 IO_INTERRUPT,
18 10 EXTINT_CLK,
11 EXTINT_IPI,
12 EXTINT_TMR,
13 EXTINT_TLA,
14 EXTINT_PFL,
15 EXTINT_DSD,
16 EXTINT_VRT,
17 EXTINT_SCP,
18 EXTINT_IUC,
19 EXTINT_CPM,
20 IOINT_QAI,
21 IOINT_QDI,
22 IOINT_DAS,
23 IOINT_C15,
24 IOINT_C70,
25 IOINT_TAP,
26 IOINT_VMR,
27 IOINT_LCS,
28 IOINT_CLW,
29 IOINT_CTC,
30 IOINT_APB,
31 NMI_NMI,
19 NR_IRQS, 32 NR_IRQS,
20}; 33};
21 34
22#endif /* __KERNEL__ */ 35typedef void (*ext_int_handler_t)(unsigned int, unsigned int, unsigned long);
23#endif 36
37int register_external_interrupt(u16 code, ext_int_handler_t handler);
38int unregister_external_interrupt(u16 code, ext_int_handler_t handler);
39void service_subclass_irq_register(void);
40void service_subclass_irq_unregister(void);
41
42#endif /* _ASM_IRQ_H */
diff --git a/arch/s390/include/asm/irqflags.h b/arch/s390/include/asm/irqflags.h
index 15b3ac253898..865d6d891ace 100644
--- a/arch/s390/include/asm/irqflags.h
+++ b/arch/s390/include/asm/irqflags.h
@@ -8,8 +8,8 @@
8 8
9#include <linux/types.h> 9#include <linux/types.h>
10 10
11/* store then or system mask. */ 11/* store then OR system mask. */
12#define __raw_local_irq_stosm(__or) \ 12#define __arch_local_irq_stosm(__or) \
13({ \ 13({ \
14 unsigned long __mask; \ 14 unsigned long __mask; \
15 asm volatile( \ 15 asm volatile( \
@@ -18,8 +18,8 @@
18 __mask; \ 18 __mask; \
19}) 19})
20 20
21/* store then and system mask. */ 21/* store then AND system mask. */
22#define __raw_local_irq_stnsm(__and) \ 22#define __arch_local_irq_stnsm(__and) \
23({ \ 23({ \
24 unsigned long __mask; \ 24 unsigned long __mask; \
25 asm volatile( \ 25 asm volatile( \
@@ -29,39 +29,44 @@
29}) 29})
30 30
31/* set system mask. */ 31/* set system mask. */
32#define __raw_local_irq_ssm(__mask) \ 32static inline void __arch_local_irq_ssm(unsigned long flags)
33({ \ 33{
34 asm volatile("ssm %0" : : "Q" (__mask) : "memory"); \ 34 asm volatile("ssm %0" : : "Q" (flags) : "memory");
35}) 35}
36 36
37/* interrupt control.. */ 37static inline unsigned long arch_local_save_flags(void)
38static inline unsigned long raw_local_irq_enable(void)
39{ 38{
40 return __raw_local_irq_stosm(0x03); 39 return __arch_local_irq_stosm(0x00);
41} 40}
42 41
43static inline unsigned long raw_local_irq_disable(void) 42static inline unsigned long arch_local_irq_save(void)
44{ 43{
45 return __raw_local_irq_stnsm(0xfc); 44 return __arch_local_irq_stnsm(0xfc);
46} 45}
47 46
48#define raw_local_save_flags(x) \ 47static inline void arch_local_irq_disable(void)
49do { \ 48{
50 typecheck(unsigned long, x); \ 49 arch_local_irq_save();
51 (x) = __raw_local_irq_stosm(0x00); \ 50}
52} while (0)
53 51
54static inline void raw_local_irq_restore(unsigned long flags) 52static inline void arch_local_irq_enable(void)
55{ 53{
56 __raw_local_irq_ssm(flags); 54 __arch_local_irq_stosm(0x03);
57} 55}
58 56
59static inline int raw_irqs_disabled_flags(unsigned long flags) 57static inline void arch_local_irq_restore(unsigned long flags)
58{
59 __arch_local_irq_ssm(flags);
60}
61
62static inline bool arch_irqs_disabled_flags(unsigned long flags)
60{ 63{
61 return !(flags & (3UL << (BITS_PER_LONG - 8))); 64 return !(flags & (3UL << (BITS_PER_LONG - 8)));
62} 65}
63 66
64/* For spinlocks etc */ 67static inline bool arch_irqs_disabled(void)
65#define raw_local_irq_save(x) ((x) = raw_local_irq_disable()) 68{
69 return arch_irqs_disabled_flags(arch_local_save_flags());
70}
66 71
67#endif /* __ASM_IRQFLAGS_H */ 72#endif /* __ASM_IRQFLAGS_H */
diff --git a/arch/s390/include/asm/jump_label.h b/arch/s390/include/asm/jump_label.h
new file mode 100644
index 000000000000..95a6cf2b5b67
--- /dev/null
+++ b/arch/s390/include/asm/jump_label.h
@@ -0,0 +1,37 @@
1#ifndef _ASM_S390_JUMP_LABEL_H
2#define _ASM_S390_JUMP_LABEL_H
3
4#include <linux/types.h>
5
6#define JUMP_LABEL_NOP_SIZE 6
7
8#ifdef CONFIG_64BIT
9#define ASM_PTR ".quad"
10#define ASM_ALIGN ".balign 8"
11#else
12#define ASM_PTR ".long"
13#define ASM_ALIGN ".balign 4"
14#endif
15
16static __always_inline bool arch_static_branch(struct jump_label_key *key)
17{
18 asm goto("0: brcl 0,0\n"
19 ".pushsection __jump_table, \"aw\"\n"
20 ASM_ALIGN "\n"
21 ASM_PTR " 0b, %l[label], %0\n"
22 ".popsection\n"
23 : : "X" (key) : : label);
24 return false;
25label:
26 return true;
27}
28
29typedef unsigned long jump_label_t;
30
31struct jump_entry {
32 jump_label_t code;
33 jump_label_t target;
34 jump_label_t key;
35};
36
37#endif
diff --git a/arch/s390/include/asm/kprobes.h b/arch/s390/include/asm/kprobes.h
index 330f68caffe4..a231a9439c4b 100644
--- a/arch/s390/include/asm/kprobes.h
+++ b/arch/s390/include/asm/kprobes.h
@@ -31,7 +31,6 @@
31#include <linux/ptrace.h> 31#include <linux/ptrace.h>
32#include <linux/percpu.h> 32#include <linux/percpu.h>
33 33
34#define __ARCH_WANT_KPROBES_INSN_SLOT
35struct pt_regs; 34struct pt_regs;
36struct kprobe; 35struct kprobe;
37 36
@@ -58,23 +57,12 @@ typedef u16 kprobe_opcode_t;
58/* Architecture specific copy of original instruction */ 57/* Architecture specific copy of original instruction */
59struct arch_specific_insn { 58struct arch_specific_insn {
60 /* copy of original instruction */ 59 /* copy of original instruction */
61 kprobe_opcode_t *insn; 60 kprobe_opcode_t insn[MAX_INSN_SIZE];
62 int fixup;
63 int ilen;
64 int reg;
65}; 61};
66 62
67struct ins_replace_args {
68 kprobe_opcode_t *ptr;
69 kprobe_opcode_t old;
70 kprobe_opcode_t new;
71};
72struct prev_kprobe { 63struct prev_kprobe {
73 struct kprobe *kp; 64 struct kprobe *kp;
74 unsigned long status; 65 unsigned long status;
75 unsigned long saved_psw;
76 unsigned long kprobe_saved_imask;
77 unsigned long kprobe_saved_ctl[3];
78}; 66};
79 67
80/* per-cpu kprobe control block */ 68/* per-cpu kprobe control block */
@@ -82,17 +70,13 @@ struct kprobe_ctlblk {
82 unsigned long kprobe_status; 70 unsigned long kprobe_status;
83 unsigned long kprobe_saved_imask; 71 unsigned long kprobe_saved_imask;
84 unsigned long kprobe_saved_ctl[3]; 72 unsigned long kprobe_saved_ctl[3];
85 struct pt_regs jprobe_saved_regs;
86 unsigned long jprobe_saved_r14;
87 unsigned long jprobe_saved_r15;
88 struct prev_kprobe prev_kprobe; 73 struct prev_kprobe prev_kprobe;
74 struct pt_regs jprobe_saved_regs;
89 kprobe_opcode_t jprobes_stack[MAX_STACK_SIZE]; 75 kprobe_opcode_t jprobes_stack[MAX_STACK_SIZE];
90}; 76};
91 77
92void arch_remove_kprobe(struct kprobe *p); 78void arch_remove_kprobe(struct kprobe *p);
93void kretprobe_trampoline(void); 79void kretprobe_trampoline(void);
94int is_prohibited_opcode(kprobe_opcode_t *instruction);
95void get_instruction_type(struct arch_specific_insn *ainsn);
96 80
97int kprobe_fault_handler(struct pt_regs *regs, int trapnr); 81int kprobe_fault_handler(struct pt_regs *regs, int trapnr);
98int kprobe_exceptions_notify(struct notifier_block *self, 82int kprobe_exceptions_notify(struct notifier_block *self,
diff --git a/arch/s390/include/asm/kvm_virtio.h b/arch/s390/include/asm/kvm_virtio.h
index acdfdff26611..72f614181eff 100644
--- a/arch/s390/include/asm/kvm_virtio.h
+++ b/arch/s390/include/asm/kvm_virtio.h
@@ -54,4 +54,11 @@ struct kvm_vqconfig {
54 * This is pagesize for historical reasons. */ 54 * This is pagesize for historical reasons. */
55#define KVM_S390_VIRTIO_RING_ALIGN 4096 55#define KVM_S390_VIRTIO_RING_ALIGN 4096
56 56
57
58/* These values are supposed to be in ext_params on an interrupt */
59#define VIRTIO_PARAM_MASK 0xff
60#define VIRTIO_PARAM_VRING_INTERRUPT 0x0
61#define VIRTIO_PARAM_CONFIG_CHANGED 0x1
62#define VIRTIO_PARAM_DEV_ADD 0x2
63
57#endif 64#endif
diff --git a/arch/s390/include/asm/lowcore.h b/arch/s390/include/asm/lowcore.h
index 0f97ef2d92ac..228cf0b295db 100644
--- a/arch/s390/include/asm/lowcore.h
+++ b/arch/s390/include/asm/lowcore.h
@@ -124,7 +124,7 @@ struct _lowcore {
124 /* Address space pointer. */ 124 /* Address space pointer. */
125 __u32 kernel_asce; /* 0x02ac */ 125 __u32 kernel_asce; /* 0x02ac */
126 __u32 user_asce; /* 0x02b0 */ 126 __u32 user_asce; /* 0x02b0 */
127 __u32 user_exec_asce; /* 0x02b4 */ 127 __u32 current_pid; /* 0x02b4 */
128 128
129 /* SMP info area */ 129 /* SMP info area */
130 __u32 cpu_nr; /* 0x02b8 */ 130 __u32 cpu_nr; /* 0x02b8 */
@@ -150,9 +150,10 @@ struct _lowcore {
150 */ 150 */
151 __u32 ipib; /* 0x0e00 */ 151 __u32 ipib; /* 0x0e00 */
152 __u32 ipib_checksum; /* 0x0e04 */ 152 __u32 ipib_checksum; /* 0x0e04 */
153 __u8 pad_0x0e08[0x0f00-0x0e08]; /* 0x0e08 */
153 154
154 /* Align to the top 1k of prefix area */ 155 /* Extended facility list */
155 __u8 pad_0x0e08[0x1000-0x0e08]; /* 0x0e08 */ 156 __u64 stfle_fac_list[32]; /* 0x0f00 */
156} __packed; 157} __packed;
157 158
158#else /* CONFIG_32BIT */ 159#else /* CONFIG_32BIT */
@@ -254,7 +255,7 @@ struct _lowcore {
254 /* Address space pointer. */ 255 /* Address space pointer. */
255 __u64 kernel_asce; /* 0x0310 */ 256 __u64 kernel_asce; /* 0x0310 */
256 __u64 user_asce; /* 0x0318 */ 257 __u64 user_asce; /* 0x0318 */
257 __u64 user_exec_asce; /* 0x0320 */ 258 __u64 current_pid; /* 0x0320 */
258 259
259 /* SMP info area */ 260 /* SMP info area */
260 __u32 cpu_nr; /* 0x0328 */ 261 __u32 cpu_nr; /* 0x0328 */
@@ -285,7 +286,11 @@ struct _lowcore {
285 */ 286 */
286 __u64 ipib; /* 0x0e00 */ 287 __u64 ipib; /* 0x0e00 */
287 __u32 ipib_checksum; /* 0x0e08 */ 288 __u32 ipib_checksum; /* 0x0e08 */
288 __u8 pad_0x0e0c[0x11b8-0x0e0c]; /* 0x0e0c */ 289 __u8 pad_0x0e0c[0x0f00-0x0e0c]; /* 0x0e0c */
290
291 /* Extended facility list */
292 __u64 stfle_fac_list[32]; /* 0x0f00 */
293 __u8 pad_0x1000[0x11b8-0x1000]; /* 0x1000 */
289 294
290 /* 64 bit extparam used for pfault/diag 250: defined by architecture */ 295 /* 64 bit extparam used for pfault/diag 250: defined by architecture */
291 __u64 ext_params2; /* 0x11B8 */ 296 __u64 ext_params2; /* 0x11B8 */
diff --git a/arch/s390/include/asm/mmu.h b/arch/s390/include/asm/mmu.h
index 78522cdefdd4..82d0847896a0 100644
--- a/arch/s390/include/asm/mmu.h
+++ b/arch/s390/include/asm/mmu.h
@@ -5,19 +5,18 @@ typedef struct {
5 atomic_t attach_count; 5 atomic_t attach_count;
6 unsigned int flush_mm; 6 unsigned int flush_mm;
7 spinlock_t list_lock; 7 spinlock_t list_lock;
8 struct list_head crst_list;
9 struct list_head pgtable_list; 8 struct list_head pgtable_list;
10 unsigned long asce_bits; 9 unsigned long asce_bits;
11 unsigned long asce_limit; 10 unsigned long asce_limit;
12 unsigned long vdso_base; 11 unsigned long vdso_base;
13 int noexec; 12 /* Cloned contexts will be created with extended page tables. */
14 int has_pgste; /* The mmu context has extended page tables */ 13 unsigned int alloc_pgste:1;
15 int alloc_pgste; /* cloned contexts will have extended page tables */ 14 /* The mmu context has extended page tables. */
15 unsigned int has_pgste:1;
16} mm_context_t; 16} mm_context_t;
17 17
18#define INIT_MM_CONTEXT(name) \ 18#define INIT_MM_CONTEXT(name) \
19 .context.list_lock = __SPIN_LOCK_UNLOCKED(name.context.list_lock), \ 19 .context.list_lock = __SPIN_LOCK_UNLOCKED(name.context.list_lock), \
20 .context.crst_list = LIST_HEAD_INIT(name.context.crst_list), \
21 .context.pgtable_list = LIST_HEAD_INIT(name.context.pgtable_list), 20 .context.pgtable_list = LIST_HEAD_INIT(name.context.pgtable_list),
22 21
23#endif 22#endif
diff --git a/arch/s390/include/asm/mmu_context.h b/arch/s390/include/asm/mmu_context.h
index a6f0e7cc9cde..5682f160ff82 100644
--- a/arch/s390/include/asm/mmu_context.h
+++ b/arch/s390/include/asm/mmu_context.h
@@ -23,7 +23,7 @@ static inline int init_new_context(struct task_struct *tsk,
23#ifdef CONFIG_64BIT 23#ifdef CONFIG_64BIT
24 mm->context.asce_bits |= _ASCE_TYPE_REGION3; 24 mm->context.asce_bits |= _ASCE_TYPE_REGION3;
25#endif 25#endif
26 if (current->mm->context.alloc_pgste) { 26 if (current->mm && current->mm->context.alloc_pgste) {
27 /* 27 /*
28 * alloc_pgste indicates, that any NEW context will be created 28 * alloc_pgste indicates, that any NEW context will be created
29 * with extended page tables. The old context is unchanged. The 29 * with extended page tables. The old context is unchanged. The
@@ -35,11 +35,9 @@ static inline int init_new_context(struct task_struct *tsk,
35 * and if has_pgste is set, it will create extended page 35 * and if has_pgste is set, it will create extended page
36 * tables. 36 * tables.
37 */ 37 */
38 mm->context.noexec = 0;
39 mm->context.has_pgste = 1; 38 mm->context.has_pgste = 1;
40 mm->context.alloc_pgste = 1; 39 mm->context.alloc_pgste = 1;
41 } else { 40 } else {
42 mm->context.noexec = (user_mode == SECONDARY_SPACE_MODE);
43 mm->context.has_pgste = 0; 41 mm->context.has_pgste = 0;
44 mm->context.alloc_pgste = 0; 42 mm->context.alloc_pgste = 0;
45 } 43 }
@@ -63,10 +61,8 @@ static inline void update_mm(struct mm_struct *mm, struct task_struct *tsk)
63 S390_lowcore.user_asce = mm->context.asce_bits | __pa(pgd); 61 S390_lowcore.user_asce = mm->context.asce_bits | __pa(pgd);
64 if (user_mode != HOME_SPACE_MODE) { 62 if (user_mode != HOME_SPACE_MODE) {
65 /* Load primary space page table origin. */ 63 /* Load primary space page table origin. */
66 pgd = mm->context.noexec ? get_shadow_table(pgd) : pgd;
67 S390_lowcore.user_exec_asce = mm->context.asce_bits | __pa(pgd);
68 asm volatile(LCTL_OPCODE" 1,1,%0\n" 64 asm volatile(LCTL_OPCODE" 1,1,%0\n"
69 : : "m" (S390_lowcore.user_exec_asce) ); 65 : : "m" (S390_lowcore.user_asce) );
70 } else 66 } else
71 /* Load home space page table origin. */ 67 /* Load home space page table origin. */
72 asm volatile(LCTL_OPCODE" 13,13,%0" 68 asm volatile(LCTL_OPCODE" 13,13,%0"
diff --git a/arch/s390/include/asm/mutex.h b/arch/s390/include/asm/mutex.h
index 458c1f7fbc18..688271f5f2e4 100644
--- a/arch/s390/include/asm/mutex.h
+++ b/arch/s390/include/asm/mutex.h
@@ -7,3 +7,5 @@
7 */ 7 */
8 8
9#include <asm-generic/mutex-dec.h> 9#include <asm-generic/mutex-dec.h>
10
11#define arch_mutex_cpu_relax() barrier()
diff --git a/arch/s390/include/asm/page.h b/arch/s390/include/asm/page.h
index af650fb47206..accb372ddc7e 100644
--- a/arch/s390/include/asm/page.h
+++ b/arch/s390/include/asm/page.h
@@ -90,6 +90,7 @@ static inline void copy_page(void *to, void *from)
90 */ 90 */
91 91
92typedef struct { unsigned long pgprot; } pgprot_t; 92typedef struct { unsigned long pgprot; } pgprot_t;
93typedef struct { unsigned long pgste; } pgste_t;
93typedef struct { unsigned long pte; } pte_t; 94typedef struct { unsigned long pte; } pte_t;
94typedef struct { unsigned long pmd; } pmd_t; 95typedef struct { unsigned long pmd; } pmd_t;
95typedef struct { unsigned long pud; } pud_t; 96typedef struct { unsigned long pud; } pud_t;
@@ -97,35 +98,91 @@ typedef struct { unsigned long pgd; } pgd_t;
97typedef pte_t *pgtable_t; 98typedef pte_t *pgtable_t;
98 99
99#define pgprot_val(x) ((x).pgprot) 100#define pgprot_val(x) ((x).pgprot)
101#define pgste_val(x) ((x).pgste)
100#define pte_val(x) ((x).pte) 102#define pte_val(x) ((x).pte)
101#define pmd_val(x) ((x).pmd) 103#define pmd_val(x) ((x).pmd)
102#define pud_val(x) ((x).pud) 104#define pud_val(x) ((x).pud)
103#define pgd_val(x) ((x).pgd) 105#define pgd_val(x) ((x).pgd)
104 106
107#define __pgste(x) ((pgste_t) { (x) } )
105#define __pte(x) ((pte_t) { (x) } ) 108#define __pte(x) ((pte_t) { (x) } )
106#define __pmd(x) ((pmd_t) { (x) } ) 109#define __pmd(x) ((pmd_t) { (x) } )
110#define __pud(x) ((pud_t) { (x) } )
107#define __pgd(x) ((pgd_t) { (x) } ) 111#define __pgd(x) ((pgd_t) { (x) } )
108#define __pgprot(x) ((pgprot_t) { (x) } ) 112#define __pgprot(x) ((pgprot_t) { (x) } )
109 113
110static inline void 114static inline void page_set_storage_key(unsigned long addr,
111page_set_storage_key(unsigned long addr, unsigned int skey) 115 unsigned char skey, int mapped)
112{ 116{
113 asm volatile("sske %0,%1" : : "d" (skey), "a" (addr)); 117 if (!mapped)
118 asm volatile(".insn rrf,0xb22b0000,%0,%1,8,0"
119 : : "d" (skey), "a" (addr));
120 else
121 asm volatile("sske %0,%1" : : "d" (skey), "a" (addr));
114} 122}
115 123
116static inline unsigned int 124static inline unsigned char page_get_storage_key(unsigned long addr)
117page_get_storage_key(unsigned long addr)
118{ 125{
119 unsigned int skey; 126 unsigned char skey;
120 127
121 asm volatile("iske %0,%1" : "=d" (skey) : "a" (addr), "0" (0)); 128 asm volatile("iske %0,%1" : "=d" (skey) : "a" (addr));
122 return skey; 129 return skey;
123} 130}
124 131
132static inline int page_reset_referenced(unsigned long addr)
133{
134 unsigned int ipm;
135
136 asm volatile(
137 " rrbe 0,%1\n"
138 " ipm %0\n"
139 : "=d" (ipm) : "a" (addr) : "cc");
140 return !!(ipm & 0x20000000);
141}
142
143/* Bits int the storage key */
144#define _PAGE_CHANGED 0x02 /* HW changed bit */
145#define _PAGE_REFERENCED 0x04 /* HW referenced bit */
146#define _PAGE_FP_BIT 0x08 /* HW fetch protection bit */
147#define _PAGE_ACC_BITS 0xf0 /* HW access control bits */
148
149/*
150 * Test and clear dirty bit in storage key.
151 * We can't clear the changed bit atomically. This is a potential
152 * race against modification of the referenced bit. This function
153 * should therefore only be called if it is not mapped in any
154 * address space.
155 */
156#define __HAVE_ARCH_PAGE_TEST_AND_CLEAR_DIRTY
157static inline int page_test_and_clear_dirty(unsigned long pfn, int mapped)
158{
159 unsigned char skey;
160
161 skey = page_get_storage_key(pfn << PAGE_SHIFT);
162 if (!(skey & _PAGE_CHANGED))
163 return 0;
164 page_set_storage_key(pfn << PAGE_SHIFT, skey & ~_PAGE_CHANGED, mapped);
165 return 1;
166}
167
168/*
169 * Test and clear referenced bit in storage key.
170 */
171#define __HAVE_ARCH_PAGE_TEST_AND_CLEAR_YOUNG
172static inline int page_test_and_clear_young(unsigned long pfn)
173{
174 return page_reset_referenced(pfn << PAGE_SHIFT);
175}
176
125struct page; 177struct page;
126void arch_free_page(struct page *page, int order); 178void arch_free_page(struct page *page, int order);
127void arch_alloc_page(struct page *page, int order); 179void arch_alloc_page(struct page *page, int order);
128 180
181static inline int devmem_is_allowed(unsigned long pfn)
182{
183 return 0;
184}
185
129#define HAVE_ARCH_FREE_PAGE 186#define HAVE_ARCH_FREE_PAGE
130#define HAVE_ARCH_ALLOC_PAGE 187#define HAVE_ARCH_ALLOC_PAGE
131 188
diff --git a/arch/s390/include/asm/percpu.h b/arch/s390/include/asm/percpu.h
index f7ad8719d02d..5325c89a5843 100644
--- a/arch/s390/include/asm/percpu.h
+++ b/arch/s390/include/asm/percpu.h
@@ -1,6 +1,9 @@
1#ifndef __ARCH_S390_PERCPU__ 1#ifndef __ARCH_S390_PERCPU__
2#define __ARCH_S390_PERCPU__ 2#define __ARCH_S390_PERCPU__
3 3
4#include <linux/preempt.h>
5#include <asm/cmpxchg.h>
6
4/* 7/*
5 * s390 uses its own implementation for per cpu data, the offset of 8 * s390 uses its own implementation for per cpu data, the offset of
6 * the cpu local data area is cached in the cpu's lowcore memory. 9 * the cpu local data area is cached in the cpu's lowcore memory.
@@ -16,6 +19,71 @@
16#define ARCH_NEEDS_WEAK_PER_CPU 19#define ARCH_NEEDS_WEAK_PER_CPU
17#endif 20#endif
18 21
22#define arch_irqsafe_cpu_to_op(pcp, val, op) \
23do { \
24 typedef typeof(pcp) pcp_op_T__; \
25 pcp_op_T__ old__, new__, prev__; \
26 pcp_op_T__ *ptr__; \
27 preempt_disable(); \
28 ptr__ = __this_cpu_ptr(&(pcp)); \
29 prev__ = *ptr__; \
30 do { \
31 old__ = prev__; \
32 new__ = old__ op (val); \
33 switch (sizeof(*ptr__)) { \
34 case 8: \
35 prev__ = cmpxchg64(ptr__, old__, new__); \
36 break; \
37 default: \
38 prev__ = cmpxchg(ptr__, old__, new__); \
39 } \
40 } while (prev__ != old__); \
41 preempt_enable(); \
42} while (0)
43
44#define irqsafe_cpu_add_1(pcp, val) arch_irqsafe_cpu_to_op(pcp, val, +)
45#define irqsafe_cpu_add_2(pcp, val) arch_irqsafe_cpu_to_op(pcp, val, +)
46#define irqsafe_cpu_add_4(pcp, val) arch_irqsafe_cpu_to_op(pcp, val, +)
47#define irqsafe_cpu_add_8(pcp, val) arch_irqsafe_cpu_to_op(pcp, val, +)
48
49#define irqsafe_cpu_and_1(pcp, val) arch_irqsafe_cpu_to_op(pcp, val, &)
50#define irqsafe_cpu_and_2(pcp, val) arch_irqsafe_cpu_to_op(pcp, val, &)
51#define irqsafe_cpu_and_4(pcp, val) arch_irqsafe_cpu_to_op(pcp, val, &)
52#define irqsafe_cpu_and_8(pcp, val) arch_irqsafe_cpu_to_op(pcp, val, &)
53
54#define irqsafe_cpu_or_1(pcp, val) arch_irqsafe_cpu_to_op(pcp, val, |)
55#define irqsafe_cpu_or_2(pcp, val) arch_irqsafe_cpu_to_op(pcp, val, |)
56#define irqsafe_cpu_or_4(pcp, val) arch_irqsafe_cpu_to_op(pcp, val, |)
57#define irqsafe_cpu_or_8(pcp, val) arch_irqsafe_cpu_to_op(pcp, val, |)
58
59#define irqsafe_cpu_xor_1(pcp, val) arch_irqsafe_cpu_to_op(pcp, val, ^)
60#define irqsafe_cpu_xor_2(pcp, val) arch_irqsafe_cpu_to_op(pcp, val, ^)
61#define irqsafe_cpu_xor_4(pcp, val) arch_irqsafe_cpu_to_op(pcp, val, ^)
62#define irqsafe_cpu_xor_8(pcp, val) arch_irqsafe_cpu_to_op(pcp, val, ^)
63
64#define arch_irqsafe_cpu_cmpxchg(pcp, oval, nval) \
65({ \
66 typedef typeof(pcp) pcp_op_T__; \
67 pcp_op_T__ ret__; \
68 pcp_op_T__ *ptr__; \
69 preempt_disable(); \
70 ptr__ = __this_cpu_ptr(&(pcp)); \
71 switch (sizeof(*ptr__)) { \
72 case 8: \
73 ret__ = cmpxchg64(ptr__, oval, nval); \
74 break; \
75 default: \
76 ret__ = cmpxchg(ptr__, oval, nval); \
77 } \
78 preempt_enable(); \
79 ret__; \
80})
81
82#define irqsafe_cpu_cmpxchg_1(pcp, oval, nval) arch_irqsafe_cpu_cmpxchg(pcp, oval, nval)
83#define irqsafe_cpu_cmpxchg_2(pcp, oval, nval) arch_irqsafe_cpu_cmpxchg(pcp, oval, nval)
84#define irqsafe_cpu_cmpxchg_4(pcp, oval, nval) arch_irqsafe_cpu_cmpxchg(pcp, oval, nval)
85#define irqsafe_cpu_cmpxchg_8(pcp, oval, nval) arch_irqsafe_cpu_cmpxchg(pcp, oval, nval)
86
19#include <asm-generic/percpu.h> 87#include <asm-generic/percpu.h>
20 88
21#endif /* __ARCH_S390_PERCPU__ */ 89#endif /* __ARCH_S390_PERCPU__ */
diff --git a/arch/s390/include/asm/perf_event.h b/arch/s390/include/asm/perf_event.h
index 3840cbe77637..a75f168d2718 100644
--- a/arch/s390/include/asm/perf_event.h
+++ b/arch/s390/include/asm/perf_event.h
@@ -4,7 +4,6 @@
4 * Copyright 2009 Martin Schwidefsky, IBM Corporation. 4 * Copyright 2009 Martin Schwidefsky, IBM Corporation.
5 */ 5 */
6 6
7static inline void set_perf_event_pending(void) {} 7/* Empty, just to avoid compiling error */
8static inline void clear_perf_event_pending(void) {}
9 8
10#define PERF_EVENT_INDEX_OFFSET 0 9#define PERF_EVENT_INDEX_OFFSET 0
diff --git a/arch/s390/include/asm/pgalloc.h b/arch/s390/include/asm/pgalloc.h
index 68940d0bad91..38e71ebcd3c2 100644
--- a/arch/s390/include/asm/pgalloc.h
+++ b/arch/s390/include/asm/pgalloc.h
@@ -17,14 +17,15 @@
17#include <linux/gfp.h> 17#include <linux/gfp.h>
18#include <linux/mm.h> 18#include <linux/mm.h>
19 19
20#define check_pgt_cache() do {} while (0) 20unsigned long *crst_table_alloc(struct mm_struct *);
21
22unsigned long *crst_table_alloc(struct mm_struct *, int);
23void crst_table_free(struct mm_struct *, unsigned long *); 21void crst_table_free(struct mm_struct *, unsigned long *);
24 22
25unsigned long *page_table_alloc(struct mm_struct *); 23unsigned long *page_table_alloc(struct mm_struct *);
26void page_table_free(struct mm_struct *, unsigned long *); 24void page_table_free(struct mm_struct *, unsigned long *);
27void disable_noexec(struct mm_struct *, struct task_struct *); 25#ifdef CONFIG_HAVE_RCU_TABLE_FREE
26void page_table_free_rcu(struct mmu_gather *, unsigned long *);
27void __tlb_remove_table(void *_table);
28#endif
28 29
29static inline void clear_table(unsigned long *s, unsigned long val, size_t n) 30static inline void clear_table(unsigned long *s, unsigned long val, size_t n)
30{ 31{
@@ -48,9 +49,6 @@ static inline void clear_table(unsigned long *s, unsigned long val, size_t n)
48static inline void crst_table_init(unsigned long *crst, unsigned long entry) 49static inline void crst_table_init(unsigned long *crst, unsigned long entry)
49{ 50{
50 clear_table(crst, entry, sizeof(unsigned long)*2048); 51 clear_table(crst, entry, sizeof(unsigned long)*2048);
51 crst = get_shadow_table(crst);
52 if (crst)
53 clear_table(crst, entry, sizeof(unsigned long)*2048);
54} 52}
55 53
56#ifndef __s390x__ 54#ifndef __s390x__
@@ -67,10 +65,7 @@ static inline unsigned long pgd_entry_type(struct mm_struct *mm)
67#define pmd_free(mm, x) do { } while (0) 65#define pmd_free(mm, x) do { } while (0)
68 66
69#define pgd_populate(mm, pgd, pud) BUG() 67#define pgd_populate(mm, pgd, pud) BUG()
70#define pgd_populate_kernel(mm, pgd, pud) BUG()
71
72#define pud_populate(mm, pud, pmd) BUG() 68#define pud_populate(mm, pud, pmd) BUG()
73#define pud_populate_kernel(mm, pud, pmd) BUG()
74 69
75#else /* __s390x__ */ 70#else /* __s390x__ */
76 71
@@ -88,7 +83,7 @@ void crst_table_downgrade(struct mm_struct *, unsigned long limit);
88 83
89static inline pud_t *pud_alloc_one(struct mm_struct *mm, unsigned long address) 84static inline pud_t *pud_alloc_one(struct mm_struct *mm, unsigned long address)
90{ 85{
91 unsigned long *table = crst_table_alloc(mm, mm->context.noexec); 86 unsigned long *table = crst_table_alloc(mm);
92 if (table) 87 if (table)
93 crst_table_init(table, _REGION3_ENTRY_EMPTY); 88 crst_table_init(table, _REGION3_ENTRY_EMPTY);
94 return (pud_t *) table; 89 return (pud_t *) table;
@@ -97,43 +92,21 @@ static inline pud_t *pud_alloc_one(struct mm_struct *mm, unsigned long address)
97 92
98static inline pmd_t *pmd_alloc_one(struct mm_struct *mm, unsigned long vmaddr) 93static inline pmd_t *pmd_alloc_one(struct mm_struct *mm, unsigned long vmaddr)
99{ 94{
100 unsigned long *table = crst_table_alloc(mm, mm->context.noexec); 95 unsigned long *table = crst_table_alloc(mm);
101 if (table) 96 if (table)
102 crst_table_init(table, _SEGMENT_ENTRY_EMPTY); 97 crst_table_init(table, _SEGMENT_ENTRY_EMPTY);
103 return (pmd_t *) table; 98 return (pmd_t *) table;
104} 99}
105#define pmd_free(mm, pmd) crst_table_free(mm, (unsigned long *) pmd) 100#define pmd_free(mm, pmd) crst_table_free(mm, (unsigned long *) pmd)
106 101
107static inline void pgd_populate_kernel(struct mm_struct *mm,
108 pgd_t *pgd, pud_t *pud)
109{
110 pgd_val(*pgd) = _REGION2_ENTRY | __pa(pud);
111}
112
113static inline void pgd_populate(struct mm_struct *mm, pgd_t *pgd, pud_t *pud) 102static inline void pgd_populate(struct mm_struct *mm, pgd_t *pgd, pud_t *pud)
114{ 103{
115 pgd_populate_kernel(mm, pgd, pud); 104 pgd_val(*pgd) = _REGION2_ENTRY | __pa(pud);
116 if (mm->context.noexec) {
117 pgd = get_shadow_table(pgd);
118 pud = get_shadow_table(pud);
119 pgd_populate_kernel(mm, pgd, pud);
120 }
121}
122
123static inline void pud_populate_kernel(struct mm_struct *mm,
124 pud_t *pud, pmd_t *pmd)
125{
126 pud_val(*pud) = _REGION3_ENTRY | __pa(pmd);
127} 105}
128 106
129static inline void pud_populate(struct mm_struct *mm, pud_t *pud, pmd_t *pmd) 107static inline void pud_populate(struct mm_struct *mm, pud_t *pud, pmd_t *pmd)
130{ 108{
131 pud_populate_kernel(mm, pud, pmd); 109 pud_val(*pud) = _REGION3_ENTRY | __pa(pmd);
132 if (mm->context.noexec) {
133 pud = get_shadow_table(pud);
134 pmd = get_shadow_table(pmd);
135 pud_populate_kernel(mm, pud, pmd);
136 }
137} 110}
138 111
139#endif /* __s390x__ */ 112#endif /* __s390x__ */
@@ -141,29 +114,19 @@ static inline void pud_populate(struct mm_struct *mm, pud_t *pud, pmd_t *pmd)
141static inline pgd_t *pgd_alloc(struct mm_struct *mm) 114static inline pgd_t *pgd_alloc(struct mm_struct *mm)
142{ 115{
143 spin_lock_init(&mm->context.list_lock); 116 spin_lock_init(&mm->context.list_lock);
144 INIT_LIST_HEAD(&mm->context.crst_list);
145 INIT_LIST_HEAD(&mm->context.pgtable_list); 117 INIT_LIST_HEAD(&mm->context.pgtable_list);
146 return (pgd_t *) 118 return (pgd_t *) crst_table_alloc(mm);
147 crst_table_alloc(mm, user_mode == SECONDARY_SPACE_MODE);
148} 119}
149#define pgd_free(mm, pgd) crst_table_free(mm, (unsigned long *) pgd) 120#define pgd_free(mm, pgd) crst_table_free(mm, (unsigned long *) pgd)
150 121
151static inline void pmd_populate_kernel(struct mm_struct *mm,
152 pmd_t *pmd, pte_t *pte)
153{
154 pmd_val(*pmd) = _SEGMENT_ENTRY + __pa(pte);
155}
156
157static inline void pmd_populate(struct mm_struct *mm, 122static inline void pmd_populate(struct mm_struct *mm,
158 pmd_t *pmd, pgtable_t pte) 123 pmd_t *pmd, pgtable_t pte)
159{ 124{
160 pmd_populate_kernel(mm, pmd, pte); 125 pmd_val(*pmd) = _SEGMENT_ENTRY + __pa(pte);
161 if (mm->context.noexec) {
162 pmd = get_shadow_table(pmd);
163 pmd_populate_kernel(mm, pmd, pte + PTRS_PER_PTE);
164 }
165} 126}
166 127
128#define pmd_populate_kernel(mm, pmd, pte) pmd_populate(mm, pmd, pte)
129
167#define pmd_pgtable(pmd) \ 130#define pmd_pgtable(pmd) \
168 (pgtable_t)(pmd_val(pmd) & -sizeof(pte_t)*PTRS_PER_PTE) 131 (pgtable_t)(pmd_val(pmd) & -sizeof(pte_t)*PTRS_PER_PTE)
169 132
@@ -176,4 +139,6 @@ static inline void pmd_populate(struct mm_struct *mm,
176#define pte_free_kernel(mm, pte) page_table_free(mm, (unsigned long *) pte) 139#define pte_free_kernel(mm, pte) page_table_free(mm, (unsigned long *) pte)
177#define pte_free(mm, pte) page_table_free(mm, (unsigned long *) pte) 140#define pte_free(mm, pte) page_table_free(mm, (unsigned long *) pte)
178 141
142extern void rcu_table_freelist_finish(void);
143
179#endif /* _S390_PGALLOC_H */ 144#endif /* _S390_PGALLOC_H */
diff --git a/arch/s390/include/asm/pgtable.h b/arch/s390/include/asm/pgtable.h
index 3157441ee1da..801fbe1d837d 100644
--- a/arch/s390/include/asm/pgtable.h
+++ b/arch/s390/include/asm/pgtable.h
@@ -31,13 +31,13 @@
31#ifndef __ASSEMBLY__ 31#ifndef __ASSEMBLY__
32#include <linux/sched.h> 32#include <linux/sched.h>
33#include <linux/mm_types.h> 33#include <linux/mm_types.h>
34#include <asm/bitops.h>
35#include <asm/bug.h> 34#include <asm/bug.h>
36#include <asm/processor.h> 35#include <asm/page.h>
37 36
38extern pgd_t swapper_pg_dir[] __attribute__ ((aligned (4096))); 37extern pgd_t swapper_pg_dir[] __attribute__ ((aligned (4096)));
39extern void paging_init(void); 38extern void paging_init(void);
40extern void vmem_map_init(void); 39extern void vmem_map_init(void);
40extern void fault_init(void);
41 41
42/* 42/*
43 * The S390 doesn't have any external MMU info: the kernel page 43 * The S390 doesn't have any external MMU info: the kernel page
@@ -46,11 +46,27 @@ extern void vmem_map_init(void);
46#define update_mmu_cache(vma, address, ptep) do { } while (0) 46#define update_mmu_cache(vma, address, ptep) do { } while (0)
47 47
48/* 48/*
49 * ZERO_PAGE is a global shared page that is always zero: used 49 * ZERO_PAGE is a global shared page that is always zero; used
50 * for zero-mapped memory areas etc.. 50 * for zero-mapped memory areas etc..
51 */ 51 */
52extern char empty_zero_page[PAGE_SIZE]; 52
53#define ZERO_PAGE(vaddr) (virt_to_page(empty_zero_page)) 53extern unsigned long empty_zero_page;
54extern unsigned long zero_page_mask;
55
56#define ZERO_PAGE(vaddr) \
57 (virt_to_page((void *)(empty_zero_page + \
58 (((unsigned long)(vaddr)) &zero_page_mask))))
59
60#define is_zero_pfn is_zero_pfn
61static inline int is_zero_pfn(unsigned long pfn)
62{
63 extern unsigned long zero_pfn;
64 unsigned long offset_from_zero_pfn = pfn - zero_pfn;
65 return offset_from_zero_pfn <= (zero_page_mask >> PAGE_SHIFT);
66}
67
68#define my_zero_pfn(addr) page_to_pfn(ZERO_PAGE(addr))
69
54#endif /* !__ASSEMBLY__ */ 70#endif /* !__ASSEMBLY__ */
55 71
56/* 72/*
@@ -226,11 +242,13 @@ extern unsigned long VMALLOC_START;
226/* Software bits in the page table entry */ 242/* Software bits in the page table entry */
227#define _PAGE_SWT 0x001 /* SW pte type bit t */ 243#define _PAGE_SWT 0x001 /* SW pte type bit t */
228#define _PAGE_SWX 0x002 /* SW pte type bit x */ 244#define _PAGE_SWX 0x002 /* SW pte type bit x */
229#define _PAGE_SPECIAL 0x004 /* SW associated with special page */ 245#define _PAGE_SWC 0x004 /* SW pte changed bit (for KVM) */
246#define _PAGE_SWR 0x008 /* SW pte referenced bit (for KVM) */
247#define _PAGE_SPECIAL 0x010 /* SW associated with special page */
230#define __HAVE_ARCH_PTE_SPECIAL 248#define __HAVE_ARCH_PTE_SPECIAL
231 249
232/* Set of bits not changed in pte_modify */ 250/* Set of bits not changed in pte_modify */
233#define _PAGE_CHG_MASK (PAGE_MASK | _PAGE_SPECIAL) 251#define _PAGE_CHG_MASK (PAGE_MASK | _PAGE_SPECIAL | _PAGE_SWC | _PAGE_SWR)
234 252
235/* Six different types of pages. */ 253/* Six different types of pages. */
236#define _PAGE_TYPE_EMPTY 0x400 254#define _PAGE_TYPE_EMPTY 0x400
@@ -239,8 +257,6 @@ extern unsigned long VMALLOC_START;
239#define _PAGE_TYPE_FILE 0x601 /* bit 0x002 is used for offset !! */ 257#define _PAGE_TYPE_FILE 0x601 /* bit 0x002 is used for offset !! */
240#define _PAGE_TYPE_RO 0x200 258#define _PAGE_TYPE_RO 0x200
241#define _PAGE_TYPE_RW 0x000 259#define _PAGE_TYPE_RW 0x000
242#define _PAGE_TYPE_EX_RO 0x202
243#define _PAGE_TYPE_EX_RW 0x002
244 260
245/* 261/*
246 * Only four types for huge pages, using the invalid bit and protection bit 262 * Only four types for huge pages, using the invalid bit and protection bit
@@ -270,8 +286,6 @@ extern unsigned long VMALLOC_START;
270 * _PAGE_TYPE_FILE 11?1 -> 11?1 286 * _PAGE_TYPE_FILE 11?1 -> 11?1
271 * _PAGE_TYPE_RO 0100 -> 1100 287 * _PAGE_TYPE_RO 0100 -> 1100
272 * _PAGE_TYPE_RW 0000 -> 1000 288 * _PAGE_TYPE_RW 0000 -> 1000
273 * _PAGE_TYPE_EX_RO 0110 -> 1110
274 * _PAGE_TYPE_EX_RW 0010 -> 1010
275 * 289 *
276 * pte_none is true for bits combinations 1000, 1010, 1100, 1110 290 * pte_none is true for bits combinations 1000, 1010, 1100, 1110
277 * pte_present is true for bits combinations 0000, 0010, 0100, 0110, 1001 291 * pte_present is true for bits combinations 0000, 0010, 0100, 0110, 1001
@@ -279,16 +293,6 @@ extern unsigned long VMALLOC_START;
279 * swap pte is 1011 and 0001, 0011, 0101, 0111 are invalid. 293 * swap pte is 1011 and 0001, 0011, 0101, 0111 are invalid.
280 */ 294 */
281 295
282/* Page status table bits for virtualization */
283#define RCP_PCL_BIT 55
284#define RCP_HR_BIT 54
285#define RCP_HC_BIT 53
286#define RCP_GR_BIT 50
287#define RCP_GC_BIT 49
288
289/* User dirty bit for KVM's migration feature */
290#define KVM_UD_BIT 47
291
292#ifndef __s390x__ 296#ifndef __s390x__
293 297
294/* Bits in the segment table address-space-control-element */ 298/* Bits in the segment table address-space-control-element */
@@ -300,6 +304,7 @@ extern unsigned long VMALLOC_START;
300 304
301/* Bits in the segment table entry */ 305/* Bits in the segment table entry */
302#define _SEGMENT_ENTRY_ORIGIN 0x7fffffc0UL /* page table origin */ 306#define _SEGMENT_ENTRY_ORIGIN 0x7fffffc0UL /* page table origin */
307#define _SEGMENT_ENTRY_RO 0x200 /* page protection bit */
303#define _SEGMENT_ENTRY_INV 0x20 /* invalid segment table entry */ 308#define _SEGMENT_ENTRY_INV 0x20 /* invalid segment table entry */
304#define _SEGMENT_ENTRY_COMMON 0x10 /* common segment bit */ 309#define _SEGMENT_ENTRY_COMMON 0x10 /* common segment bit */
305#define _SEGMENT_ENTRY_PTL 0x0f /* page table length */ 310#define _SEGMENT_ENTRY_PTL 0x0f /* page table length */
@@ -307,6 +312,19 @@ extern unsigned long VMALLOC_START;
307#define _SEGMENT_ENTRY (_SEGMENT_ENTRY_PTL) 312#define _SEGMENT_ENTRY (_SEGMENT_ENTRY_PTL)
308#define _SEGMENT_ENTRY_EMPTY (_SEGMENT_ENTRY_INV) 313#define _SEGMENT_ENTRY_EMPTY (_SEGMENT_ENTRY_INV)
309 314
315/* Page status table bits for virtualization */
316#define RCP_ACC_BITS 0xf0000000UL
317#define RCP_FP_BIT 0x08000000UL
318#define RCP_PCL_BIT 0x00800000UL
319#define RCP_HR_BIT 0x00400000UL
320#define RCP_HC_BIT 0x00200000UL
321#define RCP_GR_BIT 0x00040000UL
322#define RCP_GC_BIT 0x00020000UL
323
324/* User dirty / referenced bit for KVM's migration feature */
325#define KVM_UR_BIT 0x00008000UL
326#define KVM_UC_BIT 0x00004000UL
327
310#else /* __s390x__ */ 328#else /* __s390x__ */
311 329
312/* Bits in the segment/region table address-space-control-element */ 330/* Bits in the segment/region table address-space-control-element */
@@ -349,6 +367,19 @@ extern unsigned long VMALLOC_START;
349#define _SEGMENT_ENTRY_LARGE 0x400 /* STE-format control, large page */ 367#define _SEGMENT_ENTRY_LARGE 0x400 /* STE-format control, large page */
350#define _SEGMENT_ENTRY_CO 0x100 /* change-recording override */ 368#define _SEGMENT_ENTRY_CO 0x100 /* change-recording override */
351 369
370/* Page status table bits for virtualization */
371#define RCP_ACC_BITS 0xf000000000000000UL
372#define RCP_FP_BIT 0x0800000000000000UL
373#define RCP_PCL_BIT 0x0080000000000000UL
374#define RCP_HR_BIT 0x0040000000000000UL
375#define RCP_HC_BIT 0x0020000000000000UL
376#define RCP_GR_BIT 0x0004000000000000UL
377#define RCP_GC_BIT 0x0002000000000000UL
378
379/* User dirty / referenced bit for KVM's migration feature */
380#define KVM_UR_BIT 0x0000800000000000UL
381#define KVM_UC_BIT 0x0000400000000000UL
382
352#endif /* __s390x__ */ 383#endif /* __s390x__ */
353 384
354/* 385/*
@@ -359,85 +390,54 @@ extern unsigned long VMALLOC_START;
359#define _ASCE_USER_BITS (_ASCE_SPACE_SWITCH | _ASCE_PRIVATE_SPACE | \ 390#define _ASCE_USER_BITS (_ASCE_SPACE_SWITCH | _ASCE_PRIVATE_SPACE | \
360 _ASCE_ALT_EVENT) 391 _ASCE_ALT_EVENT)
361 392
362/* Bits int the storage key */
363#define _PAGE_CHANGED 0x02 /* HW changed bit */
364#define _PAGE_REFERENCED 0x04 /* HW referenced bit */
365
366/* 393/*
367 * Page protection definitions. 394 * Page protection definitions.
368 */ 395 */
369#define PAGE_NONE __pgprot(_PAGE_TYPE_NONE) 396#define PAGE_NONE __pgprot(_PAGE_TYPE_NONE)
370#define PAGE_RO __pgprot(_PAGE_TYPE_RO) 397#define PAGE_RO __pgprot(_PAGE_TYPE_RO)
371#define PAGE_RW __pgprot(_PAGE_TYPE_RW) 398#define PAGE_RW __pgprot(_PAGE_TYPE_RW)
372#define PAGE_EX_RO __pgprot(_PAGE_TYPE_EX_RO)
373#define PAGE_EX_RW __pgprot(_PAGE_TYPE_EX_RW)
374 399
375#define PAGE_KERNEL PAGE_RW 400#define PAGE_KERNEL PAGE_RW
376#define PAGE_COPY PAGE_RO 401#define PAGE_COPY PAGE_RO
377 402
378/* 403/*
379 * Dependent on the EXEC_PROTECT option s390 can do execute protection. 404 * On s390 the page table entry has an invalid bit and a read-only bit.
380 * Write permission always implies read permission. In theory with a 405 * Read permission implies execute permission and write permission
381 * primary/secondary page table execute only can be implemented but 406 * implies read permission.
382 * it would cost an additional bit in the pte to distinguish all the
383 * different pte types. To avoid that execute permission currently
384 * implies read permission as well.
385 */ 407 */
386 /*xwr*/ 408 /*xwr*/
387#define __P000 PAGE_NONE 409#define __P000 PAGE_NONE
388#define __P001 PAGE_RO 410#define __P001 PAGE_RO
389#define __P010 PAGE_RO 411#define __P010 PAGE_RO
390#define __P011 PAGE_RO 412#define __P011 PAGE_RO
391#define __P100 PAGE_EX_RO 413#define __P100 PAGE_RO
392#define __P101 PAGE_EX_RO 414#define __P101 PAGE_RO
393#define __P110 PAGE_EX_RO 415#define __P110 PAGE_RO
394#define __P111 PAGE_EX_RO 416#define __P111 PAGE_RO
395 417
396#define __S000 PAGE_NONE 418#define __S000 PAGE_NONE
397#define __S001 PAGE_RO 419#define __S001 PAGE_RO
398#define __S010 PAGE_RW 420#define __S010 PAGE_RW
399#define __S011 PAGE_RW 421#define __S011 PAGE_RW
400#define __S100 PAGE_EX_RO 422#define __S100 PAGE_RO
401#define __S101 PAGE_EX_RO 423#define __S101 PAGE_RO
402#define __S110 PAGE_EX_RW 424#define __S110 PAGE_RW
403#define __S111 PAGE_EX_RW 425#define __S111 PAGE_RW
404
405#ifndef __s390x__
406# define PxD_SHADOW_SHIFT 1
407#else /* __s390x__ */
408# define PxD_SHADOW_SHIFT 2
409#endif /* __s390x__ */
410 426
411static inline void *get_shadow_table(void *table) 427static inline int mm_exclusive(struct mm_struct *mm)
412{ 428{
413 unsigned long addr, offset; 429 return likely(mm == current->active_mm &&
414 struct page *page; 430 atomic_read(&mm->context.attach_count) <= 1);
415
416 addr = (unsigned long) table;
417 offset = addr & ((PAGE_SIZE << PxD_SHADOW_SHIFT) - 1);
418 page = virt_to_page((void *)(addr ^ offset));
419 return (void *)(addr_t)(page->index ? (page->index | offset) : 0UL);
420} 431}
421 432
422/* 433static inline int mm_has_pgste(struct mm_struct *mm)
423 * Certain architectures need to do special things when PTEs
424 * within a page table are directly modified. Thus, the following
425 * hook is made available.
426 */
427static inline void set_pte_at(struct mm_struct *mm, unsigned long addr,
428 pte_t *ptep, pte_t entry)
429{ 434{
430 *ptep = entry; 435#ifdef CONFIG_PGSTE
431 if (mm->context.noexec) { 436 if (unlikely(mm->context.has_pgste))
432 if (!(pte_val(entry) & _PAGE_INVALID) && 437 return 1;
433 (pte_val(entry) & _PAGE_SWX)) 438#endif
434 pte_val(entry) |= _PAGE_RO; 439 return 0;
435 else
436 pte_val(entry) = _PAGE_TYPE_EMPTY;
437 ptep[PTRS_PER_PTE] = entry;
438 }
439} 440}
440
441/* 441/*
442 * pgd/pmd/pte query functions 442 * pgd/pmd/pte query functions
443 */ 443 */
@@ -550,55 +550,130 @@ static inline int pte_special(pte_t pte)
550} 550}
551 551
552#define __HAVE_ARCH_PTE_SAME 552#define __HAVE_ARCH_PTE_SAME
553#define pte_same(a,b) (pte_val(a) == pte_val(b)) 553static inline int pte_same(pte_t a, pte_t b)
554{
555 return pte_val(a) == pte_val(b);
556}
554 557
555static inline void rcp_lock(pte_t *ptep) 558static inline pgste_t pgste_get_lock(pte_t *ptep)
556{ 559{
560 unsigned long new = 0;
557#ifdef CONFIG_PGSTE 561#ifdef CONFIG_PGSTE
558 unsigned long *pgste = (unsigned long *) (ptep + PTRS_PER_PTE); 562 unsigned long old;
563
559 preempt_disable(); 564 preempt_disable();
560 while (test_and_set_bit(RCP_PCL_BIT, pgste)) 565 asm(
561 ; 566 " lg %0,%2\n"
567 "0: lgr %1,%0\n"
568 " nihh %0,0xff7f\n" /* clear RCP_PCL_BIT in old */
569 " oihh %1,0x0080\n" /* set RCP_PCL_BIT in new */
570 " csg %0,%1,%2\n"
571 " jl 0b\n"
572 : "=&d" (old), "=&d" (new), "=Q" (ptep[PTRS_PER_PTE])
573 : "Q" (ptep[PTRS_PER_PTE]) : "cc");
562#endif 574#endif
575 return __pgste(new);
563} 576}
564 577
565static inline void rcp_unlock(pte_t *ptep) 578static inline void pgste_set_unlock(pte_t *ptep, pgste_t pgste)
566{ 579{
567#ifdef CONFIG_PGSTE 580#ifdef CONFIG_PGSTE
568 unsigned long *pgste = (unsigned long *) (ptep + PTRS_PER_PTE); 581 asm(
569 clear_bit(RCP_PCL_BIT, pgste); 582 " nihh %1,0xff7f\n" /* clear RCP_PCL_BIT */
583 " stg %1,%0\n"
584 : "=Q" (ptep[PTRS_PER_PTE])
585 : "d" (pgste_val(pgste)), "Q" (ptep[PTRS_PER_PTE]) : "cc");
570 preempt_enable(); 586 preempt_enable();
571#endif 587#endif
572} 588}
573 589
574/* forward declaration for SetPageUptodate in page-flags.h*/ 590static inline pgste_t pgste_update_all(pte_t *ptep, pgste_t pgste)
575static inline void page_clear_dirty(struct page *page);
576#include <linux/page-flags.h>
577
578static inline void ptep_rcp_copy(pte_t *ptep)
579{ 591{
580#ifdef CONFIG_PGSTE 592#ifdef CONFIG_PGSTE
581 struct page *page = virt_to_page(pte_val(*ptep)); 593 unsigned long address, bits;
582 unsigned int skey; 594 unsigned char skey;
583 unsigned long *pgste = (unsigned long *) (ptep + PTRS_PER_PTE); 595
584 596 address = pte_val(*ptep) & PAGE_MASK;
585 skey = page_get_storage_key(page_to_phys(page)); 597 skey = page_get_storage_key(address);
586 if (skey & _PAGE_CHANGED) { 598 bits = skey & (_PAGE_CHANGED | _PAGE_REFERENCED);
587 set_bit_simple(RCP_GC_BIT, pgste); 599 /* Clear page changed & referenced bit in the storage key */
588 set_bit_simple(KVM_UD_BIT, pgste); 600 if (bits) {
589 } 601 skey ^= bits;
590 if (skey & _PAGE_REFERENCED) 602 page_set_storage_key(address, skey, 1);
591 set_bit_simple(RCP_GR_BIT, pgste);
592 if (test_and_clear_bit_simple(RCP_HC_BIT, pgste)) {
593 SetPageDirty(page);
594 set_bit_simple(KVM_UD_BIT, pgste);
595 } 603 }
596 if (test_and_clear_bit_simple(RCP_HR_BIT, pgste)) 604 /* Transfer page changed & referenced bit to guest bits in pgste */
597 SetPageReferenced(page); 605 pgste_val(pgste) |= bits << 48; /* RCP_GR_BIT & RCP_GC_BIT */
606 /* Get host changed & referenced bits from pgste */
607 bits |= (pgste_val(pgste) & (RCP_HR_BIT | RCP_HC_BIT)) >> 52;
608 /* Clear host bits in pgste. */
609 pgste_val(pgste) &= ~(RCP_HR_BIT | RCP_HC_BIT);
610 pgste_val(pgste) &= ~(RCP_ACC_BITS | RCP_FP_BIT);
611 /* Copy page access key and fetch protection bit to pgste */
612 pgste_val(pgste) |=
613 (unsigned long) (skey & (_PAGE_ACC_BITS | _PAGE_FP_BIT)) << 56;
614 /* Transfer changed and referenced to kvm user bits */
615 pgste_val(pgste) |= bits << 45; /* KVM_UR_BIT & KVM_UC_BIT */
616 /* Transfer changed & referenced to pte sofware bits */
617 pte_val(*ptep) |= bits << 1; /* _PAGE_SWR & _PAGE_SWC */
618#endif
619 return pgste;
620
621}
622
623static inline pgste_t pgste_update_young(pte_t *ptep, pgste_t pgste)
624{
625#ifdef CONFIG_PGSTE
626 int young;
627
628 young = page_reset_referenced(pte_val(*ptep) & PAGE_MASK);
629 /* Transfer page referenced bit to pte software bit (host view) */
630 if (young || (pgste_val(pgste) & RCP_HR_BIT))
631 pte_val(*ptep) |= _PAGE_SWR;
632 /* Clear host referenced bit in pgste. */
633 pgste_val(pgste) &= ~RCP_HR_BIT;
634 /* Transfer page referenced bit to guest bit in pgste */
635 pgste_val(pgste) |= (unsigned long) young << 50; /* set RCP_GR_BIT */
636#endif
637 return pgste;
638
639}
640
641static inline void pgste_set_pte(pte_t *ptep, pgste_t pgste)
642{
643#ifdef CONFIG_PGSTE
644 unsigned long address;
645 unsigned long okey, nkey;
646
647 address = pte_val(*ptep) & PAGE_MASK;
648 okey = nkey = page_get_storage_key(address);
649 nkey &= ~(_PAGE_ACC_BITS | _PAGE_FP_BIT);
650 /* Set page access key and fetch protection bit from pgste */
651 nkey |= (pgste_val(pgste) & (RCP_ACC_BITS | RCP_FP_BIT)) >> 56;
652 if (okey != nkey)
653 page_set_storage_key(address, nkey, 1);
598#endif 654#endif
599} 655}
600 656
601/* 657/*
658 * Certain architectures need to do special things when PTEs
659 * within a page table are directly modified. Thus, the following
660 * hook is made available.
661 */
662static inline void set_pte_at(struct mm_struct *mm, unsigned long addr,
663 pte_t *ptep, pte_t entry)
664{
665 pgste_t pgste;
666
667 if (mm_has_pgste(mm)) {
668 pgste = pgste_get_lock(ptep);
669 pgste_set_pte(ptep, pgste);
670 *ptep = entry;
671 pgste_set_unlock(ptep, pgste);
672 } else
673 *ptep = entry;
674}
675
676/*
602 * query functions pte_write/pte_dirty/pte_young only work if 677 * query functions pte_write/pte_dirty/pte_young only work if
603 * pte_present() is true. Undefined behaviour if not.. 678 * pte_present() is true. Undefined behaviour if not..
604 */ 679 */
@@ -609,19 +684,19 @@ static inline int pte_write(pte_t pte)
609 684
610static inline int pte_dirty(pte_t pte) 685static inline int pte_dirty(pte_t pte)
611{ 686{
612 /* A pte is neither clean nor dirty on s/390. The dirty bit 687#ifdef CONFIG_PGSTE
613 * is in the storage key. See page_test_and_clear_dirty for 688 if (pte_val(pte) & _PAGE_SWC)
614 * details. 689 return 1;
615 */ 690#endif
616 return 0; 691 return 0;
617} 692}
618 693
619static inline int pte_young(pte_t pte) 694static inline int pte_young(pte_t pte)
620{ 695{
621 /* A pte is neither young nor old on s/390. The young bit 696#ifdef CONFIG_PGSTE
622 * is in the storage key. See page_test_and_clear_young for 697 if (pte_val(pte) & _PAGE_SWR)
623 * details. 698 return 1;
624 */ 699#endif
625 return 0; 700 return 0;
626} 701}
627 702
@@ -629,64 +704,30 @@ static inline int pte_young(pte_t pte)
629 * pgd/pmd/pte modification functions 704 * pgd/pmd/pte modification functions
630 */ 705 */
631 706
632#ifndef __s390x__ 707static inline void pgd_clear(pgd_t *pgd)
633
634#define pgd_clear(pgd) do { } while (0)
635#define pud_clear(pud) do { } while (0)
636
637#else /* __s390x__ */
638
639static inline void pgd_clear_kernel(pgd_t * pgd)
640{ 708{
709#ifdef __s390x__
641 if ((pgd_val(*pgd) & _REGION_ENTRY_TYPE_MASK) == _REGION_ENTRY_TYPE_R2) 710 if ((pgd_val(*pgd) & _REGION_ENTRY_TYPE_MASK) == _REGION_ENTRY_TYPE_R2)
642 pgd_val(*pgd) = _REGION2_ENTRY_EMPTY; 711 pgd_val(*pgd) = _REGION2_ENTRY_EMPTY;
712#endif
643} 713}
644 714
645static inline void pgd_clear(pgd_t * pgd) 715static inline void pud_clear(pud_t *pud)
646{
647 pgd_t *shadow = get_shadow_table(pgd);
648
649 pgd_clear_kernel(pgd);
650 if (shadow)
651 pgd_clear_kernel(shadow);
652}
653
654static inline void pud_clear_kernel(pud_t *pud)
655{ 716{
717#ifdef __s390x__
656 if ((pud_val(*pud) & _REGION_ENTRY_TYPE_MASK) == _REGION_ENTRY_TYPE_R3) 718 if ((pud_val(*pud) & _REGION_ENTRY_TYPE_MASK) == _REGION_ENTRY_TYPE_R3)
657 pud_val(*pud) = _REGION3_ENTRY_EMPTY; 719 pud_val(*pud) = _REGION3_ENTRY_EMPTY;
720#endif
658} 721}
659 722
660static inline void pud_clear(pud_t *pud) 723static inline void pmd_clear(pmd_t *pmdp)
661{
662 pud_t *shadow = get_shadow_table(pud);
663
664 pud_clear_kernel(pud);
665 if (shadow)
666 pud_clear_kernel(shadow);
667}
668
669#endif /* __s390x__ */
670
671static inline void pmd_clear_kernel(pmd_t * pmdp)
672{ 724{
673 pmd_val(*pmdp) = _SEGMENT_ENTRY_EMPTY; 725 pmd_val(*pmdp) = _SEGMENT_ENTRY_EMPTY;
674} 726}
675 727
676static inline void pmd_clear(pmd_t *pmd)
677{
678 pmd_t *shadow = get_shadow_table(pmd);
679
680 pmd_clear_kernel(pmd);
681 if (shadow)
682 pmd_clear_kernel(shadow);
683}
684
685static inline void pte_clear(struct mm_struct *mm, unsigned long addr, pte_t *ptep) 728static inline void pte_clear(struct mm_struct *mm, unsigned long addr, pte_t *ptep)
686{ 729{
687 pte_val(*ptep) = _PAGE_TYPE_EMPTY; 730 pte_val(*ptep) = _PAGE_TYPE_EMPTY;
688 if (mm->context.noexec)
689 pte_val(ptep[PTRS_PER_PTE]) = _PAGE_TYPE_EMPTY;
690} 731}
691 732
692/* 733/*
@@ -716,35 +757,27 @@ static inline pte_t pte_mkwrite(pte_t pte)
716 757
717static inline pte_t pte_mkclean(pte_t pte) 758static inline pte_t pte_mkclean(pte_t pte)
718{ 759{
719 /* The only user of pte_mkclean is the fork() code. 760#ifdef CONFIG_PGSTE
720 We must *not* clear the *physical* page dirty bit 761 pte_val(pte) &= ~_PAGE_SWC;
721 just because fork() wants to clear the dirty bit in 762#endif
722 *one* of the page's mappings. So we just do nothing. */
723 return pte; 763 return pte;
724} 764}
725 765
726static inline pte_t pte_mkdirty(pte_t pte) 766static inline pte_t pte_mkdirty(pte_t pte)
727{ 767{
728 /* We do not explicitly set the dirty bit because the
729 * sske instruction is slow. It is faster to let the
730 * next instruction set the dirty bit.
731 */
732 return pte; 768 return pte;
733} 769}
734 770
735static inline pte_t pte_mkold(pte_t pte) 771static inline pte_t pte_mkold(pte_t pte)
736{ 772{
737 /* S/390 doesn't keep its dirty/referenced bit in the pte. 773#ifdef CONFIG_PGSTE
738 * There is no point in clearing the real referenced bit. 774 pte_val(pte) &= ~_PAGE_SWR;
739 */ 775#endif
740 return pte; 776 return pte;
741} 777}
742 778
743static inline pte_t pte_mkyoung(pte_t pte) 779static inline pte_t pte_mkyoung(pte_t pte)
744{ 780{
745 /* S/390 doesn't keep its dirty/referenced bit in the pte.
746 * There is no point in setting the real referenced bit.
747 */
748 return pte; 781 return pte;
749} 782}
750 783
@@ -754,62 +787,88 @@ static inline pte_t pte_mkspecial(pte_t pte)
754 return pte; 787 return pte;
755} 788}
756 789
757#ifdef CONFIG_PGSTE 790#ifdef CONFIG_HUGETLB_PAGE
791static inline pte_t pte_mkhuge(pte_t pte)
792{
793 /*
794 * PROT_NONE needs to be remapped from the pte type to the ste type.
795 * The HW invalid bit is also different for pte and ste. The pte
796 * invalid bit happens to be the same as the ste _SEGMENT_ENTRY_LARGE
797 * bit, so we don't have to clear it.
798 */
799 if (pte_val(pte) & _PAGE_INVALID) {
800 if (pte_val(pte) & _PAGE_SWT)
801 pte_val(pte) |= _HPAGE_TYPE_NONE;
802 pte_val(pte) |= _SEGMENT_ENTRY_INV;
803 }
804 /*
805 * Clear SW pte bits SWT and SWX, there are no SW bits in a segment
806 * table entry.
807 */
808 pte_val(pte) &= ~(_PAGE_SWT | _PAGE_SWX);
809 /*
810 * Also set the change-override bit because we don't need dirty bit
811 * tracking for hugetlbfs pages.
812 */
813 pte_val(pte) |= (_SEGMENT_ENTRY_LARGE | _SEGMENT_ENTRY_CO);
814 return pte;
815}
816#endif
817
758/* 818/*
759 * Get (and clear) the user dirty bit for a PTE. 819 * Get (and clear) the user dirty bit for a pte.
760 */ 820 */
761static inline int kvm_s390_test_and_clear_page_dirty(struct mm_struct *mm, 821static inline int ptep_test_and_clear_user_dirty(struct mm_struct *mm,
762 pte_t *ptep) 822 pte_t *ptep)
763{ 823{
764 int dirty; 824 pgste_t pgste;
765 unsigned long *pgste; 825 int dirty = 0;
766 struct page *page; 826
767 unsigned int skey; 827 if (mm_has_pgste(mm)) {
768 828 pgste = pgste_get_lock(ptep);
769 if (!mm->context.has_pgste) 829 pgste = pgste_update_all(ptep, pgste);
770 return -EINVAL; 830 dirty = !!(pgste_val(pgste) & KVM_UC_BIT);
771 rcp_lock(ptep); 831 pgste_val(pgste) &= ~KVM_UC_BIT;
772 pgste = (unsigned long *) (ptep + PTRS_PER_PTE); 832 pgste_set_unlock(ptep, pgste);
773 page = virt_to_page(pte_val(*ptep)); 833 return dirty;
774 skey = page_get_storage_key(page_to_phys(page));
775 if (skey & _PAGE_CHANGED) {
776 set_bit_simple(RCP_GC_BIT, pgste);
777 set_bit_simple(KVM_UD_BIT, pgste);
778 } 834 }
779 if (test_and_clear_bit_simple(RCP_HC_BIT, pgste)) {
780 SetPageDirty(page);
781 set_bit_simple(KVM_UD_BIT, pgste);
782 }
783 dirty = test_and_clear_bit_simple(KVM_UD_BIT, pgste);
784 if (skey & _PAGE_CHANGED)
785 page_clear_dirty(page);
786 rcp_unlock(ptep);
787 return dirty; 835 return dirty;
788} 836}
789#endif 837
838/*
839 * Get (and clear) the user referenced bit for a pte.
840 */
841static inline int ptep_test_and_clear_user_young(struct mm_struct *mm,
842 pte_t *ptep)
843{
844 pgste_t pgste;
845 int young = 0;
846
847 if (mm_has_pgste(mm)) {
848 pgste = pgste_get_lock(ptep);
849 pgste = pgste_update_young(ptep, pgste);
850 young = !!(pgste_val(pgste) & KVM_UR_BIT);
851 pgste_val(pgste) &= ~KVM_UR_BIT;
852 pgste_set_unlock(ptep, pgste);
853 }
854 return young;
855}
790 856
791#define __HAVE_ARCH_PTEP_TEST_AND_CLEAR_YOUNG 857#define __HAVE_ARCH_PTEP_TEST_AND_CLEAR_YOUNG
792static inline int ptep_test_and_clear_young(struct vm_area_struct *vma, 858static inline int ptep_test_and_clear_young(struct vm_area_struct *vma,
793 unsigned long addr, pte_t *ptep) 859 unsigned long addr, pte_t *ptep)
794{ 860{
795#ifdef CONFIG_PGSTE 861 pgste_t pgste;
796 unsigned long physpage; 862 pte_t pte;
797 int young;
798 unsigned long *pgste;
799 863
800 if (!vma->vm_mm->context.has_pgste) 864 if (mm_has_pgste(vma->vm_mm)) {
801 return 0; 865 pgste = pgste_get_lock(ptep);
802 physpage = pte_val(*ptep) & PAGE_MASK; 866 pgste = pgste_update_young(ptep, pgste);
803 pgste = (unsigned long *) (ptep + PTRS_PER_PTE); 867 pte = *ptep;
804 868 *ptep = pte_mkold(pte);
805 young = ((page_get_storage_key(physpage) & _PAGE_REFERENCED) != 0); 869 pgste_set_unlock(ptep, pgste);
806 rcp_lock(ptep); 870 return pte_young(pte);
807 if (young) 871 }
808 set_bit_simple(RCP_GR_BIT, pgste);
809 young |= test_and_clear_bit_simple(RCP_HR_BIT, pgste);
810 rcp_unlock(ptep);
811 return young;
812#endif
813 return 0; 872 return 0;
814} 873}
815 874
@@ -821,10 +880,7 @@ static inline int ptep_clear_flush_young(struct vm_area_struct *vma,
821 * On s390 reference bits are in storage key and never in TLB 880 * On s390 reference bits are in storage key and never in TLB
822 * With virtualization we handle the reference bit, without we 881 * With virtualization we handle the reference bit, without we
823 * we can simply return */ 882 * we can simply return */
824#ifdef CONFIG_PGSTE
825 return ptep_test_and_clear_young(vma, address, ptep); 883 return ptep_test_and_clear_young(vma, address, ptep);
826#endif
827 return 0;
828} 884}
829 885
830static inline void __ptep_ipte(unsigned long address, pte_t *ptep) 886static inline void __ptep_ipte(unsigned long address, pte_t *ptep)
@@ -844,25 +900,6 @@ static inline void __ptep_ipte(unsigned long address, pte_t *ptep)
844 } 900 }
845} 901}
846 902
847static inline void ptep_invalidate(struct mm_struct *mm,
848 unsigned long address, pte_t *ptep)
849{
850 if (mm->context.has_pgste) {
851 rcp_lock(ptep);
852 __ptep_ipte(address, ptep);
853 ptep_rcp_copy(ptep);
854 pte_val(*ptep) = _PAGE_TYPE_EMPTY;
855 rcp_unlock(ptep);
856 return;
857 }
858 __ptep_ipte(address, ptep);
859 pte_val(*ptep) = _PAGE_TYPE_EMPTY;
860 if (mm->context.noexec) {
861 __ptep_ipte(address, ptep + PTRS_PER_PTE);
862 pte_val(*(ptep + PTRS_PER_PTE)) = _PAGE_TYPE_EMPTY;
863 }
864}
865
866/* 903/*
867 * This is hard to understand. ptep_get_and_clear and ptep_clear_flush 904 * This is hard to understand. ptep_get_and_clear and ptep_clear_flush
868 * both clear the TLB for the unmapped pte. The reason is that 905 * both clear the TLB for the unmapped pte. The reason is that
@@ -877,24 +914,72 @@ static inline void ptep_invalidate(struct mm_struct *mm,
877 * is a nop. 914 * is a nop.
878 */ 915 */
879#define __HAVE_ARCH_PTEP_GET_AND_CLEAR 916#define __HAVE_ARCH_PTEP_GET_AND_CLEAR
880#define ptep_get_and_clear(__mm, __address, __ptep) \ 917static inline pte_t ptep_get_and_clear(struct mm_struct *mm,
881({ \ 918 unsigned long address, pte_t *ptep)
882 pte_t __pte = *(__ptep); \ 919{
883 (__mm)->context.flush_mm = 1; \ 920 pgste_t pgste;
884 if (atomic_read(&(__mm)->context.attach_count) > 1 || \ 921 pte_t pte;
885 (__mm) != current->active_mm) \ 922
886 ptep_invalidate(__mm, __address, __ptep); \ 923 mm->context.flush_mm = 1;
887 else \ 924 if (mm_has_pgste(mm))
888 pte_clear((__mm), (__address), (__ptep)); \ 925 pgste = pgste_get_lock(ptep);
889 __pte; \ 926
890}) 927 pte = *ptep;
928 if (!mm_exclusive(mm))
929 __ptep_ipte(address, ptep);
930 pte_val(*ptep) = _PAGE_TYPE_EMPTY;
931
932 if (mm_has_pgste(mm)) {
933 pgste = pgste_update_all(&pte, pgste);
934 pgste_set_unlock(ptep, pgste);
935 }
936 return pte;
937}
938
939#define __HAVE_ARCH_PTEP_MODIFY_PROT_TRANSACTION
940static inline pte_t ptep_modify_prot_start(struct mm_struct *mm,
941 unsigned long address,
942 pte_t *ptep)
943{
944 pte_t pte;
945
946 mm->context.flush_mm = 1;
947 if (mm_has_pgste(mm))
948 pgste_get_lock(ptep);
949
950 pte = *ptep;
951 if (!mm_exclusive(mm))
952 __ptep_ipte(address, ptep);
953 return pte;
954}
955
956static inline void ptep_modify_prot_commit(struct mm_struct *mm,
957 unsigned long address,
958 pte_t *ptep, pte_t pte)
959{
960 *ptep = pte;
961 if (mm_has_pgste(mm))
962 pgste_set_unlock(ptep, *(pgste_t *)(ptep + PTRS_PER_PTE));
963}
891 964
892#define __HAVE_ARCH_PTEP_CLEAR_FLUSH 965#define __HAVE_ARCH_PTEP_CLEAR_FLUSH
893static inline pte_t ptep_clear_flush(struct vm_area_struct *vma, 966static inline pte_t ptep_clear_flush(struct vm_area_struct *vma,
894 unsigned long address, pte_t *ptep) 967 unsigned long address, pte_t *ptep)
895{ 968{
896 pte_t pte = *ptep; 969 pgste_t pgste;
897 ptep_invalidate(vma->vm_mm, address, ptep); 970 pte_t pte;
971
972 if (mm_has_pgste(vma->vm_mm))
973 pgste = pgste_get_lock(ptep);
974
975 pte = *ptep;
976 __ptep_ipte(address, ptep);
977 pte_val(*ptep) = _PAGE_TYPE_EMPTY;
978
979 if (mm_has_pgste(vma->vm_mm)) {
980 pgste = pgste_update_all(&pte, pgste);
981 pgste_set_unlock(ptep, pgste);
982 }
898 return pte; 983 return pte;
899} 984}
900 985
@@ -907,76 +992,67 @@ static inline pte_t ptep_clear_flush(struct vm_area_struct *vma,
907 */ 992 */
908#define __HAVE_ARCH_PTEP_GET_AND_CLEAR_FULL 993#define __HAVE_ARCH_PTEP_GET_AND_CLEAR_FULL
909static inline pte_t ptep_get_and_clear_full(struct mm_struct *mm, 994static inline pte_t ptep_get_and_clear_full(struct mm_struct *mm,
910 unsigned long addr, 995 unsigned long address,
911 pte_t *ptep, int full) 996 pte_t *ptep, int full)
912{ 997{
913 pte_t pte = *ptep; 998 pgste_t pgste;
999 pte_t pte;
1000
1001 if (mm_has_pgste(mm))
1002 pgste = pgste_get_lock(ptep);
914 1003
915 if (full) 1004 pte = *ptep;
916 pte_clear(mm, addr, ptep); 1005 if (!full)
917 else 1006 __ptep_ipte(address, ptep);
918 ptep_invalidate(mm, addr, ptep); 1007 pte_val(*ptep) = _PAGE_TYPE_EMPTY;
1008
1009 if (mm_has_pgste(mm)) {
1010 pgste = pgste_update_all(&pte, pgste);
1011 pgste_set_unlock(ptep, pgste);
1012 }
919 return pte; 1013 return pte;
920} 1014}
921 1015
922#define __HAVE_ARCH_PTEP_SET_WRPROTECT 1016#define __HAVE_ARCH_PTEP_SET_WRPROTECT
923#define ptep_set_wrprotect(__mm, __addr, __ptep) \ 1017static inline pte_t ptep_set_wrprotect(struct mm_struct *mm,
924({ \ 1018 unsigned long address, pte_t *ptep)
925 pte_t __pte = *(__ptep); \ 1019{
926 if (pte_write(__pte)) { \ 1020 pgste_t pgste;
927 (__mm)->context.flush_mm = 1; \ 1021 pte_t pte = *ptep;
928 if (atomic_read(&(__mm)->context.attach_count) > 1 || \
929 (__mm) != current->active_mm) \
930 ptep_invalidate(__mm, __addr, __ptep); \
931 set_pte_at(__mm, __addr, __ptep, pte_wrprotect(__pte)); \
932 } \
933})
934 1022
935#define __HAVE_ARCH_PTEP_SET_ACCESS_FLAGS 1023 if (pte_write(pte)) {
936#define ptep_set_access_flags(__vma, __addr, __ptep, __entry, __dirty) \ 1024 mm->context.flush_mm = 1;
937({ \ 1025 if (mm_has_pgste(mm))
938 int __changed = !pte_same(*(__ptep), __entry); \ 1026 pgste = pgste_get_lock(ptep);
939 if (__changed) { \
940 ptep_invalidate((__vma)->vm_mm, __addr, __ptep); \
941 set_pte_at((__vma)->vm_mm, __addr, __ptep, __entry); \
942 } \
943 __changed; \
944})
945 1027
946/* 1028 if (!mm_exclusive(mm))
947 * Test and clear dirty bit in storage key. 1029 __ptep_ipte(address, ptep);
948 * We can't clear the changed bit atomically. This is a potential 1030 *ptep = pte_wrprotect(pte);
949 * race against modification of the referenced bit. This function
950 * should therefore only be called if it is not mapped in any
951 * address space.
