aboutsummaryrefslogtreecommitdiffstats
path: root/arch/x86
diff options
context:
space:
mode:
Diffstat (limited to 'arch/x86')
-rw-r--r--arch/x86/Kconfig87
-rw-r--r--arch/x86/Kconfig.debug20
-rw-r--r--arch/x86/boot/tty.c2
-rw-r--r--arch/x86/boot/video-vga.c4
-rw-r--r--arch/x86/boot/video.c2
-rw-r--r--arch/x86/configs/i386_defconfig4
-rw-r--r--arch/x86/configs/x86_64_defconfig4
-rw-r--r--arch/x86/ia32/ia32_signal.c109
-rw-r--r--arch/x86/include/asm/acpi.h1
-rw-r--r--arch/x86/include/asm/amd_iommu_types.h24
-rw-r--r--arch/x86/include/asm/apic.h1
-rw-r--r--arch/x86/include/asm/bigsmp/apic.h2
-rw-r--r--arch/x86/include/asm/bitops.h10
-rw-r--r--arch/x86/include/asm/bug.h2
-rw-r--r--arch/x86/include/asm/byteorder.h74
-rw-r--r--arch/x86/include/asm/cpufeature.h5
-rw-r--r--arch/x86/include/asm/dma-mapping.h6
-rw-r--r--arch/x86/include/asm/ds.h6
-rw-r--r--arch/x86/include/asm/emergency-restart.h4
-rw-r--r--arch/x86/include/asm/es7000/apic.h79
-rw-r--r--arch/x86/include/asm/es7000/wakecpu.h41
-rw-r--r--arch/x86/include/asm/genapic_32.h19
-rw-r--r--arch/x86/include/asm/genapic_64.h2
-rw-r--r--arch/x86/include/asm/hypervisor.h26
-rw-r--r--arch/x86/include/asm/ia32.h18
-rw-r--r--arch/x86/include/asm/idle.h5
-rw-r--r--arch/x86/include/asm/io.h37
-rw-r--r--arch/x86/include/asm/io_apic.h10
-rw-r--r--arch/x86/include/asm/iomap.h30
-rw-r--r--arch/x86/include/asm/iommu.h1
-rw-r--r--arch/x86/include/asm/irq.h4
-rw-r--r--arch/x86/include/asm/irq_regs_32.h2
-rw-r--r--arch/x86/include/asm/kexec.h31
-rw-r--r--arch/x86/include/asm/mach-default/mach_apic.h2
-rw-r--r--arch/x86/include/asm/mach-default/mach_wakecpu.h24
-rw-r--r--arch/x86/include/asm/mach-default/smpboot_hooks.h8
-rw-r--r--arch/x86/include/asm/mach-generic/mach_apic.h1
-rw-r--r--arch/x86/include/asm/mach-generic/mach_wakecpu.h12
-rw-r--r--arch/x86/include/asm/mmu_context_32.h13
-rw-r--r--arch/x86/include/asm/mmzone_32.h4
-rw-r--r--arch/x86/include/asm/msr-index.h2
-rw-r--r--arch/x86/include/asm/msr.h12
-rw-r--r--arch/x86/include/asm/numaq/wakecpu.h24
-rw-r--r--arch/x86/include/asm/pci.h2
-rw-r--r--arch/x86/include/asm/pci_64.h14
-rw-r--r--arch/x86/include/asm/pgtable-2level.h50
-rw-r--r--arch/x86/include/asm/pgtable-3level.h1
-rw-r--r--arch/x86/include/asm/pgtable.h28
-rw-r--r--arch/x86/include/asm/pgtable_32.h9
-rw-r--r--arch/x86/include/asm/pgtable_64.h28
-rw-r--r--arch/x86/include/asm/prctl.h3
-rw-r--r--arch/x86/include/asm/processor.h4
-rw-r--r--arch/x86/include/asm/ptrace.h2
-rw-r--r--arch/x86/include/asm/reboot.h5
-rw-r--r--arch/x86/include/asm/setup.h7
-rw-r--r--arch/x86/include/asm/sigframe.h70
-rw-r--r--arch/x86/include/asm/signal.h6
-rw-r--r--arch/x86/include/asm/sparsemem.h2
-rw-r--r--arch/x86/include/asm/syscalls.h16
-rw-r--r--arch/x86/include/asm/system.h6
-rw-r--r--arch/x86/include/asm/thread_info.h2
-rw-r--r--arch/x86/include/asm/topology.h2
-rw-r--r--arch/x86/include/asm/trampoline.h7
-rw-r--r--arch/x86/include/asm/traps.h11
-rw-r--r--arch/x86/include/asm/tsc.h8
-rw-r--r--arch/x86/include/asm/uaccess.h4
-rw-r--r--arch/x86/include/asm/uaccess_64.h2
-rw-r--r--arch/x86/include/asm/unistd_64.h4
-rw-r--r--arch/x86/include/asm/uv/bios.h34
-rw-r--r--arch/x86/include/asm/uv/uv_hub.h103
-rw-r--r--arch/x86/include/asm/vmi.h8
-rw-r--r--arch/x86/include/asm/vmware.h27
-rw-r--r--arch/x86/include/asm/xen/hypercall.h6
-rw-r--r--arch/x86/include/asm/xen/hypervisor.h39
-rw-r--r--arch/x86/include/asm/xen/page.h5
-rw-r--r--arch/x86/kernel/Makefile9
-rw-r--r--arch/x86/kernel/acpi/boot.c12
-rw-r--r--arch/x86/kernel/amd_iommu.c56
-rw-r--r--arch/x86/kernel/amd_iommu_init.c13
-rw-r--r--arch/x86/kernel/apic.c129
-rw-r--r--arch/x86/kernel/apm_32.c4
-rw-r--r--arch/x86/kernel/asm-offsets_32.c2
-rw-r--r--arch/x86/kernel/asm-offsets_64.c4
-rw-r--r--arch/x86/kernel/bios_uv.c58
-rw-r--r--arch/x86/kernel/check.c161
-rw-r--r--arch/x86/kernel/cpu/Makefile1
-rw-r--r--arch/x86/kernel/cpu/common.c8
-rw-r--r--arch/x86/kernel/cpu/cpufreq/powernow-k8.c18
-rw-r--r--arch/x86/kernel/cpu/cpufreq/powernow-k8.h17
-rw-r--r--arch/x86/kernel/cpu/hypervisor.c58
-rw-r--r--arch/x86/kernel/cpu/intel_cacheinfo.c17
-rw-r--r--arch/x86/kernel/cpu/mcheck/mce_64.c3
-rw-r--r--arch/x86/kernel/cpu/mtrr/main.c346
-rw-r--r--arch/x86/kernel/cpu/vmware.c112
-rw-r--r--arch/x86/kernel/crash.c70
-rw-r--r--arch/x86/kernel/ds.c81
-rw-r--r--arch/x86/kernel/dumpstack.c319
-rw-r--r--arch/x86/kernel/dumpstack.h39
-rw-r--r--arch/x86/kernel/dumpstack_32.c302
-rw-r--r--arch/x86/kernel/dumpstack_64.c282
-rw-r--r--arch/x86/kernel/e820.c16
-rw-r--r--arch/x86/kernel/early-quirks.c18
-rw-r--r--arch/x86/kernel/early_printk.c47
-rw-r--r--arch/x86/kernel/entry_32.S1
-rw-r--r--arch/x86/kernel/entry_64.S193
-rw-r--r--arch/x86/kernel/es7000_32.c71
-rw-r--r--arch/x86/kernel/genapic_64.c4
-rw-r--r--arch/x86/kernel/genx2apic_uv_x.c111
-rw-r--r--arch/x86/kernel/head.c1
-rw-r--r--arch/x86/kernel/head32.c3
-rw-r--r--arch/x86/kernel/head64.c3
-rw-r--r--arch/x86/kernel/hpet.c8
-rw-r--r--arch/x86/kernel/i387.c2
-rw-r--r--arch/x86/kernel/init_task.c1
-rw-r--r--arch/x86/kernel/io_apic.c65
-rw-r--r--arch/x86/kernel/irq_64.c24
-rw-r--r--arch/x86/kernel/kvmclock.c2
-rw-r--r--arch/x86/kernel/machine_kexec_32.c104
-rw-r--r--arch/x86/kernel/microcode_amd.c232
-rw-r--r--arch/x86/kernel/microcode_core.c25
-rw-r--r--arch/x86/kernel/microcode_intel.c8
-rw-r--r--arch/x86/kernel/mpparse.c24
-rw-r--r--arch/x86/kernel/nmi.c58
-rw-r--r--arch/x86/kernel/numaq_32.c10
-rw-r--r--arch/x86/kernel/paravirt-spinlocks.c3
-rw-r--r--arch/x86/kernel/pci-calgary_64.c2
-rw-r--r--arch/x86/kernel/pci-dma.c4
-rw-r--r--arch/x86/kernel/pci-gart_64.c6
-rw-r--r--arch/x86/kernel/process.c17
-rw-r--r--arch/x86/kernel/reboot.c135
-rw-r--r--arch/x86/kernel/relocate_kernel_32.S115
-rw-r--r--arch/x86/kernel/setup.c182
-rw-r--r--arch/x86/kernel/sigframe.h42
-rw-r--r--arch/x86/kernel/signal.c (renamed from arch/x86/kernel/signal_32.c)567
-rw-r--r--arch/x86/kernel/signal_64.c516
-rw-r--r--arch/x86/kernel/smp.c13
-rw-r--r--arch/x86/kernel/smpboot.c25
-rw-r--r--arch/x86/kernel/time_64.c4
-rw-r--r--arch/x86/kernel/tlb_32.c11
-rw-r--r--arch/x86/kernel/tlb_uv.c4
-rw-r--r--arch/x86/kernel/trampoline.c19
-rw-r--r--arch/x86/kernel/traps.c34
-rw-r--r--arch/x86/kernel/tsc.c42
-rw-r--r--arch/x86/kernel/tsc_sync.c12
-rw-r--r--arch/x86/kernel/vmi_32.c135
-rw-r--r--arch/x86/kernel/vsyscall_64.c9
-rw-r--r--arch/x86/kernel/xsave.c2
-rw-r--r--arch/x86/kvm/Kconfig2
-rw-r--r--arch/x86/kvm/i8254.c4
-rw-r--r--arch/x86/kvm/mmu.c4
-rw-r--r--arch/x86/kvm/paging_tmpl.h1
-rw-r--r--arch/x86/kvm/vmx.c7
-rw-r--r--arch/x86/kvm/vmx.h1
-rw-r--r--arch/x86/mach-generic/bigsmp.c1
-rw-r--r--arch/x86/mach-generic/default.c1
-rw-r--r--arch/x86/mach-generic/es7000.c14
-rw-r--r--arch/x86/mach-generic/probe.c16
-rw-r--r--arch/x86/mach-generic/summit.c1
-rw-r--r--arch/x86/mach-voyager/voyager_smp.c16
-rw-r--r--arch/x86/mm/fault.c11
-rw-r--r--arch/x86/mm/init_32.c32
-rw-r--r--arch/x86/mm/init_64.c2
-rw-r--r--arch/x86/mm/ioremap.c3
-rw-r--r--arch/x86/mm/numa_32.c35
-rw-r--r--arch/x86/mm/pat.c236
-rw-r--r--arch/x86/oprofile/nmi_int.c5
-rw-r--r--arch/x86/oprofile/op_model_ppro.c6
-rw-r--r--arch/x86/pci/common.c17
-rw-r--r--arch/x86/pci/direct.c4
-rw-r--r--arch/x86/pci/fixup.c25
-rw-r--r--arch/x86/pci/pci.h1
-rw-r--r--arch/x86/power/hibernate_32.c4
-rw-r--r--arch/x86/xen/enlighten.c17
-rw-r--r--arch/x86/xen/mmu.c38
-rw-r--r--arch/x86/xen/multicalls.c2
-rw-r--r--arch/x86/xen/setup.c9
-rw-r--r--arch/x86/xen/smp.c2
-rw-r--r--arch/x86/xen/xen-ops.h2
178 files changed, 3638 insertions, 3221 deletions
diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
index 4cf0ab13d187..1cbec0269d39 100644
--- a/arch/x86/Kconfig
+++ b/arch/x86/Kconfig
@@ -19,6 +19,8 @@ config X86_64
19config X86 19config X86
20 def_bool y 20 def_bool y
21 select HAVE_AOUT if X86_32 21 select HAVE_AOUT if X86_32
22 select HAVE_READQ
23 select HAVE_WRITEQ
22 select HAVE_UNSTABLE_SCHED_CLOCK 24 select HAVE_UNSTABLE_SCHED_CLOCK
23 select HAVE_IDE 25 select HAVE_IDE
24 select HAVE_OPROFILE 26 select HAVE_OPROFILE
@@ -87,6 +89,10 @@ config GENERIC_IOMAP
87config GENERIC_BUG 89config GENERIC_BUG
88 def_bool y 90 def_bool y
89 depends on BUG 91 depends on BUG
92 select GENERIC_BUG_RELATIVE_POINTERS if X86_64
93
94config GENERIC_BUG_RELATIVE_POINTERS
95 bool
90 96
91config GENERIC_HWEIGHT 97config GENERIC_HWEIGHT
92 def_bool y 98 def_bool y
@@ -167,9 +173,12 @@ config GENERIC_PENDING_IRQ
167config X86_SMP 173config X86_SMP
168 bool 174 bool
169 depends on SMP && ((X86_32 && !X86_VOYAGER) || X86_64) 175 depends on SMP && ((X86_32 && !X86_VOYAGER) || X86_64)
170 select USE_GENERIC_SMP_HELPERS
171 default y 176 default y
172 177
178config USE_GENERIC_SMP_HELPERS
179 def_bool y
180 depends on SMP
181
173config X86_32_SMP 182config X86_32_SMP
174 def_bool y 183 def_bool y
175 depends on X86_32 && SMP 184 depends on X86_32 && SMP
@@ -239,21 +248,13 @@ config X86_FIND_SMP_CONFIG
239 def_bool y 248 def_bool y
240 depends on X86_MPPARSE || X86_VOYAGER 249 depends on X86_MPPARSE || X86_VOYAGER
241 250
242if ACPI
243config X86_MPPARSE 251config X86_MPPARSE
244 def_bool y 252 bool "Enable MPS table" if ACPI
245 bool "Enable MPS table" 253 default y
246 depends on X86_LOCAL_APIC 254 depends on X86_LOCAL_APIC
247 help 255 help
248 For old smp systems that do not have proper acpi support. Newer systems 256 For old smp systems that do not have proper acpi support. Newer systems
249 (esp with 64bit cpus) with acpi support, MADT and DSDT will override it 257 (esp with 64bit cpus) with acpi support, MADT and DSDT will override it
250endif
251
252if !ACPI
253config X86_MPPARSE
254 def_bool y
255 depends on X86_LOCAL_APIC
256endif
257 258
258choice 259choice
259 prompt "Subarchitecture Type" 260 prompt "Subarchitecture Type"
@@ -462,10 +463,6 @@ config X86_CYCLONE_TIMER
462 def_bool y 463 def_bool y
463 depends on X86_GENERICARCH 464 depends on X86_GENERICARCH
464 465
465config ES7000_CLUSTERED_APIC
466 def_bool y
467 depends on SMP && X86_ES7000 && MPENTIUMIII
468
469source "arch/x86/Kconfig.cpu" 466source "arch/x86/Kconfig.cpu"
470 467
471config HPET_TIMER 468config HPET_TIMER
@@ -657,6 +654,30 @@ config X86_VISWS_APIC
657 def_bool y 654 def_bool y
658 depends on X86_32 && X86_VISWS 655 depends on X86_32 && X86_VISWS
659 656
657config X86_REROUTE_FOR_BROKEN_BOOT_IRQS
658 bool "Reroute for broken boot IRQs"
659 default n
660 depends on X86_IO_APIC
661 help
662 This option enables a workaround that fixes a source of
663 spurious interrupts. This is recommended when threaded
664 interrupt handling is used on systems where the generation of
665 superfluous "boot interrupts" cannot be disabled.
666
667 Some chipsets generate a legacy INTx "boot IRQ" when the IRQ
668 entry in the chipset's IO-APIC is masked (as, e.g. the RT
669 kernel does during interrupt handling). On chipsets where this
670 boot IRQ generation cannot be disabled, this workaround keeps
671 the original IRQ line masked so that only the equivalent "boot
672 IRQ" is delivered to the CPUs. The workaround also tells the
673 kernel to set up the IRQ handler on the boot IRQ line. In this
674 way only one interrupt is delivered to the kernel. Otherwise
675 the spurious second interrupt may cause the kernel to bring
676 down (vital) interrupt lines.
677
678 Only affects "broken" chipsets. Interrupt sharing may be
679 increased on these systems.
680
660config X86_MCE 681config X86_MCE
661 bool "Machine Check Exception" 682 bool "Machine Check Exception"
662 depends on !X86_VOYAGER 683 depends on !X86_VOYAGER
@@ -953,24 +974,37 @@ config X86_PAE
953config ARCH_PHYS_ADDR_T_64BIT 974config ARCH_PHYS_ADDR_T_64BIT
954 def_bool X86_64 || X86_PAE 975 def_bool X86_64 || X86_PAE
955 976
977config DIRECT_GBPAGES
978 bool "Enable 1GB pages for kernel pagetables" if EMBEDDED
979 default y
980 depends on X86_64
981 help
982 Allow the kernel linear mapping to use 1GB pages on CPUs that
983 support it. This can improve the kernel's performance a tiny bit by
984 reducing TLB pressure. If in doubt, say "Y".
985
956# Common NUMA Features 986# Common NUMA Features
957config NUMA 987config NUMA
958 bool "Numa Memory Allocation and Scheduler Support (EXPERIMENTAL)" 988 bool "Numa Memory Allocation and Scheduler Support"
959 depends on SMP 989 depends on SMP
960 depends on X86_64 || (X86_32 && HIGHMEM64G && (X86_NUMAQ || X86_BIGSMP || X86_SUMMIT && ACPI) && EXPERIMENTAL) 990 depends on X86_64 || (X86_32 && HIGHMEM64G && (X86_NUMAQ || X86_BIGSMP || X86_SUMMIT && ACPI) && EXPERIMENTAL)
961 default n if X86_PC 991 default n if X86_PC
962 default y if (X86_NUMAQ || X86_SUMMIT || X86_BIGSMP) 992 default y if (X86_NUMAQ || X86_SUMMIT || X86_BIGSMP)
963 help 993 help
964 Enable NUMA (Non Uniform Memory Access) support. 994 Enable NUMA (Non Uniform Memory Access) support.
995
965 The kernel will try to allocate memory used by a CPU on the 996 The kernel will try to allocate memory used by a CPU on the
966 local memory controller of the CPU and add some more 997 local memory controller of the CPU and add some more
967 NUMA awareness to the kernel. 998 NUMA awareness to the kernel.
968 999
969 For 32-bit this is currently highly experimental and should be only 1000 For 64-bit this is recommended if the system is Intel Core i7
970 used for kernel development. It might also cause boot failures. 1001 (or later), AMD Opteron, or EM64T NUMA.
971 For 64-bit this is recommended on all multiprocessor Opteron systems. 1002
972 If the system is EM64T, you should say N unless your system is 1003 For 32-bit this is only needed on (rare) 32-bit-only platforms
973 EM64T NUMA. 1004 that support NUMA topologies, such as NUMAQ / Summit, or if you
1005 boot a 32-bit kernel on a 64-bit NUMA platform.
1006
1007 Otherwise, you should say N.
974 1008
975comment "NUMA (Summit) requires SMP, 64GB highmem support, ACPI" 1009comment "NUMA (Summit) requires SMP, 64GB highmem support, ACPI"
976 depends on X86_32 && X86_SUMMIT && (!HIGHMEM64G || !ACPI) 1010 depends on X86_32 && X86_SUMMIT && (!HIGHMEM64G || !ACPI)
@@ -1490,6 +1524,10 @@ config ARCH_ENABLE_MEMORY_HOTPLUG
1490 def_bool y 1524 def_bool y
1491 depends on X86_64 || (X86_32 && HIGHMEM) 1525 depends on X86_64 || (X86_32 && HIGHMEM)
1492 1526
1527config ARCH_ENABLE_MEMORY_HOTREMOVE
1528 def_bool y
1529 depends on MEMORY_HOTPLUG
1530
1493config HAVE_ARCH_EARLY_PFN_TO_NID 1531config HAVE_ARCH_EARLY_PFN_TO_NID
1494 def_bool X86_64 1532 def_bool X86_64
1495 depends on NUMA 1533 depends on NUMA
@@ -1629,13 +1667,6 @@ config APM_ALLOW_INTS
1629 many of the newer IBM Thinkpads. If you experience hangs when you 1667 many of the newer IBM Thinkpads. If you experience hangs when you
1630 suspend, try setting this to Y. Otherwise, say N. 1668 suspend, try setting this to Y. Otherwise, say N.
1631 1669
1632config APM_REAL_MODE_POWER_OFF
1633 bool "Use real mode APM BIOS call to power off"
1634 help
1635 Use real mode APM BIOS calls to switch off the computer. This is
1636 a work-around for a number of buggy BIOSes. Switch this option on if
1637 your computer crashes instead of powering off properly.
1638
1639endif # APM 1670endif # APM
1640 1671
1641source "arch/x86/kernel/cpu/cpufreq/Kconfig" 1672source "arch/x86/kernel/cpu/cpufreq/Kconfig"
diff --git a/arch/x86/Kconfig.debug b/arch/x86/Kconfig.debug
index 2a3dfbd5e677..4ee768660f75 100644
--- a/arch/x86/Kconfig.debug
+++ b/arch/x86/Kconfig.debug
@@ -114,18 +114,6 @@ config DEBUG_RODATA
114 data. This is recommended so that we can catch kernel bugs sooner. 114 data. This is recommended so that we can catch kernel bugs sooner.
115 If in doubt, say "Y". 115 If in doubt, say "Y".
116 116
117config DIRECT_GBPAGES
118 bool "Enable gbpages-mapped kernel pagetables"
119 depends on DEBUG_KERNEL && EXPERIMENTAL && X86_64
120 help
121 Enable gigabyte pages support (if the CPU supports it). This can
122 improve the kernel's performance a tiny bit by reducing TLB
123 pressure.
124
125 This is experimental code.
126
127 If in doubt, say "N".
128
129config DEBUG_RODATA_TEST 117config DEBUG_RODATA_TEST
130 bool "Testcase for the DEBUG_RODATA feature" 118 bool "Testcase for the DEBUG_RODATA feature"
131 depends on DEBUG_RODATA 119 depends on DEBUG_RODATA
@@ -307,10 +295,10 @@ config OPTIMIZE_INLINING
307 developers have marked 'inline'. Doing so takes away freedom from gcc to 295 developers have marked 'inline'. Doing so takes away freedom from gcc to
308 do what it thinks is best, which is desirable for the gcc 3.x series of 296 do what it thinks is best, which is desirable for the gcc 3.x series of
309 compilers. The gcc 4.x series have a rewritten inlining algorithm and 297 compilers. The gcc 4.x series have a rewritten inlining algorithm and
310 disabling this option will generate a smaller kernel there. Hopefully 298 enabling this option will generate a smaller kernel there. Hopefully
311 this algorithm is so good that allowing gcc4 to make the decision can 299 this algorithm is so good that allowing gcc 4.x and above to make the
312 become the default in the future, until then this option is there to 300 decision will become the default in the future. Until then this option
313 test gcc for this. 301 is there to test gcc for this.
314 302
315 If unsure, say N. 303 If unsure, say N.
316 304
diff --git a/arch/x86/boot/tty.c b/arch/x86/boot/tty.c
index 0be77b39328a..7e8e8b25f5f6 100644
--- a/arch/x86/boot/tty.c
+++ b/arch/x86/boot/tty.c
@@ -74,7 +74,7 @@ static int kbd_pending(void)
74{ 74{
75 u8 pending; 75 u8 pending;
76 asm volatile("int $0x16; setnz %0" 76 asm volatile("int $0x16; setnz %0"
77 : "=rm" (pending) 77 : "=qm" (pending)
78 : "a" (0x0100)); 78 : "a" (0x0100));
79 return pending; 79 return pending;
80} 80}
diff --git a/arch/x86/boot/video-vga.c b/arch/x86/boot/video-vga.c
index b939cb476dec..5d4742ed4aa2 100644
--- a/arch/x86/boot/video-vga.c
+++ b/arch/x86/boot/video-vga.c
@@ -34,7 +34,7 @@ static struct mode_info cga_modes[] = {
34 { VIDEO_80x25, 80, 25, 0 }, 34 { VIDEO_80x25, 80, 25, 0 },
35}; 35};
36 36
37__videocard video_vga; 37static __videocard video_vga;
38 38
39/* Set basic 80x25 mode */ 39/* Set basic 80x25 mode */
40static u8 vga_set_basic_mode(void) 40static u8 vga_set_basic_mode(void)
@@ -259,7 +259,7 @@ static int vga_probe(void)
259 return mode_count[adapter]; 259 return mode_count[adapter];
260} 260}
261 261
262__videocard video_vga = { 262static __videocard video_vga = {
263 .card_name = "VGA", 263 .card_name = "VGA",
264 .probe = vga_probe, 264 .probe = vga_probe,
265 .set_mode = vga_set_mode, 265 .set_mode = vga_set_mode,
diff --git a/arch/x86/boot/video.c b/arch/x86/boot/video.c
index 83598b23093a..3bef2c1febe9 100644
--- a/arch/x86/boot/video.c
+++ b/arch/x86/boot/video.c
@@ -226,7 +226,7 @@ static unsigned int mode_menu(void)
226 226
227#ifdef CONFIG_VIDEO_RETAIN 227#ifdef CONFIG_VIDEO_RETAIN
228/* Save screen content to the heap */ 228/* Save screen content to the heap */
229struct saved_screen { 229static struct saved_screen {
230 int x, y; 230 int x, y;
231 int curx, cury; 231 int curx, cury;
232 u16 *data; 232 u16 *data;
diff --git a/arch/x86/configs/i386_defconfig b/arch/x86/configs/i386_defconfig
index 13b8c86ae985..b30a08ed8eb4 100644
--- a/arch/x86/configs/i386_defconfig
+++ b/arch/x86/configs/i386_defconfig
@@ -77,7 +77,7 @@ CONFIG_AUDIT=y
77CONFIG_AUDITSYSCALL=y 77CONFIG_AUDITSYSCALL=y
78CONFIG_AUDIT_TREE=y 78CONFIG_AUDIT_TREE=y
79# CONFIG_IKCONFIG is not set 79# CONFIG_IKCONFIG is not set
80CONFIG_LOG_BUF_SHIFT=17 80CONFIG_LOG_BUF_SHIFT=18
81CONFIG_CGROUPS=y 81CONFIG_CGROUPS=y
82# CONFIG_CGROUP_DEBUG is not set 82# CONFIG_CGROUP_DEBUG is not set
83CONFIG_CGROUP_NS=y 83CONFIG_CGROUP_NS=y
@@ -298,7 +298,7 @@ CONFIG_KEXEC=y
298CONFIG_CRASH_DUMP=y 298CONFIG_CRASH_DUMP=y
299# CONFIG_KEXEC_JUMP is not set 299# CONFIG_KEXEC_JUMP is not set
300CONFIG_PHYSICAL_START=0x1000000 300CONFIG_PHYSICAL_START=0x1000000
301CONFIG_RELOCATABLE=y 301# CONFIG_RELOCATABLE is not set
302CONFIG_PHYSICAL_ALIGN=0x200000 302CONFIG_PHYSICAL_ALIGN=0x200000
303CONFIG_HOTPLUG_CPU=y 303CONFIG_HOTPLUG_CPU=y
304# CONFIG_COMPAT_VDSO is not set 304# CONFIG_COMPAT_VDSO is not set
diff --git a/arch/x86/configs/x86_64_defconfig b/arch/x86/configs/x86_64_defconfig
index f0a03d7a7d63..0e7dbc0a3e46 100644
--- a/arch/x86/configs/x86_64_defconfig
+++ b/arch/x86/configs/x86_64_defconfig
@@ -77,7 +77,7 @@ CONFIG_AUDIT=y
77CONFIG_AUDITSYSCALL=y 77CONFIG_AUDITSYSCALL=y
78CONFIG_AUDIT_TREE=y 78CONFIG_AUDIT_TREE=y
79# CONFIG_IKCONFIG is not set 79# CONFIG_IKCONFIG is not set
80CONFIG_LOG_BUF_SHIFT=17 80CONFIG_LOG_BUF_SHIFT=18
81CONFIG_CGROUPS=y 81CONFIG_CGROUPS=y
82# CONFIG_CGROUP_DEBUG is not set 82# CONFIG_CGROUP_DEBUG is not set
83CONFIG_CGROUP_NS=y 83CONFIG_CGROUP_NS=y
@@ -298,7 +298,7 @@ CONFIG_SCHED_HRTICK=y
298CONFIG_KEXEC=y 298CONFIG_KEXEC=y
299CONFIG_CRASH_DUMP=y 299CONFIG_CRASH_DUMP=y
300CONFIG_PHYSICAL_START=0x1000000 300CONFIG_PHYSICAL_START=0x1000000
301CONFIG_RELOCATABLE=y 301# CONFIG_RELOCATABLE is not set
302CONFIG_PHYSICAL_ALIGN=0x200000 302CONFIG_PHYSICAL_ALIGN=0x200000
303CONFIG_HOTPLUG_CPU=y 303CONFIG_HOTPLUG_CPU=y
304# CONFIG_COMPAT_VDSO is not set 304# CONFIG_COMPAT_VDSO is not set
diff --git a/arch/x86/ia32/ia32_signal.c b/arch/x86/ia32/ia32_signal.c
index 4bc02b23674b..b195f85526e3 100644
--- a/arch/x86/ia32/ia32_signal.c
+++ b/arch/x86/ia32/ia32_signal.c
@@ -32,6 +32,8 @@
32#include <asm/proto.h> 32#include <asm/proto.h>
33#include <asm/vdso.h> 33#include <asm/vdso.h>
34 34
35#include <asm/sigframe.h>
36
35#define DEBUG_SIG 0 37#define DEBUG_SIG 0
36 38
37#define _BLOCKABLE (~(sigmask(SIGKILL) | sigmask(SIGSTOP))) 39#define _BLOCKABLE (~(sigmask(SIGKILL) | sigmask(SIGSTOP)))
@@ -41,7 +43,6 @@
41 X86_EFLAGS_ZF | X86_EFLAGS_AF | X86_EFLAGS_PF | \ 43 X86_EFLAGS_ZF | X86_EFLAGS_AF | X86_EFLAGS_PF | \
42 X86_EFLAGS_CF) 44 X86_EFLAGS_CF)
43 45
44asmlinkage int do_signal(struct pt_regs *regs, sigset_t *oldset);
45void signal_fault(struct pt_regs *regs, void __user *frame, char *where); 46void signal_fault(struct pt_regs *regs, void __user *frame, char *where);
46 47
47int copy_siginfo_to_user32(compat_siginfo_t __user *to, siginfo_t *from) 48int copy_siginfo_to_user32(compat_siginfo_t __user *to, siginfo_t *from)
@@ -173,47 +174,28 @@ asmlinkage long sys32_sigaltstack(const stack_ia32_t __user *uss_ptr,
173/* 174/*
174 * Do a signal return; undo the signal stack. 175 * Do a signal return; undo the signal stack.