952 */
953#define __HAVE_ARCH_PAGE_TEST_DIRTY
954static inline int page_test_dirty(struct page *page)
955{
956 return (page_get_storage_key(page_to_phys(page)) & _PAGE_CHANGED) != 0;
957}
958 1031
959#define __HAVE_ARCH_PAGE_CLEAR_DIRTY 1032 if (mm_has_pgste(mm))
960static inline void page_clear_dirty(struct page *page) 1033 pgste_set_unlock(ptep, pgste);
961{ 1034 }
962 page_set_storage_key(page_to_phys(page), PAGE_DEFAULT_KEY); 1035 return pte;
963} 1036}
964 1037
965/* 1038#define __HAVE_ARCH_PTEP_SET_ACCESS_FLAGS
966 * Test and clear referenced bit in storage key. 1039static inline int ptep_set_access_flags(struct vm_area_struct *vma,
967 */ 1040 unsigned long address, pte_t *ptep,
968#define __HAVE_ARCH_PAGE_TEST_AND_CLEAR_YOUNG 1041 pte_t entry, int dirty)
969static inline int page_test_and_clear_young(struct page *page)
970{ 1042{
971 unsigned long physpage = page_to_phys(page); 1043 pgste_t pgste;
972 int ccode; 1044
973 1045 if (pte_same(*ptep, entry))
974 asm volatile( 1046 return 0;
975 " rrbe 0,%1\n" 1047 if (mm_has_pgste(vma->vm_mm))
976 " ipm %0\n" 1048 pgste = pgste_get_lock(ptep);
977 " srl %0,28\n" 1049
978 : "=d" (ccode) : "a" (physpage) : "cc" ); 1050 __ptep_ipte(address, ptep);
979 return ccode & 2; 1051 *ptep = entry;
1052
1053 if (mm_has_pgste(vma->vm_mm))
1054 pgste_set_unlock(ptep, pgste);
1055 return 1;
980} 1056}
981 1057
982/* 1058/*
@@ -1048,9 +1124,7 @@ static inline pmd_t *pmd_offset(pud_t *pud, unsigned long address)
1048#define pte_offset(pmd, addr) ((pte_t *) pmd_deref(*(pmd)) + pte_index(addr)) 1124#define pte_offset(pmd, addr) ((pte_t *) pmd_deref(*(pmd)) + pte_index(addr))
1049#define pte_offset_kernel(pmd, address) pte_offset(pmd,address) 1125#define pte_offset_kernel(pmd, address) pte_offset(pmd,address)
1050#define pte_offset_map(pmd, address) pte_offset_kernel(pmd, address) 1126#define pte_offset_map(pmd, address) pte_offset_kernel(pmd, address)
1051#define pte_offset_map_nested(pmd, address) pte_offset_kernel(pmd, address)
1052#define pte_unmap(pte) do { } while (0) 1127#define pte_unmap(pte) do { } while (0)
1053#define pte_unmap_nested(pte) do { } while (0)
1054 1128
1055/* 1129/*
1056 * 31 bit swap entry format: 1130 * 31 bit swap entry format:
diff --git a/arch/s390/include/asm/processor.h b/arch/s390/include/asm/processor.h
index 73e259834e10..1300c3025334 100644
--- a/arch/s390/include/asm/processor.h
+++ b/arch/s390/include/asm/processor.h
@@ -32,7 +32,6 @@ static inline void get_cpu_id(struct cpuid *ptr)
32} 32}
33 33
34extern void s390_adjust_jiffies(void); 34extern void s390_adjust_jiffies(void);
35extern void print_cpu_info(void);
36extern int get_cpu_capability(unsigned int *); 35extern int get_cpu_capability(unsigned int *);
37 36
38/* 37/*
@@ -81,11 +80,11 @@ struct thread_struct {
81 mm_segment_t mm_segment; 80 mm_segment_t mm_segment;
82 unsigned long prot_addr; /* address of protection-excep. */ 81 unsigned long prot_addr; /* address of protection-excep. */
83 unsigned int trap_no; 82 unsigned int trap_no;
84 per_struct per_info; 83 struct per_regs per_user; /* User specified PER registers */
85 /* Used to give failing instruction back to user for ieee exceptions */ 84 struct per_event per_event; /* Cause of the last PER trap */
86 unsigned long ieee_instruction_pointer;
87 /* pfault_wait is used to block the process on a pfault event */ 85 /* pfault_wait is used to block the process on a pfault event */
88 unsigned long pfault_wait; 86 unsigned long pfault_wait;
87 struct list_head list;
89}; 88};
90 89
91typedef struct thread_struct thread_struct; 90typedef struct thread_struct thread_struct;
@@ -150,11 +149,6 @@ extern int kernel_thread(int (*fn)(void *), void * arg, unsigned long flags);
150 */ 149 */
151extern unsigned long thread_saved_pc(struct task_struct *t); 150extern unsigned long thread_saved_pc(struct task_struct *t);
152 151
153/*
154 * Print register of task into buffer. Used in fs/proc/array.c.
155 */
156extern void task_show_regs(struct seq_file *m, struct task_struct *task);
157
158extern void show_code(struct pt_regs *regs); 152extern void show_code(struct pt_regs *regs);
159 153
160unsigned long get_wchan(struct task_struct *p); 154unsigned long get_wchan(struct task_struct *p);
diff --git a/arch/s390/include/asm/ptrace.h b/arch/s390/include/asm/ptrace.h
index e2c218dc68a6..9ad628a8574a 100644
--- a/arch/s390/include/asm/ptrace.h
+++ b/arch/s390/include/asm/ptrace.h
@@ -331,10 +331,60 @@ struct pt_regs
331 unsigned short ilc; 331 unsigned short ilc;
332 unsigned short svcnr; 332 unsigned short svcnr;
333}; 333};
334
335/*
336 * Program event recording (PER) register set.
337 */
338struct per_regs {
339 unsigned long control; /* PER control bits */
340 unsigned long start; /* PER starting address */
341 unsigned long end; /* PER ending address */
342};
343
344/*
345 * PER event contains information about the cause of the last PER exception.
346 */
347struct per_event {
348 unsigned short cause; /* PER code, ATMID and AI */
349 unsigned long address; /* PER address */
350 unsigned char paid; /* PER access identification */
351};
352
353/*
354 * Simplified per_info structure used to decode the ptrace user space ABI.
355 */
356struct per_struct_kernel {
357 unsigned long cr9; /* PER control bits */
358 unsigned long cr10; /* PER starting address */
359 unsigned long cr11; /* PER ending address */
360 unsigned long bits; /* Obsolete software bits */
361 unsigned long starting_addr; /* User specified start address */
362 unsigned long ending_addr; /* User specified end address */
363 unsigned short perc_atmid; /* PER trap ATMID */
364 unsigned long address; /* PER trap instruction address */
365 unsigned char access_id; /* PER trap access identification */
366};
367
368#define PER_EVENT_MASK 0xE9000000UL
369
370#define PER_EVENT_BRANCH 0x80000000UL
371#define PER_EVENT_IFETCH 0x40000000UL
372#define PER_EVENT_STORE 0x20000000UL
373#define PER_EVENT_STORE_REAL 0x08000000UL
374#define PER_EVENT_NULLIFICATION 0x01000000UL
375
376#define PER_CONTROL_MASK 0x00a00000UL
377
378#define PER_CONTROL_BRANCH_ADDRESS 0x00800000UL
379#define PER_CONTROL_ALTERATION 0x00200000UL
380
334#endif 381#endif
335 382
336/* 383/*
337 * Now for the program event recording (trace) definitions. 384 * Now for the user space program event recording (trace) definitions.
385 * The following structures are used only for the ptrace interface, don't
386 * touch or even look at it if you don't want to modify the user-space
387 * ptrace interface. In particular stay away from it for in-kernel PER.
338 */ 388 */
339typedef struct 389typedef struct
340{ 390{
@@ -481,8 +531,7 @@ struct user_regs_struct
481 * watchpoints. This is the way intel does it. 531 * watchpoints. This is the way intel does it.
482 */ 532 */
483 per_struct per_info; 533 per_struct per_info;
484 unsigned long ieee_instruction_pointer; 534 unsigned long ieee_instruction_pointer; /* obsolete, always 0 */
485 /* Used to give failing instruction back to user for ieee exceptions */
486}; 535};
487 536
488#ifdef __KERNEL__ 537#ifdef __KERNEL__
diff --git a/arch/s390/include/asm/qdio.h b/arch/s390/include/asm/qdio.h
index 2ba630276295..15c97625df8d 100644
--- a/arch/s390/include/asm/qdio.h
+++ b/arch/s390/include/asm/qdio.h
@@ -139,110 +139,47 @@ struct slib {
139 struct slibe slibe[QDIO_MAX_BUFFERS_PER_Q]; 139 struct slibe slibe[QDIO_MAX_BUFFERS_PER_Q];
140} __attribute__ ((packed, aligned(2048))); 140} __attribute__ ((packed, aligned(2048)));
141 141
142/** 142#define SBAL_EFLAGS_LAST_ENTRY 0x40
143 * struct sbal_flags - storage block address list flags 143#define SBAL_EFLAGS_CONTIGUOUS 0x20
144 * @last: last entry 144#define SBAL_EFLAGS_FIRST_FRAG 0x04
145 * @cont: contiguous storage 145#define SBAL_EFLAGS_MIDDLE_FRAG 0x08
146 * @frag: fragmentation 146#define SBAL_EFLAGS_LAST_FRAG 0x0c
147 */ 147#define SBAL_EFLAGS_MASK 0x6f
148struct sbal_flags {
149 u8 : 1;
150 u8 last : 1;
151 u8 cont : 1;
152 u8 : 1;
153 u8 frag : 2;
154 u8 : 2;
155} __attribute__ ((packed));
156
157#define SBAL_FLAGS_FIRST_FRAG 0x04000000UL
158#define SBAL_FLAGS_MIDDLE_FRAG 0x08000000UL
159#define SBAL_FLAGS_LAST_FRAG 0x0c000000UL
160#define SBAL_FLAGS_LAST_ENTRY 0x40000000UL
161#define SBAL_FLAGS_CONTIGUOUS 0x20000000UL
162 148
163#define SBAL_FLAGS0_DATA_CONTINUATION 0x20UL 149#define SBAL_SFLAGS0_PCI_REQ 0x40
150#define SBAL_SFLAGS0_DATA_CONTINUATION 0x20
164 151
165/* Awesome OpenFCP extensions */ 152/* Awesome OpenFCP extensions */
166#define SBAL_FLAGS0_TYPE_STATUS 0x00UL 153#define SBAL_SFLAGS0_TYPE_STATUS 0x00
167#define SBAL_FLAGS0_TYPE_WRITE 0x08UL 154#define SBAL_SFLAGS0_TYPE_WRITE 0x08
168#define SBAL_FLAGS0_TYPE_READ 0x10UL 155#define SBAL_SFLAGS0_TYPE_READ 0x10
169#define SBAL_FLAGS0_TYPE_WRITE_READ 0x18UL 156#define SBAL_SFLAGS0_TYPE_WRITE_READ 0x18
170#define SBAL_FLAGS0_MORE_SBALS 0x04UL 157#define SBAL_SFLAGS0_MORE_SBALS 0x04
171#define SBAL_FLAGS0_COMMAND 0x02UL 158#define SBAL_SFLAGS0_COMMAND 0x02
172#define SBAL_FLAGS0_LAST_SBAL 0x00UL 159#define SBAL_SFLAGS0_LAST_SBAL 0x00
173#define SBAL_FLAGS0_ONLY_SBAL SBAL_FLAGS0_COMMAND 160#define SBAL_SFLAGS0_ONLY_SBAL SBAL_SFLAGS0_COMMAND
174#define SBAL_FLAGS0_MIDDLE_SBAL SBAL_FLAGS0_MORE_SBALS 161#define SBAL_SFLAGS0_MIDDLE_SBAL SBAL_SFLAGS0_MORE_SBALS
175#define SBAL_FLAGS0_FIRST_SBAL SBAL_FLAGS0_MORE_SBALS | SBAL_FLAGS0_COMMAND 162#define SBAL_SFLAGS0_FIRST_SBAL (SBAL_SFLAGS0_MORE_SBALS | SBAL_SFLAGS0_COMMAND)
176#define SBAL_FLAGS0_PCI 0x40
177
178/**
179 * struct sbal_sbalf_0 - sbal flags for sbale 0
180 * @pci: PCI indicator
181 * @cont: data continuation
182 * @sbtype: storage-block type (FCP)
183 */
184struct sbal_sbalf_0 {
185 u8 : 1;
186 u8 pci : 1;
187 u8 cont : 1;
188 u8 sbtype : 2;
189 u8 : 3;
190} __attribute__ ((packed));
191
192/**
193 * struct sbal_sbalf_1 - sbal flags for sbale 1
194 * @key: storage key
195 */
196struct sbal_sbalf_1 {
197 u8 : 4;
198 u8 key : 4;
199} __attribute__ ((packed));
200
201/**
202 * struct sbal_sbalf_14 - sbal flags for sbale 14
203 * @erridx: error index
204 */
205struct sbal_sbalf_14 {
206 u8 : 4;
207 u8 erridx : 4;
208} __attribute__ ((packed));
209
210/**
211 * struct sbal_sbalf_15 - sbal flags for sbale 15
212 * @reason: reason for error state
213 */
214struct sbal_sbalf_15 {
215 u8 reason;
216} __attribute__ ((packed));
217
218/**
219 * union sbal_sbalf - storage block address list flags
220 * @i0: sbalf0
221 * @i1: sbalf1
222 * @i14: sbalf14
223 * @i15: sblaf15
224 * @value: raw value
225 */
226union sbal_sbalf {
227 struct sbal_sbalf_0 i0;
228 struct sbal_sbalf_1 i1;
229 struct sbal_sbalf_14 i14;
230 struct sbal_sbalf_15 i15;
231 u8 value;
232};
233 163
234/** 164/**
235 * struct qdio_buffer_element - SBAL entry 165 * struct qdio_buffer_element - SBAL entry
236 * @flags: flags 166 * @eflags: SBAL entry flags
167 * @scount: SBAL count
168 * @sflags: whole SBAL flags
237 * @length: length 169 * @length: length
238 * @addr: address 170 * @addr: address
239*/ 171*/
240struct qdio_buffer_element { 172struct qdio_buffer_element {
241 u32 flags; 173 u8 eflags;
174 /* private: */
175 u8 res1;
176 /* public: */
177 u8 scount;
178 u8 sflags;
242 u32 length; 179 u32 length;
243#ifdef CONFIG_32BIT 180#ifdef CONFIG_32BIT
244 /* private: */ 181 /* private: */
245 void *reserved; 182 void *res2;
246 /* public: */ 183 /* public: */
247#endif 184#endif
248 void *addr; 185 void *addr;
@@ -360,6 +297,8 @@ struct qdio_initialize {
360 unsigned int no_output_qs; 297 unsigned int no_output_qs;
361 qdio_handler_t *input_handler; 298 qdio_handler_t *input_handler;
362 qdio_handler_t *output_handler; 299 qdio_handler_t *output_handler;
300 void (*queue_start_poll) (struct ccw_device *, int, unsigned long);
301 int scan_threshold;
363 unsigned long int_parm; 302 unsigned long int_parm;
364 void **input_sbal_addr_array; 303 void **input_sbal_addr_array;
365 void **output_sbal_addr_array; 304 void **output_sbal_addr_array;
@@ -377,11 +316,13 @@ struct qdio_initialize {
377extern int qdio_allocate(struct qdio_initialize *); 316extern int qdio_allocate(struct qdio_initialize *);
378extern int qdio_establish(struct qdio_initialize *); 317extern int qdio_establish(struct qdio_initialize *);
379extern int qdio_activate(struct ccw_device *); 318extern int qdio_activate(struct ccw_device *);
380 319extern int do_QDIO(struct ccw_device *, unsigned int, int, unsigned int,
381extern int do_QDIO(struct ccw_device *cdev, unsigned int callflags, 320 unsigned int);
382 int q_nr, unsigned int bufnr, unsigned int count); 321extern int qdio_start_irq(struct ccw_device *, int);
383extern int qdio_shutdown(struct ccw_device*, int); 322extern int qdio_stop_irq(struct ccw_device *, int);
323extern int qdio_get_next_buffers(struct ccw_device *, int, int *, int *);
324extern int qdio_shutdown(struct ccw_device *, int);
384extern int qdio_free(struct ccw_device *); 325extern int qdio_free(struct ccw_device *);
385extern int qdio_get_ssqd_desc(struct ccw_device *dev, struct qdio_ssqd_desc*); 326extern int qdio_get_ssqd_desc(struct ccw_device *, struct qdio_ssqd_desc *);
386 327
387#endif /* __QDIO_H__ */ 328#endif /* __QDIO_H__ */
diff --git a/arch/s390/include/asm/qeth.h b/arch/s390/include/asm/qeth.h
index 06cbd1e8c943..90efda0b137d 100644
--- a/arch/s390/include/asm/qeth.h
+++ b/arch/s390/include/asm/qeth.h
@@ -28,39 +28,70 @@ struct qeth_arp_cache_entry {
28 __u8 reserved2[32]; 28 __u8 reserved2[32];
29} __attribute__ ((packed)); 29} __attribute__ ((packed));
30 30
31enum qeth_arp_ipaddrtype {
32 QETHARP_IP_ADDR_V4 = 1,
33 QETHARP_IP_ADDR_V6 = 2,
34};
35struct qeth_arp_entrytype {
36 __u8 mac;
37 __u8 ip;
38} __attribute__((packed));
39
40#define QETH_QARP_MEDIASPECIFIC_BYTES 32
41#define QETH_QARP_MACADDRTYPE_BYTES 1
31struct qeth_arp_qi_entry7 { 42struct qeth_arp_qi_entry7 {
32 __u8 media_specific[32]; 43 __u8 media_specific[QETH_QARP_MEDIASPECIFIC_BYTES];
33 __u8 macaddr_type; 44 struct qeth_arp_entrytype type;
34 __u8 ipaddr_type;
35 __u8 macaddr[6]; 45 __u8 macaddr[6];
36 __u8 ipaddr[4]; 46 __u8 ipaddr[4];
37} __attribute__((packed)); 47} __attribute__((packed));
38 48
49struct qeth_arp_qi_entry7_ipv6 {
50 __u8 media_specific[QETH_QARP_MEDIASPECIFIC_BYTES];
51 struct qeth_arp_entrytype type;
52 __u8 macaddr[6];
53 __u8 ipaddr[16];
54} __attribute__((packed));
55
39struct qeth_arp_qi_entry7_short { 56struct qeth_arp_qi_entry7_short {
40 __u8 macaddr_type; 57 struct qeth_arp_entrytype type;
41 __u8 ipaddr_type;
42 __u8 macaddr[6]; 58 __u8 macaddr[6];
43 __u8 ipaddr[4]; 59 __u8 ipaddr[4];
44} __attribute__((packed)); 60} __attribute__((packed));
45 61
62struct qeth_arp_qi_entry7_short_ipv6 {
63 struct qeth_arp_entrytype type;
64 __u8 macaddr[6];
65 __u8 ipaddr[16];
66} __attribute__((packed));
67
46struct qeth_arp_qi_entry5 { 68struct qeth_arp_qi_entry5 {
47 __u8 media_specific[32]; 69 __u8 media_specific[QETH_QARP_MEDIASPECIFIC_BYTES];
48 __u8 macaddr_type; 70 struct qeth_arp_entrytype type;
49 __u8 ipaddr_type;
50 __u8 ipaddr[4]; 71 __u8 ipaddr[4];
51} __attribute__((packed)); 72} __attribute__((packed));
52 73
74struct qeth_arp_qi_entry5_ipv6 {
75 __u8 media_specific[QETH_QARP_MEDIASPECIFIC_BYTES];
76 struct qeth_arp_entrytype type;
77 __u8 ipaddr[16];
78} __attribute__((packed));
79
53struct qeth_arp_qi_entry5_short { 80struct qeth_arp_qi_entry5_short {
54 __u8 macaddr_type; 81 struct qeth_arp_entrytype type;
55 __u8 ipaddr_type;
56 __u8 ipaddr[4]; 82 __u8 ipaddr[4];
57} __attribute__((packed)); 83} __attribute__((packed));
58 84
85struct qeth_arp_qi_entry5_short_ipv6 {
86 struct qeth_arp_entrytype type;
87 __u8 ipaddr[16];
88} __attribute__((packed));
59/* 89/*
60 * can be set by user if no "media specific information" is wanted 90 * can be set by user if no "media specific information" is wanted
61 * -> saves a lot of space in user space buffer 91 * -> saves a lot of space in user space buffer
62 */ 92 */
63#define QETH_QARP_STRIP_ENTRIES 0x8000 93#define QETH_QARP_STRIP_ENTRIES 0x8000
94#define QETH_QARP_WITH_IPV6 0x4000
64#define QETH_QARP_REQUEST_MASK 0x00ff 95#define QETH_QARP_REQUEST_MASK 0x00ff
65 96
66/* data sent to user space as result of query arp ioctl */ 97/* data sent to user space as result of query arp ioctl */
diff --git a/arch/s390/include/asm/rwsem.h b/arch/s390/include/asm/rwsem.h
index 423fdda2322d..d0eb4653cebd 100644
--- a/arch/s390/include/asm/rwsem.h
+++ b/arch/s390/include/asm/rwsem.h
@@ -43,29 +43,6 @@
43 43
44#ifdef __KERNEL__ 44#ifdef __KERNEL__
45 45
46#include <linux/list.h>
47#include <linux/spinlock.h>
48
49struct rwsem_waiter;
50
51extern struct rw_semaphore *rwsem_down_read_failed(struct rw_semaphore *);
52extern struct rw_semaphore *rwsem_down_write_failed(struct rw_semaphore *);
53extern struct rw_semaphore *rwsem_wake(struct rw_semaphore *);
54extern struct rw_semaphore *rwsem_downgrade_wake(struct rw_semaphore *);
55extern struct rw_semaphore *rwsem_downgrade_write(struct rw_semaphore *);
56
57/*
58 * the semaphore definition
59 */
60struct rw_semaphore {
61 signed long count;
62 spinlock_t wait_lock;
63 struct list_head wait_list;
64#ifdef CONFIG_DEBUG_LOCK_ALLOC
65 struct lockdep_map dep_map;
66#endif
67};
68
69#ifndef __s390x__ 46#ifndef __s390x__
70#define RWSEM_UNLOCKED_VALUE 0x00000000 47#define RWSEM_UNLOCKED_VALUE 0x00000000
71#define RWSEM_ACTIVE_BIAS 0x00000001 48#define RWSEM_ACTIVE_BIAS 0x00000001
@@ -81,41 +58,6 @@ struct rw_semaphore {
81#define RWSEM_ACTIVE_WRITE_BIAS (RWSEM_WAITING_BIAS + RWSEM_ACTIVE_BIAS) 58#define RWSEM_ACTIVE_WRITE_BIAS (RWSEM_WAITING_BIAS + RWSEM_ACTIVE_BIAS)
82 59
83/* 60/*
84 * initialisation
85 */
86
87#ifdef CONFIG_DEBUG_LOCK_ALLOC
88# define __RWSEM_DEP_MAP_INIT(lockname) , .dep_map = { .name = #lockname }
89#else
90# define __RWSEM_DEP_MAP_INIT(lockname)
91#endif
92
93#define __RWSEM_INITIALIZER(name) \
94 { RWSEM_UNLOCKED_VALUE, __SPIN_LOCK_UNLOCKED((name).wait.lock), \
95 LIST_HEAD_INIT((name).wait_list) __RWSEM_DEP_MAP_INIT(name) }
96
97#define DECLARE_RWSEM(name) \
98 struct rw_semaphore name = __RWSEM_INITIALIZER(name)
99
100static inline void init_rwsem(struct rw_semaphore *sem)
101{
102 sem->count = RWSEM_UNLOCKED_VALUE;
103 spin_lock_init(&sem->wait_lock);
104 INIT_LIST_HEAD(&sem->wait_list);
105}
106
107extern void __init_rwsem(struct rw_semaphore *sem, const char *name,
108 struct lock_class_key *key);
109
110#define init_rwsem(sem) \
111do { \
112 static struct lock_class_key __key; \
113 \
114 __init_rwsem((sem), #sem, &__key); \
115} while (0)
116
117
118/*
119 * lock for reading 61 * lock for reading
120 */ 62 */
121static inline void __down_read(struct rw_semaphore *sem) 63static inline void __down_read(struct rw_semaphore *sem)
@@ -377,10 +319,5 @@ static inline long rwsem_atomic_update(long delta, struct rw_semaphore *sem)
377 return new; 319 return new;
378} 320}
379 321
380static inline int rwsem_is_locked(struct rw_semaphore *sem)
381{
382 return (sem->count != 0);
383}
384
385#endif /* __KERNEL__ */ 322#endif /* __KERNEL__ */
386#endif /* _S390_RWSEM_H */ 323#endif /* _S390_RWSEM_H */
diff --git a/arch/s390/include/asm/s390_ext.h b/arch/s390/include/asm/s390_ext.h
deleted file mode 100644
index 2afc060266a2..000000000000
--- a/arch/s390/include/asm/s390_ext.h
+++ /dev/null
@@ -1,32 +0,0 @@
1#ifndef _S390_EXTINT_H
2#define _S390_EXTINT_H
3
4/*
5 * include/asm-s390/s390_ext.h
6 *
7 * S390 version
8 * Copyright IBM Corp. 1999,2007
9 * Author(s): Holger Smolinski (Holger.Smolinski@de.ibm.com),
10 * Martin Schwidefsky (schwidefsky@de.ibm.com)
11 */
12
13#include <linux/types.h>
14
15typedef void (*ext_int_handler_t)(__u16 code);
16
17typedef struct ext_int_info_t {
18 struct ext_int_info_t *next;
19 ext_int_handler_t handler;
20 __u16 code;
21} ext_int_info_t;
22
23extern ext_int_info_t *ext_int_hash[];
24
25int register_external_interrupt(__u16 code, ext_int_handler_t handler);
26int register_early_external_interrupt(__u16 code, ext_int_handler_t handler,
27 ext_int_info_t *info);
28int unregister_external_interrupt(__u16 code, ext_int_handler_t handler);
29int unregister_early_external_interrupt(__u16 code, ext_int_handler_t handler,
30 ext_int_info_t *info);
31
32#endif
diff --git a/arch/s390/include/asm/scatterlist.h b/arch/s390/include/asm/scatterlist.h
index 35d786fe93ae..6d45ef6c12a7 100644
--- a/arch/s390/include/asm/scatterlist.h
+++ b/arch/s390/include/asm/scatterlist.h
@@ -1 +1,3 @@
1#include <asm-generic/scatterlist.h> 1#include <asm-generic/scatterlist.h>
2
3#define ARCH_HAS_SG_CHAIN
diff --git a/arch/s390/include/asm/setup.h b/arch/s390/include/asm/setup.h
index 25e831d58e1e..d5e2ef10537d 100644
--- a/arch/s390/include/asm/setup.h
+++ b/arch/s390/include/asm/setup.h
@@ -73,6 +73,7 @@ extern unsigned int user_mode;
73#define MACHINE_FLAG_PFMF (1UL << 11) 73#define MACHINE_FLAG_PFMF (1UL << 11)
74#define MACHINE_FLAG_LPAR (1UL << 12) 74#define MACHINE_FLAG_LPAR (1UL << 12)
75#define MACHINE_FLAG_SPP (1UL << 13) 75#define MACHINE_FLAG_SPP (1UL << 13)
76#define MACHINE_FLAG_TOPOLOGY (1UL << 14)
76 77
77#define MACHINE_IS_VM (S390_lowcore.machine_flags & MACHINE_FLAG_VM) 78#define MACHINE_IS_VM (S390_lowcore.machine_flags & MACHINE_FLAG_VM)
78#define MACHINE_IS_KVM (S390_lowcore.machine_flags & MACHINE_FLAG_KVM) 79#define MACHINE_IS_KVM (S390_lowcore.machine_flags & MACHINE_FLAG_KVM)
@@ -90,6 +91,7 @@ extern unsigned int user_mode;
90#define MACHINE_HAS_HPAGE (0) 91#define MACHINE_HAS_HPAGE (0)
91#define MACHINE_HAS_PFMF (0) 92#define MACHINE_HAS_PFMF (0)
92#define MACHINE_HAS_SPP (0) 93#define MACHINE_HAS_SPP (0)
94#define MACHINE_HAS_TOPOLOGY (0)
93#else /* __s390x__ */ 95#else /* __s390x__ */
94#define MACHINE_HAS_IEEE (1) 96#define MACHINE_HAS_IEEE (1)
95#define MACHINE_HAS_CSP (1) 97#define MACHINE_HAS_CSP (1)
@@ -100,6 +102,7 @@ extern unsigned int user_mode;
100#define MACHINE_HAS_HPAGE (S390_lowcore.machine_flags & MACHINE_FLAG_HPAGE) 102#define MACHINE_HAS_HPAGE (S390_lowcore.machine_flags & MACHINE_FLAG_HPAGE)
101#define MACHINE_HAS_PFMF (S390_lowcore.machine_flags & MACHINE_FLAG_PFMF) 103#define MACHINE_HAS_PFMF (S390_lowcore.machine_flags & MACHINE_FLAG_PFMF)
102#define MACHINE_HAS_SPP (S390_lowcore.machine_flags & MACHINE_FLAG_SPP) 104#define MACHINE_HAS_SPP (S390_lowcore.machine_flags & MACHINE_FLAG_SPP)
105#define MACHINE_HAS_TOPOLOGY (S390_lowcore.machine_flags & MACHINE_FLAG_TOPOLOGY)
103#endif /* __s390x__ */ 106#endif /* __s390x__ */
104 107
105#define ZFCPDUMP_HSA_SIZE (32UL<<20) 108#define ZFCPDUMP_HSA_SIZE (32UL<<20)
diff --git a/arch/s390/include/asm/smp.h b/arch/s390/include/asm/smp.h
index edc03cb9cd79..045e009fc164 100644
--- a/arch/s390/include/asm/smp.h
+++ b/arch/s390/include/asm/smp.h
@@ -20,7 +20,6 @@ extern void machine_power_off_smp(void);
20 20
21extern int __cpu_disable (void); 21extern int __cpu_disable (void);
22extern void __cpu_die (unsigned int cpu); 22extern void __cpu_die (unsigned int cpu);
23extern void cpu_die (void) __attribute__ ((noreturn));
24extern int __cpu_up (unsigned int cpu); 23extern int __cpu_up (unsigned int cpu);
25 24
26extern struct mutex smp_cpu_state_mutex; 25extern struct mutex smp_cpu_state_mutex;
@@ -71,8 +70,10 @@ static inline void smp_switch_to_ipl_cpu(void (*func)(void *), void *data)
71 70
72#ifdef CONFIG_HOTPLUG_CPU 71#ifdef CONFIG_HOTPLUG_CPU
73extern int smp_rescan_cpus(void); 72extern int smp_rescan_cpus(void);
73extern void __noreturn cpu_die(void);
74#else 74#else
75static inline int smp_rescan_cpus(void) { return 0; } 75static inline int smp_rescan_cpus(void) { return 0; }
76static inline void cpu_die(void) { }
76#endif 77#endif
77 78
78#endif /* __ASM_SMP_H */ 79#endif /* __ASM_SMP_H */
diff --git a/arch/s390/include/asm/suspend.h b/arch/s390/include/asm/suspend.h
deleted file mode 100644
index dc75c616eafe..000000000000
--- a/arch/s390/include/asm/suspend.h
+++ /dev/null
@@ -1,10 +0,0 @@
1#ifndef __ASM_S390_SUSPEND_H
2#define __ASM_S390_SUSPEND_H
3
4static inline int arch_prepare_suspend(void)
5{
6 return 0;
7}
8
9#endif
10
diff --git a/arch/s390/include/asm/syscall.h b/arch/s390/include/asm/syscall.h
index 8429686951f9..5c0246b955d8 100644
--- a/arch/s390/include/asm/syscall.h
+++ b/arch/s390/include/asm/syscall.h
@@ -65,8 +65,6 @@ static inline void syscall_get_arguments(struct task_struct *task,
65 if (test_tsk_thread_flag(task, TIF_31BIT)) 65 if (test_tsk_thread_flag(task, TIF_31BIT))
66 mask = 0xffffffff; 66 mask = 0xffffffff;
67#endif 67#endif
68 if (i + n == 6)
69 args[--n] = regs->args[0] & mask;
70 while (n-- > 0) 68 while (n-- > 0)
71 if (i + n > 0) 69 if (i + n > 0)
72 args[n] = regs->gprs[2 + i + n] & mask; 70 args[n] = regs->gprs[2 + i + n] & mask;
@@ -80,8 +78,6 @@ static inline void syscall_set_arguments(struct task_struct *task,
80 const unsigned long *args) 78 const unsigned long *args)
81{ 79{
82 BUG_ON(i + n > 6); 80 BUG_ON(i + n > 6);
83 if (i + n == 6)
84 regs->args[0] = args[--n];
85 while (n-- > 0) 81 while (n-- > 0)
86 if (i + n > 0) 82 if (i + n > 0)
87 regs->gprs[2 + i + n] = args[n]; 83 regs->gprs[2 + i + n] = args[n];
diff --git a/arch/s390/include/asm/sysinfo.h b/arch/s390/include/asm/sysinfo.h
index 22bdb2a0ee5f..79d3d6e2e9c5 100644
--- a/arch/s390/include/asm/sysinfo.h
+++ b/arch/s390/include/asm/sysinfo.h
@@ -14,8 +14,13 @@
14#ifndef __ASM_S390_SYSINFO_H 14#ifndef __ASM_S390_SYSINFO_H
15#define __ASM_S390_SYSINFO_H 15#define __ASM_S390_SYSINFO_H
16 16
17#include <asm/bitsperlong.h>
18
17struct sysinfo_1_1_1 { 19struct sysinfo_1_1_1 {
18 char reserved_0[32]; 20 unsigned short :16;
21 unsigned char ccr;
22 unsigned char cai;
23 char reserved_0[28];
19 char manufacturer[16]; 24 char manufacturer[16];
20 char type[4]; 25 char type[4];
21 char reserved_1[12]; 26 char reserved_1[12];
@@ -104,6 +109,39 @@ struct sysinfo_3_2_2 {
104 char reserved_544[3552]; 109 char reserved_544[3552];
105}; 110};
106 111
112#define TOPOLOGY_CPU_BITS 64
113#define TOPOLOGY_NR_MAG 6
114
115struct topology_cpu {
116 unsigned char reserved0[4];
117 unsigned char :6;
118 unsigned char pp:2;
119 unsigned char reserved1;
120 unsigned short origin;
121 unsigned long mask[TOPOLOGY_CPU_BITS / BITS_PER_LONG];
122};
123
124struct topology_container {
125 unsigned char reserved[7];
126 unsigned char id;
127};
128
129union topology_entry {
130 unsigned char nl;
131 struct topology_cpu cpu;
132 struct topology_container container;
133};
134
135struct sysinfo_15_1_x {
136 unsigned char reserved0[2];
137 unsigned short length;
138 unsigned char mag[TOPOLOGY_NR_MAG];
139 unsigned char reserved1;
140 unsigned char mnest;
141 unsigned char reserved2[4];
142 union topology_entry tle[0];
143};
144
107static inline int stsi(void *sysinfo, int fc, int sel1, int sel2) 145static inline int stsi(void *sysinfo, int fc, int sel1, int sel2)
108{ 146{
109 register int r0 asm("0") = (fc << 28) | sel1; 147 register int r0 asm("0") = (fc << 28) | sel1;
diff --git a/arch/s390/include/asm/system.h b/arch/s390/include/asm/system.h
index cef66210c846..d382629a0172 100644
--- a/arch/s390/include/asm/system.h
+++ b/arch/s390/include/asm/system.h
@@ -14,12 +14,14 @@
14#include <asm/setup.h> 14#include <asm/setup.h>
15#include <asm/processor.h> 15#include <asm/processor.h>
16#include <asm/lowcore.h> 16#include <asm/lowcore.h>
17#include <asm/cmpxchg.h>
17 18
18#ifdef __KERNEL__ 19#ifdef __KERNEL__
19 20
20struct task_struct; 21struct task_struct;
21 22
22extern struct task_struct *__switch_to(void *, void *); 23extern struct task_struct *__switch_to(void *, void *);
24extern void update_per_regs(struct task_struct *task);
23 25
24static inline void save_fp_regs(s390_fp_regs *fpregs) 26static inline void save_fp_regs(s390_fp_regs *fpregs)
25{ 27{
@@ -85,26 +87,26 @@ static inline void restore_access_regs(unsigned int *acrs)
85 asm volatile("lam 0,15,%0" : : "Q" (*acrs)); 87 asm volatile("lam 0,15,%0" : : "Q" (*acrs));
86} 88}
87 89
88#define switch_to(prev,next,last) do { \ 90#define switch_to(prev,next,last) do { \
89 if (prev == next) \ 91 if (prev->mm) { \
90 break; \ 92 save_fp_regs(&prev->thread.fp_regs); \
91 save_fp_regs(&prev->thread.fp_regs); \ 93 save_access_regs(&prev->thread.acrs[0]); \
92 restore_fp_regs(&next->thread.fp_regs); \ 94 } \
93 save_access_regs(&prev->thread.acrs[0]); \ 95 if (next->mm) { \
94 restore_access_regs(&next->thread.acrs[0]); \ 96 restore_fp_regs(&next->thread.fp_regs); \
95 prev = __switch_to(prev,next); \ 97 restore_access_regs(&next->thread.acrs[0]); \
98 update_per_regs(next); \
99 } \
100 prev = __switch_to(prev,next); \
96} while (0) 101} while (0)
97 102
98extern void account_vtime(struct task_struct *, struct task_struct *); 103extern void account_vtime(struct task_struct *, struct task_struct *);
99extern void account_tick_vtime(struct task_struct *); 104extern void account_tick_vtime(struct task_struct *);
100extern void account_system_vtime(struct task_struct *);
101 105
102#ifdef CONFIG_PFAULT 106#ifdef CONFIG_PFAULT
103extern void pfault_irq_init(void);
104extern int pfault_init(void); 107extern int pfault_init(void);
105extern void pfault_fini(void); 108extern void pfault_fini(void);
106#else /* CONFIG_PFAULT */ 109#else /* CONFIG_PFAULT */
107#define pfault_irq_init() do { } while (0)
108#define pfault_init() ({-1;}) 110#define pfault_init() ({-1;})
109#define pfault_fini() do { } while (0) 111#define pfault_fini() do { } while (0)
110#endif /* CONFIG_PFAULT */ 112#endif /* CONFIG_PFAULT */
@@ -119,161 +121,6 @@ extern int memcpy_real(void *, void *, size_t);
119 121
120#define nop() asm volatile("nop") 122#define nop() asm volatile("nop")
121 123
122#define xchg(ptr,x) \
123({ \
124 __typeof__(*(ptr)) __ret; \
125 __ret = (__typeof__(*(ptr))) \
126 __xchg((unsigned long)(x), (void *)(ptr),sizeof(*(ptr))); \
127 __ret; \
128})
129
130extern void __xchg_called_with_bad_pointer(void);
131
132static inline unsigned long __xchg(unsigned long x, void * ptr, int size)
133{
134 unsigned long addr, old;
135 int shift;
136
137 switch (size) {
138 case 1:
139 addr = (unsigned long) ptr;
140 shift = (3 ^ (addr & 3)) << 3;
141 addr ^= addr & 3;
142 asm volatile(
143 " l %0,%4\n"
144 "0: lr 0,%0\n"
145 " nr 0,%3\n"
146 " or 0,%2\n"
147 " cs %0,0,%4\n"
148 " jl 0b\n"
149 : "=&d" (old), "=Q" (*(int *) addr)
150 : "d" (x << shift), "d" (~(255 << shift)),
151 "Q" (*(int *) addr) : "memory", "cc", "0");
152 return old >> shift;
153 case 2:
154 addr = (unsigned long) ptr;
155 shift = (2 ^ (addr & 2)) << 3;
156 addr ^= addr & 2;
157 asm volatile(
158 " l %0,%4\n"
159 "0: lr 0,%0\n"
160 " nr 0,%3\n"
161 " or 0,%2\n"
162 " cs %0,0,%4\n"
163 " jl 0b\n"
164 : "=&d" (old), "=Q" (*(int *) addr)
165 : "d" (x << shift), "d" (~(65535 << shift)),
166 "Q" (*(int *) addr) : "memory", "cc", "0");
167 return old >> shift;
168 case 4:
169 asm volatile(
170 " l %0,%3\n"
171 "0: cs %0,%2,%3\n"
172 " jl 0b\n"
173 : "=&d" (old), "=Q" (*(int *) ptr)
174 : "d" (x), "Q" (*(int *) ptr)
175 : "memory", "cc");
176 return old;
177#ifdef __s390x__
178 case 8:
179 asm volatile(
180 " lg %0,%3\n"
181 "0: csg %0,%2,%3\n"
182 " jl 0b\n"
183 : "=&d" (old), "=m" (*(long *) ptr)
184 : "d" (x), "Q" (*(long *) ptr)
185 : "memory", "cc");
186 return old;
187#endif /* __s390x__ */
188 }
189 __xchg_called_with_bad_pointer();
190 return x;
191}
192
193/*
194 * Atomic compare and exchange. Compare OLD with MEM, if identical,
195 * store NEW in MEM. Return the initial value in MEM. Success is
196 * indicated by comparing RETURN with OLD.
197 */
198
199#define __HAVE_ARCH_CMPXCHG 1
200
201#define cmpxchg(ptr, o, n) \
202 ((__typeof__(*(ptr)))__cmpxchg((ptr), (unsigned long)(o), \
203 (unsigned long)(n), sizeof(*(ptr))))
204
205extern void __cmpxchg_called_with_bad_pointer(void);
206
207static inline unsigned long
208__cmpxchg(volatile void *ptr, unsigned long old, unsigned long new, int size)
209{
210 unsigned long addr, prev, tmp;
211 int shift;
212
213 switch (size) {
214 case 1:
215 addr = (unsigned long) ptr;
216 shift = (3 ^ (addr & 3)) << 3;
217 addr ^= addr & 3;
218 asm volatile(
219 " l %0,%2\n"
220 "0: nr %0,%5\n"
221 " lr %1,%0\n"
222 " or %0,%3\n"
223 " or %1,%4\n"
224 " cs %0,%1,%2\n"
225 " jnl 1f\n"
226 " xr %1,%0\n"
227 " nr %1,%5\n"
228 " jnz 0b\n"
229 "1:"
230 : "=&d" (prev), "=&d" (tmp), "=Q" (*(int *) ptr)
231 : "d" (old << shift), "d" (new << shift),
232 "d" (~(255 << shift)), "Q" (*(int *) ptr)
233 : "memory", "cc");
234 return prev >> shift;
235 case 2:
236 addr = (unsigned long) ptr;
237 shift = (2 ^ (addr & 2)) << 3;
238 addr ^= addr & 2;
239 asm volatile(
240 " l %0,%2\n"
241 "0: nr %0,%5\n"
242 " lr %1,%0\n"
243 " or %0,%3\n"
244 " or %1,%4\n"
245 " cs %0,%1,%2\n"
246 " jnl 1f\n"
247 " xr %1,%0\n"
248 " nr %1,%5\n"
249 " jnz 0b\n"
250 "1:"
251 : "=&d" (prev), "=&d" (tmp), "=Q" (*(int *) ptr)
252 : "d" (old << shift), "d" (new << shift),
253 "d" (~(65535 << shift)), "Q" (*(int *) ptr)
254 : "memory", "cc");
255 return prev >> shift;
256 case 4:
257 asm volatile(
258 " cs %0,%3,%1\n"
259 : "=&d" (prev), "=Q" (*(int *) ptr)
260 : "0" (old), "d" (new), "Q" (*(int *) ptr)
261 : "memory", "cc");
262 return prev;
263#ifdef __s390x__
264 case 8:
265 asm volatile(
266 " csg %0,%3,%1\n"
267 : "=&d" (prev), "=Q" (*(long *) ptr)
268 : "0" (old), "d" (new), "Q" (*(long *) ptr)
269 : "memory", "cc");
270 return prev;
271#endif /* __s390x__ */
272 }
273 __cmpxchg_called_with_bad_pointer();
274 return old;
275}
276
277/* 124/*
278 * Force strict CPU ordering. 125 * Force strict CPU ordering.
279 * And yes, this is required on UP too when we're talking 126 * And yes, this is required on UP too when we're talking
@@ -352,46 +199,6 @@ __cmpxchg(volatile void *ptr, unsigned long old, unsigned long new, int size)
352 __ctl_load(__dummy, cr, cr); \ 199 __ctl_load(__dummy, cr, cr); \
353}) 200})
354 201
355#include <linux/irqflags.h>
356
357#include <asm-generic/cmpxchg-local.h>
358
359static inline unsigned long __cmpxchg_local(volatile void *ptr,
360 unsigned long old,
361 unsigned long new, int size)
362{
363 switch (size) {
364 case 1:
365 case 2:
366 case 4:
367#ifdef __s390x__
368 case 8:
369#endif
370 return __cmpxchg(ptr, old, new, size);
371 default:
372 return __cmpxchg_local_generic(ptr, old, new, size);
373 }
374
375 return old;
376}
377
378/*
379 * cmpxchg_local and cmpxchg64_local are atomic wrt current CPU. Always make
380 * them available.
381 */
382#define cmpxchg_local(ptr, o, n) \
383 ((__typeof__(*(ptr)))__cmpxchg_local((ptr), (unsigned long)(o), \
384 (unsigned long)(n), sizeof(*(ptr))))
385#ifdef __s390x__
386#define cmpxchg64_local(ptr, o, n) \
387 ({ \
388 BUILD_BUG_ON(sizeof(*(ptr)) != 8); \
389 cmpxchg_local((ptr), (o), (n)); \
390 })
391#else
392#define cmpxchg64_local(ptr, o, n) __cmpxchg64_local_generic((ptr), (o), (n))
393#endif
394
395/* 202/*
396 * Use to set psw mask except for the first byte which 203 * Use to set psw mask except for the first byte which
397 * won't be changed by this function. 204 * won't be changed by this function.
@@ -399,7 +206,7 @@ static inline unsigned long __cmpxchg_local(volatile void *ptr,
399static inline void 206static inline void
400__set_psw_mask(unsigned long mask) 207__set_psw_mask(unsigned long mask)
401{ 208{
402 __load_psw_mask(mask | (__raw_local_irq_stosm(0x00) & ~(-1UL >> 8))); 209 __load_psw_mask(mask | (arch_local_save_flags() & ~(-1UL >> 8)));
403} 210}
404 211
405#define local_mcck_enable() __set_psw_mask(psw_kernel_bits) 212#define local_mcck_enable() __set_psw_mask(psw_kernel_bits)
@@ -419,30 +226,21 @@ extern void smp_ctl_clear_bit(int cr, int bit);
419 226
420#endif /* CONFIG_SMP */ 227#endif /* CONFIG_SMP */
421 228
422static inline unsigned int stfl(void) 229#define MAX_FACILITY_BIT (256*8) /* stfle_fac_list has 256 bytes */
423{
424 asm volatile(
425 " .insn s,0xb2b10000,0(0)\n" /* stfl */
426 "0:\n"
427 EX_TABLE(0b,0b));
428 return S390_lowcore.stfl_fac_list;
429}
430 230
431static inline int __stfle(unsigned long long *list, int doublewords) 231/*
232 * The test_facility function uses the bit odering where the MSB is bit 0.
233 * That makes it easier to query facility bits with the bit number as
234 * documented in the Principles of Operation.
235 */
236static inline int test_facility(unsigned long nr)
432{ 237{
433 typedef struct { unsigned long long _[doublewords]; } addrtype; 238 unsigned char *ptr;
434 register unsigned long __nr asm("0") = doublewords - 1;
435 239
436 asm volatile(".insn s,0xb2b00000,%0" /* stfle */ 240 if (nr >= MAX_FACILITY_BIT)
437 : "=m" (*(addrtype *) list), "+d" (__nr) : : "cc"); 241 return 0;
438 return __nr + 1; 242 ptr = (unsigned char *) &S390_lowcore.stfle_fac_list + (nr >> 3);
439} 243 return (*ptr & (0x80 >> (nr & 7))) != 0;
440
441static inline int stfle(unsigned long long *list, int doublewords)
442{
443 if (!(stfl() & (1UL << 24)))
444 return -EOPNOTSUPP;
445 return __stfle(list, doublewords);
446} 244}
447 245
448static inline unsigned short stap(void) 246static inline unsigned short stap(void)
@@ -457,7 +255,7 @@ extern void (*_machine_restart)(char *command);
457extern void (*_machine_halt)(void); 255extern void (*_machine_halt)(void);
458extern void (*_machine_power_off)(void); 256extern void (*_machine_power_off)(void);
459 257
460#define arch_align_stack(x) (x) 258extern unsigned long arch_align_stack(unsigned long sp);
461 259
462static inline int tprot(unsigned long addr) 260static inline int tprot(unsigned long addr)
463{ 261{
diff --git a/arch/s390/include/asm/thread_info.h b/arch/s390/include/asm/thread_info.h
index 5baf0230b29b..ad1382f7932e 100644
--- a/arch/s390/include/asm/thread_info.h
+++ b/arch/s390/include/asm/thread_info.h
@@ -74,7 +74,7 @@ struct thread_info {
74/* how to get the thread information struct from C */ 74/* how to get the thread information struct from C */
75static inline struct thread_info *current_thread_info(void) 75static inline struct thread_info *current_thread_info(void)
76{ 76{
77 return (struct thread_info *)(S390_lowcore.kernel_stack - THREAD_SIZE); 77 return (struct thread_info *) S390_lowcore.thread_info;
78} 78}
79 79
80#define THREAD_SIZE_ORDER THREAD_ORDER 80#define THREAD_SIZE_ORDER THREAD_ORDER
@@ -88,7 +88,7 @@ static inline struct thread_info *current_thread_info(void)
88#define TIF_SIGPENDING 2 /* signal pending */ 88#define TIF_SIGPENDING 2 /* signal pending */
89#define TIF_NEED_RESCHED 3 /* rescheduling necessary */ 89#define TIF_NEED_RESCHED 3 /* rescheduling necessary */
90#define TIF_RESTART_SVC 4 /* restart svc with new svc number */ 90#define TIF_RESTART_SVC 4 /* restart svc with new svc number */
91#define TIF_SINGLE_STEP 6 /* deliver sigtrap on return to user */ 91#define TIF_PER_TRAP 6 /* deliver sigtrap on return to user */
92#define TIF_MCCK_PENDING 7 /* machine check handling is pending */ 92#define TIF_MCCK_PENDING 7 /* machine check handling is pending */
93#define TIF_SYSCALL_TRACE 8 /* syscall trace active */ 93#define TIF_SYSCALL_TRACE 8 /* syscall trace active */
94#define TIF_SYSCALL_AUDIT 9 /* syscall auditing active */ 94#define TIF_SYSCALL_AUDIT 9 /* syscall auditing active */
@@ -99,14 +99,15 @@ static inline struct thread_info *current_thread_info(void)
99#define TIF_31BIT 17 /* 32bit process */ 99#define TIF_31BIT 17 /* 32bit process */
100#define TIF_MEMDIE 18 /* is terminating due to OOM killer */ 100#define TIF_MEMDIE 18 /* is terminating due to OOM killer */
101#define TIF_RESTORE_SIGMASK 19 /* restore signal mask in do_signal() */ 101#define TIF_RESTORE_SIGMASK 19 /* restore signal mask in do_signal() */
102#define TIF_FREEZE 20 /* thread is freezing for suspend */ 102#define TIF_SINGLE_STEP 20 /* This task is single stepped */
103#define TIF_FREEZE 21 /* thread is freezing for suspend */
103 104
104#define _TIF_NOTIFY_RESUME (1<<TIF_NOTIFY_RESUME) 105#define _TIF_NOTIFY_RESUME (1<<TIF_NOTIFY_RESUME)
105#define _TIF_RESTORE_SIGMASK (1<<TIF_RESTORE_SIGMASK) 106#define _TIF_RESTORE_SIGMASK (1<<TIF_RESTORE_SIGMASK)
106#define _TIF_SIGPENDING (1<<TIF_SIGPENDING) 107#define _TIF_SIGPENDING (1<<TIF_SIGPENDING)
107#define _TIF_NEED_RESCHED (1<<TIF_NEED_RESCHED) 108#define _TIF_NEED_RESCHED (1<<TIF_NEED_RESCHED)
108#define _TIF_RESTART_SVC (1<<TIF_RESTART_SVC) 109#define _TIF_RESTART_SVC (1<<TIF_RESTART_SVC)
109#define _TIF_SINGLE_STEP (1<<TIF_SINGLE_STEP) 110#define _TIF_PER_TRAP (1<<TIF_PER_TRAP)
110#define _TIF_MCCK_PENDING (1<<TIF_MCCK_PENDING) 111#define _TIF_MCCK_PENDING (1<<TIF_MCCK_PENDING)
111#define _TIF_SYSCALL_TRACE (1<<TIF_SYSCALL_TRACE) 112#define _TIF_SYSCALL_TRACE (1<<TIF_SYSCALL_TRACE)
112#define _TIF_SYSCALL_AUDIT (1<<TIF_SYSCALL_AUDIT) 113#define _TIF_SYSCALL_AUDIT (1<<TIF_SYSCALL_AUDIT)
@@ -114,8 +115,15 @@ static inline struct thread_info *current_thread_info(void)
114#define _TIF_SYSCALL_TRACEPOINT (1<<TIF_SYSCALL_TRACEPOINT) 115#define _TIF_SYSCALL_TRACEPOINT (1<<TIF_SYSCALL_TRACEPOINT)
115#define _TIF_POLLING_NRFLAG (1<<TIF_POLLING_NRFLAG) 116#define _TIF_POLLING_NRFLAG (1<<TIF_POLLING_NRFLAG)
116#define _TIF_31BIT (1<<TIF_31BIT) 117#define _TIF_31BIT (1<<TIF_31BIT)
118#define _TIF_SINGLE_STEP (1<<TIF_FREEZE)
117#define _TIF_FREEZE (1<<TIF_FREEZE) 119#define _TIF_FREEZE (1<<TIF_FREEZE)
118 120
121#ifdef CONFIG_64BIT
122#define is_32bit_task() (test_thread_flag(TIF_31BIT))
123#else
124#define is_32bit_task() (1)
125#endif
126
119#endif /* __KERNEL__ */ 127#endif /* __KERNEL__ */
120 128
121#define PREEMPT_ACTIVE 0x4000000 129#define PREEMPT_ACTIVE 0x4000000
diff --git a/arch/s390/include/asm/timex.h b/arch/s390/include/asm/timex.h
index 09d345a701dc..88829a40af6f 100644
--- a/arch/s390/include/asm/timex.h
+++ b/arch/s390/include/asm/timex.h
@@ -11,6 +11,8 @@
11#ifndef _ASM_S390_TIMEX_H 11#ifndef _ASM_S390_TIMEX_H
12#define _ASM_S390_TIMEX_H 12#define _ASM_S390_TIMEX_H
13 13
14#include <asm/lowcore.h>
15
14/* The value of the TOD clock for 1.1.1970. */ 16/* The value of the TOD clock for 1.1.1970. */
15#define TOD_UNIX_EPOCH 0x7d91048bca000000ULL 17#define TOD_UNIX_EPOCH 0x7d91048bca000000ULL
16 18
@@ -49,6 +51,24 @@ static inline void store_clock_comparator(__u64 *time)
49 asm volatile("stckc %0" : "=Q" (*time)); 51 asm volatile("stckc %0" : "=Q" (*time));
50} 52}
51 53
54void clock_comparator_work(void);
55
56static inline unsigned long long local_tick_disable(void)
57{
58 unsigned long long old;
59
60 old = S390_lowcore.clock_comparator;
61 S390_lowcore.clock_comparator = -1ULL;
62 set_clock_comparator(S390_lowcore.clock_comparator);
63 return old;
64}
65
66static inline void local_tick_enable(unsigned long long comp)
67{
68 S390_lowcore.clock_comparator = comp;
69 set_clock_comparator(S390_lowcore.clock_comparator);
70}
71
52#define CLOCK_TICK_RATE 1193180 /* Underlying HZ */ 72#define CLOCK_TICK_RATE 1193180 /* Underlying HZ */
53 73
54typedef unsigned long long cycles_t; 74typedef unsigned long long cycles_t;
diff --git a/arch/s390/include/asm/tlb.h b/arch/s390/include/asm/tlb.h
index fd1c00d08bf5..c687a2c83462 100644
--- a/arch/s390/include/asm/tlb.h
+++ b/arch/s390/include/asm/tlb.h
@@ -22,70 +22,77 @@
22 */ 22 */
23 23
24#include <linux/mm.h> 24#include <linux/mm.h>
25#include <linux/pagemap.h>
25#include <linux/swap.h> 26#include <linux/swap.h>
26#include <asm/processor.h> 27#include <asm/processor.h>
27#include <asm/pgalloc.h> 28#include <asm/pgalloc.h>
28#include <asm/smp.h>
29#include <asm/tlbflush.h> 29#include <asm/tlbflush.h>
30 30
31#ifndef CONFIG_SMP
32#define TLB_NR_PTRS 1
33#else
34#define TLB_NR_PTRS 508
35#endif
36
37struct mmu_gather { 31struct mmu_gather {
38 struct mm_struct *mm; 32 struct mm_struct *mm;
33#ifdef CONFIG_HAVE_RCU_TABLE_FREE
34 struct mmu_table_batch *batch;
35#endif
39 unsigned int fullmm; 36 unsigned int fullmm;
40 unsigned int nr_ptes; 37 unsigned int need_flush;
41 unsigned int nr_pxds;
42 void *array[TLB_NR_PTRS];
43}; 38};
44 39
45DECLARE_PER_CPU(struct mmu_gather, mmu_gathers); 40#ifdef CONFIG_HAVE_RCU_TABLE_FREE
41struct mmu_table_batch {
42 struct rcu_head rcu;
43 unsigned int nr;
44 void *tables[0];
45};
46 46
47static inline struct mmu_gather *tlb_gather_mmu(struct mm_struct *mm, 47#define MAX_TABLE_BATCH \
48 unsigned int full_mm_flush) 48 ((PAGE_SIZE - sizeof(struct mmu_table_batch)) / sizeof(void *))
49{
50 struct mmu_gather *tlb = &get_cpu_var(mmu_gathers);
51 49
50extern void tlb_table_flush(struct mmu_gather *tlb);
51extern void tlb_remove_table(struct mmu_gather *tlb, void *table);
52#endif
53
54static inline void tlb_gather_mmu(struct mmu_gather *tlb,
55 struct mm_struct *mm,
56 unsigned int full_mm_flush)
57{
52 tlb->mm = mm; 58 tlb->mm = mm;
53 tlb->fullmm = full_mm_flush; 59 tlb->fullmm = full_mm_flush;
54 tlb->nr_ptes = 0; 60 tlb->need_flush = 0;
55 tlb->nr_pxds = TLB_NR_PTRS; 61#ifdef CONFIG_HAVE_RCU_TABLE_FREE
62 tlb->batch = NULL;
63#endif
56 if (tlb->fullmm) 64 if (tlb->fullmm)
57 __tlb_flush_mm(mm); 65 __tlb_flush_mm(mm);
58 return tlb;
59} 66}
60 67
61static inline void tlb_flush_mmu(struct mmu_gather *tlb, 68static inline void tlb_flush_mmu(struct mmu_gather *tlb)
62 unsigned long start, unsigned long end)
63{ 69{
64 if (!tlb->fullmm && (tlb->nr_ptes > 0 || tlb->nr_pxds < TLB_NR_PTRS)) 70 if (!tlb->need_flush)
65 __tlb_flush_mm(tlb->mm); 71 return;
66 while (tlb->nr_ptes > 0) 72 tlb->need_flush = 0;
67 pte_free(tlb->mm, tlb->array[--tlb->nr_ptes]); 73 __tlb_flush_mm(tlb->mm);
68 while (tlb->nr_pxds < TLB_NR_PTRS) 74#ifdef CONFIG_HAVE_RCU_TABLE_FREE
69 /* pgd_free frees the pointer as region or segment table */ 75 tlb_table_flush(tlb);
70 pgd_free(tlb->mm, tlb->array[tlb->nr_pxds++]); 76#endif
71} 77}
72 78
73static inline void tlb_finish_mmu(struct mmu_gather *tlb, 79static inline void tlb_finish_mmu(struct mmu_gather *tlb,
74 unsigned long start, unsigned long end) 80 unsigned long start, unsigned long end)
75{ 81{
76 tlb_flush_mmu(tlb, start, end); 82 tlb_flush_mmu(tlb);
77
78 /* keep the page table cache within bounds */
79 check_pgt_cache();
80
81 put_cpu_var(mmu_gathers);
82} 83}
83 84
84/* 85/*
85 * Release the page cache reference for a pte removed by 86 * Release the page cache reference for a pte removed by
86 * tlb_ptep_clear_flush. In both flush modes the tlb fo a page cache page 87 * tlb_ptep_clear_flush. In both flush modes the tlb for a page cache page
87 * has already been freed, so just do free_page_and_swap_cache. 88 * has already been freed, so just do free_page_and_swap_cache.
88 */ 89 */
90static inline int __tlb_remove_page(struct mmu_gather *tlb, struct page *page)
91{
92 free_page_and_swap_cache(page);
93 return 1; /* avoid calling tlb_flush_mmu */
94}
95
89static inline void tlb_remove_page(struct mmu_gather *tlb, struct page *page) 96static inline void tlb_remove_page(struct mmu_gather *tlb, struct page *page)
90{ 97{
91 free_page_and_swap_cache(page); 98 free_page_and_swap_cache(page);
@@ -98,12 +105,11 @@ static inline void tlb_remove_page(struct mmu_gather *tlb, struct page *page)
98static inline void pte_free_tlb(struct mmu_gather *tlb, pgtable_t pte, 105static inline void pte_free_tlb(struct mmu_gather *tlb, pgtable_t pte,
99 unsigned long address) 106 unsigned long address)
100{ 107{
101 if (!tlb->fullmm) { 108#ifdef CONFIG_HAVE_RCU_TABLE_FREE
102 tlb->array[tlb->nr_ptes++] = pte; 109 if (!tlb->fullmm)
103 if (tlb->nr_ptes >= tlb->nr_pxds) 110 return page_table_free_rcu(tlb, (unsigned long *) pte);
104 tlb_flush_mmu(tlb, 0, 0); 111#endif
105 } else 112 page_table_free(tlb->mm, (unsigned long *) pte);
106 pte_free(tlb->mm, pte);
107} 113}
108 114
109/* 115/*
@@ -119,12 +125,11 @@ static inline void pmd_free_tlb(struct mmu_gather *tlb, pmd_t *pmd,
119#ifdef __s390x__ 125#ifdef __s390x__
120 if (tlb->mm->context.asce_limit <= (1UL << 31)) 126 if (tlb->mm->context.asce_limit <= (1UL << 31))
121 return; 127 return;
122 if (!tlb->fullmm) { 128#ifdef CONFIG_HAVE_RCU_TABLE_FREE
123 tlb->array[--tlb->nr_pxds] = pmd; 129 if (!tlb->fullmm)
124 if (tlb->nr_ptes >= tlb->nr_pxds) 130 return tlb_remove_table(tlb, pmd);
125 tlb_flush_mmu(tlb, 0, 0); 131#endif
126 } else 132 crst_table_free(tlb->mm, (unsigned long *) pmd);
127 pmd_free(tlb->mm, pmd);
128#endif 133#endif
129} 134}
130 135
@@ -141,12 +146,11 @@ static inline void pud_free_tlb(struct mmu_gather *tlb, pud_t *pud,
141#ifdef __s390x__ 146#ifdef __s390x__
142 if (tlb->mm->context.asce_limit <= (1UL << 42)) 147 if (tlb->mm->context.asce_limit <= (1UL << 42))
143 return; 148 return;
144 if (!tlb->fullmm) { 149#ifdef CONFIG_HAVE_RCU_TABLE_FREE
145 tlb->array[--tlb->nr_pxds] = pud; 150 if (!tlb->fullmm)
146 if (tlb->nr_ptes >= tlb->nr_pxds) 151 return tlb_remove_table(tlb, pud);
147 tlb_flush_mmu(tlb, 0, 0); 152#endif
148 } else 153 crst_table_free(tlb->mm, (unsigned long *) pud);
149 pud_free(tlb->mm, pud);
150#endif 154#endif
151} 155}
152 156
diff --git a/arch/s390/include/asm/tlbflush.h b/arch/s390/include/asm/tlbflush.h
index 29d5d6d4becc..b7a4f2eb0057 100644
--- a/arch/s390/include/asm/tlbflush.h
+++ b/arch/s390/include/asm/tlbflush.h
@@ -50,7 +50,7 @@ static inline void __tlb_flush_full(struct mm_struct *mm)
50 /* 50 /*
51 * If the process only ran on the local cpu, do a local flush. 51 * If the process only ran on the local cpu, do a local flush.
52 */ 52 */
53 local_cpumask = cpumask_of_cpu(smp_processor_id()); 53 cpumask_copy(&local_cpumask, cpumask_of(smp_processor_id()));
54 if (cpumask_equal(mm_cpumask(mm), &local_cpumask)) 54 if (cpumask_equal(mm_cpumask(mm), &local_cpumask))
55 __tlb_flush_local(); 55 __tlb_flush_local();
56 else 56 else
@@ -80,16 +80,11 @@ static inline void __tlb_flush_mm(struct mm_struct * mm)
80 * on all cpus instead of doing a local flush if the mm 80 * on all cpus instead of doing a local flush if the mm
81 * only ran on the local cpu. 81 * only ran on the local cpu.
82 */ 82 */
83 if (MACHINE_HAS_IDTE) { 83 if (MACHINE_HAS_IDTE)
84 if (mm->context.noexec)
85 __tlb_flush_idte((unsigned long)
86 get_shadow_table(mm->pgd) |
87 mm->context.asce_bits);
88 __tlb_flush_idte((unsigned long) mm->pgd | 84 __tlb_flush_idte((unsigned long) mm->pgd |
89 mm->context.asce_bits); 85 mm->context.asce_bits);
90 return; 86 else
91 } 87 __tlb_flush_full(mm);
92 __tlb_flush_full(mm);
93} 88}
94 89
95static inline void __tlb_flush_mm_cond(struct mm_struct * mm) 90static inline void __tlb_flush_mm_cond(struct mm_struct * mm)
diff --git a/arch/s390/include/asm/topology.h b/arch/s390/include/asm/topology.h
index 831bd033ea77..005d77d8ae2a 100644
--- a/arch/s390/include/asm/topology.h
+++ b/arch/s390/include/asm/topology.h
@@ -2,19 +2,38 @@
2#define _ASM_S390_TOPOLOGY_H 2#define _ASM_S390_TOPOLOGY_H
3 3
4#include <linux/cpumask.h> 4#include <linux/cpumask.h>
5 5#include <asm/sysinfo.h>
6#define mc_capable() (1)
7
8const struct cpumask *cpu_coregroup_mask(unsigned int cpu);
9 6
10extern unsigned char cpu_core_id[NR_CPUS]; 7extern unsigned char cpu_core_id[NR_CPUS];
11extern cpumask_t cpu_core_map[NR_CPUS]; 8extern cpumask_t cpu_core_map[NR_CPUS];
12 9
10static inline const struct cpumask *cpu_coregroup_mask(int cpu)
11{
12 return &cpu_core_map[cpu];
13}
14
13#define topology_core_id(cpu) (cpu_core_id[cpu]) 15#define topology_core_id(cpu) (cpu_core_id[cpu])
14#define topology_core_cpumask(cpu) (&cpu_core_map[cpu]) 16#define topology_core_cpumask(cpu) (&cpu_core_map[cpu])
17#define mc_capable() (1)
18
19#ifdef CONFIG_SCHED_BOOK
20
21extern unsigned char cpu_book_id[NR_CPUS];
22extern cpumask_t cpu_book_map[NR_CPUS];
23
24static inline const struct cpumask *cpu_book_mask(int cpu)
25{
26 return &cpu_book_map[cpu];
27}
28
29#define topology_book_id(cpu) (cpu_book_id[cpu])
30#define topology_book_cpumask(cpu) (&cpu_book_map[cpu])
31
32#endif /* CONFIG_SCHED_BOOK */
15 33
16int topology_set_cpu_management(int fc); 34int topology_set_cpu_management(int fc);
17void topology_schedule_update(void); 35void topology_schedule_update(void);
36void store_topology(struct sysinfo_15_1_x *info);
18 37
19#define POLARIZATION_UNKNWN (-1) 38#define POLARIZATION_UNKNWN (-1)
20#define POLARIZATION_HRZ (0) 39#define POLARIZATION_HRZ (0)
@@ -30,6 +49,8 @@ static inline void s390_init_cpu_topology(void)
30}; 49};
31#endif 50#endif
32 51
52#define SD_BOOK_INIT SD_CPU_INIT
53
33#include <asm-generic/topology.h> 54#include <asm-generic/topology.h>
34 55
35#endif /* _ASM_S390_TOPOLOGY_H */ 56#endif /* _ASM_S390_TOPOLOGY_H */
diff --git a/arch/s390/include/asm/types.h b/arch/s390/include/asm/types.h
index 04d6b95a89c6..eeb52ccf499f 100644
--- a/arch/s390/include/asm/types.h
+++ b/arch/s390/include/asm/types.h
@@ -30,14 +30,6 @@ typedef __signed__ long saddr_t;
30 30
31#ifndef __ASSEMBLY__ 31#ifndef __ASSEMBLY__
32 32
33typedef u64 dma64_addr_t;
34#ifdef __s390x__
35/* DMA addresses come in 32-bit and 64-bit flavours. */
36typedef u64 dma_addr_t;
37#else
38typedef u32 dma_addr_t;
39#endif
40
41#ifndef __s390x__ 33#ifndef __s390x__
42typedef union { 34typedef union {
43 unsigned long long pair; 35 unsigned long long pair;
diff --git a/arch/s390/include/asm/uaccess.h b/arch/s390/include/asm/uaccess.h
index d6b1ed0ec52b..2b23885e81e9 100644
--- a/arch/s390/include/asm/uaccess.h
+++ b/arch/s390/include/asm/uaccess.h
@@ -49,12 +49,13 @@
49 49
50#define segment_eq(a,b) ((a).ar4 == (b).ar4) 50#define segment_eq(a,b) ((a).ar4 == (b).ar4)
51 51
52#define __access_ok(addr, size) \
53({ \
54 __chk_user_ptr(addr); \
55 1; \
56})
52 57
53static inline int __access_ok(const void __user *addr, unsigned long size) 58#define access_ok(type, addr, size) __access_ok(addr, size)
54{
55 return 1;
56}
57#define access_ok(type,addr,size) __access_ok(addr,size)
58 59
59/* 60/*
60 * The exception table consists of pairs of addresses: the first is the 61 * The exception table consists of pairs of addresses: the first is the
@@ -83,8 +84,8 @@ struct uaccess_ops {
83 size_t (*clear_user)(size_t, void __user *); 84 size_t (*clear_user)(size_t, void __user *);
84 size_t (*strnlen_user)(size_t, const char __user *); 85 size_t (*strnlen_user)(size_t, const char __user *);
85 size_t (*strncpy_from_user)(size_t, const char __user *, char *); 86 size_t (*strncpy_from_user)(size_t, const char __user *, char *);
86 int (*futex_atomic_op)(int op, int __user *, int oparg, int *old); 87 int (*futex_atomic_op)(int op, u32 __user *, int oparg, int *old);
87 int (*futex_atomic_cmpxchg)(int __user *, int old, int new); 88 int (*futex_atomic_cmpxchg)(u32 *, u32 __user *, u32 old, u32 new);
88}; 89};
89 90
90extern struct uaccess_ops uaccess; 91extern struct uaccess_ops uaccess;
diff --git a/arch/s390/include/asm/unistd.h b/arch/s390/include/asm/unistd.h
index 1049ef27c15e..404bdb9671b4 100644
--- a/arch/s390/include/asm/unistd.h
+++ b/arch/s390/include/asm/unistd.h
@@ -272,7 +272,12 @@
272#define __NR_fanotify_init 332 272#define __NR_fanotify_init 332
273#define __NR_fanotify_mark 333 273#define __NR_fanotify_mark 333
274#define __NR_prlimit64 334 274#define __NR_prlimit64 334
275#define NR_syscalls 335 275#define __NR_name_to_handle_at 335
276#define __NR_open_by_handle_at 336
277#define __NR_clock_adjtime 337
278#define __NR_syncfs 338
279#define __NR_setns 339
280#define NR_syscalls 340
276 281
277/* 282/*
278 * There are some system calls that are not present on 64 bit, some 283 * There are some system calls that are not present on 64 bit, some
@@ -381,6 +386,7 @@
381 386
382/* Ignore system calls that are also reachable via sys_socket */ 387/* Ignore system calls that are also reachable via sys_socket */
383#define __IGNORE_recvmmsg 388#define __IGNORE_recvmmsg
389#define __IGNORE_sendmmsg
384 390
385#define __ARCH_WANT_IPC_PARSE_VERSION 391#define __ARCH_WANT_IPC_PARSE_VERSION
386#define __ARCH_WANT_OLD_READDIR 392#define __ARCH_WANT_OLD_READDIR
diff --git a/arch/s390/kernel/Makefile b/arch/s390/kernel/Makefile
index 64230bc392fa..df3732249baa 100644
--- a/arch/s390/kernel/Makefile
+++ b/arch/s390/kernel/Makefile
@@ -20,10 +20,10 @@ CFLAGS_ptrace.o += -DUTS_MACHINE='"$(UTS_MACHINE)"'
20 20
21CFLAGS_sysinfo.o += -Iinclude/math-emu -Iarch/s390/math-emu -w 21CFLAGS_sysinfo.o += -Iinclude/math-emu -Iarch/s390/math-emu -w
22 22
23obj-y := bitmap.o traps.o time.o process.o base.o early.o setup.o \ 23obj-y := bitmap.o traps.o time.o process.o base.o early.o setup.o vtime.o \
24 processor.o sys_s390.o ptrace.o signal.o cpcmd.o ebcdic.o \ 24 processor.o sys_s390.o ptrace.o signal.o cpcmd.o ebcdic.o nmi.o \
25 s390_ext.o debug.o irq.o ipl.o dis.o diag.o mem_detect.o \ 25 debug.o irq.o ipl.o dis.o diag.o mem_detect.o sclp.o vdso.o \
26 vdso.o vtime.o sysinfo.o nmi.o sclp.o 26 sysinfo.o jump_label.o
27 27
28obj-y += $(if $(CONFIG_64BIT),entry64.o,entry.o) 28obj-y += $(if $(CONFIG_64BIT),entry64.o,entry.o)
29obj-y += $(if $(CONFIG_64BIT),reipl64.o,reipl.o) 29obj-y += $(if $(CONFIG_64BIT),reipl64.o,reipl.o)
diff --git a/arch/s390/kernel/asm-offsets.c b/arch/s390/kernel/asm-offsets.c
index 5232278d79ad..edfbd17d7082 100644
--- a/arch/s390/kernel/asm-offsets.c
+++ b/arch/s390/kernel/asm-offsets.c
@@ -23,14 +23,16 @@ int main(void)
23{ 23{
24 DEFINE(__THREAD_info, offsetof(struct task_struct, stack)); 24 DEFINE(__THREAD_info, offsetof(struct task_struct, stack));
25 DEFINE(__THREAD_ksp, offsetof(struct task_struct, thread.ksp)); 25 DEFINE(__THREAD_ksp, offsetof(struct task_struct, thread.ksp));
26 DEFINE(__THREAD_per, offsetof(struct task_struct, thread.per_info));
27 DEFINE(__THREAD_mm_segment, offsetof(struct task_struct, thread.mm_segment)); 26 DEFINE(__THREAD_mm_segment, offsetof(struct task_struct, thread.mm_segment));
28 BLANK(); 27 BLANK();
29 DEFINE(__TASK_pid, offsetof(struct task_struct, pid)); 28 DEFINE(__TASK_pid, offsetof(struct task_struct, pid));
30 BLANK(); 29 BLANK();
31 DEFINE(__PER_atmid, offsetof(per_struct, lowcore.words.perc_atmid)); 30 DEFINE(__THREAD_per_cause,
32 DEFINE(__PER_address, offsetof(per_struct, lowcore.words.address)); 31 offsetof(struct task_struct, thread.per_event.cause));
33 DEFINE(__PER_access_id, offsetof(per_struct, lowcore.words.access_id)); 32 DEFINE(__THREAD_per_address,
33 offsetof(struct task_struct, thread.per_event.address));
34 DEFINE(__THREAD_per_paid,
35 offsetof(struct task_struct, thread.per_event.paid));
34 BLANK(); 36 BLANK();
35 DEFINE(__TI_task, offsetof(struct thread_info, task)); 37 DEFINE(__TI_task, offsetof(struct thread_info, task));
36 DEFINE(__TI_domain, offsetof(struct thread_info, exec_domain)); 38 DEFINE(__TI_domain, offsetof(struct thread_info, exec_domain));
@@ -66,9 +68,9 @@ int main(void)
66 DEFINE(__VDSO_ECTG_BASE, offsetof(struct vdso_per_cpu_data, ectg_timer_base)); 68 DEFINE(__VDSO_ECTG_BASE, offsetof(struct vdso_per_cpu_data, ectg_timer_base));
67 DEFINE(__VDSO_ECTG_USER, offsetof(struct vdso_per_cpu_data, ectg_user_time)); 69 DEFINE(__VDSO_ECTG_USER, offsetof(struct vdso_per_cpu_data, ectg_user_time));
68 /* constants used by the vdso */ 70 /* constants used by the vdso */
69 DEFINE(CLOCK_REALTIME, CLOCK_REALTIME); 71 DEFINE(__CLOCK_REALTIME, CLOCK_REALTIME);
70 DEFINE(CLOCK_MONOTONIC, CLOCK_MONOTONIC); 72 DEFINE(__CLOCK_MONOTONIC, CLOCK_MONOTONIC);
71 DEFINE(CLOCK_REALTIME_RES, MONOTONIC_RES_NSEC); 73 DEFINE(__CLOCK_REALTIME_RES, MONOTONIC_RES_NSEC);
72 BLANK(); 74 BLANK();
73 /* constants for SIGP */ 75 /* constants for SIGP */
74 DEFINE(__SIGP_STOP, sigp_stop); 76 DEFINE(__SIGP_STOP, sigp_stop);
@@ -84,9 +86,10 @@ int main(void)
84 DEFINE(__LC_SVC_INT_CODE, offsetof(struct _lowcore, svc_code)); 86 DEFINE(__LC_SVC_INT_CODE, offsetof(struct _lowcore, svc_code));
85 DEFINE(__LC_PGM_ILC, offsetof(struct _lowcore, pgm_ilc)); 87 DEFINE(__LC_PGM_ILC, offsetof(struct _lowcore, pgm_ilc));
86 DEFINE(__LC_PGM_INT_CODE, offsetof(struct _lowcore, pgm_code)); 88 DEFINE(__LC_PGM_INT_CODE, offsetof(struct _lowcore, pgm_code));
87 DEFINE(__LC_PER_ATMID, offsetof(struct _lowcore, per_perc_atmid)); 89 DEFINE(__LC_TRANS_EXC_CODE, offsetof(struct _lowcore, trans_exc_code));
90 DEFINE(__LC_PER_CAUSE, offsetof(struct _lowcore, per_perc_atmid));
88 DEFINE(__LC_PER_ADDRESS, offsetof(struct _lowcore, per_address)); 91 DEFINE(__LC_PER_ADDRESS, offsetof(struct _lowcore, per_address));
89 DEFINE(__LC_PER_ACCESS_ID, offsetof(struct _lowcore, per_access_id)); 92 DEFINE(__LC_PER_PAID, offsetof(struct _lowcore, per_access_id));
90 DEFINE(__LC_AR_MODE_ID, offsetof(struct _lowcore, ar_access_id)); 93 DEFINE(__LC_AR_MODE_ID, offsetof(struct _lowcore, ar_access_id));
91 DEFINE(__LC_SUBCHANNEL_ID, offsetof(struct _lowcore, subchannel_id)); 94 DEFINE(__LC_SUBCHANNEL_ID, offsetof(struct _lowcore, subchannel_id));
92 DEFINE(__LC_SUBCHANNEL_NR, offsetof(struct _lowcore, subchannel_nr)); 95 DEFINE(__LC_SUBCHANNEL_NR, offsetof(struct _lowcore, subchannel_nr));
@@ -121,13 +124,11 @@ int main(void)
121 DEFINE(__LC_LAST_UPDATE_TIMER, offsetof(struct _lowcore, last_update_timer)); 124 DEFINE(__LC_LAST_UPDATE_TIMER, offsetof(struct _lowcore, last_update_timer));
122 DEFINE(__LC_LAST_UPDATE_CLOCK, offsetof(struct _lowcore, last_update_clock)); 125 DEFINE(__LC_LAST_UPDATE_CLOCK, offsetof(struct _lowcore, last_update_clock));
123 DEFINE(__LC_CURRENT, offsetof(struct _lowcore, current_task)); 126 DEFINE(__LC_CURRENT, offsetof(struct _lowcore, current_task));
127 DEFINE(__LC_CURRENT_PID, offsetof(struct _lowcore, current_pid));
124 DEFINE(__LC_THREAD_INFO, offsetof(struct _lowcore, thread_info)); 128 DEFINE(__LC_THREAD_INFO, offsetof(struct _lowcore, thread_info));
125 DEFINE(__LC_KERNEL_STACK, offsetof(struct _lowcore, kernel_stack)); 129 DEFINE(__LC_KERNEL_STACK, offsetof(struct _lowcore, kernel_stack));
126 DEFINE(__LC_ASYNC_STACK, offsetof(struct _lowcore, async_stack)); 130 DEFINE(__LC_ASYNC_STACK, offsetof(struct _lowcore, async_stack));
127 DEFINE(__LC_PANIC_STACK, offsetof(struct _lowcore, panic_stack)); 131 DEFINE(__LC_PANIC_STACK, offsetof(struct _lowcore, panic_stack));
128 DEFINE(__LC_KERNEL_ASCE, offsetof(struct _lowcore, kernel_asce));
129 DEFINE(__LC_USER_ASCE, offsetof(struct _lowcore, user_asce));
130 DEFINE(__LC_USER_EXEC_ASCE, offsetof(struct _lowcore, user_exec_asce));
131 DEFINE(__LC_INT_CLOCK, offsetof(struct _lowcore, int_clock)); 132 DEFINE(__LC_INT_CLOCK, offsetof(struct _lowcore, int_clock));
132 DEFINE(__LC_MCCK_CLOCK, offsetof(struct _lowcore, mcck_clock)); 133 DEFINE(__LC_MCCK_CLOCK, offsetof(struct _lowcore, mcck_clock));
133 DEFINE(__LC_MACHINE_FLAGS, offsetof(struct _lowcore, machine_flags)); 134 DEFINE(__LC_MACHINE_FLAGS, offsetof(struct _lowcore, machine_flags));
@@ -142,10 +143,8 @@ int main(void)
142 DEFINE(__LC_GPREGS_SAVE_AREA, offsetof(struct _lowcore, gpregs_save_area)); 143 DEFINE(__LC_GPREGS_SAVE_AREA, offsetof(struct _lowcore, gpregs_save_area));
143 DEFINE(__LC_CREGS_SAVE_AREA, offsetof(struct _lowcore, cregs_save_area)); 144 DEFINE(__LC_CREGS_SAVE_AREA, offsetof(struct _lowcore, cregs_save_area));
144#ifdef CONFIG_32BIT 145#ifdef CONFIG_32BIT
145 DEFINE(__LC_PFAULT_INTPARM, offsetof(struct _lowcore, ext_params));
146 DEFINE(SAVE_AREA_BASE, offsetof(struct _lowcore, extended_save_area_addr)); 146 DEFINE(SAVE_AREA_BASE, offsetof(struct _lowcore, extended_save_area_addr));
147#else /* CONFIG_32BIT */ 147#else /* CONFIG_32BIT */
148 DEFINE(__LC_PFAULT_INTPARM, offsetof(struct _lowcore, ext_params2));
149 DEFINE(__LC_EXT_PARAMS2, offsetof(struct _lowcore, ext_params2)); 148 DEFINE(__LC_EXT_PARAMS2, offsetof(struct _lowcore, ext_params2));
150 DEFINE(SAVE_AREA_BASE, offsetof(struct _lowcore, floating_pt_save_area)); 149 DEFINE(SAVE_AREA_BASE, offsetof(struct _lowcore, floating_pt_save_area));
151 DEFINE(__LC_PASTE, offsetof(struct _lowcore, paste)); 150 DEFINE(__LC_PASTE, offsetof(struct _lowcore, paste));
diff --git a/arch/s390/kernel/compat_linux.c b/arch/s390/kernel/compat_linux.c
index 1e6449c79ab6..53acaa86dd94 100644
--- a/arch/s390/kernel/compat_linux.c
+++ b/arch/s390/kernel/compat_linux.c
@@ -25,7 +25,6 @@
25#include <linux/resource.h> 25#include <linux/resource.h>
26#include <linux/times.h> 26#include <linux/times.h>
27#include <linux/smp.h> 27#include <linux/smp.h>
28#include <linux/smp_lock.h>
29#include <linux/sem.h> 28#include <linux/sem.h>
30#include <linux/msg.h> 29#include <linux/msg.h>
31#include <linux/shm.h> 30#include <linux/shm.h>
diff --git a/arch/s390/kernel/compat_ptrace.h b/arch/s390/kernel/compat_ptrace.h
index 123dd660d7fb..12b823833510 100644
--- a/arch/s390/kernel/compat_ptrace.h
+++ b/arch/s390/kernel/compat_ptrace.h
@@ -4,40 +4,19 @@
4#include <asm/ptrace.h> /* needed for NUM_CR_WORDS */ 4#include <asm/ptrace.h> /* needed for NUM_CR_WORDS */
5#include "compat_linux.h" /* needed for psw_compat_t */ 5#include "compat_linux.h" /* needed for psw_compat_t */
6 6
7typedef struct { 7struct compat_per_struct_kernel {
8 __u32 cr[NUM_CR_WORDS]; 8 __u32 cr9; /* PER control bits */
9} per_cr_words32; 9 __u32 cr10; /* PER starting address */
10 10 __u32 cr11; /* PER ending address */
11typedef struct { 11 __u32 bits; /* Obsolete software bits */
12 __u16 perc_atmid; /* 0x096 */ 12 __u32 starting_addr; /* User specified start address */
13 __u32 address; /* 0x098 */ 13 __u32 ending_addr; /* User specified end address */
14 __u8 access_id; /* 0x0a1 */ 14 __u16 perc_atmid; /* PER trap ATMID */
15} per_lowcore_words32; 15 __u32 address; /* PER trap instruction address */
16 16 __u8 access_id; /* PER trap access identification */
17typedef struct { 17};
18 union {
19 per_cr_words32 words;
20 } control_regs;
21 /*
22 * Use these flags instead of setting em_instruction_fetch
23 * directly they are used so that single stepping can be
24 * switched on & off while not affecting other tracing
25 */
26 unsigned single_step : 1;
27 unsigned instruction_fetch : 1;
28 unsigned : 30;
29 /*
30 * These addresses are copied into cr10 & cr11 if single
31 * stepping is switched off
32 */
33 __u32 starting_addr;
34 __u32 ending_addr;
35 union {
36 per_lowcore_words32 words;
37 } lowcore;
38} per_struct32;
39 18
40struct user_regs_struct32 19struct compat_user_regs_struct
41{ 20{
42 psw_compat_t psw; 21 psw_compat_t psw;
43 u32 gprs[NUM_GPRS]; 22 u32 gprs[NUM_GPRS];
@@ -50,15 +29,14 @@ struct user_regs_struct32
50 * itself as there is no "official" ptrace interface for hardware 29 * itself as there is no "official" ptrace interface for hardware
51 * watchpoints. This is the way intel does it. 30 * watchpoints. This is the way intel does it.
52 */ 31 */
53 per_struct32 per_info; 32 struct compat_per_struct_kernel per_info;
54 u32 ieee_instruction_pointer; 33 u32 ieee_instruction_pointer; /* obsolete, always 0 */
55 /* Used to give failing instruction back to user for ieee exceptions */
56}; 34};
57 35
58struct user32 { 36struct compat_user {
59 /* We start with the registers, to mimic the way that "memory" 37 /* We start with the registers, to mimic the way that "memory"
60 is returned from the ptrace(3,...) function. */ 38 is returned from the ptrace(3,...) function. */
61 struct user_regs_struct32 regs; /* Where the registers are actually stored */ 39 struct compat_user_regs_struct regs;
62 /* The rest of this junk is to help gdb figure out what goes where */ 40 /* The rest of this junk is to help gdb figure out what goes where */
63 u32 u_tsize; /* Text segment size (pages). */ 41 u32 u_tsize; /* Text segment size (pages). */
64 u32 u_dsize; /* Data segment size (pages). */ 42 u32 u_dsize; /* Data segment size (pages). */
@@ -80,6 +58,6 @@ typedef struct
80 __u32 len; 58 __u32 len;
81 __u32 kernel_addr; 59 __u32 kernel_addr;
82 __u32 process_addr; 60 __u32 process_addr;
83} ptrace_area_emu31; 61} compat_ptrace_area;
84 62
85#endif /* _PTRACE32_H */ 63#endif /* _PTRACE32_H */
diff --git a/arch/s390/kernel/compat_wrapper.S b/arch/s390/kernel/compat_wrapper.S
index 8e60fb23b90d..1f5eb789c3a7 100644
--- a/arch/s390/kernel/compat_wrapper.S
+++ b/arch/s390/kernel/compat_wrapper.S
@@ -1877,3 +1877,36 @@ sys_prlimit64_wrapper:
1877 llgtr %r4,%r4 # const struct rlimit64 __user * 1877 llgtr %r4,%r4 # const struct rlimit64 __user *
1878 llgtr %r5,%r5 # struct rlimit64 __user * 1878 llgtr %r5,%r5 # struct rlimit64 __user *
1879 jg sys_prlimit64 # branch to system call 1879 jg sys_prlimit64 # branch to system call
1880
1881 .globl sys_name_to_handle_at_wrapper
1882sys_name_to_handle_at_wrapper:
1883 lgfr %r2,%r2 # int
1884 llgtr %r3,%r3 # const char __user *
1885 llgtr %r4,%r4 # struct file_handle __user *
1886 llgtr %r5,%r5 # int __user *
1887 lgfr %r6,%r6 # int
1888 jg sys_name_to_handle_at
1889
1890 .globl compat_sys_open_by_handle_at_wrapper
1891compat_sys_open_by_handle_at_wrapper:
1892 lgfr %r2,%r2 # int
1893 llgtr %r3,%r3 # struct file_handle __user *
1894 lgfr %r4,%r4 # int
1895 jg compat_sys_open_by_handle_at
1896
1897 .globl compat_sys_clock_adjtime_wrapper
1898compat_sys_clock_adjtime_wrapper:
1899 lgfr %r2,%r2 # clockid_t (int)
1900 llgtr %r3,%r3 # struct compat_timex __user *
1901 jg compat_sys_clock_adjtime
1902
1903 .globl sys_syncfs_wrapper
1904sys_syncfs_wrapper:
1905 lgfr %r2,%r2 # int
1906 jg sys_syncfs
1907
1908 .globl sys_setns_wrapper
1909sys_setns_wrapper:
1910 lgfr %r2,%r2 # int
1911 lgfr %r3,%r3 # int
1912 jg sys_setns
diff --git a/arch/s390/kernel/debug.c b/arch/s390/kernel/debug.c
index 98192261491d..5ad6bc078bfd 100644
--- a/arch/s390/kernel/debug.c
+++ b/arch/s390/kernel/debug.c
@@ -174,6 +174,7 @@ static const struct file_operations debug_file_ops = {
174 .write = debug_input, 174 .write = debug_input,
175 .open = debug_open, 175 .open = debug_open,
176 .release = debug_close, 176 .release = debug_close,
177 .llseek = no_llseek,
177}; 178};
178 179
179static struct dentry *debug_debugfs_root_entry; 180static struct dentry *debug_debugfs_root_entry;
diff --git a/arch/s390/kernel/diag.c b/arch/s390/kernel/diag.c
index c032d11da8a1..8237fc07ac79 100644
--- a/arch/s390/kernel/diag.c
+++ b/arch/s390/kernel/diag.c
@@ -9,27 +9,6 @@
9#include <asm/diag.h> 9#include <asm/diag.h>
10 10
11/* 11/*
12 * Diagnose 10: Release pages
13 */
14void diag10(unsigned long addr)
15{
16 if (addr >= 0x7ff00000)
17 return;
18 asm volatile(
19#ifdef CONFIG_64BIT
20 " sam31\n"
21 " diag %0,%0,0x10\n"
22 "0: sam64\n"
23#else
24 " diag %0,%0,0x10\n"
25 "0:\n"
26#endif
27 EX_TABLE(0b, 0b)
28 : : "a" (addr));
29}
30EXPORT_SYMBOL(diag10);
31
32/*
33 * Diagnose 14: Input spool file manipulation 12 * Diagnose 14: Input spool file manipulation
34 */ 13 */
35int diag14(unsigned long rx, unsigned long ry1, unsigned long subcode) 14int diag14(unsigned long rx, unsigned long ry1, unsigned long subcode)
diff --git a/arch/s390/kernel/dis.c b/arch/s390/kernel/dis.c
index b39b27d68b45..1ca3d1d6a86c 100644
--- a/arch/s390/kernel/dis.c
+++ b/arch/s390/kernel/dis.c
@@ -30,9 +30,9 @@
30#include <asm/atomic.h> 30#include <asm/atomic.h>
31#include <asm/mathemu.h> 31#include <asm/mathemu.h>
32#include <asm/cpcmd.h> 32#include <asm/cpcmd.h>
33#include <asm/s390_ext.h>
34#include <asm/lowcore.h> 33#include <asm/lowcore.h>
35#include <asm/debug.h> 34#include <asm/debug.h>
35#include <asm/irq.h>
36 36
37#ifndef CONFIG_64BIT 37#ifndef CONFIG_64BIT
38#define ONELONG "%08lx: " 38#define ONELONG "%08lx: "
@@ -113,7 +113,7 @@ enum {
113 INSTR_INVALID, 113 INSTR_INVALID,
114 INSTR_E, 114 INSTR_E,
115 INSTR_RIE_R0IU, INSTR_RIE_R0UU, INSTR_RIE_RRP, INSTR_RIE_RRPU, 115 INSTR_RIE_R0IU, INSTR_RIE_R0UU, INSTR_RIE_RRP, INSTR_RIE_RRPU,
116 INSTR_RIE_RRUUU, INSTR_RIE_RUPI, INSTR_RIE_RUPU, 116 INSTR_RIE_RRUUU, INSTR_RIE_RUPI, INSTR_RIE_RUPU, INSTR_RIE_RRI0,
117 INSTR_RIL_RI, INSTR_RIL_RP, INSTR_RIL_RU, INSTR_RIL_UP, 117 INSTR_RIL_RI, INSTR_RIL_RP, INSTR_RIL_RU, INSTR_RIL_UP,
118 INSTR_RIS_R0RDU, INSTR_RIS_R0UU, INSTR_RIS_RURDI, INSTR_RIS_RURDU, 118 INSTR_RIS_R0RDU, INSTR_RIS_R0UU, INSTR_RIS_RURDI, INSTR_RIS_RURDU,
119 INSTR_RI_RI, INSTR_RI_RP, INSTR_RI_RU, INSTR_RI_UP, 119 INSTR_RI_RI, INSTR_RI_RP, INSTR_RI_RU, INSTR_RI_UP,
@@ -122,13 +122,14 @@ enum {
122 INSTR_RRE_RR, INSTR_RRE_RR_OPT, 122 INSTR_RRE_RR, INSTR_RRE_RR_OPT,
123 INSTR_RRF_0UFF, INSTR_RRF_F0FF, INSTR_RRF_F0FF2, INSTR_RRF_F0FR, 123 INSTR_RRF_0UFF, INSTR_RRF_F0FF, INSTR_RRF_F0FF2, INSTR_RRF_F0FR,
124 INSTR_RRF_FFRU, INSTR_RRF_FUFF, INSTR_RRF_M0RR, INSTR_RRF_R0RR, 124 INSTR_RRF_FFRU, INSTR_RRF_FUFF, INSTR_RRF_M0RR, INSTR_RRF_R0RR,
125 INSTR_RRF_RURR, INSTR_RRF_U0FF, INSTR_RRF_U0RF, INSTR_RRF_U0RR, 125 INSTR_RRF_R0RR2, INSTR_RRF_RURR, INSTR_RRF_U0FF, INSTR_RRF_U0RF,
126 INSTR_RRF_UUFF, INSTR_RRR_F0FF, INSTR_RRS_RRRDU, 126 INSTR_RRF_U0RR, INSTR_RRF_UUFF, INSTR_RRR_F0FF, INSTR_RRS_RRRDU,
127 INSTR_RR_FF, INSTR_RR_R0, INSTR_RR_RR, INSTR_RR_U0, INSTR_RR_UR, 127 INSTR_RR_FF, INSTR_RR_R0, INSTR_RR_RR, INSTR_RR_U0, INSTR_RR_UR,
128 INSTR_RSE_CCRD, INSTR_RSE_RRRD, INSTR_RSE_RURD, 128 INSTR_RSE_CCRD, INSTR_RSE_RRRD, INSTR_RSE_RURD,
129 INSTR_RSI_RRP, 129 INSTR_RSI_RRP,
130 INSTR_RSL_R0RD, 130 INSTR_RSL_R0RD,
131 INSTR_RSY_AARD, INSTR_RSY_CCRD, INSTR_RSY_RRRD, INSTR_RSY_RURD, 131 INSTR_RSY_AARD, INSTR_RSY_CCRD, INSTR_RSY_RRRD, INSTR_RSY_RURD,
132 INSTR_RSY_RDRM,
132 INSTR_RS_AARD, INSTR_RS_CCRD, INSTR_RS_R0RD, INSTR_RS_RRRD, 133 INSTR_RS_AARD, INSTR_RS_CCRD, INSTR_RS_R0RD, INSTR_RS_RRRD,
133 INSTR_RS_RURD, 134 INSTR_RS_RURD,
134 INSTR_RXE_FRRD, INSTR_RXE_RRRD, 135 INSTR_RXE_FRRD, INSTR_RXE_RRRD,
@@ -139,7 +140,7 @@ enum {
139 INSTR_SIY_IRD, INSTR_SIY_URD, 140 INSTR_SIY_IRD, INSTR_SIY_URD,
140 INSTR_SI_URD, 141 INSTR_SI_URD,
141 INSTR_SSE_RDRD, 142 INSTR_SSE_RDRD,
142 INSTR_SSF_RRDRD, 143 INSTR_SSF_RRDRD, INSTR_SSF_RRDRD2,
143 INSTR_SS_L0RDRD, INSTR_SS_LIRDRD, INSTR_SS_LLRDRD, INSTR_SS_RRRDRD, 144 INSTR_SS_L0RDRD, INSTR_SS_LIRDRD, INSTR_SS_LLRDRD, INSTR_SS_RRRDRD,
144 INSTR_SS_RRRDRD2, INSTR_SS_RRRDRD3, 145 INSTR_SS_RRRDRD2, INSTR_SS_RRRDRD3,
145 INSTR_S_00, INSTR_S_RD, 146 INSTR_S_00, INSTR_S_RD,
@@ -152,7 +153,7 @@ struct operand {
152}; 153};
153 154
154struct insn { 155struct insn {
155 const char name[6]; 156 const char name[5];
156 unsigned char opfrag; 157 unsigned char opfrag;
157 unsigned char format; 158 unsigned char format;
158}; 159};
@@ -217,6 +218,7 @@ static const unsigned char formats[][7] = {
217 [INSTR_RIE_RRP] = { 0xff, R_8,R_12,J16_16,0,0,0 }, 218 [INSTR_RIE_RRP] = { 0xff, R_8,R_12,J16_16,0,0,0 },
218 [INSTR_RIE_RRUUU] = { 0xff, R_8,R_12,U8_16,U8_24,U8_32,0 }, 219 [INSTR_RIE_RRUUU] = { 0xff, R_8,R_12,U8_16,U8_24,U8_32,0 },
219 [INSTR_RIE_RUPI] = { 0xff, R_8,I8_32,U4_12,J16_16,0,0 }, 220 [INSTR_RIE_RUPI] = { 0xff, R_8,I8_32,U4_12,J16_16,0,0 },
221 [INSTR_RIE_RRI0] = { 0xff, R_8,R_12,I16_16,0,0,0 },
220 [INSTR_RIL_RI] = { 0x0f, R_8,I32_16,0,0,0,0 }, 222 [INSTR_RIL_RI] = { 0x0f, R_8,I32_16,0,0,0,0 },
221 [INSTR_RIL_RP] = { 0x0f, R_8,J32_16,0,0,0,0 }, 223 [INSTR_RIL_RP] = { 0x0f, R_8,J32_16,0,0,0,0 },
222 [INSTR_RIL_RU] = { 0x0f, R_8,U32_16,0,0,0,0 }, 224 [INSTR_RIL_RU] = { 0x0f, R_8,U32_16,0,0,0,0 },
@@ -248,6 +250,7 @@ static const unsigned char formats[][7] = {
248 [INSTR_RRF_FUFF] = { 0xff, F_24,F_16,F_28,U4_20,0,0 }, 250 [INSTR_RRF_FUFF] = { 0xff, F_24,F_16,F_28,U4_20,0,0 },
249 [INSTR_RRF_M0RR] = { 0xff, R_24,R_28,M_16,0,0,0 }, 251 [INSTR_RRF_M0RR] = { 0xff, R_24,R_28,M_16,0,0,0 },
250 [INSTR_RRF_R0RR] = { 0xff, R_24,R_16,R_28,0,0,0 }, 252 [INSTR_RRF_R0RR] = { 0xff, R_24,R_16,R_28,0,0,0 },
253 [INSTR_RRF_R0RR2] = { 0xff, R_24,R_28,R_16,0,0,0 },
251 [INSTR_RRF_RURR] = { 0xff, R_24,R_28,R_16,U4_20,0,0 }, 254 [INSTR_RRF_RURR] = { 0xff, R_24,R_28,R_16,U4_20,0,0 },
252 [INSTR_RRF_U0FF] = { 0xff, F_24,U4_16,F_28,0,0,0 }, 255 [INSTR_RRF_U0FF] = { 0xff, F_24,U4_16,F_28,0,0,0 },
253 [INSTR_RRF_U0RF] = { 0xff, R_24,U4_16,F_28,0,0,0 }, 256 [INSTR_RRF_U0RF] = { 0xff, R_24,U4_16,F_28,0,0,0 },
@@ -269,6 +272,7 @@ static const unsigned char formats[][7] = {
269 [INSTR_RSY_CCRD] = { 0xff, C_8,C_12,D20_20,B_16,0,0 }, 272 [INSTR_RSY_CCRD] = { 0xff, C_8,C_12,D20_20,B_16,0,0 },
270 [INSTR_RSY_RRRD] = { 0xff, R_8,R_12,D20_20,B_16,0,0 }, 273 [INSTR_RSY_RRRD] = { 0xff, R_8,R_12,D20_20,B_16,0,0 },
271 [INSTR_RSY_RURD] = { 0xff, R_8,U4_12,D20_20,B_16,0,0 }, 274 [INSTR_RSY_RURD] = { 0xff, R_8,U4_12,D20_20,B_16,0,0 },
275 [INSTR_RSY_RDRM] = { 0xff, R_8,D20_20,B_16,U4_12,0,0 },
272 [INSTR_RS_AARD] = { 0xff, A_8,A_12,D_20,B_16,0,0 }, 276 [INSTR_RS_AARD] = { 0xff, A_8,A_12,D_20,B_16,0,0 },
273 [INSTR_RS_CCRD] = { 0xff, C_8,C_12,D_20,B_16,0,0 }, 277 [INSTR_RS_CCRD] = { 0xff, C_8,C_12,D_20,B_16,0,0 },
274 [INSTR_RS_R0RD] = { 0xff, R_8,D_20,B_16,0,0,0 }, 278 [INSTR_RS_R0RD] = { 0xff, R_8,D_20,B_16,0,0,0 },
@@ -290,6 +294,7 @@ static const unsigned char formats[][7] = {
290 [INSTR_SI_URD] = { 0xff, D_20,B_16,U8_8,0,0,0 }, 294 [INSTR_SI_URD] = { 0xff, D_20,B_16,U8_8,0,0,0 },
291 [INSTR_SSE_RDRD] = { 0xff, D_20,B_16,D_36,B_32,0,0 }, 295 [INSTR_SSE_RDRD] = { 0xff, D_20,B_16,D_36,B_32,0,0 },
292 [INSTR_SSF_RRDRD] = { 0x00, D_20,B_16,D_36,B_32,R_8,0 }, 296 [INSTR_SSF_RRDRD] = { 0x00, D_20,B_16,D_36,B_32,R_8,0 },
297 [INSTR_SSF_RRDRD2]= { 0x00, R_8,D_20,B_16,D_36,B_32,0 },
293 [INSTR_SS_L0RDRD] = { 0xff, D_20,L8_8,B_16,D_36,B_32,0 }, 298 [INSTR_SS_L0RDRD] = { 0xff, D_20,L8_8,B_16,D_36,B_32,0 },
294 [INSTR_SS_LIRDRD] = { 0xff, D_20,L4_8,B_16,D_36,B_32,U4_12 }, 299 [INSTR_SS_LIRDRD] = { 0xff, D_20,L4_8,B_16,D_36,B_32,U4_12 },
295 [INSTR_SS_LLRDRD] = { 0xff, D_20,L4_8,B_16,D_36,L4_12,B_32 }, 300 [INSTR_SS_LLRDRD] = { 0xff, D_20,L4_8,B_16,D_36,L4_12,B_32 },
@@ -300,6 +305,36 @@ static const unsigned char formats[][7] = {
300 [INSTR_S_RD] = { 0xff, D_20,B_16,0,0,0,0 }, 305 [INSTR_S_RD] = { 0xff, D_20,B_16,0,0,0,0 },
301}; 306};
302 307
308enum {
309 LONG_INSN_ALGHSIK,
310 LONG_INSN_ALHSIK,
311 LONG_INSN_CLFHSI,
312 LONG_INSN_CLGFRL,
313 LONG_INSN_CLGHRL,
314 LONG_INSN_CLGHSI,
315 LONG_INSN_CLHHSI,
316 LONG_INSN_LLGFRL,
317 LONG_INSN_LLGHRL,
318 LONG_INSN_POPCNT,
319 LONG_INSN_RISBHG,
320 LONG_INSN_RISBLG,
321};
322
323static char *long_insn_name[] = {
324 [LONG_INSN_ALGHSIK] = "alghsik",
325 [LONG_INSN_ALHSIK] = "alhsik",
326 [LONG_INSN_CLFHSI] = "clfhsi",
327 [LONG_INSN_CLGFRL] = "clgfrl",
328 [LONG_INSN_CLGHRL] = "clghrl",
329 [LONG_INSN_CLGHSI] = "clghsi",
330 [LONG_INSN_CLHHSI] = "clhhsi",
331 [LONG_INSN_LLGFRL] = "llgfrl",
332 [LONG_INSN_LLGHRL] = "llghrl",
333 [LONG_INSN_POPCNT] = "popcnt",
334 [LONG_INSN_RISBHG] = "risbhg",
335 [LONG_INSN_RISBLG] = "risblk",
336};
337
303static struct insn opcode[] = { 338static struct insn opcode[] = {
304#ifdef CONFIG_64BIT 339#ifdef CONFIG_64BIT
305 { "lmd", 0xef, INSTR_SS_RRRDRD3 }, 340 { "lmd", 0xef, INSTR_SS_RRRDRD3 },
@@ -637,6 +672,7 @@ static struct insn opcode_b2[] = {
637 { "rp", 0x77, INSTR_S_RD }, 672 { "rp", 0x77, INSTR_S_RD },
638 { "stcke", 0x78, INSTR_S_RD }, 673 { "stcke", 0x78, INSTR_S_RD },
639 { "sacf", 0x79, INSTR_S_RD }, 674 { "sacf", 0x79, INSTR_S_RD },
675 { "spp", 0x80, INSTR_S_RD },
640 { "stsi", 0x7d, INSTR_S_RD }, 676 { "stsi", 0x7d, INSTR_S_RD },
641 { "srnm", 0x99, INSTR_S_RD }, 677 { "srnm", 0x99, INSTR_S_RD },
642 { "stfpc", 0x9c, INSTR_S_RD }, 678 { "stfpc", 0x9c, INSTR_S_RD },
@@ -881,6 +917,35 @@ static struct insn opcode_b9[] = {
881 { "pfmf", 0xaf, INSTR_RRE_RR }, 917 { "pfmf", 0xaf, INSTR_RRE_RR },
882 { "trte", 0xbf, INSTR_RRF_M0RR }, 918 { "trte", 0xbf, INSTR_RRF_M0RR },
883 { "trtre", 0xbd, INSTR_RRF_M0RR }, 919 { "trtre", 0xbd, INSTR_RRF_M0RR },
920 { "ahhhr", 0xc8, INSTR_RRF_R0RR2 },
921 { "shhhr", 0xc9, INSTR_RRF_R0RR2 },
922 { "alhhh", 0xca, INSTR_RRF_R0RR2 },
923 { "alhhl", 0xca, INSTR_RRF_R0RR2 },
924 { "slhhh", 0xcb, INSTR_RRF_R0RR2 },
925 { "chhr ", 0xcd, INSTR_RRE_RR },
926 { "clhhr", 0xcf, INSTR_RRE_RR },
927 { "ahhlr", 0xd8, INSTR_RRF_R0RR2 },
928 { "shhlr", 0xd9, INSTR_RRF_R0RR2 },
929 { "slhhl", 0xdb, INSTR_RRF_R0RR2 },
930 { "chlr", 0xdd, INSTR_RRE_RR },
931 { "clhlr", 0xdf, INSTR_RRE_RR },
932 { { 0, LONG_INSN_POPCNT }, 0xe1, INSTR_RRE_RR },
933 { "locgr", 0xe2, INSTR_RRF_M0RR },
934 { "ngrk", 0xe4, INSTR_RRF_R0RR2 },
935 { "ogrk", 0xe6, INSTR_RRF_R0RR2 },
936 { "xgrk", 0xe7, INSTR_RRF_R0RR2 },
937 { "agrk", 0xe8, INSTR_RRF_R0RR2 },
938 { "sgrk", 0xe9, INSTR_RRF_R0RR2 },
939 { "algrk", 0xea, INSTR_RRF_R0RR2 },
940 { "slgrk", 0xeb, INSTR_RRF_R0RR2 },
941 { "locr", 0xf2, INSTR_RRF_M0RR },
942 { "nrk", 0xf4, INSTR_RRF_R0RR2 },
943 { "ork", 0xf6, INSTR_RRF_R0RR2 },
944 { "xrk", 0xf7, INSTR_RRF_R0RR2 },
945 { "ark", 0xf8, INSTR_RRF_R0RR2 },
946 { "srk", 0xf9, INSTR_RRF_R0RR2 },
947 { "alrk", 0xfa, INSTR_RRF_R0RR2 },
948 { "slrk", 0xfb, INSTR_RRF_R0RR2 },
884#endif 949#endif
885 { "kmac", 0x1e, INSTR_RRE_RR }, 950 { "kmac", 0x1e, INSTR_RRE_RR },
886 { "lrvr", 0x1f, INSTR_RRE_RR }, 951 { "lrvr", 0x1f, INSTR_RRE_RR },
@@ -949,9 +1014,9 @@ static struct insn opcode_c4[] = {
949 { "lgfrl", 0x0c, INSTR_RIL_RP }, 1014 { "lgfrl", 0x0c, INSTR_RIL_RP },
950 { "lhrl", 0x05, INSTR_RIL_RP }, 1015 { "lhrl", 0x05, INSTR_RIL_RP },
951 { "lghrl", 0x04, INSTR_RIL_RP }, 1016 { "lghrl", 0x04, INSTR_RIL_RP },
952 { "llgfrl", 0x0e, INSTR_RIL_RP }, 1017 { { 0, LONG_INSN_LLGFRL }, 0x0e, INSTR_RIL_RP },
953 { "llhrl", 0x02, INSTR_RIL_RP }, 1018 { "llhrl", 0x02, INSTR_RIL_RP },
954 { "llghrl", 0x06, INSTR_RIL_RP }, 1019 { { 0, LONG_INSN_LLGHRL }, 0x06, INSTR_RIL_RP },
955 { "strl", 0x0f, INSTR_RIL_RP }, 1020 { "strl", 0x0f, INSTR_RIL_RP },
956 { "stgrl", 0x0b, INSTR_RIL_RP }, 1021 { "stgrl", 0x0b, INSTR_RIL_RP },
957 { "sthrl", 0x07, INSTR_RIL_RP }, 1022 { "sthrl", 0x07, INSTR_RIL_RP },
@@ -968,9 +1033,9 @@ static struct insn opcode_c6[] = {
968 { "cghrl", 0x04, INSTR_RIL_RP }, 1033 { "cghrl", 0x04, INSTR_RIL_RP },
969 { "clrl", 0x0f, INSTR_RIL_RP }, 1034 { "clrl", 0x0f, INSTR_RIL_RP },
970 { "clgrl", 0x0a, INSTR_RIL_RP }, 1035 { "clgrl", 0x0a, INSTR_RIL_RP },
971 { "clgfrl", 0x0e, INSTR_RIL_RP }, 1036 { { 0, LONG_INSN_CLGFRL }, 0x0e, INSTR_RIL_RP },
972 { "clhrl", 0x07, INSTR_RIL_RP }, 1037 { "clhrl", 0x07, INSTR_RIL_RP },
973 { "clghrl", 0x06, INSTR_RIL_RP }, 1038 { { 0, LONG_INSN_CLGHRL }, 0x06, INSTR_RIL_RP },
974 { "pfdrl", 0x02, INSTR_RIL_UP }, 1039 { "pfdrl", 0x02, INSTR_RIL_UP },
975 { "exrl", 0x00, INSTR_RIL_RP }, 1040 { "exrl", 0x00, INSTR_RIL_RP },
976#endif 1041#endif
@@ -982,6 +1047,20 @@ static struct insn opcode_c8[] = {
982 { "mvcos", 0x00, INSTR_SSF_RRDRD }, 1047 { "mvcos", 0x00, INSTR_SSF_RRDRD },
983 { "ectg", 0x01, INSTR_SSF_RRDRD }, 1048 { "ectg", 0x01, INSTR_SSF_RRDRD },
984 { "csst", 0x02, INSTR_SSF_RRDRD }, 1049 { "csst", 0x02, INSTR_SSF_RRDRD },
1050 { "lpd", 0x04, INSTR_SSF_RRDRD2 },
1051 { "lpdg ", 0x05, INSTR_SSF_RRDRD2 },
1052#endif
1053 { "", 0, INSTR_INVALID }
1054};
1055
1056static struct insn opcode_cc[] = {
1057#ifdef CONFIG_64BIT
1058 { "brcth", 0x06, INSTR_RIL_RP },
1059 { "aih", 0x08, INSTR_RIL_RI },
1060 { "alsih", 0x0a, INSTR_RIL_RI },
1061 { "alsih", 0x0b, INSTR_RIL_RI },
1062 { "cih", 0x0d, INSTR_RIL_RI },
1063 { "clih ", 0x0f, INSTR_RIL_RI },
985#endif 1064#endif
986 { "", 0, INSTR_INVALID } 1065 { "", 0, INSTR_INVALID }
987}; 1066};
@@ -1063,6 +1142,16 @@ static struct insn opcode_e3[] = {
1063 { "mfy", 0x5c, INSTR_RXY_RRRD }, 1142 { "mfy", 0x5c, INSTR_RXY_RRRD },
1064 { "mhy", 0x7c, INSTR_RXY_RRRD }, 1143 { "mhy", 0x7c, INSTR_RXY_RRRD },
1065 { "pfd", 0x36, INSTR_RXY_URRD }, 1144 { "pfd", 0x36, INSTR_RXY_URRD },
1145 { "lbh", 0xc0, INSTR_RXY_RRRD },
1146 { "llch", 0xc2, INSTR_RXY_RRRD },
1147 { "stch", 0xc3, INSTR_RXY_RRRD },
1148 { "lhh", 0xc4, INSTR_RXY_RRRD },
1149 { "llhh", 0xc6, INSTR_RXY_RRRD },
1150 { "sthh", 0xc7, INSTR_RXY_RRRD },
1151 { "lfh", 0xca, INSTR_RXY_RRRD },
1152 { "stfh", 0xcb, INSTR_RXY_RRRD },
1153 { "chf", 0xcd, INSTR_RXY_RRRD },
1154 { "clhf", 0xcf, INSTR_RXY_RRRD },
1066#endif 1155#endif
1067 { "lrv", 0x1e, INSTR_RXY_RRRD }, 1156 { "lrv", 0x1e, INSTR_RXY_RRRD },
1068 { "lrvh", 0x1f, INSTR_RXY_RRRD }, 1157 { "lrvh", 0x1f, INSTR_RXY_RRRD },
@@ -1080,9 +1169,9 @@ static struct insn opcode_e5[] = {
1080 { "chhsi", 0x54, INSTR_SIL_RDI }, 1169 { "chhsi", 0x54, INSTR_SIL_RDI },
1081 { "chsi", 0x5c, INSTR_SIL_RDI }, 1170 { "chsi", 0x5c, INSTR_SIL_RDI },
1082 { "cghsi", 0x58, INSTR_SIL_RDI }, 1171 { "cghsi", 0x58, INSTR_SIL_RDI },
1083 { "clhhsi", 0x55, INSTR_SIL_RDU }, 1172 { { 0, LONG_INSN_CLHHSI }, 0x55, INSTR_SIL_RDU },
1084 { "clfhsi", 0x5d, INSTR_SIL_RDU }, 1173 { { 0, LONG_INSN_CLFHSI }, 0x5d, INSTR_SIL_RDU },
1085 { "clghsi", 0x59, INSTR_SIL_RDU }, 1174 { { 0, LONG_INSN_CLGHSI }, 0x59, INSTR_SIL_RDU },
1086 { "mvhhi", 0x44, INSTR_SIL_RDI }, 1175 { "mvhhi", 0x44, INSTR_SIL_RDI },
1087 { "mvhi", 0x4c, INSTR_SIL_RDI }, 1176 { "mvhi", 0x4c, INSTR_SIL_RDI },
1088 { "mvghi", 0x48, INSTR_SIL_RDI }, 1177 { "mvghi", 0x48, INSTR_SIL_RDI },
@@ -1137,6 +1226,24 @@ static struct insn opcode_eb[] = {
1137 { "alsi", 0x6e, INSTR_SIY_IRD }, 1226 { "alsi", 0x6e, INSTR_SIY_IRD },
1138 { "algsi", 0x7e, INSTR_SIY_IRD }, 1227 { "algsi", 0x7e, INSTR_SIY_IRD },
1139 { "ecag", 0x4c, INSTR_RSY_RRRD }, 1228 { "ecag", 0x4c, INSTR_RSY_RRRD },
1229 { "srak", 0xdc, INSTR_RSY_RRRD },
1230 { "slak", 0xdd, INSTR_RSY_RRRD },
1231 { "srlk", 0xde, INSTR_RSY_RRRD },
1232 { "sllk", 0xdf, INSTR_RSY_RRRD },
1233 { "locg", 0xe2, INSTR_RSY_RDRM },
1234 { "stocg", 0xe3, INSTR_RSY_RDRM },
1235 { "lang", 0xe4, INSTR_RSY_RRRD },
1236 { "laog", 0xe6, INSTR_RSY_RRRD },
1237 { "laxg", 0xe7, INSTR_RSY_RRRD },
1238 { "laag", 0xe8, INSTR_RSY_RRRD },
1239 { "laalg", 0xea, INSTR_RSY_RRRD },
1240 { "loc", 0xf2, INSTR_RSY_RDRM },
1241 { "stoc", 0xf3, INSTR_RSY_RDRM },
1242 { "lan", 0xf4, INSTR_RSY_RRRD },
1243 { "lao", 0xf6, INSTR_RSY_RRRD },
1244 { "lax", 0xf7, INSTR_RSY_RRRD },
1245 { "laa", 0xf8, INSTR_RSY_RRRD },
1246 { "laal", 0xfa, INSTR_RSY_RRRD },
1140#endif 1247#endif
1141 { "rll", 0x1d, INSTR_RSY_RRRD }, 1248 { "rll", 0x1d, INSTR_RSY_RRRD },
1142 { "mvclu", 0x8e, INSTR_RSY_RRRD }, 1249 { "mvclu", 0x8e, INSTR_RSY_RRRD },
@@ -1172,6 +1279,12 @@ static struct insn opcode_ec[] = {
1172 { "rxsbg", 0x57, INSTR_RIE_RRUUU }, 1279 { "rxsbg", 0x57, INSTR_RIE_RRUUU },
1173 { "rosbg", 0x56, INSTR_RIE_RRUUU }, 1280 { "rosbg", 0x56, INSTR_RIE_RRUUU },
1174 { "risbg", 0x55, INSTR_RIE_RRUUU }, 1281 { "risbg", 0x55, INSTR_RIE_RRUUU },
1282 { { 0, LONG_INSN_RISBLG }, 0x51, INSTR_RIE_RRUUU },
1283 { { 0, LONG_INSN_RISBHG }, 0x5D, INSTR_RIE_RRUUU },
1284 { "ahik", 0xd8, INSTR_RIE_RRI0 },
1285 { "aghik", 0xd9, INSTR_RIE_RRI0 },
1286 { { 0, LONG_INSN_ALHSIK }, 0xda, INSTR_RIE_RRI0 },
1287 { { 0, LONG_INSN_ALGHSIK }, 0xdb, INSTR_RIE_RRI0 },
1175#endif 1288#endif
1176 { "", 0, INSTR_INVALID } 1289 { "", 0, INSTR_INVALID }
1177}; 1290};
@@ -1321,6 +1434,9 @@ static struct insn *find_insn(unsigned char *code)
1321 case 0xc8: 1434 case 0xc8:
1322 table = opcode_c8; 1435 table = opcode_c8;
1323 break; 1436 break;
1437 case 0xcc:
1438 table = opcode_cc;
1439 break;
1324 case 0xe3: 1440 case 0xe3:
1325 table = opcode_e3; 1441 table = opcode_e3;
1326 opfrag = code[5]; 1442 opfrag = code[5];
@@ -1367,7 +1483,11 @@ static int print_insn(char *buffer, unsigned char *code, unsigned long addr)
1367 ptr = buffer; 1483 ptr = buffer;
1368 insn = find_insn(code); 1484 insn = find_insn(code);
1369 if (insn) { 1485 if (insn) {
1370 ptr += sprintf(ptr, "%.5s\t", insn->name); 1486 if (insn->name[0] == '\0')
1487 ptr += sprintf(ptr, "%s\t",
1488 long_insn_name[(int) insn->name[1]]);
1489 else
1490 ptr += sprintf(ptr, "%.5s\t", insn->name);
1371 /* Extract the operands. */ 1491 /* Extract the operands. */
1372 separator = 0; 1492 separator = 0;
1373 for (ops = formats[insn->format] + 1, i = 0; 1493 for (ops = formats[insn->format] + 1, i = 0;
diff --git a/arch/s390/kernel/early.c b/arch/s390/kernel/early.c
index c00856ad4e5a..068f8465c4ee 100644
--- a/arch/s390/kernel/early.c
+++ b/arch/s390/kernel/early.c
@@ -94,6 +94,7 @@ static noinline __init void create_kernel_nss(void)
94 unsigned int sinitrd_pfn, einitrd_pfn; 94 unsigned int sinitrd_pfn, einitrd_pfn;
95#endif 95#endif
96 int response; 96 int response;
97 int hlen;
97 size_t len; 98 size_t len;
98 char *savesys_ptr; 99 char *savesys_ptr;
99 char defsys_cmd[DEFSYS_CMD_SIZE]; 100 char defsys_cmd[DEFSYS_CMD_SIZE];
@@ -124,24 +125,27 @@ static noinline __init void create_kernel_nss(void)
124 end_pfn = PFN_UP(__pa(&_end)); 125 end_pfn = PFN_UP(__pa(&_end));
125 min_size = end_pfn << 2; 126 min_size = end_pfn << 2;
126 127
127 sprintf(defsys_cmd, "DEFSYS %s 00000-%.5X EW %.5X-%.5X SR %.5X-%.5X", 128 hlen = snprintf(defsys_cmd, DEFSYS_CMD_SIZE,
128 kernel_nss_name, stext_pfn - 1, stext_pfn, eshared_pfn - 1, 129 "DEFSYS %s 00000-%.5X EW %.5X-%.5X SR %.5X-%.5X",
129 eshared_pfn, end_pfn); 130 kernel_nss_name, stext_pfn - 1, stext_pfn,
131 eshared_pfn - 1, eshared_pfn, end_pfn);
130 132
131#ifdef CONFIG_BLK_DEV_INITRD 133#ifdef CONFIG_BLK_DEV_INITRD
132 if (INITRD_START && INITRD_SIZE) { 134 if (INITRD_START && INITRD_SIZE) {
133 sinitrd_pfn = PFN_DOWN(__pa(INITRD_START)); 135 sinitrd_pfn = PFN_DOWN(__pa(INITRD_START));
134 einitrd_pfn = PFN_UP(__pa(INITRD_START + INITRD_SIZE)); 136 einitrd_pfn = PFN_UP(__pa(INITRD_START + INITRD_SIZE));
135 min_size = einitrd_pfn << 2; 137 min_size = einitrd_pfn << 2;
136 sprintf(defsys_cmd, "%s EW %.5X-%.5X", defsys_cmd, 138 hlen += snprintf(defsys_cmd + hlen, DEFSYS_CMD_SIZE - hlen,
137 sinitrd_pfn, einitrd_pfn); 139 " EW %.5X-%.5X", sinitrd_pfn, einitrd_pfn);
138 } 140 }
139#endif 141#endif
140 142
141 sprintf(defsys_cmd, "%s EW MINSIZE=%.7iK PARMREGS=0-13", 143 snprintf(defsys_cmd + hlen, DEFSYS_CMD_SIZE - hlen,
142 defsys_cmd, min_size); 144 " EW MINSIZE=%.7iK PARMREGS=0-13", min_size);
143 sprintf(savesys_cmd, "SAVESYS %s \n IPL %s", 145 defsys_cmd[DEFSYS_CMD_SIZE - 1] = '\0';
144 kernel_nss_name, kernel_nss_name); 146 snprintf(savesys_cmd, SAVESYS_CMD_SIZE, "SAVESYS %s \n IPL %s",
147 kernel_nss_name, kernel_nss_name);
148 savesys_cmd[SAVESYS_CMD_SIZE - 1] = '\0';
145 149
146 __cpcmd(defsys_cmd, NULL, 0, &response); 150 __cpcmd(defsys_cmd, NULL, 0, &response);
147 151
@@ -208,7 +212,8 @@ static noinline __init void init_kernel_storage_key(void)
208 end_pfn = PFN_UP(__pa(&_end)); 212 end_pfn = PFN_UP(__pa(&_end));
209 213
210 for (init_pfn = 0 ; init_pfn < end_pfn; init_pfn++) 214 for (init_pfn = 0 ; init_pfn < end_pfn; init_pfn++)
211 page_set_storage_key(init_pfn << PAGE_SHIFT, PAGE_DEFAULT_KEY); 215 page_set_storage_key(init_pfn << PAGE_SHIFT,
216 PAGE_DEFAULT_KEY, 0);
212} 217}
213 218
214static __initdata struct sysinfo_3_2_2 vmms __aligned(PAGE_SIZE); 219static __initdata struct sysinfo_3_2_2 vmms __aligned(PAGE_SIZE);
@@ -255,13 +260,33 @@ static noinline __init void setup_lowcore_early(void)
255 s390_base_pgm_handler_fn = early_pgm_check_handler; 260 s390_base_pgm_handler_fn = early_pgm_check_handler;
256} 261}
257 262
263static noinline __init void setup_facility_list(void)
264{
265 unsigned long nr;
266
267 S390_lowcore.stfl_fac_list = 0;
268 asm volatile(
269 " .insn s,0xb2b10000,0(0)\n" /* stfl */
270 "0:\n"
271 EX_TABLE(0b,0b) : "=m" (S390_lowcore.stfl_fac_list));
272 memcpy(&S390_lowcore.stfle_fac_list, &S390_lowcore.stfl_fac_list, 4);
273 nr = 4; /* # bytes stored by stfl */
274 if (test_facility(7)) {
275 /* More facility bits available with stfle */
276 register unsigned long reg0 asm("0") = MAX_FACILITY_BIT/64 - 1;
277 asm volatile(".insn s,0xb2b00000,%0" /* stfle */
278 : "=m" (S390_lowcore.stfle_fac_list), "+d" (reg0)
279 : : "cc");
280 nr = (reg0 + 1) * 8; /* # bytes stored by stfle */
281 }
282 memset((char *) S390_lowcore.stfle_fac_list + nr, 0,
283 MAX_FACILITY_BIT/8 - nr);
284}
285
258static noinline __init void setup_hpage(void) 286static noinline __init void setup_hpage(void)
259{ 287{
260#ifndef CONFIG_DEBUG_PAGEALLOC 288#ifndef CONFIG_DEBUG_PAGEALLOC
261 unsigned int facilities; 289 if (!test_facility(2) || !test_facility(8))
262
263 facilities = stfl();
264 if (!(facilities & (1UL << 23)) || !(facilities & (1UL << 29)))
265 return; 290 return;
266 S390_lowcore.machine_flags |= MACHINE_FLAG_HPAGE; 291 S390_lowcore.machine_flags |= MACHINE_FLAG_HPAGE;
267 __ctl_set_bit(0, 23); 292 __ctl_set_bit(0, 23);
@@ -355,18 +380,15 @@ static __init void detect_diag44(void)
355static __init void detect_machine_facilities(void) 380static __init void detect_machine_facilities(void)
356{ 381{
357#ifdef CONFIG_64BIT 382#ifdef CONFIG_64BIT
358 unsigned int facilities; 383 if (test_facility(3))
359 unsigned long long facility_bits;
360
361 facilities = stfl();
362 if (facilities & (1 << 28))
363 S390_lowcore.machine_flags |= MACHINE_FLAG_IDTE; 384 S390_lowcore.machine_flags |= MACHINE_FLAG_IDTE;
364 if (facilities & (1 << 23)) 385 if (test_facility(8))
365 S390_lowcore.machine_flags |= MACHINE_FLAG_PFMF; 386 S390_lowcore.machine_flags |= MACHINE_FLAG_PFMF;
366 if (facilities & (1 << 4)) 387 if (test_facility(11))
388 S390_lowcore.machine_flags |= MACHINE_FLAG_TOPOLOGY;
389 if (test_facility(27))
367 S390_lowcore.machine_flags |= MACHINE_FLAG_MVCOS; 390 S390_lowcore.machine_flags |= MACHINE_FLAG_MVCOS;
368 if ((stfle(&facility_bits, 1) > 0) && 391 if (test_facility(40))
369 (facility_bits & (1ULL << (63 - 40))))
370 S390_lowcore.machine_flags |= MACHINE_FLAG_SPP; 392 S390_lowcore.machine_flags |= MACHINE_FLAG_SPP;
371#endif 393#endif
372} 394}
@@ -447,6 +469,7 @@ void __init startup_init(void)
447 lockdep_off(); 469 lockdep_off();
448 sort_main_extable(); 470 sort_main_extable();
449 setup_lowcore_early(); 471 setup_lowcore_early();
472 setup_facility_list();
450 detect_machine_type(); 473 detect_machine_type();
451 ipl_update_parameters(); 474 ipl_update_parameters();
452 setup_boot_command_line(); 475 setup_boot_command_line();
diff --git a/arch/s390/kernel/entry.S b/arch/s390/kernel/entry.S
index bea9ee37ac9d..0476174dfff5 100644
--- a/arch/s390/kernel/entry.S
+++ b/arch/s390/kernel/entry.S
@@ -9,7 +9,6 @@
9 * Heiko Carstens <heiko.carstens@de.ibm.com> 9 * Heiko Carstens <heiko.carstens@de.ibm.com>
10 */ 10 */
11 11
12#include <linux/sys.h>
13#include <linux/linkage.h> 12#include <linux/linkage.h>
14#include <linux/init.h> 13#include <linux/init.h>
15#include <asm/cache.h> 14#include <asm/cache.h>
@@ -49,7 +48,7 @@ SP_SVCNR = STACK_FRAME_OVERHEAD + __PT_SVCNR
49SP_SIZE = STACK_FRAME_OVERHEAD + __PT_SIZE 48SP_SIZE = STACK_FRAME_OVERHEAD + __PT_SIZE
50 49
51_TIF_WORK_SVC = (_TIF_SIGPENDING | _TIF_NOTIFY_RESUME | _TIF_NEED_RESCHED | \ 50_TIF_WORK_SVC = (_TIF_SIGPENDING | _TIF_NOTIFY_RESUME | _TIF_NEED_RESCHED | \
52 _TIF_MCCK_PENDING | _TIF_RESTART_SVC | _TIF_SINGLE_STEP ) 51 _TIF_MCCK_PENDING | _TIF_RESTART_SVC | _TIF_PER_TRAP )
53_TIF_WORK_INT = (_TIF_SIGPENDING | _TIF_NOTIFY_RESUME | _TIF_NEED_RESCHED | \ 52_TIF_WORK_INT = (_TIF_SIGPENDING | _TIF_NOTIFY_RESUME | _TIF_NEED_RESCHED | \
54 _TIF_MCCK_PENDING) 53 _TIF_MCCK_PENDING)
55_TIF_SYSCALL = (_TIF_SYSCALL_TRACE>>8 | _TIF_SYSCALL_AUDIT>>8 | \ 54_TIF_SYSCALL = (_TIF_SYSCALL_TRACE>>8 | _TIF_SYSCALL_AUDIT>>8 | \
@@ -72,25 +71,9 @@ STACK_SIZE = 1 << STACK_SHIFT
72 l %r1,BASED(.Ltrace_irq_off_caller) 71 l %r1,BASED(.Ltrace_irq_off_caller)
73 basr %r14,%r1 72 basr %r14,%r1
74 .endm 73 .endm
75
76 .macro TRACE_IRQS_CHECK_ON
77 tm SP_PSW(%r15),0x03 # irqs enabled?
78 bz BASED(0f)
79 TRACE_IRQS_ON
800:
81 .endm
82
83 .macro TRACE_IRQS_CHECK_OFF
84 tm SP_PSW(%r15),0x03 # irqs enabled?
85 bz BASED(0f)
86 TRACE_IRQS_OFF
870:
88 .endm
89#else 74#else
90#define TRACE_IRQS_ON 75#define TRACE_IRQS_ON
91#define TRACE_IRQS_OFF 76#define TRACE_IRQS_OFF
92#define TRACE_IRQS_CHECK_ON
93#define TRACE_IRQS_CHECK_OFF
94#endif 77#endif
95 78
96#ifdef CONFIG_LOCKDEP 79#ifdef CONFIG_LOCKDEP
@@ -126,31 +109,36 @@ STACK_SIZE = 1 << STACK_SHIFT
1261: stm %r10,%r11,\lc_sum 1091: stm %r10,%r11,\lc_sum
127 .endm 110 .endm
128 111
129 .macro SAVE_ALL_BASE savearea 112 .macro SAVE_ALL_SVC psworg,savearea
130 stm %r12,%r15,\savearea 113 stm %r12,%r15,\savearea
131 l %r13,__LC_SVC_NEW_PSW+4 # load &system_call to %r13 114 l %r13,__LC_SVC_NEW_PSW+4 # load &system_call to %r13
115 l %r15,__LC_KERNEL_STACK # problem state -> load ksp
116 s %r15,BASED(.Lc_spsize) # make room for registers & psw
132 .endm 117 .endm
133 118
134 .macro SAVE_ALL_SVC psworg,savearea 119 .macro SAVE_ALL_BASE savearea
135 la %r12,\psworg 120 stm %r12,%r15,\savearea
136 l %r15,__LC_KERNEL_STACK # problem state -> load ksp 121 l %r13,__LC_SVC_NEW_PSW+4 # load &system_call to %r13
137 .endm 122 .endm
138 123
139 .macro SAVE_ALL_SYNC psworg,savearea 124 .macro SAVE_ALL_PGM psworg,savearea
140 la %r12,\psworg
141 tm \psworg+1,0x01 # test problem state bit 125 tm \psworg+1,0x01 # test problem state bit
142 bz BASED(2f) # skip stack setup save
143 l %r15,__LC_KERNEL_STACK # problem state -> load ksp
144#ifdef CONFIG_CHECK_STACK 126#ifdef CONFIG_CHECK_STACK
145 b BASED(3f) 127 bnz BASED(1f)
1462: tml %r15,STACK_SIZE - CONFIG_STACK_GUARD 128 tml %r15,STACK_SIZE - CONFIG_STACK_GUARD
147 bz BASED(stack_overflow) 129 bnz BASED(2f)
1483: 130 la %r12,\psworg
131 b BASED(stack_overflow)
132#else
133 bz BASED(2f)
149#endif 134#endif
1502: 1351: l %r15,__LC_KERNEL_STACK # problem state -> load ksp
1362: s %r15,BASED(.Lc_spsize) # make room for registers & psw
151 .endm 137 .endm
152 138
153 .macro SAVE_ALL_ASYNC psworg,savearea 139 .macro SAVE_ALL_ASYNC psworg,savearea
140 stm %r12,%r15,\savearea
141 l %r13,__LC_SVC_NEW_PSW+4 # load &system_call to %r13
154 la %r12,\psworg 142 la %r12,\psworg
155 tm \psworg+1,0x01 # test problem state bit 143 tm \psworg+1,0x01 # test problem state bit
156 bnz BASED(1f) # from user -> load async stack 144 bnz BASED(1f) # from user -> load async stack
@@ -165,27 +153,23 @@ STACK_SIZE = 1 << STACK_SHIFT
1650: l %r14,__LC_ASYNC_STACK # are we already on the async stack ? 1530: l %r14,__LC_ASYNC_STACK # are we already on the async stack ?
166 slr %r14,%r15 154 slr %r14,%r15
167 sra %r14,STACK_SHIFT 155 sra %r14,STACK_SHIFT
168 be BASED(2f)
1691: l %r15,__LC_ASYNC_STACK
170#ifdef CONFIG_CHECK_STACK 156#ifdef CONFIG_CHECK_STACK
171 b BASED(3f) 157 bnz BASED(1f)
1722: tml %r15,STACK_SIZE - CONFIG_STACK_GUARD 158 tml %r15,STACK_SIZE - CONFIG_STACK_GUARD
173 bz BASED(stack_overflow) 159 bnz BASED(2f)
1743: 160 b BASED(stack_overflow)
161#else
162 bz BASED(2f)
175#endif 163#endif
1762: 1641: l %r15,__LC_ASYNC_STACK
1652: s %r15,BASED(.Lc_spsize) # make room for registers & psw
177 .endm 166 .endm
178 167
179 .macro CREATE_STACK_FRAME psworg,savearea 168 .macro CREATE_STACK_FRAME savearea
180 s %r15,BASED(.Lc_spsize) # make room for registers & psw 169 xc __SF_BACKCHAIN(4,%r15),__SF_BACKCHAIN(%r15)
181 mvc SP_PSW(8,%r15),0(%r12) # move user PSW to stack
182 st %r2,SP_ORIG_R2(%r15) # store original content of gpr 2 170 st %r2,SP_ORIG_R2(%r15) # store original content of gpr 2
183 icm %r12,12,__LC_SVC_ILC
184 stm %r0,%r11,SP_R0(%r15) # store gprs %r0-%r11 to kernel stack
185 st %r12,SP_ILC(%r15)
186 mvc SP_R12(16,%r15),\savearea # move %r12-%r15 to stack 171 mvc SP_R12(16,%r15),\savearea # move %r12-%r15 to stack
187 la %r12,0 172 stm %r0,%r11,SP_R0(%r15) # store gprs %r0-%r11 to kernel stack
188 st %r12,__SF_BACKCHAIN(%r15) # clear back chain
189 .endm 173 .endm
190 174
191 .macro RESTORE_ALL psworg,sync 175 .macro RESTORE_ALL psworg,sync
@@ -198,6 +182,14 @@ STACK_SIZE = 1 << STACK_SHIFT
198 lpsw \psworg # back to caller 182 lpsw \psworg # back to caller
199 .endm 183 .endm
200 184
185 .macro REENABLE_IRQS
186 mvc __SF_EMPTY(1,%r15),SP_PSW(%r15)
187 ni __SF_EMPTY(%r15),0xbf
188 ssm __SF_EMPTY(%r15)
189 .endm
190
191 .section .kprobes.text, "ax"
192
201/* 193/*
202 * Scheduler resume function, called by switch_to 194 * Scheduler resume function, called by switch_to
203 * gpr2 = (task_struct *) prev 195 * gpr2 = (task_struct *) prev
@@ -208,31 +200,22 @@ STACK_SIZE = 1 << STACK_SHIFT
208 .globl __switch_to 200 .globl __switch_to
209__switch_to: 201__switch_to:
210 basr %r1,0 202 basr %r1,0
211__switch_to_base: 2030: l %r4,__THREAD_info(%r2) # get thread_info of prev
212 tm __THREAD_per(%r3),0xe8 # new process is using per ? 204 l %r5,__THREAD_info(%r3) # get thread_info of next
213 bz __switch_to_noper-__switch_to_base(%r1) # if not we're fine
214 stctl %c9,%c11,__SF_EMPTY(%r15) # We are using per stuff
215 clc __THREAD_per(12,%r3),__SF_EMPTY(%r15)
216 be __switch_to_noper-__switch_to_base(%r1) # we got away w/o bashing TLB's
217 lctl %c9,%c11,__THREAD_per(%r3) # Nope we didn't
218__switch_to_noper:
219 l %r4,__THREAD_info(%r2) # get thread_info of prev
220 tm __TI_flags+3(%r4),_TIF_MCCK_PENDING # machine check pending? 205 tm __TI_flags+3(%r4),_TIF_MCCK_PENDING # machine check pending?
221 bz __switch_to_no_mcck-__switch_to_base(%r1) 206 bz 1f-0b(%r1)
222 ni __TI_flags+3(%r4),255-_TIF_MCCK_PENDING # clear flag in prev 207 ni __TI_flags+3(%r4),255-_TIF_MCCK_PENDING # clear flag in prev
223 l %r4,__THREAD_info(%r3) # get thread_info of next 208 oi __TI_flags+3(%r5),_TIF_MCCK_PENDING # set it in next
224 oi __TI_flags+3(%r4),_TIF_MCCK_PENDING # set it in next 2091: stm %r6,%r15,__SF_GPRS(%r15) # store gprs of prev task
225__switch_to_no_mcck: 210 st %r15,__THREAD_ksp(%r2) # store kernel stack of prev
226 stm %r6,%r15,__SF_GPRS(%r15)# store __switch_to registers of prev task 211 l %r15,__THREAD_ksp(%r3) # load kernel stack of next
227 st %r15,__THREAD_ksp(%r2) # store kernel stack to prev->tss.ksp 212 lctl %c4,%c4,__TASK_pid(%r3) # load pid to control reg. 4
228 l %r15,__THREAD_ksp(%r3) # load kernel stack from next->tss.ksp 213 lm %r6,%r15,__SF_GPRS(%r15) # load gprs of next task
229 lm %r6,%r15,__SF_GPRS(%r15)# load __switch_to registers of next task 214 st %r3,__LC_CURRENT # store task struct of next
230 st %r3,__LC_CURRENT # __LC_CURRENT = current task struct 215 mvc __LC_CURRENT_PID(4,%r0),__TASK_pid(%r3) # store pid of next
231 lctl %c4,%c4,__TASK_pid(%r3) # load pid to control reg. 4 216 st %r5,__LC_THREAD_INFO # store thread info of next
232 l %r3,__THREAD_info(%r3) # load thread_info from task struct 217 ahi %r5,STACK_SIZE # end of kernel stack of next
233 st %r3,__LC_THREAD_INFO 218 st %r5,__LC_KERNEL_STACK # store end of kernel stack
234 ahi %r3,STACK_SIZE
235 st %r3,__LC_KERNEL_STACK # __LC_KERNEL_STACK = new kernel stack
236 br %r14 219 br %r14
237 220
238__critical_start: 221__critical_start:
@@ -245,10 +228,11 @@ __critical_start:
245system_call: 228system_call:
246 stpt __LC_SYNC_ENTER_TIMER 229 stpt __LC_SYNC_ENTER_TIMER
247sysc_saveall: 230sysc_saveall:
248 SAVE_ALL_BASE __LC_SAVE_AREA
249 SAVE_ALL_SVC __LC_SVC_OLD_PSW,__LC_SAVE_AREA 231 SAVE_ALL_SVC __LC_SVC_OLD_PSW,__LC_SAVE_AREA
250 CREATE_STACK_FRAME __LC_SVC_OLD_PSW,__LC_SAVE_AREA 232 CREATE_STACK_FRAME __LC_SAVE_AREA
251 lh %r7,0x8a # get svc number from lowcore 233 mvc SP_PSW(8,%r15),__LC_SVC_OLD_PSW
234 mvc SP_ILC(4,%r15),__LC_SVC_ILC
235 l %r12,__LC_THREAD_INFO # load pointer to thread_info struct
252sysc_vtime: 236sysc_vtime:
253 UPDATE_VTIME __LC_EXIT_TIMER,__LC_SYNC_ENTER_TIMER,__LC_USER_TIMER 237 UPDATE_VTIME __LC_EXIT_TIMER,__LC_SYNC_ENTER_TIMER,__LC_USER_TIMER
254sysc_stime: 238sysc_stime:
@@ -256,21 +240,20 @@ sysc_stime:
256sysc_update: 240sysc_update:
257 mvc __LC_LAST_UPDATE_TIMER(8),__LC_SYNC_ENTER_TIMER 241 mvc __LC_LAST_UPDATE_TIMER(8),__LC_SYNC_ENTER_TIMER
258sysc_do_svc: 242sysc_do_svc:
259 l %r9,__LC_THREAD_INFO # load pointer to thread_info struct 243 xr %r7,%r7
260 ltr %r7,%r7 # test for svc 0 244 icm %r7,3,SP_SVCNR(%r15) # load svc number and test for svc 0
261 bnz BASED(sysc_nr_ok) # svc number > 0 245 bnz BASED(sysc_nr_ok) # svc number > 0
262 # svc 0: system call number in %r1 246 # svc 0: system call number in %r1
263 cl %r1,BASED(.Lnr_syscalls) 247 cl %r1,BASED(.Lnr_syscalls)
264 bnl BASED(sysc_nr_ok) 248 bnl BASED(sysc_nr_ok)
249 sth %r1,SP_SVCNR(%r15)
265 lr %r7,%r1 # copy svc number to %r7 250 lr %r7,%r1 # copy svc number to %r7
266sysc_nr_ok: 251sysc_nr_ok:
267 mvc SP_ARGS(4,%r15),SP_R7(%r15)
268sysc_do_restart:
269 sth %r7,SP_SVCNR(%r15)
270 sll %r7,2 # svc number *4 252 sll %r7,2 # svc number *4
271 l %r8,BASED(.Lsysc_table) 253 l %r10,BASED(.Lsysc_table)
272 tm __TI_flags+2(%r9),_TIF_SYSCALL 254 tm __TI_flags+2(%r12),_TIF_SYSCALL
273 l %r8,0(%r7,%r8) # get system call addr. 255 mvc SP_ARGS(4,%r15),SP_R7(%r15)
256 l %r8,0(%r7,%r10) # get system call addr.
274 bnz BASED(sysc_tracesys) 257 bnz BASED(sysc_tracesys)
275 basr %r14,%r8 # call sys_xxxx 258 basr %r14,%r8 # call sys_xxxx
276 st %r2,SP_R2(%r15) # store return value (change R2 on stack) 259 st %r2,SP_R2(%r15) # store return value (change R2 on stack)
@@ -278,7 +261,7 @@ sysc_do_restart:
278sysc_return: 261sysc_return:
279 LOCKDEP_SYS_EXIT 262 LOCKDEP_SYS_EXIT
280sysc_tif: 263sysc_tif:
281 tm __TI_flags+3(%r9),_TIF_WORK_SVC 264 tm __TI_flags+3(%r12),_TIF_WORK_SVC
282 bnz BASED(sysc_work) # there is work to do (signals etc.) 265 bnz BASED(sysc_work) # there is work to do (signals etc.)
283sysc_restore: 266sysc_restore:
284 RESTORE_ALL __LC_RETURN_PSW,1 267 RESTORE_ALL __LC_RETURN_PSW,1
@@ -295,17 +278,17 @@ sysc_work:
295# One of the work bits is on. Find out which one. 278# One of the work bits is on. Find out which one.
296# 279#
297sysc_work_tif: 280sysc_work_tif:
298 tm __TI_flags+3(%r9),_TIF_MCCK_PENDING 281 tm __TI_flags+3(%r12),_TIF_MCCK_PENDING
299 bo BASED(sysc_mcck_pending) 282 bo BASED(sysc_mcck_pending)
300 tm __TI_flags+3(%r9),_TIF_NEED_RESCHED 283 tm __TI_flags+3(%r12),_TIF_NEED_RESCHED
301 bo BASED(sysc_reschedule) 284 bo BASED(sysc_reschedule)
302 tm __TI_flags+3(%r9),_TIF_SIGPENDING 285 tm __TI_flags+3(%r12),_TIF_SIGPENDING
303 bo BASED(sysc_sigpending) 286 bo BASED(sysc_sigpending)
304 tm __TI_flags+3(%r9),_TIF_NOTIFY_RESUME 287 tm __TI_flags+3(%r12),_TIF_NOTIFY_RESUME
305 bo BASED(sysc_notify_resume) 288 bo BASED(sysc_notify_resume)
306 tm __TI_flags+3(%r9),_TIF_RESTART_SVC 289 tm __TI_flags+3(%r12),_TIF_RESTART_SVC
307 bo BASED(sysc_restart) 290 bo BASED(sysc_restart)
308 tm __TI_flags+3(%r9),_TIF_SINGLE_STEP 291 tm __TI_flags+3(%r12),_TIF_PER_TRAP
309 bo BASED(sysc_singlestep) 292 bo BASED(sysc_singlestep)
310 b BASED(sysc_return) # beware of critical section cleanup 293 b BASED(sysc_return) # beware of critical section cleanup
311 294
@@ -329,13 +312,13 @@ sysc_mcck_pending:
329# _TIF_SIGPENDING is set, call do_signal 312# _TIF_SIGPENDING is set, call do_signal
330# 313#
331sysc_sigpending: 314sysc_sigpending:
332 ni __TI_flags+3(%r9),255-_TIF_SINGLE_STEP # clear TIF_SINGLE_STEP 315 ni __TI_flags+3(%r12),255-_TIF_PER_TRAP # clear TIF_PER_TRAP
333 la %r2,SP_PTREGS(%r15) # load pt_regs 316 la %r2,SP_PTREGS(%r15) # load pt_regs
334 l %r1,BASED(.Ldo_signal) 317 l %r1,BASED(.Ldo_signal)
335 basr %r14,%r1 # call do_signal 318 basr %r14,%r1 # call do_signal
336 tm __TI_flags+3(%r9),_TIF_RESTART_SVC 319 tm __TI_flags+3(%r12),_TIF_RESTART_SVC
337 bo BASED(sysc_restart) 320 bo BASED(sysc_restart)
338 tm __TI_flags+3(%r9),_TIF_SINGLE_STEP 321 tm __TI_flags+3(%r12),_TIF_PER_TRAP
339 bo BASED(sysc_singlestep) 322 bo BASED(sysc_singlestep)
340 b BASED(sysc_return) 323 b BASED(sysc_return)
341 324
@@ -353,23 +336,23 @@ sysc_notify_resume:
353# _TIF_RESTART_SVC is set, set up registers and restart svc 336# _TIF_RESTART_SVC is set, set up registers and restart svc
354# 337#
355sysc_restart: 338sysc_restart:
356 ni __TI_flags+3(%r9),255-_TIF_RESTART_SVC # clear TIF_RESTART_SVC 339 ni __TI_flags+3(%r12),255-_TIF_RESTART_SVC # clear TIF_RESTART_SVC
357 l %r7,SP_R2(%r15) # load new svc number 340 l %r7,SP_R2(%r15) # load new svc number
358 mvc SP_R2(4,%r15),SP_ORIG_R2(%r15) # restore first argument 341 mvc SP_R2(4,%r15),SP_ORIG_R2(%r15) # restore first argument
359 lm %r2,%r6,SP_R2(%r15) # load svc arguments 342 lm %r2,%r6,SP_R2(%r15) # load svc arguments
360 b BASED(sysc_do_restart) # restart svc 343 sth %r7,SP_SVCNR(%r15)
344 b BASED(sysc_nr_ok) # restart svc
361 345
362# 346#
363# _TIF_SINGLE_STEP is set, call do_single_step 347# _TIF_PER_TRAP is set, call do_per_trap
364# 348#
365sysc_singlestep: 349sysc_singlestep:
366 ni __TI_flags+3(%r9),255-_TIF_SINGLE_STEP # clear TIF_SINGLE_STEP 350 ni __TI_flags+3(%r12),255-_TIF_PER_TRAP # clear TIF_PER_TRAP
367 mvi SP_SVCNR(%r15),0xff # set trap indication to pgm check 351 xc SP_SVCNR(2,%r15),SP_SVCNR(%r15) # clear svc number
368 mvi SP_SVCNR+1(%r15),0xff
369 la %r2,SP_PTREGS(%r15) # address of register-save area 352 la %r2,SP_PTREGS(%r15) # address of register-save area
370 l %r1,BASED(.Lhandle_per) # load adr. of per handler 353 l %r1,BASED(.Lhandle_per) # load adr. of per handler
371 la %r14,BASED(sysc_return) # load adr. of system return 354 la %r14,BASED(sysc_return) # load adr. of system return
372 br %r1 # branch to do_single_step 355 br %r1 # branch to do_per_trap
373 356
374# 357#
375# call tracehook_report_syscall_entry/tracehook_report_syscall_exit before 358# call tracehook_report_syscall_entry/tracehook_report_syscall_exit before
@@ -379,22 +362,23 @@ sysc_tracesys:
379 l %r1,BASED(.Ltrace_entry) 362 l %r1,BASED(.Ltrace_entry)
380 la %r2,SP_PTREGS(%r15) # load pt_regs 363 la %r2,SP_PTREGS(%r15) # load pt_regs
381 la %r3,0 364 la %r3,0
382 srl %r7,2 365 xr %r0,%r0
383 st %r7,SP_R2(%r15) 366 icm %r0,3,SP_SVCNR(%r15)
367 st %r0,SP_R2(%r15)
384 basr %r14,%r1 368 basr %r14,%r1
385 cl %r2,BASED(.Lnr_syscalls) 369 cl %r2,BASED(.Lnr_syscalls)
386 bnl BASED(sysc_tracenogo) 370 bnl BASED(sysc_tracenogo)
387 l %r8,BASED(.Lsysc_table)
388 lr %r7,%r2 371 lr %r7,%r2
389 sll %r7,2 # svc number *4 372 sll %r7,2 # svc number *4
390 l %r8,0(%r7,%r8) 373 l %r8,0(%r7,%r10)
391sysc_tracego: 374sysc_tracego:
392 lm %r3,%r6,SP_R3(%r15) 375 lm %r3,%r6,SP_R3(%r15)
376 mvc SP_ARGS(4,%r15),SP_R7(%r15)
393 l %r2,SP_ORIG_R2(%r15) 377 l %r2,SP_ORIG_R2(%r15)
394 basr %r14,%r8 # call sys_xxx 378 basr %r14,%r8 # call sys_xxx
395 st %r2,SP_R2(%r15) # store return value 379 st %r2,SP_R2(%r15) # store return value
396sysc_tracenogo: 380sysc_tracenogo:
397 tm __TI_flags+2(%r9),_TIF_SYSCALL 381 tm __TI_flags+2(%r12),_TIF_SYSCALL
398 bz BASED(sysc_return) 382 bz BASED(sysc_return)
399 l %r1,BASED(.Ltrace_exit) 383 l %r1,BASED(.Ltrace_exit)
400 la %r2,SP_PTREGS(%r15) # load pt_regs 384 la %r2,SP_PTREGS(%r15) # load pt_regs
@@ -407,7 +391,7 @@ sysc_tracenogo:
407 .globl ret_from_fork 391 .globl ret_from_fork
408ret_from_fork: 392ret_from_fork:
409 l %r13,__LC_SVC_NEW_PSW+4 393 l %r13,__LC_SVC_NEW_PSW+4
410 l %r9,__LC_THREAD_INFO # load pointer to thread_info struct 394 l %r12,__LC_THREAD_INFO # load pointer to thread_info struct
411 tm SP_PSW+1(%r15),0x01 # forking a kernel thread ? 395 tm SP_PSW+1(%r15),0x01 # forking a kernel thread ?
412 bo BASED(0f) 396 bo BASED(0f)
413 st %r15,SP_R15(%r15) # store stack pointer for new kthread 397 st %r15,SP_R15(%r15) # store stack pointer for new kthread
@@ -440,13 +424,11 @@ kernel_execve:
440 br %r14 424 br %r14
441 # execve succeeded. 425 # execve succeeded.
4420: stnsm __SF_EMPTY(%r15),0xfc # disable interrupts 4260: stnsm __SF_EMPTY(%r15),0xfc # disable interrupts
443 TRACE_IRQS_OFF
444 l %r15,__LC_KERNEL_STACK # load ksp 427 l %r15,__LC_KERNEL_STACK # load ksp
445 s %r15,BASED(.Lc_spsize) # make room for registers & psw 428 s %r15,BASED(.Lc_spsize) # make room for registers & psw
446 l %r9,__LC_THREAD_INFO
447 mvc SP_PTREGS(__PT_SIZE,%r15),0(%r12) # copy pt_regs 429 mvc SP_PTREGS(__PT_SIZE,%r15),0(%r12) # copy pt_regs
430 l %r12,__LC_THREAD_INFO
448 xc __SF_BACKCHAIN(4,%r15),__SF_BACKCHAIN(%r15) 431 xc __SF_BACKCHAIN(4,%r15),__SF_BACKCHAIN(%r15)
449 TRACE_IRQS_ON
450 stosm __SF_EMPTY(%r15),0x03 # reenable interrupts 432 stosm __SF_EMPTY(%r15),0x03 # reenable interrupts
451 l %r1,BASED(.Lexecve_tail) 433 l %r1,BASED(.Lexecve_tail)
452 basr %r14,%r1 434 basr %r14,%r1
@@ -475,27 +457,28 @@ pgm_check_handler:
475 SAVE_ALL_BASE __LC_SAVE_AREA 457 SAVE_ALL_BASE __LC_SAVE_AREA
476 tm __LC_PGM_INT_CODE+1,0x80 # check whether we got a per exception 458 tm __LC_PGM_INT_CODE+1,0x80 # check whether we got a per exception
477 bnz BASED(pgm_per) # got per exception -> special case 459 bnz BASED(pgm_per) # got per exception -> special case
478 SAVE_ALL_SYNC __LC_PGM_OLD_PSW,__LC_SAVE_AREA 460 SAVE_ALL_PGM __LC_PGM_OLD_PSW,__LC_SAVE_AREA
479 CREATE_STACK_FRAME __LC_PGM_OLD_PSW,__LC_SAVE_AREA 461 CREATE_STACK_FRAME __LC_SAVE_AREA
462 xc SP_ILC(4,%r15),SP_ILC(%r15)
463 mvc SP_PSW(8,%r15),__LC_PGM_OLD_PSW
464 l %r12,__LC_THREAD_INFO # load pointer to thread_info struct
480 tm SP_PSW+1(%r15),0x01 # interrupting from user ? 465 tm SP_PSW+1(%r15),0x01 # interrupting from user ?
481 bz BASED(pgm_no_vtime) 466 bz BASED(pgm_no_vtime)
482 UPDATE_VTIME __LC_EXIT_TIMER,__LC_SYNC_ENTER_TIMER,__LC_USER_TIMER 467 UPDATE_VTIME __LC_EXIT_TIMER,__LC_SYNC_ENTER_TIMER,__LC_USER_TIMER
483 UPDATE_VTIME __LC_LAST_UPDATE_TIMER,__LC_EXIT_TIMER,__LC_SYSTEM_TIMER 468 UPDATE_VTIME __LC_LAST_UPDATE_TIMER,__LC_EXIT_TIMER,__LC_SYSTEM_TIMER
484 mvc __LC_LAST_UPDATE_TIMER(8),__LC_SYNC_ENTER_TIMER 469 mvc __LC_LAST_UPDATE_TIMER(8),__LC_SYNC_ENTER_TIMER
485pgm_no_vtime: 470pgm_no_vtime:
486 TRACE_IRQS_CHECK_OFF
487 l %r9,__LC_THREAD_INFO # load pointer to thread_info struct
488 l %r3,__LC_PGM_ILC # load program interruption code 471 l %r3,__LC_PGM_ILC # load program interruption code
472 l %r4,__LC_TRANS_EXC_CODE
473 REENABLE_IRQS
489 la %r8,0x7f 474 la %r8,0x7f
490 nr %r8,%r3 475 nr %r8,%r3
491pgm_do_call:
492 l %r7,BASED(.Ljump_table)
493 sll %r8,2 476 sll %r8,2
494 l %r7,0(%r8,%r7) # load address of handler routine 477 l %r1,BASED(.Ljump_table)
478 l %r1,0(%r8,%r1) # load address of handler routine
495 la %r2,SP_PTREGS(%r15) # address of register-save area 479 la %r2,SP_PTREGS(%r15) # address of register-save area
496 basr %r14,%r7 # branch to interrupt-handler 480 basr %r14,%r1 # branch to interrupt-handler
497pgm_exit: 481pgm_exit:
498 TRACE_IRQS_CHECK_ON
499 b BASED(sysc_return) 482 b BASED(sysc_return)
500 483
501# 484#
@@ -515,55 +498,54 @@ pgm_per:
515# Normal per exception 498# Normal per exception
516# 499#
517pgm_per_std: 500pgm_per_std:
518 SAVE_ALL_SYNC __LC_PGM_OLD_PSW,__LC_SAVE_AREA 501 SAVE_ALL_PGM __LC_PGM_OLD_PSW,__LC_SAVE_AREA
519 CREATE_STACK_FRAME __LC_PGM_OLD_PSW,__LC_SAVE_AREA 502 CREATE_STACK_FRAME __LC_SAVE_AREA
503 mvc SP_PSW(8,%r15),__LC_PGM_OLD_PSW
504 l %r12,__LC_THREAD_INFO # load pointer to thread_info struct
520 tm SP_PSW+1(%r15),0x01 # interrupting from user ? 505 tm SP_PSW+1(%r15),0x01 # interrupting from user ?
521 bz BASED(pgm_no_vtime2) 506 bz BASED(pgm_no_vtime2)
522 UPDATE_VTIME __LC_EXIT_TIMER,__LC_SYNC_ENTER_TIMER,__LC_USER_TIMER 507 UPDATE_VTIME __LC_EXIT_TIMER,__LC_SYNC_ENTER_TIMER,__LC_USER_TIMER
523 UPDATE_VTIME __LC_LAST_UPDATE_TIMER,__LC_EXIT_TIMER,__LC_SYSTEM_TIMER 508 UPDATE_VTIME __LC_LAST_UPDATE_TIMER,__LC_EXIT_TIMER,__LC_SYSTEM_TIMER
524 mvc __LC_LAST_UPDATE_TIMER(8),__LC_SYNC_ENTER_TIMER 509 mvc __LC_LAST_UPDATE_TIMER(8),__LC_SYNC_ENTER_TIMER
525pgm_no_vtime2: 510pgm_no_vtime2:
526 TRACE_IRQS_CHECK_OFF 511 l %r1,__TI_task(%r12)
527 l %r9,__LC_THREAD_INFO # load pointer to thread_info struct
528 l %r1,__TI_task(%r9)
529 tm SP_PSW+1(%r15),0x01 # kernel per event ? 512 tm SP_PSW+1(%r15),0x01 # kernel per event ?
530 bz BASED(kernel_per) 513 bz BASED(kernel_per)
531 mvc __THREAD_per+__PER_atmid(2,%r1),__LC_PER_ATMID 514 mvc __THREAD_per_cause(2,%r1),__LC_PER_CAUSE
532 mvc __THREAD_per+__PER_address(4,%r1),__LC_PER_ADDRESS 515 mvc __THREAD_per_address(4,%r1),__LC_PER_ADDRESS
533 mvc __THREAD_per+__PER_access_id(1,%r1),__LC_PER_ACCESS_ID 516 mvc __THREAD_per_paid(1,%r1),__LC_PER_PAID
534 oi __TI_flags+3(%r9),_TIF_SINGLE_STEP # set TIF_SINGLE_STEP 517 oi __TI_flags+3(%r12),_TIF_PER_TRAP # set TIF_PER_TRAP
535 l %r3,__LC_PGM_ILC # load program interruption code 518 l %r3,__LC_PGM_ILC # load program interruption code
519 l %r4,__LC_TRANS_EXC_CODE
520 REENABLE_IRQS
536 la %r8,0x7f 521 la %r8,0x7f
537 nr %r8,%r3 # clear per-event-bit and ilc 522 nr %r8,%r3 # clear per-event-bit and ilc
538 be BASED(pgm_exit2) # only per or per+check ? 523 be BASED(pgm_exit2) # only per or per+check ?