175 */ 176 */
177#define COPY(x) { \
178 err |= __get_user(regs->x, &sc->x); \
179}
176 180
177struct sigframe 181#define COPY_SEG_CPL3(seg) { \
178{ 182 unsigned short tmp; \
179 u32 pretcode; 183 err |= __get_user(tmp, &sc->seg); \
180 int sig; 184 regs->seg = tmp | 3; \
181 struct sigcontext_ia32 sc;
182 struct _fpstate_ia32 fpstate_unused; /* look at kernel/sigframe.h */
183 unsigned int extramask[_COMPAT_NSIG_WORDS-1];
184 char retcode[8];
185 /* fp state follows here */
186};
187
188struct rt_sigframe
189{
190 u32 pretcode;
191 int sig;
192 u32 pinfo;
193 u32 puc;
194 compat_siginfo_t info;
195 struct ucontext_ia32 uc;
196 char retcode[8];
197 /* fp state follows here */
198};
199
200#define COPY(x) { \
201 unsigned int reg; \
202 err |= __get_user(reg, &sc->x); \
203 regs->x = reg; \
204} 185}
205 186
206#define RELOAD_SEG(seg,mask) \ 187#define RELOAD_SEG(seg) { \
207 { unsigned int cur; \ 188 unsigned int cur, pre; \
208 unsigned short pre; \ 189 err |= __get_user(pre, &sc->seg); \
209 err |= __get_user(pre, &sc->seg); \ 190 savesegment(seg, cur); \
210 savesegment(seg, cur); \ 191 pre |= 3; \
211 pre |= mask; \ 192 if (pre != cur) \
212 if (pre != cur) loadsegment(seg, pre); } 193 loadsegment(seg, pre); \
194}
213 195
214static int ia32_restore_sigcontext(struct pt_regs *regs, 196static int ia32_restore_sigcontext(struct pt_regs *regs,
215 struct sigcontext_ia32 __user *sc, 197 struct sigcontext_ia32 __user *sc,
216 unsigned int *peax) 198 unsigned int *pax)
217{ 199{
218 unsigned int tmpflags, gs, oldgs, err = 0; 200 unsigned int tmpflags, gs, oldgs, err = 0;
219 void __user *buf; 201 void __user *buf;
@@ -240,18 +222,16 @@ static int ia32_restore_sigcontext(struct pt_regs *regs,
240 if (gs != oldgs) 222 if (gs != oldgs)
241 load_gs_index(gs); 223 load_gs_index(gs);
242 224
243 RELOAD_SEG(fs, 3); 225 RELOAD_SEG(fs);
244 RELOAD_SEG(ds, 3); 226 RELOAD_SEG(ds);
245 RELOAD_SEG(es, 3); 227 RELOAD_SEG(es);
246 228
247 COPY(di); COPY(si); COPY(bp); COPY(sp); COPY(bx); 229 COPY(di); COPY(si); COPY(bp); COPY(sp); COPY(bx);
248 COPY(dx); COPY(cx); COPY(ip); 230 COPY(dx); COPY(cx); COPY(ip);
249 /* Don't touch extended registers */ 231 /* Don't touch extended registers */
250 232
251 err |= __get_user(regs->cs, &sc->cs); 233 COPY_SEG_CPL3(cs);
252 regs->cs |= 3; 234 COPY_SEG_CPL3(ss);
253 err |= __get_user(regs->ss, &sc->ss);
254 regs->ss |= 3;
255 235
256 err |= __get_user(tmpflags, &sc->flags); 236 err |= __get_user(tmpflags, &sc->flags);
257 regs->flags = (regs->flags & ~FIX_EFLAGS) | (tmpflags & FIX_EFLAGS); 237 regs->flags = (regs->flags & ~FIX_EFLAGS) | (tmpflags & FIX_EFLAGS);
@@ -262,15 +242,13 @@ static int ia32_restore_sigcontext(struct pt_regs *regs,
262 buf = compat_ptr(tmp); 242 buf = compat_ptr(tmp);
263 err |= restore_i387_xstate_ia32(buf); 243 err |= restore_i387_xstate_ia32(buf);
264 244
265 err |= __get_user(tmp, &sc->ax); 245 err |= __get_user(*pax, &sc->ax);
266 *peax = tmp;
267
268 return err; 246 return err;
269} 247}
270 248
271asmlinkage long sys32_sigreturn(struct pt_regs *regs) 249asmlinkage long sys32_sigreturn(struct pt_regs *regs)
272{ 250{
273 struct sigframe __user *frame = (struct sigframe __user *)(regs->sp-8); 251 struct sigframe_ia32 __user *frame = (struct sigframe_ia32 __user *)(regs->sp-8);
274 sigset_t set; 252 sigset_t set;
275 unsigned int ax; 253 unsigned int ax;
276 254
@@ -300,12 +278,12 @@ badframe:
300 278
301asmlinkage long sys32_rt_sigreturn(struct pt_regs *regs) 279asmlinkage long sys32_rt_sigreturn(struct pt_regs *regs)
302{ 280{
303 struct rt_sigframe __user *frame; 281 struct rt_sigframe_ia32 __user *frame;
304 sigset_t set; 282 sigset_t set;
305 unsigned int ax; 283 unsigned int ax;
306 struct pt_regs tregs; 284 struct pt_regs tregs;
307 285
308 frame = (struct rt_sigframe __user *)(regs->sp - 4); 286 frame = (struct rt_sigframe_ia32 __user *)(regs->sp - 4);
309 287
310 if (!access_ok(VERIFY_READ, frame, sizeof(*frame))) 288 if (!access_ok(VERIFY_READ, frame, sizeof(*frame)))
311 goto badframe; 289 goto badframe;
@@ -359,20 +337,15 @@ static int ia32_setup_sigcontext(struct sigcontext_ia32 __user *sc,
359 err |= __put_user(regs->dx, &sc->dx); 337 err |= __put_user(regs->dx, &sc->dx);
360 err |= __put_user(regs->cx, &sc->cx); 338 err |= __put_user(regs->cx, &sc->cx);
361 err |= __put_user(regs->ax, &sc->ax); 339 err |= __put_user(regs->ax, &sc->ax);
362 err |= __put_user(regs->cs, &sc->cs);
363 err |= __put_user(regs->ss, &sc->ss);
364 err |= __put_user(current->thread.trap_no, &sc->trapno); 340 err |= __put_user(current->thread.trap_no, &sc->trapno);
365 err |= __put_user(current->thread.error_code, &sc->err); 341 err |= __put_user(current->thread.error_code, &sc->err);
366 err |= __put_user(regs->ip, &sc->ip); 342 err |= __put_user(regs->ip, &sc->ip);
343 err |= __put_user(regs->cs, (unsigned int __user *)&sc->cs);
367 err |= __put_user(regs->flags, &sc->flags); 344 err |= __put_user(regs->flags, &sc->flags);
368 err |= __put_user(regs->sp, &sc->sp_at_signal); 345 err |= __put_user(regs->sp, &sc->sp_at_signal);
346 err |= __put_user(regs->ss, (unsigned int __user *)&sc->ss);
369 347
370 tmp = save_i387_xstate_ia32(fpstate); 348 err |= __put_user(ptr_to_compat(fpstate), &sc->fpstate);
371 if (tmp < 0)
372 err = -EFAULT;
373 else
374 err |= __put_user(ptr_to_compat(tmp ? fpstate : NULL),
375 &sc->fpstate);
376 349
377 /* non-iBCS2 extensions.. */ 350 /* non-iBCS2 extensions.. */
378 err |= __put_user(mask, &sc->oldmask); 351 err |= __put_user(mask, &sc->oldmask);
@@ -400,7 +373,7 @@ static void __user *get_sigframe(struct k_sigaction *ka, struct pt_regs *regs,
400 } 373 }
401 374
402 /* This is the legacy signal stack switching. */ 375 /* This is the legacy signal stack switching. */
403 else if ((regs->ss & 0xffff) != __USER_DS && 376 else if ((regs->ss & 0xffff) != __USER32_DS &&
404 !(ka->sa.sa_flags & SA_RESTORER) && 377 !(ka->sa.sa_flags & SA_RESTORER) &&
405 ka->sa.sa_restorer) 378 ka->sa.sa_restorer)
406 sp = (unsigned long) ka->sa.sa_restorer; 379 sp = (unsigned long) ka->sa.sa_restorer;
@@ -408,6 +381,8 @@ static void __user *get_sigframe(struct k_sigaction *ka, struct pt_regs *regs,
408 if (used_math()) { 381 if (used_math()) {
409 sp = sp - sig_xstate_ia32_size; 382 sp = sp - sig_xstate_ia32_size;
410 *fpstate = (struct _fpstate_ia32 *) sp; 383 *fpstate = (struct _fpstate_ia32 *) sp;
384 if (save_i387_xstate_ia32(*fpstate) < 0)
385 return (void __user *) -1L;
411 } 386 }
412 387
413 sp -= frame_size; 388 sp -= frame_size;
@@ -420,7 +395,7 @@ static void __user *get_sigframe(struct k_sigaction *ka, struct pt_regs *regs,
420int ia32_setup_frame(int sig, struct k_sigaction *ka, 395int ia32_setup_frame(int sig, struct k_sigaction *ka,
421 compat_sigset_t *set, struct pt_regs *regs) 396 compat_sigset_t *set, struct pt_regs *regs)
422{ 397{
423 struct sigframe __user *frame; 398 struct sigframe_ia32 __user *frame;
424 void __user *restorer; 399 void __user *restorer;
425 int err = 0; 400 int err = 0;
426 void __user *fpstate = NULL; 401 void __user *fpstate = NULL;
@@ -430,12 +405,10 @@ int ia32_setup_frame(int sig, struct k_sigaction *ka,
430 u16 poplmovl; 405 u16 poplmovl;
431 u32 val; 406 u32 val;
432 u16 int80; 407 u16 int80;
433 u16 pad;
434 } __attribute__((packed)) code = { 408 } __attribute__((packed)) code = {
435 0xb858, /* popl %eax ; movl $...,%eax */ 409 0xb858, /* popl %eax ; movl $...,%eax */
436 __NR_ia32_sigreturn, 410 __NR_ia32_sigreturn,
437 0x80cd, /* int $0x80 */ 411 0x80cd, /* int $0x80 */
438 0,
439 }; 412 };
440 413
441 frame = get_sigframe(ka, regs, sizeof(*frame), &fpstate); 414 frame = get_sigframe(ka, regs, sizeof(*frame), &fpstate);
@@ -471,7 +444,7 @@ int ia32_setup_frame(int sig, struct k_sigaction *ka,
471 * These are actually not used anymore, but left because some 444 * These are actually not used anymore, but left because some
472 * gdb versions depend on them as a marker. 445 * gdb versions depend on them as a marker.
473 */ 446 */
474 err |= __copy_to_user(frame->retcode, &code, 8); 447 err |= __put_user(*((u64 *)&code), (u64 *)frame->retcode);
475 if (err) 448 if (err)
476 return -EFAULT; 449 return -EFAULT;
477 450
@@ -501,7 +474,7 @@ int ia32_setup_frame(int sig, struct k_sigaction *ka,
501int ia32_setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info, 474int ia32_setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info,
502 compat_sigset_t *set, struct pt_regs *regs) 475 compat_sigset_t *set, struct pt_regs *regs)
503{ 476{
504 struct rt_sigframe __user *frame; 477 struct rt_sigframe_ia32 __user *frame;
505 void __user *restorer; 478 void __user *restorer;
506 int err = 0; 479 int err = 0;
507 void __user *fpstate = NULL; 480 void __user *fpstate = NULL;
@@ -511,8 +484,7 @@ int ia32_setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info,
511 u8 movl; 484 u8 movl;
512 u32 val; 485 u32 val;
513 u16 int80; 486 u16 int80;
514 u16 pad; 487 u8 pad;
515 u8 pad2;
516 } __attribute__((packed)) code = { 488 } __attribute__((packed)) code = {
517 0xb8, 489 0xb8,
518 __NR_ia32_rt_sigreturn, 490 __NR_ia32_rt_sigreturn,
@@ -559,7 +531,7 @@ int ia32_setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info,
559 * Not actually used anymore, but left because some gdb 531 * Not actually used anymore, but left because some gdb
560 * versions need it. 532 * versions need it.
561 */ 533 */
562 err |= __copy_to_user(frame->retcode, &code, 8); 534 err |= __put_user(*((u64 *)&code), (u64 *)frame->retcode);
563 if (err) 535 if (err)
564 return -EFAULT; 536 return -EFAULT;
565 537
@@ -572,11 +544,6 @@ int ia32_setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info,
572 regs->dx = (unsigned long) &frame->info; 544 regs->dx = (unsigned long) &frame->info;
573 regs->cx = (unsigned long) &frame->uc; 545 regs->cx = (unsigned long) &frame->uc;
574 546
575 /* Make -mregparm=3 work */
576 regs->ax = sig;
577 regs->dx = (unsigned long) &frame->info;
578 regs->cx = (unsigned long) &frame->uc;
579
580 loadsegment(ds, __USER32_DS); 547 loadsegment(ds, __USER32_DS);
581 loadsegment(es, __USER32_DS); 548 loadsegment(es, __USER32_DS);
582 549
diff --git a/arch/x86/include/asm/acpi.h b/arch/x86/include/asm/acpi.h
index 8d676d8ecde9..9830681446ad 100644
--- a/arch/x86/include/asm/acpi.h
+++ b/arch/x86/include/asm/acpi.h
@@ -113,7 +113,6 @@ static inline void acpi_disable_pci(void)
113 acpi_pci_disabled = 1; 113 acpi_pci_disabled = 1;
114 acpi_noirq_set(); 114 acpi_noirq_set();
115} 115}
116extern int acpi_irq_balance_set(char *str);
117 116
118/* routines for saving/restoring kernel state */ 117/* routines for saving/restoring kernel state */
119extern int acpi_save_state_mem(void); 118extern int acpi_save_state_mem(void);
diff --git a/arch/x86/include/asm/amd_iommu_types.h b/arch/x86/include/asm/amd_iommu_types.h
index 1a30c0440c6b..ac302a2fa339 100644
--- a/arch/x86/include/asm/amd_iommu_types.h
+++ b/arch/x86/include/asm/amd_iommu_types.h
@@ -251,13 +251,6 @@ struct amd_iommu {
251 /* Pointer to PCI device of this IOMMU */ 251 /* Pointer to PCI device of this IOMMU */
252 struct pci_dev *dev; 252 struct pci_dev *dev;
253 253
254 /*
255 * Capability pointer. There could be more than one IOMMU per PCI
256 * device function if there are more than one AMD IOMMU capability
257 * pointers.
258 */
259 u16 cap_ptr;
260
261 /* physical address of MMIO space */ 254 /* physical address of MMIO space */
262 u64 mmio_phys; 255 u64 mmio_phys;
263 /* virtual address of MMIO space */ 256 /* virtual address of MMIO space */
@@ -266,6 +259,13 @@ struct amd_iommu {
266 /* capabilities of that IOMMU read from ACPI */ 259 /* capabilities of that IOMMU read from ACPI */
267 u32 cap; 260 u32 cap;
268 261
262 /*
263 * Capability pointer. There could be more than one IOMMU per PCI
264 * device function if there are more than one AMD IOMMU capability
265 * pointers.
266 */
267 u16 cap_ptr;
268
269 /* pci domain of this IOMMU */ 269 /* pci domain of this IOMMU */
270 u16 pci_seg; 270 u16 pci_seg;
271 271
@@ -284,19 +284,19 @@ struct amd_iommu {
284 /* size of command buffer */ 284 /* size of command buffer */
285 u32 cmd_buf_size; 285 u32 cmd_buf_size;
286 286
287 /* event buffer virtual address */
288 u8 *evt_buf;
289 /* size of event buffer */ 287 /* size of event buffer */
290 u32 evt_buf_size; 288 u32 evt_buf_size;
289 /* event buffer virtual address */
290 u8 *evt_buf;
291 /* MSI number for event interrupt */ 291 /* MSI number for event interrupt */
292 u16 evt_msi_num; 292 u16 evt_msi_num;
293 293
294 /* if one, we need to send a completion wait command */
295 int need_sync;
296
297 /* true if interrupts for this IOMMU are already enabled */ 294 /* true if interrupts for this IOMMU are already enabled */
298 bool int_enabled; 295 bool int_enabled;
299 296
297 /* if one, we need to send a completion wait command */
298 int need_sync;
299
300 /* default dma_ops domain for that IOMMU */ 300 /* default dma_ops domain for that IOMMU */
301 struct dma_ops_domain *default_dom; 301 struct dma_ops_domain *default_dom;
302}; 302};
diff --git a/arch/x86/include/asm/apic.h b/arch/x86/include/asm/apic.h
index 3b1510b4fc57..25caa0738af5 100644
--- a/arch/x86/include/asm/apic.h
+++ b/arch/x86/include/asm/apic.h
@@ -193,6 +193,7 @@ extern u8 setup_APIC_eilvt_ibs(u8 vector, u8 msg_type, u8 mask);
193static inline void lapic_shutdown(void) { } 193static inline void lapic_shutdown(void) { }
194#define local_apic_timer_c2_ok 1 194#define local_apic_timer_c2_ok 1
195static inline void init_apic_mappings(void) { } 195static inline void init_apic_mappings(void) { }
196static inline void disable_local_APIC(void) { }
196 197
197#endif /* !CONFIG_X86_LOCAL_APIC */ 198#endif /* !CONFIG_X86_LOCAL_APIC */
198 199
diff --git a/arch/x86/include/asm/bigsmp/apic.h b/arch/x86/include/asm/bigsmp/apic.h
index 1d9543b9d358..ce547f24a1cd 100644
--- a/arch/x86/include/asm/bigsmp/apic.h
+++ b/arch/x86/include/asm/bigsmp/apic.h
@@ -24,8 +24,6 @@ static inline cpumask_t target_cpus(void)
24#define INT_DELIVERY_MODE (dest_Fixed) 24#define INT_DELIVERY_MODE (dest_Fixed)
25#define INT_DEST_MODE (0) /* phys delivery to target proc */ 25#define INT_DEST_MODE (0) /* phys delivery to target proc */
26#define NO_BALANCE_IRQ (0) 26#define NO_BALANCE_IRQ (0)
27#define WAKE_SECONDARY_VIA_INIT
28
29 27
30static inline unsigned long check_apicid_used(physid_mask_t bitmap, int apicid) 28static inline unsigned long check_apicid_used(physid_mask_t bitmap, int apicid)
31{ 29{
diff --git a/arch/x86/include/asm/bitops.h b/arch/x86/include/asm/bitops.h
index 360010322711..9fa9dcdf344b 100644
--- a/arch/x86/include/asm/bitops.h
+++ b/arch/x86/include/asm/bitops.h
@@ -168,7 +168,15 @@ static inline void __change_bit(int nr, volatile unsigned long *addr)
168 */ 168 */
169static inline void change_bit(int nr, volatile unsigned long *addr) 169static inline void change_bit(int nr, volatile unsigned long *addr)
170{ 170{
171 asm volatile(LOCK_PREFIX "btc %1,%0" : ADDR : "Ir" (nr)); 171 if (IS_IMMEDIATE(nr)) {
172 asm volatile(LOCK_PREFIX "xorb %1,%0"
173 : CONST_MASK_ADDR(nr, addr)
174 : "iq" ((u8)CONST_MASK(nr)));
175 } else {
176 asm volatile(LOCK_PREFIX "btc %1,%0"
177 : BITOP_ADDR(addr)
178 : "Ir" (nr));
179 }
172} 180}
173 181
174/** 182/**
diff --git a/arch/x86/include/asm/bug.h b/arch/x86/include/asm/bug.h
index 3def2065fcea..d9cf1cd156d2 100644
--- a/arch/x86/include/asm/bug.h
+++ b/arch/x86/include/asm/bug.h
@@ -9,7 +9,7 @@
9#ifdef CONFIG_X86_32 9#ifdef CONFIG_X86_32
10# define __BUG_C0 "2:\t.long 1b, %c0\n" 10# define __BUG_C0 "2:\t.long 1b, %c0\n"
11#else 11#else
12# define __BUG_C0 "2:\t.quad 1b, %c0\n" 12# define __BUG_C0 "2:\t.long 1b - 2b, %c0 - 2b\n"
13#endif 13#endif
14 14
15#define BUG() \ 15#define BUG() \
diff --git a/arch/x86/include/asm/byteorder.h b/arch/x86/include/asm/byteorder.h
index e02ae2d89acf..f110ad417df3 100644
--- a/arch/x86/include/asm/byteorder.h
+++ b/arch/x86/include/asm/byteorder.h
@@ -4,26 +4,33 @@
4#include <asm/types.h> 4#include <asm/types.h>
5#include <linux/compiler.h> 5#include <linux/compiler.h>
6 6
7#ifdef __GNUC__ 7#define __LITTLE_ENDIAN
8 8
9#ifdef __i386__ 9static inline __attribute_const__ __u32 __arch_swab32(__u32 val)
10
11static inline __attribute_const__ __u32 ___arch__swab32(__u32 x)
12{ 10{
13#ifdef CONFIG_X86_BSWAP 11#ifdef __i386__
14 asm("bswap %0" : "=r" (x) : "0" (x)); 12# ifdef CONFIG_X86_BSWAP
15#else 13 asm("bswap %0" : "=r" (val) : "0" (val));
14# else
16 asm("xchgb %b0,%h0\n\t" /* swap lower bytes */ 15 asm("xchgb %b0,%h0\n\t" /* swap lower bytes */
17 "rorl $16,%0\n\t" /* swap words */ 16 "rorl $16,%0\n\t" /* swap words */
18 "xchgb %b0,%h0" /* swap higher bytes */ 17 "xchgb %b0,%h0" /* swap higher bytes */
19 : "=q" (x) 18 : "=q" (val)
20 : "0" (x)); 19 : "0" (val));
20# endif
21
22#else /* __i386__ */
23 asm("bswapl %0"
24 : "=r" (val)
25 : "0" (val));
21#endif 26#endif
22 return x; 27 return val;
23} 28}
29#define __arch_swab32 __arch_swab32
24 30
25static inline __attribute_const__ __u64 ___arch__swab64(__u64 val) 31static inline __attribute_const__ __u64 __arch_swab64(__u64 val)
26{ 32{
33#ifdef __i386__
27 union { 34 union {
28 struct { 35 struct {
29 __u32 a; 36 __u32 a;
@@ -32,50 +39,27 @@ static inline __attribute_const__ __u64 ___arch__swab64(__u64 val)
32 __u64 u; 39 __u64 u;
33 } v; 40 } v;
34 v.u = val; 41 v.u = val;
35#ifdef CONFIG_X86_BSWAP 42# ifdef CONFIG_X86_BSWAP
36 asm("bswapl %0 ; bswapl %1 ; xchgl %0,%1" 43 asm("bswapl %0 ; bswapl %1 ; xchgl %0,%1"
37 : "=r" (v.s.a), "=r" (v.s.b) 44 : "=r" (v.s.a), "=r" (v.s.b)
38 : "0" (v.s.a), "1" (v.s.b)); 45 : "0" (v.s.a), "1" (v.s.b));
39#else 46# else
40 v.s.a = ___arch__swab32(v.s.a); 47 v.s.a = __arch_swab32(v.s.a);
41 v.s.b = ___arch__swab32(v.s.b); 48 v.s.b = __arch_swab32(v.s.b);
42 asm("xchgl %0,%1" 49 asm("xchgl %0,%1"
43 : "=r" (v.s.a), "=r" (v.s.b) 50 : "=r" (v.s.a), "=r" (v.s.b)
44 : "0" (v.s.a), "1" (v.s.b)); 51 : "0" (v.s.a), "1" (v.s.b));
45#endif 52# endif
46 return v.u; 53 return v.u;
47}
48
49#else /* __i386__ */ 54#else /* __i386__ */
50
51static inline __attribute_const__ __u64 ___arch__swab64(__u64 x)
52{
53 asm("bswapq %0" 55 asm("bswapq %0"
54 : "=r" (x) 56 : "=r" (val)
55 : "0" (x)); 57 : "0" (val));
56 return x; 58 return val;
57}
58
59static inline __attribute_const__ __u32 ___arch__swab32(__u32 x)
60{
61 asm("bswapl %0"
62 : "=r" (x)
63 : "0" (x));
64 return x;
65}
66
67#endif 59#endif
60}
61#define __arch_swab64 __arch_swab64
68 62
69/* Do not define swab16. Gcc is smart enough to recognize "C" version and 63#include <linux/byteorder.h>
70 convert it into rotation or exhange. */
71
72#define __arch__swab64(x) ___arch__swab64(x)
73#define __arch__swab32(x) ___arch__swab32(x)
74
75#define __BYTEORDER_HAS_U64__
76
77#endif /* __GNUC__ */
78
79#include <linux/byteorder/little_endian.h>
80 64
81#endif /* _ASM_X86_BYTEORDER_H */ 65#endif /* _ASM_X86_BYTEORDER_H */
diff --git a/arch/x86/include/asm/cpufeature.h b/arch/x86/include/asm/cpufeature.h
index cfdf8c2c5c31..ea408dcba513 100644
--- a/arch/x86/include/asm/cpufeature.h
+++ b/arch/x86/include/asm/cpufeature.h
@@ -80,7 +80,6 @@
80#define X86_FEATURE_UP (3*32+ 9) /* smp kernel running on up */ 80#define X86_FEATURE_UP (3*32+ 9) /* smp kernel running on up */
81#define X86_FEATURE_FXSAVE_LEAK (3*32+10) /* "" FXSAVE leaks FOP/FIP/FOP */ 81#define X86_FEATURE_FXSAVE_LEAK (3*32+10) /* "" FXSAVE leaks FOP/FIP/FOP */
82#define X86_FEATURE_ARCH_PERFMON (3*32+11) /* Intel Architectural PerfMon */ 82#define X86_FEATURE_ARCH_PERFMON (3*32+11) /* Intel Architectural PerfMon */
83#define X86_FEATURE_NOPL (3*32+20) /* The NOPL (0F 1F) instructions */
84#define X86_FEATURE_PEBS (3*32+12) /* Precise-Event Based Sampling */ 83#define X86_FEATURE_PEBS (3*32+12) /* Precise-Event Based Sampling */
85#define X86_FEATURE_BTS (3*32+13) /* Branch Trace Store */ 84#define X86_FEATURE_BTS (3*32+13) /* Branch Trace Store */
86#define X86_FEATURE_SYSCALL32 (3*32+14) /* "" syscall in ia32 userspace */ 85#define X86_FEATURE_SYSCALL32 (3*32+14) /* "" syscall in ia32 userspace */
@@ -92,6 +91,8 @@
92#define X86_FEATURE_NOPL (3*32+20) /* The NOPL (0F 1F) instructions */ 91#define X86_FEATURE_NOPL (3*32+20) /* The NOPL (0F 1F) instructions */
93#define X86_FEATURE_AMDC1E (3*32+21) /* AMD C1E detected */ 92#define X86_FEATURE_AMDC1E (3*32+21) /* AMD C1E detected */
94#define X86_FEATURE_XTOPOLOGY (3*32+22) /* cpu topology enum extensions */ 93#define X86_FEATURE_XTOPOLOGY (3*32+22) /* cpu topology enum extensions */
94#define X86_FEATURE_TSC_RELIABLE (3*32+23) /* TSC is known to be reliable */
95#define X86_FEATURE_NONSTOP_TSC (3*32+24) /* TSC does not stop in C states */
95 96
96/* Intel-defined CPU features, CPUID level 0x00000001 (ecx), word 4 */ 97/* Intel-defined CPU features, CPUID level 0x00000001 (ecx), word 4 */
97#define X86_FEATURE_XMM3 (4*32+ 0) /* "pni" SSE-3 */ 98#define X86_FEATURE_XMM3 (4*32+ 0) /* "pni" SSE-3 */
@@ -117,6 +118,7 @@
117#define X86_FEATURE_XSAVE (4*32+26) /* XSAVE/XRSTOR/XSETBV/XGETBV */ 118#define X86_FEATURE_XSAVE (4*32+26) /* XSAVE/XRSTOR/XSETBV/XGETBV */
118#define X86_FEATURE_OSXSAVE (4*32+27) /* "" XSAVE enabled in the OS */ 119#define X86_FEATURE_OSXSAVE (4*32+27) /* "" XSAVE enabled in the OS */
119#define X86_FEATURE_AVX (4*32+28) /* Advanced Vector Extensions */ 120#define X86_FEATURE_AVX (4*32+28) /* Advanced Vector Extensions */
121#define X86_FEATURE_HYPERVISOR (4*32+31) /* Running on a hypervisor */
120 122
121/* VIA/Cyrix/Centaur-defined CPU features, CPUID level 0xC0000001, word 5 */ 123/* VIA/Cyrix/Centaur-defined CPU features, CPUID level 0xC0000001, word 5 */
122#define X86_FEATURE_XSTORE (5*32+ 2) /* "rng" RNG present (xstore) */ 124#define X86_FEATURE_XSTORE (5*32+ 2) /* "rng" RNG present (xstore) */
@@ -237,6 +239,7 @@ extern const char * const x86_power_flags[32];
237#define cpu_has_xmm4_2 boot_cpu_has(X86_FEATURE_XMM4_2) 239#define cpu_has_xmm4_2 boot_cpu_has(X86_FEATURE_XMM4_2)
238#define cpu_has_x2apic boot_cpu_has(X86_FEATURE_X2APIC) 240#define cpu_has_x2apic boot_cpu_has(X86_FEATURE_X2APIC)
239#define cpu_has_xsave boot_cpu_has(X86_FEATURE_XSAVE) 241#define cpu_has_xsave boot_cpu_has(X86_FEATURE_XSAVE)
242#define cpu_has_hypervisor boot_cpu_has(X86_FEATURE_HYPERVISOR)
240 243
241#if defined(CONFIG_X86_INVLPG) || defined(CONFIG_X86_64) 244#if defined(CONFIG_X86_INVLPG) || defined(CONFIG_X86_64)
242# define cpu_has_invlpg 1 245# define cpu_has_invlpg 1
diff --git a/arch/x86/include/asm/dma-mapping.h b/arch/x86/include/asm/dma-mapping.h
index 7f225a4b2a26..097794ff6b79 100644
--- a/arch/x86/include/asm/dma-mapping.h
+++ b/arch/x86/include/asm/dma-mapping.h
@@ -71,15 +71,13 @@ static inline struct dma_mapping_ops *get_dma_ops(struct device *dev)
71/* Make sure we keep the same behaviour */ 71/* Make sure we keep the same behaviour */
72static inline int dma_mapping_error(struct device *dev, dma_addr_t dma_addr) 72static inline int dma_mapping_error(struct device *dev, dma_addr_t dma_addr)
73{ 73{
74#ifdef CONFIG_X86_32 74#ifdef CONFIG_X86_64
75 return 0;
76#else
77 struct dma_mapping_ops *ops = get_dma_ops(dev); 75 struct dma_mapping_ops *ops = get_dma_ops(dev);
78 if (ops->mapping_error) 76 if (ops->mapping_error)
79 return ops->mapping_error(dev, dma_addr); 77 return ops->mapping_error(dev, dma_addr);
80 78
81 return (dma_addr == bad_dma_address);
82#endif 79#endif
80 return (dma_addr == bad_dma_address);
83} 81}
84 82
85#define dma_alloc_noncoherent(d, s, h, f) dma_alloc_coherent(d, s, h, f) 83#define dma_alloc_noncoherent(d, s, h, f) dma_alloc_coherent(d, s, h, f)
diff --git a/arch/x86/include/asm/ds.h b/arch/x86/include/asm/ds.h
index 72c5a190bf48..a95008457ea4 100644
--- a/arch/x86/include/asm/ds.h
+++ b/arch/x86/include/asm/ds.h
@@ -23,12 +23,13 @@
23#ifndef _ASM_X86_DS_H 23#ifndef _ASM_X86_DS_H
24#define _ASM_X86_DS_H 24#define _ASM_X86_DS_H
25 25
26#ifdef CONFIG_X86_DS
27 26
28#include <linux/types.h> 27#include <linux/types.h>
29#include <linux/init.h> 28#include <linux/init.h>
30 29
31 30
31#ifdef CONFIG_X86_DS
32
32struct task_struct; 33struct task_struct;
33 34
34/* 35/*
@@ -232,7 +233,8 @@ extern void ds_free(struct ds_context *context);
232 233
233#else /* CONFIG_X86_DS */ 234#else /* CONFIG_X86_DS */
234 235
235#define ds_init_intel(config) do {} while (0) 236struct cpuinfo_x86;
237static inline void __cpuinit ds_init_intel(struct cpuinfo_x86 *ignored) {}
236 238
237#endif /* CONFIG_X86_DS */ 239#endif /* CONFIG_X86_DS */
238#endif /* _ASM_X86_DS_H */ 240#endif /* _ASM_X86_DS_H */
diff --git a/arch/x86/include/asm/emergency-restart.h b/arch/x86/include/asm/emergency-restart.h
index 94826cf87455..cc70c1c78ca4 100644
--- a/arch/x86/include/asm/emergency-restart.h
+++ b/arch/x86/include/asm/emergency-restart.h
@@ -8,7 +8,9 @@ enum reboot_type {
8 BOOT_BIOS = 'b', 8 BOOT_BIOS = 'b',
9#endif 9#endif
10 BOOT_ACPI = 'a', 10 BOOT_ACPI = 'a',
11 BOOT_EFI = 'e' 11 BOOT_EFI = 'e',
12 BOOT_CF9 = 'p',
13 BOOT_CF9_COND = 'q',
12}; 14};
13 15
14extern enum reboot_type reboot_type; 16extern enum reboot_type reboot_type;
diff --git a/arch/x86/include/asm/es7000/apic.h b/arch/x86/include/asm/es7000/apic.h
index 380f0b4f17ed..e24ef876915f 100644
--- a/arch/x86/include/asm/es7000/apic.h
+++ b/arch/x86/include/asm/es7000/apic.h
@@ -9,31 +9,27 @@ static inline int apic_id_registered(void)
9 return (1); 9 return (1);
10} 10}
11 11
12static inline cpumask_t target_cpus(void) 12static inline cpumask_t target_cpus_cluster(void)
13{ 13{
14#if defined CONFIG_ES7000_CLUSTERED_APIC
15 return CPU_MASK_ALL; 14 return CPU_MASK_ALL;
16#else 15}
16
17static inline cpumask_t target_cpus(void)
18{
17 return cpumask_of_cpu(smp_processor_id()); 19 return cpumask_of_cpu(smp_processor_id());
18#endif
19} 20}
20 21
21#if defined CONFIG_ES7000_CLUSTERED_APIC 22#define APIC_DFR_VALUE_CLUSTER (APIC_DFR_CLUSTER)
22#define APIC_DFR_VALUE (APIC_DFR_CLUSTER) 23#define INT_DELIVERY_MODE_CLUSTER (dest_LowestPrio)
23#define INT_DELIVERY_MODE (dest_LowestPrio) 24#define INT_DEST_MODE_CLUSTER (1) /* logical delivery broadcast to all procs */
24#define INT_DEST_MODE (1) /* logical delivery broadcast to all procs */ 25#define NO_BALANCE_IRQ_CLUSTER (1)
25#define NO_BALANCE_IRQ (1) 26
26#undef WAKE_SECONDARY_VIA_INIT
27#define WAKE_SECONDARY_VIA_MIP
28#else
29#define APIC_DFR_VALUE (APIC_DFR_FLAT) 27#define APIC_DFR_VALUE (APIC_DFR_FLAT)
30#define INT_DELIVERY_MODE (dest_Fixed) 28#define INT_DELIVERY_MODE (dest_Fixed)
31#define INT_DEST_MODE (0) /* phys delivery to target procs */ 29#define INT_DEST_MODE (0) /* phys delivery to target procs */
32#define NO_BALANCE_IRQ (0) 30#define NO_BALANCE_IRQ (0)
33#undef APIC_DEST_LOGICAL 31#undef APIC_DEST_LOGICAL
34#define APIC_DEST_LOGICAL 0x0 32#define APIC_DEST_LOGICAL 0x0
35#define WAKE_SECONDARY_VIA_INIT
36#endif
37 33
38static inline unsigned long check_apicid_used(physid_mask_t bitmap, int apicid) 34static inline unsigned long check_apicid_used(physid_mask_t bitmap, int apicid)
39{ 35{
@@ -60,6 +56,16 @@ static inline unsigned long calculate_ldr(int cpu)
60 * an APIC. See e.g. "AP-388 82489DX User's Manual" (Intel 56 * an APIC. See e.g. "AP-388 82489DX User's Manual" (Intel
61 * document number 292116). So here it goes... 57 * document number 292116). So here it goes...
62 */ 58 */
59static inline void init_apic_ldr_cluster(void)
60{
61 unsigned long val;
62 int cpu = smp_processor_id();
63
64 apic_write(APIC_DFR, APIC_DFR_VALUE_CLUSTER);
65 val = calculate_ldr(cpu);
66 apic_write(APIC_LDR, val);
67}
68
63static inline void init_apic_ldr(void) 69static inline void init_apic_ldr(void)
64{ 70{
65 unsigned long val; 71 unsigned long val;
@@ -70,10 +76,6 @@ static inline void init_apic_ldr(void)
70 apic_write(APIC_LDR, val); 76 apic_write(APIC_LDR, val);
71} 77}
72 78
73#ifndef CONFIG_X86_GENERICARCH
74extern void enable_apic_mode(void);
75#endif
76
77extern int apic_version [MAX_APICS]; 79extern int apic_version [MAX_APICS];
78static inline void setup_apic_routing(void) 80static inline void setup_apic_routing(void)
79{ 81{
@@ -144,7 +146,7 @@ static inline int check_phys_apicid_present(int cpu_physical_apicid)
144 return (1); 146 return (1);
145} 147}
146 148
147static inline unsigned int cpu_mask_to_apicid(cpumask_t cpumask) 149static inline unsigned int cpu_mask_to_apicid_cluster(cpumask_t cpumask)
148{ 150{
149 int num_bits_set; 151 int num_bits_set;
150 int cpus_found = 0; 152 int cpus_found = 0;
@@ -154,11 +156,7 @@ static inline unsigned int cpu_mask_to_apicid(cpumask_t cpumask)
154 num_bits_set = cpus_weight(cpumask); 156 num_bits_set = cpus_weight(cpumask);
155 /* Return id to all */ 157 /* Return id to all */
156 if (num_bits_set == NR_CPUS) 158 if (num_bits_set == NR_CPUS)
157#if defined CONFIG_ES7000_CLUSTERED_APIC
158 return 0xFF; 159 return 0xFF;
159#else
160 return cpu_to_logical_apicid(0);
161#endif
162 /* 160 /*
163 * The cpus in the mask must all be on the apic cluster. If are not 161 * The cpus in the mask must all be on the apic cluster. If are not
164 * on the same apicid cluster return default value of TARGET_CPUS. 162 * on the same apicid cluster return default value of TARGET_CPUS.
@@ -171,11 +169,40 @@ static inline unsigned int cpu_mask_to_apicid(cpumask_t cpumask)
171 if (apicid_cluster(apicid) != 169 if (apicid_cluster(apicid) !=
172 apicid_cluster(new_apicid)){ 170 apicid_cluster(new_apicid)){
173 printk ("%s: Not a valid mask!\n", __func__); 171 printk ("%s: Not a valid mask!\n", __func__);
174#if defined CONFIG_ES7000_CLUSTERED_APIC
175 return 0xFF; 172 return 0xFF;
176#else 173 }
174 apicid = new_apicid;
175 cpus_found++;
176 }
177 cpu++;
178 }
179 return apicid;
180}
181
182static inline unsigned int cpu_mask_to_apicid(cpumask_t cpumask)
183{
184 int num_bits_set;
185 int cpus_found = 0;
186 int cpu;
187 int apicid;
188
189 num_bits_set = cpus_weight(cpumask);
190 /* Return id to all */
191 if (num_bits_set == NR_CPUS)
192 return cpu_to_logical_apicid(0);
193 /*
194 * The cpus in the mask must all be on the apic cluster. If are not
195 * on the same apicid cluster return default value of TARGET_CPUS.
196 */
197 cpu = first_cpu(cpumask);
198 apicid = cpu_to_logical_apicid(cpu);
199 while (cpus_found < num_bits_set) {
200 if (cpu_isset(cpu, cpumask)) {
201 int new_apicid = cpu_to_logical_apicid(cpu);
202 if (apicid_cluster(apicid) !=
203 apicid_cluster(new_apicid)){
204 printk ("%s: Not a valid mask!\n", __func__);
177 return cpu_to_logical_apicid(0); 205 return cpu_to_logical_apicid(0);
178#endif
179 } 206 }
180 apicid = new_apicid; 207 apicid = new_apicid;
181 cpus_found++; 208 cpus_found++;
diff --git a/arch/x86/include/asm/es7000/wakecpu.h b/arch/x86/include/asm/es7000/wakecpu.h
index 398493461913..78f0daaee436 100644
--- a/arch/x86/include/asm/es7000/wakecpu.h
+++ b/arch/x86/include/asm/es7000/wakecpu.h
@@ -1,36 +1,12 @@
1#ifndef __ASM_ES7000_WAKECPU_H 1#ifndef __ASM_ES7000_WAKECPU_H
2#define __ASM_ES7000_WAKECPU_H 2#define __ASM_ES7000_WAKECPU_H
3 3
4/* 4#define TRAMPOLINE_PHYS_LOW 0x467
5 * This file copes with machines that wakeup secondary CPUs by the 5#define TRAMPOLINE_PHYS_HIGH 0x469
6 * INIT, INIT, STARTUP sequence.
7 */
8
9#ifdef CONFIG_ES7000_CLUSTERED_APIC
10#define WAKE_SECONDARY_VIA_MIP
11#else
12#define WAKE_SECONDARY_VIA_INIT
13#endif
14
15#ifdef WAKE_SECONDARY_VIA_MIP
16extern int es7000_start_cpu(int cpu, unsigned long eip);
17static inline int
18wakeup_secondary_cpu(int phys_apicid, unsigned long start_eip)
19{
20 int boot_error = 0;
21 boot_error = es7000_start_cpu(phys_apicid, start_eip);
22 return boot_error;
23}
24#endif
25
26#define TRAMPOLINE_LOW phys_to_virt(0x467)
27#define TRAMPOLINE_HIGH phys_to_virt(0x469)
28
29#define boot_cpu_apicid boot_cpu_physical_apicid
30 6
31static inline void wait_for_init_deassert(atomic_t *deassert) 7static inline void wait_for_init_deassert(atomic_t *deassert)
32{ 8{
33#ifdef WAKE_SECONDARY_VIA_INIT 9#ifndef CONFIG_ES7000_CLUSTERED_APIC
34 while (!atomic_read(deassert)) 10 while (!atomic_read(deassert))
35 cpu_relax(); 11 cpu_relax();
36#endif 12#endif
@@ -50,9 +26,12 @@ static inline void restore_NMI_vector(unsigned short *high, unsigned short *low)
50{ 26{
51} 27}
52 28
53#define inquire_remote_apic(apicid) do { \ 29extern void __inquire_remote_apic(int apicid);
54 if (apic_verbosity >= APIC_DEBUG) \ 30
55 __inquire_remote_apic(apicid); \ 31static inline void inquire_remote_apic(int apicid)
56 } while (0) 32{
33 if (apic_verbosity >= APIC_DEBUG)
34 __inquire_remote_apic(apicid);
35}
57 36
58#endif /* __ASM_MACH_WAKECPU_H */ 37#endif /* __ASM_MACH_WAKECPU_H */
diff --git a/arch/x86/include/asm/genapic_32.h b/arch/x86/include/asm/genapic_32.h
index 5cbd4fcc06fd..0ac17d33a8c7 100644
--- a/arch/x86/include/asm/genapic_32.h
+++ b/arch/x86/include/asm/genapic_32.h
@@ -2,6 +2,7 @@
2#define _ASM_X86_GENAPIC_32_H 2#define _ASM_X86_GENAPIC_32_H
3 3
4#include <asm/mpspec.h> 4#include <asm/mpspec.h>
5#include <asm/atomic.h>
5 6
6/* 7/*
7 * Generic APIC driver interface. 8 * Generic APIC driver interface.
@@ -65,6 +66,14 @@ struct genapic {
65 void (*send_IPI_allbutself)(int vector); 66 void (*send_IPI_allbutself)(int vector);
66 void (*send_IPI_all)(int vector); 67 void (*send_IPI_all)(int vector);
67#endif 68#endif
69 int (*wakeup_cpu)(int apicid, unsigned long start_eip);
70 int trampoline_phys_low;
71 int trampoline_phys_high;
72 void (*wait_for_init_deassert)(atomic_t *deassert);
73 void (*smp_callin_clear_local_apic)(void);
74 void (*store_NMI_vector)(unsigned short *high, unsigned short *low);
75 void (*restore_NMI_vector)(unsigned short *high, unsigned short *low);
76 void (*inquire_remote_apic)(int apicid);
68}; 77};
69 78
70#define APICFUNC(x) .x = x, 79#define APICFUNC(x) .x = x,
@@ -105,16 +114,24 @@ struct genapic {
105 APICFUNC(get_apic_id) \ 114 APICFUNC(get_apic_id) \
106 .apic_id_mask = APIC_ID_MASK, \ 115 .apic_id_mask = APIC_ID_MASK, \
107 APICFUNC(cpu_mask_to_apicid) \ 116 APICFUNC(cpu_mask_to_apicid) \
108 APICFUNC(vector_allocation_domain) \ 117 APICFUNC(vector_allocation_domain) \
109 APICFUNC(acpi_madt_oem_check) \ 118 APICFUNC(acpi_madt_oem_check) \
110 IPIFUNC(send_IPI_mask) \ 119 IPIFUNC(send_IPI_mask) \
111 IPIFUNC(send_IPI_allbutself) \ 120 IPIFUNC(send_IPI_allbutself) \
112 IPIFUNC(send_IPI_all) \ 121 IPIFUNC(send_IPI_all) \
113 APICFUNC(enable_apic_mode) \ 122 APICFUNC(enable_apic_mode) \
114 APICFUNC(phys_pkg_id) \ 123 APICFUNC(phys_pkg_id) \
124 .trampoline_phys_low = TRAMPOLINE_PHYS_LOW, \
125 .trampoline_phys_high = TRAMPOLINE_PHYS_HIGH, \
126 APICFUNC(wait_for_init_deassert) \
127 APICFUNC(smp_callin_clear_local_apic) \
128 APICFUNC(store_NMI_vector) \
129 APICFUNC(restore_NMI_vector) \
130 APICFUNC(inquire_remote_apic) \
115} 131}
116 132
117extern struct genapic *genapic; 133extern struct genapic *genapic;
134extern void es7000_update_genapic_to_cluster(void);
118 135
119enum uv_system_type {UV_NONE, UV_LEGACY_APIC, UV_X2APIC, UV_NON_UNIQUE_APIC}; 136enum uv_system_type {UV_NONE, UV_LEGACY_APIC, UV_X2APIC, UV_NON_UNIQUE_APIC};
120#define get_uv_system_type() UV_NONE 137#define get_uv_system_type() UV_NONE
diff --git a/arch/x86/include/asm/genapic_64.h b/arch/x86/include/asm/genapic_64.h
index 13c4e96199ea..2cae011668b7 100644
--- a/arch/x86/include/asm/genapic_64.h
+++ b/arch/x86/include/asm/genapic_64.h
@@ -32,6 +32,8 @@ struct genapic {
32 unsigned int (*get_apic_id)(unsigned long x); 32 unsigned int (*get_apic_id)(unsigned long x);
33 unsigned long (*set_apic_id)(unsigned int id); 33 unsigned long (*set_apic_id)(unsigned int id);
34 unsigned long apic_id_mask; 34 unsigned long apic_id_mask;
35 /* wakeup_secondary_cpu */
36 int (*wakeup_cpu)(int apicid, unsigned long start_eip);
35}; 37};
36 38
37extern struct genapic *genapic; 39extern struct genapic *genapic;
diff --git a/arch/x86/include/asm/hypervisor.h b/arch/x86/include/asm/hypervisor.h
new file mode 100644
index 000000000000..369f5c5d09a1
--- /dev/null
+++ b/arch/x86/include/asm/hypervisor.h
@@ -0,0 +1,26 @@
1/*
2 * Copyright (C) 2008, VMware, Inc.
3 *
4 * This program is free software; you can redistribute it and/or modify
5 * it under the terms of the GNU General Public License as published by
6 * the Free Software Foundation; either version 2 of the License, or
7 * (at your option) any later version.
8 *
9 * This program is distributed in the hope that it will be useful, but
10 * WITHOUT ANY WARRANTY; without even the implied warranty of
11 * MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or
12 * NON INFRINGEMENT. See the GNU General Public License for more
13 * details.
14 *
15 * You should have received a copy of the GNU General Public License
16 * along with this program; if not, write to the Free Software
17 * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
18 *
19 */
20#ifndef ASM_X86__HYPERVISOR_H
21#define ASM_X86__HYPERVISOR_H
22
23extern unsigned long get_hypervisor_tsc_freq(void);
24extern void init_hypervisor(struct cpuinfo_x86 *c);
25
26#endif
diff --git a/arch/x86/include/asm/ia32.h b/arch/x86/include/asm/ia32.h
index 97989c0e534c..50ca486fd88c 100644
--- a/arch/x86/include/asm/ia32.h
+++ b/arch/x86/include/asm/ia32.h
@@ -129,24 +129,6 @@ typedef struct compat_siginfo {
129 } _sifields; 129 } _sifields;
130} compat_siginfo_t; 130} compat_siginfo_t;
131 131
132struct sigframe32 {
133 u32 pretcode;
134 int sig;
135 struct sigcontext_ia32 sc;
136 struct _fpstate_ia32 fpstate;
137 unsigned int extramask[_COMPAT_NSIG_WORDS-1];
138};
139
140struct rt_sigframe32 {
141 u32 pretcode;
142 int sig;
143 u32 pinfo;
144 u32 puc;
145 compat_siginfo_t info;
146 struct ucontext_ia32 uc;
147 struct _fpstate_ia32 fpstate;
148};
149
150struct ustat32 { 132struct ustat32 {
151 __u32 f_tfree; 133 __u32 f_tfree;
152 compat_ino_t f_tinode; 134 compat_ino_t f_tinode;
diff --git a/arch/x86/include/asm/idle.h b/arch/x86/include/asm/idle.h
index 44c89c3a23e9..38d87379e270 100644
--- a/arch/x86/include/asm/idle.h
+++ b/arch/x86/include/asm/idle.h
@@ -8,8 +8,13 @@ struct notifier_block;
8void idle_notifier_register(struct notifier_block *n); 8void idle_notifier_register(struct notifier_block *n);
9void idle_notifier_unregister(struct notifier_block *n); 9void idle_notifier_unregister(struct notifier_block *n);
10 10
11#ifdef CONFIG_X86_64
11void enter_idle(void); 12void enter_idle(void);
12void exit_idle(void); 13void exit_idle(void);
14#else /* !CONFIG_X86_64 */
15static inline void enter_idle(void) { }
16static inline void exit_idle(void) { }
17#endif /* CONFIG_X86_64 */
13 18
14void c1e_remove_cpu(int cpu); 19void c1e_remove_cpu(int cpu);
15 20
diff --git a/arch/x86/include/asm/io.h b/arch/x86/include/asm/io.h
index ac2abc88cd95..33513b9a67f3 100644
--- a/arch/x86/include/asm/io.h
+++ b/arch/x86/include/asm/io.h
@@ -4,6 +4,7 @@
4#define ARCH_HAS_IOREMAP_WC 4#define ARCH_HAS_IOREMAP_WC
5 5
6#include <linux/compiler.h> 6#include <linux/compiler.h>
7#include <asm-generic/int-ll64.h>
7 8
8#define build_mmio_read(name, size, type, reg, barrier) \ 9#define build_mmio_read(name, size, type, reg, barrier) \
9static inline type name(const volatile void __iomem *addr) \ 10static inline type name(const volatile void __iomem *addr) \
@@ -45,20 +46,40 @@ build_mmio_write(__writel, "l", unsigned int, "r", )
45#define mmiowb() barrier() 46#define mmiowb() barrier()
46 47
47#ifdef CONFIG_X86_64 48#ifdef CONFIG_X86_64
49
48build_mmio_read(readq, "q", unsigned long, "=r", :"memory") 50build_mmio_read(readq, "q", unsigned long, "=r", :"memory")
49build_mmio_read(__readq, "q", unsigned long, "=r", )
50build_mmio_write(writeq, "q", unsigned long, "r", :"memory") 51build_mmio_write(writeq, "q", unsigned long, "r", :"memory")
51build_mmio_write(__writeq, "q", unsigned long, "r", )
52 52
53#define readq_relaxed(a) __readq(a) 53#else
54#define __raw_readq __readq 54
55#define __raw_writeq writeq 55static inline __u64 readq(const volatile void __iomem *addr)
56{
57 const volatile u32 __iomem *p = addr;
58 u32 low, high;
59
60 low = readl(p);
61 high = readl(p + 1);
62
63 return low + ((u64)high << 32);
64}
65
66static inline void writeq(__u64 val, volatile void __iomem *addr)
67{
68 writel(val, addr);
69 writel(val >> 32, addr+4);
70}
56 71
57/* Let people know we have them */
58#define readq readq
59#define writeq writeq
60#endif 72#endif
61 73
74#define readq_relaxed(a) readq(a)
75
76#define __raw_readq(a) readq(a)
77#define __raw_writeq(val, addr) writeq(val, addr)
78
79/* Let people know that we have them */
80#define readq readq
81#define writeq writeq
82
62extern int iommu_bio_merge; 83extern int iommu_bio_merge;
63 84
64#ifdef CONFIG_X86_32 85#ifdef CONFIG_X86_32
diff --git a/arch/x86/include/asm/io_apic.h b/arch/x86/include/asm/io_apic.h
index 6afd9933a7dd..e475e009ae5d 100644
--- a/arch/x86/include/asm/io_apic.h
+++ b/arch/x86/include/asm/io_apic.h
@@ -156,11 +156,21 @@ extern int sis_apic_bug;
156/* 1 if "noapic" boot option passed */ 156/* 1 if "noapic" boot option passed */
157extern int skip_ioapic_setup; 157extern int skip_ioapic_setup;
158 158
159/* 1 if "noapic" boot option passed */
160extern int noioapicquirk;
161
162/* -1 if "noapic" boot option passed */
163extern int noioapicreroute;
164
159/* 1 if the timer IRQ uses the '8259A Virtual Wire' mode */ 165/* 1 if the timer IRQ uses the '8259A Virtual Wire' mode */
160extern int timer_through_8259; 166extern int timer_through_8259;
161 167
162static inline void disable_ioapic_setup(void) 168static inline void disable_ioapic_setup(void)
163{ 169{
170#ifdef CONFIG_PCI
171 noioapicquirk = 1;
172 noioapicreroute = -1;
173#endif
164 skip_ioapic_setup = 1; 174 skip_ioapic_setup = 1;
165} 175}
166 176
diff --git a/arch/x86/include/asm/iomap.h b/arch/x86/include/asm/iomap.h
new file mode 100644
index 000000000000..c1f06289b14b
--- /dev/null
+++ b/arch/x86/include/asm/iomap.h
@@ -0,0 +1,30 @@
1/*
2 * Copyright © 2008 Ingo Molnar
3 *
4 * This program is free software; you can redistribute it and/or modify
5 * it under the terms of the GNU General Public License as published by
6 * the Free Software Foundation; either version 2 of the License, or
7 * (at your option) any later version.