539 l %r7,BASED(.Ljump_table)
540 sll %r8,2 524 sll %r8,2
541 l %r7,0(%r8,%r7) # load address of handler routine 525 l %r1,BASED(.Ljump_table)
526 l %r1,0(%r8,%r1) # load address of handler routine
542 la %r2,SP_PTREGS(%r15) # address of register-save area 527 la %r2,SP_PTREGS(%r15) # address of register-save area
543 basr %r14,%r7 # branch to interrupt-handler 528 basr %r14,%r1 # branch to interrupt-handler
544pgm_exit2: 529pgm_exit2:
545 TRACE_IRQS_ON
546 stosm __SF_EMPTY(%r15),0x03 # reenable interrupts
547 b BASED(sysc_return) 530 b BASED(sysc_return)
548 531
549# 532#
550# it was a single stepped SVC that is causing all the trouble 533# it was a single stepped SVC that is causing all the trouble
551# 534#
552pgm_svcper: 535pgm_svcper:
553 SAVE_ALL_SYNC __LC_SVC_OLD_PSW,__LC_SAVE_AREA 536 SAVE_ALL_PGM __LC_SVC_OLD_PSW,__LC_SAVE_AREA
554 CREATE_STACK_FRAME __LC_SVC_OLD_PSW,__LC_SAVE_AREA 537 CREATE_STACK_FRAME __LC_SAVE_AREA
538 mvc SP_PSW(8,%r15),__LC_SVC_OLD_PSW
539 mvc SP_ILC(4,%r15),__LC_SVC_ILC
540 l %r12,__LC_THREAD_INFO # load pointer to thread_info struct
555 UPDATE_VTIME __LC_EXIT_TIMER,__LC_SYNC_ENTER_TIMER,__LC_USER_TIMER 541 UPDATE_VTIME __LC_EXIT_TIMER,__LC_SYNC_ENTER_TIMER,__LC_USER_TIMER
556 UPDATE_VTIME __LC_LAST_UPDATE_TIMER,__LC_EXIT_TIMER,__LC_SYSTEM_TIMER 542 UPDATE_VTIME __LC_LAST_UPDATE_TIMER,__LC_EXIT_TIMER,__LC_SYSTEM_TIMER
557 mvc __LC_LAST_UPDATE_TIMER(8),__LC_SYNC_ENTER_TIMER 543 mvc __LC_LAST_UPDATE_TIMER(8),__LC_SYNC_ENTER_TIMER
558 lh %r7,0x8a # get svc number from lowcore 544 l %r8,__TI_task(%r12)
559 l %r9,__LC_THREAD_INFO # load pointer to thread_info struct 545 mvc __THREAD_per_cause(2,%r8),__LC_PER_CAUSE
560 TRACE_IRQS_OFF 546 mvc __THREAD_per_address(4,%r8),__LC_PER_ADDRESS
561 l %r8,__TI_task(%r9) 547 mvc __THREAD_per_paid(1,%r8),__LC_PER_PAID
562 mvc __THREAD_per+__PER_atmid(2,%r8),__LC_PER_ATMID 548 oi __TI_flags+3(%r12),_TIF_PER_TRAP # set TIF_PER_TRAP
563 mvc __THREAD_per+__PER_address(4,%r8),__LC_PER_ADDRESS
564 mvc __THREAD_per+__PER_access_id(1,%r8),__LC_PER_ACCESS_ID
565 oi __TI_flags+3(%r9),_TIF_SINGLE_STEP # set TIF_SINGLE_STEP
566 TRACE_IRQS_ON
567 stosm __SF_EMPTY(%r15),0x03 # reenable interrupts 549 stosm __SF_EMPTY(%r15),0x03 # reenable interrupts
568 lm %r2,%r6,SP_R2(%r15) # load svc arguments 550 lm %r2,%r6,SP_R2(%r15) # load svc arguments
569 b BASED(sysc_do_svc) 551 b BASED(sysc_do_svc)
@@ -572,8 +554,8 @@ pgm_svcper:
572# per was called from kernel, must be kprobes 554# per was called from kernel, must be kprobes
573# 555#
574kernel_per: 556kernel_per:
575 mvi SP_SVCNR(%r15),0xff # set trap indication to pgm check 557 REENABLE_IRQS
576 mvi SP_SVCNR+1(%r15),0xff 558 xc SP_SVCNR(2,%r15),SP_SVCNR(%r15)
577 la %r2,SP_PTREGS(%r15) # address of register-save area 559 la %r2,SP_PTREGS(%r15) # address of register-save area
578 l %r1,BASED(.Lhandle_per) # load adr. of per handler 560 l %r1,BASED(.Lhandle_per) # load adr. of per handler
579 basr %r14,%r1 # branch to do_single_step 561 basr %r14,%r1 # branch to do_single_step
@@ -587,9 +569,10 @@ kernel_per:
587io_int_handler: 569io_int_handler:
588 stck __LC_INT_CLOCK 570 stck __LC_INT_CLOCK
589 stpt __LC_ASYNC_ENTER_TIMER 571 stpt __LC_ASYNC_ENTER_TIMER
590 SAVE_ALL_BASE __LC_SAVE_AREA+16
591 SAVE_ALL_ASYNC __LC_IO_OLD_PSW,__LC_SAVE_AREA+16 572 SAVE_ALL_ASYNC __LC_IO_OLD_PSW,__LC_SAVE_AREA+16
592 CREATE_STACK_FRAME __LC_IO_OLD_PSW,__LC_SAVE_AREA+16 573 CREATE_STACK_FRAME __LC_SAVE_AREA+16
574 mvc SP_PSW(8,%r15),0(%r12) # move user PSW to stack
575 l %r12,__LC_THREAD_INFO # load pointer to thread_info struct
593 tm SP_PSW+1(%r15),0x01 # interrupting from user ? 576 tm SP_PSW+1(%r15),0x01 # interrupting from user ?
594 bz BASED(io_no_vtime) 577 bz BASED(io_no_vtime)
595 UPDATE_VTIME __LC_EXIT_TIMER,__LC_ASYNC_ENTER_TIMER,__LC_USER_TIMER 578 UPDATE_VTIME __LC_EXIT_TIMER,__LC_ASYNC_ENTER_TIMER,__LC_USER_TIMER
@@ -597,7 +580,6 @@ io_int_handler:
597 mvc __LC_LAST_UPDATE_TIMER(8),__LC_ASYNC_ENTER_TIMER 580 mvc __LC_LAST_UPDATE_TIMER(8),__LC_ASYNC_ENTER_TIMER
598io_no_vtime: 581io_no_vtime:
599 TRACE_IRQS_OFF 582 TRACE_IRQS_OFF
600 l %r9,__LC_THREAD_INFO # load pointer to thread_info struct
601 l %r1,BASED(.Ldo_IRQ) # load address of do_IRQ 583 l %r1,BASED(.Ldo_IRQ) # load address of do_IRQ
602 la %r2,SP_PTREGS(%r15) # address of register-save area 584 la %r2,SP_PTREGS(%r15) # address of register-save area
603 basr %r14,%r1 # branch to standard irq handler 585 basr %r14,%r1 # branch to standard irq handler
@@ -605,7 +587,7 @@ io_return:
605 LOCKDEP_SYS_EXIT 587 LOCKDEP_SYS_EXIT
606 TRACE_IRQS_ON 588 TRACE_IRQS_ON
607io_tif: 589io_tif:
608 tm __TI_flags+3(%r9),_TIF_WORK_INT 590 tm __TI_flags+3(%r12),_TIF_WORK_INT
609 bnz BASED(io_work) # there is work to do (signals etc.) 591 bnz BASED(io_work) # there is work to do (signals etc.)
610io_restore: 592io_restore:
611 RESTORE_ALL __LC_RETURN_PSW,0 593 RESTORE_ALL __LC_RETURN_PSW,0
@@ -623,9 +605,9 @@ io_work:
623 bo BASED(io_work_user) # yes -> do resched & signal 605 bo BASED(io_work_user) # yes -> do resched & signal
624#ifdef CONFIG_PREEMPT 606#ifdef CONFIG_PREEMPT
625 # check for preemptive scheduling 607 # check for preemptive scheduling
626 icm %r0,15,__TI_precount(%r9) 608 icm %r0,15,__TI_precount(%r12)
627 bnz BASED(io_restore) # preemption disabled 609 bnz BASED(io_restore) # preemption disabled
628 tm __TI_flags+3(%r9),_TIF_NEED_RESCHED 610 tm __TI_flags+3(%r12),_TIF_NEED_RESCHED
629 bno BASED(io_restore) 611 bno BASED(io_restore)
630 # switch to kernel stack 612 # switch to kernel stack
631 l %r1,SP_R15(%r15) 613 l %r1,SP_R15(%r15)
@@ -659,13 +641,13 @@ io_work_user:
659# and _TIF_MCCK_PENDING 641# and _TIF_MCCK_PENDING
660# 642#
661io_work_tif: 643io_work_tif:
662 tm __TI_flags+3(%r9),_TIF_MCCK_PENDING 644 tm __TI_flags+3(%r12),_TIF_MCCK_PENDING
663 bo BASED(io_mcck_pending) 645 bo BASED(io_mcck_pending)
664 tm __TI_flags+3(%r9),_TIF_NEED_RESCHED 646 tm __TI_flags+3(%r12),_TIF_NEED_RESCHED
665 bo BASED(io_reschedule) 647 bo BASED(io_reschedule)
666 tm __TI_flags+3(%r9),_TIF_SIGPENDING 648 tm __TI_flags+3(%r12),_TIF_SIGPENDING
667 bo BASED(io_sigpending) 649 bo BASED(io_sigpending)
668 tm __TI_flags+3(%r9),_TIF_NOTIFY_RESUME 650 tm __TI_flags+3(%r12),_TIF_NOTIFY_RESUME
669 bo BASED(io_notify_resume) 651 bo BASED(io_notify_resume)
670 b BASED(io_return) # beware of critical section cleanup 652 b BASED(io_return) # beware of critical section cleanup
671 653
@@ -725,19 +707,20 @@ io_notify_resume:
725ext_int_handler: 707ext_int_handler:
726 stck __LC_INT_CLOCK 708 stck __LC_INT_CLOCK
727 stpt __LC_ASYNC_ENTER_TIMER 709 stpt __LC_ASYNC_ENTER_TIMER
728 SAVE_ALL_BASE __LC_SAVE_AREA+16
729 SAVE_ALL_ASYNC __LC_EXT_OLD_PSW,__LC_SAVE_AREA+16 710 SAVE_ALL_ASYNC __LC_EXT_OLD_PSW,__LC_SAVE_AREA+16
730 CREATE_STACK_FRAME __LC_EXT_OLD_PSW,__LC_SAVE_AREA+16 711 CREATE_STACK_FRAME __LC_SAVE_AREA+16
712 mvc SP_PSW(8,%r15),0(%r12) # move user PSW to stack
713 l %r12,__LC_THREAD_INFO # load pointer to thread_info struct
731 tm SP_PSW+1(%r15),0x01 # interrupting from user ? 714 tm SP_PSW+1(%r15),0x01 # interrupting from user ?
732 bz BASED(ext_no_vtime) 715 bz BASED(ext_no_vtime)
733 UPDATE_VTIME __LC_EXIT_TIMER,__LC_ASYNC_ENTER_TIMER,__LC_USER_TIMER 716 UPDATE_VTIME __LC_EXIT_TIMER,__LC_ASYNC_ENTER_TIMER,__LC_USER_TIMER
734 UPDATE_VTIME __LC_LAST_UPDATE_TIMER,__LC_EXIT_TIMER,__LC_SYSTEM_TIMER 717 UPDATE_VTIME __LC_LAST_UPDATE_TIMER,__LC_EXIT_TIMER,__LC_SYSTEM_TIMER
735 mvc __LC_LAST_UPDATE_TIMER(8),__LC_ASYNC_ENTER_TIMER 718 mvc __LC_LAST_UPDATE_TIMER(8),__LC_ASYNC_ENTER_TIMER
736ext_no_vtime: 719ext_no_vtime:
737 l %r9,__LC_THREAD_INFO # load pointer to thread_info struct
738 TRACE_IRQS_OFF 720 TRACE_IRQS_OFF
739 la %r2,SP_PTREGS(%r15) # address of register-save area 721 la %r2,SP_PTREGS(%r15) # address of register-save area
740 lh %r3,__LC_EXT_INT_CODE # get interruption code 722 l %r3,__LC_CPU_ADDRESS # get cpu address + interruption code
723 l %r4,__LC_EXT_PARAMS # get external parameters
741 l %r1,BASED(.Ldo_extint) 724 l %r1,BASED(.Ldo_extint)
742 basr %r14,%r1 725 basr %r14,%r1
743 b BASED(io_return) 726 b BASED(io_return)
@@ -788,7 +771,10 @@ mcck_int_main:
788 sra %r14,PAGE_SHIFT 771 sra %r14,PAGE_SHIFT
789 be BASED(0f) 772 be BASED(0f)
790 l %r15,__LC_PANIC_STACK # load panic stack 773 l %r15,__LC_PANIC_STACK # load panic stack
7910: CREATE_STACK_FRAME __LC_MCK_OLD_PSW,__LC_SAVE_AREA+32 7740: s %r15,BASED(.Lc_spsize) # make room for registers & psw
775 CREATE_STACK_FRAME __LC_SAVE_AREA+32
776 mvc SP_PSW(8,%r15),0(%r12)
777 l %r12,__LC_THREAD_INFO # load pointer to thread_info struct
792 tm __LC_MCCK_CODE+2,0x08 # mwp of old psw valid? 778 tm __LC_MCCK_CODE+2,0x08 # mwp of old psw valid?
793 bno BASED(mcck_no_vtime) # no -> skip cleanup critical 779 bno BASED(mcck_no_vtime) # no -> skip cleanup critical
794 tm SP_PSW+1(%r15),0x01 # interrupting from user ? 780 tm SP_PSW+1(%r15),0x01 # interrupting from user ?
@@ -797,7 +783,6 @@ mcck_int_main:
797 UPDATE_VTIME __LC_LAST_UPDATE_TIMER,__LC_EXIT_TIMER,__LC_SYSTEM_TIMER 783 UPDATE_VTIME __LC_LAST_UPDATE_TIMER,__LC_EXIT_TIMER,__LC_SYSTEM_TIMER
798 mvc __LC_LAST_UPDATE_TIMER(8),__LC_MCCK_ENTER_TIMER 784 mvc __LC_LAST_UPDATE_TIMER(8),__LC_MCCK_ENTER_TIMER
799mcck_no_vtime: 785mcck_no_vtime:
800 l %r9,__LC_THREAD_INFO # load pointer to thread_info struct
801 la %r2,SP_PTREGS(%r15) # load pt_regs 786 la %r2,SP_PTREGS(%r15) # load pt_regs
802 l %r1,BASED(.Ls390_mcck) 787 l %r1,BASED(.Ls390_mcck)
803 basr %r14,%r1 # call machine check handler 788 basr %r14,%r1 # call machine check handler
@@ -809,7 +794,7 @@ mcck_no_vtime:
809 xc __SF_BACKCHAIN(4,%r1),__SF_BACKCHAIN(%r1) # clear back chain 794 xc __SF_BACKCHAIN(4,%r1),__SF_BACKCHAIN(%r1) # clear back chain
810 lr %r15,%r1 795 lr %r15,%r1
811 stosm __SF_EMPTY(%r15),0x04 # turn dat on 796 stosm __SF_EMPTY(%r15),0x04 # turn dat on
812 tm __TI_flags+3(%r9),_TIF_MCCK_PENDING 797 tm __TI_flags+3(%r12),_TIF_MCCK_PENDING
813 bno BASED(mcck_return) 798 bno BASED(mcck_return)
814 TRACE_IRQS_OFF 799 TRACE_IRQS_OFF
815 l %r1,BASED(.Ls390_handle_mcck) 800 l %r1,BASED(.Ls390_handle_mcck)
@@ -852,7 +837,7 @@ restart_base:
852 stosm __SF_EMPTY(%r15),0x04 # now we can turn dat on 837 stosm __SF_EMPTY(%r15),0x04 # now we can turn dat on
853 basr %r14,0 838 basr %r14,0
854 l %r14,restart_addr-.(%r14) 839 l %r14,restart_addr-.(%r14)
855 br %r14 # branch to start_secondary 840 basr %r14,%r14 # branch to start_secondary
856restart_addr: 841restart_addr:
857 .long start_secondary 842 .long start_secondary
858 .align 8 843 .align 8
@@ -874,6 +859,8 @@ restart_crash:
874restart_go: 859restart_go:
875#endif 860#endif
876 861
862 .section .kprobes.text, "ax"
863
877#ifdef CONFIG_CHECK_STACK 864#ifdef CONFIG_CHECK_STACK
878/* 865/*
879 * The synchronous or the asynchronous stack overflowed. We are dead. 866 * The synchronous or the asynchronous stack overflowed. We are dead.
@@ -956,12 +943,13 @@ cleanup_system_call:
956 bh BASED(0f) 943 bh BASED(0f)
957 mvc __LC_SAVE_AREA(16),0(%r12) 944 mvc __LC_SAVE_AREA(16),0(%r12)
9580: st %r13,4(%r12) 9450: st %r13,4(%r12)
959 st %r12,__LC_SAVE_AREA+48 # argh 946 l %r15,__LC_KERNEL_STACK # problem state -> load ksp
960 SAVE_ALL_SYNC __LC_SVC_OLD_PSW,__LC_SAVE_AREA 947 s %r15,BASED(.Lc_spsize) # make room for registers & psw
961 CREATE_STACK_FRAME __LC_SVC_OLD_PSW,__LC_SAVE_AREA
962 l %r12,__LC_SAVE_AREA+48 # argh
963 st %r15,12(%r12) 948 st %r15,12(%r12)
964 lh %r7,0x8a 949 CREATE_STACK_FRAME __LC_SAVE_AREA
950 mvc SP_PSW(8,%r15),__LC_SVC_OLD_PSW
951 mvc SP_ILC(4,%r15),__LC_SVC_ILC
952 mvc 0(4,%r12),__LC_THREAD_INFO
965cleanup_vtime: 953cleanup_vtime:
966 clc __LC_RETURN_PSW+4(4),BASED(cleanup_system_call_insn+12) 954 clc __LC_RETURN_PSW+4(4),BASED(cleanup_system_call_insn+12)
967 bhe BASED(cleanup_stime) 955 bhe BASED(cleanup_stime)
@@ -1059,7 +1047,7 @@ cleanup_io_restore_insn:
1059.Ldo_signal: .long do_signal 1047.Ldo_signal: .long do_signal
1060.Ldo_notify_resume: 1048.Ldo_notify_resume:
1061 .long do_notify_resume 1049 .long do_notify_resume
1062.Lhandle_per: .long do_single_step 1050.Lhandle_per: .long do_per_trap
1063.Ldo_execve: .long do_execve 1051.Ldo_execve: .long do_execve
1064.Lexecve_tail: .long execve_tail 1052.Lexecve_tail: .long execve_tail
1065.Ljump_table: .long pgm_check_table 1053.Ljump_table: .long pgm_check_table
diff --git a/arch/s390/kernel/entry.h b/arch/s390/kernel/entry.h
index ff579b6bde06..17a6f83a2d67 100644
--- a/arch/s390/kernel/entry.h
+++ b/arch/s390/kernel/entry.h
@@ -5,21 +5,21 @@
5#include <linux/signal.h> 5#include <linux/signal.h>
6#include <asm/ptrace.h> 6#include <asm/ptrace.h>
7 7
8typedef void pgm_check_handler_t(struct pt_regs *, long); 8typedef void pgm_check_handler_t(struct pt_regs *, long, unsigned long);
9extern pgm_check_handler_t *pgm_check_table[128]; 9extern pgm_check_handler_t *pgm_check_table[128];
10pgm_check_handler_t do_protection_exception; 10pgm_check_handler_t do_protection_exception;
11pgm_check_handler_t do_dat_exception; 11pgm_check_handler_t do_dat_exception;
12 12
13extern int sysctl_userprocess_debug; 13extern int sysctl_userprocess_debug;
14 14
15void do_single_step(struct pt_regs *regs); 15void do_per_trap(struct pt_regs *regs);
16void syscall_trace(struct pt_regs *regs, int entryexit); 16void syscall_trace(struct pt_regs *regs, int entryexit);
17void kernel_stack_overflow(struct pt_regs * regs); 17void kernel_stack_overflow(struct pt_regs * regs);
18void do_signal(struct pt_regs *regs); 18void do_signal(struct pt_regs *regs);
19int handle_signal32(unsigned long sig, struct k_sigaction *ka, 19int handle_signal32(unsigned long sig, struct k_sigaction *ka,
20 siginfo_t *info, sigset_t *oldset, struct pt_regs *regs); 20 siginfo_t *info, sigset_t *oldset, struct pt_regs *regs);
21 21
22void do_extint(struct pt_regs *regs, unsigned short code); 22void do_extint(struct pt_regs *regs, unsigned int, unsigned int, unsigned long);
23int __cpuinit start_secondary(void *cpuvoid); 23int __cpuinit start_secondary(void *cpuvoid);
24void __init startup_init(void); 24void __init startup_init(void);
25void die(const char * str, struct pt_regs * regs, long err); 25void die(const char * str, struct pt_regs * regs, long err);
diff --git a/arch/s390/kernel/entry64.S b/arch/s390/kernel/entry64.S
index 8bccec15ea90..d61967e2eab0 100644
--- a/arch/s390/kernel/entry64.S
+++ b/arch/s390/kernel/entry64.S
@@ -51,7 +51,7 @@ STACK_SHIFT = PAGE_SHIFT + THREAD_ORDER
51STACK_SIZE = 1 << STACK_SHIFT 51STACK_SIZE = 1 << STACK_SHIFT
52 52
53_TIF_WORK_SVC = (_TIF_SIGPENDING | _TIF_NOTIFY_RESUME | _TIF_NEED_RESCHED | \ 53_TIF_WORK_SVC = (_TIF_SIGPENDING | _TIF_NOTIFY_RESUME | _TIF_NEED_RESCHED | \
54 _TIF_MCCK_PENDING | _TIF_RESTART_SVC | _TIF_SINGLE_STEP ) 54 _TIF_MCCK_PENDING | _TIF_RESTART_SVC | _TIF_PER_TRAP )
55_TIF_WORK_INT = (_TIF_SIGPENDING | _TIF_NOTIFY_RESUME | _TIF_NEED_RESCHED | \ 55_TIF_WORK_INT = (_TIF_SIGPENDING | _TIF_NOTIFY_RESUME | _TIF_NEED_RESCHED | \
56 _TIF_MCCK_PENDING) 56 _TIF_MCCK_PENDING)
57_TIF_SYSCALL = (_TIF_SYSCALL_TRACE>>8 | _TIF_SYSCALL_AUDIT>>8 | \ 57_TIF_SYSCALL = (_TIF_SYSCALL_TRACE>>8 | _TIF_SYSCALL_AUDIT>>8 | \
@@ -79,25 +79,9 @@ _TIF_SYSCALL = (_TIF_SYSCALL_TRACE>>8 | _TIF_SYSCALL_AUDIT>>8 | \
79 basr %r2,%r0 79 basr %r2,%r0
80 brasl %r14,trace_hardirqs_off_caller 80 brasl %r14,trace_hardirqs_off_caller
81 .endm 81 .endm
82
83 .macro TRACE_IRQS_CHECK_ON
84 tm SP_PSW(%r15),0x03 # irqs enabled?
85 jz 0f
86 TRACE_IRQS_ON
870:
88 .endm
89
90 .macro TRACE_IRQS_CHECK_OFF
91 tm SP_PSW(%r15),0x03 # irqs enabled?
92 jz 0f
93 TRACE_IRQS_OFF
940:
95 .endm
96#else 82#else
97#define TRACE_IRQS_ON 83#define TRACE_IRQS_ON
98#define TRACE_IRQS_OFF 84#define TRACE_IRQS_OFF
99#define TRACE_IRQS_CHECK_ON
100#define TRACE_IRQS_CHECK_OFF
101#endif 85#endif
102 86
103#ifdef CONFIG_LOCKDEP 87#ifdef CONFIG_LOCKDEP
@@ -207,6 +191,14 @@ _TIF_SYSCALL = (_TIF_SYSCALL_TRACE>>8 | _TIF_SYSCALL_AUDIT>>8 | \
2070: 1910:
208 .endm 192 .endm
209 193
194 .macro REENABLE_IRQS
195 mvc __SF_EMPTY(1,%r15),SP_PSW(%r15)
196 ni __SF_EMPTY(%r15),0xbf
197 ssm __SF_EMPTY(%r15)
198 .endm
199
200 .section .kprobes.text, "ax"
201
210/* 202/*
211 * Scheduler resume function, called by switch_to 203 * Scheduler resume function, called by switch_to
212 * gpr2 = (task_struct *) prev 204 * gpr2 = (task_struct *) prev
@@ -216,30 +208,22 @@ _TIF_SYSCALL = (_TIF_SYSCALL_TRACE>>8 | _TIF_SYSCALL_AUDIT>>8 | \
216 */ 208 */
217 .globl __switch_to 209 .globl __switch_to
218__switch_to: 210__switch_to:
219 tm __THREAD_per+4(%r3),0xe8 # is the new process using per ? 211 lg %r4,__THREAD_info(%r2) # get thread_info of prev
220 jz __switch_to_noper # if not we're fine 212 lg %r5,__THREAD_info(%r3) # get thread_info of next
221 stctg %c9,%c11,__SF_EMPTY(%r15)# We are using per stuff
222 clc __THREAD_per(24,%r3),__SF_EMPTY(%r15)
223 je __switch_to_noper # we got away without bashing TLB's
224 lctlg %c9,%c11,__THREAD_per(%r3) # Nope we didn't
225__switch_to_noper:
226 lg %r4,__THREAD_info(%r2) # get thread_info of prev
227 tm __TI_flags+7(%r4),_TIF_MCCK_PENDING # machine check pending? 213 tm __TI_flags+7(%r4),_TIF_MCCK_PENDING # machine check pending?
228 jz __switch_to_no_mcck 214 jz 0f
229 ni __TI_flags+7(%r4),255-_TIF_MCCK_PENDING # clear flag in prev 215 ni __TI_flags+7(%r4),255-_TIF_MCCK_PENDING # clear flag in prev
230 lg %r4,__THREAD_info(%r3) # get thread_info of next 216 oi __TI_flags+7(%r5),_TIF_MCCK_PENDING # set it in next
231 oi __TI_flags+7(%r4),_TIF_MCCK_PENDING # set it in next 2170: stmg %r6,%r15,__SF_GPRS(%r15) # store gprs of prev task
232__switch_to_no_mcck: 218 stg %r15,__THREAD_ksp(%r2) # store kernel stack of prev
233 stmg %r6,%r15,__SF_GPRS(%r15)# store __switch_to registers of prev task 219 lg %r15,__THREAD_ksp(%r3) # load kernel stack of next
234 stg %r15,__THREAD_ksp(%r2) # store kernel stack to prev->tss.ksp 220 lctl %c4,%c4,__TASK_pid(%r3) # load pid to control reg. 4
235 lg %r15,__THREAD_ksp(%r3) # load kernel stack from next->tss.ksp 221 lmg %r6,%r15,__SF_GPRS(%r15) # load gprs of next task
236 lmg %r6,%r15,__SF_GPRS(%r15)# load __switch_to registers of next task 222 stg %r3,__LC_CURRENT # store task struct of next
237 stg %r3,__LC_CURRENT # __LC_CURRENT = current task struct 223 mvc __LC_CURRENT_PID+4(4,%r0),__TASK_pid(%r3) # store pid of next
238 lctl %c4,%c4,__TASK_pid(%r3) # load pid to control reg. 4 224 stg %r5,__LC_THREAD_INFO # store thread info of next
239 lg %r3,__THREAD_info(%r3) # load thread_info from task struct 225 aghi %r5,STACK_SIZE # end of kernel stack of next
240 stg %r3,__LC_THREAD_INFO 226 stg %r5,__LC_KERNEL_STACK # store end of kernel stack
241 aghi %r3,STACK_SIZE
242 stg %r3,__LC_KERNEL_STACK # __LC_KERNEL_STACK = new kernel stack
243 br %r14 227 br %r14
244 228
245__critical_start: 229__critical_start:
@@ -256,7 +240,6 @@ sysc_saveall:
256 CREATE_STACK_FRAME __LC_SAVE_AREA 240 CREATE_STACK_FRAME __LC_SAVE_AREA
257 mvc SP_PSW(16,%r15),__LC_SVC_OLD_PSW 241 mvc SP_PSW(16,%r15),__LC_SVC_OLD_PSW
258 mvc SP_ILC(4,%r15),__LC_SVC_ILC 242 mvc SP_ILC(4,%r15),__LC_SVC_ILC
259 stg %r7,SP_ARGS(%r15)
260 lg %r12,__LC_THREAD_INFO # load pointer to thread_info struct 243 lg %r12,__LC_THREAD_INFO # load pointer to thread_info struct
261sysc_vtime: 244sysc_vtime:
262 UPDATE_VTIME __LC_EXIT_TIMER,__LC_SYNC_ENTER_TIMER,__LC_USER_TIMER 245 UPDATE_VTIME __LC_EXIT_TIMER,__LC_SYNC_ENTER_TIMER,__LC_USER_TIMER
@@ -284,6 +267,7 @@ sysc_nr_ok:
284sysc_noemu: 267sysc_noemu:
285#endif 268#endif
286 tm __TI_flags+6(%r12),_TIF_SYSCALL 269 tm __TI_flags+6(%r12),_TIF_SYSCALL
270 mvc SP_ARGS(8,%r15),SP_R7(%r15)
287 lgf %r8,0(%r7,%r10) # load address of system call routine 271 lgf %r8,0(%r7,%r10) # load address of system call routine
288 jnz sysc_tracesys 272 jnz sysc_tracesys
289 basr %r14,%r8 # call sys_xxxx 273 basr %r14,%r8 # call sys_xxxx
@@ -319,7 +303,7 @@ sysc_work_tif:
319 jo sysc_notify_resume 303 jo sysc_notify_resume
320 tm __TI_flags+7(%r12),_TIF_RESTART_SVC 304 tm __TI_flags+7(%r12),_TIF_RESTART_SVC
321 jo sysc_restart 305 jo sysc_restart
322 tm __TI_flags+7(%r12),_TIF_SINGLE_STEP 306 tm __TI_flags+7(%r12),_TIF_PER_TRAP
323 jo sysc_singlestep 307 jo sysc_singlestep
324 j sysc_return # beware of critical section cleanup 308 j sysc_return # beware of critical section cleanup
325 309
@@ -341,12 +325,12 @@ sysc_mcck_pending:
341# _TIF_SIGPENDING is set, call do_signal 325# _TIF_SIGPENDING is set, call do_signal
342# 326#
343sysc_sigpending: 327sysc_sigpending:
344 ni __TI_flags+7(%r12),255-_TIF_SINGLE_STEP # clear TIF_SINGLE_STEP 328 ni __TI_flags+7(%r12),255-_TIF_PER_TRAP # clear TIF_PER_TRAP
345 la %r2,SP_PTREGS(%r15) # load pt_regs 329 la %r2,SP_PTREGS(%r15) # load pt_regs
346 brasl %r14,do_signal # call do_signal 330 brasl %r14,do_signal # call do_signal
347 tm __TI_flags+7(%r12),_TIF_RESTART_SVC 331 tm __TI_flags+7(%r12),_TIF_RESTART_SVC
348 jo sysc_restart 332 jo sysc_restart
349 tm __TI_flags+7(%r12),_TIF_SINGLE_STEP 333 tm __TI_flags+7(%r12),_TIF_PER_TRAP
350 jo sysc_singlestep 334 jo sysc_singlestep
351 j sysc_return 335 j sysc_return
352 336
@@ -371,14 +355,14 @@ sysc_restart:
371 j sysc_nr_ok # restart svc 355 j sysc_nr_ok # restart svc
372 356
373# 357#
374# _TIF_SINGLE_STEP is set, call do_single_step 358# _TIF_PER_TRAP is set, call do_per_trap
375# 359#
376sysc_singlestep: 360sysc_singlestep:
377 ni __TI_flags+7(%r12),255-_TIF_SINGLE_STEP # clear TIF_SINGLE_STEP 361 ni __TI_flags+7(%r12),255-_TIF_PER_TRAP # clear TIF_PER_TRAP
378 xc SP_SVCNR(2,%r15),SP_SVCNR(%r15) # clear svc number 362 xc SP_SVCNR(2,%r15),SP_SVCNR(%r15) # clear svc number
379 la %r2,SP_PTREGS(%r15) # address of register-save area 363 la %r2,SP_PTREGS(%r15) # address of register-save area
380 larl %r14,sysc_return # load adr. of system return 364 larl %r14,sysc_return # load adr. of system return
381 jg do_single_step # branch to do_sigtrap 365 jg do_per_trap
382 366
383# 367#
384# call tracehook_report_syscall_entry/tracehook_report_syscall_exit before 368# call tracehook_report_syscall_entry/tracehook_report_syscall_exit before
@@ -397,6 +381,7 @@ sysc_tracesys:
397 lgf %r8,0(%r7,%r10) 381 lgf %r8,0(%r7,%r10)
398sysc_tracego: 382sysc_tracego:
399 lmg %r3,%r6,SP_R3(%r15) 383 lmg %r3,%r6,SP_R3(%r15)
384 mvc SP_ARGS(8,%r15),SP_R7(%r15)
400 lg %r2,SP_ORIG_R2(%r15) 385 lg %r2,SP_ORIG_R2(%r15)
401 basr %r14,%r8 # call sys_xxx 386 basr %r14,%r8 # call sys_xxx
402 stg %r2,SP_R2(%r15) # store return value 387 stg %r2,SP_R2(%r15) # store return value
@@ -443,14 +428,12 @@ kernel_execve:
443 br %r14 428 br %r14
444 # execve succeeded. 429 # execve succeeded.
4450: stnsm __SF_EMPTY(%r15),0xfc # disable interrupts 4300: stnsm __SF_EMPTY(%r15),0xfc # disable interrupts
446# TRACE_IRQS_OFF
447 lg %r15,__LC_KERNEL_STACK # load ksp 431 lg %r15,__LC_KERNEL_STACK # load ksp
448 aghi %r15,-SP_SIZE # make room for registers & psw 432 aghi %r15,-SP_SIZE # make room for registers & psw
449 lg %r13,__LC_SVC_NEW_PSW+8 433 lg %r13,__LC_SVC_NEW_PSW+8
450 mvc SP_PTREGS(__PT_SIZE,%r15),0(%r12) # copy pt_regs 434 mvc SP_PTREGS(__PT_SIZE,%r15),0(%r12) # copy pt_regs
451 lg %r12,__LC_THREAD_INFO 435 lg %r12,__LC_THREAD_INFO
452 xc __SF_BACKCHAIN(8,%r15),__SF_BACKCHAIN(%r15) 436 xc __SF_BACKCHAIN(8,%r15),__SF_BACKCHAIN(%r15)
453# TRACE_IRQS_ON
454 stosm __SF_EMPTY(%r15),0x03 # reenable interrupts 437 stosm __SF_EMPTY(%r15),0x03 # reenable interrupts
455 brasl %r14,execve_tail 438 brasl %r14,execve_tail
456 j sysc_return 439 j sysc_return
@@ -490,19 +473,18 @@ pgm_check_handler:
490 LAST_BREAK 473 LAST_BREAK
491pgm_no_vtime: 474pgm_no_vtime:
492 HANDLE_SIE_INTERCEPT 475 HANDLE_SIE_INTERCEPT
493 TRACE_IRQS_CHECK_OFF
494 stg %r11,SP_ARGS(%r15) 476 stg %r11,SP_ARGS(%r15)
495 lgf %r3,__LC_PGM_ILC # load program interruption code 477 lgf %r3,__LC_PGM_ILC # load program interruption code
478 lg %r4,__LC_TRANS_EXC_CODE
479 REENABLE_IRQS
496 lghi %r8,0x7f 480 lghi %r8,0x7f
497 ngr %r8,%r3 481 ngr %r8,%r3
498pgm_do_call:
499 sll %r8,3 482 sll %r8,3
500 larl %r1,pgm_check_table 483 larl %r1,pgm_check_table
501 lg %r1,0(%r8,%r1) # load address of handler routine 484 lg %r1,0(%r8,%r1) # load address of handler routine
502 la %r2,SP_PTREGS(%r15) # address of register-save area 485 la %r2,SP_PTREGS(%r15) # address of register-save area
503 basr %r14,%r1 # branch to interrupt-handler 486 basr %r14,%r1 # branch to interrupt-handler
504pgm_exit: 487pgm_exit:
505 TRACE_IRQS_CHECK_ON
506 j sysc_return 488 j sysc_return
507 489
508# 490#
@@ -533,15 +515,16 @@ pgm_per_std:
533 LAST_BREAK 515 LAST_BREAK
534pgm_no_vtime2: 516pgm_no_vtime2:
535 HANDLE_SIE_INTERCEPT 517 HANDLE_SIE_INTERCEPT
536 TRACE_IRQS_CHECK_OFF
537 lg %r1,__TI_task(%r12) 518 lg %r1,__TI_task(%r12)
538 tm SP_PSW+1(%r15),0x01 # kernel per event ? 519 tm SP_PSW+1(%r15),0x01 # kernel per event ?
539 jz kernel_per 520 jz kernel_per
540 mvc __THREAD_per+__PER_atmid(2,%r1),__LC_PER_ATMID 521 mvc __THREAD_per_cause(2,%r1),__LC_PER_CAUSE
541 mvc __THREAD_per+__PER_address(8,%r1),__LC_PER_ADDRESS 522 mvc __THREAD_per_address(8,%r1),__LC_PER_ADDRESS
542 mvc __THREAD_per+__PER_access_id(1,%r1),__LC_PER_ACCESS_ID 523 mvc __THREAD_per_paid(1,%r1),__LC_PER_PAID
543 oi __TI_flags+7(%r12),_TIF_SINGLE_STEP # set TIF_SINGLE_STEP 524 oi __TI_flags+7(%r12),_TIF_PER_TRAP # set TIF_PER_TRAP
544 lgf %r3,__LC_PGM_ILC # load program interruption code 525 lgf %r3,__LC_PGM_ILC # load program interruption code
526 lg %r4,__LC_TRANS_EXC_CODE
527 REENABLE_IRQS
545 lghi %r8,0x7f 528 lghi %r8,0x7f
546 ngr %r8,%r3 # clear per-event-bit and ilc 529 ngr %r8,%r3 # clear per-event-bit and ilc
547 je pgm_exit2 530 je pgm_exit2
@@ -551,8 +534,6 @@ pgm_no_vtime2:
551 la %r2,SP_PTREGS(%r15) # address of register-save area 534 la %r2,SP_PTREGS(%r15) # address of register-save area
552 basr %r14,%r1 # branch to interrupt-handler 535 basr %r14,%r1 # branch to interrupt-handler
553pgm_exit2: 536pgm_exit2:
554 TRACE_IRQS_ON
555 stosm __SF_EMPTY(%r15),0x03 # reenable interrupts
556 j sysc_return 537 j sysc_return
557 538
558# 539#
@@ -568,13 +549,11 @@ pgm_svcper:
568 UPDATE_VTIME __LC_LAST_UPDATE_TIMER,__LC_EXIT_TIMER,__LC_SYSTEM_TIMER 549 UPDATE_VTIME __LC_LAST_UPDATE_TIMER,__LC_EXIT_TIMER,__LC_SYSTEM_TIMER
569 mvc __LC_LAST_UPDATE_TIMER(8),__LC_SYNC_ENTER_TIMER 550 mvc __LC_LAST_UPDATE_TIMER(8),__LC_SYNC_ENTER_TIMER
570 LAST_BREAK 551 LAST_BREAK
571 TRACE_IRQS_OFF
572 lg %r8,__TI_task(%r12) 552 lg %r8,__TI_task(%r12)
573 mvc __THREAD_per+__PER_atmid(2,%r8),__LC_PER_ATMID 553 mvc __THREAD_per_cause(2,%r8),__LC_PER_CAUSE
574 mvc __THREAD_per+__PER_address(8,%r8),__LC_PER_ADDRESS 554 mvc __THREAD_per_address(8,%r8),__LC_PER_ADDRESS
575 mvc __THREAD_per+__PER_access_id(1,%r8),__LC_PER_ACCESS_ID 555 mvc __THREAD_per_paid(1,%r8),__LC_PER_PAID
576 oi __TI_flags+7(%r12),_TIF_SINGLE_STEP # set TIF_SINGLE_STEP 556 oi __TI_flags+7(%r12),_TIF_PER_TRAP # set TIF_PER_TRAP
577 TRACE_IRQS_ON
578 stosm __SF_EMPTY(%r15),0x03 # reenable interrupts 557 stosm __SF_EMPTY(%r15),0x03 # reenable interrupts
579 lmg %r2,%r6,SP_R2(%r15) # load svc arguments 558 lmg %r2,%r6,SP_R2(%r15) # load svc arguments
580 j sysc_do_svc 559 j sysc_do_svc
@@ -583,9 +562,10 @@ pgm_svcper:
583# per was called from kernel, must be kprobes 562# per was called from kernel, must be kprobes
584# 563#
585kernel_per: 564kernel_per:
565 REENABLE_IRQS
586 xc SP_SVCNR(2,%r15),SP_SVCNR(%r15) # clear svc number 566 xc SP_SVCNR(2,%r15),SP_SVCNR(%r15) # clear svc number
587 la %r2,SP_PTREGS(%r15) # address of register-save area 567 la %r2,SP_PTREGS(%r15) # address of register-save area
588 brasl %r14,do_single_step 568 brasl %r14,do_per_trap
589 j pgm_exit 569 j pgm_exit
590 570
591/* 571/*
@@ -743,8 +723,11 @@ ext_int_handler:
743ext_no_vtime: 723ext_no_vtime:
744 HANDLE_SIE_INTERCEPT 724 HANDLE_SIE_INTERCEPT
745 TRACE_IRQS_OFF 725 TRACE_IRQS_OFF
726 lghi %r1,4096
746 la %r2,SP_PTREGS(%r15) # address of register-save area 727 la %r2,SP_PTREGS(%r15) # address of register-save area
747 llgh %r3,__LC_EXT_INT_CODE # get interruption code 728 llgf %r3,__LC_CPU_ADDRESS # get cpu address + interruption code
729 llgf %r4,__LC_EXT_PARAMS # get external parameter
730 lg %r5,__LC_EXT_PARAMS2-4096(%r1) # get 64 bit external parameter
748 brasl %r14,do_extint 731 brasl %r14,do_extint
749 j io_return 732 j io_return
750 733
@@ -859,7 +842,7 @@ restart_base:
859 mvc __LC_SYSTEM_TIMER(8),__TI_system_timer(%r1) 842 mvc __LC_SYSTEM_TIMER(8),__TI_system_timer(%r1)
860 xc __LC_STEAL_TIMER(8),__LC_STEAL_TIMER 843 xc __LC_STEAL_TIMER(8),__LC_STEAL_TIMER
861 stosm __SF_EMPTY(%r15),0x04 # now we can turn dat on 844 stosm __SF_EMPTY(%r15),0x04 # now we can turn dat on
862 jg start_secondary 845 brasl %r14,start_secondary
863 .align 8 846 .align 8
864restart_vtime: 847restart_vtime:
865 .long 0x7fffffff,0xffffffff 848 .long 0x7fffffff,0xffffffff
@@ -879,6 +862,8 @@ restart_crash:
879restart_go: 862restart_go:
880#endif 863#endif
881 864
865 .section .kprobes.text, "ax"
866
882#ifdef CONFIG_CHECK_STACK 867#ifdef CONFIG_CHECK_STACK
883/* 868/*
884 * The synchronous or the asynchronous stack overflowed. We are dead. 869 * The synchronous or the asynchronous stack overflowed. We are dead.
@@ -966,7 +951,6 @@ cleanup_system_call:
966 CREATE_STACK_FRAME __LC_SAVE_AREA 951 CREATE_STACK_FRAME __LC_SAVE_AREA
967 mvc SP_PSW(16,%r15),__LC_SVC_OLD_PSW 952 mvc SP_PSW(16,%r15),__LC_SVC_OLD_PSW
968 mvc SP_ILC(4,%r15),__LC_SVC_ILC 953 mvc SP_ILC(4,%r15),__LC_SVC_ILC
969 stg %r7,SP_ARGS(%r15)
970 mvc 8(8,%r12),__LC_THREAD_INFO 954 mvc 8(8,%r12),__LC_THREAD_INFO
971cleanup_vtime: 955cleanup_vtime:
972 clc __LC_RETURN_PSW+8(8),BASED(cleanup_system_call_insn+24) 956 clc __LC_RETURN_PSW+8(8),BASED(cleanup_system_call_insn+24)
diff --git a/arch/s390/kernel/ftrace.c b/arch/s390/kernel/ftrace.c
index 6a83d0581317..78bdf0e5dff7 100644
--- a/arch/s390/kernel/ftrace.c
+++ b/arch/s390/kernel/ftrace.c
@@ -4,7 +4,7 @@
4 * Copyright IBM Corp. 2009 4 * Copyright IBM Corp. 2009
5 * 5 *
6 * Author(s): Heiko Carstens <heiko.carstens@de.ibm.com>, 6 * Author(s): Heiko Carstens <heiko.carstens@de.ibm.com>,
7 * 7 * Martin Schwidefsky <schwidefsky@de.ibm.com>
8 */ 8 */
9 9
10#include <linux/hardirq.h> 10#include <linux/hardirq.h>
@@ -12,176 +12,144 @@
12#include <linux/ftrace.h> 12#include <linux/ftrace.h>
13#include <linux/kernel.h> 13#include <linux/kernel.h>
14#include <linux/types.h> 14#include <linux/types.h>
15#include <linux/kprobes.h>
15#include <trace/syscall.h> 16#include <trace/syscall.h>
16#include <asm/asm-offsets.h> 17#include <asm/asm-offsets.h>
17 18
19#ifdef CONFIG_64BIT
20#define MCOUNT_OFFSET_RET 12
21#else
22#define MCOUNT_OFFSET_RET 22
23#endif
24
18#ifdef CONFIG_DYNAMIC_FTRACE 25#ifdef CONFIG_DYNAMIC_FTRACE
19 26
20void ftrace_disable_code(void); 27void ftrace_disable_code(void);
21void ftrace_disable_return(void); 28void ftrace_enable_insn(void);
22void ftrace_call_code(void);
23void ftrace_nop_code(void);
24
25#define FTRACE_INSN_SIZE 4
26 29
27#ifdef CONFIG_64BIT 30#ifdef CONFIG_64BIT
28 31/*
32 * The 64-bit mcount code looks like this:
33 * stg %r14,8(%r15) # offset 0
34 * > larl %r1,<&counter> # offset 6
35 * > brasl %r14,_mcount # offset 12
36 * lg %r14,8(%r15) # offset 18
37 * Total length is 24 bytes. The middle two instructions of the mcount
38 * block get overwritten by ftrace_make_nop / ftrace_make_call.
39 * The 64-bit enabled ftrace code block looks like this:
40 * stg %r14,8(%r15) # offset 0
41 * > lg %r1,__LC_FTRACE_FUNC # offset 6
42 * > lgr %r0,%r0 # offset 12
43 * > basr %r14,%r1 # offset 16
44 * lg %r14,8(%15) # offset 18
45 * The return points of the mcount/ftrace function have the same offset 18.
46 * The 64-bit disable ftrace code block looks like this:
47 * stg %r14,8(%r15) # offset 0
48 * > jg .+18 # offset 6
49 * > lgr %r0,%r0 # offset 12
50 * > basr %r14,%r1 # offset 16
51 * lg %r14,8(%15) # offset 18
52 * The jg instruction branches to offset 24 to skip as many instructions
53 * as possible.
54 */
29asm( 55asm(
30 " .align 4\n" 56 " .align 4\n"
31 "ftrace_disable_code:\n" 57 "ftrace_disable_code:\n"
32 " j 0f\n" 58 " jg 0f\n"
33 " .word 0x0024\n"
34 " lg %r1,"__stringify(__LC_FTRACE_FUNC)"\n"
35 " basr %r14,%r1\n"
36 "ftrace_disable_return:\n"
37 " lg %r14,8(15)\n"
38 " lgr %r0,%r0\n" 59 " lgr %r0,%r0\n"
39 "0:\n"); 60 " basr %r14,%r1\n"
40 61 "0:\n"
41asm(
42 " .align 4\n" 62 " .align 4\n"
43 "ftrace_nop_code:\n" 63 "ftrace_enable_insn:\n"
44 " j .+"__stringify(MCOUNT_INSN_SIZE)"\n"); 64 " lg %r1,"__stringify(__LC_FTRACE_FUNC)"\n");
45 65
46asm( 66#define FTRACE_INSN_SIZE 6
47 " .align 4\n"
48 "ftrace_call_code:\n"
49 " stg %r14,8(%r15)\n");
50 67
51#else /* CONFIG_64BIT */ 68#else /* CONFIG_64BIT */
52 69/*
70 * The 31-bit mcount code looks like this:
71 * st %r14,4(%r15) # offset 0
72 * > bras %r1,0f # offset 4
73 * > .long _mcount # offset 8
74 * > .long <&counter> # offset 12
75 * > 0: l %r14,0(%r1) # offset 16
76 * > l %r1,4(%r1) # offset 20
77 * basr %r14,%r14 # offset 24
78 * l %r14,4(%r15) # offset 26
79 * Total length is 30 bytes. The twenty bytes starting from offset 4
80 * to offset 24 get overwritten by ftrace_make_nop / ftrace_make_call.
81 * The 31-bit enabled ftrace code block looks like this:
82 * st %r14,4(%r15) # offset 0
83 * > l %r14,__LC_FTRACE_FUNC # offset 4
84 * > j 0f # offset 8
85 * > .fill 12,1,0x07 # offset 12
86 * 0: basr %r14,%r14 # offset 24
87 * l %r14,4(%r14) # offset 26
88 * The return points of the mcount/ftrace function have the same offset 26.
89 * The 31-bit disabled ftrace code block looks like this:
90 * st %r14,4(%r15) # offset 0
91 * > j .+26 # offset 4
92 * > j 0f # offset 8
93 * > .fill 12,1,0x07 # offset 12
94 * 0: basr %r14,%r14 # offset 24
95 * l %r14,4(%r14) # offset 26
96 * The j instruction branches to offset 30 to skip as many instructions
97 * as possible.
98 */
53asm( 99asm(
54 " .align 4\n" 100 " .align 4\n"
55 "ftrace_disable_code:\n" 101 "ftrace_disable_code:\n"
102 " j 1f\n"
56 " j 0f\n" 103 " j 0f\n"
57 " l %r1,"__stringify(__LC_FTRACE_FUNC)"\n" 104 " .fill 12,1,0x07\n"
58 " basr %r14,%r1\n" 105 "0: basr %r14,%r14\n"
59 "ftrace_disable_return:\n" 106 "1:\n"
60 " l %r14,4(%r15)\n"
61 " j 0f\n"
62 " bcr 0,%r7\n"
63 " bcr 0,%r7\n"
64 " bcr 0,%r7\n"
65 " bcr 0,%r7\n"
66 " bcr 0,%r7\n"
67 " bcr 0,%r7\n"
68 "0:\n");
69
70asm(
71 " .align 4\n" 107 " .align 4\n"
72 "ftrace_nop_code:\n" 108 "ftrace_enable_insn:\n"
73 " j .+"__stringify(MCOUNT_INSN_SIZE)"\n"); 109 " l %r14,"__stringify(__LC_FTRACE_FUNC)"\n");
74 110
75asm( 111#define FTRACE_INSN_SIZE 4
76 " .align 4\n"
77 "ftrace_call_code:\n"
78 " st %r14,4(%r15)\n");
79 112
80#endif /* CONFIG_64BIT */ 113#endif /* CONFIG_64BIT */
81 114
82static int ftrace_modify_code(unsigned long ip,
83 void *old_code, int old_size,
84 void *new_code, int new_size)
85{
86 unsigned char replaced[MCOUNT_INSN_SIZE];
87
88 /*
89 * Note: Due to modules code can disappear and change.
90 * We need to protect against faulting as well as code
91 * changing. We do this by using the probe_kernel_*
92 * functions.
93 * This however is just a simple sanity check.
94 */
95 if (probe_kernel_read(replaced, (void *)ip, old_size))
96 return -EFAULT;
97 if (memcmp(replaced, old_code, old_size) != 0)
98 return -EINVAL;
99 if (probe_kernel_write((void *)ip, new_code, new_size))
100 return -EPERM;
101 return 0;
102}
103
104static int ftrace_make_initial_nop(struct module *mod, struct dyn_ftrace *rec,
105 unsigned long addr)
106{
107 return ftrace_modify_code(rec->ip,
108 ftrace_call_code, FTRACE_INSN_SIZE,
109 ftrace_disable_code, MCOUNT_INSN_SIZE);
110}
111 115
112int ftrace_make_nop(struct module *mod, struct dyn_ftrace *rec, 116int ftrace_make_nop(struct module *mod, struct dyn_ftrace *rec,
113 unsigned long addr) 117 unsigned long addr)
114{ 118{
115 if (addr == MCOUNT_ADDR) 119 if (probe_kernel_write((void *) rec->ip, ftrace_disable_code,
116 return ftrace_make_initial_nop(mod, rec, addr); 120 MCOUNT_INSN_SIZE))
117 return ftrace_modify_code(rec->ip, 121 return -EPERM;
118 ftrace_call_code, FTRACE_INSN_SIZE, 122 return 0;
119 ftrace_nop_code, FTRACE_INSN_SIZE);
120} 123}
121 124
122int ftrace_make_call(struct dyn_ftrace *rec, unsigned long addr) 125int ftrace_make_call(struct dyn_ftrace *rec, unsigned long addr)
123{ 126{
124 return ftrace_modify_code(rec->ip, 127 if (probe_kernel_write((void *) rec->ip, ftrace_enable_insn,
125 ftrace_nop_code, FTRACE_INSN_SIZE, 128 FTRACE_INSN_SIZE))
126 ftrace_call_code, FTRACE_INSN_SIZE); 129 return -EPERM;
130 return 0;
127} 131}
128 132
129int ftrace_update_ftrace_func(ftrace_func_t func) 133int ftrace_update_ftrace_func(ftrace_func_t func)
130{ 134{
131 ftrace_dyn_func = (unsigned long)func;
132 return 0; 135 return 0;
133} 136}
134 137
135int __init ftrace_dyn_arch_init(void *data) 138int __init ftrace_dyn_arch_init(void *data)
136{ 139{
137 *(unsigned long *)data = 0; 140 *(unsigned long *) data = 0;
138 return 0; 141 return 0;
139} 142}
140 143
141#endif /* CONFIG_DYNAMIC_FTRACE */ 144#endif /* CONFIG_DYNAMIC_FTRACE */
142 145
143#ifdef CONFIG_FUNCTION_GRAPH_TRACER 146#ifdef CONFIG_FUNCTION_GRAPH_TRACER
144#ifdef CONFIG_DYNAMIC_FTRACE
145/*
146 * Patch the kernel code at ftrace_graph_caller location:
147 * The instruction there is branch relative on condition. The condition mask
148 * is either all ones (always branch aka disable ftrace_graph_caller) or all
149 * zeroes (nop aka enable ftrace_graph_caller).
150 * Instruction format for brc is a7m4xxxx where m is the condition mask.
151 */
152int ftrace_enable_ftrace_graph_caller(void)
153{
154 unsigned short opcode = 0xa704;
155
156 return probe_kernel_write(ftrace_graph_caller, &opcode, sizeof(opcode));
157}
158
159int ftrace_disable_ftrace_graph_caller(void)
160{
161 unsigned short opcode = 0xa7f4;
162
163 return probe_kernel_write(ftrace_graph_caller, &opcode, sizeof(opcode));
164}
165
166static inline unsigned long ftrace_mcount_call_adjust(unsigned long addr)
167{
168 return addr - (ftrace_disable_return - ftrace_disable_code);
169}
170
171#else /* CONFIG_DYNAMIC_FTRACE */
172
173static inline unsigned long ftrace_mcount_call_adjust(unsigned long addr)
174{
175 return addr - MCOUNT_OFFSET_RET;
176}
177
178#endif /* CONFIG_DYNAMIC_FTRACE */
179
180/* 147/*
181 * Hook the return address and push it in the stack of return addresses 148 * Hook the return address and push it in the stack of return addresses
182 * in current thread info. 149 * in current thread info.
183 */ 150 */
184unsigned long prepare_ftrace_return(unsigned long ip, unsigned long parent) 151unsigned long __kprobes prepare_ftrace_return(unsigned long parent,
152 unsigned long ip)
185{ 153{
186 struct ftrace_graph_ent trace; 154 struct ftrace_graph_ent trace;
187 155
@@ -189,14 +157,42 @@ unsigned long prepare_ftrace_return(unsigned long ip, unsigned long parent)
189 goto out; 157 goto out;
190 if (ftrace_push_return_trace(parent, ip, &trace.depth, 0) == -EBUSY) 158 if (ftrace_push_return_trace(parent, ip, &trace.depth, 0) == -EBUSY)
191 goto out; 159 goto out;
192 trace.func = ftrace_mcount_call_adjust(ip) & PSW_ADDR_INSN; 160 trace.func = (ip & PSW_ADDR_INSN) - MCOUNT_OFFSET_RET;
193 /* Only trace if the calling function expects to. */ 161 /* Only trace if the calling function expects to. */
194 if (!ftrace_graph_entry(&trace)) { 162 if (!ftrace_graph_entry(&trace)) {
195 current->curr_ret_stack--; 163 current->curr_ret_stack--;
196 goto out; 164 goto out;
197 } 165 }
198 parent = (unsigned long)return_to_handler; 166 parent = (unsigned long) return_to_handler;
199out: 167out:
200 return parent; 168 return parent;
201} 169}
170
171#ifdef CONFIG_DYNAMIC_FTRACE
172/*
173 * Patch the kernel code at ftrace_graph_caller location. The instruction
174 * there is branch relative and save to prepare_ftrace_return. To disable
175 * the call to prepare_ftrace_return we patch the bras offset to point
176 * directly after the instructions. To enable the call we calculate
177 * the original offset to prepare_ftrace_return and put it back.
178 */
179int ftrace_enable_ftrace_graph_caller(void)
180{
181 unsigned short offset;
182
183 offset = ((void *) prepare_ftrace_return -
184 (void *) ftrace_graph_caller) / 2;
185 return probe_kernel_write(ftrace_graph_caller + 2,
186 &offset, sizeof(offset));
187}
188
189int ftrace_disable_ftrace_graph_caller(void)
190{
191 static unsigned short offset = 0x0002;
192
193 return probe_kernel_write(ftrace_graph_caller + 2,
194 &offset, sizeof(offset));
195}
196
197#endif /* CONFIG_DYNAMIC_FTRACE */
202#endif /* CONFIG_FUNCTION_GRAPH_TRACER */ 198#endif /* CONFIG_FUNCTION_GRAPH_TRACER */
diff --git a/arch/s390/kernel/head.S b/arch/s390/kernel/head.S
index db1696e210af..fb317bf2c378 100644
--- a/arch/s390/kernel/head.S
+++ b/arch/s390/kernel/head.S
@@ -460,7 +460,7 @@ startup:
460#ifndef CONFIG_MARCH_G5 460#ifndef CONFIG_MARCH_G5
461 # check capabilities against MARCH_{G5,Z900,Z990,Z9_109,Z10} 461 # check capabilities against MARCH_{G5,Z900,Z990,Z9_109,Z10}
462 xc __LC_STFL_FAC_LIST(8),__LC_STFL_FAC_LIST 462 xc __LC_STFL_FAC_LIST(8),__LC_STFL_FAC_LIST
463 stfl __LC_STFL_FAC_LIST # store facility list 463 .insn s,0xb2b10000,__LC_STFL_FAC_LIST # store facility list
464 tm __LC_STFL_FAC_LIST,0x01 # stfle available ? 464 tm __LC_STFL_FAC_LIST,0x01 # stfle available ?
465 jz 0f 465 jz 0f
466 la %r0,0 466 la %r0,0
@@ -488,7 +488,9 @@ startup:
488 .align 16 488 .align 16
4892: .long 0x000a0000,0x8badcccc 4892: .long 0x000a0000,0x8badcccc
490#if defined(CONFIG_64BIT) 490#if defined(CONFIG_64BIT)
491#if defined(CONFIG_MARCH_Z10) 491#if defined(CONFIG_MARCH_Z196)
492 .long 0xc100efe3, 0xf46c0000
493#elif defined(CONFIG_MARCH_Z10)
492 .long 0xc100efe3, 0xf0680000 494 .long 0xc100efe3, 0xf0680000
493#elif defined(CONFIG_MARCH_Z9_109) 495#elif defined(CONFIG_MARCH_Z9_109)
494 .long 0xc100efc3, 0x00000000 496 .long 0xc100efc3, 0x00000000
@@ -498,7 +500,9 @@ startup:
498 .long 0xc0000000, 0x00000000 500 .long 0xc0000000, 0x00000000
499#endif 501#endif
500#else 502#else
501#if defined(CONFIG_MARCH_Z10) 503#if defined(CONFIG_MARCH_Z196)
504 .long 0x8100c880, 0x00000000
505#elif defined(CONFIG_MARCH_Z10)
502 .long 0x8100c880, 0x00000000 506 .long 0x8100c880, 0x00000000
503#elif defined(CONFIG_MARCH_Z9_109) 507#elif defined(CONFIG_MARCH_Z9_109)
504 .long 0x8100c880, 0x00000000 508 .long 0x8100c880, 0x00000000
diff --git a/arch/s390/kernel/irq.c b/arch/s390/kernel/irq.c
index 026a37a94fc9..e3264f6a9720 100644
--- a/arch/s390/kernel/irq.c
+++ b/arch/s390/kernel/irq.c
@@ -1,28 +1,66 @@
1/* 1/*
2 * arch/s390/kernel/irq.c 2 * Copyright IBM Corp. 2004,2011
3 * 3 * Author(s): Martin Schwidefsky <schwidefsky@de.ibm.com>,
4 * Copyright IBM Corp. 2004,2007 4 * Holger Smolinski <Holger.Smolinski@de.ibm.com>,
5 * Author(s): Martin Schwidefsky (schwidefsky@de.ibm.com), 5 * Thomas Spatzier <tspat@de.ibm.com>,
6 * Thomas Spatzier (tspat@de.ibm.com)
7 * 6 *
8 * This file contains interrupt related functions. 7 * This file contains interrupt related functions.
9 */ 8 */
10 9
11#include <linux/module.h>
12#include <linux/kernel.h>
13#include <linux/kernel_stat.h> 10#include <linux/kernel_stat.h>
14#include <linux/interrupt.h> 11#include <linux/interrupt.h>
15#include <linux/seq_file.h> 12#include <linux/seq_file.h>
16#include <linux/cpu.h>
17#include <linux/proc_fs.h> 13#include <linux/proc_fs.h>
18#include <linux/profile.h> 14#include <linux/profile.h>
15#include <linux/module.h>
16#include <linux/kernel.h>
17#include <linux/ftrace.h>
18#include <linux/errno.h>
19#include <linux/slab.h>
20#include <linux/cpu.h>
21#include <asm/irq_regs.h>
22#include <asm/cputime.h>
23#include <asm/lowcore.h>
24#include <asm/irq.h>
25#include "entry.h"
26
27struct irq_class {
28 char *name;
29 char *desc;
30};
31
32static const struct irq_class intrclass_names[] = {
33 {.name = "EXT" },
34 {.name = "I/O" },
35 {.name = "CLK", .desc = "[EXT] Clock Comparator" },
36 {.name = "IPI", .desc = "[EXT] Signal Processor" },
37 {.name = "TMR", .desc = "[EXT] CPU Timer" },
38 {.name = "TAL", .desc = "[EXT] Timing Alert" },
39 {.name = "PFL", .desc = "[EXT] Pseudo Page Fault" },
40 {.name = "DSD", .desc = "[EXT] DASD Diag" },
41 {.name = "VRT", .desc = "[EXT] Virtio" },
42 {.name = "SCP", .desc = "[EXT] Service Call" },
43 {.name = "IUC", .desc = "[EXT] IUCV" },
44 {.name = "CPM", .desc = "[EXT] CPU Measurement" },
45 {.name = "QAI", .desc = "[I/O] QDIO Adapter Interrupt" },
46 {.name = "QDI", .desc = "[I/O] QDIO Interrupt" },
47 {.name = "DAS", .desc = "[I/O] DASD" },
48 {.name = "C15", .desc = "[I/O] 3215" },
49 {.name = "C70", .desc = "[I/O] 3270" },
50 {.name = "TAP", .desc = "[I/O] Tape" },
51 {.name = "VMR", .desc = "[I/O] Unit Record Devices" },
52 {.name = "LCS", .desc = "[I/O] LCS" },
53 {.name = "CLW", .desc = "[I/O] CLAW" },
54 {.name = "CTC", .desc = "[I/O] CTC" },
55 {.name = "APB", .desc = "[I/O] AP Bus" },
56 {.name = "NMI", .desc = "[NMI] Machine Check" },
57};
19 58
20/* 59/*
21 * show_interrupts is needed by /proc/interrupts. 60 * show_interrupts is needed by /proc/interrupts.
22 */ 61 */
23int show_interrupts(struct seq_file *p, void *v) 62int show_interrupts(struct seq_file *p, void *v)
24{ 63{
25 static const char *intrclass_names[] = { "EXT", "I/O", };
26 int i = *(loff_t *) v, j; 64 int i = *(loff_t *) v, j;
27 65
28 get_online_cpus(); 66 get_online_cpus();
@@ -34,15 +72,16 @@ int show_interrupts(struct seq_file *p, void *v)
34 } 72 }
35 73
36 if (i < NR_IRQS) { 74 if (i < NR_IRQS) {
37 seq_printf(p, "%s: ", intrclass_names[i]); 75 seq_printf(p, "%s: ", intrclass_names[i].name);
38#ifndef CONFIG_SMP 76#ifndef CONFIG_SMP
39 seq_printf(p, "%10u ", kstat_irqs(i)); 77 seq_printf(p, "%10u ", kstat_irqs(i));
40#else 78#else
41 for_each_online_cpu(j) 79 for_each_online_cpu(j)
42 seq_printf(p, "%10u ", kstat_cpu(j).irqs[i]); 80 seq_printf(p, "%10u ", kstat_cpu(j).irqs[i]);
43#endif 81#endif
82 if (intrclass_names[i].desc)
83 seq_printf(p, " %s", intrclass_names[i].desc);
44 seq_putc(p, '\n'); 84 seq_putc(p, '\n');
45
46 } 85 }
47 put_online_cpus(); 86 put_online_cpus();
48 return 0; 87 return 0;
@@ -52,8 +91,7 @@ int show_interrupts(struct seq_file *p, void *v)
52 * For compatibilty only. S/390 specific setup of interrupts et al. is done 91 * For compatibilty only. S/390 specific setup of interrupts et al. is done
53 * much later in init_channel_subsystem(). 92 * much later in init_channel_subsystem().
54 */ 93 */
55void __init 94void __init init_IRQ(void)
56init_IRQ(void)
57{ 95{
58 /* nothing... */ 96 /* nothing... */
59} 97}
@@ -104,3 +142,116 @@ void init_irq_proc(void)
104 create_prof_cpu_mask(root_irq_dir); 142 create_prof_cpu_mask(root_irq_dir);
105} 143}
106#endif 144#endif
145
146/*
147 * ext_int_hash[index] is the start of the list for all external interrupts
148 * that hash to this index. With the current set of external interrupts
149 * (0x1202 external call, 0x1004 cpu timer, 0x2401 hwc console, 0x4000
150 * iucv and 0x2603 pfault) this is always the first element.
151 */
152
153struct ext_int_info {
154 struct ext_int_info *next;
155 ext_int_handler_t handler;
156 u16 code;
157};
158
159static struct ext_int_info *ext_int_hash[256];
160
161static inline int ext_hash(u16 code)
162{
163 return (code + (code >> 9)) & 0xff;
164}
165
166int register_external_interrupt(u16 code, ext_int_handler_t handler)
167{
168 struct ext_int_info *p;
169 int index;
170
171 p = kmalloc(sizeof(*p), GFP_ATOMIC);
172 if (!p)
173 return -ENOMEM;
174 p->code = code;
175 p->handler = handler;
176 index = ext_hash(code);
177 p->next = ext_int_hash[index];
178 ext_int_hash[index] = p;
179 return 0;
180}
181EXPORT_SYMBOL(register_external_interrupt);
182
183int unregister_external_interrupt(u16 code, ext_int_handler_t handler)
184{
185 struct ext_int_info *p, *q;
186 int index;
187
188 index = ext_hash(code);
189 q = NULL;
190 p = ext_int_hash[index];
191 while (p) {
192 if (p->code == code && p->handler == handler)
193 break;
194 q = p;
195 p = p->next;
196 }
197 if (!p)
198 return -ENOENT;
199 if (q)
200 q->next = p->next;
201 else
202 ext_int_hash[index] = p->next;
203 kfree(p);
204 return 0;
205}
206EXPORT_SYMBOL(unregister_external_interrupt);
207
208void __irq_entry do_extint(struct pt_regs *regs, unsigned int ext_int_code,
209 unsigned int param32, unsigned long param64)
210{
211 struct pt_regs *old_regs;
212 unsigned short code;
213 struct ext_int_info *p;
214 int index;
215
216 code = (unsigned short) ext_int_code;
217 old_regs = set_irq_regs(regs);
218 s390_idle_check(regs, S390_lowcore.int_clock,
219 S390_lowcore.async_enter_timer);
220 irq_enter();
221 if (S390_lowcore.int_clock >= S390_lowcore.clock_comparator)
222 /* Serve timer interrupts first. */
223 clock_comparator_work();
224 kstat_cpu(smp_processor_id()).irqs[EXTERNAL_INTERRUPT]++;
225 if (code != 0x1004)
226 __get_cpu_var(s390_idle).nohz_delay = 1;
227 index = ext_hash(code);
228 for (p = ext_int_hash[index]; p; p = p->next) {
229 if (likely(p->code == code))
230 p->handler(ext_int_code, param32, param64);
231 }
232 irq_exit();
233 set_irq_regs(old_regs);
234}
235
236static DEFINE_SPINLOCK(sc_irq_lock);
237static int sc_irq_refcount;
238
239void service_subclass_irq_register(void)
240{
241 spin_lock(&sc_irq_lock);
242 if (!sc_irq_refcount)
243 ctl_set_bit(0, 9);
244 sc_irq_refcount++;
245 spin_unlock(&sc_irq_lock);
246}
247EXPORT_SYMBOL(service_subclass_irq_register);
248
249void service_subclass_irq_unregister(void)
250{
251 spin_lock(&sc_irq_lock);
252 sc_irq_refcount--;
253 if (!sc_irq_refcount)
254 ctl_clear_bit(0, 9);
255 spin_unlock(&sc_irq_lock);
256}
257EXPORT_SYMBOL(service_subclass_irq_unregister);
diff --git a/arch/s390/kernel/jump_label.c b/arch/s390/kernel/jump_label.c
new file mode 100644
index 000000000000..44cc06bedf77
--- /dev/null
+++ b/arch/s390/kernel/jump_label.c
@@ -0,0 +1,59 @@
1/*
2 * Jump label s390 support
3 *
4 * Copyright IBM Corp. 2011
5 * Author(s): Jan Glauber <jang@linux.vnet.ibm.com>
6 */
7#include <linux/module.h>
8#include <linux/uaccess.h>
9#include <linux/stop_machine.h>
10#include <linux/jump_label.h>
11#include <asm/ipl.h>
12
13#ifdef HAVE_JUMP_LABEL
14
15struct insn {
16 u16 opcode;
17 s32 offset;
18} __packed;
19
20struct insn_args {
21 unsigned long *target;
22 struct insn *insn;
23 ssize_t size;
24};
25
26static int __arch_jump_label_transform(void *data)
27{
28 struct insn_args *args = data;
29 int rc;
30
31 rc = probe_kernel_write(args->target, args->insn, args->size);
32 WARN_ON_ONCE(rc < 0);
33 return 0;
34}
35
36void arch_jump_label_transform(struct jump_entry *entry,
37 enum jump_label_type type)
38{
39 struct insn_args args;
40 struct insn insn;
41
42 if (type == JUMP_LABEL_ENABLE) {
43 /* brcl 15,offset */
44 insn.opcode = 0xc0f4;
45 insn.offset = (entry->target - entry->code) >> 1;
46 } else {
47 /* brcl 0,0 */
48 insn.opcode = 0xc004;
49 insn.offset = 0;
50 }
51
52 args.target = (void *) entry->code;
53 args.insn = &insn;
54 args.size = JUMP_LABEL_NOP_SIZE;
55
56 stop_machine(__arch_jump_label_transform, &args, NULL);
57}
58
59#endif
diff --git a/arch/s390/kernel/kprobes.c b/arch/s390/kernel/kprobes.c
index 2a3d2bf6f083..1d05d669107c 100644
--- a/arch/s390/kernel/kprobes.c
+++ b/arch/s390/kernel/kprobes.c
@@ -30,35 +30,16 @@
30#include <asm/sections.h> 30#include <asm/sections.h>
31#include <linux/module.h> 31#include <linux/module.h>
32#include <linux/slab.h> 32#include <linux/slab.h>
33#include <linux/hardirq.h>
33 34
34DEFINE_PER_CPU(struct kprobe *, current_kprobe) = NULL; 35DEFINE_PER_CPU(struct kprobe *, current_kprobe);
35DEFINE_PER_CPU(struct kprobe_ctlblk, kprobe_ctlblk); 36DEFINE_PER_CPU(struct kprobe_ctlblk, kprobe_ctlblk);
36 37
37struct kretprobe_blackpoint kretprobe_blacklist[] = {{NULL, NULL}}; 38struct kretprobe_blackpoint kretprobe_blacklist[] = { };
38 39
39int __kprobes arch_prepare_kprobe(struct kprobe *p) 40static int __kprobes is_prohibited_opcode(kprobe_opcode_t *insn)
40{ 41{
41 /* Make sure the probe isn't going on a difficult instruction */ 42 switch (insn[0] >> 8) {
42 if (is_prohibited_opcode((kprobe_opcode_t *) p->addr))
43 return -EINVAL;
44
45 if ((unsigned long)p->addr & 0x01)
46 return -EINVAL;
47
48 /* Use the get_insn_slot() facility for correctness */
49 if (!(p->ainsn.insn = get_insn_slot()))
50 return -ENOMEM;
51
52 memcpy(p->ainsn.insn, p->addr, MAX_INSN_SIZE * sizeof(kprobe_opcode_t));
53
54 get_instruction_type(&p->ainsn);
55 p->opcode = *p->addr;
56 return 0;
57}
58
59int __kprobes is_prohibited_opcode(kprobe_opcode_t *instruction)
60{
61 switch (*(__u8 *) instruction) {
62 case 0x0c: /* bassm */ 43 case 0x0c: /* bassm */
63 case 0x0b: /* bsm */ 44 case 0x0b: /* bsm */
64 case 0x83: /* diag */ 45 case 0x83: /* diag */
@@ -67,7 +48,7 @@ int __kprobes is_prohibited_opcode(kprobe_opcode_t *instruction)
67 case 0xad: /* stosm */ 48 case 0xad: /* stosm */
68 return -EINVAL; 49 return -EINVAL;
69 } 50 }
70 switch (*(__u16 *) instruction) { 51 switch (insn[0]) {
71 case 0x0101: /* pr */ 52 case 0x0101: /* pr */
72 case 0xb25a: /* bsa */ 53 case 0xb25a: /* bsa */
73 case 0xb240: /* bakr */ 54 case 0xb240: /* bakr */
@@ -80,93 +61,92 @@ int __kprobes is_prohibited_opcode(kprobe_opcode_t *instruction)
80 return 0; 61 return 0;
81} 62}
82 63
83void __kprobes get_instruction_type(struct arch_specific_insn *ainsn) 64static int __kprobes get_fixup_type(kprobe_opcode_t *insn)
84{ 65{
85 /* default fixup method */ 66 /* default fixup method */
86 ainsn->fixup = FIXUP_PSW_NORMAL; 67 int fixup = FIXUP_PSW_NORMAL;
87
88 /* save r1 operand */
89 ainsn->reg = (*ainsn->insn & 0xf0) >> 4;
90
91 /* save the instruction length (pop 5-5) in bytes */
92 switch (*(__u8 *) (ainsn->insn) >> 6) {
93 case 0:
94 ainsn->ilen = 2;
95 break;
96 case 1:
97 case 2:
98 ainsn->ilen = 4;
99 break;
100 case 3:
101 ainsn->ilen = 6;
102 break;
103 }
104 68
105 switch (*(__u8 *) ainsn->insn) { 69 switch (insn[0] >> 8) {
106 case 0x05: /* balr */ 70 case 0x05: /* balr */
107 case 0x0d: /* basr */ 71 case 0x0d: /* basr */
108 ainsn->fixup = FIXUP_RETURN_REGISTER; 72 fixup = FIXUP_RETURN_REGISTER;
109 /* if r2 = 0, no branch will be taken */ 73 /* if r2 = 0, no branch will be taken */
110 if ((*ainsn->insn & 0x0f) == 0) 74 if ((insn[0] & 0x0f) == 0)
111 ainsn->fixup |= FIXUP_BRANCH_NOT_TAKEN; 75 fixup |= FIXUP_BRANCH_NOT_TAKEN;
112 break; 76 break;
113 case 0x06: /* bctr */ 77 case 0x06: /* bctr */
114 case 0x07: /* bcr */ 78 case 0x07: /* bcr */
115 ainsn->fixup = FIXUP_BRANCH_NOT_TAKEN; 79 fixup = FIXUP_BRANCH_NOT_TAKEN;
116 break; 80 break;
117 case 0x45: /* bal */ 81 case 0x45: /* bal */
118 case 0x4d: /* bas */ 82 case 0x4d: /* bas */
119 ainsn->fixup = FIXUP_RETURN_REGISTER; 83 fixup = FIXUP_RETURN_REGISTER;
120 break; 84 break;
121 case 0x47: /* bc */ 85 case 0x47: /* bc */
122 case 0x46: /* bct */ 86 case 0x46: /* bct */
123 case 0x86: /* bxh */ 87 case 0x86: /* bxh */
124 case 0x87: /* bxle */ 88 case 0x87: /* bxle */
125 ainsn->fixup = FIXUP_BRANCH_NOT_TAKEN; 89 fixup = FIXUP_BRANCH_NOT_TAKEN;
126 break; 90 break;
127 case 0x82: /* lpsw */ 91 case 0x82: /* lpsw */
128 ainsn->fixup = FIXUP_NOT_REQUIRED; 92 fixup = FIXUP_NOT_REQUIRED;
129 break; 93 break;
130 case 0xb2: /* lpswe */ 94 case 0xb2: /* lpswe */
131 if (*(((__u8 *) ainsn->insn) + 1) == 0xb2) { 95 if ((insn[0] & 0xff) == 0xb2)
132 ainsn->fixup = FIXUP_NOT_REQUIRED; 96 fixup = FIXUP_NOT_REQUIRED;
133 }
134 break; 97 break;
135 case 0xa7: /* bras */ 98 case 0xa7: /* bras */
136 if ((*ainsn->insn & 0x0f) == 0x05) { 99 if ((insn[0] & 0x0f) == 0x05)
137 ainsn->fixup |= FIXUP_RETURN_REGISTER; 100 fixup |= FIXUP_RETURN_REGISTER;
138 }
139 break; 101 break;
140 case 0xc0: 102 case 0xc0:
141 if ((*ainsn->insn & 0x0f) == 0x00 /* larl */ 103 if ((insn[0] & 0x0f) == 0x00 || /* larl */
142 || (*ainsn->insn & 0x0f) == 0x05) /* brasl */ 104 (insn[0] & 0x0f) == 0x05) /* brasl */
143 ainsn->fixup |= FIXUP_RETURN_REGISTER; 105 fixup |= FIXUP_RETURN_REGISTER;
144 break; 106 break;
145 case 0xeb: 107 case 0xeb:
146 if (*(((__u8 *) ainsn->insn) + 5 ) == 0x44 || /* bxhg */ 108 if ((insn[2] & 0xff) == 0x44 || /* bxhg */
147 *(((__u8 *) ainsn->insn) + 5) == 0x45) {/* bxleg */ 109 (insn[2] & 0xff) == 0x45) /* bxleg */
148 ainsn->fixup = FIXUP_BRANCH_NOT_TAKEN; 110 fixup = FIXUP_BRANCH_NOT_TAKEN;
149 }
150 break; 111 break;
151 case 0xe3: /* bctg */ 112 case 0xe3: /* bctg */
152 if (*(((__u8 *) ainsn->insn) + 5) == 0x46) { 113 if ((insn[2] & 0xff) == 0x46)
153 ainsn->fixup = FIXUP_BRANCH_NOT_TAKEN; 114 fixup = FIXUP_BRANCH_NOT_TAKEN;
154 }
155 break; 115 break;
156 } 116 }
117 return fixup;
157} 118}
158 119
120int __kprobes arch_prepare_kprobe(struct kprobe *p)
121{
122 if ((unsigned long) p->addr & 0x01)
123 return -EINVAL;
124
125 /* Make sure the probe isn't going on a difficult instruction */
126 if (is_prohibited_opcode(p->addr))
127 return -EINVAL;
128
129 p->opcode = *p->addr;
130 memcpy(p->ainsn.insn, p->addr, ((p->opcode >> 14) + 3) & -2);
131
132 return 0;
133}
134
135struct ins_replace_args {
136 kprobe_opcode_t *ptr;
137 kprobe_opcode_t opcode;
138};
139
159static int __kprobes swap_instruction(void *aref) 140static int __kprobes swap_instruction(void *aref)
160{ 141{
161 struct kprobe_ctlblk *kcb = get_kprobe_ctlblk(); 142 struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
162 unsigned long status = kcb->kprobe_status; 143 unsigned long status = kcb->kprobe_status;
163 struct ins_replace_args *args = aref; 144 struct ins_replace_args *args = aref;
164 int rc;
165 145
166 kcb->kprobe_status = KPROBE_SWAP_INST; 146 kcb->kprobe_status = KPROBE_SWAP_INST;
167 rc = probe_kernel_write(args->ptr, &args->new, sizeof(args->new)); 147 probe_kernel_write(args->ptr, &args->opcode, sizeof(args->opcode));
168 kcb->kprobe_status = status; 148 kcb->kprobe_status = status;
169 return rc; 149 return 0;
170} 150}
171 151
172void __kprobes arch_arm_kprobe(struct kprobe *p) 152void __kprobes arch_arm_kprobe(struct kprobe *p)
@@ -174,8 +154,7 @@ void __kprobes arch_arm_kprobe(struct kprobe *p)
174 struct ins_replace_args args; 154 struct ins_replace_args args;
175 155
176 args.ptr = p->addr; 156 args.ptr = p->addr;
177 args.old = p->opcode; 157 args.opcode = BREAKPOINT_INSTRUCTION;
178 args.new = BREAKPOINT_INSTRUCTION;
179 stop_machine(swap_instruction, &args, NULL); 158 stop_machine(swap_instruction, &args, NULL);
180} 159}
181 160
@@ -184,64 +163,69 @@ void __kprobes arch_disarm_kprobe(struct kprobe *p)
184 struct ins_replace_args args; 163 struct ins_replace_args args;
185 164
186 args.ptr = p->addr; 165 args.ptr = p->addr;
187 args.old = BREAKPOINT_INSTRUCTION; 166 args.opcode = p->opcode;
188 args.new = p->opcode;
189 stop_machine(swap_instruction, &args, NULL); 167 stop_machine(swap_instruction, &args, NULL);
190} 168}
191 169
192void __kprobes arch_remove_kprobe(struct kprobe *p) 170void __kprobes arch_remove_kprobe(struct kprobe *p)
193{ 171{
194 if (p->ainsn.insn) {
195 free_insn_slot(p->ainsn.insn, 0);
196 p->ainsn.insn = NULL;
197 }
198} 172}
199 173
200static void __kprobes prepare_singlestep(struct kprobe *p, struct pt_regs *regs) 174static void __kprobes enable_singlestep(struct kprobe_ctlblk *kcb,
175 struct pt_regs *regs,
176 unsigned long ip)
201{ 177{
202 per_cr_bits kprobe_per_regs[1]; 178 struct per_regs per_kprobe;
203 179
204 memset(kprobe_per_regs, 0, sizeof(per_cr_bits)); 180 /* Set up the PER control registers %cr9-%cr11 */
205 regs->psw.addr = (unsigned long)p->ainsn.insn | PSW_ADDR_AMODE; 181 per_kprobe.control = PER_EVENT_IFETCH;
182 per_kprobe.start = ip;
183 per_kprobe.end = ip;
206 184
207 /* Set up the per control reg info, will pass to lctl */ 185 /* Save control regs and psw mask */
208 kprobe_per_regs[0].em_instruction_fetch = 1; 186 __ctl_store(kcb->kprobe_saved_ctl, 9, 11);
209 kprobe_per_regs[0].starting_addr = (unsigned long)p->ainsn.insn; 187 kcb->kprobe_saved_imask = regs->psw.mask &
210 kprobe_per_regs[0].ending_addr = (unsigned long)p->ainsn.insn + 1; 188 (PSW_MASK_PER | PSW_MASK_IO | PSW_MASK_EXT);
211 189
212 /* Set the PER control regs, turns on single step for this address */ 190 /* Set PER control regs, turns on single step for the given address */
213 __ctl_load(kprobe_per_regs, 9, 11); 191 __ctl_load(per_kprobe, 9, 11);
214 regs->psw.mask |= PSW_MASK_PER; 192 regs->psw.mask |= PSW_MASK_PER;
215 regs->psw.mask &= ~(PSW_MASK_IO | PSW_MASK_EXT | PSW_MASK_MCHECK); 193 regs->psw.mask &= ~(PSW_MASK_IO | PSW_MASK_EXT);
194 regs->psw.addr = ip | PSW_ADDR_AMODE;
216} 195}
217 196
218static void __kprobes save_previous_kprobe(struct kprobe_ctlblk *kcb) 197static void __kprobes disable_singlestep(struct kprobe_ctlblk *kcb,
198 struct pt_regs *regs,
199 unsigned long ip)
219{ 200{
220 kcb->prev_kprobe.kp = kprobe_running(); 201 /* Restore control regs and psw mask, set new psw address */
221 kcb->prev_kprobe.status = kcb->kprobe_status; 202 __ctl_load(kcb->kprobe_saved_ctl, 9, 11);
222 kcb->prev_kprobe.kprobe_saved_imask = kcb->kprobe_saved_imask; 203 regs->psw.mask &= ~PSW_MASK_PER;
223 memcpy(kcb->prev_kprobe.kprobe_saved_ctl, kcb->kprobe_saved_ctl, 204 regs->psw.mask |= kcb->kprobe_saved_imask;
224 sizeof(kcb->kprobe_saved_ctl)); 205 regs->psw.addr = ip | PSW_ADDR_AMODE;
225} 206}
226 207
227static void __kprobes restore_previous_kprobe(struct kprobe_ctlblk *kcb) 208/*
209 * Activate a kprobe by storing its pointer to current_kprobe. The
210 * previous kprobe is stored in kcb->prev_kprobe. A stack of up to
211 * two kprobes can be active, see KPROBE_REENTER.
212 */
213static void __kprobes push_kprobe(struct kprobe_ctlblk *kcb, struct kprobe *p)
228{ 214{
229 __get_cpu_var(current_kprobe) = kcb->prev_kprobe.kp; 215 kcb->prev_kprobe.kp = __get_cpu_var(current_kprobe);
230 kcb->kprobe_status = kcb->prev_kprobe.status; 216 kcb->prev_kprobe.status = kcb->kprobe_status;
231 kcb->kprobe_saved_imask = kcb->prev_kprobe.kprobe_saved_imask; 217 __get_cpu_var(current_kprobe) = p;
232 memcpy(kcb->kprobe_saved_ctl, kcb->prev_kprobe.kprobe_saved_ctl,
233 sizeof(kcb->kprobe_saved_ctl));
234} 218}
235 219
236static void __kprobes set_current_kprobe(struct kprobe *p, struct pt_regs *regs, 220/*
237 struct kprobe_ctlblk *kcb) 221 * Deactivate a kprobe by backing up to the previous state. If the
222 * current state is KPROBE_REENTER prev_kprobe.kp will be non-NULL,
223 * for any other state prev_kprobe.kp will be NULL.
224 */
225static void __kprobes pop_kprobe(struct kprobe_ctlblk *kcb)
238{ 226{
239 __get_cpu_var(current_kprobe) = p; 227 __get_cpu_var(current_kprobe) = kcb->prev_kprobe.kp;
240 /* Save the interrupt and per flags */ 228 kcb->kprobe_status = kcb->prev_kprobe.status;
241 kcb->kprobe_saved_imask = regs->psw.mask &
242 (PSW_MASK_PER | PSW_MASK_IO | PSW_MASK_EXT | PSW_MASK_MCHECK);
243 /* Save the control regs that govern PER */
244 __ctl_store(kcb->kprobe_saved_ctl, 9, 11);
245} 229}
246 230
247void __kprobes arch_prepare_kretprobe(struct kretprobe_instance *ri, 231void __kprobes arch_prepare_kretprobe(struct kretprobe_instance *ri,
@@ -250,79 +234,104 @@ void __kprobes arch_prepare_kretprobe(struct kretprobe_instance *ri,
250 ri->ret_addr = (kprobe_opcode_t *) regs->gprs[14]; 234 ri->ret_addr = (kprobe_opcode_t *) regs->gprs[14];
251 235
252 /* Replace the return addr with trampoline addr */ 236 /* Replace the return addr with trampoline addr */
253 regs->gprs[14] = (unsigned long)&kretprobe_trampoline; 237 regs->gprs[14] = (unsigned long) &kretprobe_trampoline;
238}
239
240static void __kprobes kprobe_reenter_check(struct kprobe_ctlblk *kcb,
241 struct kprobe *p)
242{
243 switch (kcb->kprobe_status) {
244 case KPROBE_HIT_SSDONE:
245 case KPROBE_HIT_ACTIVE:
246 kprobes_inc_nmissed_count(p);
247 break;
248 case KPROBE_HIT_SS:
249 case KPROBE_REENTER:
250 default:
251 /*
252 * A kprobe on the code path to single step an instruction
253 * is a BUG. The code path resides in the .kprobes.text
254 * section and is executed with interrupts disabled.
255 */
256 printk(KERN_EMERG "Invalid kprobe detected at %p.\n", p->addr);
257 dump_kprobe(p);
258 BUG();
259 }
254} 260}
255 261
256static int __kprobes kprobe_handler(struct pt_regs *regs) 262static int __kprobes kprobe_handler(struct pt_regs *regs)
257{ 263{
258 struct kprobe *p;
259 int ret = 0;
260 unsigned long *addr = (unsigned long *)
261 ((regs->psw.addr & PSW_ADDR_INSN) - 2);
262 struct kprobe_ctlblk *kcb; 264 struct kprobe_ctlblk *kcb;
265 struct kprobe *p;
263 266
264 /* 267 /*
265 * We don't want to be preempted for the entire 268 * We want to disable preemption for the entire duration of kprobe
266 * duration of kprobe processing 269 * processing. That includes the calls to the pre/post handlers
270 * and single stepping the kprobe instruction.
267 */ 271 */
268 preempt_disable(); 272 preempt_disable();
269 kcb = get_kprobe_ctlblk(); 273 kcb = get_kprobe_ctlblk();
274 p = get_kprobe((void *)((regs->psw.addr & PSW_ADDR_INSN) - 2));
270 275
271 /* Check we're not actually recursing */ 276 if (p) {
272 if (kprobe_running()) { 277 if (kprobe_running()) {
273 p = get_kprobe(addr); 278 /*
274 if (p) { 279 * We have hit a kprobe while another is still
275 if (kcb->kprobe_status == KPROBE_HIT_SS && 280 * active. This can happen in the pre and post
276 *p->ainsn.insn == BREAKPOINT_INSTRUCTION) { 281 * handler. Single step the instruction of the
277 regs->psw.mask &= ~PSW_MASK_PER; 282 * new probe but do not call any handler function
278 regs->psw.mask |= kcb->kprobe_saved_imask; 283 * of this secondary kprobe.
279 goto no_kprobe; 284 * push_kprobe and pop_kprobe saves and restores
280 } 285 * the currently active kprobe.
281 /* We have reentered the kprobe_handler(), since
282 * another probe was hit while within the handler.
283 * We here save the original kprobes variables and
284 * just single step on the instruction of the new probe
285 * without calling any user handlers.
286 */ 286 */
287 save_previous_kprobe(kcb); 287 kprobe_reenter_check(kcb, p);
288 set_current_kprobe(p, regs, kcb); 288 push_kprobe(kcb, p);
289 kprobes_inc_nmissed_count(p);
290 prepare_singlestep(p, regs);
291 kcb->kprobe_status = KPROBE_REENTER; 289 kcb->kprobe_status = KPROBE_REENTER;
292 return 1;
293 } else { 290 } else {
294 p = __get_cpu_var(current_kprobe); 291 /*
295 if (p->break_handler && p->break_handler(p, regs)) { 292 * If we have no pre-handler or it returned 0, we
296 goto ss_probe; 293 * continue with single stepping. If we have a
297 } 294 * pre-handler and it returned non-zero, it prepped
295 * for calling the break_handler below on re-entry
296 * for jprobe processing, so get out doing nothing
297 * more here.
298 */
299 push_kprobe(kcb, p);
300 kcb->kprobe_status = KPROBE_HIT_ACTIVE;
301 if (p->pre_handler && p->pre_handler(p, regs))
302 return 1;
303 kcb->kprobe_status = KPROBE_HIT_SS;
298 } 304 }
299 goto no_kprobe; 305 enable_singlestep(kcb, regs, (unsigned long) p->ainsn.insn);
300 }
301
302 p = get_kprobe(addr);
303 if (!p)
304 /*
305 * No kprobe at this address. The fault has not been
306 * caused by a kprobe breakpoint. The race of breakpoint
307 * vs. kprobe remove does not exist because on s390 we
308 * use stop_machine to arm/disarm the breakpoints.
309 */
310 goto no_kprobe;
311
312 kcb->kprobe_status = KPROBE_HIT_ACTIVE;
313 set_current_kprobe(p, regs, kcb);
314 if (p->pre_handler && p->pre_handler(p, regs))
315 /* handler has already set things up, so skip ss setup */
316 return 1; 306 return 1;
317 307 } else if (kprobe_running()) {
318ss_probe: 308 p = __get_cpu_var(current_kprobe);
319 prepare_singlestep(p, regs); 309 if (p->break_handler && p->break_handler(p, regs)) {
320 kcb->kprobe_status = KPROBE_HIT_SS; 310 /*
321 return 1; 311 * Continuation after the jprobe completed and
322 312 * caused the jprobe_return trap. The jprobe
323no_kprobe: 313 * break_handler "returns" to the original
314 * function that still has the kprobe breakpoint
315 * installed. We continue with single stepping.
316 */
317 kcb->kprobe_status = KPROBE_HIT_SS;
318 enable_singlestep(kcb, regs,
319 (unsigned long) p->ainsn.insn);
320 return 1;
321 } /* else:
322 * No kprobe at this address and the current kprobe
323 * has no break handler (no jprobe!). The kernel just
324 * exploded, let the standard trap handler pick up the
325 * pieces.
326 */
327 } /* else:
328 * No kprobe at this address and no active kprobe. The trap has
329 * not been caused by a kprobe breakpoint. The race of breakpoint
330 * vs. kprobe remove does not exist because on s390 as we use
331 * stop_machine to arm/disarm the breakpoints.
332 */
324 preempt_enable_no_resched(); 333 preempt_enable_no_resched();
325 return ret; 334 return 0;
326} 335}
327 336
328/* 337/*
@@ -343,11 +352,12 @@ static void __used kretprobe_trampoline_holder(void)
343static int __kprobes trampoline_probe_handler(struct kprobe *p, 352static int __kprobes trampoline_probe_handler(struct kprobe *p,
344 struct pt_regs *regs) 353 struct pt_regs *regs)
345{ 354{
346 struct kretprobe_instance *ri = NULL; 355 struct kretprobe_instance *ri;
347 struct hlist_head *head, empty_rp; 356 struct hlist_head *head, empty_rp;
348 struct hlist_node *node, *tmp; 357 struct hlist_node *node, *tmp;
349 unsigned long flags, orig_ret_address = 0; 358 unsigned long flags, orig_ret_address;
350 unsigned long trampoline_address = (unsigned long)&kretprobe_trampoline; 359 unsigned long trampoline_address;
360 kprobe_opcode_t *correct_ret_addr;
351 361
352 INIT_HLIST_HEAD(&empty_rp); 362 INIT_HLIST_HEAD(&empty_rp);
353 kretprobe_hash_lock(current, &head, &flags); 363 kretprobe_hash_lock(current, &head, &flags);
@@ -365,30 +375,55 @@ static int __kprobes trampoline_probe_handler(struct kprobe *p,
365 * real return address, and all the rest will point to 375 * real return address, and all the rest will point to
366 * kretprobe_trampoline 376 * kretprobe_trampoline
367 */ 377 */
378 ri = NULL;
379 orig_ret_address = 0;
380 correct_ret_addr = NULL;
381 trampoline_address = (unsigned long) &kretprobe_trampoline;
368 hlist_for_each_entry_safe(ri, node, tmp, head, hlist) { 382 hlist_for_each_entry_safe(ri, node, tmp, head, hlist) {
369 if (ri->task != current) 383 if (ri->task != current)
370 /* another task is sharing our hash bucket */ 384 /* another task is sharing our hash bucket */
371 continue; 385 continue;
372 386
373 if (ri->rp && ri->rp->handler) 387 orig_ret_address = (unsigned long) ri->ret_addr;
388
389 if (orig_ret_address != trampoline_address)
390 /*
391 * This is the real return address. Any other
392 * instances associated with this task are for
393 * other calls deeper on the call stack
394 */
395 break;
396 }
397
398 kretprobe_assert(ri, orig_ret_address, trampoline_address);
399
400 correct_ret_addr = ri->ret_addr;
401 hlist_for_each_entry_safe(ri, node, tmp, head, hlist) {
402 if (ri->task != current)
403 /* another task is sharing our hash bucket */
404 continue;
405
406 orig_ret_address = (unsigned long) ri->ret_addr;
407
408 if (ri->rp && ri->rp->handler) {
409 ri->ret_addr = correct_ret_addr;
374 ri->rp->handler(ri, regs); 410 ri->rp->handler(ri, regs);
411 }
375 412
376 orig_ret_address = (unsigned long)ri->ret_addr;
377 recycle_rp_inst(ri, &empty_rp); 413 recycle_rp_inst(ri, &empty_rp);
378 414
379 if (orig_ret_address != trampoline_address) { 415 if (orig_ret_address != trampoline_address)
380 /* 416 /*
381 * This is the real return address. Any other 417 * This is the real return address. Any other
382 * instances associated with this task are for 418 * instances associated with this task are for
383 * other calls deeper on the call stack 419 * other calls deeper on the call stack
384 */ 420 */
385 break; 421 break;
386 }
387 } 422 }
388 kretprobe_assert(ri, orig_ret_address, trampoline_address); 423
389 regs->psw.addr = orig_ret_address | PSW_ADDR_AMODE; 424 regs->psw.addr = orig_ret_address | PSW_ADDR_AMODE;
390 425
391 reset_current_kprobe(); 426 pop_kprobe(get_kprobe_ctlblk());
392 kretprobe_hash_unlock(current, &flags); 427 kretprobe_hash_unlock(current, &flags);
393 preempt_enable_no_resched(); 428 preempt_enable_no_resched();
394 429
@@ -415,55 +450,42 @@ static int __kprobes trampoline_probe_handler(struct kprobe *p,
415static void __kprobes resume_execution(struct kprobe *p, struct pt_regs *regs) 450static void __kprobes resume_execution(struct kprobe *p, struct pt_regs *regs)
416{ 451{
417 struct kprobe_ctlblk *kcb = get_kprobe_ctlblk(); 452 struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
453 unsigned long ip = regs->psw.addr & PSW_ADDR_INSN;
454 int fixup = get_fixup_type(p->ainsn.insn);
418 455
419 regs->psw.addr &= PSW_ADDR_INSN; 456 if (fixup & FIXUP_PSW_NORMAL)
420 457 ip += (unsigned long) p->addr - (unsigned long) p->ainsn.insn;
421 if (p->ainsn.fixup & FIXUP_PSW_NORMAL)
422 regs->psw.addr = (unsigned long)p->addr +
423 ((unsigned long)regs->psw.addr -
424 (unsigned long)p->ainsn.insn);
425 458
426 if (p->ainsn.fixup & FIXUP_BRANCH_NOT_TAKEN) 459 if (fixup & FIXUP_BRANCH_NOT_TAKEN) {
427 if ((unsigned long)regs->psw.addr - 460 int ilen = ((p->ainsn.insn[0] >> 14) + 3) & -2;
428 (unsigned long)p->ainsn.insn == p->ainsn.ilen) 461 if (ip - (unsigned long) p->ainsn.insn == ilen)
429 regs->psw.addr = (unsigned long)p->addr + p->ainsn.ilen; 462 ip = (unsigned long) p->addr + ilen;
463 }
430 464
431 if (p->ainsn.fixup & FIXUP_RETURN_REGISTER) 465 if (fixup & FIXUP_RETURN_REGISTER) {
432 regs->gprs[p->ainsn.reg] = ((unsigned long)p->addr + 466 int reg = (p->ainsn.insn[0] & 0xf0) >> 4;
433 (regs->gprs[p->ainsn.reg] - 467 regs->gprs[reg] += (unsigned long) p->addr -
434 (unsigned long)p->ainsn.insn)) 468 (unsigned long) p->ainsn.insn;
435 | PSW_ADDR_AMODE; 469 }
436 470
437 regs->psw.addr |= PSW_ADDR_AMODE; 471 disable_singlestep(kcb, regs, ip);
438 /* turn off PER mode */
439 regs->psw.mask &= ~PSW_MASK_PER;
440 /* Restore the original per control regs */
441 __ctl_load(kcb->kprobe_saved_ctl, 9, 11);
442 regs->psw.mask |= kcb->kprobe_saved_imask;
443} 472}
444 473
445static int __kprobes post_kprobe_handler(struct pt_regs *regs) 474static int __kprobes post_kprobe_handler(struct pt_regs *regs)
446{ 475{
447 struct kprobe *cur = kprobe_running();
448 struct kprobe_ctlblk *kcb = get_kprobe_ctlblk(); 476 struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
477 struct kprobe *p = kprobe_running();
449 478
450 if (!cur) 479 if (!p)
451 return 0; 480 return 0;
452 481
453 if ((kcb->kprobe_status != KPROBE_REENTER) && cur->post_handler) { 482 if (kcb->kprobe_status != KPROBE_REENTER && p->post_handler) {
454 kcb->kprobe_status = KPROBE_HIT_SSDONE; 483 kcb->kprobe_status = KPROBE_HIT_SSDONE;
455 cur->post_handler(cur, regs, 0); 484 p->post_handler(p, regs, 0);
456 } 485 }
457 486
458 resume_execution(cur, regs); 487 resume_execution(p, regs);
459 488 pop_kprobe(kcb);
460 /*Restore back the original saved kprobes variables and continue. */
461 if (kcb->kprobe_status == KPROBE_REENTER) {
462 restore_previous_kprobe(kcb);
463 goto out;
464 }
465 reset_current_kprobe();
466out:
467 preempt_enable_no_resched(); 489 preempt_enable_no_resched();
468 490
469 /* 491 /*
@@ -471,17 +493,16 @@ out:
471 * will have PER set, in which case, continue the remaining processing 493 * will have PER set, in which case, continue the remaining processing
472 * of do_single_step, as if this is not a probe hit. 494 * of do_single_step, as if this is not a probe hit.
473 */ 495 */
474 if (regs->psw.mask & PSW_MASK_PER) { 496 if (regs->psw.mask & PSW_MASK_PER)
475 return 0; 497 return 0;
476 }
477 498
478 return 1; 499 return 1;
479} 500}
480 501
481int __kprobes kprobe_fault_handler(struct pt_regs *regs, int trapnr) 502static int __kprobes kprobe_trap_handler(struct pt_regs *regs, int trapnr)
482{ 503{
483 struct kprobe *cur = kprobe_running();
484 struct kprobe_ctlblk *kcb = get_kprobe_ctlblk(); 504 struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
505 struct kprobe *p = kprobe_running();
485 const struct exception_table_entry *entry; 506 const struct exception_table_entry *entry;
486 507
487 switch(kcb->kprobe_status) { 508 switch(kcb->kprobe_status) {
@@ -497,13 +518,8 @@ int __kprobes kprobe_fault_handler(struct pt_regs *regs, int trapnr)
497 * and allow the page fault handler to continue as a 518 * and allow the page fault handler to continue as a
498 * normal page fault. 519 * normal page fault.
499 */ 520 */
500 regs->psw.addr = (unsigned long)cur->addr | PSW_ADDR_AMODE; 521 disable_singlestep(kcb, regs, (unsigned long) p->addr);
501 regs->psw.mask &= ~PSW_MASK_PER; 522 pop_kprobe(kcb);
502 regs->psw.mask |= kcb->kprobe_saved_imask;
503 if (kcb->kprobe_status == KPROBE_REENTER)
504 restore_previous_kprobe(kcb);
505 else
506 reset_current_kprobe();
507 preempt_enable_no_resched(); 523 preempt_enable_no_resched();
508 break; 524 break;
509 case KPROBE_HIT_ACTIVE: 525 case KPROBE_HIT_ACTIVE:
@@ -513,7 +529,7 @@ int __kprobes kprobe_fault_handler(struct pt_regs *regs, int trapnr)
513 * we can also use npre/npostfault count for accouting 529 * we can also use npre/npostfault count for accouting
514 * these specific fault cases. 530 * these specific fault cases.
515 */ 531 */
516 kprobes_inc_nmissed_count(cur); 532 kprobes_inc_nmissed_count(p);
517 533
518 /* 534 /*
519 * We come here because instructions in the pre/post 535 * We come here because instructions in the pre/post
@@ -522,7 +538,7 @@ int __kprobes kprobe_fault_handler(struct pt_regs *regs, int trapnr)
522 * copy_from_user(), get_user() etc. Let the 538 * copy_from_user(), get_user() etc. Let the
523 * user-specified handler try to fix it first. 539 * user-specified handler try to fix it first.
524 */ 540 */
525 if (cur->fault_handler && cur->fault_handler(cur, regs, trapnr)) 541 if (p->fault_handler && p->fault_handler(p, regs, trapnr))
526 return 1; 542 return 1;
527 543
528 /* 544 /*
@@ -546,57 +562,71 @@ int __kprobes kprobe_fault_handler(struct pt_regs *regs, int trapnr)
546 return 0; 562 return 0;
547} 563}
548 564
565int __kprobes kprobe_fault_handler(struct pt_regs *regs, int trapnr)
566{
567 int ret;
568
569 if (regs->psw.mask & (PSW_MASK_IO | PSW_MASK_EXT))
570 local_irq_disable();
571 ret = kprobe_trap_handler(regs, trapnr);
572 if (regs->psw.mask & (PSW_MASK_IO | PSW_MASK_EXT))
573 local_irq_restore(regs->psw.mask & ~PSW_MASK_PER);
574 return ret;
575}
576
549/* 577/*
550 * Wrapper routine to for handling exceptions. 578 * Wrapper routine to for handling exceptions.
551 */ 579 */
552int __kprobes kprobe_exceptions_notify(struct notifier_block *self, 580int __kprobes kprobe_exceptions_notify(struct notifier_block *self,
553 unsigned long val, void *data) 581 unsigned long val, void *data)
554{ 582{
555 struct die_args *args = (struct die_args *)data; 583 struct die_args *args = (struct die_args *) data;
584 struct pt_regs *regs = args->regs;
556 int ret = NOTIFY_DONE; 585 int ret = NOTIFY_DONE;
557 586
587 if (regs->psw.mask & (PSW_MASK_IO | PSW_MASK_EXT))
588 local_irq_disable();
589
558 switch (val) { 590 switch (val) {
559 case DIE_BPT: 591 case DIE_BPT:
560 if (kprobe_handler(args->regs)) 592 if (kprobe_handler(regs))
561 ret = NOTIFY_STOP; 593 ret = NOTIFY_STOP;
562 break; 594 break;
563 case DIE_SSTEP: 595 case DIE_SSTEP:
564 if (post_kprobe_handler(args->regs)) 596 if (post_kprobe_handler(regs))
565 ret = NOTIFY_STOP; 597 ret = NOTIFY_STOP;
566 break; 598 break;
567 case DIE_TRAP: 599 case DIE_TRAP:
568 /* kprobe_running() needs smp_processor_id() */ 600 if (!preemptible() && kprobe_running() &&
569 preempt_disable(); 601 kprobe_trap_handler(regs, args->trapnr))
570 if (kprobe_running() &&
571 kprobe_fault_handler(args->regs, args->trapnr))
572 ret = NOTIFY_STOP; 602 ret = NOTIFY_STOP;
573 preempt_enable();
574 break; 603 break;
575 default: 604 default:
576 break; 605 break;
577 } 606 }
607
608 if (regs->psw.mask & (PSW_MASK_IO | PSW_MASK_EXT))
609 local_irq_restore(regs->psw.mask & ~PSW_MASK_PER);
610
578 return ret; 611 return ret;
579} 612}
580 613
581int __kprobes setjmp_pre_handler(struct kprobe *p, struct pt_regs *regs) 614int __kprobes setjmp_pre_handler(struct kprobe *p, struct pt_regs *regs)
582{ 615{
583 struct jprobe *jp = container_of(p, struct jprobe, kp); 616 struct jprobe *jp = container_of(p, struct jprobe, kp);
584 unsigned long addr;
585 struct kprobe_ctlblk *kcb = get_kprobe_ctlblk(); 617 struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
618 unsigned long stack;
586 619
587 memcpy(&kcb->jprobe_saved_regs, regs, sizeof(struct pt_regs)); 620 memcpy(&kcb->jprobe_saved_regs, regs, sizeof(struct pt_regs));
588 621
589 /* setup return addr to the jprobe handler routine */ 622 /* setup return addr to the jprobe handler routine */
590 regs->psw.addr = (unsigned long)(jp->entry) | PSW_ADDR_AMODE; 623 regs->psw.addr = (unsigned long) jp->entry | PSW_ADDR_AMODE;
624 regs->psw.mask &= ~(PSW_MASK_IO | PSW_MASK_EXT);
591 625
592 /* r14 is the function return address */
593 kcb->jprobe_saved_r14 = (unsigned long)regs->gprs[14];
594 /* r15 is the stack pointer */ 626 /* r15 is the stack pointer */
595 kcb->jprobe_saved_r15 = (unsigned long)regs->gprs[15]; 627 stack = (unsigned long) regs->gprs[15];
596 addr = (unsigned long)kcb->jprobe_saved_r15;
597 628
598 memcpy(kcb->jprobes_stack, (kprobe_opcode_t *) addr, 629 memcpy(kcb->jprobes_stack, (void *) stack, MIN_STACK_SIZE(stack));
599 MIN_STACK_SIZE(addr));
600 return 1; 630 return 1;
601} 631}
602 632
@@ -613,30 +643,29 @@ void __kprobes jprobe_return_end(void)
613int __kprobes longjmp_break_handler(struct kprobe *p, struct pt_regs *regs) 643int __kprobes longjmp_break_handler(struct kprobe *p, struct pt_regs *regs)
614{ 644{
615 struct kprobe_ctlblk *kcb = get_kprobe_ctlblk(); 645 struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
616 unsigned long stack_addr = (unsigned long)(kcb->jprobe_saved_r15); 646 unsigned long stack;
647
648 stack = (unsigned long) kcb->jprobe_saved_regs.gprs[15];
617 649
618 /* Put the regs back */ 650 /* Put the regs back */
619 memcpy(regs, &kcb->jprobe_saved_regs, sizeof(struct pt_regs)); 651 memcpy(regs, &kcb->jprobe_saved_regs, sizeof(struct pt_regs));
620 /* put the stack back */ 652 /* put the stack back */
621 memcpy((kprobe_opcode_t *) stack_addr, kcb->jprobes_stack, 653 memcpy((void *) stack, kcb->jprobes_stack, MIN_STACK_SIZE(stack));
622 MIN_STACK_SIZE(stack_addr));
623 preempt_enable_no_resched(); 654 preempt_enable_no_resched();
624 return 1; 655 return 1;
625} 656}
626 657
627static struct kprobe trampoline_p = { 658static struct kprobe trampoline = {
628 .addr = (kprobe_opcode_t *) & kretprobe_trampoline, 659 .addr = (kprobe_opcode_t *) &kretprobe_trampoline,
629 .pre_handler = trampoline_probe_handler 660 .pre_handler = trampoline_probe_handler
630}; 661};
631 662
632int __init arch_init_kprobes(void) 663int __init arch_init_kprobes(void)
633{ 664{
634 return register_kprobe(&trampoline_p); 665 return register_kprobe(&trampoline);
635} 666}
636 667
637int __kprobes arch_trampoline_kprobe(struct kprobe *p) 668int __kprobes arch_trampoline_kprobe(struct kprobe *p)
638{ 669{
639 if (p->addr == (kprobe_opcode_t *) & kretprobe_trampoline) 670 return p->addr == (kprobe_opcode_t *) &kretprobe_trampoline;
640 return 1;
641 return 0;
642} 671}
diff --git a/arch/s390/kernel/machine_kexec.c b/arch/s390/kernel/machine_kexec.c
index a922d51df6bf..b09b9c62573e 100644
--- a/arch/s390/kernel/machine_kexec.c
+++ b/arch/s390/kernel/machine_kexec.c
@@ -12,6 +12,7 @@
12#include <linux/kexec.h> 12#include <linux/kexec.h>
13#include <linux/delay.h> 13#include <linux/delay.h>
14#include <linux/reboot.h> 14#include <linux/reboot.h>
15#include <linux/ftrace.h>
15#include <asm/cio.h> 16#include <asm/cio.h>
16#include <asm/setup.h> 17#include <asm/setup.h>
17#include <asm/pgtable.h> 18#include <asm/pgtable.h>
@@ -71,6 +72,7 @@ static void __machine_kexec(void *data)
71 72
72void machine_kexec(struct kimage *image) 73void machine_kexec(struct kimage *image)
73{ 74{
75 tracer_disable();
74 smp_send_stop(); 76 smp_send_stop();
75 smp_switch_to_ipl_cpu(__machine_kexec, image); 77 smp_switch_to_ipl_cpu(__machine_kexec, image);
76} 78}
diff --git a/arch/s390/kernel/mcount.S b/arch/s390/kernel/mcount.S
index dfe015d7398c..1e6a55795628 100644
--- a/arch/s390/kernel/mcount.S
+++ b/arch/s390/kernel/mcount.S
@@ -7,6 +7,8 @@
7 7
8#include <asm/asm-offsets.h> 8#include <asm/asm-offsets.h>
9 9
10 .section .kprobes.text, "ax"
11
10 .globl ftrace_stub 12 .globl ftrace_stub
11ftrace_stub: 13ftrace_stub:
12 br %r14 14 br %r14
@@ -16,22 +18,12 @@ _mcount:
16#ifdef CONFIG_DYNAMIC_FTRACE 18#ifdef CONFIG_DYNAMIC_FTRACE
17 br %r14 19 br %r14
18 20
19 .data
20 .globl ftrace_dyn_func
21ftrace_dyn_func:
22 .long ftrace_stub
23 .previous
24
25 .globl ftrace_caller 21 .globl ftrace_caller
26ftrace_caller: 22ftrace_caller:
27#endif 23#endif
28 stm %r2,%r5,16(%r15) 24 stm %r2,%r5,16(%r15)
29 bras %r1,2f 25 bras %r1,2f
30#ifdef CONFIG_DYNAMIC_FTRACE
310: .long ftrace_dyn_func
32#else
330: .long ftrace_trace_function 260: .long ftrace_trace_function
34#endif
351: .long function_trace_stop 271: .long function_trace_stop
362: l %r2,1b-0b(%r1) 282: l %r2,1b-0b(%r1)
37 icm %r2,0xf,0(%r2) 29 icm %r2,0xf,0(%r2)
@@ -47,21 +39,15 @@ ftrace_caller:
47 l %r14,0(%r14) 39 l %r14,0(%r14)
48 basr %r14,%r14 40 basr %r14,%r14
49#ifdef CONFIG_FUNCTION_GRAPH_TRACER 41#ifdef CONFIG_FUNCTION_GRAPH_TRACER
50#ifdef CONFIG_DYNAMIC_FTRACE 42 l %r2,100(%r15)
43 l %r3,152(%r15)
51 .globl ftrace_graph_caller 44 .globl ftrace_graph_caller
52ftrace_graph_caller: 45ftrace_graph_caller:
53 # This unconditional branch gets runtime patched. Change only if 46# The bras instruction gets runtime patched to call prepare_ftrace_return.
54 # you know what you are doing. See ftrace_enable_graph_caller(). 47# See ftrace_enable_ftrace_graph_caller. The patched instruction is:
55 j 1f 48# bras %r14,prepare_ftrace_return
56#endif 49 bras %r14,0f
57 bras %r1,0f 500: st %r2,100(%r15)
58 .long prepare_ftrace_return
590: l %r2,152(%r15)
60 l %r4,0(%r1)
61 l %r3,100(%r15)
62 basr %r14,%r4
63 st %r2,100(%r15)
641:
65#endif 51#endif
66 ahi %r15,96 52 ahi %r15,96
67 l %r14,56(%r15) 53 l %r14,56(%r15)
diff --git a/arch/s390/kernel/mcount64.S b/arch/s390/kernel/mcount64.S
index c37211c6092b..e73667286ac0 100644
--- a/arch/s390/kernel/mcount64.S
+++ b/arch/s390/kernel/mcount64.S
@@ -7,6 +7,8 @@
7 7
8#include <asm/asm-offsets.h> 8#include <asm/asm-offsets.h>
9 9
10 .section .kprobes.text, "ax"
11
10 .globl ftrace_stub 12 .globl ftrace_stub
11ftrace_stub: 13ftrace_stub:
12 br %r14 14 br %r14
@@ -16,12 +18,6 @@ _mcount:
16#ifdef CONFIG_DYNAMIC_FTRACE 18#ifdef CONFIG_DYNAMIC_FTRACE
17 br %r14 19 br %r14
18 20
19 .data
20 .globl ftrace_dyn_func
21ftrace_dyn_func:
22 .quad ftrace_stub
23 .previous
24
25 .globl ftrace_caller 21 .globl ftrace_caller
26ftrace_caller: 22ftrace_caller:
27#endif 23#endif
@@ -35,26 +31,19 @@ ftrace_caller:
35 stg %r1,__SF_BACKCHAIN(%r15) 31 stg %r1,__SF_BACKCHAIN(%r15)
36 lgr %r2,%r14 32 lgr %r2,%r14
37 lg %r3,168(%r15) 33 lg %r3,168(%r15)
38#ifdef CONFIG_DYNAMIC_FTRACE
39 larl %r14,ftrace_dyn_func
40#else
41 larl %r14,ftrace_trace_function 34 larl %r14,ftrace_trace_function
42#endif
43 lg %r14,0(%r14) 35 lg %r14,0(%r14)
44 basr %r14,%r14 36 basr %r14,%r14
45#ifdef CONFIG_FUNCTION_GRAPH_TRACER 37#ifdef CONFIG_FUNCTION_GRAPH_TRACER
46#ifdef CONFIG_DYNAMIC_FTRACE 38 lg %r2,168(%r15)
39 lg %r3,272(%r15)
47 .globl ftrace_graph_caller 40 .globl ftrace_graph_caller
48ftrace_graph_caller: 41ftrace_graph_caller:
49 # This unconditional branch gets runtime patched. Change only if 42# The bras instruction gets runtime patched to call prepare_ftrace_return.
50 # you know what you are doing. See ftrace_enable_graph_caller(). 43# See ftrace_enable_ftrace_graph_caller. The patched instruction is:
51 j 0f 44# bras %r14,prepare_ftrace_return
52#endif 45 bras %r14,0f
53 lg %r2,272(%r15) 460: stg %r2,168(%r15)
54 lg %r3,168(%r15)
55 brasl %r14,prepare_ftrace_return
56 stg %r2,168(%r15)
570:
58#endif 47#endif
59 aghi %r15,160 48 aghi %r15,160
60 lmg %r2,%r5,32(%r15) 49 lmg %r2,%r5,32(%r15)
diff --git a/arch/s390/kernel/mem_detect.c b/arch/s390/kernel/mem_detect.c
index 559af0d07878..0fbe4e32f7ba 100644
--- a/arch/s390/kernel/mem_detect.c
+++ b/arch/s390/kernel/mem_detect.c
@@ -54,11 +54,11 @@ void detect_memory_layout(struct mem_chunk chunk[])
54 * right thing and we don't get scheduled away with low address 54 * right thing and we don't get scheduled away with low address
55 * protection disabled. 55 * protection disabled.
56 */ 56 */
57 flags = __raw_local_irq_stnsm(0xf8); 57 flags = __arch_local_irq_stnsm(0xf8);
58 __ctl_store(cr0, 0, 0); 58 __ctl_store(cr0, 0, 0);
59 __ctl_clear_bit(0, 28); 59 __ctl_clear_bit(0, 28);
60 find_memory_chunks(chunk); 60 find_memory_chunks(chunk);
61 __ctl_load(cr0, 0, 0); 61 __ctl_load(cr0, 0, 0);
62 __raw_local_irq_ssm(flags); 62 arch_local_irq_restore(flags);
63} 63}
64EXPORT_SYMBOL(detect_memory_layout); 64EXPORT_SYMBOL(detect_memory_layout);
diff --git a/arch/s390/kernel/nmi.c b/arch/s390/kernel/nmi.c
index ac151399ef34..fab88431a06f 100644
--- a/arch/s390/kernel/nmi.c
+++ b/arch/s390/kernel/nmi.c
@@ -8,6 +8,7 @@
8 * Heiko Carstens <heiko.carstens@de.ibm.com>, 8 * Heiko Carstens <heiko.carstens@de.ibm.com>,
9 */ 9 */
10 10
11#include <linux/kernel_stat.h>
11#include <linux/init.h> 12#include <linux/init.h>
12#include <linux/errno.h> 13#include <linux/errno.h>
13#include <linux/hardirq.h> 14#include <linux/hardirq.h>
@@ -95,7 +96,6 @@ EXPORT_SYMBOL_GPL(s390_handle_mcck);
95static int notrace s390_revalidate_registers(struct mci *mci) 96static int notrace s390_revalidate_registers(struct mci *mci)
96{ 97{
97 int kill_task; 98 int kill_task;
98 u64 tmpclock;
99 u64 zero; 99 u64 zero;
100 void *fpt_save_area, *fpt_creg_save_area; 100 void *fpt_save_area, *fpt_creg_save_area;
101 101
@@ -214,11 +214,10 @@ static int notrace s390_revalidate_registers(struct mci *mci)
214 : "0", "cc"); 214 : "0", "cc");
215#endif 215#endif
216 /* Revalidate clock comparator register */ 216 /* Revalidate clock comparator register */
217 asm volatile( 217 if (S390_lowcore.clock_comparator == -1)
218 " stck 0(%1)\n" 218 set_clock_comparator(S390_lowcore.mcck_clock);
219 " sckc 0(%1)" 219 else
220 : "=m" (tmpclock) : "a" (&(tmpclock)) : "cc", "memory"); 220 set_clock_comparator(S390_lowcore.clock_comparator);
221
222 /* Check if old PSW is valid */ 221 /* Check if old PSW is valid */
223 if (!mci->wp) 222 if (!mci->wp)
224 /* 223 /*
@@ -257,7 +256,7 @@ void notrace s390_do_machine_check(struct pt_regs *regs)
257 nmi_enter(); 256 nmi_enter();
258 s390_idle_check(regs, S390_lowcore.mcck_clock, 257 s390_idle_check(regs, S390_lowcore.mcck_clock,
259 S390_lowcore.mcck_enter_timer); 258 S390_lowcore.mcck_enter_timer);
260 259 kstat_cpu(smp_processor_id()).irqs[NMI_NMI]++;
261 mci = (struct mci *) &S390_lowcore.mcck_interruption_code; 260 mci = (struct mci *) &S390_lowcore.mcck_interruption_code;
262 mcck = &__get_cpu_var(cpu_mcck); 261 mcck = &__get_cpu_var(cpu_mcck);
263 umode = user_mode(regs); 262 umode = user_mode(regs);
diff --git a/arch/s390/kernel/process.c b/arch/s390/kernel/process.c
index d3a2d1c6438e..541a7509faeb 100644
--- a/arch/s390/kernel/process.c
+++ b/arch/s390/kernel/process.c
@@ -9,38 +9,27 @@
9 9
10#include <linux/compiler.h> 10#include <linux/compiler.h>
11#include <linux/cpu.h> 11#include <linux/cpu.h>
12#include <linux/errno.h>
13#include <linux/sched.h> 12#include <linux/sched.h>
14#include <linux/kernel.h> 13#include <linux/kernel.h>
15#include <linux/mm.h> 14#include <linux/mm.h>
16#include <linux/fs.h>
17#include <linux/smp.h> 15#include <linux/smp.h>
18#include <linux/stddef.h>
19#include <linux/slab.h> 16#include <linux/slab.h>
20#include <linux/unistd.h>
21#include <linux/ptrace.h>
22#include <linux/vmalloc.h>
23#include <linux/user.h>
24#include <linux/interrupt.h> 17#include <linux/interrupt.h>
25#include <linux/delay.h>
26#include <linux/reboot.h>
27#include <linux/init.h>
28#include <linux/module.h>
29#include <linux/notifier.h>
30#include <linux/tick.h> 18#include <linux/tick.h>
31#include <linux/elfcore.h> 19#include <linux/personality.h>
32#include <linux/kernel_stat.h>
33#include <linux/syscalls.h> 20#include <linux/syscalls.h>
34#include <linux/compat.h> 21#include <linux/compat.h>
35#include <asm/compat.h> 22#include <linux/kprobes.h>
36#include <asm/uaccess.h> 23#include <linux/random.h>
37#include <asm/pgtable.h> 24#include <linux/module.h>
38#include <asm/system.h> 25#include <asm/system.h>
39#include <asm/io.h> 26#include <asm/io.h>
40#include <asm/processor.h> 27#include <asm/processor.h>
41#include <asm/irq.h> 28#include <asm/irq.h>
42#include <asm/timer.h> 29#include <asm/timer.h>
43#include <asm/nmi.h> 30#include <asm/nmi.h>
31#include <asm/compat.h>
32#include <asm/smp.h>
44#include "entry.h" 33#include "entry.h"
45 34
46asmlinkage void ret_from_fork(void) asm ("ret_from_fork"); 35asmlinkage void ret_from_fork(void) asm ("ret_from_fork");
@@ -75,18 +64,13 @@ unsigned long thread_saved_pc(struct task_struct *tsk)
75 */ 64 */
76static void default_idle(void) 65static void default_idle(void)
77{ 66{
78 /* CPU is going idle. */ 67 if (cpu_is_offline(smp_processor_id()))
68 cpu_die();
79 local_irq_disable(); 69 local_irq_disable();
80 if (need_resched()) { 70 if (need_resched()) {
81 local_irq_enable(); 71 local_irq_enable();
82 return; 72 return;
83 } 73 }
84#ifdef CONFIG_HOTPLUG_CPU
85 if (cpu_is_offline(smp_processor_id())) {
86 preempt_enable_no_resched();
87 cpu_die();
88 }
89#endif
90 local_mcck_disable(); 74 local_mcck_disable();
91 if (test_thread_flag(TIF_MCCK_PENDING)) { 75 if (test_thread_flag(TIF_MCCK_PENDING)) {
92 local_mcck_enable(); 76 local_mcck_enable();
@@ -116,15 +100,17 @@ void cpu_idle(void)
116 } 100 }
117} 101}
118 102
119extern void kernel_thread_starter(void); 103extern void __kprobes kernel_thread_starter(void);
120 104
121asm( 105asm(
122 ".align 4\n" 106 ".section .kprobes.text, \"ax\"\n"
107 ".global kernel_thread_starter\n"
123 "kernel_thread_starter:\n" 108 "kernel_thread_starter:\n"
124 " la 2,0(10)\n" 109 " la 2,0(10)\n"
125 " basr 14,9\n" 110 " basr 14,9\n"
126 " la 2,0\n" 111 " la 2,0\n"
127 " br 11\n"); 112 " br 11\n"
113 ".previous\n");
128 114
129int kernel_thread(int (*fn)(void *), void * arg, unsigned long flags) 115int kernel_thread(int (*fn)(void *), void * arg, unsigned long flags)
130{ 116{
@@ -214,8 +200,10 @@ int copy_thread(unsigned long clone_flags, unsigned long new_stackp,
214 /* start new process with ar4 pointing to the correct address space */ 200 /* start new process with ar4 pointing to the correct address space */
215 p->thread.mm_segment = get_fs(); 201 p->thread.mm_segment = get_fs();
216 /* Don't copy debug registers */ 202 /* Don't copy debug registers */
217 memset(&p->thread.per_info, 0, sizeof(p->thread.per_info)); 203 memset(&p->thread.per_user, 0, sizeof(p->thread.per_user));
204 memset(&p->thread.per_event, 0, sizeof(p->thread.per_event));
218 clear_tsk_thread_flag(p, TIF_SINGLE_STEP); 205 clear_tsk_thread_flag(p, TIF_SINGLE_STEP);
206 clear_tsk_thread_flag(p, TIF_PER_TRAP);
219 /* Initialize per thread user and system timer values */ 207 /* Initialize per thread user and system timer values */
220 ti = task_thread_info(p); 208 ti = task_thread_info(p);
221 ti->user_timer = 0; 209 ti->user_timer = 0;
@@ -331,3 +319,39 @@ unsigned long get_wchan(struct task_struct *p)
331 } 319 }
332 return 0; 320 return 0;
333} 321}
322
323unsigned long arch_align_stack(unsigned long sp)
324{
325 if (!(current->personality & ADDR_NO_RANDOMIZE) && randomize_va_space)
326 sp -= get_random_int() & ~PAGE_MASK;
327 return sp & ~0xf;
328}
329
330static inline unsigned long brk_rnd(void)
331{
332 /* 8MB for 32bit, 1GB for 64bit */
333 if (is_32bit_task())
334 return (get_random_int() & 0x7ffUL) << PAGE_SHIFT;
335 else
336 return (get_random_int() & 0x3ffffUL) << PAGE_SHIFT;
337}
338
339unsigned long arch_randomize_brk(struct mm_struct *mm)
340{
341 unsigned long ret = PAGE_ALIGN(mm->brk + brk_rnd());
342
343 if (ret < mm->brk)
344 return mm->brk;
345 return ret;
346}
347
348unsigned long randomize_et_dyn(unsigned long base)
349{
350 unsigned long ret = PAGE_ALIGN(base + brk_rnd());
351
352 if (!(current->flags & PF_RANDOMIZE))
353 return base;
354 if (ret < base)
355 return base;
356 return ret;
357}
diff --git a/arch/s390/kernel/processor.c b/arch/s390/kernel/processor.c
index ecb2d02b02e4..311e9d712888 100644
--- a/arch/s390/kernel/processor.c
+++ b/arch/s390/kernel/processor.c
@@ -13,7 +13,7 @@
13#include <linux/smp.h> 13#include <linux/smp.h>
14#include <linux/seq_file.h> 14#include <linux/seq_file.h>
15#include <linux/delay.h> 15#include <linux/delay.h>
16 16#include <linux/cpu.h>
17#include <asm/elf.h> 17#include <asm/elf.h>
18#include <asm/lowcore.h> 18#include <asm/lowcore.h>
19#include <asm/param.h> 19#include <asm/param.h>
@@ -35,17 +35,6 @@ void __cpuinit cpu_init(void)
35} 35}
36 36
37/* 37/*
38 * print_cpu_info - print basic information about a cpu
39 */
40void __cpuinit print_cpu_info(void)
41{
42 struct cpuid *id = &per_cpu(cpu_id, smp_processor_id());
43
44 pr_info("Processor %d started, address %d, identification %06X\n",
45 S390_lowcore.cpu_nr, S390_lowcore.cpu_addr, id->ident);
46}
47
48/*
49 * show_cpuinfo - Get information on one CPU for use by procfs. 38 * show_cpuinfo - Get information on one CPU for use by procfs.
50 */ 39 */
51static int show_cpuinfo(struct seq_file *m, void *v) 40static int show_cpuinfo(struct seq_file *m, void *v)
@@ -57,9 +46,8 @@ static int show_cpuinfo(struct seq_file *m, void *v)
57 unsigned long n = (unsigned long) v - 1; 46 unsigned long n = (unsigned long) v - 1;
58 int i; 47 int i;
59 48
60 s390_adjust_jiffies();
61 preempt_disable();
62 if (!n) { 49 if (!n) {
50 s390_adjust_jiffies();
63 seq_printf(m, "vendor_id : IBM/S390\n" 51 seq_printf(m, "vendor_id : IBM/S390\n"
64 "# processors : %i\n" 52 "# processors : %i\n"
65 "bogomips per cpu: %lu.%02lu\n", 53 "bogomips per cpu: %lu.%02lu\n",
@@ -71,7 +59,7 @@ static int show_cpuinfo(struct seq_file *m, void *v)
71 seq_printf(m, "%s ", hwcap_str[i]); 59 seq_printf(m, "%s ", hwcap_str[i]);
72 seq_puts(m, "\n"); 60 seq_puts(m, "\n");
73 } 61 }
74 62 get_online_cpus();
75 if (cpu_online(n)) { 63 if (cpu_online(n)) {
76 struct cpuid *id = &per_cpu(cpu_id, n); 64 struct cpuid *id = &per_cpu(cpu_id, n);
77 seq_printf(m, "processor %li: " 65 seq_printf(m, "processor %li: "
@@ -80,7 +68,7 @@ static int show_cpuinfo(struct seq_file *m, void *v)
80 "machine = %04X\n", 68 "machine = %04X\n",
81 n, id->version, id->ident, id->machine); 69 n, id->version, id->ident, id->machine);
82 } 70 }
83 preempt_enable(); 71 put_online_cpus();
84 return 0; 72 return 0;
85} 73}
86 74
diff --git a/arch/s390/kernel/ptrace.c b/arch/s390/kernel/ptrace.c
index 83339d33c4b1..ef86ad243986 100644
--- a/arch/s390/kernel/ptrace.c
+++ b/arch/s390/kernel/ptrace.c
@@ -1,25 +1,9 @@
1/* 1/*
2 * arch/s390/kernel/ptrace.c 2 * Ptrace user space interface.
3 * 3 *
4 * S390 version 4 * Copyright IBM Corp. 1999,2010
5 * Copyright (C) 1999,2000 IBM Deutschland Entwicklung GmbH, IBM Corporation 5 * Author(s): Denis Joseph Barrow
6 * Author(s): Denis Joseph Barrow (djbarrow@de.ibm.com,barrow_dj@yahoo.com),
7 * Martin Schwidefsky (schwidefsky@de.ibm.com) 6 * Martin Schwidefsky (schwidefsky@de.ibm.com)
8 *
9 * Based on PowerPC version
10 * Copyright (C) 1995-1996 Gary Thomas (gdt@linuxppc.org)
11 *
12 * Derived from "arch/m68k/kernel/ptrace.c"
13 * Copyright (C) 1994 by Hamish Macdonald
14 * Taken from linux/kernel/ptrace.c and modified for M680x0.
15 * linux/kernel/ptrace.c is by Ross Biro 1/23/92, edited by Linus Torvalds
16 *
17 * Modified by Cort Dougan (cort@cs.nmt.edu)
18 *
19 *
20 * This file is subject to the terms and conditions of the GNU General
21 * Public License. See the file README.legal in the main directory of
22 * this archive for more details.
23 */ 7 */
24 8
25#include <linux/kernel.h> 9#include <linux/kernel.h>
@@ -61,76 +45,58 @@ enum s390_regset {
61 REGSET_GENERAL_EXTENDED, 45 REGSET_GENERAL_EXTENDED,
62}; 46};
63 47
64static void 48void update_per_regs(struct task_struct *task)
65FixPerRegisters(struct task_struct *task)
66{ 49{
67 struct pt_regs *regs; 50 static const struct per_regs per_single_step = {
68 per_struct *per_info; 51 .control = PER_EVENT_IFETCH,
69 per_cr_words cr_words; 52 .start = 0,
70 53 .end = PSW_ADDR_INSN,
71 regs = task_pt_regs(task); 54 };
72 per_info = (per_struct *) &task->thread.per_info; 55 struct pt_regs *regs = task_pt_regs(task);
73 per_info->control_regs.bits.em_instruction_fetch = 56 struct thread_struct *thread = &task->thread;
74 per_info->single_step | per_info->instruction_fetch; 57 const struct per_regs *new;
75 58 struct per_regs old;
76 if (per_info->single_step) { 59
77 per_info->control_regs.bits.starting_addr = 0; 60 /* TIF_SINGLE_STEP overrides the user specified PER registers. */
78#ifdef CONFIG_COMPAT 61 new = test_tsk_thread_flag(task, TIF_SINGLE_STEP) ?
79 if (is_compat_task()) 62 &per_single_step : &thread->per_user;
80 per_info->control_regs.bits.ending_addr = 0x7fffffffUL; 63
81 else 64 /* Take care of the PER enablement bit in the PSW. */
82#endif 65 if (!(new->control & PER_EVENT_MASK)) {
83 per_info->control_regs.bits.ending_addr = PSW_ADDR_INSN;
84 } else {
85 per_info->control_regs.bits.starting_addr =
86 per_info->starting_addr;
87 per_info->control_regs.bits.ending_addr =
88 per_info->ending_addr;
89 }
90 /*
91 * if any of the control reg tracing bits are on
92 * we switch on per in the psw
93 */
94 if (per_info->control_regs.words.cr[0] & PER_EM_MASK)
95 regs->psw.mask |= PSW_MASK_PER;
96 else
97 regs->psw.mask &= ~PSW_MASK_PER; 66 regs->psw.mask &= ~PSW_MASK_PER;
98 67 return;
99 if (per_info->control_regs.bits.em_storage_alteration)
100 per_info->control_regs.bits.storage_alt_space_ctl = 1;
101 else
102 per_info->control_regs.bits.storage_alt_space_ctl = 0;
103
104 if (task == current) {
105 __ctl_store(cr_words, 9, 11);
106 if (memcmp(&cr_words, &per_info->control_regs.words,
107 sizeof(cr_words)) != 0)
108 __ctl_load(per_info->control_regs.words, 9, 11);
109 } 68 }
69 regs->psw.mask |= PSW_MASK_PER;
70 __ctl_store(old, 9, 11);
71 if (memcmp(new, &old, sizeof(struct per_regs)) != 0)
72 __ctl_load(*new, 9, 11);
110} 73}
111 74
112void user_enable_single_step(struct task_struct *task) 75void user_enable_single_step(struct task_struct *task)
113{ 76{
114 task->thread.per_info.single_step = 1; 77 set_tsk_thread_flag(task, TIF_SINGLE_STEP);
115 FixPerRegisters(task); 78 if (task == current)
79 update_per_regs(task);
116} 80}
117 81
118void user_disable_single_step(struct task_struct *task) 82void user_disable_single_step(struct task_struct *task)
119{ 83{
120 task->thread.per_info.single_step = 0; 84 clear_tsk_thread_flag(task, TIF_SINGLE_STEP);
121 FixPerRegisters(task); 85 if (task == current)
86 update_per_regs(task);
122} 87}
123 88
124/* 89/*
125 * Called by kernel/ptrace.c when detaching.. 90 * Called by kernel/ptrace.c when detaching..
126 * 91 *
127 * Make sure single step bits etc are not set. 92 * Clear all debugging related fields.
128 */ 93 */
129void 94void ptrace_disable(struct task_struct *task)
130ptrace_disable(struct task_struct *child)
131{ 95{
132 /* make sure the single step bit is not set. */ 96 memset(&task->thread.per_user, 0, sizeof(task->thread.per_user));
133 user_disable_single_step(child); 97 memset(&task->thread.per_event, 0, sizeof(task->thread.per_event));
98 clear_tsk_thread_flag(task, TIF_SINGLE_STEP);
99 clear_tsk_thread_flag(task, TIF_PER_TRAP);
134} 100}
135 101
136#ifndef CONFIG_64BIT 102#ifndef CONFIG_64BIT
@@ -139,6 +105,47 @@ ptrace_disable(struct task_struct *child)
139# define __ADDR_MASK 7 105# define __ADDR_MASK 7
140#endif 106#endif
141 107
108static inline unsigned long __peek_user_per(struct task_struct *child,
109 addr_t addr)
110{
111 struct per_struct_kernel *dummy = NULL;
112
113 if (addr == (addr_t) &dummy->cr9)
114 /* Control bits of the active per set. */
115 return test_thread_flag(TIF_SINGLE_STEP) ?
116 PER_EVENT_IFETCH : child->thread.per_user.control;
117 else if (addr == (addr_t) &dummy->cr10)
118 /* Start address of the active per set. */
119 return test_thread_flag(TIF_SINGLE_STEP) ?
120 0 : child->thread.per_user.start;
121 else if (addr == (addr_t) &dummy->cr11)
122 /* End address of the active per set. */
123 return test_thread_flag(TIF_SINGLE_STEP) ?
124 PSW_ADDR_INSN : child->thread.per_user.end;
125 else if (addr == (addr_t) &dummy->bits)
126 /* Single-step bit. */
127 return test_thread_flag(TIF_SINGLE_STEP) ?
128 (1UL << (BITS_PER_LONG - 1)) : 0;
129 else if (addr == (addr_t) &dummy->starting_addr)
130 /* Start address of the user specified per set. */
131 return child->thread.per_user.start;
132 else if (addr == (addr_t) &dummy->ending_addr)
133 /* End address of the user specified per set. */
134 return child->thread.per_user.end;
135 else if (addr == (addr_t) &dummy->perc_atmid)
136 /* PER code, ATMID and AI of the last PER trap */
137 return (unsigned long)
138 child->thread.per_event.cause << (BITS_PER_LONG - 16);
139 else if (addr == (addr_t) &dummy->address)
140 /* Address of the last PER trap */
141 return child->thread.per_event.address;
142 else if (addr == (addr_t) &dummy->access_id)
143 /* Access id of the last PER trap */
144 return (unsigned long)
145 child->thread.per_event.paid << (BITS_PER_LONG - 8);
146 return 0;
147}
148
142/* 149/*
143 * Read the word at offset addr from the user area of a process. The 150 * Read the word at offset addr from the user area of a process. The
144 * trouble here is that the information is littered over different 151 * trouble here is that the information is littered over different
@@ -204,10 +211,10 @@ static unsigned long __peek_user(struct task_struct *child, addr_t addr)
204 211
205 } else if (addr < (addr_t) (&dummy->regs.per_info + 1)) { 212 } else if (addr < (addr_t) (&dummy->regs.per_info + 1)) {
206 /* 213 /*
207 * per_info is found in the thread structure 214 * Handle access to the per_info structure.
208 */ 215 */
209 offset = addr - (addr_t) &dummy->regs.per_info; 216 addr -= (addr_t) &dummy->regs.per_info;
210 tmp = *(addr_t *)((addr_t) &child->thread.per_info + offset); 217 tmp = __peek_user_per(child, addr);
211 218
212 } else 219 } else
213 tmp = 0; 220 tmp = 0;
@@ -237,6 +244,35 @@ peek_user(struct task_struct *child, addr_t addr, addr_t data)
237 return put_user(tmp, (addr_t __user *) data); 244 return put_user(tmp, (addr_t __user *) data);
238} 245}
239 246
247static inline void __poke_user_per(struct task_struct *child,
248 addr_t addr, addr_t data)
249{
250 struct per_struct_kernel *dummy = NULL;
251
252 /*
253 * There are only three fields in the per_info struct that the
254 * debugger user can write to.
255 * 1) cr9: the debugger wants to set a new PER event mask
256 * 2) starting_addr: the debugger wants to set a new starting
257 * address to use with the PER event mask.
258 * 3) ending_addr: the debugger wants to set a new ending
259 * address to use with the PER event mask.
260 * The user specified PER event mask and the start and end
261 * addresses are used only if single stepping is not in effect.
262 * Writes to any other field in per_info are ignored.
263 */
264 if (addr == (addr_t) &dummy->cr9)
265 /* PER event mask of the user specified per set. */
266 child->thread.per_user.control =
267 data & (PER_EVENT_MASK | PER_CONTROL_MASK);
268 else if (addr == (addr_t) &dummy->starting_addr)
269 /* Starting address of the user specified per set. */
270 child->thread.per_user.start = data;
271 else if (addr == (addr_t) &dummy->ending_addr)
272 /* Ending address of the user specified per set. */
273 child->thread.per_user.end = data;
274}
275
240/* 276/*
241 * Write a word to the user area of a process at location addr. This 277 * Write a word to the user area of a process at location addr. This
242 * operation does have an additional problem compared to peek_user. 278 * operation does have an additional problem compared to peek_user.
@@ -311,19 +347,17 @@ static int __poke_user(struct task_struct *child, addr_t addr, addr_t data)
311 347
312 } else if (addr < (addr_t) (&dummy->regs.per_info + 1)) { 348 } else if (addr < (addr_t) (&dummy->regs.per_info + 1)) {
313 /* 349 /*
314 * per_info is found in the thread structure 350 * Handle access to the per_info structure.
315 */ 351 */
316 offset = addr - (addr_t) &dummy->regs.per_info; 352 addr -= (addr_t) &dummy->regs.per_info;
317 *(addr_t *)((addr_t) &child->thread.per_info + offset) = data; 353 __poke_user_per(child, addr, data);
318 354
319 } 355 }
320 356
321 FixPerRegisters(child);
322 return 0; 357 return 0;
323} 358}
324 359
325static int 360static int poke_user(struct task_struct *child, addr_t addr, addr_t data)
326poke_user(struct task_struct *child, addr_t addr, addr_t data)
327{ 361{
328 addr_t mask; 362 addr_t mask;
329 363
@@ -343,7 +377,8 @@ poke_user(struct task_struct *child, addr_t addr, addr_t data)
343 return __poke_user(child, addr, data); 377 return __poke_user(child, addr, data);
344} 378}
345 379
346long arch_ptrace(struct task_struct *child, long request, long addr, long data) 380long arch_ptrace(struct task_struct *child, long request,
381 unsigned long addr, unsigned long data)
347{ 382{
348 ptrace_area parea; 383 ptrace_area parea;
349 int copied, ret; 384 int copied, ret;
@@ -409,12 +444,53 @@ long arch_ptrace(struct task_struct *child, long request, long addr, long data)
409 */ 444 */
410 445
411/* 446/*
447 * Same as peek_user_per but for a 31 bit program.
448 */
449static inline __u32 __peek_user_per_compat(struct task_struct *child,
450 addr_t addr)
451{
452 struct compat_per_struct_kernel *dummy32 = NULL;
453
454 if (addr == (addr_t) &dummy32->cr9)
455 /* Control bits of the active per set. */
456 return (__u32) test_thread_flag(TIF_SINGLE_STEP) ?
457 PER_EVENT_IFETCH : child->thread.per_user.control;
458 else if (addr == (addr_t) &dummy32->cr10)
459 /* Start address of the active per set. */
460 return (__u32) test_thread_flag(TIF_SINGLE_STEP) ?
461 0 : child->thread.per_user.start;
462 else if (addr == (addr_t) &dummy32->cr11)
463 /* End address of the active per set. */
464 return test_thread_flag(TIF_SINGLE_STEP) ?
465 PSW32_ADDR_INSN : child->thread.per_user.end;
466 else if (addr == (addr_t) &dummy32->bits)
467 /* Single-step bit. */
468 return (__u32) test_thread_flag(TIF_SINGLE_STEP) ?
469 0x80000000 : 0;
470 else if (addr == (addr_t) &dummy32->starting_addr)
471 /* Start address of the user specified per set. */
472 return (__u32) child->thread.per_user.start;
473 else if (addr == (addr_t) &dummy32->ending_addr)
474 /* End address of the user specified per set. */
475 return (__u32) child->thread.per_user.end;
476 else if (addr == (addr_t) &dummy32->perc_atmid)
477 /* PER code, ATMID and AI of the last PER trap */
478 return (__u32) child->thread.per_event.cause << 16;
479 else if (addr == (addr_t) &dummy32->address)
480 /* Address of the last PER trap */
481 return (__u32) child->thread.per_event.address;
482 else if (addr == (addr_t) &dummy32->access_id)
483 /* Access id of the last PER trap */
484 return (__u32) child->thread.per_event.paid << 24;
485 return 0;
486}
487
488/*
412 * Same as peek_user but for a 31 bit program. 489 * Same as peek_user but for a 31 bit program.
413 */ 490 */
414static u32 __peek_user_compat(struct task_struct *child, addr_t addr) 491static u32 __peek_user_compat(struct task_struct *child, addr_t addr)
415{ 492{
416 struct user32 *dummy32 = NULL; 493 struct compat_user *dummy32 = NULL;
417 per_struct32 *dummy_per32 = NULL;
418 addr_t offset; 494 addr_t offset;
419 __u32 tmp; 495 __u32 tmp;
420 496
@@ -464,19 +540,10 @@ static u32 __peek_user_compat(struct task_struct *child, addr_t addr)
464 540
465 } else if (addr < (addr_t) (&dummy32->regs.per_info + 1)) { 541 } else if (addr < (addr_t) (&dummy32->regs.per_info + 1)) {
466 /* 542 /*
467 * per_info is found in the thread structure 543 * Handle access to the per_info structure.
468 */ 544 */
469 offset = addr - (addr_t) &dummy32->regs.per_info; 545 addr -= (addr_t) &dummy32->regs.per_info;
470 /* This is magic. See per_struct and per_struct32. */ 546 tmp = __peek_user_per_compat(child, addr);
471 if ((offset >= (addr_t) &dummy_per32->control_regs &&
472 offset < (addr_t) (&dummy_per32->control_regs + 1)) ||
473 (offset >= (addr_t) &dummy_per32->starting_addr &&
474 offset <= (addr_t) &dummy_per32->ending_addr) ||
475 offset == (addr_t) &dummy_per32->lowcore.words.address)
476 offset = offset*2 + 4;
477 else
478 offset = offset*2;
479 tmp = *(__u32 *)((addr_t) &child->thread.per_info + offset);
480 547
481 } else 548 } else
482 tmp = 0; 549 tmp = 0;
@@ -497,13 +564,32 @@ static int peek_user_compat(struct task_struct *child,
497} 564}
498 565
499/* 566/*
567 * Same as poke_user_per but for a 31 bit program.
568 */
569static inline void __poke_user_per_compat(struct task_struct *child,
570 addr_t addr, __u32 data)
571{
572 struct compat_per_struct_kernel *dummy32 = NULL;
573
574 if (addr == (addr_t) &dummy32->cr9)
575 /* PER event mask of the user specified per set. */
576 child->thread.per_user.control =
577 data & (PER_EVENT_MASK | PER_CONTROL_MASK);
578 else if (addr == (addr_t) &dummy32->starting_addr)
579 /* Starting address of the user specified per set. */
580 child->thread.per_user.start = data;
581 else if (addr == (addr_t) &dummy32->ending_addr)
582 /* Ending address of the user specified per set. */
583 child->thread.per_user.end = data;
584}
585
586/*
500 * Same as poke_user but for a 31 bit program. 587 * Same as poke_user but for a 31 bit program.
501 */ 588 */
502static int __poke_user_compat(struct task_struct *child, 589static int __poke_user_compat(struct task_struct *child,
503 addr_t addr, addr_t data) 590 addr_t addr, addr_t data)
504{ 591{
505 struct user32 *dummy32 = NULL; 592 struct compat_user *dummy32 = NULL;
506 per_struct32 *dummy_per32 = NULL;
507 __u32 tmp = (__u32) data; 593 __u32 tmp = (__u32) data;
508 addr_t offset; 594 addr_t offset;
509 595
@@ -560,37 +646,20 @@ static int __poke_user_compat(struct task_struct *child,
560 646
561 } else if (addr < (addr_t) (&dummy32->regs.per_info + 1)) { 647 } else if (addr < (addr_t) (&dummy32->regs.per_info + 1)) {
562 /* 648 /*
563 * per_info is found in the thread structure. 649 * Handle access to the per_info structure.
564 */
565 offset = addr - (addr_t) &dummy32->regs.per_info;
566 /*
567 * This is magic. See per_struct and per_struct32.
568 * By incident the offsets in per_struct are exactly
569 * twice the offsets in per_struct32 for all fields.
570 * The 8 byte fields need special handling though,
571 * because the second half (bytes 4-7) is needed and
572 * not the first half.
573 */ 650 */
574 if ((offset >= (addr_t) &dummy_per32->control_regs && 651 addr -= (addr_t) &dummy32->regs.per_info;
575 offset < (addr_t) (&dummy_per32->control_regs + 1)) || 652 __poke_user_per_compat(child, addr, data);
576 (offset >= (addr_t) &dummy_per32->starting_addr &&
577 offset <= (addr_t) &dummy_per32->ending_addr) ||
578 offset == (addr_t) &dummy_per32->lowcore.words.address)
579 offset = offset*2 + 4;
580 else
581 offset = offset*2;
582 *(__u32 *)((addr_t) &child->thread.per_info + offset) = tmp;
583
584 } 653 }
585 654
586 FixPerRegisters(child);
587 return 0; 655 return 0;
588} 656}
589 657
590static int poke_user_compat(struct task_struct *child, 658static int poke_user_compat(struct task_struct *child,
591 addr_t addr, addr_t data) 659 addr_t addr, addr_t data)
592{ 660{
593 if (!is_compat_task() || (addr & 3) || addr > sizeof(struct user32) - 3) 661 if (!is_compat_task() || (addr & 3) ||
662 addr > sizeof(struct compat_user) - 3)
594 return -EIO; 663 return -EIO;
595 664
596 return __poke_user_compat(child, addr, data); 665 return __poke_user_compat(child, addr, data);
@@ -601,7 +670,7 @@ long compat_arch_ptrace(struct task_struct *child, compat_long_t request,
601{ 670{
602 unsigned long addr = caddr; 671 unsigned long addr = caddr;
603 unsigned long data = cdata; 672 unsigned long data = cdata;
604 ptrace_area_emu31 parea; 673 compat_ptrace_area parea;
605 int copied, ret; 674 int copied, ret;
606 675
607 switch (request) { 676 switch (request) {
diff --git a/arch/s390/kernel/reipl64.S b/arch/s390/kernel/reipl64.S
index 5e73dee63baa..9eabbc90795d 100644
--- a/arch/s390/kernel/reipl64.S
+++ b/arch/s390/kernel/reipl64.S
@@ -78,7 +78,7 @@ do_reipl_asm: basr %r13,0
78 * in the ESA psw. 78 * in the ESA psw.
79 * Bit 31 of the addresses has to be 0 for the 79 * Bit 31 of the addresses has to be 0 for the
80 * 31bit lpswe instruction a fact they appear to have 80 * 31bit lpswe instruction a fact they appear to have
81 * ommited from the pop. 81 * omitted from the pop.
82 */ 82 */
83.Lnewpsw: .quad 0x0000000080000000 83.Lnewpsw: .quad 0x0000000080000000
84 .quad .Lpg1 84 .quad .Lpg1
diff --git a/arch/s390/kernel/s390_ext.c b/arch/s390/kernel/s390_ext.c
deleted file mode 100644
index 9ce641b5291f..000000000000
--- a/arch/s390/kernel/s390_ext.c
+++ /dev/null
@@ -1,142 +0,0 @@
1/*
2 * arch/s390/kernel/s390_ext.c
3 *
4 * S390 version
5 * Copyright (C) 1999,2000 IBM Deutschland Entwicklung GmbH, IBM Corporation
6 * Author(s): Holger Smolinski (Holger.Smolinski@de.ibm.com),
7 * Martin Schwidefsky (schwidefsky@de.ibm.com)
8 */
9
10#include <linux/module.h>
11#include <linux/kernel.h>
12#include <linux/slab.h>
13#include <linux/ftrace.h>
14#include <linux/errno.h>
15#include <linux/kernel_stat.h>
16#include <linux/interrupt.h>
17#include <asm/cputime.h>
18#include <asm/lowcore.h>
19#include <asm/s390_ext.h>
20#include <asm/irq_regs.h>
21#include <asm/irq.h>
22#include "entry.h"
23
24/*
25 * ext_int_hash[index] is the start of the list for all external interrupts
26 * that hash to this index. With the current set of external interrupts
27 * (0x1202 external call, 0x1004 cpu timer, 0x2401 hwc console, 0x4000
28 * iucv and 0x2603 pfault) this is always the first element.
29 */
30ext_int_info_t *ext_int_hash[256] = { NULL, };
31
32static inline int ext_hash(__u16 code)
33{
34 return (code + (code >> 9)) & 0xff;
35}
36
37int register_external_interrupt(__u16 code, ext_int_handler_t handler)
38{
39 ext_int_info_t *p;
40 int index;
41
42 p = kmalloc(sizeof(ext_int_info_t), GFP_ATOMIC);
43 if (p == NULL)
44 return -ENOMEM;
45 p->code = code;
46 p->handler = handler;
47 index = ext_hash(code);
48 p->next = ext_int_hash[index];
49 ext_int_hash[index] = p;
50 return 0;
51}
52
53int register_early_external_interrupt(__u16 code, ext_int_handler_t handler,
54 ext_int_info_t *p)
55{
56 int index;
57
58 if (p == NULL)
59 return -EINVAL;
60 p->code = code;
61 p->handler = handler;
62 index = ext_hash(code);
63 p->next = ext_int_hash[index];
64 ext_int_hash[index] = p;
65 return 0;
66}
67
68int unregister_external_interrupt(__u16 code, ext_int_handler_t handler)
69{
70 ext_int_info_t *p, *q;
71 int index;
72
73 index = ext_hash(code);
74 q = NULL;
75 p = ext_int_hash[index];
76 while (p != NULL) {
77 if (p->code == code && p->handler == handler)
78 break;
79 q = p;
80 p = p->next;
81 }
82 if (p == NULL)
83 return -ENOENT;
84 if (q != NULL)
85 q->next = p->next;
86 else
87 ext_int_hash[index] = p->next;
88 kfree(p);
89 return 0;
90}
91
92int unregister_early_external_interrupt(__u16 code, ext_int_handler_t handler,
93 ext_int_info_t *p)
94{
95 ext_int_info_t *q;
96 int index;
97
98 if (p == NULL || p->code != code || p->handler != handler)
99 return -EINVAL;
100 index = ext_hash(code);
101 q = ext_int_hash[index];
102 if (p != q) {
103 while (q != NULL) {
104 if (q->next == p)
105 break;
106 q = q->next;
107 }
108 if (q == NULL)
109 return -ENOENT;
110 q->next = p->next;
111 } else
112 ext_int_hash[index] = p->next;
113 return 0;
114}
115
116void __irq_entry do_extint(struct pt_regs *regs, unsigned short code)
117{
118 ext_int_info_t *p;
119 int index;
120 struct pt_regs *old_regs;
121
122 old_regs = set_irq_regs(regs);
123 s390_idle_check(regs, S390_lowcore.int_clock,
124 S390_lowcore.async_enter_timer);
125 irq_enter();
126 if (S390_lowcore.int_clock >= S390_lowcore.clock_comparator)
127 /* Serve timer interrupts first. */
128 clock_comparator_work();
129 kstat_cpu(smp_processor_id()).irqs[EXTERNAL_INTERRUPT]++;
130 if (code != 0x1004)
131 __get_cpu_var(s390_idle).nohz_delay = 1;
132 index = ext_hash(code);
133 for (p = ext_int_hash[index]; p; p = p->next) {
134 if (likely(p->code == code))
135 p->handler(code);
136 }
137 irq_exit();
138 set_irq_regs(old_regs);
139}
140
141EXPORT_SYMBOL(register_external_interrupt);
142EXPORT_SYMBOL(unregister_external_interrupt);
diff --git a/arch/s390/kernel/setup.c b/arch/s390/kernel/setup.c
index c8e8e1354e1d..0c35dee10b00 100644
--- a/arch/s390/kernel/setup.c
+++ b/arch/s390/kernel/setup.c
@@ -102,16 +102,6 @@ EXPORT_SYMBOL(lowcore_ptr);
102 102
103#include <asm/setup.h> 103#include <asm/setup.h>
104 104
105static struct resource code_resource = {
106 .name = "Kernel code",
107 .flags = IORESOURCE_BUSY | IORESOURCE_MEM,
108};
109
110static struct resource data_resource = {
111 .name = "Kernel data",
112 .flags = IORESOURCE_BUSY | IORESOURCE_MEM,
113};
114
115/* 105/*
116 * condev= and conmode= setup parameter. 106 * condev= and conmode= setup parameter.
117 */ 107 */
@@ -315,8 +305,7 @@ static int set_amode_and_uaccess(unsigned long user_amode,
315 */ 305 */
316static int __init early_parse_switch_amode(char *p) 306static int __init early_parse_switch_amode(char *p)
317{ 307{
318 if (user_mode != SECONDARY_SPACE_MODE) 308 user_mode = PRIMARY_SPACE_MODE;
319 user_mode = PRIMARY_SPACE_MODE;
320 return 0; 309 return 0;
321} 310}
322early_param("switch_amode", early_parse_switch_amode); 311early_param("switch_amode", early_parse_switch_amode);
@@ -325,10 +314,6 @@ static int __init early_parse_user_mode(char *p)
325{ 314{
326 if (p && strcmp(p, "primary") == 0) 315 if (p && strcmp(p, "primary") == 0)
327 user_mode = PRIMARY_SPACE_MODE; 316 user_mode = PRIMARY_SPACE_MODE;
328#ifdef CONFIG_S390_EXEC_PROTECT
329 else if (p && strcmp(p, "secondary") == 0)
330 user_mode = SECONDARY_SPACE_MODE;
331#endif
332 else if (!p || strcmp(p, "home") == 0) 317 else if (!p || strcmp(p, "home") == 0)
333 user_mode = HOME_SPACE_MODE; 318 user_mode = HOME_SPACE_MODE;
334 else 319 else
@@ -337,31 +322,9 @@ static int __init early_parse_user_mode(char *p)
337} 322}
338early_param("user_mode", early_parse_user_mode); 323early_param("user_mode", early_parse_user_mode);
339 324
340#ifdef CONFIG_S390_EXEC_PROTECT
341/*
342 * Enable execute protection?
343 */
344static int __init early_parse_noexec(char *p)
345{
346 if (!strncmp(p, "off", 3))
347 return 0;
348 user_mode = SECONDARY_SPACE_MODE;
349 return 0;
350}
351early_param("noexec", early_parse_noexec);
352#endif /* CONFIG_S390_EXEC_PROTECT */
353
354static void setup_addressing_mode(void) 325static void setup_addressing_mode(void)
355{ 326{
356 if (user_mode == SECONDARY_SPACE_MODE) { 327 if (user_mode == PRIMARY_SPACE_MODE) {
357 if (set_amode_and_uaccess(PSW_ASC_SECONDARY,
358 PSW32_ASC_SECONDARY))
359 pr_info("Execute protection active, "
360 "mvcos available\n");
361 else
362 pr_info("Execute protection active, "
363 "mvcos not available\n");
364 } else if (user_mode == PRIMARY_SPACE_MODE) {
365 if (set_amode_and_uaccess(PSW_ASC_PRIMARY, PSW32_ASC_PRIMARY)) 328 if (set_amode_and_uaccess(PSW_ASC_PRIMARY, PSW32_ASC_PRIMARY))
366 pr_info("Address spaces switched, " 329 pr_info("Address spaces switched, "
367 "mvcos available\n"); 330 "mvcos available\n");
@@ -409,6 +372,9 @@ setup_lowcore(void)
409 lc->current_task = (unsigned long) init_thread_union.thread_info.task; 372 lc->current_task = (unsigned long) init_thread_union.thread_info.task;
410 lc->thread_info = (unsigned long) &init_thread_union; 373 lc->thread_info = (unsigned long) &init_thread_union;
411 lc->machine_flags = S390_lowcore.machine_flags; 374 lc->machine_flags = S390_lowcore.machine_flags;
375 lc->stfl_fac_list = S390_lowcore.stfl_fac_list;
376 memcpy(lc->stfle_fac_list, S390_lowcore.stfle_fac_list,
377 MAX_FACILITY_BIT/8);
412#ifndef CONFIG_64BIT 378#ifndef CONFIG_64BIT
413 if (MACHINE_HAS_IEEE) { 379 if (MACHINE_HAS_IEEE) {
414 lc->extended_save_area_addr = (__u32) 380 lc->extended_save_area_addr = (__u32)
@@ -433,21 +399,43 @@ setup_lowcore(void)
433 lowcore_ptr[0] = lc; 399 lowcore_ptr[0] = lc;
434} 400}
435 401
436static void __init 402static struct resource code_resource = {
437setup_resources(void) 403 .name = "Kernel code",
404 .flags = IORESOURCE_BUSY | IORESOURCE_MEM,
405};
406
407static struct resource data_resource = {
408 .name = "Kernel data",
409 .flags = IORESOURCE_BUSY | IORESOURCE_MEM,
410};
411
412static struct resource bss_resource = {
413 .name = "Kernel bss",
414 .flags = IORESOURCE_BUSY | IORESOURCE_MEM,
415};
416
417static struct resource __initdata *standard_resources[] = {
418 &code_resource,
419 &data_resource,
420 &bss_resource,
421};
422
423static void __init setup_resources(void)
438{ 424{
439 struct resource *res, *sub_res; 425 struct resource *res, *std_res, *sub_res;
440 int i; 426 int i, j;
441 427
442 code_resource.start = (unsigned long) &_text; 428 code_resource.start = (unsigned long) &_text;
443 code_resource.end = (unsigned long) &_etext - 1; 429 code_resource.end = (unsigned long) &_etext - 1;
444 data_resource.start = (unsigned long) &_etext; 430 data_resource.start = (unsigned long) &_etext;
445 data_resource.end = (unsigned long) &_edata - 1; 431 data_resource.end = (unsigned long) &_edata - 1;
432 bss_resource.start = (unsigned long) &__bss_start;
433 bss_resource.end = (unsigned long) &__bss_stop - 1;
446 434
447 for (i = 0; i < MEMORY_CHUNKS; i++) { 435 for (i = 0; i < MEMORY_CHUNKS; i++) {
448 if (!memory_chunk[i].size) 436 if (!memory_chunk[i].size)
449 continue; 437 continue;
450 res = alloc_bootmem_low(sizeof(struct resource)); 438 res = alloc_bootmem_low(sizeof(*res));
451 res->flags = IORESOURCE_BUSY | IORESOURCE_MEM; 439 res->flags = IORESOURCE_BUSY | IORESOURCE_MEM;
452 switch (memory_chunk[i].type) { 440 switch (memory_chunk[i].type) {
453 case CHUNK_READ_WRITE: 441 case CHUNK_READ_WRITE:
@@ -461,40 +449,24 @@ setup_resources(void)
461 res->name = "reserved"; 449 res->name = "reserved";
462 } 450 }
463 res->start = memory_chunk[i].addr; 451 res->start = memory_chunk[i].addr;
464 res->end = memory_chunk[i].addr + memory_chunk[i].size - 1; 452 res->end = res->start + memory_chunk[i].size - 1;
465 request_resource(&iomem_resource, res); 453 request_resource(&iomem_resource, res);
466 454
467 if (code_resource.start >= res->start && 455 for (j = 0; j < ARRAY_SIZE(standard_resources); j++) {
468 code_resource.start <= res->end && 456 std_res = standard_resources[j];
469 code_resource.end > res->end) { 457 if (std_res->start < res->start ||
470 sub_res = alloc_bootmem_low(sizeof(struct resource)); 458 std_res->start > res->end)
471 memcpy(sub_res, &code_resource, 459 continue;
472 sizeof(struct resource)); 460 if (std_res->end > res->end) {
473 sub_res->end = res->end; 461 sub_res = alloc_bootmem_low(sizeof(*sub_res));
474 code_resource.start = res->end + 1; 462 *sub_res = *std_res;
475 request_resource(res, sub_res); 463 sub_res->end = res->end;
476 } 464 std_res->start = res->end + 1;
477 465 request_resource(res, sub_res);
478 if (code_resource.start >= res->start && 466 } else {
479 code_resource.start <= res->end && 467 request_resource(res, std_res);
480 code_resource.end <= res->end) 468 }
481 request_resource(res, &code_resource);
482
483 if (data_resource.start >= res->start &&
484 data_resource.start <= res->end &&
485 data_resource.end > res->end) {
486 sub_res = alloc_bootmem_low(sizeof(struct resource));
487 memcpy(sub_res, &data_resource,
488 sizeof(struct resource));
489 sub_res->end = res->end;
490 data_resource.start = res->end + 1;
491 request_resource(res, sub_res);
492 } 469 }
493
494 if (data_resource.start >= res->start &&
495 data_resource.start <= res->end &&
496 data_resource.end <= res->end)
497 request_resource(res, &data_resource);
498 } 470 }
499} 471}
500 472
@@ -627,7 +599,8 @@ setup_memory(void)
627 add_active_range(0, start_chunk, end_chunk); 599 add_active_range(0, start_chunk, end_chunk);
628 pfn = max(start_chunk, start_pfn); 600 pfn = max(start_chunk, start_pfn);
629 for (; pfn < end_chunk; pfn++) 601 for (; pfn < end_chunk; pfn++)
630 page_set_storage_key(PFN_PHYS(pfn), PAGE_DEFAULT_KEY); 602 page_set_storage_key(PFN_PHYS(pfn),
603 PAGE_DEFAULT_KEY, 0);
631 } 604 }
632 605
633 psw_set_key(PAGE_DEFAULT_KEY); 606 psw_set_key(PAGE_DEFAULT_KEY);
@@ -674,12 +647,9 @@ setup_memory(void)
674static void __init setup_hwcaps(void) 647static void __init setup_hwcaps(void)
675{ 648{
676 static const int stfl_bits[6] = { 0, 2, 7, 17, 19, 21 }; 649 static const int stfl_bits[6] = { 0, 2, 7, 17, 19, 21 };
677 unsigned long long facility_list_extended;
678 unsigned int facility_list;
679 struct cpuid cpu_id; 650 struct cpuid cpu_id;
680 int i; 651 int i;
681 652
682 facility_list = stfl();
683 /* 653 /*
684 * The store facility list bits numbers as found in the principles 654 * The store facility list bits numbers as found in the principles
685 * of operation are numbered with bit 1UL<<31 as number 0 to 655 * of operation are numbered with bit 1UL<<31 as number 0 to
@@ -699,11 +669,10 @@ static void __init setup_hwcaps(void)
699 * HWCAP_S390_ETF3EH bit 8 (22 && 30). 669 * HWCAP_S390_ETF3EH bit 8 (22 && 30).