8 *
9 * This program is distributed in the hope that it will be useful, but
10 * WITHOUT ANY WARRANTY; without even the implied warranty of
11 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
12 * General Public License for more details.
13 *
14 * You should have received a copy of the GNU General Public License along
15 * with this program; if not, write to the Free Software Foundation, Inc.,
16 * 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA.
17 */
18
19#include <linux/fs.h>
20#include <linux/mm.h>
21#include <linux/uaccess.h>
22#include <asm/cacheflush.h>
23#include <asm/pgtable.h>
24#include <asm/tlbflush.h>
25
26void *
27iomap_atomic_prot_pfn(unsigned long pfn, enum km_type type, pgprot_t prot);
28
29void
30iounmap_atomic(void *kvaddr, enum km_type type);
diff --git a/arch/x86/include/asm/iommu.h b/arch/x86/include/asm/iommu.h
index e4a552d44465..0b500c5b6446 100644
--- a/arch/x86/include/asm/iommu.h
+++ b/arch/x86/include/asm/iommu.h
@@ -6,7 +6,6 @@ extern void no_iommu_init(void);
6extern struct dma_mapping_ops nommu_dma_ops; 6extern struct dma_mapping_ops nommu_dma_ops;
7extern int force_iommu, no_iommu; 7extern int force_iommu, no_iommu;
8extern int iommu_detected; 8extern int iommu_detected;
9extern int dmar_disabled;
10 9
11extern unsigned long iommu_nr_pages(unsigned long addr, unsigned long len); 10extern unsigned long iommu_nr_pages(unsigned long addr, unsigned long len);
12 11
diff --git a/arch/x86/include/asm/irq.h b/arch/x86/include/asm/irq.h
index bae0eda95486..28e409fc73f3 100644
--- a/arch/x86/include/asm/irq.h
+++ b/arch/x86/include/asm/irq.h
@@ -31,10 +31,6 @@ static inline int irq_canonicalize(int irq)
31# endif 31# endif
32#endif 32#endif
33 33
34#ifdef CONFIG_IRQBALANCE
35extern int irqbalance_disable(char *str);
36#endif
37
38#ifdef CONFIG_HOTPLUG_CPU 34#ifdef CONFIG_HOTPLUG_CPU
39#include <linux/cpumask.h> 35#include <linux/cpumask.h>
40extern void fixup_irqs(cpumask_t map); 36extern void fixup_irqs(cpumask_t map);
diff --git a/arch/x86/include/asm/irq_regs_32.h b/arch/x86/include/asm/irq_regs_32.h
index af2f02d27fc7..86afd7473457 100644
--- a/arch/x86/include/asm/irq_regs_32.h
+++ b/arch/x86/include/asm/irq_regs_32.h
@@ -9,6 +9,8 @@
9 9
10#include <asm/percpu.h> 10#include <asm/percpu.h>
11 11
12#define ARCH_HAS_OWN_IRQ_REGS
13
12DECLARE_PER_CPU(struct pt_regs *, irq_regs); 14DECLARE_PER_CPU(struct pt_regs *, irq_regs);
13 15
14static inline struct pt_regs *get_irq_regs(void) 16static inline struct pt_regs *get_irq_regs(void)
diff --git a/arch/x86/include/asm/kexec.h b/arch/x86/include/asm/kexec.h
index a1f22771a15a..c61d8b2ab8b9 100644
--- a/arch/x86/include/asm/kexec.h
+++ b/arch/x86/include/asm/kexec.h
@@ -5,21 +5,8 @@
5# define PA_CONTROL_PAGE 0 5# define PA_CONTROL_PAGE 0
6# define VA_CONTROL_PAGE 1 6# define VA_CONTROL_PAGE 1
7# define PA_PGD 2 7# define PA_PGD 2
8# define VA_PGD 3 8# define PA_SWAP_PAGE 3
9# define PA_PTE_0 4 9# define PAGES_NR 4
10# define VA_PTE_0 5
11# define PA_PTE_1 6
12# define VA_PTE_1 7
13# define PA_SWAP_PAGE 8
14# ifdef CONFIG_X86_PAE
15# define PA_PMD_0 9
16# define VA_PMD_0 10
17# define PA_PMD_1 11
18# define VA_PMD_1 12
19# define PAGES_NR 13
20# else
21# define PAGES_NR 9
22# endif
23#else 10#else
24# define PA_CONTROL_PAGE 0 11# define PA_CONTROL_PAGE 0
25# define VA_CONTROL_PAGE 1 12# define VA_CONTROL_PAGE 1
@@ -170,6 +157,20 @@ relocate_kernel(unsigned long indirection_page,
170 unsigned long start_address) ATTRIB_NORET; 157 unsigned long start_address) ATTRIB_NORET;
171#endif 158#endif
172 159
160#ifdef CONFIG_X86_32
161#define ARCH_HAS_KIMAGE_ARCH
162
163struct kimage_arch {
164 pgd_t *pgd;
165#ifdef CONFIG_X86_PAE
166 pmd_t *pmd0;
167 pmd_t *pmd1;
168#endif
169 pte_t *pte0;
170 pte_t *pte1;
171};
172#endif
173
173#endif /* __ASSEMBLY__ */ 174#endif /* __ASSEMBLY__ */
174 175
175#endif /* _ASM_X86_KEXEC_H */ 176#endif /* _ASM_X86_KEXEC_H */
diff --git a/arch/x86/include/asm/mach-default/mach_apic.h b/arch/x86/include/asm/mach-default/mach_apic.h
index ff3a6c236c00..6cb3a467e067 100644
--- a/arch/x86/include/asm/mach-default/mach_apic.h
+++ b/arch/x86/include/asm/mach-default/mach_apic.h
@@ -32,11 +32,13 @@ static inline cpumask_t target_cpus(void)
32#define vector_allocation_domain (genapic->vector_allocation_domain) 32#define vector_allocation_domain (genapic->vector_allocation_domain)
33#define read_apic_id() (GET_APIC_ID(apic_read(APIC_ID))) 33#define read_apic_id() (GET_APIC_ID(apic_read(APIC_ID)))
34#define send_IPI_self (genapic->send_IPI_self) 34#define send_IPI_self (genapic->send_IPI_self)
35#define wakeup_secondary_cpu (genapic->wakeup_cpu)
35extern void setup_apic_routing(void); 36extern void setup_apic_routing(void);
36#else 37#else
37#define INT_DELIVERY_MODE dest_LowestPrio 38#define INT_DELIVERY_MODE dest_LowestPrio
38#define INT_DEST_MODE 1 /* logical delivery broadcast to all procs */ 39#define INT_DEST_MODE 1 /* logical delivery broadcast to all procs */
39#define TARGET_CPUS (target_cpus()) 40#define TARGET_CPUS (target_cpus())
41#define wakeup_secondary_cpu wakeup_secondary_cpu_via_init
40/* 42/*
41 * Set up the logical destination ID. 43 * Set up the logical destination ID.
42 * 44 *
diff --git a/arch/x86/include/asm/mach-default/mach_wakecpu.h b/arch/x86/include/asm/mach-default/mach_wakecpu.h
index 9d80db91e992..ceb013660146 100644
--- a/arch/x86/include/asm/mach-default/mach_wakecpu.h
+++ b/arch/x86/include/asm/mach-default/mach_wakecpu.h
@@ -1,17 +1,8 @@
1#ifndef _ASM_X86_MACH_DEFAULT_MACH_WAKECPU_H 1#ifndef _ASM_X86_MACH_DEFAULT_MACH_WAKECPU_H
2#define _ASM_X86_MACH_DEFAULT_MACH_WAKECPU_H 2#define _ASM_X86_MACH_DEFAULT_MACH_WAKECPU_H
3 3
4/* 4#define TRAMPOLINE_PHYS_LOW (0x467)
5 * This file copes with machines that wakeup secondary CPUs by the 5#define TRAMPOLINE_PHYS_HIGH (0x469)
6 * INIT, INIT, STARTUP sequence.
7 */
8
9#define WAKE_SECONDARY_VIA_INIT
10
11#define TRAMPOLINE_LOW phys_to_virt(0x467)
12#define TRAMPOLINE_HIGH phys_to_virt(0x469)
13
14#define boot_cpu_apicid boot_cpu_physical_apicid
15 6
16static inline void wait_for_init_deassert(atomic_t *deassert) 7static inline void wait_for_init_deassert(atomic_t *deassert)
17{ 8{
@@ -33,9 +24,12 @@ static inline void restore_NMI_vector(unsigned short *high, unsigned short *low)
33{ 24{
34} 25}
35 26
36#define inquire_remote_apic(apicid) do { \ 27extern void __inquire_remote_apic(int apicid);
37 if (apic_verbosity >= APIC_DEBUG) \ 28
38 __inquire_remote_apic(apicid); \ 29static inline void inquire_remote_apic(int apicid)
39 } while (0) 30{
31 if (apic_verbosity >= APIC_DEBUG)
32 __inquire_remote_apic(apicid);
33}
40 34
41#endif /* _ASM_X86_MACH_DEFAULT_MACH_WAKECPU_H */ 35#endif /* _ASM_X86_MACH_DEFAULT_MACH_WAKECPU_H */
diff --git a/arch/x86/include/asm/mach-default/smpboot_hooks.h b/arch/x86/include/asm/mach-default/smpboot_hooks.h
index dbab36d64d48..23bf52103b89 100644
--- a/arch/x86/include/asm/mach-default/smpboot_hooks.h
+++ b/arch/x86/include/asm/mach-default/smpboot_hooks.h
@@ -13,9 +13,11 @@ static inline void smpboot_setup_warm_reset_vector(unsigned long start_eip)
13 CMOS_WRITE(0xa, 0xf); 13 CMOS_WRITE(0xa, 0xf);
14 local_flush_tlb(); 14 local_flush_tlb();
15 pr_debug("1.\n"); 15 pr_debug("1.\n");
16 *((volatile unsigned short *) TRAMPOLINE_HIGH) = start_eip >> 4; 16 *((volatile unsigned short *)phys_to_virt(TRAMPOLINE_PHYS_HIGH)) =
17 start_eip >> 4;
17 pr_debug("2.\n"); 18 pr_debug("2.\n");
18 *((volatile unsigned short *) TRAMPOLINE_LOW) = start_eip & 0xf; 19 *((volatile unsigned short *)phys_to_virt(TRAMPOLINE_PHYS_LOW)) =
20 start_eip & 0xf;
19 pr_debug("3.\n"); 21 pr_debug("3.\n");
20} 22}
21 23
@@ -32,7 +34,7 @@ static inline void smpboot_restore_warm_reset_vector(void)
32 */ 34 */
33 CMOS_WRITE(0, 0xf); 35 CMOS_WRITE(0, 0xf);
34 36
35 *((volatile long *) phys_to_virt(0x467)) = 0; 37 *((volatile long *)phys_to_virt(TRAMPOLINE_PHYS_LOW)) = 0;
36} 38}
37 39
38static inline void __init smpboot_setup_io_apic(void) 40static inline void __init smpboot_setup_io_apic(void)
diff --git a/arch/x86/include/asm/mach-generic/mach_apic.h b/arch/x86/include/asm/mach-generic/mach_apic.h
index 5180bd7478fb..e430f47df667 100644
--- a/arch/x86/include/asm/mach-generic/mach_apic.h
+++ b/arch/x86/include/asm/mach-generic/mach_apic.h
@@ -27,6 +27,7 @@
27#define vector_allocation_domain (genapic->vector_allocation_domain) 27#define vector_allocation_domain (genapic->vector_allocation_domain)
28#define enable_apic_mode (genapic->enable_apic_mode) 28#define enable_apic_mode (genapic->enable_apic_mode)
29#define phys_pkg_id (genapic->phys_pkg_id) 29#define phys_pkg_id (genapic->phys_pkg_id)
30#define wakeup_secondary_cpu (genapic->wakeup_cpu)
30 31
31extern void generic_bigsmp_probe(void); 32extern void generic_bigsmp_probe(void);
32 33
diff --git a/arch/x86/include/asm/mach-generic/mach_wakecpu.h b/arch/x86/include/asm/mach-generic/mach_wakecpu.h
new file mode 100644
index 000000000000..1ab16b168c8a
--- /dev/null
+++ b/arch/x86/include/asm/mach-generic/mach_wakecpu.h
@@ -0,0 +1,12 @@
1#ifndef _ASM_X86_MACH_GENERIC_MACH_WAKECPU_H
2#define _ASM_X86_MACH_GENERIC_MACH_WAKECPU_H
3
4#define TRAMPOLINE_PHYS_LOW (genapic->trampoline_phys_low)
5#define TRAMPOLINE_PHYS_HIGH (genapic->trampoline_phys_high)
6#define wait_for_init_deassert (genapic->wait_for_init_deassert)
7#define smp_callin_clear_local_apic (genapic->smp_callin_clear_local_apic)
8#define store_NMI_vector (genapic->store_NMI_vector)
9#define restore_NMI_vector (genapic->restore_NMI_vector)
10#define inquire_remote_apic (genapic->inquire_remote_apic)
11
12#endif /* _ASM_X86_MACH_GENERIC_MACH_APIC_H */
diff --git a/arch/x86/include/asm/mmu_context_32.h b/arch/x86/include/asm/mmu_context_32.h
index 8e10015781fb..7e98ce1d2c0e 100644
--- a/arch/x86/include/asm/mmu_context_32.h
+++ b/arch/x86/include/asm/mmu_context_32.h
@@ -4,9 +4,8 @@
4static inline void enter_lazy_tlb(struct mm_struct *mm, struct task_struct *tsk) 4static inline void enter_lazy_tlb(struct mm_struct *mm, struct task_struct *tsk)
5{ 5{
6#ifdef CONFIG_SMP 6#ifdef CONFIG_SMP
7 unsigned cpu = smp_processor_id(); 7 if (x86_read_percpu(cpu_tlbstate.state) == TLBSTATE_OK)
8 if (per_cpu(cpu_tlbstate, cpu).state == TLBSTATE_OK) 8 x86_write_percpu(cpu_tlbstate.state, TLBSTATE_LAZY);
9 per_cpu(cpu_tlbstate, cpu).state = TLBSTATE_LAZY;
10#endif 9#endif
11} 10}
12 11
@@ -20,8 +19,8 @@ static inline void switch_mm(struct mm_struct *prev,
20 /* stop flush ipis for the previous mm */ 19 /* stop flush ipis for the previous mm */
21 cpu_clear(cpu, prev->cpu_vm_mask); 20 cpu_clear(cpu, prev->cpu_vm_mask);
22#ifdef CONFIG_SMP 21#ifdef CONFIG_SMP
23 per_cpu(cpu_tlbstate, cpu).state = TLBSTATE_OK; 22 x86_write_percpu(cpu_tlbstate.state, TLBSTATE_OK);
24 per_cpu(cpu_tlbstate, cpu).active_mm = next; 23 x86_write_percpu(cpu_tlbstate.active_mm, next);
25#endif 24#endif
26 cpu_set(cpu, next->cpu_vm_mask); 25 cpu_set(cpu, next->cpu_vm_mask);
27 26
@@ -36,8 +35,8 @@ static inline void switch_mm(struct mm_struct *prev,
36 } 35 }
37#ifdef CONFIG_SMP 36#ifdef CONFIG_SMP
38 else { 37 else {
39 per_cpu(cpu_tlbstate, cpu).state = TLBSTATE_OK; 38 x86_write_percpu(cpu_tlbstate.state, TLBSTATE_OK);
40 BUG_ON(per_cpu(cpu_tlbstate, cpu).active_mm != next); 39 BUG_ON(x86_read_percpu(cpu_tlbstate.active_mm) != next);
41 40
42 if (!cpu_test_and_set(cpu, next->cpu_vm_mask)) { 41 if (!cpu_test_and_set(cpu, next->cpu_vm_mask)) {
43 /* We were in lazy tlb mode and leave_mm disabled 42 /* We were in lazy tlb mode and leave_mm disabled
diff --git a/arch/x86/include/asm/mmzone_32.h b/arch/x86/include/asm/mmzone_32.h
index 485bdf059ffb..07f1af494ca5 100644
--- a/arch/x86/include/asm/mmzone_32.h
+++ b/arch/x86/include/asm/mmzone_32.h
@@ -34,10 +34,14 @@ static inline void get_memcfg_numa(void)
34 34
35extern int early_pfn_to_nid(unsigned long pfn); 35extern int early_pfn_to_nid(unsigned long pfn);
36 36
37extern void resume_map_numa_kva(pgd_t *pgd);
38
37#else /* !CONFIG_NUMA */ 39#else /* !CONFIG_NUMA */
38 40
39#define get_memcfg_numa get_memcfg_numa_flat 41#define get_memcfg_numa get_memcfg_numa_flat
40 42
43static inline void resume_map_numa_kva(pgd_t *pgd) {}
44
41#endif /* CONFIG_NUMA */ 45#endif /* CONFIG_NUMA */
42 46
43#ifdef CONFIG_DISCONTIGMEM 47#ifdef CONFIG_DISCONTIGMEM
diff --git a/arch/x86/include/asm/msr-index.h b/arch/x86/include/asm/msr-index.h
index e38859d577a1..cb58643947b9 100644
--- a/arch/x86/include/asm/msr-index.h
+++ b/arch/x86/include/asm/msr-index.h
@@ -85,7 +85,9 @@
85/* AMD64 MSRs. Not complete. See the architecture manual for a more 85/* AMD64 MSRs. Not complete. See the architecture manual for a more
86 complete list. */ 86 complete list. */
87 87
88#define MSR_AMD64_PATCH_LEVEL 0x0000008b
88#define MSR_AMD64_NB_CFG 0xc001001f 89#define MSR_AMD64_NB_CFG 0xc001001f
90#define MSR_AMD64_PATCH_LOADER 0xc0010020
89#define MSR_AMD64_IBSFETCHCTL 0xc0011030 91#define MSR_AMD64_IBSFETCHCTL 0xc0011030
90#define MSR_AMD64_IBSFETCHLINAD 0xc0011031 92#define MSR_AMD64_IBSFETCHLINAD 0xc0011031
91#define MSR_AMD64_IBSFETCHPHYSAD 0xc0011032 93#define MSR_AMD64_IBSFETCHPHYSAD 0xc0011032
diff --git a/arch/x86/include/asm/msr.h b/arch/x86/include/asm/msr.h
index c2a812ebde89..4640ddd58fb9 100644
--- a/arch/x86/include/asm/msr.h
+++ b/arch/x86/include/asm/msr.h
@@ -22,10 +22,10 @@ static inline unsigned long long native_read_tscp(unsigned int *aux)
22} 22}
23 23
24/* 24/*
25 * i386 calling convention returns 64-bit value in edx:eax, while 25 * both i386 and x86_64 returns 64-bit value in edx:eax, but gcc's "A"
26 * x86_64 returns at rax. Also, the "A" constraint does not really 26 * constraint has different meanings. For i386, "A" means exactly
27 * mean rdx:rax in x86_64, so we need specialized behaviour for each 27 * edx:eax, while for x86_64 it doesn't mean rdx:rax or edx:eax. Instead,
28 * architecture 28 * it means rax *or* rdx.