700 */ 670 */
701 for (i = 0; i < 6; i++) 671 for (i = 0; i < 6; i++)
702 if (facility_list & (1UL << (31 - stfl_bits[i]))) 672 if (test_facility(stfl_bits[i]))
703 elf_hwcap |= 1UL << i; 673 elf_hwcap |= 1UL << i;
704 674
705 if ((facility_list & (1UL << (31 - 22))) 675 if (test_facility(22) && test_facility(30))
706 && (facility_list & (1UL << (31 - 30))))
707 elf_hwcap |= HWCAP_S390_ETF3EH; 676 elf_hwcap |= HWCAP_S390_ETF3EH;
708 677
709 /* 678 /*
@@ -712,19 +681,15 @@ static void __init setup_hwcaps(void)
712 * and 1ULL<<0 as bit 63. Bits 0-31 contain the same information 681 * and 1ULL<<0 as bit 63. Bits 0-31 contain the same information
713 * as stored by stfl, bits 32-xxx contain additional facilities. 682 * as stored by stfl, bits 32-xxx contain additional facilities.
714 * How many facility words are stored depends on the number of 683 * How many facility words are stored depends on the number of
715 * doublewords passed to the instruction. The additional facilites 684 * doublewords passed to the instruction. The additional facilities
716 * are: 685 * are:
717 * Bit 42: decimal floating point facility is installed 686 * Bit 42: decimal floating point facility is installed
718 * Bit 44: perform floating point operation facility is installed 687 * Bit 44: perform floating point operation facility is installed
719 * translated to: 688 * translated to:
720 * HWCAP_S390_DFP bit 6 (42 && 44). 689 * HWCAP_S390_DFP bit 6 (42 && 44).
721 */ 690 */
722 if ((elf_hwcap & (1UL << 2)) && 691 if ((elf_hwcap & (1UL << 2)) && test_facility(42) && test_facility(44))
723 __stfle(&facility_list_extended, 1) > 0) { 692 elf_hwcap |= HWCAP_S390_DFP;
724 if ((facility_list_extended & (1ULL << (63 - 42)))
725 && (facility_list_extended & (1ULL << (63 - 44))))
726 elf_hwcap |= HWCAP_S390_DFP;
727 }
728 693
729 /* 694 /*
730 * Huge page support HWCAP_S390_HPAGE is bit 7. 695 * Huge page support HWCAP_S390_HPAGE is bit 7.
@@ -765,6 +730,9 @@ static void __init setup_hwcaps(void)
765 case 0x2098: 730 case 0x2098:
766 strcpy(elf_platform, "z10"); 731 strcpy(elf_platform, "z10");
767 break; 732 break;
733 case 0x2817:
734 strcpy(elf_platform, "z196");
735 break;
768 } 736 }
769} 737}
770 738
diff --git a/arch/s390/kernel/signal.c b/arch/s390/kernel/signal.c
index ee7ac8b11782..abbb3c3c7aab 100644
--- a/arch/s390/kernel/signal.c
+++ b/arch/s390/kernel/signal.c
@@ -505,7 +505,7 @@ void do_signal(struct pt_regs *regs)
505 * Let tracing know that we've done the handler setup. 505 * Let tracing know that we've done the handler setup.
506 */ 506 */
507 tracehook_signal_handler(signr, &info, &ka, regs, 507 tracehook_signal_handler(signr, &info, &ka, regs,
508 current->thread.per_info.single_step); 508 test_thread_flag(TIF_SINGLE_STEP));
509 } 509 }
510 return; 510 return;
511 } 511 }
diff --git a/arch/s390/kernel/smp.c b/arch/s390/kernel/smp.c
index 8127ebd59c4d..1d55c95f617c 100644
--- a/arch/s390/kernel/smp.c
+++ b/arch/s390/kernel/smp.c
@@ -23,6 +23,7 @@
23#define KMSG_COMPONENT "cpu" 23#define KMSG_COMPONENT "cpu"
24#define pr_fmt(fmt) KMSG_COMPONENT ": " fmt 24#define pr_fmt(fmt) KMSG_COMPONENT ": " fmt
25 25
26#include <linux/workqueue.h>
26#include <linux/module.h> 27#include <linux/module.h>
27#include <linux/init.h> 28#include <linux/init.h>
28#include <linux/mm.h> 29#include <linux/mm.h>
@@ -43,7 +44,6 @@
43#include <asm/sigp.h> 44#include <asm/sigp.h>
44#include <asm/pgalloc.h> 45#include <asm/pgalloc.h>
45#include <asm/irq.h> 46#include <asm/irq.h>
46#include <asm/s390_ext.h>
47#include <asm/cpcmd.h> 47#include <asm/cpcmd.h>
48#include <asm/tlbflush.h> 48#include <asm/tlbflush.h>
49#include <asm/timer.h> 49#include <asm/timer.h>
@@ -156,18 +156,20 @@ void smp_send_stop(void)
156 * cpus are handled. 156 * cpus are handled.
157 */ 157 */
158 158
159static void do_ext_call_interrupt(__u16 code) 159static void do_ext_call_interrupt(unsigned int ext_int_code,
160 unsigned int param32, unsigned long param64)
160{ 161{
161 unsigned long bits; 162 unsigned long bits;
162 163
164 kstat_cpu(smp_processor_id()).irqs[EXTINT_IPI]++;
163 /* 165 /*
164 * handle bit signal external calls 166 * handle bit signal external calls
165 *
166 * For the ec_schedule signal we have to do nothing. All the work
167 * is done automatically when we return from the interrupt.
168 */ 167 */
169 bits = xchg(&S390_lowcore.ext_call_fast, 0); 168 bits = xchg(&S390_lowcore.ext_call_fast, 0);
170 169
170 if (test_bit(ec_schedule, &bits))
171 scheduler_ipi();
172
171 if (test_bit(ec_call_function, &bits)) 173 if (test_bit(ec_call_function, &bits))
172 generic_smp_call_function_interrupt(); 174 generic_smp_call_function_interrupt();
173 175
@@ -260,7 +262,7 @@ void smp_ctl_set_bit(int cr, int bit)
260 262
261 memset(&parms.orvals, 0, sizeof(parms.orvals)); 263 memset(&parms.orvals, 0, sizeof(parms.orvals));
262 memset(&parms.andvals, 0xff, sizeof(parms.andvals)); 264 memset(&parms.andvals, 0xff, sizeof(parms.andvals));
263 parms.orvals[cr] = 1 << bit; 265 parms.orvals[cr] = 1UL << bit;
264 on_each_cpu(smp_ctl_bit_callback, &parms, 1); 266 on_each_cpu(smp_ctl_bit_callback, &parms, 1);
265} 267}
266EXPORT_SYMBOL(smp_ctl_set_bit); 268EXPORT_SYMBOL(smp_ctl_set_bit);
@@ -274,7 +276,7 @@ void smp_ctl_clear_bit(int cr, int bit)
274 276
275 memset(&parms.orvals, 0, sizeof(parms.orvals)); 277 memset(&parms.orvals, 0, sizeof(parms.orvals));
276 memset(&parms.andvals, 0xff, sizeof(parms.andvals)); 278 memset(&parms.andvals, 0xff, sizeof(parms.andvals));
277 parms.andvals[cr] = ~(1L << bit); 279 parms.andvals[cr] = ~(1UL << bit);
278 on_each_cpu(smp_ctl_bit_callback, &parms, 1); 280 on_each_cpu(smp_ctl_bit_callback, &parms, 1);
279} 281}
280EXPORT_SYMBOL(smp_ctl_clear_bit); 282EXPORT_SYMBOL(smp_ctl_clear_bit);
@@ -332,7 +334,7 @@ static int smp_rescan_cpus_sigp(cpumask_t avail)
332 smp_cpu_polarization[logical_cpu] = POLARIZATION_UNKNWN; 334 smp_cpu_polarization[logical_cpu] = POLARIZATION_UNKNWN;
333 if (!cpu_stopped(logical_cpu)) 335 if (!cpu_stopped(logical_cpu))
334 continue; 336 continue;
335 cpu_set(logical_cpu, cpu_present_map); 337 set_cpu_present(logical_cpu, true);
336 smp_cpu_state[logical_cpu] = CPU_STATE_CONFIGURED; 338 smp_cpu_state[logical_cpu] = CPU_STATE_CONFIGURED;
337 logical_cpu = cpumask_next(logical_cpu, &avail); 339 logical_cpu = cpumask_next(logical_cpu, &avail);
338 if (logical_cpu >= nr_cpu_ids) 340 if (logical_cpu >= nr_cpu_ids)
@@ -364,7 +366,7 @@ static int smp_rescan_cpus_sclp(cpumask_t avail)
364 continue; 366 continue;
365 __cpu_logical_map[logical_cpu] = cpu_id; 367 __cpu_logical_map[logical_cpu] = cpu_id;
366 smp_cpu_polarization[logical_cpu] = POLARIZATION_UNKNWN; 368 smp_cpu_polarization[logical_cpu] = POLARIZATION_UNKNWN;
367 cpu_set(logical_cpu, cpu_present_map); 369 set_cpu_present(logical_cpu, true);
368 if (cpu >= info->configured) 370 if (cpu >= info->configured)
369 smp_cpu_state[logical_cpu] = CPU_STATE_STANDBY; 371 smp_cpu_state[logical_cpu] = CPU_STATE_STANDBY;
370 else 372 else
@@ -382,7 +384,7 @@ static int __smp_rescan_cpus(void)
382{ 384{
383 cpumask_t avail; 385 cpumask_t avail;
384 386
385 cpus_xor(avail, cpu_possible_map, cpu_present_map); 387 cpumask_xor(&avail, cpu_possible_mask, cpu_present_mask);
386 if (smp_use_sigp_detection) 388 if (smp_use_sigp_detection)
387 return smp_rescan_cpus_sigp(avail); 389 return smp_rescan_cpus_sigp(avail);
388 else 390 else
@@ -464,29 +466,29 @@ int __cpuinit start_secondary(void *cpuvoid)
464 notify_cpu_starting(smp_processor_id()); 466 notify_cpu_starting(smp_processor_id());
465 /* Mark this cpu as online */ 467 /* Mark this cpu as online */
466 ipi_call_lock(); 468 ipi_call_lock();
467 cpu_set(smp_processor_id(), cpu_online_map); 469 set_cpu_online(smp_processor_id(), true);
468 ipi_call_unlock(); 470 ipi_call_unlock();
469 /* Switch on interrupts */ 471 /* Switch on interrupts */
470 local_irq_enable(); 472 local_irq_enable();
471 /* Print info about this processor */
472 print_cpu_info();
473 /* cpu_idle will call schedule for us */ 473 /* cpu_idle will call schedule for us */
474 cpu_idle(); 474 cpu_idle();
475 return 0; 475 return 0;
476} 476}
477 477
478static void __init smp_create_idle(unsigned int cpu) 478struct create_idle {
479 struct work_struct work;
480 struct task_struct *idle;
481 struct completion done;
482 int cpu;
483};
484
485static void __cpuinit smp_fork_idle(struct work_struct *work)
479{ 486{
480 struct task_struct *p; 487 struct create_idle *c_idle;
481 488
482 /* 489 c_idle = container_of(work, struct create_idle, work);
483 * don't care about the psw and regs settings since we'll never 490 c_idle->idle = fork_idle(c_idle->cpu);
484 * reschedule the forked task. 491 complete(&c_idle->done);
485 */
486 p = fork_idle(cpu);
487 if (IS_ERR(p))
488 panic("failed fork for CPU %u: %li", cpu, PTR_ERR(p));
489 current_set[cpu] = p;
490} 492}
491 493
492static int __cpuinit smp_alloc_lowcore(int cpu) 494static int __cpuinit smp_alloc_lowcore(int cpu)
@@ -550,6 +552,7 @@ static void smp_free_lowcore(int cpu)
550int __cpuinit __cpu_up(unsigned int cpu) 552int __cpuinit __cpu_up(unsigned int cpu)
551{ 553{
552 struct _lowcore *cpu_lowcore; 554 struct _lowcore *cpu_lowcore;
555 struct create_idle c_idle;
553 struct task_struct *idle; 556 struct task_struct *idle;
554 struct stack_frame *sf; 557 struct stack_frame *sf;
555 u32 lowcore; 558 u32 lowcore;
@@ -557,6 +560,19 @@ int __cpuinit __cpu_up(unsigned int cpu)
557 560
558 if (smp_cpu_state[cpu] != CPU_STATE_CONFIGURED) 561 if (smp_cpu_state[cpu] != CPU_STATE_CONFIGURED)
559 return -EIO; 562 return -EIO;
563 idle = current_set[cpu];
564 if (!idle) {
565 c_idle.done = COMPLETION_INITIALIZER_ONSTACK(c_idle.done);
566 INIT_WORK_ONSTACK(&c_idle.work, smp_fork_idle);
567 c_idle.cpu = cpu;
568 schedule_work(&c_idle.work);
569 wait_for_completion(&c_idle.done);
570 if (IS_ERR(c_idle.idle))
571 return PTR_ERR(c_idle.idle);
572 idle = c_idle.idle;
573 current_set[cpu] = c_idle.idle;
574 }
575 init_idle(idle, cpu);
560 if (smp_alloc_lowcore(cpu)) 576 if (smp_alloc_lowcore(cpu))
561 return -ENOMEM; 577 return -ENOMEM;
562 do { 578 do {
@@ -571,7 +587,6 @@ int __cpuinit __cpu_up(unsigned int cpu)
571 while (sigp_p(lowcore, cpu, sigp_set_prefix) == sigp_busy) 587 while (sigp_p(lowcore, cpu, sigp_set_prefix) == sigp_busy)
572 udelay(10); 588 udelay(10);
573 589
574 idle = current_set[cpu];
575 cpu_lowcore = lowcore_ptr[cpu]; 590 cpu_lowcore = lowcore_ptr[cpu];
576 cpu_lowcore->kernel_stack = (unsigned long) 591 cpu_lowcore->kernel_stack = (unsigned long)
577 task_stack_page(idle) + THREAD_SIZE; 592 task_stack_page(idle) + THREAD_SIZE;
@@ -593,6 +608,8 @@ int __cpuinit __cpu_up(unsigned int cpu)
593 cpu_lowcore->kernel_asce = S390_lowcore.kernel_asce; 608 cpu_lowcore->kernel_asce = S390_lowcore.kernel_asce;
594 cpu_lowcore->machine_flags = S390_lowcore.machine_flags; 609 cpu_lowcore->machine_flags = S390_lowcore.machine_flags;
595 cpu_lowcore->ftrace_func = S390_lowcore.ftrace_func; 610 cpu_lowcore->ftrace_func = S390_lowcore.ftrace_func;
611 memcpy(cpu_lowcore->stfle_fac_list, S390_lowcore.stfle_fac_list,
612 MAX_FACILITY_BIT/8);
596 eieio(); 613 eieio();
597 614
598 while (sigp(cpu, sigp_restart) == sigp_busy) 615 while (sigp(cpu, sigp_restart) == sigp_busy)
@@ -626,7 +643,7 @@ int __cpu_disable(void)
626 struct ec_creg_mask_parms cr_parms; 643 struct ec_creg_mask_parms cr_parms;
627 int cpu = smp_processor_id(); 644 int cpu = smp_processor_id();
628 645
629 cpu_clear(cpu, cpu_online_map); 646 set_cpu_online(cpu, false);
630 647
631 /* Disable pfault pseudo page faults on this cpu. */ 648 /* Disable pfault pseudo page faults on this cpu. */
632 pfault_fini(); 649 pfault_fini();
@@ -636,8 +653,8 @@ int __cpu_disable(void)
636 653
637 /* disable all external interrupts */ 654 /* disable all external interrupts */
638 cr_parms.orvals[0] = 0; 655 cr_parms.orvals[0] = 0;
639 cr_parms.andvals[0] = ~(1 << 15 | 1 << 14 | 1 << 13 | 1 << 12 | 656 cr_parms.andvals[0] = ~(1 << 15 | 1 << 14 | 1 << 13 | 1 << 11 |
640 1 << 11 | 1 << 10 | 1 << 6 | 1 << 4); 657 1 << 10 | 1 << 9 | 1 << 6 | 1 << 4);
641 /* disable all I/O interrupts */ 658 /* disable all I/O interrupts */
642 cr_parms.orvals[6] = 0; 659 cr_parms.orvals[6] = 0;
643 cr_parms.andvals[6] = ~(1 << 31 | 1 << 30 | 1 << 29 | 1 << 28 | 660 cr_parms.andvals[6] = ~(1 << 31 | 1 << 30 | 1 << 29 | 1 << 28 |
@@ -661,10 +678,9 @@ void __cpu_die(unsigned int cpu)
661 udelay(10); 678 udelay(10);
662 smp_free_lowcore(cpu); 679 smp_free_lowcore(cpu);
663 atomic_dec(&init_mm.context.attach_count); 680 atomic_dec(&init_mm.context.attach_count);
664 pr_info("Processor %d stopped\n", cpu);
665} 681}
666 682
667void cpu_die(void) 683void __noreturn cpu_die(void)
668{ 684{
669 idle_task_exit(); 685 idle_task_exit();
670 while (sigp(smp_processor_id(), sigp_stop) == sigp_busy) 686 while (sigp(smp_processor_id(), sigp_stop) == sigp_busy)
@@ -681,14 +697,12 @@ void __init smp_prepare_cpus(unsigned int max_cpus)
681#endif 697#endif
682 unsigned long async_stack, panic_stack; 698 unsigned long async_stack, panic_stack;
683 struct _lowcore *lowcore; 699 struct _lowcore *lowcore;
684 unsigned int cpu;
685 700
686 smp_detect_cpus(); 701 smp_detect_cpus();
687 702
688 /* request the 0x1201 emergency signal external interrupt */ 703 /* request the 0x1201 emergency signal external interrupt */
689 if (register_external_interrupt(0x1201, do_ext_call_interrupt) != 0) 704 if (register_external_interrupt(0x1201, do_ext_call_interrupt) != 0)
690 panic("Couldn't request external interrupt 0x1201"); 705 panic("Couldn't request external interrupt 0x1201");
691 print_cpu_info();
692 706
693 /* Reallocate current lowcore, but keep its contents. */ 707 /* Reallocate current lowcore, but keep its contents. */
694 lowcore = (void *) __get_free_pages(GFP_KERNEL | GFP_DMA, LC_ORDER); 708 lowcore = (void *) __get_free_pages(GFP_KERNEL | GFP_DMA, LC_ORDER);
@@ -716,9 +730,6 @@ void __init smp_prepare_cpus(unsigned int max_cpus)
716 if (vdso_alloc_per_cpu(smp_processor_id(), &S390_lowcore)) 730 if (vdso_alloc_per_cpu(smp_processor_id(), &S390_lowcore))
717 BUG(); 731 BUG();
718#endif 732#endif
719 for_each_possible_cpu(cpu)
720 if (cpu != smp_processor_id())
721 smp_create_idle(cpu);
722} 733}
723 734
724void __init smp_prepare_boot_cpu(void) 735void __init smp_prepare_boot_cpu(void)
@@ -726,8 +737,8 @@ void __init smp_prepare_boot_cpu(void)
726 BUG_ON(smp_processor_id() != 0); 737 BUG_ON(smp_processor_id() != 0);
727 738
728 current_thread_info()->cpu = 0; 739 current_thread_info()->cpu = 0;
729 cpu_set(0, cpu_present_map); 740 set_cpu_present(0, true);
730 cpu_set(0, cpu_online_map); 741 set_cpu_online(0, true);
731 S390_lowcore.percpu_offset = __per_cpu_offset[0]; 742 S390_lowcore.percpu_offset = __per_cpu_offset[0];
732 current_set[0] = current; 743 current_set[0] = current;
733 smp_cpu_state[0] = CPU_STATE_CONFIGURED; 744 smp_cpu_state[0] = CPU_STATE_CONFIGURED;
@@ -1004,21 +1015,21 @@ int __ref smp_rescan_cpus(void)
1004 1015
1005 get_online_cpus(); 1016 get_online_cpus();
1006 mutex_lock(&smp_cpu_state_mutex); 1017 mutex_lock(&smp_cpu_state_mutex);
1007 newcpus = cpu_present_map; 1018 cpumask_copy(&newcpus, cpu_present_mask);
1008 rc = __smp_rescan_cpus(); 1019 rc = __smp_rescan_cpus();
1009 if (rc) 1020 if (rc)
1010 goto out; 1021 goto out;
1011 cpus_andnot(newcpus, cpu_present_map, newcpus); 1022 cpumask_andnot(&newcpus, cpu_present_mask, &newcpus);
1012 for_each_cpu_mask(cpu, newcpus) { 1023 for_each_cpu(cpu, &newcpus) {
1013 rc = smp_add_present_cpu(cpu); 1024 rc = smp_add_present_cpu(cpu);
1014 if (rc) 1025 if (rc)
1015 cpu_clear(cpu, cpu_present_map); 1026 set_cpu_present(cpu, false);
1016 } 1027 }
1017 rc = 0; 1028 rc = 0;
1018out: 1029out:
1019 mutex_unlock(&smp_cpu_state_mutex); 1030 mutex_unlock(&smp_cpu_state_mutex);
1020 put_online_cpus(); 1031 put_online_cpus();
1021 if (!cpus_empty(newcpus)) 1032 if (!cpumask_empty(&newcpus))
1022 topology_schedule_update(); 1033 topology_schedule_update();
1023 return rc; 1034 return rc;
1024} 1035}
diff --git a/arch/s390/kernel/switch_cpu.S b/arch/s390/kernel/switch_cpu.S
index 469f11b574fa..20530dd2eab1 100644
--- a/arch/s390/kernel/switch_cpu.S
+++ b/arch/s390/kernel/switch_cpu.S
@@ -46,7 +46,9 @@ smp_restart_cpu:
46 ltr %r4,%r4 /* New stack ? */ 46 ltr %r4,%r4 /* New stack ? */
47 jz 1f 47 jz 1f
48 lr %r15,%r4 48 lr %r15,%r4
491: basr %r14,%r2 491: lr %r14,%r2 /* r14: Function to call */
50 lr %r2,%r3 /* r2 : Parameter for function*/
51 basr %r14,%r14 /* Call function */
50 52
51.gprregs_addr: 53.gprregs_addr:
52 .long .gprregs 54 .long .gprregs
diff --git a/arch/s390/kernel/switch_cpu64.S b/arch/s390/kernel/switch_cpu64.S
index d94aacc898cb..5be3f43898f9 100644
--- a/arch/s390/kernel/switch_cpu64.S
+++ b/arch/s390/kernel/switch_cpu64.S
@@ -42,7 +42,9 @@ smp_restart_cpu:
42 ltgr %r4,%r4 /* New stack ? */ 42 ltgr %r4,%r4 /* New stack ? */
43 jz 1f 43 jz 1f
44 lgr %r15,%r4 44 lgr %r15,%r4
451: basr %r14,%r2 451: lgr %r14,%r2 /* r14: Function to call */
46 lgr %r2,%r3 /* r2 : Parameter for function*/
47 basr %r14,%r14 /* Call function */
46 48
47 .section .data,"aw",@progbits 49 .section .data,"aw",@progbits
48.gprregs: 50.gprregs:
diff --git a/arch/s390/kernel/syscalls.S b/arch/s390/kernel/syscalls.S
index a8fee1b14395..6ee39ef8fe4a 100644
--- a/arch/s390/kernel/syscalls.S
+++ b/arch/s390/kernel/syscalls.S
@@ -343,3 +343,8 @@ SYSCALL(sys_perf_event_open,sys_perf_event_open,sys_perf_event_open_wrapper)
343SYSCALL(sys_fanotify_init,sys_fanotify_init,sys_fanotify_init_wrapper) 343SYSCALL(sys_fanotify_init,sys_fanotify_init,sys_fanotify_init_wrapper)
344SYSCALL(sys_fanotify_mark,sys_fanotify_mark,sys_fanotify_mark_wrapper) 344SYSCALL(sys_fanotify_mark,sys_fanotify_mark,sys_fanotify_mark_wrapper)
345SYSCALL(sys_prlimit64,sys_prlimit64,sys_prlimit64_wrapper) 345SYSCALL(sys_prlimit64,sys_prlimit64,sys_prlimit64_wrapper)
346SYSCALL(sys_name_to_handle_at,sys_name_to_handle_at,sys_name_to_handle_at_wrapper) /* 335 */
347SYSCALL(sys_open_by_handle_at,sys_open_by_handle_at,compat_sys_open_by_handle_at_wrapper)
348SYSCALL(sys_clock_adjtime,sys_clock_adjtime,compat_sys_clock_adjtime_wrapper)
349SYSCALL(sys_syncfs,sys_syncfs,sys_syncfs_wrapper)
350SYSCALL(sys_setns,sys_setns,sys_setns_wrapper)
diff --git a/arch/s390/kernel/sysinfo.c b/arch/s390/kernel/sysinfo.c
index a0ffc7717ed6..5c9e439bf3f6 100644
--- a/arch/s390/kernel/sysinfo.c
+++ b/arch/s390/kernel/sysinfo.c
@@ -15,6 +15,7 @@
15#include <asm/ebcdic.h> 15#include <asm/ebcdic.h>
16#include <asm/sysinfo.h> 16#include <asm/sysinfo.h>
17#include <asm/cpcmd.h> 17#include <asm/cpcmd.h>
18#include <asm/topology.h>
18 19
19/* Sigh, math-emu. Don't ask. */ 20/* Sigh, math-emu. Don't ask. */
20#include <asm/sfp-util.h> 21#include <asm/sfp-util.h>
@@ -74,6 +75,44 @@ static int stsi_1_1_1(struct sysinfo_1_1_1 *info, char *page, int len)
74 "Model Temp. Capacity: %-16.16s %08u\n", 75 "Model Temp. Capacity: %-16.16s %08u\n",
75 info->model_temp_cap, 76 info->model_temp_cap,
76 *(u32 *) info->model_temp_cap_rating); 77 *(u32 *) info->model_temp_cap_rating);
78 if (info->cai) {
79 len += sprintf(page + len,
80 "Capacity Adj. Ind.: %d\n",
81 info->cai);
82 len += sprintf(page + len, "Capacity Ch. Reason: %d\n",
83 info->ccr);
84 }
85 return len;
86}
87
88static int stsi_15_1_x(struct sysinfo_15_1_x *info, char *page, int len)
89{
90 static int max_mnest;
91 int i, rc;
92
93 len += sprintf(page + len, "\n");
94 if (!MACHINE_HAS_TOPOLOGY)
95 return len;
96 if (max_mnest) {
97 stsi(info, 15, 1, max_mnest);
98 } else {
99 for (max_mnest = 6; max_mnest > 1; max_mnest--) {
100 rc = stsi(info, 15, 1, max_mnest);
101 if (rc != -ENOSYS)
102 break;
103 }
104 }
105 len += sprintf(page + len, "CPU Topology HW: ");
106 for (i = 0; i < TOPOLOGY_NR_MAG; i++)
107 len += sprintf(page + len, " %d", info->mag[i]);
108 len += sprintf(page + len, "\n");
109#ifdef CONFIG_SCHED_MC
110 store_topology(info);
111 len += sprintf(page + len, "CPU Topology SW: ");
112 for (i = 0; i < TOPOLOGY_NR_MAG; i++)
113 len += sprintf(page + len, " %d", info->mag[i]);
114 len += sprintf(page + len, "\n");
115#endif
77 return len; 116 return len;
78} 117}
79 118
@@ -87,7 +126,6 @@ static int stsi_1_2_2(struct sysinfo_1_2_2 *info, char *page, int len)
87 ext = (struct sysinfo_1_2_2_extension *) 126 ext = (struct sysinfo_1_2_2_extension *)
88 ((unsigned long) info + info->acc_offset); 127 ((unsigned long) info + info->acc_offset);
89 128
90 len += sprintf(page + len, "\n");
91 len += sprintf(page + len, "CPUs Total: %d\n", 129 len += sprintf(page + len, "CPUs Total: %d\n",
92 info->cpus_total); 130 info->cpus_total);
93 len += sprintf(page + len, "CPUs Configured: %d\n", 131 len += sprintf(page + len, "CPUs Configured: %d\n",
@@ -217,6 +255,9 @@ static int proc_read_sysinfo(char *page, char **start,
217 len = stsi_1_1_1((struct sysinfo_1_1_1 *) info, page, len); 255 len = stsi_1_1_1((struct sysinfo_1_1_1 *) info, page, len);
218 256
219 if (level >= 1) 257 if (level >= 1)
258 len = stsi_15_1_x((struct sysinfo_15_1_x *) info, page, len);
259
260 if (level >= 1)
220 len = stsi_1_2_2((struct sysinfo_1_2_2 *) info, page, len); 261 len = stsi_1_2_2((struct sysinfo_1_2_2 *) info, page, len);
221 262
222 if (level >= 2) 263 if (level >= 2)
diff --git a/arch/s390/kernel/time.c b/arch/s390/kernel/time.c
index 2896cac9c14a..dff933065ab6 100644
--- a/arch/s390/kernel/time.c
+++ b/arch/s390/kernel/time.c
@@ -15,6 +15,7 @@
15#define KMSG_COMPONENT "time" 15#define KMSG_COMPONENT "time"
16#define pr_fmt(fmt) KMSG_COMPONENT ": " fmt 16#define pr_fmt(fmt) KMSG_COMPONENT ": " fmt
17 17
18#include <linux/kernel_stat.h>
18#include <linux/errno.h> 19#include <linux/errno.h>
19#include <linux/module.h> 20#include <linux/module.h>
20#include <linux/sched.h> 21#include <linux/sched.h>
@@ -37,9 +38,9 @@
37#include <linux/clocksource.h> 38#include <linux/clocksource.h>
38#include <linux/clockchips.h> 39#include <linux/clockchips.h>
39#include <linux/gfp.h> 40#include <linux/gfp.h>
41#include <linux/kprobes.h>
40#include <asm/uaccess.h> 42#include <asm/uaccess.h>
41#include <asm/delay.h> 43#include <asm/delay.h>
42#include <asm/s390_ext.h>
43#include <asm/div64.h> 44#include <asm/div64.h>
44#include <asm/vdso.h> 45#include <asm/vdso.h>
45#include <asm/irq.h> 46#include <asm/irq.h>
@@ -60,7 +61,7 @@ static DEFINE_PER_CPU(struct clock_event_device, comparators);
60/* 61/*
61 * Scheduler clock - returns current time in nanosec units. 62 * Scheduler clock - returns current time in nanosec units.
62 */ 63 */
63unsigned long long notrace sched_clock(void) 64unsigned long long notrace __kprobes sched_clock(void)
64{ 65{
65 return (get_clock_monotonic() * 125) >> 9; 66 return (get_clock_monotonic() * 125) >> 9;
66} 67}
@@ -155,8 +156,11 @@ void init_cpu_timer(void)
155 __ctl_set_bit(0, 4); 156 __ctl_set_bit(0, 4);
156} 157}
157 158
158static void clock_comparator_interrupt(__u16 code) 159static void clock_comparator_interrupt(unsigned int ext_int_code,
160 unsigned int param32,
161 unsigned long param64)
159{ 162{
163 kstat_cpu(smp_processor_id()).irqs[EXTINT_CLK]++;
160 if (S390_lowcore.clock_comparator == -1ULL) 164 if (S390_lowcore.clock_comparator == -1ULL)
161 set_clock_comparator(S390_lowcore.clock_comparator); 165 set_clock_comparator(S390_lowcore.clock_comparator);
162} 166}
@@ -164,14 +168,14 @@ static void clock_comparator_interrupt(__u16 code)
164static void etr_timing_alert(struct etr_irq_parm *); 168static void etr_timing_alert(struct etr_irq_parm *);
165static void stp_timing_alert(struct stp_irq_parm *); 169static void stp_timing_alert(struct stp_irq_parm *);
166 170
167static void timing_alert_interrupt(__u16 code) 171static void timing_alert_interrupt(unsigned int ext_int_code,
172 unsigned int param32, unsigned long param64)
168{ 173{
169 if (S390_lowcore.ext_params & 0x00c40000) 174 kstat_cpu(smp_processor_id()).irqs[EXTINT_TLA]++;
170 etr_timing_alert((struct etr_irq_parm *) 175 if (param32 & 0x00c40000)
171 &S390_lowcore.ext_params); 176 etr_timing_alert((struct etr_irq_parm *) &param32);
172 if (S390_lowcore.ext_params & 0x00038000) 177 if (param32 & 0x00038000)
173 stp_timing_alert((struct stp_irq_parm *) 178 stp_timing_alert((struct stp_irq_parm *) &param32);
174 &S390_lowcore.ext_params);
175} 179}
176 180
177static void etr_reset(void); 181static void etr_reset(void);
@@ -719,7 +723,7 @@ static void clock_sync_cpu(struct clock_sync_data *sync)
719} 723}
720 724
721/* 725/*
722 * Sync the TOD clock using the port refered to by aibp. This port 726 * Sync the TOD clock using the port referred to by aibp. This port
723 * has to be enabled and the other port has to be disabled. The 727 * has to be enabled and the other port has to be disabled. The
724 * last eacr update has to be more than 1.6 seconds in the past. 728 * last eacr update has to be more than 1.6 seconds in the past.
725 */ 729 */
@@ -805,7 +809,7 @@ static int etr_sync_clock_stop(struct etr_aib *aib, int port)
805 etr_sync.etr_port = port; 809 etr_sync.etr_port = port;
806 get_online_cpus(); 810 get_online_cpus();
807 atomic_set(&etr_sync.cpus, num_online_cpus() - 1); 811 atomic_set(&etr_sync.cpus, num_online_cpus() - 1);
808 rc = stop_machine(etr_sync_clock, &etr_sync, &cpu_online_map); 812 rc = stop_machine(etr_sync_clock, &etr_sync, cpu_online_mask);
809 put_online_cpus(); 813 put_online_cpus();
810 return rc; 814 return rc;
811} 815}
@@ -1007,7 +1011,7 @@ static void etr_work_fn(struct work_struct *work)
1007 eacr = etr_handle_update(&aib, eacr); 1011 eacr = etr_handle_update(&aib, eacr);
1008 1012
1009 /* 1013 /*
1010 * Select ports to enable. The prefered synchronization mode is PPS. 1014 * Select ports to enable. The preferred synchronization mode is PPS.
1011 * If a port can be enabled depends on a number of things: 1015 * If a port can be enabled depends on a number of things:
1012 * 1) The port needs to be online and uptodate. A port is not 1016 * 1) The port needs to be online and uptodate. A port is not
1013 * disabled just because it is not uptodate, but it is only 1017 * disabled just because it is not uptodate, but it is only
@@ -1086,7 +1090,7 @@ static void etr_work_fn(struct work_struct *work)
1086 /* 1090 /*
1087 * Update eacr and try to synchronize the clock. If the update 1091 * Update eacr and try to synchronize the clock. If the update
1088 * of eacr caused a stepping port switch (or if we have to 1092 * of eacr caused a stepping port switch (or if we have to
1089 * assume that a stepping port switch has occured) or the 1093 * assume that a stepping port switch has occurred) or the
1090 * clock syncing failed, reset the sync check control bit 1094 * clock syncing failed, reset the sync check control bit
1091 * and set up a timer to try again after 0.5 seconds 1095 * and set up a timer to try again after 0.5 seconds
1092 */ 1096 */
@@ -1574,7 +1578,7 @@ static void stp_work_fn(struct work_struct *work)
1574 memset(&stp_sync, 0, sizeof(stp_sync)); 1578 memset(&stp_sync, 0, sizeof(stp_sync));
1575 get_online_cpus(); 1579 get_online_cpus();
1576 atomic_set(&stp_sync.cpus, num_online_cpus() - 1); 1580 atomic_set(&stp_sync.cpus, num_online_cpus() - 1);
1577 stop_machine(stp_sync_clock, &stp_sync, &cpu_online_map); 1581 stop_machine(stp_sync_clock, &stp_sync, cpu_online_mask);
1578 put_online_cpus(); 1582 put_online_cpus();
1579 1583
1580 if (!check_sync_clock()) 1584 if (!check_sync_clock())
diff --git a/arch/s390/kernel/topology.c b/arch/s390/kernel/topology.c
index bcef00766a64..0cd340b72632 100644
--- a/arch/s390/kernel/topology.c
+++ b/arch/s390/kernel/topology.c
@@ -17,159 +17,140 @@
17#include <linux/smp.h> 17#include <linux/smp.h>
18#include <linux/cpuset.h> 18#include <linux/cpuset.h>
19#include <asm/delay.h> 19#include <asm/delay.h>
20#include <asm/s390_ext.h>
21#include <asm/sysinfo.h>
22
23#define CPU_BITS 64
24#define NR_MAG 6
25 20
26#define PTF_HORIZONTAL (0UL) 21#define PTF_HORIZONTAL (0UL)
27#define PTF_VERTICAL (1UL) 22#define PTF_VERTICAL (1UL)
28#define PTF_CHECK (2UL) 23#define PTF_CHECK (2UL)
29 24
30struct tl_cpu { 25struct mask_info {
31 unsigned char reserved0[4]; 26 struct mask_info *next;
32 unsigned char :6;
33 unsigned char pp:2;
34 unsigned char reserved1;
35 unsigned short origin;
36 unsigned long mask[CPU_BITS / BITS_PER_LONG];
37};
38
39struct tl_container {
40 unsigned char reserved[7];
41 unsigned char id;
42};
43
44union tl_entry {
45 unsigned char nl;
46 struct tl_cpu cpu;
47 struct tl_container container;
48};
49
50struct tl_info {
51 unsigned char reserved0[2];
52 unsigned short length;
53 unsigned char mag[NR_MAG];
54 unsigned char reserved1;
55 unsigned char mnest;
56 unsigned char reserved2[4];
57 union tl_entry tle[0];
58};
59
60struct core_info {
61 struct core_info *next;
62 unsigned char id; 27 unsigned char id;
63 cpumask_t mask; 28 cpumask_t mask;
64}; 29};
65 30
66static int topology_enabled; 31static int topology_enabled = 1;
67static void topology_work_fn(struct work_struct *work); 32static void topology_work_fn(struct work_struct *work);
68static struct tl_info *tl_info; 33static struct sysinfo_15_1_x *tl_info;
69static struct core_info core_info;
70static int machine_has_topology;
71static struct timer_list topology_timer; 34static struct timer_list topology_timer;
72static void set_topology_timer(void); 35static void set_topology_timer(void);
73static DECLARE_WORK(topology_work, topology_work_fn); 36static DECLARE_WORK(topology_work, topology_work_fn);
74/* topology_lock protects the core linked list */ 37/* topology_lock protects the core linked list */
75static DEFINE_SPINLOCK(topology_lock); 38static DEFINE_SPINLOCK(topology_lock);
76 39
40static struct mask_info core_info;
77cpumask_t cpu_core_map[NR_CPUS]; 41cpumask_t cpu_core_map[NR_CPUS];
78unsigned char cpu_core_id[NR_CPUS]; 42unsigned char cpu_core_id[NR_CPUS];
79 43
80static cpumask_t cpu_coregroup_map(unsigned int cpu) 44#ifdef CONFIG_SCHED_BOOK
45static struct mask_info book_info;
46cpumask_t cpu_book_map[NR_CPUS];
47unsigned char cpu_book_id[NR_CPUS];
48#endif
49
50static cpumask_t cpu_group_map(struct mask_info *info, unsigned int cpu)
81{ 51{
82 struct core_info *core = &core_info;
83 unsigned long flags;
84 cpumask_t mask; 52 cpumask_t mask;
85 53
86 cpus_clear(mask); 54 cpumask_clear(&mask);
87 if (!topology_enabled || !machine_has_topology) 55 if (!topology_enabled || !MACHINE_HAS_TOPOLOGY) {
88 return cpu_possible_map; 56 cpumask_copy(&mask, cpumask_of(cpu));
89 spin_lock_irqsave(&topology_lock, flags); 57 return mask;
90 while (core) { 58 }
91 if (cpu_isset(cpu, core->mask)) { 59 while (info) {
92 mask = core->mask; 60 if (cpumask_test_cpu(cpu, &info->mask)) {
61 mask = info->mask;
93 break; 62 break;
94 } 63 }
95 core = core->next; 64 info = info->next;
96 } 65 }
97 spin_unlock_irqrestore(&topology_lock, flags); 66 if (cpumask_empty(&mask))
98 if (cpus_empty(mask)) 67 cpumask_copy(&mask, cpumask_of(cpu));
99 mask = cpumask_of_cpu(cpu);
100 return mask; 68 return mask;
101} 69}
102 70
103const struct cpumask *cpu_coregroup_mask(unsigned int cpu) 71static void add_cpus_to_mask(struct topology_cpu *tl_cpu,
104{ 72 struct mask_info *book, struct mask_info *core)
105 return &cpu_core_map[cpu];
106}
107
108static void add_cpus_to_core(struct tl_cpu *tl_cpu, struct core_info *core)
109{ 73{
110 unsigned int cpu; 74 unsigned int cpu;
111 75
112 for (cpu = find_first_bit(&tl_cpu->mask[0], CPU_BITS); 76 for (cpu = find_first_bit(&tl_cpu->mask[0], TOPOLOGY_CPU_BITS);
113 cpu < CPU_BITS; 77 cpu < TOPOLOGY_CPU_BITS;
114 cpu = find_next_bit(&tl_cpu->mask[0], CPU_BITS, cpu + 1)) 78 cpu = find_next_bit(&tl_cpu->mask[0], TOPOLOGY_CPU_BITS, cpu + 1))
115 { 79 {
116 unsigned int rcpu, lcpu; 80 unsigned int rcpu, lcpu;
117 81
118 rcpu = CPU_BITS - 1 - cpu + tl_cpu->origin; 82 rcpu = TOPOLOGY_CPU_BITS - 1 - cpu + tl_cpu->origin;
119 for_each_present_cpu(lcpu) { 83 for_each_present_cpu(lcpu) {
120 if (cpu_logical_map(lcpu) == rcpu) { 84 if (cpu_logical_map(lcpu) != rcpu)
121 cpu_set(lcpu, core->mask); 85 continue;
122 cpu_core_id[lcpu] = core->id; 86#ifdef CONFIG_SCHED_BOOK
123 smp_cpu_polarization[lcpu] = tl_cpu->pp; 87 cpumask_set_cpu(lcpu, &book->mask);
124 } 88 cpu_book_id[lcpu] = book->id;
89#endif
90 cpumask_set_cpu(lcpu, &core->mask);
91 cpu_core_id[lcpu] = core->id;
92 smp_cpu_polarization[lcpu] = tl_cpu->pp;
125 } 93 }
126 } 94 }
127} 95}
128 96
129static void clear_cores(void) 97static void clear_masks(void)
130{ 98{
131 struct core_info *core = &core_info; 99 struct mask_info *info;
132 100
133 while (core) { 101 info = &core_info;
134 cpus_clear(core->mask); 102 while (info) {
135 core = core->next; 103 cpumask_clear(&info->mask);
104 info = info->next;
136 } 105 }
106#ifdef CONFIG_SCHED_BOOK
107 info = &book_info;
108 while (info) {
109 cpumask_clear(&info->mask);
110 info = info->next;
111 }
112#endif
137} 113}
138 114
139static union tl_entry *next_tle(union tl_entry *tle) 115static union topology_entry *next_tle(union topology_entry *tle)
140{ 116{
141 if (tle->nl) 117 if (!tle->nl)
142 return (union tl_entry *)((struct tl_container *)tle + 1); 118 return (union topology_entry *)((struct topology_cpu *)tle + 1);
143 else 119 return (union topology_entry *)((struct topology_container *)tle + 1);
144 return (union tl_entry *)((struct tl_cpu *)tle + 1);
145} 120}
146 121
147static void tl_to_cores(struct tl_info *info) 122static void tl_to_cores(struct sysinfo_15_1_x *info)
148{ 123{
149 union tl_entry *tle, *end; 124#ifdef CONFIG_SCHED_BOOK
150 struct core_info *core = &core_info; 125 struct mask_info *book = &book_info;
126#else
127 struct mask_info *book = NULL;
128#endif
129 struct mask_info *core = &core_info;
130 union topology_entry *tle, *end;
131
151 132
152 spin_lock_irq(&topology_lock); 133 spin_lock_irq(&topology_lock);
153 clear_cores(); 134 clear_masks();
154 tle = info->tle; 135 tle = info->tle;
155 end = (union tl_entry *)((unsigned long)info + info->length); 136 end = (union topology_entry *)((unsigned long)info + info->length);
156 while (tle < end) { 137 while (tle < end) {
157 switch (tle->nl) { 138 switch (tle->nl) {
158 case 5: 139#ifdef CONFIG_SCHED_BOOK
159 case 4:
160 case 3:
161 case 2: 140 case 2:
141 book = book->next;
142 book->id = tle->container.id;
162 break; 143 break;
144#endif
163 case 1: 145 case 1:
164 core = core->next; 146 core = core->next;
165 core->id = tle->container.id; 147 core->id = tle->container.id;
166 break; 148 break;
167 case 0: 149 case 0:
168 add_cpus_to_core(&tle->cpu, core); 150 add_cpus_to_mask(&tle->cpu, book, core);
169 break; 151 break;
170 default: 152 default:
171 clear_cores(); 153 clear_masks();
172 machine_has_topology = 0;
173 goto out; 154 goto out;
174 } 155 }
175 tle = next_tle(tle); 156 tle = next_tle(tle);
@@ -206,7 +187,7 @@ int topology_set_cpu_management(int fc)
206 int cpu; 187 int cpu;
207 int rc; 188 int rc;
208 189
209 if (!machine_has_topology) 190 if (!MACHINE_HAS_TOPOLOGY)
210 return -EOPNOTSUPP; 191 return -EOPNOTSUPP;
211 if (fc) 192 if (fc)
212 rc = ptf(PTF_VERTICAL); 193 rc = ptf(PTF_VERTICAL);
@@ -221,24 +202,43 @@ int topology_set_cpu_management(int fc)
221 202
222static void update_cpu_core_map(void) 203static void update_cpu_core_map(void)
223{ 204{
205 unsigned long flags;
224 int cpu; 206 int cpu;
225 207
226 for_each_possible_cpu(cpu) 208 spin_lock_irqsave(&topology_lock, flags);
227 cpu_core_map[cpu] = cpu_coregroup_map(cpu); 209 for_each_possible_cpu(cpu) {
210 cpu_core_map[cpu] = cpu_group_map(&core_info, cpu);
211#ifdef CONFIG_SCHED_BOOK
212 cpu_book_map[cpu] = cpu_group_map(&book_info, cpu);
213#endif
214 }
215 spin_unlock_irqrestore(&topology_lock, flags);
216}
217
218void store_topology(struct sysinfo_15_1_x *info)
219{
220#ifdef CONFIG_SCHED_BOOK
221 int rc;
222
223 rc = stsi(info, 15, 1, 3);
224 if (rc != -ENOSYS)
225 return;
226#endif
227 stsi(info, 15, 1, 2);
228} 228}
229 229
230int arch_update_cpu_topology(void) 230int arch_update_cpu_topology(void)
231{ 231{
232 struct tl_info *info = tl_info; 232 struct sysinfo_15_1_x *info = tl_info;
233 struct sys_device *sysdev; 233 struct sys_device *sysdev;
234 int cpu; 234 int cpu;
235 235
236 if (!machine_has_topology) { 236 if (!MACHINE_HAS_TOPOLOGY) {
237 update_cpu_core_map(); 237 update_cpu_core_map();
238 topology_update_polarization_simple(); 238 topology_update_polarization_simple();
239 return 0; 239 return 0;
240 } 240 }
241 stsi(info, 15, 1, 2); 241 store_topology(info);
242 tl_to_cores(info); 242 tl_to_cores(info);
243 update_cpu_core_map(); 243 update_cpu_core_map();
244 for_each_online_cpu(cpu) { 244 for_each_online_cpu(cpu) {
@@ -275,9 +275,9 @@ static void set_topology_timer(void)
275 275
276static int __init early_parse_topology(char *p) 276static int __init early_parse_topology(char *p)
277{ 277{
278 if (strncmp(p, "on", 2)) 278 if (strncmp(p, "off", 3))
279 return 0; 279 return 0;
280 topology_enabled = 1; 280 topology_enabled = 0;
281 return 0; 281 return 0;
282} 282}
283early_param("topology", early_parse_topology); 283early_param("topology", early_parse_topology);
@@ -287,7 +287,7 @@ static int __init init_topology_update(void)
287 int rc; 287 int rc;
288 288
289 rc = 0; 289 rc = 0;
290 if (!machine_has_topology) { 290 if (!MACHINE_HAS_TOPOLOGY) {
291 topology_update_polarization_simple(); 291 topology_update_polarization_simple();
292 goto out; 292 goto out;
293 } 293 }
@@ -299,41 +299,37 @@ out:
299} 299}
300__initcall(init_topology_update); 300__initcall(init_topology_update);
301 301
302static void alloc_masks(struct sysinfo_15_1_x *info, struct mask_info *mask,
303 int offset)
304{
305 int i, nr_masks;
306
307 nr_masks = info->mag[TOPOLOGY_NR_MAG - offset];
308 for (i = 0; i < info->mnest - offset; i++)
309 nr_masks *= info->mag[TOPOLOGY_NR_MAG - offset - 1 - i];
310 nr_masks = max(nr_masks, 1);
311 for (i = 0; i < nr_masks; i++) {
312 mask->next = alloc_bootmem(sizeof(struct mask_info));
313 mask = mask->next;
314 }
315}
316
302void __init s390_init_cpu_topology(void) 317void __init s390_init_cpu_topology(void)
303{ 318{
304 unsigned long long facility_bits; 319 struct sysinfo_15_1_x *info;
305 struct tl_info *info;
306 struct core_info *core;
307 int nr_cores;
308 int i; 320 int i;
309 321
310 if (stfle(&facility_bits, 1) <= 0) 322 if (!MACHINE_HAS_TOPOLOGY)
311 return;
312 if (!(facility_bits & (1ULL << 52)) || !(facility_bits & (1ULL << 61)))
313 return; 323 return;
314 machine_has_topology = 1;
315
316 tl_info = alloc_bootmem_pages(PAGE_SIZE); 324 tl_info = alloc_bootmem_pages(PAGE_SIZE);
317 info = tl_info; 325 info = tl_info;
318 stsi(info, 15, 1, 2); 326 store_topology(info);
319
320 nr_cores = info->mag[NR_MAG - 2];
321 for (i = 0; i < info->mnest - 2; i++)
322 nr_cores *= info->mag[NR_MAG - 3 - i];
323
324 pr_info("The CPU configuration topology of the machine is:"); 327 pr_info("The CPU configuration topology of the machine is:");
325 for (i = 0; i < NR_MAG; i++) 328 for (i = 0; i < TOPOLOGY_NR_MAG; i++)
326 printk(" %d", info->mag[i]); 329 printk(" %d", info->mag[i]);
327 printk(" / %d\n", info->mnest); 330 printk(" / %d\n", info->mnest);
328 331 alloc_masks(info, &core_info, 2);
329 core = &core_info; 332#ifdef CONFIG_SCHED_BOOK
330 for (i = 0; i < nr_cores; i++) { 333 alloc_masks(info, &book_info, 3);
331 core->next = alloc_bootmem(sizeof(struct core_info)); 334#endif
332 core = core->next;
333 if (!core)
334 goto error;
335 }
336 return;
337error:
338 machine_has_topology = 0;
339} 335}
diff --git a/arch/s390/kernel/traps.c b/arch/s390/kernel/traps.c
index 5d8f0f3d0250..a65d2e82f61d 100644
--- a/arch/s390/kernel/traps.c
+++ b/arch/s390/kernel/traps.c
@@ -39,7 +39,6 @@
39#include <asm/atomic.h> 39#include <asm/atomic.h>
40#include <asm/mathemu.h> 40#include <asm/mathemu.h>
41#include <asm/cpcmd.h> 41#include <asm/cpcmd.h>
42#include <asm/s390_ext.h>
43#include <asm/lowcore.h> 42#include <asm/lowcore.h>
44#include <asm/debug.h> 43#include <asm/debug.h>
45#include "entry.h" 44#include "entry.h"
@@ -237,43 +236,6 @@ void show_regs(struct pt_regs *regs)
237 show_last_breaking_event(regs); 236 show_last_breaking_event(regs);
238} 237}
239 238
240/* This is called from fs/proc/array.c */
241void task_show_regs(struct seq_file *m, struct task_struct *task)
242{
243 struct pt_regs *regs;
244
245 regs = task_pt_regs(task);
246 seq_printf(m, "task: %p, ksp: %p\n",
247 task, (void *)task->thread.ksp);
248 seq_printf(m, "User PSW : %p %p\n",
249 (void *) regs->psw.mask, (void *)regs->psw.addr);
250
251 seq_printf(m, "User GPRS: " FOURLONG,
252 regs->gprs[0], regs->gprs[1],
253 regs->gprs[2], regs->gprs[3]);
254 seq_printf(m, " " FOURLONG,
255 regs->gprs[4], regs->gprs[5],
256 regs->gprs[6], regs->gprs[7]);
257 seq_printf(m, " " FOURLONG,
258 regs->gprs[8], regs->gprs[9],
259 regs->gprs[10], regs->gprs[11]);
260 seq_printf(m, " " FOURLONG,
261 regs->gprs[12], regs->gprs[13],
262 regs->gprs[14], regs->gprs[15]);
263 seq_printf(m, "User ACRS: %08x %08x %08x %08x\n",
264 task->thread.acrs[0], task->thread.acrs[1],
265 task->thread.acrs[2], task->thread.acrs[3]);
266 seq_printf(m, " %08x %08x %08x %08x\n",
267 task->thread.acrs[4], task->thread.acrs[5],
268 task->thread.acrs[6], task->thread.acrs[7]);
269 seq_printf(m, " %08x %08x %08x %08x\n",
270 task->thread.acrs[8], task->thread.acrs[9],
271 task->thread.acrs[10], task->thread.acrs[11]);
272 seq_printf(m, " %08x %08x %08x %08x\n",
273 task->thread.acrs[12], task->thread.acrs[13],
274 task->thread.acrs[14], task->thread.acrs[15]);
275}
276
277static DEFINE_SPINLOCK(die_lock); 239static DEFINE_SPINLOCK(die_lock);
278 240
279void die(const char * str, struct pt_regs * regs, long err) 241void die(const char * str, struct pt_regs * regs, long err)
@@ -329,27 +291,19 @@ int is_valid_bugaddr(unsigned long addr)
329 return 1; 291 return 1;
330} 292}
331 293
332static void __kprobes inline do_trap(long interruption_code, int signr, 294static inline void __kprobes do_trap(long pgm_int_code, int signr, char *str,
333 char *str, struct pt_regs *regs, 295 struct pt_regs *regs, siginfo_t *info)
334 siginfo_t *info)
335{ 296{
336 /* 297 if (notify_die(DIE_TRAP, str, regs, pgm_int_code,
337 * We got all needed information from the lowcore and can 298 pgm_int_code, signr) == NOTIFY_STOP)
338 * now safely switch on interrupts.
339 */
340 if (regs->psw.mask & PSW_MASK_PSTATE)
341 local_irq_enable();
342
343 if (notify_die(DIE_TRAP, str, regs, interruption_code,
344 interruption_code, signr) == NOTIFY_STOP)
345 return; 299 return;
346 300
347 if (regs->psw.mask & PSW_MASK_PSTATE) { 301 if (regs->psw.mask & PSW_MASK_PSTATE) {
348 struct task_struct *tsk = current; 302 struct task_struct *tsk = current;
349 303
350 tsk->thread.trap_no = interruption_code & 0xffff; 304 tsk->thread.trap_no = pgm_int_code & 0xffff;
351 force_sig_info(signr, info, tsk); 305 force_sig_info(signr, info, tsk);
352 report_user_fault(regs, interruption_code, signr); 306 report_user_fault(regs, pgm_int_code, signr);
353 } else { 307 } else {
354 const struct exception_table_entry *fixup; 308 const struct exception_table_entry *fixup;
355 fixup = search_exception_tables(regs->psw.addr & PSW_ADDR_INSN); 309 fixup = search_exception_tables(regs->psw.addr & PSW_ADDR_INSN);
@@ -361,77 +315,77 @@ static void __kprobes inline do_trap(long interruption_code, int signr,
361 btt = report_bug(regs->psw.addr & PSW_ADDR_INSN, regs); 315 btt = report_bug(regs->psw.addr & PSW_ADDR_INSN, regs);
362 if (btt == BUG_TRAP_TYPE_WARN) 316 if (btt == BUG_TRAP_TYPE_WARN)
363 return; 317 return;
364 die(str, regs, interruption_code); 318 die(str, regs, pgm_int_code);
365 } 319 }
366 } 320 }
367} 321}
368 322
369static inline void __user *get_check_address(struct pt_regs *regs) 323static inline void __user *get_psw_address(struct pt_regs *regs,
324 long pgm_int_code)
370{ 325{
371 return (void __user *)((regs->psw.addr-S390_lowcore.pgm_ilc) & PSW_ADDR_INSN); 326 return (void __user *)
327 ((regs->psw.addr - (pgm_int_code >> 16)) & PSW_ADDR_INSN);
372} 328}
373 329
374void __kprobes do_single_step(struct pt_regs *regs) 330void __kprobes do_per_trap(struct pt_regs *regs)
375{ 331{
376 if (notify_die(DIE_SSTEP, "sstep", regs, 0, 0, 332 if (notify_die(DIE_SSTEP, "sstep", regs, 0, 0, SIGTRAP) == NOTIFY_STOP)
377 SIGTRAP) == NOTIFY_STOP){
378 return; 333 return;
379 }
380 if (tracehook_consider_fatal_signal(current, SIGTRAP)) 334 if (tracehook_consider_fatal_signal(current, SIGTRAP))
381 force_sig(SIGTRAP, current); 335 force_sig(SIGTRAP, current);
382} 336}
383 337
384static void default_trap_handler(struct pt_regs * regs, long interruption_code) 338static void default_trap_handler(struct pt_regs *regs, long pgm_int_code,
339 unsigned long trans_exc_code)
385{ 340{
386 if (regs->psw.mask & PSW_MASK_PSTATE) { 341 if (regs->psw.mask & PSW_MASK_PSTATE) {
387 local_irq_enable(); 342 report_user_fault(regs, pgm_int_code, SIGSEGV);
388 report_user_fault(regs, interruption_code, SIGSEGV);
389 do_exit(SIGSEGV); 343 do_exit(SIGSEGV);
390 } else 344 } else
391 die("Unknown program exception", regs, interruption_code); 345 die("Unknown program exception", regs, pgm_int_code);
392} 346}
393 347
394#define DO_ERROR_INFO(signr, str, name, sicode, siaddr) \ 348#define DO_ERROR_INFO(name, signr, sicode, str) \
395static void name(struct pt_regs * regs, long interruption_code) \ 349static void name(struct pt_regs *regs, long pgm_int_code, \
350 unsigned long trans_exc_code) \
396{ \ 351{ \
397 siginfo_t info; \ 352 siginfo_t info; \
398 info.si_signo = signr; \ 353 info.si_signo = signr; \
399 info.si_errno = 0; \ 354 info.si_errno = 0; \
400 info.si_code = sicode; \ 355 info.si_code = sicode; \
401 info.si_addr = siaddr; \ 356 info.si_addr = get_psw_address(regs, pgm_int_code); \
402 do_trap(interruption_code, signr, str, regs, &info); \ 357 do_trap(pgm_int_code, signr, str, regs, &info); \
403} 358}
404 359
405DO_ERROR_INFO(SIGILL, "addressing exception", addressing_exception, 360DO_ERROR_INFO(addressing_exception, SIGILL, ILL_ILLADR,
406 ILL_ILLADR, get_check_address(regs)) 361 "addressing exception")
407DO_ERROR_INFO(SIGILL, "execute exception", execute_exception, 362DO_ERROR_INFO(execute_exception, SIGILL, ILL_ILLOPN,
408 ILL_ILLOPN, get_check_address(regs)) 363 "execute exception")
409DO_ERROR_INFO(SIGFPE, "fixpoint divide exception", divide_exception, 364DO_ERROR_INFO(divide_exception, SIGFPE, FPE_INTDIV,
410 FPE_INTDIV, get_check_address(regs)) 365 "fixpoint divide exception")
411DO_ERROR_INFO(SIGFPE, "fixpoint overflow exception", overflow_exception, 366DO_ERROR_INFO(overflow_exception, SIGFPE, FPE_INTOVF,
412 FPE_INTOVF, get_check_address(regs)) 367 "fixpoint overflow exception")
413DO_ERROR_INFO(SIGFPE, "HFP overflow exception", hfp_overflow_exception, 368DO_ERROR_INFO(hfp_overflow_exception, SIGFPE, FPE_FLTOVF,
414 FPE_FLTOVF, get_check_address(regs)) 369 "HFP overflow exception")
415DO_ERROR_INFO(SIGFPE, "HFP underflow exception", hfp_underflow_exception, 370DO_ERROR_INFO(hfp_underflow_exception, SIGFPE, FPE_FLTUND,
416 FPE_FLTUND, get_check_address(regs)) 371 "HFP underflow exception")
417DO_ERROR_INFO(SIGFPE, "HFP significance exception", hfp_significance_exception, 372DO_ERROR_INFO(hfp_significance_exception, SIGFPE, FPE_FLTRES,
418 FPE_FLTRES, get_check_address(regs)) 373 "HFP significance exception")
419DO_ERROR_INFO(SIGFPE, "HFP divide exception", hfp_divide_exception, 374DO_ERROR_INFO(hfp_divide_exception, SIGFPE, FPE_FLTDIV,
420 FPE_FLTDIV, get_check_address(regs)) 375 "HFP divide exception")
421DO_ERROR_INFO(SIGFPE, "HFP square root exception", hfp_sqrt_exception, 376DO_ERROR_INFO(hfp_sqrt_exception, SIGFPE, FPE_FLTINV,
422 FPE_FLTINV, get_check_address(regs)) 377 "HFP square root exception")
423DO_ERROR_INFO(SIGILL, "operand exception", operand_exception, 378DO_ERROR_INFO(operand_exception, SIGILL, ILL_ILLOPN,
424 ILL_ILLOPN, get_check_address(regs)) 379 "operand exception")
425DO_ERROR_INFO(SIGILL, "privileged operation", privileged_op, 380DO_ERROR_INFO(privileged_op, SIGILL, ILL_PRVOPC,
426 ILL_PRVOPC, get_check_address(regs)) 381 "privileged operation")
427DO_ERROR_INFO(SIGILL, "special operation exception", special_op_exception, 382DO_ERROR_INFO(special_op_exception, SIGILL, ILL_ILLOPN,
428 ILL_ILLOPN, get_check_address(regs)) 383 "special operation exception")
429DO_ERROR_INFO(SIGILL, "translation exception", translation_exception, 384DO_ERROR_INFO(translation_exception, SIGILL, ILL_ILLOPN,
430 ILL_ILLOPN, get_check_address(regs)) 385 "translation exception")
431 386
432static inline void 387static inline void do_fp_trap(struct pt_regs *regs, void __user *location,
433do_fp_trap(struct pt_regs *regs, void __user *location, 388 int fpc, long pgm_int_code)
434 int fpc, long interruption_code)
435{ 389{
436 siginfo_t si; 390 siginfo_t si;
437 391
@@ -453,26 +407,19 @@ do_fp_trap(struct pt_regs *regs, void __user *location,
453 else if (fpc & 0x0800) /* inexact */ 407 else if (fpc & 0x0800) /* inexact */
454 si.si_code = FPE_FLTRES; 408 si.si_code = FPE_FLTRES;
455 } 409 }
456 current->thread.ieee_instruction_pointer = (addr_t) location; 410 do_trap(pgm_int_code, SIGFPE,
457 do_trap(interruption_code, SIGFPE,
458 "floating point exception", regs, &si); 411 "floating point exception", regs, &si);
459} 412}
460 413
461static void illegal_op(struct pt_regs * regs, long interruption_code) 414static void __kprobes illegal_op(struct pt_regs *regs, long pgm_int_code,
415 unsigned long trans_exc_code)
462{ 416{
463 siginfo_t info; 417 siginfo_t info;
464 __u8 opcode[6]; 418 __u8 opcode[6];
465 __u16 __user *location; 419 __u16 __user *location;
466 int signal = 0; 420 int signal = 0;
467 421
468 location = get_check_address(regs); 422 location = get_psw_address(regs, pgm_int_code);
469
470 /*
471 * We got all needed information from the lowcore and can
472 * now safely switch on interrupts.
473 */
474 if (regs->psw.mask & PSW_MASK_PSTATE)
475 local_irq_enable();
476 423
477 if (regs->psw.mask & PSW_MASK_PSTATE) { 424 if (regs->psw.mask & PSW_MASK_PSTATE) {
478 if (get_user(*((__u16 *) opcode), (__u16 __user *) location)) 425 if (get_user(*((__u16 *) opcode), (__u16 __user *) location))
@@ -512,7 +459,7 @@ static void illegal_op(struct pt_regs * regs, long interruption_code)
512 * If we get an illegal op in kernel mode, send it through the 459 * If we get an illegal op in kernel mode, send it through the
513 * kprobes notifier. If kprobes doesn't pick it up, SIGILL 460 * kprobes notifier. If kprobes doesn't pick it up, SIGILL
514 */ 461 */
515 if (notify_die(DIE_BPT, "bpt", regs, interruption_code, 462 if (notify_die(DIE_BPT, "bpt", regs, pgm_int_code,
516 3, SIGTRAP) != NOTIFY_STOP) 463 3, SIGTRAP) != NOTIFY_STOP)
517 signal = SIGILL; 464 signal = SIGILL;
518 } 465 }
@@ -520,13 +467,13 @@ static void illegal_op(struct pt_regs * regs, long interruption_code)
520#ifdef CONFIG_MATHEMU 467#ifdef CONFIG_MATHEMU
521 if (signal == SIGFPE) 468 if (signal == SIGFPE)
522 do_fp_trap(regs, location, 469 do_fp_trap(regs, location,
523 current->thread.fp_regs.fpc, interruption_code); 470 current->thread.fp_regs.fpc, pgm_int_code);
524 else if (signal == SIGSEGV) { 471 else if (signal == SIGSEGV) {
525 info.si_signo = signal; 472 info.si_signo = signal;
526 info.si_errno = 0; 473 info.si_errno = 0;
527 info.si_code = SEGV_MAPERR; 474 info.si_code = SEGV_MAPERR;
528 info.si_addr = (void __user *) location; 475 info.si_addr = (void __user *) location;
529 do_trap(interruption_code, signal, 476 do_trap(pgm_int_code, signal,
530 "user address fault", regs, &info); 477 "user address fault", regs, &info);
531 } else 478 } else
532#endif 479#endif
@@ -535,28 +482,22 @@ static void illegal_op(struct pt_regs * regs, long interruption_code)
535 info.si_errno = 0; 482 info.si_errno = 0;
536 info.si_code = ILL_ILLOPC; 483 info.si_code = ILL_ILLOPC;
537 info.si_addr = (void __user *) location; 484 info.si_addr = (void __user *) location;
538 do_trap(interruption_code, signal, 485 do_trap(pgm_int_code, signal,
539 "illegal operation", regs, &info); 486 "illegal operation", regs, &info);
540 } 487 }
541} 488}
542 489
543 490
544#ifdef CONFIG_MATHEMU 491#ifdef CONFIG_MATHEMU
545asmlinkage void 492asmlinkage void specification_exception(struct pt_regs *regs,
546specification_exception(struct pt_regs * regs, long interruption_code) 493 long pgm_int_code,
494 unsigned long trans_exc_code)
547{ 495{
548 __u8 opcode[6]; 496 __u8 opcode[6];
549 __u16 __user *location = NULL; 497 __u16 __user *location = NULL;
550 int signal = 0; 498 int signal = 0;
551 499
552 location = (__u16 __user *) get_check_address(regs); 500 location = (__u16 __user *) get_psw_address(regs, pgm_int_code);
553
554 /*
555 * We got all needed information from the lowcore and can
556 * now safely switch on interrupts.
557 */
558 if (regs->psw.mask & PSW_MASK_PSTATE)
559 local_irq_enable();
560 501
561 if (regs->psw.mask & PSW_MASK_PSTATE) { 502 if (regs->psw.mask & PSW_MASK_PSTATE) {
562 get_user(*((__u16 *) opcode), location); 503 get_user(*((__u16 *) opcode), location);
@@ -592,35 +533,29 @@ specification_exception(struct pt_regs * regs, long interruption_code)
592 533
593 if (signal == SIGFPE) 534 if (signal == SIGFPE)
594 do_fp_trap(regs, location, 535 do_fp_trap(regs, location,
595 current->thread.fp_regs.fpc, interruption_code); 536 current->thread.fp_regs.fpc, pgm_int_code);
596 else if (signal) { 537 else if (signal) {
597 siginfo_t info; 538 siginfo_t info;
598 info.si_signo = signal; 539 info.si_signo = signal;
599 info.si_errno = 0; 540 info.si_errno = 0;
600 info.si_code = ILL_ILLOPN; 541 info.si_code = ILL_ILLOPN;
601 info.si_addr = location; 542 info.si_addr = location;
602 do_trap(interruption_code, signal, 543 do_trap(pgm_int_code, signal,
603 "specification exception", regs, &info); 544 "specification exception", regs, &info);
604 } 545 }
605} 546}
606#else 547#else
607DO_ERROR_INFO(SIGILL, "specification exception", specification_exception, 548DO_ERROR_INFO(specification_exception, SIGILL, ILL_ILLOPN,
608 ILL_ILLOPN, get_check_address(regs)); 549 "specification exception");
609#endif 550#endif
610 551
611static void data_exception(struct pt_regs * regs, long interruption_code) 552static void data_exception(struct pt_regs *regs, long pgm_int_code,
553 unsigned long trans_exc_code)
612{ 554{
613 __u16 __user *location; 555 __u16 __user *location;
614 int signal = 0; 556 int signal = 0;
615 557
616 location = get_check_address(regs); 558 location = get_psw_address(regs, pgm_int_code);
617
618 /*
619 * We got all needed information from the lowcore and can
620 * now safely switch on interrupts.
621 */
622 if (regs->psw.mask & PSW_MASK_PSTATE)
623 local_irq_enable();
624 559
625 if (MACHINE_HAS_IEEE) 560 if (MACHINE_HAS_IEEE)
626 asm volatile("stfpc %0" : "=m" (current->thread.fp_regs.fpc)); 561 asm volatile("stfpc %0" : "=m" (current->thread.fp_regs.fpc));
@@ -686,19 +621,19 @@ static void data_exception(struct pt_regs * regs, long interruption_code)
686 signal = SIGILL; 621 signal = SIGILL;
687 if (signal == SIGFPE) 622 if (signal == SIGFPE)
688 do_fp_trap(regs, location, 623 do_fp_trap(regs, location,
689 current->thread.fp_regs.fpc, interruption_code); 624 current->thread.fp_regs.fpc, pgm_int_code);
690 else if (signal) { 625 else if (signal) {
691 siginfo_t info; 626 siginfo_t info;
692 info.si_signo = signal; 627 info.si_signo = signal;
693 info.si_errno = 0; 628 info.si_errno = 0;
694 info.si_code = ILL_ILLOPN; 629 info.si_code = ILL_ILLOPN;
695 info.si_addr = location; 630 info.si_addr = location;
696 do_trap(interruption_code, signal, 631 do_trap(pgm_int_code, signal, "data exception", regs, &info);
697 "data exception", regs, &info);
698 } 632 }
699} 633}
700 634
701static void space_switch_exception(struct pt_regs * regs, long int_code) 635static void space_switch_exception(struct pt_regs *regs, long pgm_int_code,
636 unsigned long trans_exc_code)
702{ 637{
703 siginfo_t info; 638 siginfo_t info;
704 639
@@ -709,11 +644,11 @@ static void space_switch_exception(struct pt_regs * regs, long int_code)
709 info.si_signo = SIGILL; 644 info.si_signo = SIGILL;
710 info.si_errno = 0; 645 info.si_errno = 0;
711 info.si_code = ILL_PRVOPC; 646 info.si_code = ILL_PRVOPC;
712 info.si_addr = get_check_address(regs); 647 info.si_addr = get_psw_address(regs, pgm_int_code);
713 do_trap(int_code, SIGILL, "space switch event", regs, &info); 648 do_trap(pgm_int_code, SIGILL, "space switch event", regs, &info);
714} 649}
715 650
716asmlinkage void kernel_stack_overflow(struct pt_regs * regs) 651asmlinkage void __kprobes kernel_stack_overflow(struct pt_regs * regs)
717{ 652{
718 bust_spinlocks(1); 653 bust_spinlocks(1);
719 printk("Kernel stack overflow.\n"); 654 printk("Kernel stack overflow.\n");
@@ -758,5 +693,6 @@ void __init trap_init(void)
758 pgm_check_table[0x15] = &operand_exception; 693 pgm_check_table[0x15] = &operand_exception;
759 pgm_check_table[0x1C] = &space_switch_exception; 694 pgm_check_table[0x1C] = &space_switch_exception;
760 pgm_check_table[0x1D] = &hfp_sqrt_exception; 695 pgm_check_table[0x1D] = &hfp_sqrt_exception;
761 pfault_irq_init(); 696 /* Enable machine checks early. */
697 local_mcck_enable();
762} 698}
diff --git a/arch/s390/kernel/vdso.c b/arch/s390/kernel/vdso.c
index 6b83870507d5..d73630b4fe1d 100644
--- a/arch/s390/kernel/vdso.c
+++ b/arch/s390/kernel/vdso.c
@@ -84,11 +84,7 @@ struct vdso_data *vdso_data = &vdso_data_store.data;
84 */ 84 */
85static void vdso_init_data(struct vdso_data *vd) 85static void vdso_init_data(struct vdso_data *vd)
86{ 86{
87 unsigned int facility_list; 87 vd->ectg_available = user_mode != HOME_SPACE_MODE && test_facility(31);
88
89 facility_list = stfl();
90 vd->ectg_available =
91 user_mode != HOME_SPACE_MODE && (facility_list & 1);
92} 88}
93 89
94#ifdef CONFIG_64BIT 90#ifdef CONFIG_64BIT
@@ -207,7 +203,6 @@ int arch_setup_additional_pages(struct linux_binprm *bprm, int uses_interp)
207 if (!uses_interp) 203 if (!uses_interp)
208 return 0; 204 return 0;
209 205
210 vdso_base = mm->mmap_base;
211#ifdef CONFIG_64BIT 206#ifdef CONFIG_64BIT
212 vdso_pagelist = vdso64_pagelist; 207 vdso_pagelist = vdso64_pagelist;
213 vdso_pages = vdso64_pages; 208 vdso_pages = vdso64_pages;
@@ -237,8 +232,7 @@ int arch_setup_additional_pages(struct linux_binprm *bprm, int uses_interp)
237 * fail and end up putting it elsewhere. 232 * fail and end up putting it elsewhere.
238 */ 233 */
239 down_write(&mm->mmap_sem); 234 down_write(&mm->mmap_sem);
240 vdso_base = get_unmapped_area(NULL, vdso_base, 235 vdso_base = get_unmapped_area(NULL, 0, vdso_pages << PAGE_SHIFT, 0, 0);
241 vdso_pages << PAGE_SHIFT, 0, 0);
242 if (IS_ERR_VALUE(vdso_base)) { 236 if (IS_ERR_VALUE(vdso_base)) {
243 rc = vdso_base; 237 rc = vdso_base;
244 goto out_up; 238 goto out_up;
@@ -343,17 +337,17 @@ static int __init vdso_init(void)
343} 337}
344arch_initcall(vdso_init); 338arch_initcall(vdso_init);
345 339
346int in_gate_area_no_task(unsigned long addr) 340int in_gate_area_no_mm(unsigned long addr)
347{ 341{
348 return 0; 342 return 0;
349} 343}
350 344
351int in_gate_area(struct task_struct *task, unsigned long addr) 345int in_gate_area(struct mm_struct *mm, unsigned long addr)
352{ 346{
353 return 0; 347 return 0;
354} 348}
355 349
356struct vm_area_struct *get_gate_vma(struct task_struct *tsk) 350struct vm_area_struct *get_gate_vma(struct mm_struct *mm)
357{ 351{
358 return NULL; 352 return NULL;
359} 353}
diff --git a/arch/s390/kernel/vdso32/Makefile b/arch/s390/kernel/vdso32/Makefile
index d13e8755a8cc..8ad2b34ad151 100644
--- a/arch/s390/kernel/vdso32/Makefile
+++ b/arch/s390/kernel/vdso32/Makefile
@@ -22,6 +22,9 @@ obj-y += vdso32_wrapper.o
22extra-y += vdso32.lds 22extra-y += vdso32.lds
23CPPFLAGS_vdso32.lds += -P -C -U$(ARCH) 23CPPFLAGS_vdso32.lds += -P -C -U$(ARCH)
24 24
25# Disable gcov profiling for VDSO code
26GCOV_PROFILE := n
27
25# Force dependency (incbin is bad) 28# Force dependency (incbin is bad)
26$(obj)/vdso32_wrapper.o : $(obj)/vdso32.so 29$(obj)/vdso32_wrapper.o : $(obj)/vdso32.so
27 30
diff --git a/arch/s390/kernel/vdso32/clock_getres.S b/arch/s390/kernel/vdso32/clock_getres.S
index 9532c4e6a9d2..36aaa25d05da 100644
--- a/arch/s390/kernel/vdso32/clock_getres.S
+++ b/arch/s390/kernel/vdso32/clock_getres.S
@@ -19,9 +19,9 @@
19 .type __kernel_clock_getres,@function 19 .type __kernel_clock_getres,@function
20__kernel_clock_getres: 20__kernel_clock_getres:
21 .cfi_startproc 21 .cfi_startproc
22 chi %r2,CLOCK_REALTIME 22 chi %r2,__CLOCK_REALTIME
23 je 0f 23 je 0f
24 chi %r2,CLOCK_MONOTONIC 24 chi %r2,__CLOCK_MONOTONIC
25 jne 3f 25 jne 3f
260: ltr %r3,%r3 260: ltr %r3,%r3
27 jz 2f /* res == NULL */ 27 jz 2f /* res == NULL */
@@ -34,6 +34,6 @@ __kernel_clock_getres:
343: lhi %r1,__NR_clock_getres /* fallback to svc */ 343: lhi %r1,__NR_clock_getres /* fallback to svc */
35 svc 0 35 svc 0
36 br %r14 36 br %r14
374: .long CLOCK_REALTIME_RES 374: .long __CLOCK_REALTIME_RES
38 .cfi_endproc 38 .cfi_endproc
39 .size __kernel_clock_getres,.-__kernel_clock_getres 39 .size __kernel_clock_getres,.-__kernel_clock_getres
diff --git a/arch/s390/kernel/vdso32/clock_gettime.S b/arch/s390/kernel/vdso32/clock_gettime.S
index 969643954273..b2224e0b974c 100644
--- a/arch/s390/kernel/vdso32/clock_gettime.S
+++ b/arch/s390/kernel/vdso32/clock_gettime.S
@@ -21,9 +21,9 @@ __kernel_clock_gettime:
21 .cfi_startproc 21 .cfi_startproc
22 basr %r5,0 22 basr %r5,0
230: al %r5,21f-0b(%r5) /* get &_vdso_data */ 230: al %r5,21f-0b(%r5) /* get &_vdso_data */
24 chi %r2,CLOCK_REALTIME 24 chi %r2,__CLOCK_REALTIME
25 je 10f 25 je 10f
26 chi %r2,CLOCK_MONOTONIC 26 chi %r2,__CLOCK_MONOTONIC
27 jne 19f 27 jne 19f
28 28
29 /* CLOCK_MONOTONIC */ 29 /* CLOCK_MONOTONIC */
diff --git a/arch/s390/kernel/vdso64/Makefile b/arch/s390/kernel/vdso64/Makefile
index 449352dda9cd..2a8ddfd12a5b 100644
--- a/arch/s390/kernel/vdso64/Makefile
+++ b/arch/s390/kernel/vdso64/Makefile
@@ -22,6 +22,9 @@ obj-y += vdso64_wrapper.o
22extra-y += vdso64.lds 22extra-y += vdso64.lds
23CPPFLAGS_vdso64.lds += -P -C -U$(ARCH) 23CPPFLAGS_vdso64.lds += -P -C -U$(ARCH)
24 24
25# Disable gcov profiling for VDSO code
26GCOV_PROFILE := n
27
25# Force dependency (incbin is bad) 28# Force dependency (incbin is bad)
26$(obj)/vdso64_wrapper.o : $(obj)/vdso64.so 29$(obj)/vdso64_wrapper.o : $(obj)/vdso64.so
27 30
diff --git a/arch/s390/kernel/vdso64/clock_getres.S b/arch/s390/kernel/vdso64/clock_getres.S
index 9ce8caafdb4e..176e1f75f9aa 100644
--- a/arch/s390/kernel/vdso64/clock_getres.S
+++ b/arch/s390/kernel/vdso64/clock_getres.S
@@ -19,9 +19,9 @@
19 .type __kernel_clock_getres,@function 19 .type __kernel_clock_getres,@function
20__kernel_clock_getres: 20__kernel_clock_getres:
21 .cfi_startproc 21 .cfi_startproc
22 cghi %r2,CLOCK_REALTIME 22 cghi %r2,__CLOCK_REALTIME
23 je 0f 23 je 0f
24 cghi %r2,CLOCK_MONOTONIC 24 cghi %r2,__CLOCK_MONOTONIC
25 je 0f 25 je 0f
26 cghi %r2,-2 /* CLOCK_THREAD_CPUTIME_ID for this thread */ 26 cghi %r2,-2 /* CLOCK_THREAD_CPUTIME_ID for this thread */
27 jne 2f 27 jne 2f
@@ -39,6 +39,6 @@ __kernel_clock_getres:
392: lghi %r1,__NR_clock_getres /* fallback to svc */ 392: lghi %r1,__NR_clock_getres /* fallback to svc */
40 svc 0 40 svc 0
41 br %r14 41 br %r14
423: .quad CLOCK_REALTIME_RES 423: .quad __CLOCK_REALTIME_RES
43 .cfi_endproc 43 .cfi_endproc
44 .size __kernel_clock_getres,.-__kernel_clock_getres 44 .size __kernel_clock_getres,.-__kernel_clock_getres
diff --git a/arch/s390/kernel/vdso64/clock_gettime.S b/arch/s390/kernel/vdso64/clock_gettime.S
index f40467884a03..d46c95ed5f19 100644
--- a/arch/s390/kernel/vdso64/clock_gettime.S
+++ b/arch/s390/kernel/vdso64/clock_gettime.S
@@ -20,11 +20,11 @@
20__kernel_clock_gettime: 20__kernel_clock_gettime:
21 .cfi_startproc 21 .cfi_startproc
22 larl %r5,_vdso_data 22 larl %r5,_vdso_data
23 cghi %r2,CLOCK_REALTIME 23 cghi %r2,__CLOCK_REALTIME
24 je 4f 24 je 4f
25 cghi %r2,-2 /* CLOCK_THREAD_CPUTIME_ID for this thread */ 25 cghi %r2,-2 /* CLOCK_THREAD_CPUTIME_ID for this thread */
26 je 9f 26 je 9f
27 cghi %r2,CLOCK_MONOTONIC 27 cghi %r2,__CLOCK_MONOTONIC
28 jne 12f 28 jne 12f
29 29
30 /* CLOCK_MONOTONIC */ 30 /* CLOCK_MONOTONIC */
diff --git a/arch/s390/kernel/vmlinux.lds.S b/arch/s390/kernel/vmlinux.lds.S
index a68ac10213b2..56fe6bc81fee 100644
--- a/arch/s390/kernel/vmlinux.lds.S
+++ b/arch/s390/kernel/vmlinux.lds.S
@@ -77,7 +77,7 @@ SECTIONS
77 . = ALIGN(PAGE_SIZE); 77 . = ALIGN(PAGE_SIZE);
78 INIT_DATA_SECTION(0x100) 78 INIT_DATA_SECTION(0x100)
79 79
80 PERCPU(PAGE_SIZE) 80 PERCPU_SECTION(0x100)
81 . = ALIGN(PAGE_SIZE); 81 . = ALIGN(PAGE_SIZE);
82 __init_end = .; /* freed after init ends here */ 82 __init_end = .; /* freed after init ends here */
83 83
diff --git a/arch/s390/kernel/vtime.c b/arch/s390/kernel/vtime.c
index 3479f1b0d4e0..2d6228f60cd6 100644
--- a/arch/s390/kernel/vtime.c
+++ b/arch/s390/kernel/vtime.c
@@ -19,11 +19,13 @@
19#include <linux/kernel_stat.h> 19#include <linux/kernel_stat.h>
20#include <linux/rcupdate.h> 20#include <linux/rcupdate.h>
21#include <linux/posix-timers.h> 21#include <linux/posix-timers.h>
22#include <linux/cpu.h>
23#include <linux/kprobes.h>
22 24
23#include <asm/s390_ext.h>
24#include <asm/timer.h> 25#include <asm/timer.h>
25#include <asm/irq_regs.h> 26#include <asm/irq_regs.h>
26#include <asm/cputime.h> 27#include <asm/cputime.h>
28#include <asm/irq.h>
27 29
28static DEFINE_PER_CPU(struct vtimer_queue, virt_cpu_timer); 30static DEFINE_PER_CPU(struct vtimer_queue, virt_cpu_timer);
29 31
@@ -42,7 +44,7 @@ static inline void set_vtimer(__u64 expires)
42 __u64 timer; 44 __u64 timer;
43 45
44 asm volatile (" STPT %0\n" /* Store current cpu timer value */ 46 asm volatile (" STPT %0\n" /* Store current cpu timer value */
45 " SPT %1" /* Set new value immediatly afterwards */ 47 " SPT %1" /* Set new value immediately afterwards */
46 : "=m" (timer) : "m" (expires) ); 48 : "=m" (timer) : "m" (expires) );
47 S390_lowcore.system_timer += S390_lowcore.last_update_timer - timer; 49 S390_lowcore.system_timer += S390_lowcore.last_update_timer - timer;
48 S390_lowcore.last_update_timer = expires; 50 S390_lowcore.last_update_timer = expires;
@@ -121,7 +123,7 @@ void account_system_vtime(struct task_struct *tsk)
121} 123}
122EXPORT_SYMBOL_GPL(account_system_vtime); 124EXPORT_SYMBOL_GPL(account_system_vtime);
123 125
124void vtime_start_cpu(__u64 int_clock, __u64 enter_timer) 126void __kprobes vtime_start_cpu(__u64 int_clock, __u64 enter_timer)
125{ 127{
126 struct s390_idle_data *idle = &__get_cpu_var(s390_idle); 128 struct s390_idle_data *idle = &__get_cpu_var(s390_idle);
127 struct vtimer_queue *vq = &__get_cpu_var(virt_cpu_timer); 129 struct vtimer_queue *vq = &__get_cpu_var(virt_cpu_timer);
@@ -161,7 +163,7 @@ void vtime_start_cpu(__u64 int_clock, __u64 enter_timer)
161 idle->sequence++; 163 idle->sequence++;
162} 164}
163 165
164void vtime_stop_cpu(void) 166void __kprobes vtime_stop_cpu(void)
165{ 167{
166 struct s390_idle_data *idle = &__get_cpu_var(s390_idle); 168 struct s390_idle_data *idle = &__get_cpu_var(s390_idle);
167 struct vtimer_queue *vq = &__get_cpu_var(virt_cpu_timer); 169 struct vtimer_queue *vq = &__get_cpu_var(virt_cpu_timer);
@@ -314,13 +316,15 @@ static void do_callbacks(struct list_head *cb_list)
314/* 316/*
315 * Handler for the virtual CPU timer. 317 * Handler for the virtual CPU timer.
316 */ 318 */
317static void do_cpu_timer_interrupt(__u16 error_code) 319static void do_cpu_timer_interrupt(unsigned int ext_int_code,
320 unsigned int param32, unsigned long param64)
318{ 321{
319 struct vtimer_queue *vq; 322 struct vtimer_queue *vq;
320 struct vtimer_list *event, *tmp; 323 struct vtimer_list *event, *tmp;
321 struct list_head cb_list; /* the callback queue */ 324 struct list_head cb_list; /* the callback queue */
322 __u64 elapsed, next; 325 __u64 elapsed, next;
323 326
327 kstat_cpu(smp_processor_id()).irqs[EXTINT_TMR]++;
324 INIT_LIST_HEAD(&cb_list); 328 INIT_LIST_HEAD(&cb_list);
325 vq = &__get_cpu_var(virt_cpu_timer); 329 vq = &__get_cpu_var(virt_cpu_timer);
326 330
@@ -565,6 +569,23 @@ void init_cpu_vtimer(void)
565 __ctl_set_bit(0,10); 569 __ctl_set_bit(0,10);
566} 570}
567 571
572static int __cpuinit s390_nohz_notify(struct notifier_block *self,
573 unsigned long action, void *hcpu)
574{
575 struct s390_idle_data *idle;
576 long cpu = (long) hcpu;
577
578 idle = &per_cpu(s390_idle, cpu);
579 switch (action) {
580 case CPU_DYING:
581 case CPU_DYING_FROZEN:
582 idle->nohz_delay = 0;
583 default:
584 break;
585 }
586 return NOTIFY_OK;
587}
588
568void __init vtime_init(void) 589void __init vtime_init(void)
569{ 590{
570 /* request the cpu timer external interrupt */ 591 /* request the cpu timer external interrupt */
@@ -573,5 +594,6 @@ void __init vtime_init(void)
573 594
574 /* Enable cpu timer interrupts on the boot cpu. */ 595 /* Enable cpu timer interrupts on the boot cpu. */
575 init_cpu_vtimer(); 596 init_cpu_vtimer();
597 cpu_notifier(s390_nohz_notify, 0);
576} 598}
577 599
diff --git a/arch/s390/kvm/Kconfig b/arch/s390/kvm/Kconfig
index a7251580891c..f66a1bdbb61d 100644
--- a/arch/s390/kvm/Kconfig
+++ b/arch/s390/kvm/Kconfig
@@ -4,8 +4,8 @@
4source "virt/kvm/Kconfig" 4source "virt/kvm/Kconfig"
5 5
6menuconfig VIRTUALIZATION 6menuconfig VIRTUALIZATION
7 bool "Virtualization" 7 def_bool y
8 default y 8 prompt "Virtualization"
9 ---help--- 9 ---help---
10 Say Y here to get to see options for using your Linux host to run other 10 Say Y here to get to see options for using your Linux host to run other
11 operating systems inside virtual machines (guests). 11 operating systems inside virtual machines (guests).
@@ -16,7 +16,8 @@ menuconfig VIRTUALIZATION
16if VIRTUALIZATION 16if VIRTUALIZATION
17 17
18config KVM 18config KVM
19 tristate "Kernel-based Virtual Machine (KVM) support" 19 def_tristate y
20 prompt "Kernel-based Virtual Machine (KVM) support"
20 depends on HAVE_KVM && EXPERIMENTAL 21 depends on HAVE_KVM && EXPERIMENTAL
21 select PREEMPT_NOTIFIERS 22 select PREEMPT_NOTIFIERS
22 select ANON_INODES 23 select ANON_INODES
diff --git a/arch/s390/kvm/Makefile b/arch/s390/kvm/Makefile
index e5221ec0b8e3..860d26514c08 100644
--- a/arch/s390/kvm/Makefile
+++ b/arch/s390/kvm/Makefile
@@ -8,7 +8,7 @@
8 8
9common-objs = $(addprefix ../../../virt/kvm/, kvm_main.o) 9common-objs = $(addprefix ../../../virt/kvm/, kvm_main.o)
10 10
11EXTRA_CFLAGS += -Ivirt/kvm -Iarch/s390/kvm 11ccflags-y := -Ivirt/kvm -Iarch/s390/kvm
12 12
13kvm-objs := $(common-objs) kvm-s390.o sie64a.o intercept.o interrupt.o priv.o sigp.o diag.o 13kvm-objs := $(common-objs) kvm-s390.o sie64a.o intercept.o interrupt.o priv.o sigp.o diag.o
14obj-$(CONFIG_KVM) += kvm.o 14obj-$(CONFIG_KVM) += kvm.o
diff --git a/arch/s390/kvm/kvm-s390.c b/arch/s390/kvm/kvm-s390.c
index 4fe68650535c..67345ae7ce8d 100644
--- a/arch/s390/kvm/kvm-s390.c
+++ b/arch/s390/kvm/kvm-s390.c
@@ -164,24 +164,18 @@ long kvm_arch_vm_ioctl(struct file *filp,
164 return r; 164 return r;
165} 165}
166 166
167struct kvm *kvm_arch_create_vm(void) 167int kvm_arch_init_vm(struct kvm *kvm)
168{ 168{
169 struct kvm *kvm;
170 int rc; 169 int rc;
171 char debug_name[16]; 170 char debug_name[16];
172 171
173 rc = s390_enable_sie(); 172 rc = s390_enable_sie();
174 if (rc) 173 if (rc)
175 goto out_nokvm; 174 goto out_err;
176
177 rc = -ENOMEM;
178 kvm = kzalloc(sizeof(struct kvm), GFP_KERNEL);
179 if (!kvm)
180 goto out_nokvm;
181 175
182 kvm->arch.sca = (struct sca_block *) get_zeroed_page(GFP_KERNEL); 176 kvm->arch.sca = (struct sca_block *) get_zeroed_page(GFP_KERNEL);
183 if (!kvm->arch.sca) 177 if (!kvm->arch.sca)
184 goto out_nosca; 178 goto out_err;
185 179
186 sprintf(debug_name, "kvm-%u", current->pid); 180 sprintf(debug_name, "kvm-%u", current->pid);
187 181
@@ -195,13 +189,11 @@ struct kvm *kvm_arch_create_vm(void)
195 debug_register_view(kvm->arch.dbf, &debug_sprintf_view); 189 debug_register_view(kvm->arch.dbf, &debug_sprintf_view);
196 VM_EVENT(kvm, 3, "%s", "vm created"); 190 VM_EVENT(kvm, 3, "%s", "vm created");
197 191
198 return kvm; 192 return 0;
199out_nodbf: 193out_nodbf:
200 free_page((unsigned long)(kvm->arch.sca)); 194 free_page((unsigned long)(kvm->arch.sca));
201out_nosca: 195out_err:
202 kfree(kvm); 196 return rc;
203out_nokvm:
204 return ERR_PTR(rc);
205} 197}
206 198
207void kvm_arch_vcpu_destroy(struct kvm_vcpu *vcpu) 199void kvm_arch_vcpu_destroy(struct kvm_vcpu *vcpu)
@@ -240,11 +232,8 @@ void kvm_arch_sync_events(struct kvm *kvm)
240void kvm_arch_destroy_vm(struct kvm *kvm) 232void kvm_arch_destroy_vm(struct kvm *kvm)
241{ 233{
242 kvm_free_vcpus(kvm); 234 kvm_free_vcpus(kvm);
243 kvm_free_physmem(kvm);
244 free_page((unsigned long)(kvm->arch.sca)); 235 free_page((unsigned long)(kvm->arch.sca));
245 debug_unregister(kvm->arch.dbf); 236 debug_unregister(kvm->arch.dbf);
246 cleanup_srcu_struct(&kvm->srcu);
247 kfree(kvm);
248} 237}
249 238
250/* Section: vcpu related */ 239/* Section: vcpu related */
@@ -732,7 +721,7 @@ static int __init kvm_s390_init(void)
732 721
733 /* 722 /*
734 * guests can ask for up to 255+1 double words, we need a full page 723 * guests can ask for up to 255+1 double words, we need a full page
735 * to hold the maximum amount of facilites. On the other hand, we 724 * to hold the maximum amount of facilities. On the other hand, we
736 * only set facilities that are known to work in KVM. 725 * only set facilities that are known to work in KVM.
737 */ 726 */
738 facilities = (unsigned long long *) get_zeroed_page(GFP_KERNEL|GFP_DMA); 727 facilities = (unsigned long long *) get_zeroed_page(GFP_KERNEL|GFP_DMA);
@@ -740,8 +729,9 @@ static int __init kvm_s390_init(void)
740 kvm_exit(); 729 kvm_exit();
741 return -ENOMEM; 730 return -ENOMEM;
742 } 731 }
743 stfle(facilities, 1); 732 memcpy(facilities, S390_lowcore.stfle_fac_list, 16);
744 facilities[0] &= 0xff00fff3f0700000ULL; 733 facilities[0] &= 0xff00fff3f47c0000ULL;
734 facilities[1] &= 0x201c000000000000ULL;
745 return 0; 735 return 0;
746} 736}
747 737
diff --git a/arch/s390/kvm/priv.c b/arch/s390/kvm/priv.c
index 44205507717c..73c47bd95db3 100644
--- a/arch/s390/kvm/priv.c
+++ b/arch/s390/kvm/priv.c
@@ -154,12 +154,12 @@ static int handle_chsc(struct kvm_vcpu *vcpu)
154 154
155static int handle_stfl(struct kvm_vcpu *vcpu) 155static int handle_stfl(struct kvm_vcpu *vcpu)
156{ 156{
157 unsigned int facility_list = stfl(); 157 unsigned int facility_list;
158 int rc; 158 int rc;
159 159
160 vcpu->stat.instruction_stfl++; 160 vcpu->stat.instruction_stfl++;
161 /* only pass the facility bits, which we can handle */ 161 /* only pass the facility bits, which we can handle */
162 facility_list &= 0xff00fff3; 162 facility_list = S390_lowcore.stfl_fac_list & 0xff00fff3;
163 163
164 rc = copy_to_guest(vcpu, offsetof(struct _lowcore, stfl_fac_list), 164 rc = copy_to_guest(vcpu, offsetof(struct _lowcore, stfl_fac_list),
165 &facility_list, sizeof(facility_list)); 165 &facility_list, sizeof(facility_list));
@@ -311,7 +311,7 @@ int kvm_s390_handle_b2(struct kvm_vcpu *vcpu)
311 311
312 /* 312 /*
313 * a lot of B2 instructions are priviledged. We first check for 313 * a lot of B2 instructions are priviledged. We first check for
314 * the priviledges ones, that we can handle in the kernel. If the 314 * the privileged ones, that we can handle in the kernel. If the
315 * kernel can handle this instruction, we check for the problem 315 * kernel can handle this instruction, we check for the problem
316 * state bit and (a) handle the instruction or (b) send a code 2 316 * state bit and (a) handle the instruction or (b) send a code 2
317 * program check. 317 * program check.
diff --git a/arch/s390/kvm/sie64a.S b/arch/s390/kvm/sie64a.S
index 7e9d30d567b0..5faa1b1b23fa 100644
--- a/arch/s390/kvm/sie64a.S
+++ b/arch/s390/kvm/sie64a.S
@@ -48,10 +48,10 @@ sie_irq_handler:
48 tm __TI_flags+7(%r2),_TIF_EXIT_SIE 48 tm __TI_flags+7(%r2),_TIF_EXIT_SIE
49 jz 0f 49 jz 0f
50 larl %r2,sie_exit # work pending, leave sie 50 larl %r2,sie_exit # work pending, leave sie
51 stg %r2,__LC_RETURN_PSW+8 51 stg %r2,SPI_PSW+8(0,%r15)
52 br %r14 52 br %r14
530: larl %r2,sie_reenter # re-enter with guest id 530: larl %r2,sie_reenter # re-enter with guest id
54 stg %r2,__LC_RETURN_PSW+8 54 stg %r2,SPI_PSW+8(0,%r15)
551: br %r14 551: br %r14
56 56
57/* 57/*
@@ -93,4 +93,6 @@ sie_err:
93 93
94 .section __ex_table,"a" 94 .section __ex_table,"a"
95 .quad sie_inst,sie_err 95 .quad sie_inst,sie_err
96 .quad sie_exit,sie_err
97 .quad sie_reenter,sie_err
96 .previous 98 .previous
diff --git a/arch/s390/lib/delay.c b/arch/s390/lib/delay.c
index 752b362bf651..a65229d91c92 100644
--- a/arch/s390/lib/delay.c
+++ b/arch/s390/lib/delay.c
@@ -12,6 +12,7 @@
12#include <linux/module.h> 12#include <linux/module.h>
13#include <linux/irqflags.h> 13#include <linux/irqflags.h>
14#include <linux/interrupt.h> 14#include <linux/interrupt.h>
15#include <asm/div64.h>
15 16
16void __delay(unsigned long loops) 17void __delay(unsigned long loops)
17{ 18{
@@ -29,21 +30,24 @@ static void __udelay_disabled(unsigned long long usecs)
29{ 30{
30 unsigned long mask, cr0, cr0_saved; 31 unsigned long mask, cr0, cr0_saved;
31 u64 clock_saved; 32 u64 clock_saved;
33 u64 end;
32 34
35 mask = psw_kernel_bits | PSW_MASK_WAIT | PSW_MASK_EXT;
36 end = get_clock() + (usecs << 12);
33 clock_saved = local_tick_disable(); 37 clock_saved = local_tick_disable();
34 set_clock_comparator(get_clock() + (usecs << 12));
35 __ctl_store(cr0_saved, 0, 0); 38 __ctl_store(cr0_saved, 0, 0);
36 cr0 = (cr0_saved & 0xffff00e0) | 0x00000800; 39 cr0 = (cr0_saved & 0xffff00e0) | 0x00000800;
37 __ctl_load(cr0 , 0, 0); 40 __ctl_load(cr0 , 0, 0);
38 mask = psw_kernel_bits | PSW_MASK_WAIT | PSW_MASK_EXT;
39 lockdep_off(); 41 lockdep_off();
40 trace_hardirqs_on(); 42 do {
41 __load_psw_mask(mask); 43 set_clock_comparator(end);
42 local_irq_disable(); 44 trace_hardirqs_on();
45 __load_psw_mask(mask);
46 local_irq_disable();
47 } while (get_clock() < end);
43 lockdep_on(); 48 lockdep_on();
44 __ctl_load(cr0_saved, 0, 0); 49 __ctl_load(cr0_saved, 0, 0);
45 local_tick_enable(clock_saved); 50 local_tick_enable(clock_saved);
46 set_clock_comparator(S390_lowcore.clock_comparator);
47} 51}
48 52
49static void __udelay_enabled(unsigned long long usecs) 53static void __udelay_enabled(unsigned long long usecs)
@@ -66,7 +70,6 @@ static void __udelay_enabled(unsigned long long usecs)
66 if (clock_saved) 70 if (clock_saved)
67 local_tick_enable(clock_saved); 71 local_tick_enable(clock_saved);
68 } while (get_clock() < end); 72 } while (get_clock() < end);
69 set_clock_comparator(S390_lowcore.clock_comparator);
70} 73}
71 74
72/* 75/*
@@ -114,3 +117,17 @@ void udelay_simple(unsigned long long usecs)
114 while (get_clock() < end) 117 while (get_clock() < end)
115 cpu_relax(); 118 cpu_relax();
116} 119}
120
121void __ndelay(unsigned long long nsecs)
122{
123 u64 end;
124
125 nsecs <<= 9;
126 do_div(nsecs, 125);
127 end = get_clock() + nsecs;
128 if (nsecs & ~0xfffUL)
129 __udelay(nsecs >> 12);
130 while (get_clock() < end)
131 barrier();
132}
133EXPORT_SYMBOL(__ndelay);
diff --git a/arch/s390/lib/uaccess.h b/arch/s390/lib/uaccess.h
index 126011df14f1..1d2536cb630b 100644
--- a/arch/s390/lib/uaccess.h
+++ b/arch/s390/lib/uaccess.h
@@ -12,12 +12,12 @@ extern size_t copy_from_user_std(size_t, const void __user *, void *);
12extern size_t copy_to_user_std(size_t, void __user *, const void *); 12extern size_t copy_to_user_std(size_t, void __user *, const void *);
13extern size_t strnlen_user_std(size_t, const char __user *); 13extern size_t strnlen_user_std(size_t, const char __user *);
14extern size_t strncpy_from_user_std(size_t, const char __user *, char *); 14extern size_t strncpy_from_user_std(size_t, const char __user *, char *);
15extern int futex_atomic_cmpxchg_std(int __user *, int, int); 15extern int futex_atomic_cmpxchg_std(u32 *, u32 __user *, u32, u32);
16extern int futex_atomic_op_std(int, int __user *, int, int *); 16extern int futex_atomic_op_std(int, u32 __user *, int, int *);
17 17
18extern size_t copy_from_user_pt(size_t, const void __user *, void *); 18extern size_t copy_from_user_pt(size_t, const void __user *, void *);
19extern size_t copy_to_user_pt(size_t, void __user *, const void *); 19extern size_t copy_to_user_pt(size_t, void __user *, const void *);
20extern int futex_atomic_op_pt(int, int __user *, int, int *); 20extern int futex_atomic_op_pt(int, u32 __user *, int, int *);
21extern int futex_atomic_cmpxchg_pt(int __user *, int, int); 21extern int futex_atomic_cmpxchg_pt(u32 *, u32 __user *, u32, u32);
22 22
23#endif /* __ARCH_S390_LIB_UACCESS_H */ 23#endif /* __ARCH_S390_LIB_UACCESS_H */
diff --git a/arch/s390/lib/uaccess_pt.c b/arch/s390/lib/uaccess_pt.c
index 404f2de296dc..74833831417f 100644
--- a/arch/s390/lib/uaccess_pt.c
+++ b/arch/s390/lib/uaccess_pt.c
@@ -302,7 +302,7 @@ fault:
302 : "0" (-EFAULT), "d" (oparg), "a" (uaddr), \ 302 : "0" (-EFAULT), "d" (oparg), "a" (uaddr), \
303 "m" (*uaddr) : "cc" ); 303 "m" (*uaddr) : "cc" );
304 304
305static int __futex_atomic_op_pt(int op, int __user *uaddr, int oparg, int *old) 305static int __futex_atomic_op_pt(int op, u32 __user *uaddr, int oparg, int *old)
306{ 306{
307 int oldval = 0, newval, ret; 307 int oldval = 0, newval, ret;
308 308
@@ -335,7 +335,7 @@ static int __futex_atomic_op_pt(int op, int __user *uaddr, int oparg, int *old)
335 return ret; 335 return ret;
336} 336}
337 337
338int futex_atomic_op_pt(int op, int __user *uaddr, int oparg, int *old) 338int futex_atomic_op_pt(int op, u32 __user *uaddr, int oparg, int *old)
339{ 339{
340 int ret; 340 int ret;
341 341
@@ -354,26 +354,29 @@ int futex_atomic_op_pt(int op, int __user *uaddr, int oparg, int *old)
354 return ret; 354 return ret;
355} 355}
356 356
357static int __futex_atomic_cmpxchg_pt(int __user *uaddr, int oldval, int newval) 357static int __futex_atomic_cmpxchg_pt(u32 *uval, u32 __user *uaddr,
358 u32 oldval, u32 newval)
358{ 359{
359 int ret; 360 int ret;
360 361
361 asm volatile("0: cs %1,%4,0(%5)\n" 362 asm volatile("0: cs %1,%4,0(%5)\n"
362 "1: lr %0,%1\n" 363 "1: la %0,0\n"
363 "2:\n" 364 "2:\n"
364 EX_TABLE(0b,2b) EX_TABLE(1b,2b) 365 EX_TABLE(0b,2b) EX_TABLE(1b,2b)
365 : "=d" (ret), "+d" (oldval), "=m" (*uaddr) 366 : "=d" (ret), "+d" (oldval), "=m" (*uaddr)
366 : "0" (-EFAULT), "d" (newval), "a" (uaddr), "m" (*uaddr) 367 : "0" (-EFAULT), "d" (newval), "a" (uaddr), "m" (*uaddr)
367 : "cc", "memory" ); 368 : "cc", "memory" );
369 *uval = oldval;
368 return ret; 370 return ret;
369} 371}
370 372
371int futex_atomic_cmpxchg_pt(int __user *uaddr, int oldval, int newval) 373int futex_atomic_cmpxchg_pt(u32 *uval, u32 __user *uaddr,
374 u32 oldval, u32 newval)
372{ 375{
373 int ret; 376 int ret;
374 377
375 if (segment_eq(get_fs(), KERNEL_DS)) 378 if (segment_eq(get_fs(), KERNEL_DS))
376 return __futex_atomic_cmpxchg_pt(uaddr, oldval, newval); 379 return __futex_atomic_cmpxchg_pt(uval, uaddr, oldval, newval);
377 spin_lock(&current->mm->page_table_lock); 380 spin_lock(&current->mm->page_table_lock);
378 uaddr = (int __user *) __dat_user_addr((unsigned long) uaddr); 381 uaddr = (int __user *) __dat_user_addr((unsigned long) uaddr);
379 if (!uaddr) { 382 if (!uaddr) {
@@ -382,7 +385,7 @@ int futex_atomic_cmpxchg_pt(int __user *uaddr, int oldval, int newval)
382 } 385 }
383 get_page(virt_to_page(uaddr)); 386 get_page(virt_to_page(uaddr));
384 spin_unlock(&current->mm->page_table_lock); 387 spin_unlock(&current->mm->page_table_lock);
385 ret = __futex_atomic_cmpxchg_pt(uaddr, oldval, newval); 388 ret = __futex_atomic_cmpxchg_pt(uval, uaddr, oldval, newval);
386 put_page(virt_to_page(uaddr)); 389 put_page(virt_to_page(uaddr));
387 return ret; 390 return ret;
388} 391}
diff --git a/arch/s390/lib/uaccess_std.c b/arch/s390/lib/uaccess_std.c
index 07deaeee14c8..bb1a7eed42ce 100644
--- a/arch/s390/lib/uaccess_std.c
+++ b/arch/s390/lib/uaccess_std.c
@@ -125,9 +125,9 @@ static size_t copy_in_user_std(size_t size, void __user *to,
125 unsigned long tmp1; 125 unsigned long tmp1;
126 126
127 asm volatile( 127 asm volatile(
128 " sacf 256\n"
128 " "AHI" %0,-1\n" 129 " "AHI" %0,-1\n"
129 " jo 5f\n" 130 " jo 5f\n"
130 " sacf 256\n"
131 " bras %3,3f\n" 131 " bras %3,3f\n"
132 "0:"AHI" %0,257\n" 132 "0:"AHI" %0,257\n"
133 "1: mvc 0(1,%1),0(%2)\n" 133 "1: mvc 0(1,%1),0(%2)\n"
@@ -142,9 +142,8 @@ static size_t copy_in_user_std(size_t size, void __user *to,
142 "3:"AHI" %0,-256\n" 142 "3:"AHI" %0,-256\n"
143 " jnm 2b\n" 143 " jnm 2b\n"
144 "4: ex %0,1b-0b(%3)\n" 144 "4: ex %0,1b-0b(%3)\n"
145 " sacf 0\n"
146 "5: "SLR" %0,%0\n" 145 "5: "SLR" %0,%0\n"
147 "6:\n" 146 "6: sacf 0\n"
148 EX_TABLE(1b,6b) EX_TABLE(2b,0b) EX_TABLE(4b,0b) 147 EX_TABLE(1b,6b) EX_TABLE(2b,0b) EX_TABLE(4b,0b)
149 : "+a" (size), "+a" (to), "+a" (from), "=a" (tmp1) 148 : "+a" (size), "+a" (to), "+a" (from), "=a" (tmp1)
150 : : "cc", "memory"); 149 : : "cc", "memory");
@@ -156,9 +155,9 @@ static size_t clear_user_std(size_t size, void __user *to)
156 unsigned long tmp1, tmp2; 155 unsigned long tmp1, tmp2;
157 156
158 asm volatile( 157 asm volatile(
158 " sacf 256\n"
159 " "AHI" %0,-1\n" 159 " "AHI" %0,-1\n"
160 " jo 5f\n" 160 " jo 5f\n"
161 " sacf 256\n"
162 " bras %3,3f\n" 161 " bras %3,3f\n"
163 " xc 0(1,%1),0(%1)\n" 162 " xc 0(1,%1),0(%1)\n"
164 "0:"AHI" %0,257\n" 163 "0:"AHI" %0,257\n"
@@ -178,9 +177,8 @@ static size_t clear_user_std(size_t size, void __user *to)
178 "3:"AHI" %0,-256\n" 177 "3:"AHI" %0,-256\n"
179 " jnm 2b\n" 178 " jnm 2b\n"
180 "4: ex %0,0(%3)\n" 179 "4: ex %0,0(%3)\n"
181 " sacf 0\n"
182 "5: "SLR" %0,%0\n" 180 "5: "SLR" %0,%0\n"
183 "6:\n" 181 "6: sacf 0\n"
184 EX_TABLE(1b,6b) EX_TABLE(2b,0b) EX_TABLE(4b,0b) 182 EX_TABLE(1b,6b) EX_TABLE(2b,0b) EX_TABLE(4b,0b)
185 : "+a" (size), "+a" (to), "=a" (tmp1), "=a" (tmp2) 183 : "+a" (size), "+a" (to), "=a" (tmp1), "=a" (tmp2)
186 : : "cc", "memory"); 184 : : "cc", "memory");
@@ -257,7 +255,7 @@ size_t strncpy_from_user_std(size_t size, const char __user *src, char *dst)
257 : "0" (-EFAULT), "d" (oparg), "a" (uaddr), \ 255 : "0" (-EFAULT), "d" (oparg), "a" (uaddr), \
258 "m" (*uaddr) : "cc"); 256 "m" (*uaddr) : "cc");
259 257
260int futex_atomic_op_std(int op, int __user *uaddr, int oparg, int *old) 258int futex_atomic_op_std(int op, u32 __user *uaddr, int oparg, int *old)
261{ 259{
262 int oldval = 0, newval, ret; 260 int oldval = 0, newval, ret;
263 261
@@ -289,19 +287,21 @@ int futex_atomic_op_std(int op, int __user *uaddr, int oparg, int *old)
289 return ret; 287 return ret;
290} 288}
291 289
292int futex_atomic_cmpxchg_std(int __user *uaddr, int oldval, int newval) 290int futex_atomic_cmpxchg_std(u32 *uval, u32 __user *uaddr,
291 u32 oldval, u32 newval)
293{ 292{
294 int ret; 293 int ret;
295 294
296 asm volatile( 295 asm volatile(
297 " sacf 256\n" 296 " sacf 256\n"
298 "0: cs %1,%4,0(%5)\n" 297 "0: cs %1,%4,0(%5)\n"
299 "1: lr %0,%1\n" 298 "1: la %0,0\n"
300 "2: sacf 0\n" 299 "2: sacf 0\n"
301 EX_TABLE(0b,2b) EX_TABLE(1b,2b) 300 EX_TABLE(0b,2b) EX_TABLE(1b,2b)
302 : "=d" (ret), "+d" (oldval), "=m" (*uaddr) 301 : "=d" (ret), "+d" (oldval), "=m" (*uaddr)
303 : "0" (-EFAULT), "d" (newval), "a" (uaddr), "m" (*uaddr) 302 : "0" (-EFAULT), "d" (newval), "a" (uaddr), "m" (*uaddr)
304 : "cc", "memory" ); 303 : "cc", "memory" );
304 *uval = oldval;
305 return ret; 305 return ret;
306} 306}
307 307
diff --git a/arch/s390/math-emu/Makefile b/arch/s390/math-emu/Makefile
index c84890341052..51d399549f60 100644
--- a/arch/s390/math-emu/Makefile
+++ b/arch/s390/math-emu/Makefile
@@ -4,4 +4,4 @@
4 4
5obj-$(CONFIG_MATHEMU) := math.o 5obj-$(CONFIG_MATHEMU) := math.o
6 6
7EXTRA_CFLAGS := -I$(src) -Iinclude/math-emu -w 7ccflags-y := -I$(src) -Iinclude/math-emu -w
diff --git a/arch/s390/mm/Makefile b/arch/s390/mm/Makefile
index eec054484419..d98fe9004a52 100644
--- a/arch/s390/mm/Makefile
+++ b/arch/s390/mm/Makefile
@@ -3,6 +3,7 @@
3# 3#
4 4
5obj-y := init.o fault.o extmem.o mmap.o vmem.o pgtable.o maccess.o \ 5obj-y := init.o fault.o extmem.o mmap.o vmem.o pgtable.o maccess.o \
6 page-states.o 6 page-states.o gup.o
7obj-$(CONFIG_CMM) += cmm.o 7obj-$(CONFIG_CMM) += cmm.o
8obj-$(CONFIG_HUGETLB_PAGE) += hugetlbpage.o 8obj-$(CONFIG_HUGETLB_PAGE) += hugetlbpage.o
9obj-$(CONFIG_DEBUG_SET_MODULE_RONX) += pageattr.o
diff --git a/arch/s390/mm/cmm.c b/arch/s390/mm/cmm.c
index a9550dca3e4b..1f1dba9dcf58 100644
--- a/arch/s390/mm/cmm.c
+++ b/arch/s390/mm/cmm.c
@@ -23,7 +23,10 @@
23#include <asm/pgalloc.h> 23#include <asm/pgalloc.h>
24#include <asm/diag.h> 24#include <asm/diag.h>
25 25
26static char *sender = "VMRMSVM"; 26#ifdef CONFIG_CMM_IUCV
27static char *cmm_default_sender = "VMRMSVM";
28#endif
29static char *sender;
27module_param(sender, charp, 0400); 30module_param(sender, charp, 0400);
28MODULE_PARM_DESC(sender, 31MODULE_PARM_DESC(sender,
29 "Guest name that may send SMSG messages (default VMRMSVM)"); 32 "Guest name that may send SMSG messages (default VMRMSVM)");
@@ -88,7 +91,7 @@ static long cmm_alloc_pages(long nr, long *counter,
88 } else 91 } else
89 free_page((unsigned long) npa); 92 free_page((unsigned long) npa);
90 } 93 }
91 diag10(addr); 94 diag10_range(addr >> PAGE_SHIFT, 1);
92 pa->pages[pa->index++] = addr; 95 pa->pages[pa->index++] = addr;
93 (*counter)++; 96 (*counter)++;
94 spin_unlock(&cmm_lock); 97 spin_unlock(&cmm_lock);
@@ -440,6 +443,8 @@ static int __init cmm_init(void)
440 int len = strlen(sender); 443 int len = strlen(sender);
441 while (len--) 444 while (len--)
442 sender[len] = toupper(sender[len]); 445 sender[len] = toupper(sender[len]);
446 } else {
447 sender = cmm_default_sender;
443 } 448 }
444 449
445 rc = smsg_register_callback(SMSG_PREFIX, cmm_smsg_target); 450 rc = smsg_register_callback(SMSG_PREFIX, cmm_smsg_target);
diff --git a/arch/s390/mm/extmem.c b/arch/s390/mm/extmem.c
index 3cc95dd0a3a6..075ddada4911 100644
--- a/arch/s390/mm/extmem.c
+++ b/arch/s390/mm/extmem.c
@@ -412,6 +412,7 @@ __segment_load (char *name, int do_nonshared, unsigned long *addr, unsigned long
412 struct dcss_segment *seg; 412 struct dcss_segment *seg;
413 int rc, diag_cc; 413 int rc, diag_cc;
414 414
415 start_addr = end_addr = 0;
415 seg = kmalloc(sizeof(*seg), GFP_KERNEL | GFP_DMA); 416 seg = kmalloc(sizeof(*seg), GFP_KERNEL | GFP_DMA);
416 if (seg == NULL) { 417 if (seg == NULL) {
417 rc = -ENOMEM; 418 rc = -ENOMEM;
@@ -573,6 +574,7 @@ segment_modify_shared (char *name, int do_nonshared)
573 unsigned long start_addr, end_addr, dummy; 574 unsigned long start_addr, end_addr, dummy;
574 int rc, diag_cc; 575 int rc, diag_cc;
575 576
577 start_addr = end_addr = 0;
576 mutex_lock(&dcss_lock); 578 mutex_lock(&dcss_lock);
577 seg = segment_by_name (name); 579 seg = segment_by_name (name);
578 if (seg == NULL) { 580 if (seg == NULL) {
@@ -681,8 +683,6 @@ void
681segment_save(char *name) 683segment_save(char *name)
682{ 684{
683 struct dcss_segment *seg; 685 struct dcss_segment *seg;
684 int startpfn = 0;
685 int endpfn = 0;
686 char cmd1[160]; 686 char cmd1[160];
687 char cmd2[80]; 687 char cmd2[80];
688 int i, response; 688 int i, response;
@@ -698,8 +698,6 @@ segment_save(char *name)
698 goto out; 698 goto out;
699 } 699 }
700 700
701 startpfn = seg->start_addr >> PAGE_SHIFT;
702 endpfn = (seg->end) >> PAGE_SHIFT;
703 sprintf(cmd1, "DEFSEG %s", name); 701 sprintf(cmd1, "DEFSEG %s", name);
704 for (i=0; i<seg->segcnt; i++) { 702 for (i=0; i<seg->segcnt; i++) {
705 sprintf(cmd1+strlen(cmd1), " %lX-%lX %s", 703 sprintf(cmd1+strlen(cmd1), " %lX-%lX %s",
diff --git a/arch/s390/mm/fault.c b/arch/s390/mm/fault.c
index 2505b2ea0ef1..fe103e891e7a 100644
--- a/arch/s390/mm/fault.c
+++ b/arch/s390/mm/fault.c
@@ -10,6 +10,7 @@
10 * Copyright (C) 1995 Linus Torvalds 10 * Copyright (C) 1995 Linus Torvalds
11 */ 11 */
12 12
13#include <linux/kernel_stat.h>
13#include <linux/perf_event.h> 14#include <linux/perf_event.h>
14#include <linux/signal.h> 15#include <linux/signal.h>
15#include <linux/sched.h> 16#include <linux/sched.h>
@@ -33,7 +34,7 @@
33#include <asm/asm-offsets.h> 34#include <asm/asm-offsets.h>
34#include <asm/system.h> 35#include <asm/system.h>
35#include <asm/pgtable.h> 36#include <asm/pgtable.h>
36#include <asm/s390_ext.h> 37#include <asm/irq.h>
37#include <asm/mmu_context.h> 38#include <asm/mmu_context.h>
38#include <asm/compat.h> 39#include <asm/compat.h>
39#include "../kernel/entry.h" 40#include "../kernel/entry.h"
@@ -52,6 +53,14 @@
52#define VM_FAULT_BADMAP 0x020000 53#define VM_FAULT_BADMAP 0x020000
53#define VM_FAULT_BADACCESS 0x040000 54#define VM_FAULT_BADACCESS 0x040000
54 55
56static unsigned long store_indication;
57
58void fault_init(void)
59{
60 if (test_facility(2) && test_facility(75))
61 store_indication = 0xc00;
62}
63
55static inline int notify_page_fault(struct pt_regs *regs) 64static inline int notify_page_fault(struct pt_regs *regs)
56{ 65{
57 int ret = 0; 66 int ret = 0;
@@ -199,42 +208,22 @@ static noinline void do_sigbus(struct pt_regs *regs, long int_code,
199 unsigned long trans_exc_code) 208 unsigned long trans_exc_code)
200{ 209{
201 struct task_struct *tsk = current; 210 struct task_struct *tsk = current;
211 unsigned long address;
212 struct siginfo si;
202 213
203 /* 214 /*
204 * Send a sigbus, regardless of whether we were in kernel 215 * Send a sigbus, regardless of whether we were in kernel
205 * or user mode. 216 * or user mode.
206 */ 217 */
207 tsk->thread.prot_addr = trans_exc_code & __FAIL_ADDR_MASK; 218 address = trans_exc_code & __FAIL_ADDR_MASK;
219 tsk->thread.prot_addr = address;
208 tsk->thread.trap_no = int_code; 220 tsk->thread.trap_no = int_code;
209 force_sig(SIGBUS, tsk); 221 si.si_signo = SIGBUS;
210} 222 si.si_errno = 0;
211 223 si.si_code = BUS_ADRERR;
212#ifdef CONFIG_S390_EXEC_PROTECT 224 si.si_addr = (void __user *) address;
213static noinline int signal_return(struct pt_regs *regs, long int_code, 225 force_sig_info(SIGBUS, &si, tsk);
214 unsigned long trans_exc_code)
215{
216 u16 instruction;
217 int rc;
218
219 rc = __get_user(instruction, (u16 __user *) regs->psw.addr);
220
221 if (!rc && instruction == 0x0a77) {
222 clear_tsk_thread_flag(current, TIF_SINGLE_STEP);
223 if (is_compat_task())
224 sys32_sigreturn();
225 else
226 sys_sigreturn();
227 } else if (!rc && instruction == 0x0aad) {
228 clear_tsk_thread_flag(current, TIF_SINGLE_STEP);
229 if (is_compat_task())
230 sys32_rt_sigreturn();
231 else
232 sys_rt_sigreturn();
233 } else
234 do_sigsegv(regs, int_code, SEGV_MAPERR, trans_exc_code);
235 return 0;
236} 226}
237#endif /* CONFIG_S390_EXEC_PROTECT */
238 227
239static noinline void do_fault_error(struct pt_regs *regs, long int_code, 228static noinline void do_fault_error(struct pt_regs *regs, long int_code,
240 unsigned long trans_exc_code, int fault) 229 unsigned long trans_exc_code, int fault)
@@ -243,13 +232,6 @@ static noinline void do_fault_error(struct pt_regs *regs, long int_code,
243 232
244 switch (fault) { 233 switch (fault) {
245 case VM_FAULT_BADACCESS: 234 case VM_FAULT_BADACCESS:
246#ifdef CONFIG_S390_EXEC_PROTECT
247 if ((regs->psw.mask & PSW_MASK_ASC) == PSW_ASC_SECONDARY &&
248 (trans_exc_code & 3) == 0) {
249 signal_return(regs, int_code, trans_exc_code);
250 break;
251 }
252#endif /* CONFIG_S390_EXEC_PROTECT */
253 case VM_FAULT_BADMAP: 235 case VM_FAULT_BADMAP:
254 /* Bad memory access. Check if it is kernel or user space. */ 236 /* Bad memory access. Check if it is kernel or user space. */
255 if (regs->psw.mask & PSW_MASK_PSTATE) { 237 if (regs->psw.mask & PSW_MASK_PSTATE) {
@@ -263,13 +245,17 @@ static noinline void do_fault_error(struct pt_regs *regs, long int_code,
263 do_no_context(regs, int_code, trans_exc_code); 245 do_no_context(regs, int_code, trans_exc_code);
264 break; 246 break;
265 default: /* fault & VM_FAULT_ERROR */ 247 default: /* fault & VM_FAULT_ERROR */
266 if (fault & VM_FAULT_OOM) 248 if (fault & VM_FAULT_OOM) {
267 pagefault_out_of_memory(); 249 if (!(regs->psw.mask & PSW_MASK_PSTATE))
268 else if (fault & VM_FAULT_SIGBUS) { 250 do_no_context(regs, int_code, trans_exc_code);
269 do_sigbus(regs, int_code, trans_exc_code); 251 else
252 pagefault_out_of_memory();
253 } else if (fault & VM_FAULT_SIGBUS) {
270 /* Kernel mode? Handle exceptions or die */ 254 /* Kernel mode? Handle exceptions or die */
271 if (!(regs->psw.mask & PSW_MASK_PSTATE)) 255 if (!(regs->psw.mask & PSW_MASK_PSTATE))
272 do_no_context(regs, int_code, trans_exc_code); 256 do_no_context(regs, int_code, trans_exc_code);
257 else
258 do_sigbus(regs, int_code, trans_exc_code);
273 } else 259 } else
274 BUG(); 260 BUG();
275 break; 261 break;
@@ -294,6 +280,7 @@ static inline int do_exception(struct pt_regs *regs, int access,
294 struct mm_struct *mm; 280 struct mm_struct *mm;
295 struct vm_area_struct *vma; 281 struct vm_area_struct *vma;
296 unsigned long address; 282 unsigned long address;
283 unsigned int flags;
297 int fault; 284 int fault;
298 285
299 if (notify_page_fault(regs)) 286 if (notify_page_fault(regs))
@@ -312,13 +299,11 @@ static inline int do_exception(struct pt_regs *regs, int access,
312 goto out; 299 goto out;
313 300
314 address = trans_exc_code & __FAIL_ADDR_MASK; 301 address = trans_exc_code & __FAIL_ADDR_MASK;
315 /*
316 * When we get here, the fault happened in the current
317 * task's user address space, so we can switch on the
318 * interrupts again and then search the VMAs
319 */
320 local_irq_enable();
321 perf_sw_event(PERF_COUNT_SW_PAGE_FAULTS, 1, 0, regs, address); 302 perf_sw_event(PERF_COUNT_SW_PAGE_FAULTS, 1, 0, regs, address);
303 flags = FAULT_FLAG_ALLOW_RETRY;
304 if (access == VM_WRITE || (trans_exc_code & store_indication) == 0x400)
305 flags |= FAULT_FLAG_WRITE;
306retry:
322 down_read(&mm->mmap_sem); 307 down_read(&mm->mmap_sem);
323 308
324 fault = VM_FAULT_BADMAP; 309 fault = VM_FAULT_BADMAP;
@@ -348,25 +333,37 @@ static inline int do_exception(struct pt_regs *regs, int access,
348 * make sure we exit gracefully rather than endlessly redo 333 * make sure we exit gracefully rather than endlessly redo
349 * the fault. 334 * the fault.
350 */ 335 */
351 fault = handle_mm_fault(mm, vma, address, 336 fault = handle_mm_fault(mm, vma, address, flags);
352 (access == VM_WRITE) ? FAULT_FLAG_WRITE : 0);
353 if (unlikely(fault & VM_FAULT_ERROR)) 337 if (unlikely(fault & VM_FAULT_ERROR))
354 goto out_up; 338 goto out_up;
355 339
356 if (fault & VM_FAULT_MAJOR) { 340 /*
357 tsk->maj_flt++; 341 * Major/minor page fault accounting is only done on the
358 perf_sw_event(PERF_COUNT_SW_PAGE_FAULTS_MAJ, 1, 0, 342 * initial attempt. If we go through a retry, it is extremely
359 regs, address); 343 * likely that the page will be found in page cache at that point.
360 } else { 344 */
361 tsk->min_flt++; 345 if (flags & FAULT_FLAG_ALLOW_RETRY) {
362 perf_sw_event(PERF_COUNT_SW_PAGE_FAULTS_MIN, 1, 0, 346 if (fault & VM_FAULT_MAJOR) {
363 regs, address); 347 tsk->maj_flt++;
348 perf_sw_event(PERF_COUNT_SW_PAGE_FAULTS_MAJ, 1, 0,
349 regs, address);
350 } else {
351 tsk->min_flt++;
352 perf_sw_event(PERF_COUNT_SW_PAGE_FAULTS_MIN, 1, 0,
353 regs, address);
354 }
355 if (fault & VM_FAULT_RETRY) {
356 /* Clear FAULT_FLAG_ALLOW_RETRY to avoid any risk
357 * of starvation. */
358 flags &= ~FAULT_FLAG_ALLOW_RETRY;
359 goto retry;
360 }
364 } 361 }
365 /* 362 /*
366 * The instruction that caused the program check will 363 * The instruction that caused the program check will
367 * be repeated. Don't signal single step via SIGTRAP. 364 * be repeated. Don't signal single step via SIGTRAP.
368 */ 365 */
369 clear_tsk_thread_flag(tsk, TIF_SINGLE_STEP); 366 clear_tsk_thread_flag(tsk, TIF_PER_TRAP);
370 fault = 0; 367 fault = 0;
371out_up: 368out_up:
372 up_read(&mm->mmap_sem); 369 up_read(&mm->mmap_sem);
@@ -374,20 +371,20 @@ out:
374 return fault; 371 return fault;
375} 372}
376 373
377void __kprobes do_protection_exception(struct pt_regs *regs, long int_code) 374void __kprobes do_protection_exception(struct pt_regs *regs, long pgm_int_code,
375 unsigned long trans_exc_code)
378{ 376{
379 unsigned long trans_exc_code = S390_lowcore.trans_exc_code;
380 int fault; 377 int fault;
381 378
382 /* Protection exception is supressing, decrement psw address. */ 379 /* Protection exception is suppressing, decrement psw address. */
383 regs->psw.addr -= (int_code >> 16); 380 regs->psw.addr -= (pgm_int_code >> 16);
384 /* 381 /*
385 * Check for low-address protection. This needs to be treated 382 * Check for low-address protection. This needs to be treated
386 * as a special case because the translation exception code 383 * as a special case because the translation exception code
387 * field is not guaranteed to contain valid data in this case. 384 * field is not guaranteed to contain valid data in this case.
388 */ 385 */
389 if (unlikely(!(trans_exc_code & 4))) { 386 if (unlikely(!(trans_exc_code & 4))) {
390 do_low_address(regs, int_code, trans_exc_code); 387 do_low_address(regs, pgm_int_code, trans_exc_code);
391 return; 388 return;
392 } 389 }
393 fault = do_exception(regs, VM_WRITE, trans_exc_code); 390 fault = do_exception(regs, VM_WRITE, trans_exc_code);
@@ -395,34 +392,27 @@ void __kprobes do_protection_exception(struct pt_regs *regs, long int_code)
395 do_fault_error(regs, 4, trans_exc_code, fault); 392 do_fault_error(regs, 4, trans_exc_code, fault);
396} 393}
397 394
398void __kprobes do_dat_exception(struct pt_regs *regs, long int_code) 395void __kprobes do_dat_exception(struct pt_regs *regs, long pgm_int_code,
396 unsigned long trans_exc_code)
399{ 397{
400 unsigned long trans_exc_code = S390_lowcore.trans_exc_code;
401 int access, fault; 398 int access, fault;
402 399
403 access = VM_READ | VM_EXEC | VM_WRITE; 400 access = VM_READ | VM_EXEC | VM_WRITE;
404#ifdef CONFIG_S390_EXEC_PROTECT
405 if ((regs->psw.mask & PSW_MASK_ASC) == PSW_ASC_SECONDARY &&
406 (trans_exc_code & 3) == 0)
407 access = VM_EXEC;
408#endif
409 fault = do_exception(regs, access, trans_exc_code); 401 fault = do_exception(regs, access, trans_exc_code);
410 if (unlikely(fault)) 402 if (unlikely(fault))
411 do_fault_error(regs, int_code & 255, trans_exc_code, fault); 403 do_fault_error(regs, pgm_int_code & 255, trans_exc_code, fault);
412} 404}
413 405
414#ifdef CONFIG_64BIT 406#ifdef CONFIG_64BIT
415void __kprobes do_asce_exception(struct pt_regs *regs, long int_code) 407void __kprobes do_asce_exception(struct pt_regs *regs, long pgm_int_code,
408 unsigned long trans_exc_code)
416{ 409{
417 unsigned long trans_exc_code = S390_lowcore.trans_exc_code;
418 struct mm_struct *mm = current->mm; 410 struct mm_struct *mm = current->mm;
419 struct vm_area_struct *vma; 411 struct vm_area_struct *vma;
420 412
421 if (unlikely(!user_space_fault(trans_exc_code) || in_atomic() || !mm)) 413 if (unlikely(!user_space_fault(trans_exc_code) || in_atomic() || !mm))
422 goto no_context; 414 goto no_context;
423 415
424 local_irq_enable();
425
426 down_read(&mm->mmap_sem); 416 down_read(&mm->mmap_sem);
427 vma = find_vma(mm, trans_exc_code & __FAIL_ADDR_MASK); 417 vma = find_vma(mm, trans_exc_code & __FAIL_ADDR_MASK);
428 up_read(&mm->mmap_sem); 418 up_read(&mm->mmap_sem);
@@ -434,16 +424,16 @@ void __kprobes do_asce_exception(struct pt_regs *regs, long int_code)
434 424
435 /* User mode accesses just cause a SIGSEGV */ 425 /* User mode accesses just cause a SIGSEGV */
436 if (regs->psw.mask & PSW_MASK_PSTATE) { 426 if (regs->psw.mask & PSW_MASK_PSTATE) {
437 do_sigsegv(regs, int_code, SEGV_MAPERR, trans_exc_code); 427 do_sigsegv(regs, pgm_int_code, SEGV_MAPERR, trans_exc_code);
438 return; 428 return;
439 } 429 }
440 430
441no_context: 431no_context:
442 do_no_context(regs, int_code, trans_exc_code); 432 do_no_context(regs, pgm_int_code, trans_exc_code);
443} 433}
444#endif 434#endif
445 435
446int __handle_fault(unsigned long uaddr, unsigned long int_code, int write_user) 436int __handle_fault(unsigned long uaddr, unsigned long pgm_int_code, int write)
447{ 437{
448 struct pt_regs regs; 438 struct pt_regs regs;
449 int access, fault; 439 int access, fault;
@@ -454,14 +444,13 @@ int __handle_fault(unsigned long uaddr, unsigned long int_code, int write_user)
454 regs.psw.addr = (unsigned long) __builtin_return_address(0); 444 regs.psw.addr = (unsigned long) __builtin_return_address(0);
455 regs.psw.addr |= PSW_ADDR_AMODE; 445 regs.psw.addr |= PSW_ADDR_AMODE;
456 uaddr &= PAGE_MASK; 446 uaddr &= PAGE_MASK;
457 access = write_user ? VM_WRITE : VM_READ; 447 access = write ? VM_WRITE : VM_READ;
458 fault = do_exception(&regs, access, uaddr | 2); 448 fault = do_exception(&regs, access, uaddr | 2);
459 if (unlikely(fault)) { 449 if (unlikely(fault)) {
460 if (fault & VM_FAULT_OOM) { 450 if (fault & VM_FAULT_OOM)
461 pagefault_out_of_memory(); 451 return -EFAULT;
462 fault = 0; 452 else if (fault & VM_FAULT_SIGBUS)
463 } else if (fault & VM_FAULT_SIGBUS) 453 do_sigbus(&regs, pgm_int_code, uaddr);
464 do_sigbus(&regs, int_code, uaddr);
465 } 454 }
466 return fault ? -EFAULT : 0; 455 return fault ? -EFAULT : 0;
467} 456}
@@ -470,8 +459,7 @@ int __handle_fault(unsigned long uaddr, unsigned long int_code, int write_user)
470/* 459/*
471 * 'pfault' pseudo page faults routines. 460 * 'pfault' pseudo page faults routines.
472 */ 461 */
473static ext_int_info_t ext_int_pfault; 462static int pfault_disable;
474static int pfault_disable = 0;
475 463
476static int __init nopfault(char *str) 464static int __init nopfault(char *str)
477{ 465{
@@ -481,22 +469,28 @@ static int __init nopfault(char *str)
481 469
482__setup("nopfault", nopfault); 470__setup("nopfault", nopfault);
483 471
484typedef struct { 472struct pfault_refbk {
485 __u16 refdiagc; 473 u16 refdiagc;
486 __u16 reffcode; 474 u16 reffcode;
487 __u16 refdwlen; 475 u16 refdwlen;
488 __u16 refversn; 476 u16 refversn;
489 __u64 refgaddr; 477 u64 refgaddr;
490 __u64 refselmk; 478 u64 refselmk;
491 __u64 refcmpmk; 479 u64 refcmpmk;
492 __u64 reserved; 480 u64 reserved;
493} __attribute__ ((packed, aligned(8))) pfault_refbk_t; 481} __attribute__ ((packed, aligned(8)));
494 482
495int pfault_init(void) 483int pfault_init(void)
496{ 484{
497 pfault_refbk_t refbk = 485 struct pfault_refbk refbk = {
498 { 0x258, 0, 5, 2, __LC_CURRENT, 1ULL << 48, 1ULL << 48, 486 .refdiagc = 0x258,
499 __PF_RES_FIELD }; 487 .reffcode = 0,
488 .refdwlen = 5,
489 .refversn = 2,
490 .refgaddr = __LC_CURRENT_PID,
491 .refselmk = 1ULL << 48,
492 .refcmpmk = 1ULL << 48,
493 .reserved = __PF_RES_FIELD };
500 int rc; 494 int rc;
501 495
502 if (!MACHINE_IS_VM || pfault_disable) 496 if (!MACHINE_IS_VM || pfault_disable)
@@ -508,18 +502,20 @@ int pfault_init(void)
508 "2:\n" 502 "2:\n"
509 EX_TABLE(0b,1b) 503 EX_TABLE(0b,1b)
510 : "=d" (rc) : "a" (&refbk), "m" (refbk) : "cc"); 504 : "=d" (rc) : "a" (&refbk), "m" (refbk) : "cc");
511 __ctl_set_bit(0, 9);
512 return rc; 505 return rc;
513} 506}
514 507
515void pfault_fini(void) 508void pfault_fini(void)
516{ 509{
517 pfault_refbk_t refbk = 510 struct pfault_refbk refbk = {
518 { 0x258, 1, 5, 2, 0ULL, 0ULL, 0ULL, 0ULL }; 511 .refdiagc = 0x258,
512 .reffcode = 1,
513 .refdwlen = 5,
514 .refversn = 2,
515 };
519 516
520 if (!MACHINE_IS_VM || pfault_disable) 517 if (!MACHINE_IS_VM || pfault_disable)
521 return; 518 return;
522 __ctl_clear_bit(0,9);
523 asm volatile( 519 asm volatile(
524 " diag %0,0,0x258\n" 520 " diag %0,0,0x258\n"
525 "0:\n" 521 "0:\n"
@@ -527,10 +523,15 @@ void pfault_fini(void)
527 : : "a" (&refbk), "m" (refbk) : "cc"); 523 : : "a" (&refbk), "m" (refbk) : "cc");
528} 524}
529 525
530static void pfault_interrupt(__u16 int_code) 526static DEFINE_SPINLOCK(pfault_lock);
527static LIST_HEAD(pfault_list);
528
529static void pfault_interrupt(unsigned int ext_int_code,
530 unsigned int param32, unsigned long param64)
531{ 531{
532 struct task_struct *tsk; 532 struct task_struct *tsk;
533 __u16 subcode; 533 __u16 subcode;
534 pid_t pid;
534 535
535 /* 536 /*
536 * Get the external interruption subcode & pfault 537 * Get the external interruption subcode & pfault
@@ -538,63 +539,107 @@ static void pfault_interrupt(__u16 int_code)
538 * in the 'cpu address' field associated with the 539 * in the 'cpu address' field associated with the
539 * external interrupt. 540 * external interrupt.
540 */ 541 */
541 subcode = S390_lowcore.cpu_addr; 542 subcode = ext_int_code >> 16;
542 if ((subcode & 0xff00) != __SUBCODE_MASK) 543 if ((subcode & 0xff00) != __SUBCODE_MASK)
543 return; 544 return;
544 545 kstat_cpu(smp_processor_id()).irqs[EXTINT_PFL]++;
545 /* 546 if (subcode & 0x0080) {
546 * Get the token (= address of the task structure of the affected task). 547 /* Get the token (= pid of the affected task). */
547 */ 548 pid = sizeof(void *) == 4 ? param32 : param64;
548 tsk = *(struct task_struct **) __LC_PFAULT_INTPARM; 549 rcu_read_lock();
549 550 tsk = find_task_by_pid_ns(pid, &init_pid_ns);
551 if (tsk)
552 get_task_struct(tsk);
553 rcu_read_unlock();
554 if (!tsk)
555 return;
556 } else {
557 tsk = current;
558 }
559 spin_lock(&pfault_lock);
550 if (subcode & 0x0080) { 560 if (subcode & 0x0080) {
551 /* signal bit is set -> a page has been swapped in by VM */ 561 /* signal bit is set -> a page has been swapped in by VM */
552 if (xchg(&tsk->thread.pfault_wait, -1) != 0) { 562 if (tsk->thread.pfault_wait == 1) {
553 /* Initial interrupt was faster than the completion 563 /* Initial interrupt was faster than the completion
554 * interrupt. pfault_wait is valid. Set pfault_wait 564 * interrupt. pfault_wait is valid. Set pfault_wait
555 * back to zero and wake up the process. This can 565 * back to zero and wake up the process. This can
556 * safely be done because the task is still sleeping 566 * safely be done because the task is still sleeping
557 * and can't produce new pfaults. */ 567 * and can't produce new pfaults. */
558 tsk->thread.pfault_wait = 0; 568 tsk->thread.pfault_wait = 0;
569 list_del(&tsk->thread.list);
559 wake_up_process(tsk); 570 wake_up_process(tsk);
560 put_task_struct(tsk); 571 } else {
572 /* Completion interrupt was faster than initial
573 * interrupt. Set pfault_wait to -1 so the initial
574 * interrupt doesn't put the task to sleep. */
575 tsk->thread.pfault_wait = -1;
561 } 576 }
577 put_task_struct(tsk);
562 } else { 578 } else {
563 /* signal bit not set -> a real page is missing. */ 579 /* signal bit not set -> a real page is missing. */
564 get_task_struct(tsk); 580 if (tsk->thread.pfault_wait == -1) {
565 set_task_state(tsk, TASK_UNINTERRUPTIBLE);
566 if (xchg(&tsk->thread.pfault_wait, 1) != 0) {
567 /* Completion interrupt was faster than the initial 581 /* Completion interrupt was faster than the initial
568 * interrupt (swapped in a -1 for pfault_wait). Set 582 * interrupt (pfault_wait == -1). Set pfault_wait
569 * pfault_wait back to zero and exit. This can be 583 * back to zero and exit. */
570 * done safely because tsk is running in kernel
571 * mode and can't produce new pfaults. */
572 tsk->thread.pfault_wait = 0; 584 tsk->thread.pfault_wait = 0;
573 set_task_state(tsk, TASK_RUNNING); 585 } else {
574 put_task_struct(tsk); 586 /* Initial interrupt arrived before completion
575 } else 587 * interrupt. Let the task sleep. */
588 tsk->thread.pfault_wait = 1;
589 list_add(&tsk->thread.list, &pfault_list);
590 set_task_state(tsk, TASK_UNINTERRUPTIBLE);
576 set_tsk_need_resched(tsk); 591 set_tsk_need_resched(tsk);
592 }
577 } 593 }
594 spin_unlock(&pfault_lock);
578} 595}
579 596
580void __init pfault_irq_init(void) 597static int __cpuinit pfault_cpu_notify(struct notifier_block *self,
598 unsigned long action, void *hcpu)
581{ 599{
582 if (!MACHINE_IS_VM) 600 struct thread_struct *thread, *next;
583 return; 601 struct task_struct *tsk;
584 602
585 /* 603 switch (action) {
586 * Try to get pfault pseudo page faults going. 604 case CPU_DEAD:
587 */ 605 case CPU_DEAD_FROZEN:
588 if (register_early_external_interrupt(0x2603, pfault_interrupt, 606 spin_lock_irq(&pfault_lock);
589 &ext_int_pfault) != 0) 607 list_for_each_entry_safe(thread, next, &pfault_list, list) {
590 panic("Couldn't request external interrupt 0x2603"); 608 thread->pfault_wait = 0;
609 list_del(&thread->list);
610 tsk = container_of(thread, struct task_struct, thread);
611 wake_up_process(tsk);
612 }
613 spin_unlock_irq(&pfault_lock);
614 break;
615 default:
616 break;
617 }
618 return NOTIFY_OK;
619}
591 620
592 if (pfault_init() == 0) 621static int __init pfault_irq_init(void)
593 return; 622{
623 int rc;
594 624
595 /* Tough luck, no pfault. */ 625 if (!MACHINE_IS_VM)
626 return 0;
627 rc = register_external_interrupt(0x2603, pfault_interrupt);
628 if (rc)
629 goto out_extint;
630 rc = pfault_init() == 0 ? 0 : -EOPNOTSUPP;
631 if (rc)
632 goto out_pfault;
633 service_subclass_irq_register();
634 hotcpu_notifier(pfault_cpu_notify, 0);
635 return 0;
636
637out_pfault:
638 unregister_external_interrupt(0x2603, pfault_interrupt);
639out_extint:
596 pfault_disable = 1; 640 pfault_disable = 1;
597 unregister_early_external_interrupt(0x2603, pfault_interrupt, 641 return rc;
598 &ext_int_pfault);
599} 642}
600#endif 643early_initcall(pfault_irq_init);
644
645#endif /* CONFIG_PFAULT */
diff --git a/arch/s390/mm/gup.c b/arch/s390/mm/gup.c
new file mode 100644
index 000000000000..45b405ca2567
--- /dev/null
+++ b/arch/s390/mm/gup.c
@@ -0,0 +1,224 @@
1/*
2 * Lockless get_user_pages_fast for s390
3 *
4 * Copyright IBM Corp. 2010
5 * Author(s): Martin Schwidefsky <schwidefsky@de.ibm.com>
6 */
7#include <linux/sched.h>
8#include <linux/mm.h>
9#include <linux/hugetlb.h>
10#include <linux/vmstat.h>
11#include <linux/pagemap.h>
12#include <linux/rwsem.h>
13#include <asm/pgtable.h>
14
15/*
16 * The performance critical leaf functions are made noinline otherwise gcc
17 * inlines everything into a single function which results in too much
18 * register pressure.
19 */
20static inline int gup_pte_range(pmd_t *pmdp, pmd_t pmd, unsigned long addr,
21 unsigned long end, int write, struct page **pages, int *nr)
22{
23 unsigned long mask;
24 pte_t *ptep, pte;
25 struct page *page;
26
27 mask = (write ? _PAGE_RO : 0) | _PAGE_INVALID | _PAGE_SPECIAL;
28
29 ptep = ((pte_t *) pmd_deref(pmd)) + pte_index(addr);
30 do {
31 pte = *ptep;
32 barrier();
33 if ((pte_val(pte) & mask) != 0)
34 return 0;
35 VM_BUG_ON(!pfn_valid(pte_pfn(pte)));
36 page = pte_page(pte);
37 if (!page_cache_get_speculative(page))
38 return 0;
39 if (unlikely(pte_val(pte) != pte_val(*ptep))) {
40 put_page(page);
41 return 0;
42 }
43 pages[*nr] = page;
44 (*nr)++;
45
46 } while (ptep++, addr += PAGE_SIZE, addr != end);
47
48 return 1;
49}
50
51static inline int gup_huge_pmd(pmd_t *pmdp, pmd_t pmd, unsigned long addr,
52 unsigned long end, int write, struct page **pages, int *nr)
53{
54 unsigned long mask, result;
55 struct page *head, *page;
56 int refs;
57
58 result = write ? 0 : _SEGMENT_ENTRY_RO;
59 mask = result | _SEGMENT_ENTRY_INV;
60 if ((pmd_val(pmd) & mask) != result)
61 return 0;
62 VM_BUG_ON(!pfn_valid(pmd_val(pmd) >> PAGE_SHIFT));
63
64 refs = 0;
65 head = pmd_page(pmd);
66 page = head + ((addr & ~PMD_MASK) >> PAGE_SHIFT);
67 do {
68 VM_BUG_ON(compound_head(page) != head);
69 pages[*nr] = page;
70 (*nr)++;
71 page++;
72 refs++;
73 } while (addr += PAGE_SIZE, addr != end);
74
75 if (!page_cache_add_speculative(head, refs)) {
76 *nr -= refs;
77 return 0;
78 }
79
80 if (unlikely(pmd_val(pmd) != pmd_val(*pmdp))) {
81 *nr -= refs;
82 while (refs--)
83 put_page(head);
84 }
85
86 return 1;
87}
88
89
90static inline int gup_pmd_range(pud_t *pudp, pud_t pud, unsigned long addr,
91 unsigned long end, int write, struct page **pages, int *nr)
92{
93 unsigned long next;
94 pmd_t *pmdp, pmd;
95
96 pmdp = (pmd_t *) pudp;
97#ifdef CONFIG_64BIT
98 if ((pud_val(pud) & _REGION_ENTRY_TYPE_MASK) == _REGION_ENTRY_TYPE_R3)
99 pmdp = (pmd_t *) pud_deref(pud);
100 pmdp += pmd_index(addr);
101#endif
102 do {
103 pmd = *pmdp;
104 barrier();
105 next = pmd_addr_end(addr, end);
106 if (pmd_none(pmd))
107 return 0;
108 if (unlikely(pmd_huge(pmd))) {
109 if (!gup_huge_pmd(pmdp, pmd, addr, next,
110 write, pages, nr))
111 return 0;
112 } else if (!gup_pte_range(pmdp, pmd, addr, next,
113 write, pages, nr))
114 return 0;
115 } while (pmdp++, addr = next, addr != end);
116
117 return 1;
118}
119
120static inline int gup_pud_range(pgd_t *pgdp, pgd_t pgd, unsigned long addr,
121 unsigned long end, int write, struct page **pages, int *nr)
122{
123 unsigned long next;
124 pud_t *pudp, pud;
125
126 pudp = (pud_t *) pgdp;
127#ifdef CONFIG_64BIT
128 if ((pgd_val(pgd) & _REGION_ENTRY_TYPE_MASK) == _REGION_ENTRY_TYPE_R2)
129 pudp = (pud_t *) pgd_deref(pgd);
130 pudp += pud_index(addr);
131#endif
132 do {
133 pud = *pudp;
134 barrier();
135 next = pud_addr_end(addr, end);
136 if (pud_none(pud))
137 return 0;
138 if (!gup_pmd_range(pudp, pud, addr, next, write, pages, nr))
139 return 0;
140 } while (pudp++, addr = next, addr != end);
141
142 return 1;
143}
144
145/**
146 * get_user_pages_fast() - pin user pages in memory
147 * @start: starting user address
148 * @nr_pages: number of pages from start to pin
149 * @write: whether pages will be written to
150 * @pages: array that receives pointers to the pages pinned.
151 * Should be at least nr_pages long.
152 *
153 * Attempt to pin user pages in memory without taking mm->mmap_sem.
154 * If not successful, it will fall back to taking the lock and
155 * calling get_user_pages().
156 *
157 * Returns number of pages pinned. This may be fewer than the number
158 * requested. If nr_pages is 0 or negative, returns 0. If no pages
159 * were pinned, returns -errno.
160 */
161int get_user_pages_fast(unsigned long start, int nr_pages, int write,
162 struct page **pages)
163{
164 struct mm_struct *mm = current->mm;
165 unsigned long addr, len, end;
166 unsigned long next;
167 pgd_t *pgdp, pgd;
168 int nr = 0;
169
170 start &= PAGE_MASK;
171 addr = start;
172 len = (unsigned long) nr_pages << PAGE_SHIFT;
173 end = start + len;
174 if (end < start)
175 goto slow_irqon;
176
177 /*
178 * local_irq_disable() doesn't prevent pagetable teardown, but does
179 * prevent the pagetables from being freed on s390.
180 *
181 * So long as we atomically load page table pointers versus teardown,
182 * we can follow the address down to the the page and take a ref on it.
183 */
184 local_irq_disable();
185 pgdp = pgd_offset(mm, addr);
186 do {
187 pgd = *pgdp;
188 barrier();
189 next = pgd_addr_end(addr, end);
190 if (pgd_none(pgd))
191 goto slow;
192 if (!gup_pud_range(pgdp, pgd, addr, next, write, pages, &nr))
193 goto slow;
194 } while (pgdp++, addr = next, addr != end);
195 local_irq_enable();
196
197 VM_BUG_ON(nr != (end - start) >> PAGE_SHIFT);
198 return nr;
199
200 {
201 int ret;
202slow:
203 local_irq_enable();
204slow_irqon:
205 /* Try to get the remaining pages with get_user_pages */
206 start += nr << PAGE_SHIFT;
207 pages += nr;
208
209 down_read(&mm->mmap_sem);
210 ret = get_user_pages(current, mm, start,
211 (end - start) >> PAGE_SHIFT, write, 0, pages, NULL);
212 up_read(&mm->mmap_sem);
213
214 /* Have to be a bit careful with return values */
215 if (nr > 0) {
216 if (ret < 0)
217 ret = nr;
218 else
219 ret += nr;
220 }
221
222 return ret;
223 }
224}
diff --git a/arch/s390/mm/hugetlbpage.c b/arch/s390/mm/hugetlbpage.c
index f28c43d2f61d..a4d856db9154 100644
--- a/arch/s390/mm/hugetlbpage.c
+++ b/arch/s390/mm/hugetlbpage.c
@@ -13,7 +13,6 @@ void set_huge_pte_at(struct mm_struct *mm, unsigned long addr,
13 pte_t *pteptr, pte_t pteval) 13 pte_t *pteptr, pte_t pteval)
14{ 14{
15 pmd_t *pmdp = (pmd_t *) pteptr; 15 pmd_t *pmdp = (pmd_t *) pteptr;
16 pte_t shadow_pteval = pteval;
17 unsigned long mask; 16 unsigned long mask;
18 17
19 if (!MACHINE_HAS_HPAGE) { 18 if (!MACHINE_HAS_HPAGE) {
@@ -21,18 +20,9 @@ void set_huge_pte_at(struct mm_struct *mm, unsigned long addr,
21 mask = pte_val(pteval) & 20 mask = pte_val(pteval) &
22 (_SEGMENT_ENTRY_INV | _SEGMENT_ENTRY_RO); 21 (_SEGMENT_ENTRY_INV | _SEGMENT_ENTRY_RO);
23 pte_val(pteval) = (_SEGMENT_ENTRY + __pa(pteptr)) | mask; 22 pte_val(pteval) = (_SEGMENT_ENTRY + __pa(pteptr)) | mask;
24 if (mm->context.noexec) {
25 pteptr += PTRS_PER_PTE;
26 pte_val(shadow_pteval) =
27 (_SEGMENT_ENTRY + __pa(pteptr)) | mask;
28 }
29 } 23 }
30 24
31 pmd_val(*pmdp) = pte_val(pteval); 25 pmd_val(*pmdp) = pte_val(pteval);
32 if (mm->context.noexec) {
33 pmdp = get_shadow_table(pmdp);
34 pmd_val(*pmdp) = pte_val(shadow_pteval);
35 }
36} 26}
37 27
38int arch_prepare_hugepage(struct page *page) 28int arch_prepare_hugepage(struct page *page)
@@ -68,7 +58,7 @@ void arch_release_hugepage(struct page *page)
68 ptep = (pte_t *) page[1].index; 58 ptep = (pte_t *) page[1].index;
69 if (!ptep) 59 if (!ptep)
70 return; 60 return;
71 pte_free(&init_mm, ptep); 61 page_table_free(&init_mm, (unsigned long *) ptep);
72 page[1].index = 0; 62 page[1].index = 0;
73} 63}
74 64
diff --git a/arch/s390/mm/init.c b/arch/s390/mm/init.c
index 30eb6d02ddb8..59b663109d90 100644
--- a/arch/s390/mm/init.c
+++ b/arch/s390/mm/init.c
@@ -38,19 +38,59 @@
38#include <asm/tlbflush.h> 38#include <asm/tlbflush.h>
39#include <asm/sections.h> 39#include <asm/sections.h>
40 40
41DEFINE_PER_CPU(struct mmu_gather, mmu_gathers);
42
43pgd_t swapper_pg_dir[PTRS_PER_PGD] __attribute__((__aligned__(PAGE_SIZE))); 41pgd_t swapper_pg_dir[PTRS_PER_PGD] __attribute__((__aligned__(PAGE_SIZE)));
44 42
45char empty_zero_page[PAGE_SIZE] __attribute__((__aligned__(PAGE_SIZE))); 43unsigned long empty_zero_page, zero_page_mask;
46EXPORT_SYMBOL(empty_zero_page); 44EXPORT_SYMBOL(empty_zero_page);
47 45
46static unsigned long setup_zero_pages(void)
47{
48 struct cpuid cpu_id;
49 unsigned int order;
50 unsigned long size;
51 struct page *page;
52 int i;
53
54 get_cpu_id(&cpu_id);
55 switch (cpu_id.machine) {
56 case 0x9672: /* g5 */
57 case 0x2064: /* z900 */
58 case 0x2066: /* z900 */
59 case 0x2084: /* z990 */
60 case 0x2086: /* z990 */
61 case 0x2094: /* z9-109 */
62 case 0x2096: /* z9-109 */
63 order = 0;
64 break;
65 case 0x2097: /* z10 */
66 case 0x2098: /* z10 */
67 default:
68 order = 2;
69 break;
70 }
71
72 empty_zero_page = __get_free_pages(GFP_KERNEL | __GFP_ZERO, order);
73 if (!empty_zero_page)
74 panic("Out of memory in setup_zero_pages");
75
76 page = virt_to_page((void *) empty_zero_page);
77 split_page(page, order);
78 for (i = 1 << order; i > 0; i--) {
79 SetPageReserved(page);
80 page++;
81 }
82
83 size = PAGE_SIZE << order;
84 zero_page_mask = (size - 1) & PAGE_MASK;
85
86 return 1UL << order;
87}
88
48/* 89/*
49 * paging_init() sets up the page tables 90 * paging_init() sets up the page tables
50 */ 91 */
51void __init paging_init(void) 92void __init paging_init(void)
52{ 93{
53 static const int ssm_mask = 0x04000000L;
54 unsigned long max_zone_pfns[MAX_NR_ZONES]; 94 unsigned long max_zone_pfns[MAX_NR_ZONES];
55 unsigned long pgd_type; 95 unsigned long pgd_type;
56 96
@@ -72,18 +112,17 @@ void __init paging_init(void)
72 __ctl_load(S390_lowcore.kernel_asce, 1, 1); 112 __ctl_load(S390_lowcore.kernel_asce, 1, 1);
73 __ctl_load(S390_lowcore.kernel_asce, 7, 7); 113 __ctl_load(S390_lowcore.kernel_asce, 7, 7);
74 __ctl_load(S390_lowcore.kernel_asce, 13, 13); 114 __ctl_load(S390_lowcore.kernel_asce, 13, 13);
75 __raw_local_irq_ssm(ssm_mask); 115 arch_local_irq_restore(4UL << (BITS_PER_LONG - 8));
76 116
77 atomic_set(&init_mm.context.attach_count, 1); 117 atomic_set(&init_mm.context.attach_count, 1);
78 118
79 sparse_memory_present_with_active_regions(MAX_NUMNODES); 119 sparse_memory_present_with_active_regions(MAX_NUMNODES);
80 sparse_init(); 120 sparse_init();
81 memset(max_zone_pfns, 0, sizeof(max_zone_pfns)); 121 memset(max_zone_pfns, 0, sizeof(max_zone_pfns));
82#ifdef CONFIG_ZONE_DMA
83 max_zone_pfns[ZONE_DMA] = PFN_DOWN(MAX_DMA_ADDRESS); 122 max_zone_pfns[ZONE_DMA] = PFN_DOWN(MAX_DMA_ADDRESS);
84#endif
85 max_zone_pfns[ZONE_NORMAL] = max_low_pfn; 123 max_zone_pfns[ZONE_NORMAL] = max_low_pfn;
86 free_area_init_nodes(max_zone_pfns); 124 free_area_init_nodes(max_zone_pfns);
125 fault_init();
87} 126}
88 127
89void __init mem_init(void) 128void __init mem_init(void)
@@ -93,14 +132,12 @@ void __init mem_init(void)
93 max_mapnr = num_physpages = max_low_pfn; 132 max_mapnr = num_physpages = max_low_pfn;
94 high_memory = (void *) __va(max_low_pfn * PAGE_SIZE); 133 high_memory = (void *) __va(max_low_pfn * PAGE_SIZE);
95 134
96 /* clear the zero-page */
97 memset(empty_zero_page, 0, PAGE_SIZE);
98
99 /* Setup guest page hinting */ 135 /* Setup guest page hinting */
100 cmma_init(); 136 cmma_init();
101 137
102 /* this will put all low memory onto the freelists */ 138 /* this will put all low memory onto the freelists */
103 totalram_pages += free_all_bootmem(); 139 totalram_pages += free_all_bootmem();
140 totalram_pages -= setup_zero_pages(); /* Setup zeroed pages. */
104 141
105 reservedpages = 0; 142 reservedpages = 0;
106 143
@@ -136,7 +173,8 @@ void kernel_map_pages(struct page *page, int numpages, int enable)
136 pmd = pmd_offset(pud, address); 173 pmd = pmd_offset(pud, address);
137 pte = pte_offset_kernel(pmd, address); 174 pte = pte_offset_kernel(pmd, address);
138 if (!enable) { 175 if (!enable) {
139 ptep_invalidate(&init_mm, address, pte); 176 __ptep_ipte(address, pte);
177 pte_val(*pte) = _PAGE_TYPE_EMPTY;
140 continue; 178 continue;
141 } 179 }
142 *pte = mk_pte_phys(address, __pgprot(_PAGE_TYPE_RW)); 180 *pte = mk_pte_phys(address, __pgprot(_PAGE_TYPE_RW));
diff --git a/arch/s390/mm/maccess.c b/arch/s390/mm/maccess.c
index a8c2af8c650f..51e5cd9b906a 100644
--- a/arch/s390/mm/maccess.c
+++ b/arch/s390/mm/maccess.c
@@ -19,7 +19,7 @@
19 * using the stura instruction. 19 * using the stura instruction.