29 */ 29 */
30#ifdef CONFIG_X86_64 30#ifdef CONFIG_X86_64
31#define DECLARE_ARGS(val, low, high) unsigned low, high 31#define DECLARE_ARGS(val, low, high) unsigned low, high
@@ -181,10 +181,10 @@ static inline int rdmsrl_amd_safe(unsigned msr, unsigned long long *p)
181} 181}
182 182
183#define rdtscl(low) \ 183#define rdtscl(low) \
184 ((low) = (u32)native_read_tsc()) 184 ((low) = (u32)__native_read_tsc())
185 185
186#define rdtscll(val) \ 186#define rdtscll(val) \
187 ((val) = native_read_tsc()) 187 ((val) = __native_read_tsc())
188 188
189#define rdpmc(counter, low, high) \ 189#define rdpmc(counter, low, high) \
190do { \ 190do { \
diff --git a/arch/x86/include/asm/numaq/wakecpu.h b/arch/x86/include/asm/numaq/wakecpu.h
index c577bda5b1c5..6f499df8eddb 100644
--- a/arch/x86/include/asm/numaq/wakecpu.h
+++ b/arch/x86/include/asm/numaq/wakecpu.h
@@ -3,12 +3,8 @@
3 3
4/* This file copes with machines that wakeup secondary CPUs by NMIs */ 4/* This file copes with machines that wakeup secondary CPUs by NMIs */
5 5
6#define WAKE_SECONDARY_VIA_NMI 6#define TRAMPOLINE_PHYS_LOW (0x8)
7 7#define TRAMPOLINE_PHYS_HIGH (0xa)
8#define TRAMPOLINE_LOW phys_to_virt(0x8)
9#define TRAMPOLINE_HIGH phys_to_virt(0xa)
10
11#define boot_cpu_apicid boot_cpu_logical_apicid
12 8
13/* We don't do anything here because we use NMI's to boot instead */ 9/* We don't do anything here because we use NMI's to boot instead */
14static inline void wait_for_init_deassert(atomic_t *deassert) 10static inline void wait_for_init_deassert(atomic_t *deassert)
@@ -27,17 +23,23 @@ static inline void smp_callin_clear_local_apic(void)
27static inline void store_NMI_vector(unsigned short *high, unsigned short *low) 23static inline void store_NMI_vector(unsigned short *high, unsigned short *low)
28{ 24{
29 printk("Storing NMI vector\n"); 25 printk("Storing NMI vector\n");
30 *high = *((volatile unsigned short *) TRAMPOLINE_HIGH); 26 *high =
31 *low = *((volatile unsigned short *) TRAMPOLINE_LOW); 27 *((volatile unsigned short *)phys_to_virt(TRAMPOLINE_PHYS_HIGH));
28 *low =
29 *((volatile unsigned short *)phys_to_virt(TRAMPOLINE_PHYS_LOW));
32} 30}
33 31
34static inline void restore_NMI_vector(unsigned short *high, unsigned short *low) 32static inline void restore_NMI_vector(unsigned short *high, unsigned short *low)
35{ 33{
36 printk("Restoring NMI vector\n"); 34 printk("Restoring NMI vector\n");
37 *((volatile unsigned short *) TRAMPOLINE_HIGH) = *high; 35 *((volatile unsigned short *)phys_to_virt(TRAMPOLINE_PHYS_HIGH)) =
38 *((volatile unsigned short *) TRAMPOLINE_LOW) = *low; 36 *high;
37 *((volatile unsigned short *)phys_to_virt(TRAMPOLINE_PHYS_LOW)) =
38 *low;
39} 39}
40 40
41#define inquire_remote_apic(apicid) {} 41static inline void inquire_remote_apic(int apicid)
42{
43}
42 44
43#endif /* __ASM_NUMAQ_WAKECPU_H */ 45#endif /* __ASM_NUMAQ_WAKECPU_H */
diff --git a/arch/x86/include/asm/pci.h b/arch/x86/include/asm/pci.h
index 875b38edf193..647781298e7e 100644
--- a/arch/x86/include/asm/pci.h
+++ b/arch/x86/include/asm/pci.h
@@ -19,6 +19,8 @@ struct pci_sysdata {
19}; 19};
20 20
21extern int pci_routeirq; 21extern int pci_routeirq;
22extern int noioapicquirk;
23extern int noioapicreroute;
22 24
23/* scan a bus after allocating a pci_sysdata for it */ 25/* scan a bus after allocating a pci_sysdata for it */
24extern struct pci_bus *pci_scan_bus_on_node(int busno, struct pci_ops *ops, 26extern struct pci_bus *pci_scan_bus_on_node(int busno, struct pci_ops *ops,
diff --git a/arch/x86/include/asm/pci_64.h b/arch/x86/include/asm/pci_64.h
index 5b28995d664e..d02d936840a3 100644
--- a/arch/x86/include/asm/pci_64.h
+++ b/arch/x86/include/asm/pci_64.h
@@ -34,8 +34,6 @@ extern void pci_iommu_alloc(void);
34 */ 34 */
35#define PCI_DMA_BUS_IS_PHYS (dma_ops->is_phys) 35#define PCI_DMA_BUS_IS_PHYS (dma_ops->is_phys)
36 36
37#if defined(CONFIG_GART_IOMMU) || defined(CONFIG_CALGARY_IOMMU)
38
39#define DECLARE_PCI_UNMAP_ADDR(ADDR_NAME) \ 37#define DECLARE_PCI_UNMAP_ADDR(ADDR_NAME) \
40 dma_addr_t ADDR_NAME; 38 dma_addr_t ADDR_NAME;
41#define DECLARE_PCI_UNMAP_LEN(LEN_NAME) \ 39#define DECLARE_PCI_UNMAP_LEN(LEN_NAME) \
@@ -49,18 +47,6 @@ extern void pci_iommu_alloc(void);
49#define pci_unmap_len_set(PTR, LEN_NAME, VAL) \ 47#define pci_unmap_len_set(PTR, LEN_NAME, VAL) \
50 (((PTR)->LEN_NAME) = (VAL)) 48 (((PTR)->LEN_NAME) = (VAL))
51 49
52#else
53/* No IOMMU */
54
55#define DECLARE_PCI_UNMAP_ADDR(ADDR_NAME)
56#define DECLARE_PCI_UNMAP_LEN(LEN_NAME)
57#define pci_unmap_addr(PTR, ADDR_NAME) (0)
58#define pci_unmap_addr_set(PTR, ADDR_NAME, VAL) do { } while (0)
59#define pci_unmap_len(PTR, LEN_NAME) (0)
60#define pci_unmap_len_set(PTR, LEN_NAME, VAL) do { } while (0)
61
62#endif
63
64#endif /* __KERNEL__ */ 50#endif /* __KERNEL__ */
65 51
66#endif /* _ASM_X86_PCI_64_H */ 52#endif /* _ASM_X86_PCI_64_H */
diff --git a/arch/x86/include/asm/pgtable-2level.h b/arch/x86/include/asm/pgtable-2level.h
index b17edfd23628..e0d199fe1d83 100644
--- a/arch/x86/include/asm/pgtable-2level.h
+++ b/arch/x86/include/asm/pgtable-2level.h
@@ -56,23 +56,55 @@ static inline pte_t native_ptep_get_and_clear(pte_t *xp)
56#define pte_none(x) (!(x).pte_low) 56#define pte_none(x) (!(x).pte_low)
57 57
58/* 58/*
59 * Bits 0, 6 and 7 are taken, split up the 29 bits of offset 59 * Bits _PAGE_BIT_PRESENT, _PAGE_BIT_FILE and _PAGE_BIT_PROTNONE are taken,
60 * into this range: 60 * split up the 29 bits of offset into this range:
61 */ 61 */
62#define PTE_FILE_MAX_BITS 29 62#define PTE_FILE_MAX_BITS 29
63#define PTE_FILE_SHIFT1 (_PAGE_BIT_PRESENT + 1)
64#if _PAGE_BIT_FILE < _PAGE_BIT_PROTNONE
65#define PTE_FILE_SHIFT2 (_PAGE_BIT_FILE + 1)
66#define PTE_FILE_SHIFT3 (_PAGE_BIT_PROTNONE + 1)
67#else
68#define PTE_FILE_SHIFT2 (_PAGE_BIT_PROTNONE + 1)
69#define PTE_FILE_SHIFT3 (_PAGE_BIT_FILE + 1)
70#endif
71#define PTE_FILE_BITS1 (PTE_FILE_SHIFT2 - PTE_FILE_SHIFT1 - 1)
72#define PTE_FILE_BITS2 (PTE_FILE_SHIFT3 - PTE_FILE_SHIFT2 - 1)
63 73
64#define pte_to_pgoff(pte) \ 74#define pte_to_pgoff(pte) \
65 ((((pte).pte_low >> 1) & 0x1f) + (((pte).pte_low >> 8) << 5)) 75 ((((pte).pte_low >> PTE_FILE_SHIFT1) \
76 & ((1U << PTE_FILE_BITS1) - 1)) \
77 + ((((pte).pte_low >> PTE_FILE_SHIFT2) \
78 & ((1U << PTE_FILE_BITS2) - 1)) << PTE_FILE_BITS1) \
79 + (((pte).pte_low >> PTE_FILE_SHIFT3) \
80 << (PTE_FILE_BITS1 + PTE_FILE_BITS2)))
66 81
67#define pgoff_to_pte(off) \ 82#define pgoff_to_pte(off) \
68 ((pte_t) { .pte_low = (((off) & 0x1f) << 1) + \ 83 ((pte_t) { .pte_low = \
69 (((off) >> 5) << 8) + _PAGE_FILE }) 84 (((off) & ((1U << PTE_FILE_BITS1) - 1)) << PTE_FILE_SHIFT1) \
85 + ((((off) >> PTE_FILE_BITS1) & ((1U << PTE_FILE_BITS2) - 1)) \
86 << PTE_FILE_SHIFT2) \
87 + (((off) >> (PTE_FILE_BITS1 + PTE_FILE_BITS2)) \
88 << PTE_FILE_SHIFT3) \
89 + _PAGE_FILE })
70 90
71/* Encode and de-code a swap entry */ 91/* Encode and de-code a swap entry */
72#define __swp_type(x) (((x).val >> 1) & 0x1f) 92#if _PAGE_BIT_FILE < _PAGE_BIT_PROTNONE
73#define __swp_offset(x) ((x).val >> 8) 93#define SWP_TYPE_BITS (_PAGE_BIT_FILE - _PAGE_BIT_PRESENT - 1)
74#define __swp_entry(type, offset) \ 94#define SWP_OFFSET_SHIFT (_PAGE_BIT_PROTNONE + 1)
75 ((swp_entry_t) { ((type) << 1) | ((offset) << 8) }) 95#else
96#define SWP_TYPE_BITS (_PAGE_BIT_PROTNONE - _PAGE_BIT_PRESENT - 1)
97#define SWP_OFFSET_SHIFT (_PAGE_BIT_FILE + 1)
98#endif
99
100#define MAX_SWAPFILES_CHECK() BUILD_BUG_ON(MAX_SWAPFILES_SHIFT > SWP_TYPE_BITS)
101
102#define __swp_type(x) (((x).val >> (_PAGE_BIT_PRESENT + 1)) \
103 & ((1U << SWP_TYPE_BITS) - 1))
104#define __swp_offset(x) ((x).val >> SWP_OFFSET_SHIFT)
105#define __swp_entry(type, offset) ((swp_entry_t) { \
106 ((type) << (_PAGE_BIT_PRESENT + 1)) \
107 | ((offset) << SWP_OFFSET_SHIFT) })
76#define __pte_to_swp_entry(pte) ((swp_entry_t) { (pte).pte_low }) 108#define __pte_to_swp_entry(pte) ((swp_entry_t) { (pte).pte_low })
77#define __swp_entry_to_pte(x) ((pte_t) { .pte = (x).val }) 109#define __swp_entry_to_pte(x) ((pte_t) { .pte = (x).val })
78 110
diff --git a/arch/x86/include/asm/pgtable-3level.h b/arch/x86/include/asm/pgtable-3level.h
index 52597aeadfff..447da43cddb3 100644
--- a/arch/x86/include/asm/pgtable-3level.h
+++ b/arch/x86/include/asm/pgtable-3level.h
@@ -166,6 +166,7 @@ static inline int pte_none(pte_t pte)
166#define PTE_FILE_MAX_BITS 32 166#define PTE_FILE_MAX_BITS 32
167 167
168/* Encode and de-code a swap entry */ 168/* Encode and de-code a swap entry */
169#define MAX_SWAPFILES_CHECK() BUILD_BUG_ON(MAX_SWAPFILES_SHIFT > 5)
169#define __swp_type(x) (((x).val) & 0x1f) 170#define __swp_type(x) (((x).val) & 0x1f)
170#define __swp_offset(x) ((x).val >> 5) 171#define __swp_offset(x) ((x).val >> 5)
171#define __swp_entry(type, offset) ((swp_entry_t){(type) | (offset) << 5}) 172#define __swp_entry(type, offset) ((swp_entry_t){(type) | (offset) << 5})
diff --git a/arch/x86/include/asm/pgtable.h b/arch/x86/include/asm/pgtable.h
index c012f3b11671..83e69f4a37f0 100644
--- a/arch/x86/include/asm/pgtable.h
+++ b/arch/x86/include/asm/pgtable.h
@@ -10,7 +10,6 @@
10#define _PAGE_BIT_PCD 4 /* page cache disabled */ 10#define _PAGE_BIT_PCD 4 /* page cache disabled */
11#define _PAGE_BIT_ACCESSED 5 /* was accessed (raised by CPU) */ 11#define _PAGE_BIT_ACCESSED 5 /* was accessed (raised by CPU) */
12#define _PAGE_BIT_DIRTY 6 /* was written to (raised by CPU) */ 12#define _PAGE_BIT_DIRTY 6 /* was written to (raised by CPU) */
13#define _PAGE_BIT_FILE 6
14#define _PAGE_BIT_PSE 7 /* 4 MB (or 2MB) page */ 13#define _PAGE_BIT_PSE 7 /* 4 MB (or 2MB) page */
15#define _PAGE_BIT_PAT 7 /* on 4KB pages */ 14#define _PAGE_BIT_PAT 7 /* on 4KB pages */
16#define _PAGE_BIT_GLOBAL 8 /* Global TLB entry PPro+ */ 15#define _PAGE_BIT_GLOBAL 8 /* Global TLB entry PPro+ */
@@ -22,6 +21,12 @@
22#define _PAGE_BIT_CPA_TEST _PAGE_BIT_UNUSED1 21#define _PAGE_BIT_CPA_TEST _PAGE_BIT_UNUSED1
23#define _PAGE_BIT_NX 63 /* No execute: only valid after cpuid check */ 22#define _PAGE_BIT_NX 63 /* No execute: only valid after cpuid check */
24 23
24/* If _PAGE_BIT_PRESENT is clear, we use these: */
25/* - if the user mapped it with PROT_NONE; pte_present gives true */
26#define _PAGE_BIT_PROTNONE _PAGE_BIT_GLOBAL
27/* - set: nonlinear file mapping, saved PTE; unset:swap */
28#define _PAGE_BIT_FILE _PAGE_BIT_DIRTY
29
25#define _PAGE_PRESENT (_AT(pteval_t, 1) << _PAGE_BIT_PRESENT) 30#define _PAGE_PRESENT (_AT(pteval_t, 1) << _PAGE_BIT_PRESENT)
26#define _PAGE_RW (_AT(pteval_t, 1) << _PAGE_BIT_RW) 31#define _PAGE_RW (_AT(pteval_t, 1) << _PAGE_BIT_RW)
27#define _PAGE_USER (_AT(pteval_t, 1) << _PAGE_BIT_USER) 32#define _PAGE_USER (_AT(pteval_t, 1) << _PAGE_BIT_USER)
@@ -46,11 +51,8 @@
46#define _PAGE_NX (_AT(pteval_t, 0)) 51#define _PAGE_NX (_AT(pteval_t, 0))
47#endif 52#endif
48 53
49/* If _PAGE_PRESENT is clear, we use these: */ 54#define _PAGE_FILE (_AT(pteval_t, 1) << _PAGE_BIT_FILE)
50#define _PAGE_FILE _PAGE_DIRTY /* nonlinear file mapping, 55#define _PAGE_PROTNONE (_AT(pteval_t, 1) << _PAGE_BIT_PROTNONE)
51 * saved PTE; unset:swap */
52#define _PAGE_PROTNONE _PAGE_PSE /* if the user mapped it with PROT_NONE;
53 pte_present gives true */
54 56
55#define _PAGE_TABLE (_PAGE_PRESENT | _PAGE_RW | _PAGE_USER | \ 57#define _PAGE_TABLE (_PAGE_PRESENT | _PAGE_RW | _PAGE_USER | \
56 _PAGE_ACCESSED | _PAGE_DIRTY) 58 _PAGE_ACCESSED | _PAGE_DIRTY)
@@ -158,8 +160,19 @@
158#define PGD_IDENT_ATTR 0x001 /* PRESENT (no other attributes) */ 160#define PGD_IDENT_ATTR 0x001 /* PRESENT (no other attributes) */
159#endif 161#endif
160 162
163/*
164 * Macro to mark a page protection value as UC-
165 */
166#define pgprot_noncached(prot) \
167 ((boot_cpu_data.x86 > 3) \
168 ? (__pgprot(pgprot_val(prot) | _PAGE_CACHE_UC_MINUS)) \
169 : (prot))
170
161#ifndef __ASSEMBLY__ 171#ifndef __ASSEMBLY__
162 172
173#define pgprot_writecombine pgprot_writecombine
174extern pgprot_t pgprot_writecombine(pgprot_t prot);
175
163/* 176/*
164 * ZERO_PAGE is a global shared page that is always zero: used 177 * ZERO_PAGE is a global shared page that is always zero: used
165 * for zero-mapped memory areas etc.. 178 * for zero-mapped memory areas etc..
@@ -329,6 +342,9 @@ static inline pgprot_t pgprot_modify(pgprot_t oldprot, pgprot_t newprot)
329#define canon_pgprot(p) __pgprot(pgprot_val(p) & __supported_pte_mask) 342#define canon_pgprot(p) __pgprot(pgprot_val(p) & __supported_pte_mask)
330 343
331#ifndef __ASSEMBLY__ 344#ifndef __ASSEMBLY__
345/* Indicate that x86 has its own track and untrack pfn vma functions */
346#define __HAVE_PFNMAP_TRACKING
347
332#define __HAVE_PHYS_MEM_ACCESS_PROT 348#define __HAVE_PHYS_MEM_ACCESS_PROT
333struct file; 349struct file;
334pgprot_t phys_mem_access_prot(struct file *file, unsigned long pfn, 350pgprot_t phys_mem_access_prot(struct file *file, unsigned long pfn,
diff --git a/arch/x86/include/asm/pgtable_32.h b/arch/x86/include/asm/pgtable_32.h
index f9d5889b336b..72b020deb46b 100644
--- a/arch/x86/include/asm/pgtable_32.h
+++ b/arch/x86/include/asm/pgtable_32.h
@@ -101,15 +101,6 @@ extern unsigned long pg0[];
101#endif 101#endif
102 102
103/* 103/*
104 * Macro to mark a page protection value as "uncacheable".
105 * On processors which do not support it, this is a no-op.
106 */
107#define pgprot_noncached(prot) \
108 ((boot_cpu_data.x86 > 3) \
109 ? (__pgprot(pgprot_val(prot) | _PAGE_PCD | _PAGE_PWT)) \
110 : (prot))
111
112/*
113 * Conversion functions: convert a page and protection to a page entry, 104 * Conversion functions: convert a page and protection to a page entry,
114 * and a page entry and page directory to the page they refer to. 105 * and a page entry and page directory to the page they refer to.
115 */ 106 */
diff --git a/arch/x86/include/asm/pgtable_64.h b/arch/x86/include/asm/pgtable_64.h
index 545a0e042bb2..ba09289accaa 100644
--- a/arch/x86/include/asm/pgtable_64.h
+++ b/arch/x86/include/asm/pgtable_64.h
@@ -146,7 +146,7 @@ static inline void native_pgd_clear(pgd_t *pgd)
146#define PGDIR_MASK (~(PGDIR_SIZE - 1)) 146#define PGDIR_MASK (~(PGDIR_SIZE - 1))
147 147
148 148
149#define MAXMEM _AC(0x00003fffffffffff, UL) 149#define MAXMEM _AC(__AC(1, UL) << MAX_PHYSMEM_BITS, UL)
150#define VMALLOC_START _AC(0xffffc20000000000, UL) 150#define VMALLOC_START _AC(0xffffc20000000000, UL)
151#define VMALLOC_END _AC(0xffffe1ffffffffff, UL) 151#define VMALLOC_END _AC(0xffffe1ffffffffff, UL)
152#define VMEMMAP_START _AC(0xffffe20000000000, UL) 152#define VMEMMAP_START _AC(0xffffe20000000000, UL)
@@ -177,12 +177,6 @@ static inline int pmd_bad(pmd_t pmd)
177#define pages_to_mb(x) ((x) >> (20 - PAGE_SHIFT)) /* FIXME: is this right? */ 177#define pages_to_mb(x) ((x) >> (20 - PAGE_SHIFT)) /* FIXME: is this right? */
178 178
179/* 179/*
180 * Macro to mark a page protection value as "uncacheable".
181 */
182#define pgprot_noncached(prot) \
183 (__pgprot(pgprot_val((prot)) | _PAGE_PCD | _PAGE_PWT))
184
185/*
186 * Conversion functions: convert a page and protection to a page entry, 180 * Conversion functions: convert a page and protection to a page entry,
187 * and a page entry and page directory to the page they refer to. 181 * and a page entry and page directory to the page they refer to.
188 */ 182 */
@@ -250,10 +244,22 @@ static inline int pud_large(pud_t pte)
250extern int direct_gbpages; 244extern int direct_gbpages;
251 245
252/* Encode and de-code a swap entry */ 246/* Encode and de-code a swap entry */
253#define __swp_type(x) (((x).val >> 1) & 0x3f) 247#if _PAGE_BIT_FILE < _PAGE_BIT_PROTNONE
254#define __swp_offset(x) ((x).val >> 8) 248#define SWP_TYPE_BITS (_PAGE_BIT_FILE - _PAGE_BIT_PRESENT - 1)
255#define __swp_entry(type, offset) ((swp_entry_t) { ((type) << 1) | \ 249#define SWP_OFFSET_SHIFT (_PAGE_BIT_PROTNONE + 1)
256 ((offset) << 8) }) 250#else
251#define SWP_TYPE_BITS (_PAGE_BIT_PROTNONE - _PAGE_BIT_PRESENT - 1)
252#define SWP_OFFSET_SHIFT (_PAGE_BIT_FILE + 1)
253#endif
254
255#define MAX_SWAPFILES_CHECK() BUILD_BUG_ON(MAX_SWAPFILES_SHIFT > SWP_TYPE_BITS)
256
257#define __swp_type(x) (((x).val >> (_PAGE_BIT_PRESENT + 1)) \
258 & ((1U << SWP_TYPE_BITS) - 1))
259#define __swp_offset(x) ((x).val >> SWP_OFFSET_SHIFT)
260#define __swp_entry(type, offset) ((swp_entry_t) { \
261 ((type) << (_PAGE_BIT_PRESENT + 1)) \
262 | ((offset) << SWP_OFFSET_SHIFT) })
257#define __pte_to_swp_entry(pte) ((swp_entry_t) { pte_val((pte)) }) 263#define __pte_to_swp_entry(pte) ((swp_entry_t) { pte_val((pte)) })
258#define __swp_entry_to_pte(x) ((pte_t) { .pte = (x).val }) 264#define __swp_entry_to_pte(x) ((pte_t) { .pte = (x).val })
259 265
diff --git a/arch/x86/include/asm/prctl.h b/arch/x86/include/asm/prctl.h
index fe681147a4f7..a8894647dd9a 100644
--- a/arch/x86/include/asm/prctl.h
+++ b/arch/x86/include/asm/prctl.h
@@ -6,5 +6,8 @@
6#define ARCH_GET_FS 0x1003 6#define ARCH_GET_FS 0x1003
7#define ARCH_GET_GS 0x1004 7#define ARCH_GET_GS 0x1004
8 8
9#ifdef CONFIG_X86_64
10extern long sys_arch_prctl(int, unsigned long);
11#endif /* CONFIG_X86_64 */
9 12
10#endif /* _ASM_X86_PRCTL_H */ 13#endif /* _ASM_X86_PRCTL_H */
diff --git a/arch/x86/include/asm/processor.h b/arch/x86/include/asm/processor.h
index 5ca01e383269..a570eafa4755 100644
--- a/arch/x86/include/asm/processor.h
+++ b/arch/x86/include/asm/processor.h
@@ -110,6 +110,7 @@ struct cpuinfo_x86 {
110 /* Index into per_cpu list: */ 110 /* Index into per_cpu list: */
111 u16 cpu_index; 111 u16 cpu_index;
112#endif 112#endif
113 unsigned int x86_hyper_vendor;
113} __attribute__((__aligned__(SMP_CACHE_BYTES))); 114} __attribute__((__aligned__(SMP_CACHE_BYTES)));
114 115
115#define X86_VENDOR_INTEL 0 116#define X86_VENDOR_INTEL 0
@@ -123,6 +124,9 @@ struct cpuinfo_x86 {
123 124
124#define X86_VENDOR_UNKNOWN 0xff 125#define X86_VENDOR_UNKNOWN 0xff
125 126
127#define X86_HYPER_VENDOR_NONE 0
128#define X86_HYPER_VENDOR_VMWARE 1
129
126/* 130/*
127 * capabilities of CPUs 131 * capabilities of CPUs
128 */ 132 */
diff --git a/arch/x86/include/asm/ptrace.h b/arch/x86/include/asm/ptrace.h
index d1531c8480b7..eefb0594b058 100644
--- a/arch/x86/include/asm/ptrace.h
+++ b/arch/x86/include/asm/ptrace.h
@@ -271,8 +271,6 @@ extern int do_get_thread_area(struct task_struct *p, int idx,
271extern int do_set_thread_area(struct task_struct *p, int idx, 271extern int do_set_thread_area(struct task_struct *p, int idx,
272 struct user_desc __user *info, int can_allocate); 272 struct user_desc __user *info, int can_allocate);
273 273
274#define __ARCH_WANT_COMPAT_SYS_PTRACE
275
276#endif /* __KERNEL__ */ 274#endif /* __KERNEL__ */
277 275
278#endif /* !__ASSEMBLY__ */ 276#endif /* !__ASSEMBLY__ */
diff --git a/arch/x86/include/asm/reboot.h b/arch/x86/include/asm/reboot.h
index df7710354f85..562d4fd31ba8 100644
--- a/arch/x86/include/asm/reboot.h
+++ b/arch/x86/include/asm/reboot.h
@@ -1,6 +1,8 @@
1#ifndef _ASM_X86_REBOOT_H 1#ifndef _ASM_X86_REBOOT_H
2#define _ASM_X86_REBOOT_H 2#define _ASM_X86_REBOOT_H
3 3
4#include <linux/kdebug.h>
5
4struct pt_regs; 6struct pt_regs;
5 7
6struct machine_ops { 8struct machine_ops {
@@ -18,4 +20,7 @@ void native_machine_crash_shutdown(struct pt_regs *regs);
18void native_machine_shutdown(void); 20void native_machine_shutdown(void);
19void machine_real_restart(const unsigned char *code, int length); 21void machine_real_restart(const unsigned char *code, int length);
20 22
23typedef void (*nmi_shootdown_cb)(int, struct die_args*);
24void nmi_shootdown_cpus(nmi_shootdown_cb callback);
25
21#endif /* _ASM_X86_REBOOT_H */ 26#endif /* _ASM_X86_REBOOT_H */
diff --git a/arch/x86/include/asm/setup.h b/arch/x86/include/asm/setup.h
index f12d37237465..4fcd53fd5f43 100644
--- a/arch/x86/include/asm/setup.h
+++ b/arch/x86/include/asm/setup.h
@@ -8,6 +8,10 @@
8/* Interrupt control for vSMPowered x86_64 systems */ 8/* Interrupt control for vSMPowered x86_64 systems */
9void vsmp_init(void); 9void vsmp_init(void);
10 10
11
12void setup_bios_corruption_check(void);
13
14
11#ifdef CONFIG_X86_VISWS 15#ifdef CONFIG_X86_VISWS
12extern void visws_early_detect(void); 16extern void visws_early_detect(void);
13extern int is_visws_box(void); 17extern int is_visws_box(void);
@@ -16,6 +20,8 @@ static inline void visws_early_detect(void) { }
16static inline int is_visws_box(void) { return 0; } 20static inline int is_visws_box(void) { return 0; }
17#endif 21#endif
18 22
23extern int wakeup_secondary_cpu_via_nmi(int apicid, unsigned long start_eip);
24extern int wakeup_secondary_cpu_via_init(int apicid, unsigned long start_eip);
19/* 25/*
20 * Any setup quirks to be performed? 26 * Any setup quirks to be performed?
21 */ 27 */
@@ -39,6 +45,7 @@ struct x86_quirks {
39 void (*smp_read_mpc_oem)(struct mp_config_oemtable *oemtable, 45 void (*smp_read_mpc_oem)(struct mp_config_oemtable *oemtable,
40 unsigned short oemsize); 46 unsigned short oemsize);
41 int (*setup_ioapic_ids)(void); 47 int (*setup_ioapic_ids)(void);
48 int (*update_genapic)(void);
42}; 49};
43 50
44extern struct x86_quirks *x86_quirks; 51extern struct x86_quirks *x86_quirks;
diff --git a/arch/x86/include/asm/sigframe.h b/arch/x86/include/asm/sigframe.h
new file mode 100644
index 000000000000..4e0fe26d27d3
--- /dev/null
+++ b/arch/x86/include/asm/sigframe.h
@@ -0,0 +1,70 @@
1#ifndef _ASM_X86_SIGFRAME_H
2#define _ASM_X86_SIGFRAME_H
3
4#include <asm/sigcontext.h>
5#include <asm/siginfo.h>
6#include <asm/ucontext.h>
7
8#ifdef CONFIG_X86_32
9#define sigframe_ia32 sigframe
10#define rt_sigframe_ia32 rt_sigframe
11#define sigcontext_ia32 sigcontext
12#define _fpstate_ia32 _fpstate
13#define ucontext_ia32 ucontext
14#else /* !CONFIG_X86_32 */
15
16#ifdef CONFIG_IA32_EMULATION
17#include <asm/ia32.h>
18#endif /* CONFIG_IA32_EMULATION */
19
20#endif /* CONFIG_X86_32 */
21
22#if defined(CONFIG_X86_32) || defined(CONFIG_IA32_EMULATION)
23struct sigframe_ia32 {
24 u32 pretcode;
25 int sig;
26 struct sigcontext_ia32 sc;
27 /*
28 * fpstate is unused. fpstate is moved/allocated after
29 * retcode[] below. This movement allows to have the FP state and the
30 * future state extensions (xsave) stay together.
31 * And at the same time retaining the unused fpstate, prevents changing
32 * the offset of extramask[] in the sigframe and thus prevent any
33 * legacy application accessing/modifying it.
34 */
35 struct _fpstate_ia32 fpstate_unused;
36#ifdef CONFIG_IA32_EMULATION
37 unsigned int extramask[_COMPAT_NSIG_WORDS-1];
38#else /* !CONFIG_IA32_EMULATION */
39 unsigned long extramask[_NSIG_WORDS-1];
40#endif /* CONFIG_IA32_EMULATION */
41 char retcode[8];
42 /* fp state follows here */
43};
44
45struct rt_sigframe_ia32 {
46 u32 pretcode;
47 int sig;
48 u32 pinfo;
49 u32 puc;
50#ifdef CONFIG_IA32_EMULATION
51 compat_siginfo_t info;
52#else /* !CONFIG_IA32_EMULATION */
53 struct siginfo info;
54#endif /* CONFIG_IA32_EMULATION */
55 struct ucontext_ia32 uc;
56 char retcode[8];
57 /* fp state follows here */
58};
59#endif /* defined(CONFIG_X86_32) || defined(CONFIG_IA32_EMULATION) */
60
61#ifdef CONFIG_X86_64
62struct rt_sigframe {
63 char __user *pretcode;
64 struct ucontext uc;
65 struct siginfo info;
66 /* fp state follows here */
67};
68#endif /* CONFIG_X86_64 */
69
70#endif /* _ASM_X86_SIGFRAME_H */
diff --git a/arch/x86/include/asm/signal.h b/arch/x86/include/asm/signal.h
index 96ac44f275da..7761a5d554bb 100644
--- a/arch/x86/include/asm/signal.h
+++ b/arch/x86/include/asm/signal.h
@@ -121,6 +121,10 @@ typedef unsigned long sigset_t;
121 121
122#ifndef __ASSEMBLY__ 122#ifndef __ASSEMBLY__
123 123
124# ifdef __KERNEL__
125extern void do_notify_resume(struct pt_regs *, void *, __u32);
126# endif /* __KERNEL__ */
127
124#ifdef __i386__ 128#ifdef __i386__
125# ifdef __KERNEL__ 129# ifdef __KERNEL__
126struct old_sigaction { 130struct old_sigaction {
@@ -141,8 +145,6 @@ struct k_sigaction {
141 struct sigaction sa; 145 struct sigaction sa;
142}; 146};
143 147
144extern void do_notify_resume(struct pt_regs *, void *, __u32);
145
146# else /* __KERNEL__ */ 148# else /* __KERNEL__ */
147/* Here we must cater to libcs that poke about in kernel headers. */ 149/* Here we must cater to libcs that poke about in kernel headers. */
148 150
diff --git a/arch/x86/include/asm/sparsemem.h b/arch/x86/include/asm/sparsemem.h
index be44f7dab395..e3cc3c063ec5 100644
--- a/arch/x86/include/asm/sparsemem.h
+++ b/arch/x86/include/asm/sparsemem.h
@@ -27,7 +27,7 @@
27#else /* CONFIG_X86_32 */ 27#else /* CONFIG_X86_32 */
28# define SECTION_SIZE_BITS 27 /* matt - 128 is convenient right now */ 28# define SECTION_SIZE_BITS 27 /* matt - 128 is convenient right now */
29# define MAX_PHYSADDR_BITS 44 29# define MAX_PHYSADDR_BITS 44
30# define MAX_PHYSMEM_BITS 44 30# define MAX_PHYSMEM_BITS 44 /* Can be max 45 bits */
31#endif 31#endif
32 32
33#endif /* CONFIG_SPARSEMEM */ 33#endif /* CONFIG_SPARSEMEM */
diff --git a/arch/x86/include/asm/syscalls.h b/arch/x86/include/asm/syscalls.h
index 87803da44010..9c6797c3e56c 100644
--- a/arch/x86/include/asm/syscalls.h
+++ b/arch/x86/include/asm/syscalls.h
@@ -19,6 +19,13 @@
19/* kernel/ioport.c */ 19/* kernel/ioport.c */
20asmlinkage long sys_ioperm(unsigned long, unsigned long, int); 20asmlinkage long sys_ioperm(unsigned long, unsigned long, int);
21 21
22/* kernel/ldt.c */
23asmlinkage int sys_modify_ldt(int, void __user *, unsigned long);
24
25/* kernel/tls.c */
26asmlinkage int sys_set_thread_area(struct user_desc __user *);
27asmlinkage int sys_get_thread_area(struct user_desc __user *);
28
22/* X86_32 only */ 29/* X86_32 only */
23#ifdef CONFIG_X86_32 30#ifdef CONFIG_X86_32
24/* kernel/process_32.c */ 31/* kernel/process_32.c */
@@ -33,14 +40,11 @@ asmlinkage int sys_sigaction(int, const struct old_sigaction __user *,
33 struct old_sigaction __user *); 40 struct old_sigaction __user *);
34asmlinkage int sys_sigaltstack(unsigned long); 41asmlinkage int sys_sigaltstack(unsigned long);
35asmlinkage unsigned long sys_sigreturn(unsigned long); 42asmlinkage unsigned long sys_sigreturn(unsigned long);
36asmlinkage int sys_rt_sigreturn(unsigned long); 43asmlinkage int sys_rt_sigreturn(struct pt_regs);
37 44
38/* kernel/ioport.c */ 45/* kernel/ioport.c */
39asmlinkage long sys_iopl(unsigned long); 46asmlinkage long sys_iopl(unsigned long);
40 47
41/* kernel/ldt.c */
42asmlinkage int sys_modify_ldt(int, void __user *, unsigned long);
43
44/* kernel/sys_i386_32.c */ 48/* kernel/sys_i386_32.c */
45asmlinkage long sys_mmap2(unsigned long, unsigned long, unsigned long, 49asmlinkage long sys_mmap2(unsigned long, unsigned long, unsigned long,
46 unsigned long, unsigned long, unsigned long); 50 unsigned long, unsigned long, unsigned long);
@@ -54,10 +58,6 @@ asmlinkage int sys_uname(struct old_utsname __user *);
54struct oldold_utsname; 58struct oldold_utsname;
55asmlinkage int sys_olduname(struct oldold_utsname __user *); 59asmlinkage int sys_olduname(struct oldold_utsname __user *);
56 60
57/* kernel/tls.c */
58asmlinkage int sys_set_thread_area(struct user_desc __user *);
59asmlinkage int sys_get_thread_area(struct user_desc __user *);
60
61/* kernel/vm86_32.c */ 61/* kernel/vm86_32.c */
62asmlinkage int sys_vm86old(struct pt_regs); 62asmlinkage int sys_vm86old(struct pt_regs);
63asmlinkage int sys_vm86(struct pt_regs); 63asmlinkage int sys_vm86(struct pt_regs);
diff --git a/arch/x86/include/asm/system.h b/arch/x86/include/asm/system.h
index 2ed3f0f44ff7..8e626ea33a1a 100644
--- a/arch/x86/include/asm/system.h
+++ b/arch/x86/include/asm/system.h
@@ -17,12 +17,12 @@
17# define AT_VECTOR_SIZE_ARCH 1 17# define AT_VECTOR_SIZE_ARCH 1
18#endif 18#endif
19 19
20#ifdef CONFIG_X86_32
21
22struct task_struct; /* one of the stranger aspects of C forward declarations */ 20struct task_struct; /* one of the stranger aspects of C forward declarations */
23struct task_struct *__switch_to(struct task_struct *prev, 21struct task_struct *__switch_to(struct task_struct *prev,
24 struct task_struct *next); 22 struct task_struct *next);
25 23
24#ifdef CONFIG_X86_32
25
26/* 26/*
27 * Saving eflags is important. It switches not only IOPL between tasks, 27 * Saving eflags is important. It switches not only IOPL between tasks,
28 * it also protects other tasks from NT leaking through sysenter etc. 28 * it also protects other tasks from NT leaking through sysenter etc.
@@ -314,6 +314,8 @@ extern void free_init_pages(char *what, unsigned long begin, unsigned long end);
314 314
315void default_idle(void); 315void default_idle(void);
316 316
317void stop_this_cpu(void *dummy);
318
317/* 319/*
318 * Force strict CPU ordering. 320 * Force strict CPU ordering.
319 * And yes, this is required on UP too when we're talking 321 * And yes, this is required on UP too when we're talking
diff --git a/arch/x86/include/asm/thread_info.h b/arch/x86/include/asm/thread_info.h
index e44d379faad2..8dbc57390d25 100644
--- a/arch/x86/include/asm/thread_info.h
+++ b/arch/x86/include/asm/thread_info.h
@@ -24,7 +24,7 @@ struct exec_domain;
24struct thread_info { 24struct thread_info {
25 struct task_struct *task; /* main task structure */ 25 struct task_struct *task; /* main task structure */
26 struct exec_domain *exec_domain; /* execution domain */ 26 struct exec_domain *exec_domain; /* execution domain */
27 unsigned long flags; /* low level flags */ 27 __u32 flags; /* low level flags */
28 __u32 status; /* thread synchronous flags */ 28 __u32 status; /* thread synchronous flags */
29 __u32 cpu; /* current CPU */ 29 __u32 cpu; /* current CPU */
30 int preempt_count; /* 0 => preemptable, 30 int preempt_count; /* 0 => preemptable,
diff --git a/arch/x86/include/asm/topology.h b/arch/x86/include/asm/topology.h
index 4850e4b02b61..ff386ff50ed7 100644
--- a/arch/x86/include/asm/topology.h
+++ b/arch/x86/include/asm/topology.h
@@ -239,7 +239,7 @@ struct pci_bus;
239void set_pci_bus_resources_arch_default(struct pci_bus *b); 239void set_pci_bus_resources_arch_default(struct pci_bus *b);
240 240
241#ifdef CONFIG_SMP 241#ifdef CONFIG_SMP
242#define mc_capable() (boot_cpu_data.x86_max_cores > 1) 242#define mc_capable() (cpus_weight(per_cpu(cpu_core_map, 0)) != nr_cpu_ids)
243#define smt_capable() (smp_num_siblings > 1) 243#define smt_capable() (smp_num_siblings > 1)
244#endif 244#endif
245 245
diff --git a/arch/x86/include/asm/trampoline.h b/arch/x86/include/asm/trampoline.h
index fa0d79facdbc..780ba0ab94f9 100644
--- a/arch/x86/include/asm/trampoline.h
+++ b/arch/x86/include/asm/trampoline.h
@@ -3,6 +3,7 @@
3 3
4#ifndef __ASSEMBLY__ 4#ifndef __ASSEMBLY__
5 5
6#ifdef CONFIG_X86_TRAMPOLINE
6/* 7/*
7 * Trampoline 80x86 program as an array. 8 * Trampoline 80x86 program as an array.
8 */ 9 */
@@ -13,8 +14,14 @@ extern unsigned char *trampoline_base;
13extern unsigned long init_rsp; 14extern unsigned long init_rsp;
14extern unsigned long initial_code; 15extern unsigned long initial_code;
15 16
17#define TRAMPOLINE_SIZE roundup(trampoline_end - trampoline_data, PAGE_SIZE)
16#define TRAMPOLINE_BASE 0x6000 18#define TRAMPOLINE_BASE 0x6000
19
17extern unsigned long setup_trampoline(void); 20extern unsigned long setup_trampoline(void);
21extern void __init reserve_trampoline_memory(void);
22#else
23static inline void reserve_trampoline_memory(void) {};
24#endif /* CONFIG_X86_TRAMPOLINE */
18 25
19#endif /* __ASSEMBLY__ */ 26#endif /* __ASSEMBLY__ */
20 27
diff --git a/arch/x86/include/asm/traps.h b/arch/x86/include/asm/traps.h
index 45dee286e45c..2ee0a3bceedf 100644
--- a/arch/x86/include/asm/traps.h
+++ b/arch/x86/include/asm/traps.h
@@ -46,6 +46,10 @@ dotraplinkage void do_coprocessor_segment_overrun(struct pt_regs *, long);
46dotraplinkage void do_invalid_TSS(struct pt_regs *, long); 46dotraplinkage void do_invalid_TSS(struct pt_regs *, long);
47dotraplinkage void do_segment_not_present(struct pt_regs *, long); 47dotraplinkage void do_segment_not_present(struct pt_regs *, long);
48dotraplinkage void do_stack_segment(struct pt_regs *, long); 48dotraplinkage void do_stack_segment(struct pt_regs *, long);
49#ifdef CONFIG_X86_64
50dotraplinkage void do_double_fault(struct pt_regs *, long);
51asmlinkage __kprobes struct pt_regs *sync_regs(struct pt_regs *);
52#endif
49dotraplinkage void do_general_protection(struct pt_regs *, long); 53dotraplinkage void do_general_protection(struct pt_regs *, long);
50dotraplinkage void do_page_fault(struct pt_regs *, unsigned long); 54dotraplinkage void do_page_fault(struct pt_regs *, unsigned long);
51dotraplinkage void do_spurious_interrupt_bug(struct pt_regs *, long); 55dotraplinkage void do_spurious_interrupt_bug(struct pt_regs *, long);
@@ -72,10 +76,13 @@ static inline int get_si_code(unsigned long condition)
72extern int panic_on_unrecovered_nmi; 76extern int panic_on_unrecovered_nmi;
73extern int kstack_depth_to_print; 77extern int kstack_depth_to_print;
74 78
75#ifdef CONFIG_X86_32
76void math_error(void __user *); 79void math_error(void __user *);
77unsigned long patch_espfix_desc(unsigned long, unsigned long);
78asmlinkage void math_emulate(long); 80asmlinkage void math_emulate(long);
81#ifdef CONFIG_X86_32
82unsigned long patch_espfix_desc(unsigned long, unsigned long);
83#else
84asmlinkage void smp_thermal_interrupt(void);
85asmlinkage void mce_threshold_interrupt(void);
79#endif 86#endif
80 87
81#endif /* _ASM_X86_TRAPS_H */ 88#endif /* _ASM_X86_TRAPS_H */
diff --git a/arch/x86/include/asm/tsc.h b/arch/x86/include/asm/tsc.h
index 9cd83a8e40d5..38ae163cc91b 100644
--- a/arch/x86/include/asm/tsc.h
+++ b/arch/x86/include/asm/tsc.h
@@ -34,8 +34,6 @@ static inline cycles_t get_cycles(void)
34 34
35static __always_inline cycles_t vget_cycles(void) 35static __always_inline cycles_t vget_cycles(void)
36{ 36{
37 cycles_t cycles;
38
39 /* 37 /*
40 * We only do VDSOs on TSC capable CPUs, so this shouldnt 38 * We only do VDSOs on TSC capable CPUs, so this shouldnt
41 * access boot_cpu_data (which is not VDSO-safe): 39 * access boot_cpu_data (which is not VDSO-safe):
@@ -44,11 +42,7 @@ static __always_inline cycles_t vget_cycles(void)
44 if (!cpu_has_tsc) 42 if (!cpu_has_tsc)
45 return 0; 43 return 0;
46#endif 44#endif
47 rdtsc_barrier(); 45 return (cycles_t)__native_read_tsc();
48 cycles = (cycles_t)__native_read_tsc();
49 rdtsc_barrier();
50
51 return cycles;
52} 46}
53 47
54extern void tsc_init(void); 48extern void tsc_init(void);
diff --git a/arch/x86/include/asm/uaccess.h b/arch/x86/include/asm/uaccess.h
index 35c54921b2e4..580c3ee6c58c 100644
--- a/arch/x86/include/asm/uaccess.h
+++ b/arch/x86/include/asm/uaccess.h
@@ -350,14 +350,14 @@ do { \
350 350
351#define __put_user_nocheck(x, ptr, size) \ 351#define __put_user_nocheck(x, ptr, size) \
352({ \ 352({ \
353 long __pu_err; \ 353 int __pu_err; \
354 __put_user_size((x), (ptr), (size), __pu_err, -EFAULT); \ 354 __put_user_size((x), (ptr), (size), __pu_err, -EFAULT); \
355 __pu_err; \ 355 __pu_err; \
356}) 356})
357 357
358#define __get_user_nocheck(x, ptr, size) \ 358#define __get_user_nocheck(x, ptr, size) \
359({ \ 359({ \
360 long __gu_err; \ 360 int __gu_err; \
361 unsigned long __gu_val; \ 361 unsigned long __gu_val; \
362 __get_user_size(__gu_val, (ptr), (size), __gu_err, -EFAULT); \ 362 __get_user_size(__gu_val, (ptr), (size), __gu_err, -EFAULT); \
363 (x) = (__force __typeof__(*(ptr)))__gu_val; \ 363 (x) = (__force __typeof__(*(ptr)))__gu_val; \
diff --git a/arch/x86/include/asm/uaccess_64.h b/arch/x86/include/asm/uaccess_64.h
index 664f15280f14..f8cfd00db450 100644
--- a/arch/x86/include/asm/uaccess_64.h
+++ b/arch/x86/include/asm/uaccess_64.h
@@ -46,7 +46,7 @@ int __copy_from_user(void *dst, const void __user *src, unsigned size)
46 return ret; 46 return ret;
47 case 10: 47 case 10:
48 __get_user_asm(*(u64 *)dst, (u64 __user *)src, 48 __get_user_asm(*(u64 *)dst, (u64 __user *)src,
49 ret, "q", "", "=r", 16); 49 ret, "q", "", "=r", 10);
50 if (unlikely(ret)) 50 if (unlikely(ret))
51 return ret; 51 return ret;
52 __get_user_asm(*(u16 *)(8 + (char *)dst), 52 __get_user_asm(*(u16 *)(8 + (char *)dst),
diff --git a/arch/x86/include/asm/unistd_64.h b/arch/x86/include/asm/unistd_64.h
index 834b2c1d89fb..d2e415e6666f 100644
--- a/arch/x86/include/asm/unistd_64.h
+++ b/arch/x86/include/asm/unistd_64.h
@@ -639,8 +639,8 @@ __SYSCALL(__NR_fallocate, sys_fallocate)
639__SYSCALL(__NR_timerfd_settime, sys_timerfd_settime) 639__SYSCALL(__NR_timerfd_settime, sys_timerfd_settime)
640#define __NR_timerfd_gettime 287 640#define __NR_timerfd_gettime 287
641__SYSCALL(__NR_timerfd_gettime, sys_timerfd_gettime) 641__SYSCALL(__NR_timerfd_gettime, sys_timerfd_gettime)
642#define __NR_paccept 288 642#define __NR_accept4 288
643__SYSCALL(__NR_paccept, sys_paccept) 643__SYSCALL(__NR_accept4, sys_accept4)
644#define __NR_signalfd4 289 644#define __NR_signalfd4 289
645__SYSCALL(__NR_signalfd4, sys_signalfd4) 645__SYSCALL(__NR_signalfd4, sys_signalfd4)
646#define __NR_eventfd2 290 646#define __NR_eventfd2 290
diff --git a/arch/x86/include/asm/uv/bios.h b/arch/x86/include/asm/uv/bios.h
index d931d3b7e6f7..7ed17ff502b9 100644
--- a/arch/x86/include/asm/uv/bios.h
+++ b/arch/x86/include/asm/uv/bios.h
@@ -32,13 +32,18 @@
32enum uv_bios_cmd { 32enum uv_bios_cmd {
33 UV_BIOS_COMMON, 33 UV_BIOS_COMMON,
34 UV_BIOS_GET_SN_INFO, 34 UV_BIOS_GET_SN_INFO,
35 UV_BIOS_FREQ_BASE 35 UV_BIOS_FREQ_BASE,
36 UV_BIOS_WATCHLIST_ALLOC,
37 UV_BIOS_WATCHLIST_FREE,
38 UV_BIOS_MEMPROTECT,
39 UV_BIOS_GET_PARTITION_ADDR
36}; 40};
37 41
38/* 42/*
39 * Status values returned from a BIOS call. 43 * Status values returned from a BIOS call.
40 */ 44 */
41enum { 45enum {
46 BIOS_STATUS_MORE_PASSES = 1,
42 BIOS_STATUS_SUCCESS = 0, 47 BIOS_STATUS_SUCCESS = 0,
43 BIOS_STATUS_UNIMPLEMENTED = -ENOSYS, 48 BIOS_STATUS_UNIMPLEMENTED = -ENOSYS,
44 BIOS_STATUS_EINVAL = -EINVAL, 49 BIOS_STATUS_EINVAL = -EINVAL,
@@ -71,6 +76,21 @@ union partition_info_u {
71 }; 76 };
72}; 77};
73 78
79union uv_watchlist_u {
80 u64 val;
81 struct {
82 u64 blade : 16,
83 size : 32,
84 filler : 16;
85 };
86};
87
88enum uv_memprotect {
89 UV_MEMPROT_RESTRICT_ACCESS,
90 UV_MEMPROT_ALLOW_AMO,
91 UV_MEMPROT_ALLOW_RW
92};
93
74/* 94/*
75 * bios calls have 6 parameters 95 * bios calls have 6 parameters
76 */ 96 */
@@ -80,14 +100,20 @@ extern s64 uv_bios_call_reentrant(enum uv_bios_cmd, u64, u64, u64, u64, u64);
80 100
81extern s64 uv_bios_get_sn_info(int, int *, long *, long *, long *); 101extern s64 uv_bios_get_sn_info(int, int *, long *, long *, long *);
82extern s64 uv_bios_freq_base(u64, u64 *); 102extern s64 uv_bios_freq_base(u64, u64 *);
103extern int uv_bios_mq_watchlist_alloc(int, unsigned long, unsigned int,
104 unsigned long *);
105extern int uv_bios_mq_watchlist_free(int, int);
106extern s64 uv_bios_change_memprotect(u64, u64, enum uv_memprotect);
107extern s64 uv_bios_reserved_page_pa(u64, u64 *, u64 *, u64 *);
83 108
84extern void uv_bios_init(void); 109extern void uv_bios_init(void);
85 110
111extern unsigned long sn_rtc_cycles_per_second;
86extern int uv_type; 112extern int uv_type;
87extern long sn_partition_id; 113extern long sn_partition_id;
88extern long uv_coherency_id; 114extern long sn_coherency_id;
89extern long uv_region_size; 115extern long sn_region_size;
90#define partition_coherence_id() (uv_coherency_id) 116#define partition_coherence_id() (sn_coherency_id)
91 117
92extern struct kobject *sgi_uv_kobj; /* /sys/firmware/sgi_uv */ 118extern struct kobject *sgi_uv_kobj; /* /sys/firmware/sgi_uv */
93 119
diff --git a/arch/x86/include/asm/uv/uv_hub.h b/arch/x86/include/asm/uv/uv_hub.h
index 7a5782610b2b..777327ef05c1 100644
--- a/arch/x86/include/asm/uv/uv_hub.h
+++ b/arch/x86/include/asm/uv/uv_hub.h
@@ -113,25 +113,37 @@
113 */ 113 */
114#define UV_MAX_NASID_VALUE (UV_MAX_NUMALINK_NODES * 2) 114#define UV_MAX_NASID_VALUE (UV_MAX_NUMALINK_NODES * 2)
115 115
116struct uv_scir_s {
117 struct timer_list timer;
118 unsigned long offset;
119 unsigned long last;
120 unsigned long idle_on;
121 unsigned long idle_off;
122 unsigned char state;
123 unsigned char enabled;
124};
125
116/* 126/*
117 * The following defines attributes of the HUB chip. These attributes are 127 * The following defines attributes of the HUB chip. These attributes are
118 * frequently referenced and are kept in the per-cpu data areas of each cpu. 128 * frequently referenced and are kept in the per-cpu data areas of each cpu.
119 * They are kept together in a struct to minimize cache misses. 129 * They are kept together in a struct to minimize cache misses.
120 */ 130 */
121struct uv_hub_info_s { 131struct uv_hub_info_s {
122 unsigned long global_mmr_base; 132 unsigned long global_mmr_base;
123 unsigned long gpa_mask; 133 unsigned long gpa_mask;
124 unsigned long gnode_upper; 134 unsigned long gnode_upper;
125 unsigned long lowmem_remap_top; 135 unsigned long lowmem_remap_top;
126 unsigned long lowmem_remap_base; 136 unsigned long lowmem_remap_base;
127 unsigned short pnode; 137 unsigned short pnode;
128 unsigned short pnode_mask; 138 unsigned short pnode_mask;
129 unsigned short coherency_domain_number; 139 unsigned short coherency_domain_number;
130 unsigned short numa_blade_id; 140 unsigned short numa_blade_id;
131 unsigned char blade_processor_id; 141 unsigned char blade_processor_id;
132 unsigned char m_val; 142 unsigned char m_val;
133 unsigned char n_val; 143 unsigned char n_val;
144 struct uv_scir_s scir;
134}; 145};
146
135DECLARE_PER_CPU(struct uv_hub_info_s, __uv_hub_info); 147DECLARE_PER_CPU(struct uv_hub_info_s, __uv_hub_info);
136#define uv_hub_info (&__get_cpu_var(__uv_hub_info)) 148#define uv_hub_info (&__get_cpu_var(__uv_hub_info))
137#define uv_cpu_hub_info(cpu) (&per_cpu(__uv_hub_info, cpu)) 149#define uv_cpu_hub_info(cpu) (&per_cpu(__uv_hub_info, cpu))
@@ -163,6 +175,30 @@ DECLARE_PER_CPU(struct uv_hub_info_s, __uv_hub_info);
163 175
164#define UV_APIC_PNODE_SHIFT 6 176#define UV_APIC_PNODE_SHIFT 6
165 177
178/* Local Bus from cpu's perspective */
179#define LOCAL_BUS_BASE 0x1c00000
180#define LOCAL_BUS_SIZE (4 * 1024 * 1024)
181
182/*
183 * System Controller Interface Reg
184 *
185 * Note there are NO leds on a UV system. This register is only
186 * used by the system controller to monitor system-wide operation.
187 * There are 64 regs per node. With Nahelem cpus (2 cores per node,
188 * 8 cpus per core, 2 threads per cpu) there are 32 cpu threads on
189 * a node.
190 *
191 * The window is located at top of ACPI MMR space
192 */
193#define SCIR_WINDOW_COUNT 64
194#define SCIR_LOCAL_MMR_BASE (LOCAL_BUS_BASE + \
195 LOCAL_BUS_SIZE - \
196 SCIR_WINDOW_COUNT)
197
198#define SCIR_CPU_HEARTBEAT 0x01 /* timer interrupt */
199#define SCIR_CPU_ACTIVITY 0x02 /* not idle */
200#define SCIR_CPU_HB_INTERVAL (HZ) /* once per second */
201
166/* 202/*
167 * Macros for converting between kernel virtual addresses, socket local physical 203 * Macros for converting between kernel virtual addresses, socket local physical
168 * addresses, and UV global physical addresses. 204 * addresses, and UV global physical addresses.
@@ -174,7 +210,7 @@ DECLARE_PER_CPU(struct uv_hub_info_s, __uv_hub_info);
174static inline unsigned long uv_soc_phys_ram_to_gpa(unsigned long paddr) 210static inline unsigned long uv_soc_phys_ram_to_gpa(unsigned long paddr)
175{ 211{
176 if (paddr < uv_hub_info->lowmem_remap_top) 212 if (paddr < uv_hub_info->lowmem_remap_top)
177 paddr += uv_hub_info->lowmem_remap_base; 213 paddr |= uv_hub_info->lowmem_remap_base;
178 return paddr | uv_hub_info->gnode_upper; 214 return paddr | uv_hub_info->gnode_upper;
179} 215}
180 216
@@ -182,19 +218,7 @@ static inline unsigned long uv_soc_phys_ram_to_gpa(unsigned long paddr)
182/* socket virtual --> UV global physical address */ 218/* socket virtual --> UV global physical address */
183static inline unsigned long uv_gpa(void *v) 219static inline unsigned long uv_gpa(void *v)
184{ 220{
185 return __pa(v) | uv_hub_info->gnode_upper; 221 return uv_soc_phys_ram_to_gpa(__pa(v));
186}
187
188/* socket virtual --> UV global physical address */
189static inline void *uv_vgpa(void *v)
190{
191 return (void *)uv_gpa(v);
192}
193
194/* UV global physical address --> socket virtual */
195static inline void *uv_va(unsigned long gpa)
196{
197 return __va(gpa & uv_hub_info->gpa_mask);
198} 222}
199 223
200/* pnode, offset --> socket virtual */ 224/* pnode, offset --> socket virtual */
@@ -277,6 +301,16 @@ static inline void uv_write_local_mmr(unsigned long offset, unsigned long val)
277 *uv_local_mmr_address(offset) = val; 301 *uv_local_mmr_address(offset) = val;
278} 302}
279 303
304static inline unsigned char uv_read_local_mmr8(unsigned long offset)
305{
306 return *((unsigned char *)uv_local_mmr_address(offset));
307}
308
309static inline void uv_write_local_mmr8(unsigned long offset, unsigned char val)
310{
311 *((unsigned char *)uv_local_mmr_address(offset)) = val;
312}
313
280/* 314/*
281 * Structures and definitions for converting between cpu, node, pnode, and blade 315 * Structures and definitions for converting between cpu, node, pnode, and blade
282 * numbers. 316 * numbers.
@@ -351,5 +385,20 @@ static inline int uv_num_possible_blades(void)
351 return uv_possible_blades; 385 return uv_possible_blades;
352} 386}
353 387
354#endif /* _ASM_X86_UV_UV_HUB_H */ 388/* Update SCIR state */
389static inline void uv_set_scir_bits(unsigned char value)
390{
391 if (uv_hub_info->scir.state != value) {
392 uv_hub_info->scir.state = value;
393 uv_write_local_mmr8(uv_hub_info->scir.offset, value);
394 }
395}
396static inline void uv_set_cpu_scir_bits(int cpu, unsigned char value)
397{
398 if (uv_cpu_hub_info(cpu)->scir.state != value) {
399 uv_cpu_hub_info(cpu)->scir.state = value;
400 uv_write_local_mmr8(uv_cpu_hub_info(cpu)->scir.offset, value);
401 }
402}
355 403
404#endif /* _ASM_X86_UV_UV_HUB_H */
diff --git a/arch/x86/include/asm/vmi.h b/arch/x86/include/asm/vmi.h
index b7c0dea119fe..61e08c0a2907 100644
--- a/arch/x86/include/asm/vmi.h
+++ b/arch/x86/include/asm/vmi.h
@@ -223,9 +223,15 @@ struct pci_header {
223} __attribute__((packed)); 223} __attribute__((packed));
224 224
225/* Function prototypes for bootstrapping */ 225/* Function prototypes for bootstrapping */
226#ifdef CONFIG_VMI
226extern void vmi_init(void); 227extern void vmi_init(void);
228extern void vmi_activate(void);
227extern void vmi_bringup(void); 229extern void vmi_bringup(void);
228extern void vmi_apply_boot_page_allocations(void); 230#else
231static inline void vmi_init(void) {}
232static inline void vmi_activate(void) {}
233static inline void vmi_bringup(void) {}
234#endif
229 235
230/* State needed to start an application processor in an SMP system. */ 236/* State needed to start an application processor in an SMP system. */
231struct vmi_ap_state { 237struct vmi_ap_state {
diff --git a/arch/x86/include/asm/vmware.h b/arch/x86/include/asm/vmware.h
new file mode 100644
index 000000000000..c11b7e100d83
--- /dev/null
+++ b/arch/x86/include/asm/vmware.h
@@ -0,0 +1,27 @@
1/*
2 * Copyright (C) 2008, VMware, Inc.
3 *
4 * This program is free software; you can redistribute it and/or modify
5 * it under the terms of the GNU General Public License as published by
6 * the Free Software Foundation; either version 2 of the License, or
7 * (at your option) any later version.
8 *
9 * This program is distributed in the hope that it will be useful, but
10 * WITHOUT ANY WARRANTY; without even the implied warranty of
11 * MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or
12 * NON INFRINGEMENT. See the GNU General Public License for more
13 * details.