20 * Returns the number of bytes copied or -EFAULT. 20 * Returns the number of bytes copied or -EFAULT.
21 */ 21 */
22static long probe_kernel_write_odd(void *dst, void *src, size_t size) 22static long probe_kernel_write_odd(void *dst, const void *src, size_t size)
23{ 23{
24 unsigned long count, aligned; 24 unsigned long count, aligned;
25 int offset, mask; 25 int offset, mask;
@@ -45,7 +45,7 @@ static long probe_kernel_write_odd(void *dst, void *src, size_t size)
45 return rc ? rc : count; 45 return rc ? rc : count;
46} 46}
47 47
48long probe_kernel_write(void *dst, void *src, size_t size) 48long probe_kernel_write(void *dst, const void *src, size_t size)
49{ 49{
50 long copied = 0; 50 long copied = 0;
51 51
@@ -71,7 +71,7 @@ int memcpy_real(void *dest, void *src, size_t count)
71 71
72 if (!count) 72 if (!count)
73 return 0; 73 return 0;
74 flags = __raw_local_irq_stnsm(0xf8UL); 74 flags = __arch_local_irq_stnsm(0xf8UL);
75 asm volatile ( 75 asm volatile (
76 "0: mvcle %1,%2,0x0\n" 76 "0: mvcle %1,%2,0x0\n"
77 "1: jo 0b\n" 77 "1: jo 0b\n"
@@ -82,6 +82,6 @@ int memcpy_real(void *dest, void *src, size_t count)
82 "+d" (_len2), "=m" (*((long *) dest)) 82 "+d" (_len2), "=m" (*((long *) dest))
83 : "m" (*((long *) src)) 83 : "m" (*((long *) src))
84 : "cc", "memory"); 84 : "cc", "memory");
85 __raw_local_irq_ssm(flags); 85 arch_local_irq_restore(flags);
86 return rc; 86 return rc;
87} 87}
diff --git a/arch/s390/mm/mmap.c b/arch/s390/mm/mmap.c
index 869efbaed3ea..c9a9f7f18188 100644
--- a/arch/s390/mm/mmap.c
+++ b/arch/s390/mm/mmap.c
@@ -27,17 +27,44 @@
27#include <linux/personality.h> 27#include <linux/personality.h>
28#include <linux/mm.h> 28#include <linux/mm.h>
29#include <linux/module.h> 29#include <linux/module.h>
30#include <linux/random.h>
30#include <asm/pgalloc.h> 31#include <asm/pgalloc.h>
31#include <asm/compat.h> 32#include <asm/compat.h>
32 33
34static unsigned long stack_maxrandom_size(void)
35{
36 if (!(current->flags & PF_RANDOMIZE))
37 return 0;
38 if (current->personality & ADDR_NO_RANDOMIZE)
39 return 0;
40 return STACK_RND_MASK << PAGE_SHIFT;
41}
42
33/* 43/*
34 * Top of mmap area (just below the process stack). 44 * Top of mmap area (just below the process stack).
35 * 45 *
36 * Leave an at least ~128 MB hole. 46 * Leave at least a ~32 MB hole.
37 */ 47 */
38#define MIN_GAP (128*1024*1024) 48#define MIN_GAP (32*1024*1024)
39#define MAX_GAP (STACK_TOP/6*5) 49#define MAX_GAP (STACK_TOP/6*5)
40 50
51static inline int mmap_is_legacy(void)
52{
53 if (current->personality & ADDR_COMPAT_LAYOUT)
54 return 1;
55 if (rlimit(RLIMIT_STACK) == RLIM_INFINITY)
56 return 1;
57 return sysctl_legacy_va_layout;
58}
59
60static unsigned long mmap_rnd(void)
61{
62 if (!(current->flags & PF_RANDOMIZE))
63 return 0;
64 /* 8MB randomization for mmap_base */
65 return (get_random_int() & 0x7ffUL) << PAGE_SHIFT;
66}
67
41static inline unsigned long mmap_base(void) 68static inline unsigned long mmap_base(void)
42{ 69{
43 unsigned long gap = rlimit(RLIMIT_STACK); 70 unsigned long gap = rlimit(RLIMIT_STACK);
@@ -46,22 +73,8 @@ static inline unsigned long mmap_base(void)
46 gap = MIN_GAP; 73 gap = MIN_GAP;
47 else if (gap > MAX_GAP) 74 else if (gap > MAX_GAP)
48 gap = MAX_GAP; 75 gap = MAX_GAP;
49 76 gap &= PAGE_MASK;
50 return STACK_TOP - (gap & PAGE_MASK); 77 return STACK_TOP - stack_maxrandom_size() - mmap_rnd() - gap;
51}
52
53static inline int mmap_is_legacy(void)
54{
55#ifdef CONFIG_64BIT
56 /*
57 * Force standard allocation for 64 bit programs.
58 */
59 if (!is_compat_task())
60 return 1;
61#endif
62 return sysctl_legacy_va_layout ||
63 (current->personality & ADDR_COMPAT_LAYOUT) ||
64 rlimit(RLIMIT_STACK) == RLIM_INFINITY;
65} 78}
66 79
67#ifndef CONFIG_64BIT 80#ifndef CONFIG_64BIT
diff --git a/arch/s390/mm/pageattr.c b/arch/s390/mm/pageattr.c
new file mode 100644
index 000000000000..d013ed39743b
--- /dev/null
+++ b/arch/s390/mm/pageattr.c
@@ -0,0 +1,61 @@
1/*
2 * Copyright IBM Corp. 2011
3 * Author(s): Jan Glauber <jang@linux.vnet.ibm.com>
4 */
5#include <linux/module.h>
6#include <linux/mm.h>
7#include <linux/hugetlb.h>
8#include <asm/pgtable.h>
9
10static void change_page_attr(unsigned long addr, int numpages,
11 pte_t (*set) (pte_t))
12{
13 pte_t *ptep, pte;
14 pmd_t *pmdp;
15 pud_t *pudp;
16 pgd_t *pgdp;
17 int i;
18
19 for (i = 0; i < numpages; i++) {
20 pgdp = pgd_offset(&init_mm, addr);
21 pudp = pud_offset(pgdp, addr);
22 pmdp = pmd_offset(pudp, addr);
23 if (pmd_huge(*pmdp)) {
24 WARN_ON_ONCE(1);
25 continue;
26 }
27 ptep = pte_offset_kernel(pmdp, addr);
28
29 pte = *ptep;
30 pte = set(pte);
31 __ptep_ipte(addr, ptep);
32 *ptep = pte;
33 addr += PAGE_SIZE;
34 }
35}
36
37int set_memory_ro(unsigned long addr, int numpages)
38{
39 change_page_attr(addr, numpages, pte_wrprotect);
40 return 0;
41}
42EXPORT_SYMBOL_GPL(set_memory_ro);
43
44int set_memory_rw(unsigned long addr, int numpages)
45{
46 change_page_attr(addr, numpages, pte_mkwrite);
47 return 0;
48}
49EXPORT_SYMBOL_GPL(set_memory_rw);
50
51/* not possible */
52int set_memory_nx(unsigned long addr, int numpages)
53{
54 return 0;
55}
56EXPORT_SYMBOL_GPL(set_memory_nx);
57
58int set_memory_x(unsigned long addr, int numpages)
59{
60 return 0;
61}
diff --git a/arch/s390/mm/pgtable.c b/arch/s390/mm/pgtable.c
index 8d999249d357..37a23c223705 100644
--- a/arch/s390/mm/pgtable.c
+++ b/arch/s390/mm/pgtable.c
@@ -15,6 +15,7 @@
15#include <linux/spinlock.h> 15#include <linux/spinlock.h>
16#include <linux/module.h> 16#include <linux/module.h>
17#include <linux/quicklist.h> 17#include <linux/quicklist.h>
18#include <linux/rcupdate.h>
18 19
19#include <asm/system.h> 20#include <asm/system.h>
20#include <asm/pgtable.h> 21#include <asm/pgtable.h>
@@ -25,30 +26,10 @@
25 26
26#ifndef CONFIG_64BIT 27#ifndef CONFIG_64BIT
27#define ALLOC_ORDER 1 28#define ALLOC_ORDER 1
28#define TABLES_PER_PAGE 4 29#define FRAG_MASK 0x0f
29#define FRAG_MASK 15UL
30#define SECOND_HALVES 10UL
31
32void clear_table_pgstes(unsigned long *table)
33{
34 clear_table(table, _PAGE_TYPE_EMPTY, PAGE_SIZE/4);
35 memset(table + 256, 0, PAGE_SIZE/4);
36 clear_table(table + 512, _PAGE_TYPE_EMPTY, PAGE_SIZE/4);
37 memset(table + 768, 0, PAGE_SIZE/4);
38}
39
40#else 30#else
41#define ALLOC_ORDER 2 31#define ALLOC_ORDER 2
42#define TABLES_PER_PAGE 2 32#define FRAG_MASK 0x03
43#define FRAG_MASK 3UL
44#define SECOND_HALVES 2UL
45
46void clear_table_pgstes(unsigned long *table)
47{
48 clear_table(table, _PAGE_TYPE_EMPTY, PAGE_SIZE/2);
49 memset(table + 256, 0, PAGE_SIZE/2);
50}
51
52#endif 33#endif
53 34
54unsigned long VMALLOC_START = VMALLOC_END - VMALLOC_SIZE; 35unsigned long VMALLOC_START = VMALLOC_END - VMALLOC_SIZE;
@@ -63,37 +44,17 @@ static int __init parse_vmalloc(char *arg)
63} 44}
64early_param("vmalloc", parse_vmalloc); 45early_param("vmalloc", parse_vmalloc);
65 46
66unsigned long *crst_table_alloc(struct mm_struct *mm, int noexec) 47unsigned long *crst_table_alloc(struct mm_struct *mm)
67{ 48{
68 struct page *page = alloc_pages(GFP_KERNEL, ALLOC_ORDER); 49 struct page *page = alloc_pages(GFP_KERNEL, ALLOC_ORDER);
69 50
70 if (!page) 51 if (!page)
71 return NULL; 52 return NULL;
72 page->index = 0;
73 if (noexec) {
74 struct page *shadow = alloc_pages(GFP_KERNEL, ALLOC_ORDER);
75 if (!shadow) {
76 __free_pages(page, ALLOC_ORDER);
77 return NULL;
78 }
79 page->index = page_to_phys(shadow);
80 }
81 spin_lock(&mm->context.list_lock);
82 list_add(&page->lru, &mm->context.crst_list);
83 spin_unlock(&mm->context.list_lock);
84 return (unsigned long *) page_to_phys(page); 53 return (unsigned long *) page_to_phys(page);
85} 54}
86 55
87void crst_table_free(struct mm_struct *mm, unsigned long *table) 56void crst_table_free(struct mm_struct *mm, unsigned long *table)
88{ 57{
89 unsigned long *shadow = get_shadow_table(table);
90 struct page *page = virt_to_page(table);
91
92 spin_lock(&mm->context.list_lock);
93 list_del(&page->lru);
94 spin_unlock(&mm->context.list_lock);
95 if (shadow)
96 free_pages((unsigned long) shadow, ALLOC_ORDER);
97 free_pages((unsigned long) table, ALLOC_ORDER); 58 free_pages((unsigned long) table, ALLOC_ORDER);
98} 59}
99 60
@@ -105,10 +66,10 @@ int crst_table_upgrade(struct mm_struct *mm, unsigned long limit)
105 66
106 BUG_ON(limit > (1UL << 53)); 67 BUG_ON(limit > (1UL << 53));
107repeat: 68repeat:
108 table = crst_table_alloc(mm, mm->context.noexec); 69 table = crst_table_alloc(mm);
109 if (!table) 70 if (!table)
110 return -ENOMEM; 71 return -ENOMEM;
111 spin_lock(&mm->page_table_lock); 72 spin_lock_bh(&mm->page_table_lock);
112 if (mm->context.asce_limit < limit) { 73 if (mm->context.asce_limit < limit) {
113 pgd = (unsigned long *) mm->pgd; 74 pgd = (unsigned long *) mm->pgd;
114 if (mm->context.asce_limit <= (1UL << 31)) { 75 if (mm->context.asce_limit <= (1UL << 31)) {
@@ -130,7 +91,7 @@ repeat:
130 mm->task_size = mm->context.asce_limit; 91 mm->task_size = mm->context.asce_limit;
131 table = NULL; 92 table = NULL;
132 } 93 }
133 spin_unlock(&mm->page_table_lock); 94 spin_unlock_bh(&mm->page_table_lock);
134 if (table) 95 if (table)
135 crst_table_free(mm, table); 96 crst_table_free(mm, table);
136 if (mm->context.asce_limit < limit) 97 if (mm->context.asce_limit < limit)
@@ -172,94 +133,175 @@ void crst_table_downgrade(struct mm_struct *mm, unsigned long limit)
172} 133}
173#endif 134#endif
174 135
136static inline unsigned int atomic_xor_bits(atomic_t *v, unsigned int bits)
137{
138 unsigned int old, new;
139
140 do {
141 old = atomic_read(v);
142 new = old ^ bits;
143 } while (atomic_cmpxchg(v, old, new) != old);
144 return new;
145}
146
175/* 147/*
176 * page table entry allocation/free routines. 148 * page table entry allocation/free routines.
177 */ 149 */
150#ifdef CONFIG_PGSTE
151static inline unsigned long *page_table_alloc_pgste(struct mm_struct *mm)
152{
153 struct page *page;
154 unsigned long *table;
155
156 page = alloc_page(GFP_KERNEL|__GFP_REPEAT);
157 if (!page)
158 return NULL;
159 pgtable_page_ctor(page);
160 atomic_set(&page->_mapcount, 3);
161 table = (unsigned long *) page_to_phys(page);
162 clear_table(table, _PAGE_TYPE_EMPTY, PAGE_SIZE/2);
163 clear_table(table + PTRS_PER_PTE, 0, PAGE_SIZE/2);
164 return table;
165}
166
167static inline void page_table_free_pgste(unsigned long *table)
168{
169 struct page *page;
170
171 page = pfn_to_page(__pa(table) >> PAGE_SHIFT);
172 pgtable_page_ctor(page);
173 atomic_set(&page->_mapcount, -1);
174 __free_page(page);
175}
176#endif
177
178unsigned long *page_table_alloc(struct mm_struct *mm) 178unsigned long *page_table_alloc(struct mm_struct *mm)
179{ 179{
180 struct page *page; 180 struct page *page;
181 unsigned long *table; 181 unsigned long *table;
182 unsigned long bits; 182 unsigned int mask, bit;
183 183
184 bits = (mm->context.noexec || mm->context.has_pgste) ? 3UL : 1UL; 184#ifdef CONFIG_PGSTE
185 spin_lock(&mm->context.list_lock); 185 if (mm_has_pgste(mm))
186 page = NULL; 186 return page_table_alloc_pgste(mm);
187#endif
188 /* Allocate fragments of a 4K page as 1K/2K page table */
189 spin_lock_bh(&mm->context.list_lock);
190 mask = FRAG_MASK;
187 if (!list_empty(&mm->context.pgtable_list)) { 191 if (!list_empty(&mm->context.pgtable_list)) {
188 page = list_first_entry(&mm->context.pgtable_list, 192 page = list_first_entry(&mm->context.pgtable_list,
189 struct page, lru); 193 struct page, lru);
190 if ((page->flags & FRAG_MASK) == ((1UL << TABLES_PER_PAGE) - 1)) 194 table = (unsigned long *) page_to_phys(page);
191 page = NULL; 195 mask = atomic_read(&page->_mapcount);
196 mask = mask | (mask >> 4);
192 } 197 }
193 if (!page) { 198 if ((mask & FRAG_MASK) == FRAG_MASK) {
194 spin_unlock(&mm->context.list_lock); 199 spin_unlock_bh(&mm->context.list_lock);
195 page = alloc_page(GFP_KERNEL|__GFP_REPEAT); 200 page = alloc_page(GFP_KERNEL|__GFP_REPEAT);
196 if (!page) 201 if (!page)
197 return NULL; 202 return NULL;
198 pgtable_page_ctor(page); 203 pgtable_page_ctor(page);
199 page->flags &= ~FRAG_MASK; 204 atomic_set(&page->_mapcount, 1);
200 table = (unsigned long *) page_to_phys(page); 205 table = (unsigned long *) page_to_phys(page);
201 if (mm->context.has_pgste) 206 clear_table(table, _PAGE_TYPE_EMPTY, PAGE_SIZE);
202 clear_table_pgstes(table); 207 spin_lock_bh(&mm->context.list_lock);
203 else
204 clear_table(table, _PAGE_TYPE_EMPTY, PAGE_SIZE);
205 spin_lock(&mm->context.list_lock);
206 list_add(&page->lru, &mm->context.pgtable_list); 208 list_add(&page->lru, &mm->context.pgtable_list);
209 } else {
210 for (bit = 1; mask & bit; bit <<= 1)
211 table += PTRS_PER_PTE;
212 mask = atomic_xor_bits(&page->_mapcount, bit);
213 if ((mask & FRAG_MASK) == FRAG_MASK)
214 list_del(&page->lru);
207 } 215 }
208 table = (unsigned long *) page_to_phys(page); 216 spin_unlock_bh(&mm->context.list_lock);
209 while (page->flags & bits) {
210 table += 256;
211 bits <<= 1;
212 }
213 page->flags |= bits;
214 if ((page->flags & FRAG_MASK) == ((1UL << TABLES_PER_PAGE) - 1))
215 list_move_tail(&page->lru, &mm->context.pgtable_list);
216 spin_unlock(&mm->context.list_lock);
217 return table; 217 return table;
218} 218}
219 219
220void page_table_free(struct mm_struct *mm, unsigned long *table) 220void page_table_free(struct mm_struct *mm, unsigned long *table)
221{ 221{
222 struct page *page; 222 struct page *page;
223 unsigned long bits; 223 unsigned int bit, mask;
224 224
225 bits = (mm->context.noexec || mm->context.has_pgste) ? 3UL : 1UL; 225#ifdef CONFIG_PGSTE
226 bits <<= (__pa(table) & (PAGE_SIZE - 1)) / 256 / sizeof(unsigned long); 226 if (mm_has_pgste(mm))
227 return page_table_free_pgste(table);
228#endif
229 /* Free 1K/2K page table fragment of a 4K page */
227 page = pfn_to_page(__pa(table) >> PAGE_SHIFT); 230 page = pfn_to_page(__pa(table) >> PAGE_SHIFT);
228 spin_lock(&mm->context.list_lock); 231 bit = 1 << ((__pa(table) & ~PAGE_MASK)/(PTRS_PER_PTE*sizeof(pte_t)));
229 page->flags ^= bits; 232 spin_lock_bh(&mm->context.list_lock);
230 if (page->flags & FRAG_MASK) { 233 if ((atomic_read(&page->_mapcount) & FRAG_MASK) != FRAG_MASK)
231 /* Page now has some free pgtable fragments. */
232 list_move(&page->lru, &mm->context.pgtable_list);
233 page = NULL;
234 } else
235 /* All fragments of the 4K page have been freed. */
236 list_del(&page->lru); 234 list_del(&page->lru);
237 spin_unlock(&mm->context.list_lock); 235 mask = atomic_xor_bits(&page->_mapcount, bit);
238 if (page) { 236 if (mask & FRAG_MASK)
237 list_add(&page->lru, &mm->context.pgtable_list);
238 spin_unlock_bh(&mm->context.list_lock);
239 if (mask == 0) {
239 pgtable_page_dtor(page); 240 pgtable_page_dtor(page);
241 atomic_set(&page->_mapcount, -1);
240 __free_page(page); 242 __free_page(page);
241 } 243 }
242} 244}
243 245
244void disable_noexec(struct mm_struct *mm, struct task_struct *tsk) 246#ifdef CONFIG_HAVE_RCU_TABLE_FREE
247
248static void __page_table_free_rcu(void *table, unsigned bit)
245{ 249{
246 struct page *page; 250 struct page *page;
247 251
248 spin_lock(&mm->context.list_lock); 252#ifdef CONFIG_PGSTE
249 /* Free shadow region and segment tables. */ 253 if (bit == FRAG_MASK)
250 list_for_each_entry(page, &mm->context.crst_list, lru) 254 return page_table_free_pgste(table);
251 if (page->index) { 255#endif
252 free_pages((unsigned long) page->index, ALLOC_ORDER); 256 /* Free 1K/2K page table fragment of a 4K page */
253 page->index = 0; 257 page = pfn_to_page(__pa(table) >> PAGE_SHIFT);
254 } 258 if (atomic_xor_bits(&page->_mapcount, bit) == 0) {
255 /* "Free" second halves of page tables. */ 259 pgtable_page_dtor(page);
256 list_for_each_entry(page, &mm->context.pgtable_list, lru) 260 atomic_set(&page->_mapcount, -1);
257 page->flags &= ~SECOND_HALVES; 261 __free_page(page);
258 spin_unlock(&mm->context.list_lock); 262 }
259 mm->context.noexec = 0;
260 update_mm(mm, tsk);
261} 263}
262 264
265void page_table_free_rcu(struct mmu_gather *tlb, unsigned long *table)
266{
267 struct mm_struct *mm;
268 struct page *page;
269 unsigned int bit, mask;
270
271 mm = tlb->mm;
272#ifdef CONFIG_PGSTE
273 if (mm_has_pgste(mm)) {
274 table = (unsigned long *) (__pa(table) | FRAG_MASK);
275 tlb_remove_table(tlb, table);
276 return;
277 }
278#endif
279 bit = 1 << ((__pa(table) & ~PAGE_MASK) / (PTRS_PER_PTE*sizeof(pte_t)));
280 page = pfn_to_page(__pa(table) >> PAGE_SHIFT);
281 spin_lock_bh(&mm->context.list_lock);
282 if ((atomic_read(&page->_mapcount) & FRAG_MASK) != FRAG_MASK)
283 list_del(&page->lru);
284 mask = atomic_xor_bits(&page->_mapcount, bit | (bit << 4));
285 if (mask & FRAG_MASK)
286 list_add_tail(&page->lru, &mm->context.pgtable_list);
287 spin_unlock_bh(&mm->context.list_lock);
288 table = (unsigned long *) (__pa(table) | (bit << 4));
289 tlb_remove_table(tlb, table);
290}
291
292void __tlb_remove_table(void *_table)
293{
294 void *table = (void *)((unsigned long) _table & PAGE_MASK);
295 unsigned type = (unsigned long) _table & ~PAGE_MASK;
296
297 if (type)
298 __page_table_free_rcu(table, type);
299 else
300 free_pages((unsigned long) table, ALLOC_ORDER);
301}
302
303#endif
304
263/* 305/*
264 * switch on pgstes for its userspace process (for kvm) 306 * switch on pgstes for its userspace process (for kvm)
265 */ 307 */
@@ -273,7 +315,7 @@ int s390_enable_sie(void)
273 return -EINVAL; 315 return -EINVAL;
274 316
275 /* Do we have pgstes? if yes, we are done */ 317 /* Do we have pgstes? if yes, we are done */
276 if (tsk->mm->context.has_pgste) 318 if (mm_has_pgste(tsk->mm))
277 return 0; 319 return 0;
278 320
279 /* lets check if we are allowed to replace the mm */ 321 /* lets check if we are allowed to replace the mm */
@@ -312,6 +354,8 @@ int s390_enable_sie(void)
312 tsk->mm = tsk->active_mm = mm; 354 tsk->mm = tsk->active_mm = mm;
313 preempt_disable(); 355 preempt_disable();
314 update_mm(mm, tsk); 356 update_mm(mm, tsk);
357 atomic_inc(&mm->context.attach_count);
358 atomic_dec(&old_mm->context.attach_count);
315 cpumask_set_cpu(smp_processor_id(), mm_cpumask(mm)); 359 cpumask_set_cpu(smp_processor_id(), mm_cpumask(mm));
316 preempt_enable(); 360 preempt_enable();
317 task_unlock(tsk); 361 task_unlock(tsk);
diff --git a/arch/s390/mm/vmem.c b/arch/s390/mm/vmem.c
index 34c43f23b28c..8c1970d1dd91 100644
--- a/arch/s390/mm/vmem.c
+++ b/arch/s390/mm/vmem.c
@@ -95,7 +95,7 @@ static int vmem_add_mem(unsigned long start, unsigned long size, int ro)
95 pu_dir = vmem_pud_alloc(); 95 pu_dir = vmem_pud_alloc();
96 if (!pu_dir) 96 if (!pu_dir)
97 goto out; 97 goto out;
98 pgd_populate_kernel(&init_mm, pg_dir, pu_dir); 98 pgd_populate(&init_mm, pg_dir, pu_dir);
99 } 99 }
100 100
101 pu_dir = pud_offset(pg_dir, address); 101 pu_dir = pud_offset(pg_dir, address);
@@ -103,7 +103,7 @@ static int vmem_add_mem(unsigned long start, unsigned long size, int ro)
103 pm_dir = vmem_pmd_alloc(); 103 pm_dir = vmem_pmd_alloc();
104 if (!pm_dir) 104 if (!pm_dir)
105 goto out; 105 goto out;
106 pud_populate_kernel(&init_mm, pu_dir, pm_dir); 106 pud_populate(&init_mm, pu_dir, pm_dir);
107 } 107 }
108 108
109 pte = mk_pte_phys(address, __pgprot(ro ? _PAGE_RO : 0)); 109 pte = mk_pte_phys(address, __pgprot(ro ? _PAGE_RO : 0));
@@ -123,7 +123,7 @@ static int vmem_add_mem(unsigned long start, unsigned long size, int ro)
123 pt_dir = vmem_pte_alloc(); 123 pt_dir = vmem_pte_alloc();
124 if (!pt_dir) 124 if (!pt_dir)
125 goto out; 125 goto out;
126 pmd_populate_kernel(&init_mm, pm_dir, pt_dir); 126 pmd_populate(&init_mm, pm_dir, pt_dir);
127 } 127 }
128 128
129 pt_dir = pte_offset_kernel(pm_dir, address); 129 pt_dir = pte_offset_kernel(pm_dir, address);
@@ -159,7 +159,7 @@ static void vmem_remove_range(unsigned long start, unsigned long size)
159 continue; 159 continue;
160 160
161 if (pmd_huge(*pm_dir)) { 161 if (pmd_huge(*pm_dir)) {
162 pmd_clear_kernel(pm_dir); 162 pmd_clear(pm_dir);
163 address += HPAGE_SIZE - PAGE_SIZE; 163 address += HPAGE_SIZE - PAGE_SIZE;
164 continue; 164 continue;
165 } 165 }
@@ -192,7 +192,7 @@ int __meminit vmemmap_populate(struct page *start, unsigned long nr, int node)
192 pu_dir = vmem_pud_alloc(); 192 pu_dir = vmem_pud_alloc();
193 if (!pu_dir) 193 if (!pu_dir)
194 goto out; 194 goto out;
195 pgd_populate_kernel(&init_mm, pg_dir, pu_dir); 195 pgd_populate(&init_mm, pg_dir, pu_dir);
196 } 196 }
197 197
198 pu_dir = pud_offset(pg_dir, address); 198 pu_dir = pud_offset(pg_dir, address);
@@ -200,7 +200,7 @@ int __meminit vmemmap_populate(struct page *start, unsigned long nr, int node)
200 pm_dir = vmem_pmd_alloc(); 200 pm_dir = vmem_pmd_alloc();
201 if (!pm_dir) 201 if (!pm_dir)
202 goto out; 202 goto out;
203 pud_populate_kernel(&init_mm, pu_dir, pm_dir); 203 pud_populate(&init_mm, pu_dir, pm_dir);
204 } 204 }
205 205
206 pm_dir = pmd_offset(pu_dir, address); 206 pm_dir = pmd_offset(pu_dir, address);
@@ -208,7 +208,7 @@ int __meminit vmemmap_populate(struct page *start, unsigned long nr, int node)
208 pt_dir = vmem_pte_alloc(); 208 pt_dir = vmem_pte_alloc();
209 if (!pt_dir) 209 if (!pt_dir)
210 goto out; 210 goto out;
211 pmd_populate_kernel(&init_mm, pm_dir, pt_dir); 211 pmd_populate(&init_mm, pm_dir, pt_dir);
212 } 212 }
213 213
214 pt_dir = pte_offset_kernel(pm_dir, address); 214 pt_dir = pte_offset_kernel(pm_dir, address);
diff --git a/arch/s390/oprofile/Makefile b/arch/s390/oprofile/Makefile
index 537b2d840e69..524c4b615821 100644
--- a/arch/s390/oprofile/Makefile
+++ b/arch/s390/oprofile/Makefile
@@ -6,4 +6,5 @@ DRIVER_OBJS = $(addprefix ../../../drivers/oprofile/, \
6 oprofilefs.o oprofile_stats.o \ 6 oprofilefs.o oprofile_stats.o \
7 timer_int.o ) 7 timer_int.o )
8 8
9oprofile-y := $(DRIVER_OBJS) init.o backtrace.o 9oprofile-y := $(DRIVER_OBJS) init.o backtrace.o
10oprofile-$(CONFIG_64BIT) += hwsampler.o
diff --git a/arch/s390/oprofile/hwsampler.c b/arch/s390/oprofile/hwsampler.c
new file mode 100644
index 000000000000..4552ce40c81a
--- /dev/null
+++ b/arch/s390/oprofile/hwsampler.c
@@ -0,0 +1,1241 @@
1/**
2 * arch/s390/oprofile/hwsampler.c
3 *
4 * Copyright IBM Corp. 2010
5 * Author: Heinz Graalfs <graalfs@de.ibm.com>
6 */
7
8#include <linux/kernel_stat.h>
9#include <linux/kernel.h>
10#include <linux/module.h>
11#include <linux/smp.h>
12#include <linux/errno.h>
13#include <linux/workqueue.h>
14#include <linux/interrupt.h>
15#include <linux/notifier.h>
16#include <linux/cpu.h>
17#include <linux/semaphore.h>
18#include <linux/oom.h>
19#include <linux/oprofile.h>
20
21#include <asm/lowcore.h>
22#include <asm/irq.h>
23
24#include "hwsampler.h"
25
26#define MAX_NUM_SDB 511
27#define MIN_NUM_SDB 1
28
29#define ALERT_REQ_MASK 0x4000000000000000ul
30#define BUFFER_FULL_MASK 0x8000000000000000ul
31
32#define EI_IEA (1 << 31) /* invalid entry address */
33#define EI_ISE (1 << 30) /* incorrect SDBT entry */
34#define EI_PRA (1 << 29) /* program request alert */
35#define EI_SACA (1 << 23) /* sampler authorization change alert */
36#define EI_LSDA (1 << 22) /* loss of sample data alert */
37
38DECLARE_PER_CPU(struct hws_cpu_buffer, sampler_cpu_buffer);
39
40struct hws_execute_parms {
41 void *buffer;
42 signed int rc;
43};
44
45DEFINE_PER_CPU(struct hws_cpu_buffer, sampler_cpu_buffer);
46EXPORT_PER_CPU_SYMBOL(sampler_cpu_buffer);
47
48static DEFINE_MUTEX(hws_sem);
49static DEFINE_MUTEX(hws_sem_oom);
50
51static unsigned char hws_flush_all;
52static unsigned int hws_oom;
53static struct workqueue_struct *hws_wq;
54
55static unsigned int hws_state;
56enum {
57 HWS_INIT = 1,
58 HWS_DEALLOCATED,
59 HWS_STOPPED,
60 HWS_STARTED,
61 HWS_STOPPING };
62
63/* set to 1 if called by kernel during memory allocation */
64static unsigned char oom_killer_was_active;
65/* size of SDBT and SDB as of allocate API */
66static unsigned long num_sdbt = 100;
67static unsigned long num_sdb = 511;
68/* sampling interval (machine cycles) */
69static unsigned long interval;
70
71static unsigned long min_sampler_rate;
72static unsigned long max_sampler_rate;
73
74static int ssctl(void *buffer)
75{
76 int cc;
77
78 /* set in order to detect a program check */
79 cc = 1;
80
81 asm volatile(
82 "0: .insn s,0xB2870000,0(%1)\n"
83 "1: ipm %0\n"
84 " srl %0,28\n"
85 "2:\n"
86 EX_TABLE(0b, 2b) EX_TABLE(1b, 2b)
87 : "+d" (cc), "+a" (buffer)
88 : "m" (*((struct hws_ssctl_request_block *)buffer))
89 : "cc", "memory");
90
91 return cc ? -EINVAL : 0 ;
92}
93
94static int qsi(void *buffer)
95{
96 int cc;
97 cc = 1;
98
99 asm volatile(
100 "0: .insn s,0xB2860000,0(%1)\n"
101 "1: lhi %0,0\n"
102 "2:\n"
103 EX_TABLE(0b, 2b) EX_TABLE(1b, 2b)
104 : "=d" (cc), "+a" (buffer)
105 : "m" (*((struct hws_qsi_info_block *)buffer))
106 : "cc", "memory");
107
108 return cc ? -EINVAL : 0;
109}
110
111static void execute_qsi(void *parms)
112{
113 struct hws_execute_parms *ep = parms;
114
115 ep->rc = qsi(ep->buffer);
116}
117
118static void execute_ssctl(void *parms)
119{
120 struct hws_execute_parms *ep = parms;
121
122 ep->rc = ssctl(ep->buffer);
123}
124
125static int smp_ctl_ssctl_stop(int cpu)
126{
127 int rc;
128 struct hws_execute_parms ep;
129 struct hws_cpu_buffer *cb;
130
131 cb = &per_cpu(sampler_cpu_buffer, cpu);
132
133 cb->ssctl.es = 0;
134 cb->ssctl.cs = 0;
135
136 ep.buffer = &cb->ssctl;
137 smp_call_function_single(cpu, execute_ssctl, &ep, 1);
138 rc = ep.rc;
139 if (rc) {
140 printk(KERN_ERR "hwsampler: CPU %d CPUMF SSCTL failed.\n", cpu);
141 dump_stack();
142 }
143
144 ep.buffer = &cb->qsi;
145 smp_call_function_single(cpu, execute_qsi, &ep, 1);
146
147 if (cb->qsi.es || cb->qsi.cs) {
148 printk(KERN_EMERG "CPUMF sampling did not stop properly.\n");
149 dump_stack();
150 }
151
152 return rc;
153}
154
155static int smp_ctl_ssctl_deactivate(int cpu)
156{
157 int rc;
158 struct hws_execute_parms ep;
159 struct hws_cpu_buffer *cb;
160
161 cb = &per_cpu(sampler_cpu_buffer, cpu);
162
163 cb->ssctl.es = 1;
164 cb->ssctl.cs = 0;
165
166 ep.buffer = &cb->ssctl;
167 smp_call_function_single(cpu, execute_ssctl, &ep, 1);
168 rc = ep.rc;
169 if (rc)
170 printk(KERN_ERR "hwsampler: CPU %d CPUMF SSCTL failed.\n", cpu);
171
172 ep.buffer = &cb->qsi;
173 smp_call_function_single(cpu, execute_qsi, &ep, 1);
174
175 if (cb->qsi.cs)
176 printk(KERN_EMERG "CPUMF sampling was not set inactive.\n");
177
178 return rc;
179}
180
181static int smp_ctl_ssctl_enable_activate(int cpu, unsigned long interval)
182{
183 int rc;
184 struct hws_execute_parms ep;
185 struct hws_cpu_buffer *cb;
186
187 cb = &per_cpu(sampler_cpu_buffer, cpu);
188
189 cb->ssctl.h = 1;
190 cb->ssctl.tear = cb->first_sdbt;
191 cb->ssctl.dear = *(unsigned long *) cb->first_sdbt;
192 cb->ssctl.interval = interval;
193 cb->ssctl.es = 1;
194 cb->ssctl.cs = 1;
195
196 ep.buffer = &cb->ssctl;
197 smp_call_function_single(cpu, execute_ssctl, &ep, 1);
198 rc = ep.rc;
199 if (rc)
200 printk(KERN_ERR "hwsampler: CPU %d CPUMF SSCTL failed.\n", cpu);
201
202 ep.buffer = &cb->qsi;
203 smp_call_function_single(cpu, execute_qsi, &ep, 1);
204 if (ep.rc)
205 printk(KERN_ERR "hwsampler: CPU %d CPUMF QSI failed.\n", cpu);
206
207 return rc;
208}
209
210static int smp_ctl_qsi(int cpu)
211{
212 struct hws_execute_parms ep;
213 struct hws_cpu_buffer *cb;
214
215 cb = &per_cpu(sampler_cpu_buffer, cpu);
216
217 ep.buffer = &cb->qsi;
218 smp_call_function_single(cpu, execute_qsi, &ep, 1);
219
220 return ep.rc;
221}
222
223static inline unsigned long *trailer_entry_ptr(unsigned long v)
224{
225 void *ret;
226
227 ret = (void *)v;
228 ret += PAGE_SIZE;
229 ret -= sizeof(struct hws_trailer_entry);
230
231 return (unsigned long *) ret;
232}
233
234/* prototypes for external interrupt handler and worker */
235static void hws_ext_handler(unsigned int ext_int_code,
236 unsigned int param32, unsigned long param64);
237
238static void worker(struct work_struct *work);
239
240static void add_samples_to_oprofile(unsigned cpu, unsigned long *,
241 unsigned long *dear);
242
243static void init_all_cpu_buffers(void)
244{
245 int cpu;
246 struct hws_cpu_buffer *cb;
247
248 for_each_online_cpu(cpu) {
249 cb = &per_cpu(sampler_cpu_buffer, cpu);
250 memset(cb, 0, sizeof(struct hws_cpu_buffer));
251 }
252}
253
254static int is_link_entry(unsigned long *s)
255{
256 return *s & 0x1ul ? 1 : 0;
257}
258
259static unsigned long *get_next_sdbt(unsigned long *s)
260{
261 return (unsigned long *) (*s & ~0x1ul);
262}
263
264static int prepare_cpu_buffers(void)
265{
266 int cpu;
267 int rc;
268 struct hws_cpu_buffer *cb;
269
270 rc = 0;
271 for_each_online_cpu(cpu) {
272 cb = &per_cpu(sampler_cpu_buffer, cpu);
273 atomic_set(&cb->ext_params, 0);
274 cb->worker_entry = 0;
275 cb->sample_overflow = 0;
276 cb->req_alert = 0;
277 cb->incorrect_sdbt_entry = 0;
278 cb->invalid_entry_address = 0;
279 cb->loss_of_sample_data = 0;
280 cb->sample_auth_change_alert = 0;
281 cb->finish = 0;
282 cb->oom = 0;
283 cb->stop_mode = 0;
284 }
285
286 return rc;
287}
288
289/*
290 * allocate_sdbt() - allocate sampler memory
291 * @cpu: the cpu for which sampler memory is allocated
292 *
293 * A 4K page is allocated for each requested SDBT.
294 * A maximum of 511 4K pages are allocated for the SDBs in each of the SDBTs.
295 * Set ALERT_REQ mask in each SDBs trailer.
296 * Returns zero if successful, <0 otherwise.
297 */
298static int allocate_sdbt(int cpu)
299{
300 int j, k, rc;
301 unsigned long *sdbt;
302 unsigned long sdb;
303 unsigned long *tail;
304 unsigned long *trailer;
305 struct hws_cpu_buffer *cb;
306
307 cb = &per_cpu(sampler_cpu_buffer, cpu);
308
309 if (cb->first_sdbt)
310 return -EINVAL;
311
312 sdbt = NULL;
313 tail = sdbt;
314
315 for (j = 0; j < num_sdbt; j++) {
316 sdbt = (unsigned long *)get_zeroed_page(GFP_KERNEL);
317
318 mutex_lock(&hws_sem_oom);
319 /* OOM killer might have been activated */
320 barrier();
321 if (oom_killer_was_active || !sdbt) {
322 if (sdbt)
323 free_page((unsigned long)sdbt);
324
325 goto allocate_sdbt_error;
326 }
327 if (cb->first_sdbt == 0)
328 cb->first_sdbt = (unsigned long)sdbt;
329
330 /* link current page to tail of chain */
331 if (tail)
332 *tail = (unsigned long)(void *)sdbt + 1;
333
334 mutex_unlock(&hws_sem_oom);
335
336 for (k = 0; k < num_sdb; k++) {
337 /* get and set SDB page */
338 sdb = get_zeroed_page(GFP_KERNEL);
339
340 mutex_lock(&hws_sem_oom);
341 /* OOM killer might have been activated */
342 barrier();
343 if (oom_killer_was_active || !sdb) {
344 if (sdb)
345 free_page(sdb);
346
347 goto allocate_sdbt_error;
348 }
349 *sdbt = sdb;
350 trailer = trailer_entry_ptr(*sdbt);
351 *trailer = ALERT_REQ_MASK;
352 sdbt++;
353 mutex_unlock(&hws_sem_oom);
354 }
355 tail = sdbt;
356 }
357 mutex_lock(&hws_sem_oom);
358 if (oom_killer_was_active)
359 goto allocate_sdbt_error;
360
361 rc = 0;
362 if (tail)
363 *tail = (unsigned long)
364 ((void *)cb->first_sdbt) + 1;
365
366allocate_sdbt_exit:
367 mutex_unlock(&hws_sem_oom);
368 return rc;
369
370allocate_sdbt_error:
371 rc = -ENOMEM;
372 goto allocate_sdbt_exit;
373}
374
375/*
376 * deallocate_sdbt() - deallocate all sampler memory
377 *
378 * For each online CPU all SDBT trees are deallocated.
379 * Returns the number of freed pages.
380 */
381static int deallocate_sdbt(void)
382{
383 int cpu;
384 int counter;
385
386 counter = 0;
387
388 for_each_online_cpu(cpu) {
389 unsigned long start;
390 unsigned long sdbt;
391 unsigned long *curr;
392 struct hws_cpu_buffer *cb;
393
394 cb = &per_cpu(sampler_cpu_buffer, cpu);
395
396 if (!cb->first_sdbt)
397 continue;
398
399 sdbt = cb->first_sdbt;
400 curr = (unsigned long *) sdbt;
401 start = sdbt;
402
403 /* we'll free the SDBT after all SDBs are processed... */
404 while (1) {
405 if (!*curr || !sdbt)
406 break;
407
408 /* watch for link entry reset if found */
409 if (is_link_entry(curr)) {
410 curr = get_next_sdbt(curr);
411 if (sdbt)
412 free_page(sdbt);
413
414 /* we are done if we reach the start */
415 if ((unsigned long) curr == start)
416 break;
417 else
418 sdbt = (unsigned long) curr;
419 } else {
420 /* process SDB pointer */
421 if (*curr) {
422 free_page(*curr);
423 curr++;
424 }
425 }
426 counter++;
427 }
428 cb->first_sdbt = 0;
429 }
430 return counter;
431}
432
433static int start_sampling(int cpu)
434{
435 int rc;
436 struct hws_cpu_buffer *cb;
437
438 cb = &per_cpu(sampler_cpu_buffer, cpu);
439 rc = smp_ctl_ssctl_enable_activate(cpu, interval);
440 if (rc) {
441 printk(KERN_INFO "hwsampler: CPU %d ssctl failed.\n", cpu);
442 goto start_exit;
443 }
444
445 rc = -EINVAL;
446 if (!cb->qsi.es) {
447 printk(KERN_INFO "hwsampler: CPU %d ssctl not enabled.\n", cpu);
448 goto start_exit;
449 }
450
451 if (!cb->qsi.cs) {
452 printk(KERN_INFO "hwsampler: CPU %d ssctl not active.\n", cpu);
453 goto start_exit;
454 }
455
456 printk(KERN_INFO
457 "hwsampler: CPU %d, CPUMF Sampling started, interval %lu.\n",
458 cpu, interval);
459
460 rc = 0;
461
462start_exit:
463 return rc;
464}
465
466static int stop_sampling(int cpu)
467{
468 unsigned long v;
469 int rc;
470 struct hws_cpu_buffer *cb;
471
472 rc = smp_ctl_qsi(cpu);
473 WARN_ON(rc);
474
475 cb = &per_cpu(sampler_cpu_buffer, cpu);
476 if (!rc && !cb->qsi.es)
477 printk(KERN_INFO "hwsampler: CPU %d, already stopped.\n", cpu);
478
479 rc = smp_ctl_ssctl_stop(cpu);
480 if (rc) {
481 printk(KERN_INFO "hwsampler: CPU %d, ssctl stop error %d.\n",
482 cpu, rc);
483 goto stop_exit;
484 }
485
486 printk(KERN_INFO "hwsampler: CPU %d, CPUMF Sampling stopped.\n", cpu);
487
488stop_exit:
489 v = cb->req_alert;
490 if (v)
491 printk(KERN_ERR "hwsampler: CPU %d CPUMF Request alert,"
492 " count=%lu.\n", cpu, v);
493
494 v = cb->loss_of_sample_data;
495 if (v)
496 printk(KERN_ERR "hwsampler: CPU %d CPUMF Loss of sample data,"
497 " count=%lu.\n", cpu, v);
498
499 v = cb->invalid_entry_address;
500 if (v)
501 printk(KERN_ERR "hwsampler: CPU %d CPUMF Invalid entry address,"
502 " count=%lu.\n", cpu, v);
503
504 v = cb->incorrect_sdbt_entry;
505 if (v)
506 printk(KERN_ERR
507 "hwsampler: CPU %d CPUMF Incorrect SDBT address,"
508 " count=%lu.\n", cpu, v);
509
510 v = cb->sample_auth_change_alert;
511 if (v)
512 printk(KERN_ERR
513 "hwsampler: CPU %d CPUMF Sample authorization change,"
514 " count=%lu.\n", cpu, v);
515
516 return rc;
517}
518
519static int check_hardware_prerequisites(void)
520{
521 if (!test_facility(68))
522 return -EOPNOTSUPP;
523 return 0;
524}
525/*
526 * hws_oom_callback() - the OOM callback function
527 *
528 * In case the callback is invoked during memory allocation for the
529 * hw sampler, all obtained memory is deallocated and a flag is set
530 * so main sampler memory allocation can exit with a failure code.
531 * In case the callback is invoked during sampling the hw sampler
532 * is deactivated for all CPUs.
533 */
534static int hws_oom_callback(struct notifier_block *nfb,
535 unsigned long dummy, void *parm)
536{
537 unsigned long *freed;
538 int cpu;
539 struct hws_cpu_buffer *cb;
540
541 freed = parm;
542
543 mutex_lock(&hws_sem_oom);
544
545 if (hws_state == HWS_DEALLOCATED) {
546 /* during memory allocation */
547 if (oom_killer_was_active == 0) {
548 oom_killer_was_active = 1;
549 *freed += deallocate_sdbt();
550 }
551 } else {
552 int i;
553 cpu = get_cpu();
554 cb = &per_cpu(sampler_cpu_buffer, cpu);
555
556 if (!cb->oom) {
557 for_each_online_cpu(i) {
558 smp_ctl_ssctl_deactivate(i);
559 cb->oom = 1;
560 }
561 cb->finish = 1;
562
563 printk(KERN_INFO
564 "hwsampler: CPU %d, OOM notify during CPUMF Sampling.\n",
565 cpu);
566 }
567 }
568
569 mutex_unlock(&hws_sem_oom);
570
571 return NOTIFY_OK;
572}
573
574static struct notifier_block hws_oom_notifier = {
575 .notifier_call = hws_oom_callback
576};
577
578static int hws_cpu_callback(struct notifier_block *nfb,
579 unsigned long action, void *hcpu)
580{
581 /* We do not have sampler space available for all possible CPUs.
582 All CPUs should be online when hw sampling is activated. */
583 return (hws_state <= HWS_DEALLOCATED) ? NOTIFY_OK : NOTIFY_BAD;
584}
585
586static struct notifier_block hws_cpu_notifier = {
587 .notifier_call = hws_cpu_callback
588};
589
590/**
591 * hwsampler_deactivate() - set hardware sampling temporarily inactive
592 * @cpu: specifies the CPU to be set inactive.
593 *
594 * Returns 0 on success, !0 on failure.
595 */
596int hwsampler_deactivate(unsigned int cpu)
597{
598 /*
599 * Deactivate hw sampling temporarily and flush the buffer
600 * by pushing all the pending samples to oprofile buffer.
601 *
602 * This function can be called under one of the following conditions:
603 * Memory unmap, task is exiting.
604 */
605 int rc;
606 struct hws_cpu_buffer *cb;
607
608 rc = 0;
609 mutex_lock(&hws_sem);
610
611 cb = &per_cpu(sampler_cpu_buffer, cpu);
612 if (hws_state == HWS_STARTED) {
613 rc = smp_ctl_qsi(cpu);
614 WARN_ON(rc);
615 if (cb->qsi.cs) {
616 rc = smp_ctl_ssctl_deactivate(cpu);
617 if (rc) {
618 printk(KERN_INFO
619 "hwsampler: CPU %d, CPUMF Deactivation failed.\n", cpu);
620 cb->finish = 1;
621 hws_state = HWS_STOPPING;
622 } else {
623 hws_flush_all = 1;
624 /* Add work to queue to read pending samples.*/
625 queue_work_on(cpu, hws_wq, &cb->worker);
626 }
627 }
628 }
629 mutex_unlock(&hws_sem);
630
631 if (hws_wq)
632 flush_workqueue(hws_wq);
633
634 return rc;
635}
636
637/**
638 * hwsampler_activate() - activate/resume hardware sampling which was deactivated
639 * @cpu: specifies the CPU to be set active.
640 *
641 * Returns 0 on success, !0 on failure.
642 */
643int hwsampler_activate(unsigned int cpu)
644{
645 /*
646 * Re-activate hw sampling. This should be called in pair with
647 * hwsampler_deactivate().
648 */
649 int rc;
650 struct hws_cpu_buffer *cb;
651
652 rc = 0;
653 mutex_lock(&hws_sem);
654
655 cb = &per_cpu(sampler_cpu_buffer, cpu);
656 if (hws_state == HWS_STARTED) {
657 rc = smp_ctl_qsi(cpu);
658 WARN_ON(rc);
659 if (!cb->qsi.cs) {
660 hws_flush_all = 0;
661 rc = smp_ctl_ssctl_enable_activate(cpu, interval);
662 if (rc) {
663 printk(KERN_ERR
664 "CPU %d, CPUMF activate sampling failed.\n",
665 cpu);
666 }
667 }
668 }
669
670 mutex_unlock(&hws_sem);
671
672 return rc;
673}
674
675static void hws_ext_handler(unsigned int ext_int_code,
676 unsigned int param32, unsigned long param64)
677{
678 struct hws_cpu_buffer *cb;
679
680 kstat_cpu(smp_processor_id()).irqs[EXTINT_CPM]++;
681 cb = &__get_cpu_var(sampler_cpu_buffer);
682 atomic_xchg(&cb->ext_params, atomic_read(&cb->ext_params) | param32);
683 if (hws_wq)
684 queue_work(hws_wq, &cb->worker);
685}
686
687static int check_qsi_on_setup(void)
688{
689 int rc;
690 unsigned int cpu;
691 struct hws_cpu_buffer *cb;
692
693 for_each_online_cpu(cpu) {
694 cb = &per_cpu(sampler_cpu_buffer, cpu);
695 rc = smp_ctl_qsi(cpu);
696 WARN_ON(rc);
697 if (rc)
698 return -EOPNOTSUPP;
699
700 if (!cb->qsi.as) {
701 printk(KERN_INFO "hwsampler: CPUMF sampling is not authorized.\n");
702 return -EINVAL;
703 }
704
705 if (cb->qsi.es) {
706 printk(KERN_WARNING "hwsampler: CPUMF is still enabled.\n");
707 rc = smp_ctl_ssctl_stop(cpu);
708 if (rc)
709 return -EINVAL;
710
711 printk(KERN_INFO
712 "CPU %d, CPUMF Sampling stopped now.\n", cpu);
713 }
714 }
715 return 0;
716}
717
718static int check_qsi_on_start(void)
719{
720 unsigned int cpu;
721 int rc;
722 struct hws_cpu_buffer *cb;
723
724 for_each_online_cpu(cpu) {
725 cb = &per_cpu(sampler_cpu_buffer, cpu);
726 rc = smp_ctl_qsi(cpu);
727 WARN_ON(rc);
728
729 if (!cb->qsi.as)
730 return -EINVAL;
731
732 if (cb->qsi.es)
733 return -EINVAL;
734
735 if (cb->qsi.cs)
736 return -EINVAL;
737 }
738 return 0;
739}
740
741static void worker_on_start(unsigned int cpu)
742{
743 struct hws_cpu_buffer *cb;
744
745 cb = &per_cpu(sampler_cpu_buffer, cpu);
746 cb->worker_entry = cb->first_sdbt;
747}
748
749static int worker_check_error(unsigned int cpu, int ext_params)
750{
751 int rc;
752 unsigned long *sdbt;
753 struct hws_cpu_buffer *cb;
754
755 rc = 0;
756 cb = &per_cpu(sampler_cpu_buffer, cpu);
757 sdbt = (unsigned long *) cb->worker_entry;
758
759 if (!sdbt || !*sdbt)
760 return -EINVAL;
761
762 if (ext_params & EI_PRA)
763 cb->req_alert++;
764
765 if (ext_params & EI_LSDA)
766 cb->loss_of_sample_data++;
767
768 if (ext_params & EI_IEA) {
769 cb->invalid_entry_address++;
770 rc = -EINVAL;
771 }
772
773 if (ext_params & EI_ISE) {
774 cb->incorrect_sdbt_entry++;
775 rc = -EINVAL;
776 }
777
778 if (ext_params & EI_SACA) {
779 cb->sample_auth_change_alert++;
780 rc = -EINVAL;
781 }
782
783 return rc;
784}
785
786static void worker_on_finish(unsigned int cpu)
787{
788 int rc, i;
789 struct hws_cpu_buffer *cb;
790
791 cb = &per_cpu(sampler_cpu_buffer, cpu);
792
793 if (cb->finish) {
794 rc = smp_ctl_qsi(cpu);
795 WARN_ON(rc);
796 if (cb->qsi.es) {
797 printk(KERN_INFO
798 "hwsampler: CPU %d, CPUMF Stop/Deactivate sampling.\n",
799 cpu);
800 rc = smp_ctl_ssctl_stop(cpu);
801 if (rc)
802 printk(KERN_INFO
803 "hwsampler: CPU %d, CPUMF Deactivation failed.\n",
804 cpu);
805
806 for_each_online_cpu(i) {
807 if (i == cpu)
808 continue;
809 if (!cb->finish) {
810 cb->finish = 1;
811 queue_work_on(i, hws_wq,
812 &cb->worker);
813 }
814 }
815 }
816 }
817}
818
819static void worker_on_interrupt(unsigned int cpu)
820{
821 unsigned long *sdbt;
822 unsigned char done;
823 struct hws_cpu_buffer *cb;
824
825 cb = &per_cpu(sampler_cpu_buffer, cpu);
826
827 sdbt = (unsigned long *) cb->worker_entry;
828
829 done = 0;
830 /* do not proceed if stop was entered,
831 * forget the buffers not yet processed */
832 while (!done && !cb->stop_mode) {
833 unsigned long *trailer;
834 struct hws_trailer_entry *te;
835 unsigned long *dear = 0;
836
837 trailer = trailer_entry_ptr(*sdbt);
838 /* leave loop if no more work to do */
839 if (!(*trailer & BUFFER_FULL_MASK)) {
840 done = 1;
841 if (!hws_flush_all)
842 continue;
843 }
844
845 te = (struct hws_trailer_entry *)trailer;
846 cb->sample_overflow += te->overflow;
847
848 add_samples_to_oprofile(cpu, sdbt, dear);
849
850 /* reset trailer */
851 xchg((unsigned char *) te, 0x40);
852
853 /* advance to next sdb slot in current sdbt */
854 sdbt++;
855 /* in case link bit is set use address w/o link bit */
856 if (is_link_entry(sdbt))
857 sdbt = get_next_sdbt(sdbt);
858
859 cb->worker_entry = (unsigned long)sdbt;
860 }
861}
862
863static void add_samples_to_oprofile(unsigned int cpu, unsigned long *sdbt,
864 unsigned long *dear)
865{
866 struct hws_data_entry *sample_data_ptr;
867 unsigned long *trailer;
868
869 trailer = trailer_entry_ptr(*sdbt);
870 if (dear) {
871 if (dear > trailer)
872 return;
873 trailer = dear;
874 }
875
876 sample_data_ptr = (struct hws_data_entry *)(*sdbt);
877
878 while ((unsigned long *)sample_data_ptr < trailer) {
879 struct pt_regs *regs = NULL;
880 struct task_struct *tsk = NULL;
881
882 /*
883 * Check sampling mode, 1 indicates basic (=customer) sampling
884 * mode.
885 */
886 if (sample_data_ptr->def != 1) {
887 /* sample slot is not yet written */
888 break;
889 } else {
890 /* make sure we don't use it twice,
891 * the next time the sampler will set it again */
892 sample_data_ptr->def = 0;
893 }
894
895 /* Get pt_regs. */
896 if (sample_data_ptr->P == 1) {
897 /* userspace sample */
898 unsigned int pid = sample_data_ptr->prim_asn;
899 rcu_read_lock();
900 tsk = pid_task(find_vpid(pid), PIDTYPE_PID);
901 if (tsk)
902 regs = task_pt_regs(tsk);
903 rcu_read_unlock();
904 } else {
905 /* kernelspace sample */
906 regs = task_pt_regs(current);
907 }
908
909 mutex_lock(&hws_sem);
910 oprofile_add_ext_hw_sample(sample_data_ptr->ia, regs, 0,
911 !sample_data_ptr->P, tsk);
912 mutex_unlock(&hws_sem);
913
914 sample_data_ptr++;
915 }
916}
917
918static void worker(struct work_struct *work)
919{
920 unsigned int cpu;
921 int ext_params;
922 struct hws_cpu_buffer *cb;
923
924 cb = container_of(work, struct hws_cpu_buffer, worker);
925 cpu = smp_processor_id();
926 ext_params = atomic_xchg(&cb->ext_params, 0);
927
928 if (!cb->worker_entry)
929 worker_on_start(cpu);
930
931 if (worker_check_error(cpu, ext_params))
932 return;
933
934 if (!cb->finish)
935 worker_on_interrupt(cpu);
936
937 if (cb->finish)
938 worker_on_finish(cpu);
939}
940
941/**
942 * hwsampler_allocate() - allocate memory for the hardware sampler
943 * @sdbt: number of SDBTs per online CPU (must be > 0)
944 * @sdb: number of SDBs per SDBT (minimum 1, maximum 511)
945 *
946 * Returns 0 on success, !0 on failure.
947 */
948int hwsampler_allocate(unsigned long sdbt, unsigned long sdb)
949{
950 int cpu, rc;
951 mutex_lock(&hws_sem);
952
953 rc = -EINVAL;
954 if (hws_state != HWS_DEALLOCATED)
955 goto allocate_exit;
956
957 if (sdbt < 1)
958 goto allocate_exit;
959
960 if (sdb > MAX_NUM_SDB || sdb < MIN_NUM_SDB)
961 goto allocate_exit;
962
963 num_sdbt = sdbt;
964 num_sdb = sdb;
965
966 oom_killer_was_active = 0;
967 register_oom_notifier(&hws_oom_notifier);
968
969 for_each_online_cpu(cpu) {
970 if (allocate_sdbt(cpu)) {
971 unregister_oom_notifier(&hws_oom_notifier);
972 goto allocate_error;
973 }
974 }
975 unregister_oom_notifier(&hws_oom_notifier);
976 if (oom_killer_was_active)
977 goto allocate_error;
978
979 hws_state = HWS_STOPPED;
980 rc = 0;
981
982allocate_exit:
983 mutex_unlock(&hws_sem);
984 return rc;
985
986allocate_error:
987 rc = -ENOMEM;
988 printk(KERN_ERR "hwsampler: CPUMF Memory allocation failed.\n");
989 goto allocate_exit;
990}
991
992/**
993 * hwsampler_deallocate() - deallocate hardware sampler memory
994 *
995 * Returns 0 on success, !0 on failure.
996 */
997int hwsampler_deallocate()
998{
999 int rc;
1000
1001 mutex_lock(&hws_sem);
1002
1003 rc = -EINVAL;
1004 if (hws_state != HWS_STOPPED)
1005 goto deallocate_exit;
1006
1007 ctl_clear_bit(0, 5); /* set bit 58 CR0 off */
1008 deallocate_sdbt();
1009
1010 hws_state = HWS_DEALLOCATED;
1011 rc = 0;
1012
1013deallocate_exit:
1014 mutex_unlock(&hws_sem);
1015
1016 return rc;
1017}
1018
1019unsigned long hwsampler_query_min_interval(void)
1020{
1021 return min_sampler_rate;
1022}
1023
1024unsigned long hwsampler_query_max_interval(void)
1025{
1026 return max_sampler_rate;
1027}
1028
1029unsigned long hwsampler_get_sample_overflow_count(unsigned int cpu)
1030{
1031 struct hws_cpu_buffer *cb;
1032
1033 cb = &per_cpu(sampler_cpu_buffer, cpu);
1034
1035 return cb->sample_overflow;
1036}
1037
1038int hwsampler_setup()
1039{
1040 int rc;
1041 int cpu;
1042 struct hws_cpu_buffer *cb;
1043
1044 mutex_lock(&hws_sem);
1045
1046 rc = -EINVAL;
1047 if (hws_state)
1048 goto setup_exit;
1049
1050 hws_state = HWS_INIT;
1051
1052 init_all_cpu_buffers();
1053
1054 rc = check_hardware_prerequisites();
1055 if (rc)
1056 goto setup_exit;
1057
1058 rc = check_qsi_on_setup();
1059 if (rc)
1060 goto setup_exit;
1061
1062 rc = -EINVAL;
1063 hws_wq = create_workqueue("hwsampler");
1064 if (!hws_wq)
1065 goto setup_exit;
1066
1067 register_cpu_notifier(&hws_cpu_notifier);
1068
1069 for_each_online_cpu(cpu) {
1070 cb = &per_cpu(sampler_cpu_buffer, cpu);
1071 INIT_WORK(&cb->worker, worker);
1072 rc = smp_ctl_qsi(cpu);
1073 WARN_ON(rc);
1074 if (min_sampler_rate != cb->qsi.min_sampl_rate) {
1075 if (min_sampler_rate) {
1076 printk(KERN_WARNING
1077 "hwsampler: different min sampler rate values.\n");
1078 if (min_sampler_rate < cb->qsi.min_sampl_rate)
1079 min_sampler_rate =
1080 cb->qsi.min_sampl_rate;
1081 } else
1082 min_sampler_rate = cb->qsi.min_sampl_rate;
1083 }
1084 if (max_sampler_rate != cb->qsi.max_sampl_rate) {
1085 if (max_sampler_rate) {
1086 printk(KERN_WARNING
1087 "hwsampler: different max sampler rate values.\n");
1088 if (max_sampler_rate > cb->qsi.max_sampl_rate)
1089 max_sampler_rate =
1090 cb->qsi.max_sampl_rate;
1091 } else
1092 max_sampler_rate = cb->qsi.max_sampl_rate;
1093 }
1094 }
1095 register_external_interrupt(0x1407, hws_ext_handler);
1096
1097 hws_state = HWS_DEALLOCATED;
1098 rc = 0;
1099
1100setup_exit:
1101 mutex_unlock(&hws_sem);
1102 return rc;
1103}
1104
1105int hwsampler_shutdown()
1106{
1107 int rc;
1108
1109 mutex_lock(&hws_sem);
1110
1111 rc = -EINVAL;
1112 if (hws_state == HWS_DEALLOCATED || hws_state == HWS_STOPPED) {
1113 mutex_unlock(&hws_sem);
1114
1115 if (hws_wq)
1116 flush_workqueue(hws_wq);
1117
1118 mutex_lock(&hws_sem);
1119
1120 if (hws_state == HWS_STOPPED) {
1121 ctl_clear_bit(0, 5); /* set bit 58 CR0 off */
1122 deallocate_sdbt();
1123 }
1124 if (hws_wq) {
1125 destroy_workqueue(hws_wq);
1126 hws_wq = NULL;
1127 }
1128
1129 unregister_external_interrupt(0x1407, hws_ext_handler);
1130 hws_state = HWS_INIT;
1131 rc = 0;
1132 }
1133 mutex_unlock(&hws_sem);
1134
1135 unregister_cpu_notifier(&hws_cpu_notifier);
1136
1137 return rc;
1138}
1139
1140/**
1141 * hwsampler_start_all() - start hardware sampling on all online CPUs
1142 * @rate: specifies the used interval when samples are taken
1143 *
1144 * Returns 0 on success, !0 on failure.
1145 */
1146int hwsampler_start_all(unsigned long rate)
1147{
1148 int rc, cpu;
1149
1150 mutex_lock(&hws_sem);
1151
1152 hws_oom = 0;
1153
1154 rc = -EINVAL;
1155 if (hws_state != HWS_STOPPED)
1156 goto start_all_exit;
1157
1158 interval = rate;
1159
1160 /* fail if rate is not valid */
1161 if (interval < min_sampler_rate || interval > max_sampler_rate)
1162 goto start_all_exit;
1163
1164 rc = check_qsi_on_start();
1165 if (rc)
1166 goto start_all_exit;
1167
1168 rc = prepare_cpu_buffers();
1169 if (rc)
1170 goto start_all_exit;
1171
1172 for_each_online_cpu(cpu) {
1173 rc = start_sampling(cpu);
1174 if (rc)
1175 break;
1176 }
1177 if (rc) {
1178 for_each_online_cpu(cpu) {
1179 stop_sampling(cpu);
1180 }
1181 goto start_all_exit;
1182 }
1183 hws_state = HWS_STARTED;
1184 rc = 0;
1185
1186start_all_exit:
1187 mutex_unlock(&hws_sem);
1188
1189 if (rc)
1190 return rc;
1191
1192 register_oom_notifier(&hws_oom_notifier);
1193 hws_oom = 1;
1194 hws_flush_all = 0;
1195 /* now let them in, 1407 CPUMF external interrupts */
1196 ctl_set_bit(0, 5); /* set CR0 bit 58 */
1197
1198 return 0;
1199}
1200
1201/**
1202 * hwsampler_stop_all() - stop hardware sampling on all online CPUs
1203 *
1204 * Returns 0 on success, !0 on failure.
1205 */
1206int hwsampler_stop_all()
1207{
1208 int tmp_rc, rc, cpu;
1209 struct hws_cpu_buffer *cb;
1210
1211 mutex_lock(&hws_sem);
1212
1213 rc = 0;
1214 if (hws_state == HWS_INIT) {
1215 mutex_unlock(&hws_sem);
1216 return rc;
1217 }
1218 hws_state = HWS_STOPPING;
1219 mutex_unlock(&hws_sem);
1220
1221 for_each_online_cpu(cpu) {
1222 cb = &per_cpu(sampler_cpu_buffer, cpu);
1223 cb->stop_mode = 1;
1224 tmp_rc = stop_sampling(cpu);
1225 if (tmp_rc)
1226 rc = tmp_rc;
1227 }
1228
1229 if (hws_wq)
1230 flush_workqueue(hws_wq);
1231
1232 mutex_lock(&hws_sem);
1233 if (hws_oom) {
1234 unregister_oom_notifier(&hws_oom_notifier);
1235 hws_oom = 0;
1236 }
1237 hws_state = HWS_STOPPED;
1238 mutex_unlock(&hws_sem);
1239
1240 return rc;
1241}
diff --git a/arch/s390/oprofile/hwsampler.h b/arch/s390/oprofile/hwsampler.h
new file mode 100644
index 000000000000..1912f3bb190c
--- /dev/null
+++ b/arch/s390/oprofile/hwsampler.h
@@ -0,0 +1,113 @@
1/*
2 * CPUMF HW sampler functions and internal structures
3 *
4 * Copyright IBM Corp. 2010
5 * Author(s): Heinz Graalfs <graalfs@de.ibm.com>
6 */
7
8#ifndef HWSAMPLER_H_
9#define HWSAMPLER_H_
10
11#include <linux/workqueue.h>
12
13struct hws_qsi_info_block /* QUERY SAMPLING information block */
14{ /* Bit(s) */
15 unsigned int b0_13:14; /* 0-13: zeros */
16 unsigned int as:1; /* 14: sampling authorisation control*/
17 unsigned int b15_21:7; /* 15-21: zeros */
18 unsigned int es:1; /* 22: sampling enable control */
19 unsigned int b23_29:7; /* 23-29: zeros */
20 unsigned int cs:1; /* 30: sampling activation control */
21 unsigned int:1; /* 31: reserved */
22 unsigned int bsdes:16; /* 4-5: size of sampling entry */
23 unsigned int:16; /* 6-7: reserved */
24 unsigned long min_sampl_rate; /* 8-15: minimum sampling interval */
25 unsigned long max_sampl_rate; /* 16-23: maximum sampling interval*/
26 unsigned long tear; /* 24-31: TEAR contents */
27 unsigned long dear; /* 32-39: DEAR contents */
28 unsigned int rsvrd0; /* 40-43: reserved */
29 unsigned int cpu_speed; /* 44-47: CPU speed */
30 unsigned long long rsvrd1; /* 48-55: reserved */
31 unsigned long long rsvrd2; /* 56-63: reserved */
32};
33
34struct hws_ssctl_request_block /* SET SAMPLING CONTROLS req block */
35{ /* bytes 0 - 7 Bit(s) */
36 unsigned int s:1; /* 0: maximum buffer indicator */
37 unsigned int h:1; /* 1: part. level reserved for VM use*/
38 unsigned long b2_53:52; /* 2-53: zeros */
39 unsigned int es:1; /* 54: sampling enable control */
40 unsigned int b55_61:7; /* 55-61: - zeros */
41 unsigned int cs:1; /* 62: sampling activation control */
42 unsigned int b63:1; /* 63: zero */
43 unsigned long interval; /* 8-15: sampling interval */
44 unsigned long tear; /* 16-23: TEAR contents */
45 unsigned long dear; /* 24-31: DEAR contents */
46 /* 32-63: */
47 unsigned long rsvrd1; /* reserved */
48 unsigned long rsvrd2; /* reserved */
49 unsigned long rsvrd3; /* reserved */
50 unsigned long rsvrd4; /* reserved */
51};
52
53struct hws_cpu_buffer {
54 unsigned long first_sdbt; /* @ of 1st SDB-Table for this CP*/
55 unsigned long worker_entry;
56 unsigned long sample_overflow; /* taken from SDB ... */
57 struct hws_qsi_info_block qsi;
58 struct hws_ssctl_request_block ssctl;
59 struct work_struct worker;
60 atomic_t ext_params;
61 unsigned long req_alert;
62 unsigned long loss_of_sample_data;
63 unsigned long invalid_entry_address;
64 unsigned long incorrect_sdbt_entry;
65 unsigned long sample_auth_change_alert;
66 unsigned int finish:1;
67 unsigned int oom:1;
68 unsigned int stop_mode:1;
69};
70
71struct hws_data_entry {
72 unsigned int def:16; /* 0-15 Data Entry Format */
73 unsigned int R:4; /* 16-19 reserved */
74 unsigned int U:4; /* 20-23 Number of unique instruct. */
75 unsigned int z:2; /* zeros */
76 unsigned int T:1; /* 26 PSW DAT mode */
77 unsigned int W:1; /* 27 PSW wait state */
78 unsigned int P:1; /* 28 PSW Problem state */
79 unsigned int AS:2; /* 29-30 PSW address-space control */
80 unsigned int I:1; /* 31 entry valid or invalid */
81 unsigned int:16;
82 unsigned int prim_asn:16; /* primary ASN */
83 unsigned long long ia; /* Instruction Address */
84 unsigned long long lpp; /* Logical-Partition Program Param. */
85 unsigned long long vpp; /* Virtual-Machine Program Param. */
86};
87
88struct hws_trailer_entry {
89 unsigned int f:1; /* 0 - Block Full Indicator */
90 unsigned int a:1; /* 1 - Alert request control */
91 unsigned long:62; /* 2 - 63: Reserved */
92 unsigned long overflow; /* 64 - sample Overflow count */
93 unsigned long timestamp; /* 16 - time-stamp */
94 unsigned long timestamp1; /* */
95 unsigned long reserved1; /* 32 -Reserved */
96 unsigned long reserved2; /* */
97 unsigned long progusage1; /* 48 - reserved for programming use */
98 unsigned long progusage2; /* */
99};
100
101int hwsampler_setup(void);
102int hwsampler_shutdown(void);
103int hwsampler_allocate(unsigned long sdbt, unsigned long sdb);
104int hwsampler_deallocate(void);
105unsigned long hwsampler_query_min_interval(void);
106unsigned long hwsampler_query_max_interval(void);
107int hwsampler_start_all(unsigned long interval);
108int hwsampler_stop_all(void);
109int hwsampler_deactivate(unsigned int cpu);
110int hwsampler_activate(unsigned int cpu);
111unsigned long hwsampler_get_sample_overflow_count(unsigned int cpu);
112
113#endif /*HWSAMPLER_H_*/
diff --git a/arch/s390/oprofile/init.c b/arch/s390/oprofile/init.c
index 7a995113b918..0e358c2cffeb 100644
--- a/arch/s390/oprofile/init.c
+++ b/arch/s390/oprofile/init.c
@@ -4,23 +4,195 @@
4 * S390 Version 4 * S390 Version
5 * Copyright (C) 2003 IBM Deutschland Entwicklung GmbH, IBM Corporation 5 * Copyright (C) 2003 IBM Deutschland Entwicklung GmbH, IBM Corporation
6 * Author(s): Thomas Spatzier (tspat@de.ibm.com) 6 * Author(s): Thomas Spatzier (tspat@de.ibm.com)
7 * Author(s): Mahesh Salgaonkar (mahesh@linux.vnet.ibm.com)
8 * Author(s): Heinz Graalfs (graalfs@linux.vnet.ibm.com)
7 * 9 *
8 * @remark Copyright 2002 OProfile authors 10 * @remark Copyright 2002-2011 OProfile authors
9 */ 11 */
10 12
11#include <linux/oprofile.h> 13#include <linux/oprofile.h>
12#include <linux/init.h> 14#include <linux/init.h>
13#include <linux/errno.h> 15#include <linux/errno.h>
16#include <linux/oprofile.h>
17#include <linux/errno.h>
18#include <linux/fs.h>
14 19
20#include "../../../drivers/oprofile/oprof.h"
15 21
16extern void s390_backtrace(struct pt_regs * const regs, unsigned int depth); 22extern void s390_backtrace(struct pt_regs * const regs, unsigned int depth);
17 23
18int __init oprofile_arch_init(struct oprofile_operations* ops) 24#ifdef CONFIG_64BIT
25
26#include "hwsampler.h"
27
28#define DEFAULT_INTERVAL 4127518
29
30#define DEFAULT_SDBT_BLOCKS 1
31#define DEFAULT_SDB_BLOCKS 511
32
33static unsigned long oprofile_hw_interval = DEFAULT_INTERVAL;
34static unsigned long oprofile_min_interval;
35static unsigned long oprofile_max_interval;
36
37static unsigned long oprofile_sdbt_blocks = DEFAULT_SDBT_BLOCKS;
38static unsigned long oprofile_sdb_blocks = DEFAULT_SDB_BLOCKS;
39
40static int hwsampler_file;
41static int hwsampler_running; /* start_mutex must be held to change */
42
43static struct oprofile_operations timer_ops;
44
45static int oprofile_hwsampler_start(void)
46{
47 int retval;
48
49 hwsampler_running = hwsampler_file;
50
51 if (!hwsampler_running)
52 return timer_ops.start();
53
54 retval = hwsampler_allocate(oprofile_sdbt_blocks, oprofile_sdb_blocks);
55 if (retval)
56 return retval;
57
58 retval = hwsampler_start_all(oprofile_hw_interval);
59 if (retval)
60 hwsampler_deallocate();
61
62 return retval;
63}
64
65static void oprofile_hwsampler_stop(void)
66{
67 if (!hwsampler_running) {
68 timer_ops.stop();
69 return;
70 }
71
72 hwsampler_stop_all();
73 hwsampler_deallocate();
74 return;
75}
76
77static ssize_t hwsampler_read(struct file *file, char __user *buf,
78 size_t count, loff_t *offset)
79{
80 return oprofilefs_ulong_to_user(hwsampler_file, buf, count, offset);
81}
82
83static ssize_t hwsampler_write(struct file *file, char const __user *buf,
84 size_t count, loff_t *offset)
85{
86 unsigned long val;
87 int retval;
88
89 if (*offset)
90 return -EINVAL;
91
92 retval = oprofilefs_ulong_from_user(&val, buf, count);
93 if (retval)
94 return retval;
95
96 if (oprofile_started)
97 /*
98 * save to do without locking as we set
99 * hwsampler_running in start() when start_mutex is
100 * held
101 */
102 return -EBUSY;
103
104 hwsampler_file = val;
105
106 return count;
107}
108
109static const struct file_operations hwsampler_fops = {
110 .read = hwsampler_read,
111 .write = hwsampler_write,
112};
113
114static int oprofile_create_hwsampling_files(struct super_block *sb,
115 struct dentry *root)
116{
117 struct dentry *hw_dir;
118
119 /* reinitialize default values */
120 hwsampler_file = 1;
121
122 hw_dir = oprofilefs_mkdir(sb, root, "hwsampling");
123 if (!hw_dir)
124 return -EINVAL;
125
126 oprofilefs_create_file(sb, hw_dir, "hwsampler", &hwsampler_fops);
127 oprofilefs_create_ulong(sb, hw_dir, "hw_interval",
128 &oprofile_hw_interval);
129 oprofilefs_create_ro_ulong(sb, hw_dir, "hw_min_interval",
130 &oprofile_min_interval);
131 oprofilefs_create_ro_ulong(sb, hw_dir, "hw_max_interval",
132 &oprofile_max_interval);
133 oprofilefs_create_ulong(sb, hw_dir, "hw_sdbt_blocks",
134 &oprofile_sdbt_blocks);
135
136 return 0;
137}
138
139static int oprofile_hwsampler_init(struct oprofile_operations *ops)
140{
141 if (hwsampler_setup())
142 return -ENODEV;
143
144 /*
145 * create hwsampler files only if hwsampler_setup() succeeds.
146 */
147 oprofile_min_interval = hwsampler_query_min_interval();
148 if (oprofile_min_interval == 0)
149 return -ENODEV;
150 oprofile_max_interval = hwsampler_query_max_interval();
151 if (oprofile_max_interval == 0)
152 return -ENODEV;
153
154 /* The initial value should be sane */
155 if (oprofile_hw_interval < oprofile_min_interval)
156 oprofile_hw_interval = oprofile_min_interval;
157 if (oprofile_hw_interval > oprofile_max_interval)
158 oprofile_hw_interval = oprofile_max_interval;
159
160 if (oprofile_timer_init(ops))
161 return -ENODEV;
162
163 printk(KERN_INFO "oprofile: using hardware sampling\n");
164
165 memcpy(&timer_ops, ops, sizeof(timer_ops));
166
167 ops->start = oprofile_hwsampler_start;
168 ops->stop = oprofile_hwsampler_stop;
169 ops->create_files = oprofile_create_hwsampling_files;
170
171 return 0;
172}
173
174static void oprofile_hwsampler_exit(void)
175{
176 oprofile_timer_exit();
177 hwsampler_shutdown();
178}
179
180#endif /* CONFIG_64BIT */
181
182int __init oprofile_arch_init(struct oprofile_operations *ops)
19{ 183{
20 ops->backtrace = s390_backtrace; 184 ops->backtrace = s390_backtrace;
185
186#ifdef CONFIG_64BIT
187 return oprofile_hwsampler_init(ops);
188#else
21 return -ENODEV; 189 return -ENODEV;
190#endif
22} 191}
23 192
24void oprofile_arch_exit(void) 193void oprofile_arch_exit(void)
25{ 194{
195#ifdef CONFIG_64BIT
196 oprofile_hwsampler_exit();
197#endif
26} 198}
generated by cgit v1.2.2 (git 2.25.0) at 2025-07-24 17:19:54 -0400