14 *
15 * You should have received a copy of the GNU General Public License
16 * along with this program; if not, write to the Free Software
17 * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
18 *
19 */
20#ifndef ASM_X86__VMWARE_H
21#define ASM_X86__VMWARE_H
22
23extern unsigned long vmware_get_tsc_khz(void);
24extern int vmware_platform(void);
25extern void vmware_set_feature_bits(struct cpuinfo_x86 *c);
26
27#endif
diff --git a/arch/x86/include/asm/xen/hypercall.h b/arch/x86/include/asm/xen/hypercall.h
index 3f6000d95fe2..5e79ca694326 100644
--- a/arch/x86/include/asm/xen/hypercall.h
+++ b/arch/x86/include/asm/xen/hypercall.h
@@ -33,8 +33,14 @@
33#ifndef _ASM_X86_XEN_HYPERCALL_H 33#ifndef _ASM_X86_XEN_HYPERCALL_H
34#define _ASM_X86_XEN_HYPERCALL_H 34#define _ASM_X86_XEN_HYPERCALL_H
35 35
36#include <linux/kernel.h>
37#include <linux/spinlock.h>
36#include <linux/errno.h> 38#include <linux/errno.h>
37#include <linux/string.h> 39#include <linux/string.h>
40#include <linux/types.h>
41
42#include <asm/page.h>
43#include <asm/pgtable.h>
38 44
39#include <xen/interface/xen.h> 45#include <xen/interface/xen.h>
40#include <xen/interface/sched.h> 46#include <xen/interface/sched.h>
diff --git a/arch/x86/include/asm/xen/hypervisor.h b/arch/x86/include/asm/xen/hypervisor.h
index a38d25ac87d2..81fbd735aec4 100644
--- a/arch/x86/include/asm/xen/hypervisor.h
+++ b/arch/x86/include/asm/xen/hypervisor.h
@@ -33,39 +33,10 @@
33#ifndef _ASM_X86_XEN_HYPERVISOR_H 33#ifndef _ASM_X86_XEN_HYPERVISOR_H
34#define _ASM_X86_XEN_HYPERVISOR_H 34#define _ASM_X86_XEN_HYPERVISOR_H
35 35
36#include <linux/types.h>
37#include <linux/kernel.h>
38
39#include <xen/interface/xen.h>
40#include <xen/interface/version.h>
41
42#include <asm/ptrace.h>
43#include <asm/page.h>
44#include <asm/desc.h>
45#if defined(__i386__)
46# ifdef CONFIG_X86_PAE
47# include <asm-generic/pgtable-nopud.h>
48# else
49# include <asm-generic/pgtable-nopmd.h>
50# endif
51#endif
52#include <asm/xen/hypercall.h>
53
54/* arch/i386/kernel/setup.c */ 36/* arch/i386/kernel/setup.c */
55extern struct shared_info *HYPERVISOR_shared_info; 37extern struct shared_info *HYPERVISOR_shared_info;
56extern struct start_info *xen_start_info; 38extern struct start_info *xen_start_info;
57 39
58/* arch/i386/mach-xen/evtchn.c */
59/* Force a proper event-channel callback from Xen. */
60extern void force_evtchn_callback(void);
61
62/* Turn jiffies into Xen system time. */
63u64 jiffies_to_st(unsigned long jiffies);
64
65
66#define MULTI_UVMFLAGS_INDEX 3
67#define MULTI_UVMDOMID_INDEX 4
68
69enum xen_domain_type { 40enum xen_domain_type {
70 XEN_NATIVE, 41 XEN_NATIVE,
71 XEN_PV_DOMAIN, 42 XEN_PV_DOMAIN,
@@ -74,9 +45,15 @@ enum xen_domain_type {
74 45
75extern enum xen_domain_type xen_domain_type; 46extern enum xen_domain_type xen_domain_type;
76 47
48#ifdef CONFIG_XEN
77#define xen_domain() (xen_domain_type != XEN_NATIVE) 49#define xen_domain() (xen_domain_type != XEN_NATIVE)
78#define xen_pv_domain() (xen_domain_type == XEN_PV_DOMAIN) 50#else
51#define xen_domain() (0)
52#endif
53
54#define xen_pv_domain() (xen_domain() && xen_domain_type == XEN_PV_DOMAIN)
55#define xen_hvm_domain() (xen_domain() && xen_domain_type == XEN_HVM_DOMAIN)
56
79#define xen_initial_domain() (xen_pv_domain() && xen_start_info->flags & SIF_INITDOMAIN) 57#define xen_initial_domain() (xen_pv_domain() && xen_start_info->flags & SIF_INITDOMAIN)
80#define xen_hvm_domain() (xen_domain_type == XEN_HVM_DOMAIN)
81 58
82#endif /* _ASM_X86_XEN_HYPERVISOR_H */ 59#endif /* _ASM_X86_XEN_HYPERVISOR_H */
diff --git a/arch/x86/include/asm/xen/page.h b/arch/x86/include/asm/xen/page.h
index bc628998a1b9..7ef617ef1df3 100644
--- a/arch/x86/include/asm/xen/page.h
+++ b/arch/x86/include/asm/xen/page.h
@@ -1,11 +1,16 @@
1#ifndef _ASM_X86_XEN_PAGE_H 1#ifndef _ASM_X86_XEN_PAGE_H
2#define _ASM_X86_XEN_PAGE_H 2#define _ASM_X86_XEN_PAGE_H
3 3
4#include <linux/kernel.h>
5#include <linux/types.h>
6#include <linux/spinlock.h>
4#include <linux/pfn.h> 7#include <linux/pfn.h>
5 8
6#include <asm/uaccess.h> 9#include <asm/uaccess.h>
10#include <asm/page.h>
7#include <asm/pgtable.h> 11#include <asm/pgtable.h>
8 12
13#include <xen/interface/xen.h>
9#include <xen/features.h> 14#include <xen/features.h>
10 15
11/* Xen machine address */ 16/* Xen machine address */
diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile
index e489ff9cb3e2..1f208aaee780 100644
--- a/arch/x86/kernel/Makefile
+++ b/arch/x86/kernel/Makefile
@@ -12,6 +12,7 @@ CFLAGS_REMOVE_tsc.o = -pg
12CFLAGS_REMOVE_rtc.o = -pg 12CFLAGS_REMOVE_rtc.o = -pg
13CFLAGS_REMOVE_paravirt-spinlocks.o = -pg 13CFLAGS_REMOVE_paravirt-spinlocks.o = -pg
14CFLAGS_REMOVE_ftrace.o = -pg 14CFLAGS_REMOVE_ftrace.o = -pg
15CFLAGS_REMOVE_early_printk.o = -pg
15endif 16endif
16 17
17# 18#
@@ -23,9 +24,9 @@ CFLAGS_vsyscall_64.o := $(PROFILING) -g0 $(nostackp)
23CFLAGS_hpet.o := $(nostackp) 24CFLAGS_hpet.o := $(nostackp)
24CFLAGS_tsc.o := $(nostackp) 25CFLAGS_tsc.o := $(nostackp)
25 26
26obj-y := process_$(BITS).o signal_$(BITS).o entry_$(BITS).o 27obj-y := process_$(BITS).o signal.o entry_$(BITS).o
27obj-y += traps.o irq.o irq_$(BITS).o dumpstack_$(BITS).o 28obj-y += traps.o irq.o irq_$(BITS).o dumpstack_$(BITS).o
28obj-y += time_$(BITS).o ioport.o ldt.o 29obj-y += time_$(BITS).o ioport.o ldt.o dumpstack.o
29obj-y += setup.o i8259.o irqinit_$(BITS).o setup_percpu.o 30obj-y += setup.o i8259.o irqinit_$(BITS).o setup_percpu.o
30obj-$(CONFIG_X86_VISWS) += visws_quirks.o 31obj-$(CONFIG_X86_VISWS) += visws_quirks.o
31obj-$(CONFIG_X86_32) += probe_roms_32.o 32obj-$(CONFIG_X86_32) += probe_roms_32.o
@@ -41,7 +42,7 @@ obj-$(CONFIG_X86_TRAMPOLINE) += trampoline.o
41obj-y += process.o 42obj-y += process.o
42obj-y += i387.o xsave.o 43obj-y += i387.o xsave.o
43obj-y += ptrace.o 44obj-y += ptrace.o
44obj-y += ds.o 45obj-$(CONFIG_X86_DS) += ds.o
45obj-$(CONFIG_X86_32) += tls.o 46obj-$(CONFIG_X86_32) += tls.o
46obj-$(CONFIG_IA32_EMULATION) += tls.o 47obj-$(CONFIG_IA32_EMULATION) += tls.o
47obj-y += step.o 48obj-y += step.o
@@ -105,6 +106,8 @@ microcode-$(CONFIG_MICROCODE_INTEL) += microcode_intel.o
105microcode-$(CONFIG_MICROCODE_AMD) += microcode_amd.o 106microcode-$(CONFIG_MICROCODE_AMD) += microcode_amd.o
106obj-$(CONFIG_MICROCODE) += microcode.o 107obj-$(CONFIG_MICROCODE) += microcode.o
107 108
109obj-$(CONFIG_X86_CHECK_BIOS_CORRUPTION) += check.o
110
108### 111###
109# 64 bit specific files 112# 64 bit specific files
110ifeq ($(CONFIG_X86_64),y) 113ifeq ($(CONFIG_X86_64),y)
diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c
index 8c1f76abae9e..65d0b72777ea 100644
--- a/arch/x86/kernel/acpi/boot.c
+++ b/arch/x86/kernel/acpi/boot.c
@@ -1343,7 +1343,6 @@ static void __init acpi_process_madt(void)
1343 error = acpi_parse_madt_ioapic_entries(); 1343 error = acpi_parse_madt_ioapic_entries();
1344 if (!error) { 1344 if (!error) {
1345 acpi_irq_model = ACPI_IRQ_MODEL_IOAPIC; 1345 acpi_irq_model = ACPI_IRQ_MODEL_IOAPIC;
1346 acpi_irq_balance_set(NULL);
1347 acpi_ioapic = 1; 1346 acpi_ioapic = 1;
1348 1347
1349 smp_found_config = 1; 1348 smp_found_config = 1;
@@ -1361,6 +1360,17 @@ static void __init acpi_process_madt(void)
1361 disable_acpi(); 1360 disable_acpi();
1362 } 1361 }
1363 } 1362 }
1363
1364 /*
1365 * ACPI supports both logical (e.g. Hyper-Threading) and physical
1366 * processors, where MPS only supports physical.
1367 */
1368 if (acpi_lapic && acpi_ioapic)
1369 printk(KERN_INFO "Using ACPI (MADT) for SMP configuration "
1370 "information\n");
1371 else if (acpi_lapic)
1372 printk(KERN_INFO "Using ACPI for processor (LAPIC) "
1373 "configuration information\n");
1364#endif 1374#endif
1365 return; 1375 return;
1366} 1376}
diff --git a/arch/x86/kernel/amd_iommu.c b/arch/x86/kernel/amd_iommu.c
index 331b318304eb..0a60d60ed036 100644
--- a/arch/x86/kernel/amd_iommu.c
+++ b/arch/x86/kernel/amd_iommu.c
@@ -187,6 +187,8 @@ static int iommu_queue_command(struct amd_iommu *iommu, struct iommu_cmd *cmd)
187 187
188 spin_lock_irqsave(&iommu->lock, flags); 188 spin_lock_irqsave(&iommu->lock, flags);
189 ret = __iommu_queue_command(iommu, cmd); 189 ret = __iommu_queue_command(iommu, cmd);
190 if (!ret)
191 iommu->need_sync = 1;
190 spin_unlock_irqrestore(&iommu->lock, flags); 192 spin_unlock_irqrestore(&iommu->lock, flags);
191 193
192 return ret; 194 return ret;
@@ -210,10 +212,13 @@ static int iommu_completion_wait(struct amd_iommu *iommu)
210 cmd.data[0] = CMD_COMPL_WAIT_INT_MASK; 212 cmd.data[0] = CMD_COMPL_WAIT_INT_MASK;
211 CMD_SET_TYPE(&cmd, CMD_COMPL_WAIT); 213 CMD_SET_TYPE(&cmd, CMD_COMPL_WAIT);
212 214
213 iommu->need_sync = 0;
214
215 spin_lock_irqsave(&iommu->lock, flags); 215 spin_lock_irqsave(&iommu->lock, flags);
216 216
217 if (!iommu->need_sync)
218 goto out;
219
220 iommu->need_sync = 0;
221
217 ret = __iommu_queue_command(iommu, &cmd); 222 ret = __iommu_queue_command(iommu, &cmd);
218 223
219 if (ret) 224 if (ret)
@@ -230,8 +235,9 @@ static int iommu_completion_wait(struct amd_iommu *iommu)
230 status &= ~MMIO_STATUS_COM_WAIT_INT_MASK; 235 status &= ~MMIO_STATUS_COM_WAIT_INT_MASK;
231 writel(status, iommu->mmio_base + MMIO_STATUS_OFFSET); 236 writel(status, iommu->mmio_base + MMIO_STATUS_OFFSET);
232 237
233 if (unlikely((i == EXIT_LOOP_COUNT) && printk_ratelimit())) 238 if (unlikely(i == EXIT_LOOP_COUNT))
234 printk(KERN_WARNING "AMD IOMMU: Completion wait loop failed\n"); 239 panic("AMD IOMMU: Completion wait loop failed\n");
240
235out: 241out:
236 spin_unlock_irqrestore(&iommu->lock, flags); 242 spin_unlock_irqrestore(&iommu->lock, flags);
237 243
@@ -254,8 +260,6 @@ static int iommu_queue_inv_dev_entry(struct amd_iommu *iommu, u16 devid)
254 260
255 ret = iommu_queue_command(iommu, &cmd); 261 ret = iommu_queue_command(iommu, &cmd);
256 262
257 iommu->need_sync = 1;
258
259 return ret; 263 return ret;
260} 264}
261 265
@@ -281,8 +285,6 @@ static int iommu_queue_inv_iommu_pages(struct amd_iommu *iommu,
281 285
282 ret = iommu_queue_command(iommu, &cmd); 286 ret = iommu_queue_command(iommu, &cmd);
283 287
284 iommu->need_sync = 1;
285
286 return ret; 288 return ret;
287} 289}
288 290
@@ -343,7 +345,7 @@ static int iommu_map(struct protection_domain *dom,
343 u64 __pte, *pte, *page; 345 u64 __pte, *pte, *page;
344 346
345 bus_addr = PAGE_ALIGN(bus_addr); 347 bus_addr = PAGE_ALIGN(bus_addr);
346 phys_addr = PAGE_ALIGN(bus_addr); 348 phys_addr = PAGE_ALIGN(phys_addr);
347 349
348 /* only support 512GB address spaces for now */ 350 /* only support 512GB address spaces for now */
349 if (bus_addr > IOMMU_MAP_SIZE_L3 || !(prot & IOMMU_PROT_MASK)) 351 if (bus_addr > IOMMU_MAP_SIZE_L3 || !(prot & IOMMU_PROT_MASK))
@@ -537,7 +539,7 @@ static void dma_ops_free_addresses(struct dma_ops_domain *dom,
537 address >>= PAGE_SHIFT; 539 address >>= PAGE_SHIFT;
538 iommu_area_free(dom->bitmap, address, pages); 540 iommu_area_free(dom->bitmap, address, pages);
539 541
540 if (address + pages >= dom->next_bit) 542 if (address >= dom->next_bit)
541 dom->need_flush = true; 543 dom->need_flush = true;
542} 544}
543 545
@@ -599,7 +601,7 @@ static void dma_ops_free_pagetable(struct dma_ops_domain *dma_dom)
599 continue; 601 continue;
600 602
601 p2 = IOMMU_PTE_PAGE(p1[i]); 603 p2 = IOMMU_PTE_PAGE(p1[i]);
602 for (j = 0; j < 512; ++i) { 604 for (j = 0; j < 512; ++j) {
603 if (!IOMMU_PTE_PRESENT(p2[j])) 605 if (!IOMMU_PTE_PRESENT(p2[j]))
604 continue; 606 continue;
605 p3 = IOMMU_PTE_PAGE(p2[j]); 607 p3 = IOMMU_PTE_PAGE(p2[j]);
@@ -762,8 +764,6 @@ static void set_device_domain(struct amd_iommu *iommu,
762 write_unlock_irqrestore(&amd_iommu_devtable_lock, flags); 764 write_unlock_irqrestore(&amd_iommu_devtable_lock, flags);
763 765
764 iommu_queue_inv_dev_entry(iommu, devid); 766 iommu_queue_inv_dev_entry(iommu, devid);
765
766 iommu->need_sync = 1;
767} 767}
768 768
769/***************************************************************************** 769/*****************************************************************************
@@ -858,6 +858,9 @@ static int get_device_resources(struct device *dev,
858 print_devid(_bdf, 1); 858 print_devid(_bdf, 1);
859 } 859 }
860 860
861 if (domain_for_device(_bdf) == NULL)
862 set_device_domain(*iommu, *domain, _bdf);
863
861 return 1; 864 return 1;
862} 865}
863 866
@@ -908,7 +911,7 @@ static void dma_ops_domain_unmap(struct amd_iommu *iommu,
908 if (address >= dom->aperture_size) 911 if (address >= dom->aperture_size)
909 return; 912 return;
910 913
911 WARN_ON(address & 0xfffULL || address > dom->aperture_size); 914 WARN_ON(address & ~PAGE_MASK || address >= dom->aperture_size);
912 915
913 pte = dom->pte_pages[IOMMU_PTE_L1_INDEX(address)]; 916 pte = dom->pte_pages[IOMMU_PTE_L1_INDEX(address)];
914 pte += IOMMU_PTE_L0_INDEX(address); 917 pte += IOMMU_PTE_L0_INDEX(address);
@@ -920,8 +923,8 @@ static void dma_ops_domain_unmap(struct amd_iommu *iommu,
920 923
921/* 924/*
922 * This function contains common code for mapping of a physically 925 * This function contains common code for mapping of a physically
923 * contiguous memory region into DMA address space. It is uses by all 926 * contiguous memory region into DMA address space. It is used by all
924 * mapping functions provided by this IOMMU driver. 927 * mapping functions provided with this IOMMU driver.
925 * Must be called with the domain lock held. 928 * Must be called with the domain lock held.
926 */ 929 */
927static dma_addr_t __map_single(struct device *dev, 930static dma_addr_t __map_single(struct device *dev,
@@ -981,7 +984,8 @@ static void __unmap_single(struct amd_iommu *iommu,
981 dma_addr_t i, start; 984 dma_addr_t i, start;
982 unsigned int pages; 985 unsigned int pages;
983 986
984 if ((dma_addr == 0) || (dma_addr + size > dma_dom->aperture_size)) 987 if ((dma_addr == bad_dma_address) ||
988 (dma_addr + size > dma_dom->aperture_size))
985 return; 989 return;
986 990
987 pages = iommu_num_pages(dma_addr, size, PAGE_SIZE); 991 pages = iommu_num_pages(dma_addr, size, PAGE_SIZE);
@@ -1031,8 +1035,7 @@ static dma_addr_t map_single(struct device *dev, phys_addr_t paddr,
1031 if (addr == bad_dma_address) 1035 if (addr == bad_dma_address)
1032 goto out; 1036 goto out;
1033 1037
1034 if (unlikely(iommu->need_sync)) 1038 iommu_completion_wait(iommu);
1035 iommu_completion_wait(iommu);
1036 1039
1037out: 1040out:
1038 spin_unlock_irqrestore(&domain->lock, flags); 1041 spin_unlock_irqrestore(&domain->lock, flags);
@@ -1060,8 +1063,7 @@ static void unmap_single(struct device *dev, dma_addr_t dma_addr,
1060 1063
1061 __unmap_single(iommu, domain->priv, dma_addr, size, dir); 1064 __unmap_single(iommu, domain->priv, dma_addr, size, dir);
1062 1065
1063 if (unlikely(iommu->need_sync)) 1066 iommu_completion_wait(iommu);
1064 iommu_completion_wait(iommu);
1065 1067
1066 spin_unlock_irqrestore(&domain->lock, flags); 1068 spin_unlock_irqrestore(&domain->lock, flags);
1067} 1069}
@@ -1127,8 +1129,7 @@ static int map_sg(struct device *dev, struct scatterlist *sglist,
1127 goto unmap; 1129 goto unmap;
1128 } 1130 }
1129 1131
1130 if (unlikely(iommu->need_sync)) 1132 iommu_completion_wait(iommu);
1131 iommu_completion_wait(iommu);
1132 1133
1133out: 1134out:
1134 spin_unlock_irqrestore(&domain->lock, flags); 1135 spin_unlock_irqrestore(&domain->lock, flags);
@@ -1173,8 +1174,7 @@ static void unmap_sg(struct device *dev, struct scatterlist *sglist,
1173 s->dma_address = s->dma_length = 0; 1174 s->dma_address = s->dma_length = 0;
1174 } 1175 }
1175 1176
1176 if (unlikely(iommu->need_sync)) 1177 iommu_completion_wait(iommu);
1177 iommu_completion_wait(iommu);
1178 1178
1179 spin_unlock_irqrestore(&domain->lock, flags); 1179 spin_unlock_irqrestore(&domain->lock, flags);
1180} 1180}
@@ -1225,8 +1225,7 @@ static void *alloc_coherent(struct device *dev, size_t size,
1225 goto out; 1225 goto out;
1226 } 1226 }
1227 1227
1228 if (unlikely(iommu->need_sync)) 1228 iommu_completion_wait(iommu);
1229 iommu_completion_wait(iommu);
1230 1229
1231out: 1230out:
1232 spin_unlock_irqrestore(&domain->lock, flags); 1231 spin_unlock_irqrestore(&domain->lock, flags);
@@ -1257,8 +1256,7 @@ static void free_coherent(struct device *dev, size_t size,
1257 1256
1258 __unmap_single(iommu, domain->priv, dma_addr, size, DMA_BIDIRECTIONAL); 1257 __unmap_single(iommu, domain->priv, dma_addr, size, DMA_BIDIRECTIONAL);
1259 1258
1260 if (unlikely(iommu->need_sync)) 1259 iommu_completion_wait(iommu);
1261 iommu_completion_wait(iommu);
1262 1260
1263 spin_unlock_irqrestore(&domain->lock, flags); 1261 spin_unlock_irqrestore(&domain->lock, flags);
1264 1262
diff --git a/arch/x86/kernel/amd_iommu_init.c b/arch/x86/kernel/amd_iommu_init.c
index 0cdcda35a05f..c6cc22815d35 100644
--- a/arch/x86/kernel/amd_iommu_init.c
+++ b/arch/x86/kernel/amd_iommu_init.c
@@ -121,7 +121,7 @@ u16 amd_iommu_last_bdf; /* largest PCI device id we have
121LIST_HEAD(amd_iommu_unity_map); /* a list of required unity mappings 121LIST_HEAD(amd_iommu_unity_map); /* a list of required unity mappings
122 we find in ACPI */ 122 we find in ACPI */
123unsigned amd_iommu_aperture_order = 26; /* size of aperture in power of 2 */ 123unsigned amd_iommu_aperture_order = 26; /* size of aperture in power of 2 */
124int amd_iommu_isolate; /* if 1, device isolation is enabled */ 124int amd_iommu_isolate = 1; /* if 1, device isolation is enabled */
125bool amd_iommu_unmap_flush; /* if true, flush on every unmap */ 125bool amd_iommu_unmap_flush; /* if true, flush on every unmap */
126 126
127LIST_HEAD(amd_iommu_list); /* list of all AMD IOMMUs in the 127LIST_HEAD(amd_iommu_list); /* list of all AMD IOMMUs in the
@@ -427,6 +427,10 @@ static u8 * __init alloc_command_buffer(struct amd_iommu *iommu)
427 memcpy_toio(iommu->mmio_base + MMIO_CMD_BUF_OFFSET, 427 memcpy_toio(iommu->mmio_base + MMIO_CMD_BUF_OFFSET,
428 &entry, sizeof(entry)); 428 &entry, sizeof(entry));
429 429
430 /* set head and tail to zero manually */
431 writel(0x00, iommu->mmio_base + MMIO_CMD_HEAD_OFFSET);
432 writel(0x00, iommu->mmio_base + MMIO_CMD_TAIL_OFFSET);
433
430 iommu_feature_enable(iommu, CONTROL_CMDBUF_EN); 434 iommu_feature_enable(iommu, CONTROL_CMDBUF_EN);
431 435
432 return cmd_buf; 436 return cmd_buf;
@@ -1074,7 +1078,8 @@ int __init amd_iommu_init(void)
1074 goto free; 1078 goto free;
1075 1079
1076 /* IOMMU rlookup table - find the IOMMU for a specific device */ 1080 /* IOMMU rlookup table - find the IOMMU for a specific device */
1077 amd_iommu_rlookup_table = (void *)__get_free_pages(GFP_KERNEL, 1081 amd_iommu_rlookup_table = (void *)__get_free_pages(
1082 GFP_KERNEL | __GFP_ZERO,
1078 get_order(rlookup_table_size)); 1083 get_order(rlookup_table_size));
1079 if (amd_iommu_rlookup_table == NULL) 1084 if (amd_iommu_rlookup_table == NULL)
1080 goto free; 1085 goto free;
@@ -1213,7 +1218,9 @@ static int __init parse_amd_iommu_options(char *str)
1213 for (; *str; ++str) { 1218 for (; *str; ++str) {
1214 if (strncmp(str, "isolate", 7) == 0) 1219 if (strncmp(str, "isolate", 7) == 0)
1215 amd_iommu_isolate = 1; 1220 amd_iommu_isolate = 1;
1216 if (strncmp(str, "fullflush", 11) == 0) 1221 if (strncmp(str, "share", 5) == 0)
1222 amd_iommu_isolate = 0;
1223 if (strncmp(str, "fullflush", 9) == 0)
1217 amd_iommu_unmap_flush = true; 1224 amd_iommu_unmap_flush = true;
1218 } 1225 }
1219 1226
diff --git a/arch/x86/kernel/apic.c b/arch/x86/kernel/apic.c
index 04a7f960bbc0..20c6e12c0475 100644
--- a/arch/x86/kernel/apic.c
+++ b/arch/x86/kernel/apic.c
@@ -441,6 +441,7 @@ static void lapic_timer_setup(enum clock_event_mode mode,
441 v = apic_read(APIC_LVTT); 441 v = apic_read(APIC_LVTT);
442 v |= (APIC_LVT_MASKED | LOCAL_TIMER_VECTOR); 442 v |= (APIC_LVT_MASKED | LOCAL_TIMER_VECTOR);
443 apic_write(APIC_LVTT, v); 443 apic_write(APIC_LVTT, v);
444 apic_write(APIC_TMICT, 0xffffffff);
444 break; 445 break;
445 case CLOCK_EVT_MODE_RESUME: 446 case CLOCK_EVT_MODE_RESUME:
446 /* Nothing to do here */ 447 /* Nothing to do here */
@@ -559,13 +560,13 @@ static int __init calibrate_by_pmtimer(long deltapm, long *delta)
559 } else { 560 } else {
560 res = (((u64)deltapm) * mult) >> 22; 561 res = (((u64)deltapm) * mult) >> 22;
561 do_div(res, 1000000); 562 do_div(res, 1000000);
562 printk(KERN_WARNING "APIC calibration not consistent " 563 pr_warning("APIC calibration not consistent "
563 "with PM Timer: %ldms instead of 100ms\n", 564 "with PM Timer: %ldms instead of 100ms\n",
564 (long)res); 565 (long)res);
565 /* Correct the lapic counter value */ 566 /* Correct the lapic counter value */
566 res = (((u64)(*delta)) * pm_100ms); 567 res = (((u64)(*delta)) * pm_100ms);
567 do_div(res, deltapm); 568 do_div(res, deltapm);
568 printk(KERN_INFO "APIC delta adjusted to PM-Timer: " 569 pr_info("APIC delta adjusted to PM-Timer: "
569 "%lu (%ld)\n", (unsigned long)res, *delta); 570 "%lu (%ld)\n", (unsigned long)res, *delta);
570 *delta = (long)res; 571 *delta = (long)res;
571 } 572 }
@@ -645,8 +646,7 @@ static int __init calibrate_APIC_clock(void)
645 */ 646 */
646 if (calibration_result < (1000000 / HZ)) { 647 if (calibration_result < (1000000 / HZ)) {
647 local_irq_enable(); 648 local_irq_enable();
648 printk(KERN_WARNING 649 pr_warning("APIC frequency too slow, disabling apic timer\n");
649 "APIC frequency too slow, disabling apic timer\n");
650 return -1; 650 return -1;
651 } 651 }
652 652
@@ -672,13 +672,9 @@ static int __init calibrate_APIC_clock(void)
672 while (lapic_cal_loops <= LAPIC_CAL_LOOPS) 672 while (lapic_cal_loops <= LAPIC_CAL_LOOPS)
673 cpu_relax(); 673 cpu_relax();
674 674
675 local_irq_disable();
676
677 /* Stop the lapic timer */ 675 /* Stop the lapic timer */
678 lapic_timer_setup(CLOCK_EVT_MODE_SHUTDOWN, levt); 676 lapic_timer_setup(CLOCK_EVT_MODE_SHUTDOWN, levt);
679 677
680 local_irq_enable();
681
682 /* Jiffies delta */ 678 /* Jiffies delta */
683 deltaj = lapic_cal_j2 - lapic_cal_j1; 679 deltaj = lapic_cal_j2 - lapic_cal_j1;
684 apic_printk(APIC_VERBOSE, "... jiffies delta = %lu\n", deltaj); 680 apic_printk(APIC_VERBOSE, "... jiffies delta = %lu\n", deltaj);
@@ -692,8 +688,7 @@ static int __init calibrate_APIC_clock(void)
692 local_irq_enable(); 688 local_irq_enable();
693 689
694 if (levt->features & CLOCK_EVT_FEAT_DUMMY) { 690 if (levt->features & CLOCK_EVT_FEAT_DUMMY) {
695 printk(KERN_WARNING 691 pr_warning("APIC timer disabled due to verification failure.\n");
696 "APIC timer disabled due to verification failure.\n");
697 return -1; 692 return -1;
698 } 693 }
699 694
@@ -714,7 +709,7 @@ void __init setup_boot_APIC_clock(void)
714 * broadcast mechanism is used. On UP systems simply ignore it. 709 * broadcast mechanism is used. On UP systems simply ignore it.
715 */ 710 */
716 if (disable_apic_timer) { 711 if (disable_apic_timer) {
717 printk(KERN_INFO "Disabling APIC timer\n"); 712 pr_info("Disabling APIC timer\n");
718 /* No broadcast on UP ! */ 713 /* No broadcast on UP ! */
719 if (num_possible_cpus() > 1) { 714 if (num_possible_cpus() > 1) {
720 lapic_clockevent.mult = 1; 715 lapic_clockevent.mult = 1;
@@ -741,7 +736,7 @@ void __init setup_boot_APIC_clock(void)
741 if (nmi_watchdog != NMI_IO_APIC) 736 if (nmi_watchdog != NMI_IO_APIC)
742 lapic_clockevent.features &= ~CLOCK_EVT_FEAT_DUMMY; 737 lapic_clockevent.features &= ~CLOCK_EVT_FEAT_DUMMY;
743 else 738 else
744 printk(KERN_WARNING "APIC timer registered as dummy," 739 pr_warning("APIC timer registered as dummy,"
745 " due to nmi_watchdog=%d!\n", nmi_watchdog); 740 " due to nmi_watchdog=%d!\n", nmi_watchdog);
746 741
747 /* Setup the lapic or request the broadcast */ 742 /* Setup the lapic or request the broadcast */
@@ -773,8 +768,7 @@ static void local_apic_timer_interrupt(void)
773 * spurious. 768 * spurious.
774 */ 769 */
775 if (!evt->event_handler) { 770 if (!evt->event_handler) {
776 printk(KERN_WARNING 771 pr_warning("Spurious LAPIC timer interrupt on cpu %d\n", cpu);
777 "Spurious LAPIC timer interrupt on cpu %d\n", cpu);
778 /* Switch it off */ 772 /* Switch it off */
779 lapic_timer_setup(CLOCK_EVT_MODE_SHUTDOWN, evt); 773 lapic_timer_setup(CLOCK_EVT_MODE_SHUTDOWN, evt);
780 return; 774 return;
@@ -814,9 +808,7 @@ void smp_apic_timer_interrupt(struct pt_regs *regs)
814 * Besides, if we don't timer interrupts ignore the global 808 * Besides, if we don't timer interrupts ignore the global
815 * interrupt lock, which is the WrongThing (tm) to do. 809 * interrupt lock, which is the WrongThing (tm) to do.
816 */ 810 */
817#ifdef CONFIG_X86_64
818 exit_idle(); 811 exit_idle();
819#endif
820 irq_enter(); 812 irq_enter();
821 local_apic_timer_interrupt(); 813 local_apic_timer_interrupt();
822 irq_exit(); 814 irq_exit();
@@ -1093,7 +1085,7 @@ static void __cpuinit lapic_setup_esr(void)
1093 unsigned int oldvalue, value, maxlvt; 1085 unsigned int oldvalue, value, maxlvt;
1094 1086
1095 if (!lapic_is_integrated()) { 1087 if (!lapic_is_integrated()) {
1096 printk(KERN_INFO "No ESR for 82489DX.\n"); 1088 pr_info("No ESR for 82489DX.\n");
1097 return; 1089 return;
1098 } 1090 }
1099 1091
@@ -1104,7 +1096,7 @@ static void __cpuinit lapic_setup_esr(void)
1104 * ESR disabled - we can't do anything useful with the 1096 * ESR disabled - we can't do anything useful with the
1105 * errors anyway - mbligh 1097 * errors anyway - mbligh
1106 */ 1098 */
1107 printk(KERN_INFO "Leaving ESR disabled.\n"); 1099 pr_info("Leaving ESR disabled.\n");
1108 return; 1100 return;
1109 } 1101 }
1110 1102
@@ -1298,7 +1290,7 @@ void check_x2apic(void)
1298 rdmsr(MSR_IA32_APICBASE, msr, msr2); 1290 rdmsr(MSR_IA32_APICBASE, msr, msr2);
1299 1291
1300 if (msr & X2APIC_ENABLE) { 1292 if (msr & X2APIC_ENABLE) {
1301 printk("x2apic enabled by BIOS, switching to x2apic ops\n"); 1293 pr_info("x2apic enabled by BIOS, switching to x2apic ops\n");
1302 x2apic_preenabled = x2apic = 1; 1294 x2apic_preenabled = x2apic = 1;
1303 apic_ops = &x2apic_ops; 1295 apic_ops = &x2apic_ops;
1304 } 1296 }
@@ -1310,12 +1302,12 @@ void enable_x2apic(void)
1310 1302
1311 rdmsr(MSR_IA32_APICBASE, msr, msr2); 1303 rdmsr(MSR_IA32_APICBASE, msr, msr2);
1312 if (!(msr & X2APIC_ENABLE)) { 1304 if (!(msr & X2APIC_ENABLE)) {
1313 printk("Enabling x2apic\n"); 1305 pr_info("Enabling x2apic\n");
1314 wrmsr(MSR_IA32_APICBASE, msr | X2APIC_ENABLE, 0); 1306 wrmsr(MSR_IA32_APICBASE, msr | X2APIC_ENABLE, 0);
1315 } 1307 }
1316} 1308}
1317 1309
1318void enable_IR_x2apic(void) 1310void __init enable_IR_x2apic(void)
1319{ 1311{
1320#ifdef CONFIG_INTR_REMAP 1312#ifdef CONFIG_INTR_REMAP
1321 int ret; 1313 int ret;
@@ -1325,9 +1317,8 @@ void enable_IR_x2apic(void)
1325 return; 1317 return;
1326 1318
1327 if (!x2apic_preenabled && disable_x2apic) { 1319 if (!x2apic_preenabled && disable_x2apic) {
1328 printk(KERN_INFO 1320 pr_info("Skipped enabling x2apic and Interrupt-remapping "
1329 "Skipped enabling x2apic and Interrupt-remapping " 1321 "because of nox2apic\n");
1330 "because of nox2apic\n");
1331 return; 1322 return;
1332 } 1323 }
1333 1324
@@ -1335,22 +1326,19 @@ void enable_IR_x2apic(void)
1335 panic("Bios already enabled x2apic, can't enforce nox2apic"); 1326 panic("Bios already enabled x2apic, can't enforce nox2apic");
1336 1327
1337 if (!x2apic_preenabled && skip_ioapic_setup) { 1328 if (!x2apic_preenabled && skip_ioapic_setup) {
1338 printk(KERN_INFO 1329 pr_info("Skipped enabling x2apic and Interrupt-remapping "
1339 "Skipped enabling x2apic and Interrupt-remapping " 1330 "because of skipping io-apic setup\n");
1340 "because of skipping io-apic setup\n");
1341 return; 1331 return;
1342 } 1332 }
1343 1333
1344 ret = dmar_table_init(); 1334 ret = dmar_table_init();
1345 if (ret) { 1335 if (ret) {
1346 printk(KERN_INFO 1336 pr_info("dmar_table_init() failed with %d:\n", ret);
1347 "dmar_table_init() failed with %d:\n", ret);
1348 1337
1349 if (x2apic_preenabled) 1338 if (x2apic_preenabled)
1350 panic("x2apic enabled by bios. But IR enabling failed"); 1339 panic("x2apic enabled by bios. But IR enabling failed");
1351 else 1340 else
1352 printk(KERN_INFO 1341 pr_info("Not enabling x2apic,Intr-remapping\n");
1353 "Not enabling x2apic,Intr-remapping\n");
1354 return; 1342 return;
1355 } 1343 }
1356 1344
@@ -1359,7 +1347,7 @@ void enable_IR_x2apic(void)
1359 1347
1360 ret = save_mask_IO_APIC_setup(); 1348 ret = save_mask_IO_APIC_setup();
1361 if (ret) { 1349 if (ret) {
1362 printk(KERN_INFO "Saving IO-APIC state failed: %d\n", ret); 1350 pr_info("Saving IO-APIC state failed: %d\n", ret);
1363 goto end; 1351 goto end;
1364 } 1352 }
1365 1353
@@ -1394,14 +1382,11 @@ end:
1394 1382
1395 if (!ret) { 1383 if (!ret) {
1396 if (!x2apic_preenabled) 1384 if (!x2apic_preenabled)
1397 printk(KERN_INFO 1385 pr_info("Enabled x2apic and interrupt-remapping\n");
1398 "Enabled x2apic and interrupt-remapping\n");
1399 else 1386 else
1400 printk(KERN_INFO 1387 pr_info("Enabled Interrupt-remapping\n");
1401 "Enabled Interrupt-remapping\n");
1402 } else 1388 } else
1403 printk(KERN_ERR 1389 pr_err("Failed to enable Interrupt-remapping and x2apic\n");
1404 "Failed to enable Interrupt-remapping and x2apic\n");
1405#else 1390#else
1406 if (!cpu_has_x2apic) 1391 if (!cpu_has_x2apic)
1407 return; 1392 return;
@@ -1410,8 +1395,8 @@ end:
1410 panic("x2apic enabled prior OS handover," 1395 panic("x2apic enabled prior OS handover,"
1411 " enable CONFIG_INTR_REMAP"); 1396 " enable CONFIG_INTR_REMAP");
1412 1397
1413 printk(KERN_INFO "Enable CONFIG_INTR_REMAP for enabling intr-remapping " 1398 pr_info("Enable CONFIG_INTR_REMAP for enabling intr-remapping "
1414 " and x2apic\n"); 1399 " and x2apic\n");
1415#endif 1400#endif
1416 1401
1417 return; 1402 return;
@@ -1428,7 +1413,7 @@ end:
1428static int __init detect_init_APIC(void) 1413static int __init detect_init_APIC(void)
1429{ 1414{
1430 if (!cpu_has_apic) { 1415 if (!cpu_has_apic) {
1431 printk(KERN_INFO "No local APIC present\n"); 1416 pr_info("No local APIC present\n");
1432 return -1; 1417 return -1;
1433 } 1418 }
1434 1419
@@ -1469,8 +1454,8 @@ static int __init detect_init_APIC(void)
1469 * "lapic" specified. 1454 * "lapic" specified.
1470 */ 1455 */
1471 if (!force_enable_local_apic) { 1456 if (!force_enable_local_apic) {
1472 printk(KERN_INFO "Local APIC disabled by BIOS -- " 1457 pr_info("Local APIC disabled by BIOS -- "
1473 "you can enable it with \"lapic\"\n"); 1458 "you can enable it with \"lapic\"\n");
1474 return -1; 1459 return -1;
1475 } 1460 }
1476 /* 1461 /*
@@ -1480,8 +1465,7 @@ static int __init detect_init_APIC(void)
1480 */ 1465 */
1481 rdmsr(MSR_IA32_APICBASE, l, h); 1466 rdmsr(MSR_IA32_APICBASE, l, h);
1482 if (!(l & MSR_IA32_APICBASE_ENABLE)) { 1467 if (!(l & MSR_IA32_APICBASE_ENABLE)) {
1483 printk(KERN_INFO 1468 pr_info("Local APIC disabled by BIOS -- reenabling.\n");
1484 "Local APIC disabled by BIOS -- reenabling.\n");
1485 l &= ~MSR_IA32_APICBASE_BASE; 1469 l &= ~MSR_IA32_APICBASE_BASE;
1486 l |= MSR_IA32_APICBASE_ENABLE | APIC_DEFAULT_PHYS_BASE; 1470 l |= MSR_IA32_APICBASE_ENABLE | APIC_DEFAULT_PHYS_BASE;
1487 wrmsr(MSR_IA32_APICBASE, l, h); 1471 wrmsr(MSR_IA32_APICBASE, l, h);
@@ -1494,7 +1478,7 @@ static int __init detect_init_APIC(void)
1494 */ 1478 */
1495 features = cpuid_edx(1); 1479 features = cpuid_edx(1);
1496 if (!(features & (1 << X86_FEATURE_APIC))) { 1480 if (!(features & (1 << X86_FEATURE_APIC))) {
1497 printk(KERN_WARNING "Could not enable APIC!\n"); 1481 pr_warning("Could not enable APIC!\n");
1498 return -1; 1482 return -1;
1499 } 1483 }
1500 set_cpu_cap(&boot_cpu_data, X86_FEATURE_APIC); 1484 set_cpu_cap(&boot_cpu_data, X86_FEATURE_APIC);
@@ -1505,14 +1489,14 @@ static int __init detect_init_APIC(void)
1505 if (l & MSR_IA32_APICBASE_ENABLE) 1489 if (l & MSR_IA32_APICBASE_ENABLE)
1506 mp_lapic_addr = l & MSR_IA32_APICBASE_BASE; 1490 mp_lapic_addr = l & MSR_IA32_APICBASE_BASE;
1507 1491
1508 printk(KERN_INFO "Found and enabled local APIC!\n"); 1492 pr_info("Found and enabled local APIC!\n");
1509 1493
1510 apic_pm_activate(); 1494 apic_pm_activate();
1511 1495
1512 return 0; 1496 return 0;
1513 1497
1514no_apic: 1498no_apic:
1515 printk(KERN_INFO "No local APIC present or hardware disabled\n"); 1499 pr_info("No local APIC present or hardware disabled\n");
1516 return -1; 1500 return -1;
1517} 1501}
1518#endif 1502#endif
@@ -1588,12 +1572,12 @@ int __init APIC_init_uniprocessor(void)
1588{ 1572{
1589#ifdef CONFIG_X86_64 1573#ifdef CONFIG_X86_64
1590 if (disable_apic) { 1574 if (disable_apic) {
1591 printk(KERN_INFO "Apic disabled\n"); 1575 pr_info("Apic disabled\n");
1592 return -1; 1576 return -1;
1593 } 1577 }
1594 if (!cpu_has_apic) { 1578 if (!cpu_has_apic) {
1595 disable_apic = 1; 1579 disable_apic = 1;
1596 printk(KERN_INFO "Apic disabled by BIOS\n"); 1580 pr_info("Apic disabled by BIOS\n");
1597 return -1; 1581 return -1;
1598 } 1582 }
1599#else 1583#else
@@ -1605,8 +1589,8 @@ int __init APIC_init_uniprocessor(void)
1605 */ 1589 */
1606 if (!cpu_has_apic && 1590 if (!cpu_has_apic &&
1607 APIC_INTEGRATED(apic_version[boot_cpu_physical_apicid])) { 1591 APIC_INTEGRATED(apic_version[boot_cpu_physical_apicid])) {
1608 printk(KERN_ERR "BIOS bug, local APIC 0x%x not detected!...\n", 1592 pr_err("BIOS bug, local APIC 0x%x not detected!...\n",
1609 boot_cpu_physical_apicid); 1593 boot_cpu_physical_apicid);
1610 clear_cpu_cap(&boot_cpu_data, X86_FEATURE_APIC); 1594 clear_cpu_cap(&boot_cpu_data, X86_FEATURE_APIC);
1611 return -1; 1595 return -1;
1612 } 1596 }
@@ -1682,9 +1666,7 @@ void smp_spurious_interrupt(struct pt_regs *regs)
1682{ 1666{
1683 u32 v; 1667 u32 v;
1684 1668
1685#ifdef CONFIG_X86_64
1686 exit_idle(); 1669 exit_idle();
1687#endif
1688 irq_enter(); 1670 irq_enter();
1689 /* 1671 /*
1690 * Check if this really is a spurious interrupt and ACK it 1672 * Check if this really is a spurious interrupt and ACK it
@@ -1699,8 +1681,8 @@ void smp_spurious_interrupt(struct pt_regs *regs)
1699 add_pda(irq_spurious_count, 1); 1681 add_pda(irq_spurious_count, 1);
1700#else 1682#else
1701 /* see sw-dev-man vol 3, chapter 7.4.13.5 */ 1683 /* see sw-dev-man vol 3, chapter 7.4.13.5 */
1702 printk(KERN_INFO "spurious APIC interrupt on CPU#%d, " 1684 pr_info("spurious APIC interrupt on CPU#%d, "
1703 "should never happen.\n", smp_processor_id()); 1685 "should never happen.\n", smp_processor_id());
1704 __get_cpu_var(irq_stat).irq_spurious_count++; 1686 __get_cpu_var(irq_stat).irq_spurious_count++;
1705#endif 1687#endif
1706 irq_exit(); 1688 irq_exit();
@@ -1713,9 +1695,7 @@ void smp_error_interrupt(struct pt_regs *regs)
1713{ 1695{
1714 u32 v, v1; 1696 u32 v, v1;
1715 1697
1716#ifdef CONFIG_X86_64
1717 exit_idle(); 1698 exit_idle();
1718#endif
1719 irq_enter(); 1699 irq_enter();
1720 /* First tickle the hardware, only then report what went on. -- REW */ 1700 /* First tickle the hardware, only then report what went on. -- REW */
1721 v = apic_read(APIC_ESR); 1701 v = apic_read(APIC_ESR);
@@ -1724,17 +1704,18 @@ void smp_error_interrupt(struct pt_regs *regs)
1724 ack_APIC_irq(); 1704 ack_APIC_irq();
1725 atomic_inc(&irq_err_count); 1705 atomic_inc(&irq_err_count);
1726 1706
1727 /* Here is what the APIC error bits mean: 1707 /*
1728 0: Send CS error 1708 * Here is what the APIC error bits mean:
1729 1: Receive CS error 1709 * 0: Send CS error
1730 2: Send accept error 1710 * 1: Receive CS error
1731 3: Receive accept error 1711 * 2: Send accept error
1732 4: Reserved 1712 * 3: Receive accept error
1733 5: Send illegal vector 1713 * 4: Reserved
1734 6: Received illegal vector 1714 * 5: Send illegal vector
1735 7: Illegal register address 1715 * 6: Received illegal vector
1736 */ 1716 * 7: Illegal register address
1737 printk(KERN_DEBUG "APIC error on CPU%d: %02x(%02x)\n", 1717 */
1718 pr_debug("APIC error on CPU%d: %02x(%02x)\n",
1738 smp_processor_id(), v , v1); 1719 smp_processor_id(), v , v1);
1739 irq_exit(); 1720 irq_exit();
1740} 1721}
@@ -1838,15 +1819,15 @@ void __cpuinit generic_processor_info(int apicid, int version)
1838 * Validate version 1819 * Validate version
1839 */ 1820 */
1840 if (version == 0x0) { 1821 if (version == 0x0) {
1841 printk(KERN_WARNING "BIOS bug, APIC version is 0 for CPU#%d! " 1822 pr_warning("BIOS bug, APIC version is 0 for CPU#%d! "
1842 "fixing up to 0x10. (tell your hw vendor)\n", 1823 "fixing up to 0x10. (tell your hw vendor)\n",
1843 version); 1824 version);
1844 version = 0x10; 1825 version = 0x10;
1845 } 1826 }
1846 apic_version[apicid] = version; 1827 apic_version[apicid] = version;
1847 1828
1848 if (num_processors >= NR_CPUS) { 1829 if (num_processors >= NR_CPUS) {
1849 printk(KERN_WARNING "WARNING: NR_CPUS limit of %i reached." 1830 pr_warning("WARNING: NR_CPUS limit of %i reached."
1850 " Processor ignored.\n", NR_CPUS); 1831 " Processor ignored.\n", NR_CPUS);
1851 return; 1832 return;
1852 } 1833 }
@@ -2209,7 +2190,7 @@ static int __init apic_set_verbosity(char *arg)
2209 else if (strcmp("verbose", arg) == 0) 2190 else if (strcmp("verbose", arg) == 0)
2210 apic_verbosity = APIC_VERBOSE; 2191 apic_verbosity = APIC_VERBOSE;
2211 else { 2192 else {
2212 printk(KERN_WARNING "APIC Verbosity level %s not recognised" 2193 pr_warning("APIC Verbosity level %s not recognised"
2213 " use apic=verbose or apic=debug\n", arg); 2194 " use apic=verbose or apic=debug\n", arg);
2214 return -EINVAL; 2195 return -EINVAL;
2215 } 2196 }
diff --git a/arch/x86/kernel/apm_32.c b/arch/x86/kernel/apm_32.c
index 5145a6e72bbb..3a26525a3f31 100644
--- a/arch/x86/kernel/apm_32.c
+++ b/arch/x86/kernel/apm_32.c
@@ -391,11 +391,7 @@ static int power_off;
391#else 391#else
392static int power_off = 1; 392static int power_off = 1;
393#endif 393#endif
394#ifdef CONFIG_APM_REAL_MODE_POWER_OFF
395static int realmode_power_off = 1;
396#else
397static int realmode_power_off; 394static int realmode_power_off;
398#endif
399#ifdef CONFIG_APM_ALLOW_INTS 395#ifdef CONFIG_APM_ALLOW_INTS
400static int allow_ints = 1; 396static int allow_ints = 1;
401#else 397#else
diff --git a/arch/x86/kernel/asm-offsets_32.c b/arch/x86/kernel/asm-offsets_32.c
index 6649d09ad88f..ee4df08feee6 100644
--- a/arch/x86/kernel/asm-offsets_32.c
+++ b/arch/x86/kernel/asm-offsets_32.c
@@ -11,7 +11,7 @@
11#include <linux/suspend.h> 11#include <linux/suspend.h>
12#include <linux/kbuild.h> 12#include <linux/kbuild.h>
13#include <asm/ucontext.h> 13#include <asm/ucontext.h>
14#include "sigframe.h" 14#include <asm/sigframe.h>
15#include <asm/pgtable.h> 15#include <asm/pgtable.h>
16#include <asm/fixmap.h> 16#include <asm/fixmap.h>
17#include <asm/processor.h> 17#include <asm/processor.h>
diff --git a/arch/x86/kernel/asm-offsets_64.c b/arch/x86/kernel/asm-offsets_64.c
index 7fcf63d22f8b..1d41d3f1edbc 100644
--- a/arch/x86/kernel/asm-offsets_64.c
+++ b/arch/x86/kernel/asm-offsets_64.c
@@ -20,6 +20,8 @@
20 20
21#include <xen/interface/xen.h> 21#include <xen/interface/xen.h>
22 22
23#include <asm/sigframe.h>
24
23#define __NO_STUBS 1 25#define __NO_STUBS 1
24#undef __SYSCALL 26#undef __SYSCALL
25#undef _ASM_X86_UNISTD_64_H 27#undef _ASM_X86_UNISTD_64_H
@@ -87,7 +89,7 @@ int main(void)
87 BLANK(); 89 BLANK();
88#undef ENTRY 90#undef ENTRY
89 DEFINE(IA32_RT_SIGFRAME_sigcontext, 91 DEFINE(IA32_RT_SIGFRAME_sigcontext,
90 offsetof (struct rt_sigframe32, uc.uc_mcontext)); 92 offsetof (struct rt_sigframe_ia32, uc.uc_mcontext));
91 BLANK(); 93 BLANK();
92#endif 94#endif
93 DEFINE(pbe_address, offsetof(struct pbe, address)); 95 DEFINE(pbe_address, offsetof(struct pbe, address));
diff --git a/arch/x86/kernel/bios_uv.c b/arch/x86/kernel/bios_uv.c
index f0dfe6f17e7e..2a0a2a3cac26 100644
--- a/arch/x86/kernel/bios_uv.c
+++ b/arch/x86/kernel/bios_uv.c
@@ -69,10 +69,10 @@ s64 uv_bios_call_reentrant(enum uv_bios_cmd which, u64 a1, u64 a2, u64 a3,
69 69
70long sn_partition_id; 70long sn_partition_id;
71EXPORT_SYMBOL_GPL(sn_partition_id); 71EXPORT_SYMBOL_GPL(sn_partition_id);
72long uv_coherency_id; 72long sn_coherency_id;
73EXPORT_SYMBOL_GPL(uv_coherency_id); 73EXPORT_SYMBOL_GPL(sn_coherency_id);
74long uv_region_size; 74long sn_region_size;
75EXPORT_SYMBOL_GPL(uv_region_size); 75EXPORT_SYMBOL_GPL(sn_region_size);
76int uv_type; 76int uv_type;
77 77
78 78
@@ -100,6 +100,56 @@ s64 uv_bios_get_sn_info(int fc, int *uvtype, long *partid, long *coher,
100 return ret; 100 return ret;
101} 101}
102 102
103int
104uv_bios_mq_watchlist_alloc(int blade, unsigned long addr, unsigned int mq_size,
105 unsigned long *intr_mmr_offset)
106{
107 union uv_watchlist_u size_blade;
108 u64 watchlist;
109 s64 ret;
110
111 size_blade.size = mq_size;
112 size_blade.blade = blade;
113
114 /*
115 * bios returns watchlist number or negative error number.
116 */
117 ret = (int)uv_bios_call_irqsave(UV_BIOS_WATCHLIST_ALLOC, addr,
118 size_blade.val, (u64)intr_mmr_offset,
119 (u64)&watchlist, 0);
120 if (ret < BIOS_STATUS_SUCCESS)
121 return ret;
122
123 return watchlist;
124}
125EXPORT_SYMBOL_GPL(uv_bios_mq_watchlist_alloc);
126
127int
128uv_bios_mq_watchlist_free(int blade, int watchlist_num)
129{
130 return (int)uv_bios_call_irqsave(UV_BIOS_WATCHLIST_FREE,
131 blade, watchlist_num, 0, 0, 0);
132}
133EXPORT_SYMBOL_GPL(uv_bios_mq_watchlist_free);
134
135s64
136uv_bios_change_memprotect(u64 paddr, u64 len, enum uv_memprotect perms)
137{
138 return uv_bios_call_irqsave(UV_BIOS_MEMPROTECT, paddr, len,
139 perms, 0, 0);
140}
141EXPORT_SYMBOL_GPL(uv_bios_change_memprotect);
142
143s64
144uv_bios_reserved_page_pa(u64 buf, u64 *cookie, u64 *addr, u64 *len)
145{
146 s64 ret;
147
148 ret = uv_bios_call_irqsave(UV_BIOS_GET_PARTITION_ADDR, (u64)cookie,
149 (u64)addr, buf, (u64)len, 0);
150 return ret;
151}
152EXPORT_SYMBOL_GPL(uv_bios_reserved_page_pa);
103 153
104s64 uv_bios_freq_base(u64 clock_type, u64 *ticks_per_second) 154s64 uv_bios_freq_base(u64 clock_type, u64 *ticks_per_second)
105{ 155{
diff --git a/arch/x86/kernel/check.c b/arch/x86/kernel/check.c
new file mode 100644
index 000000000000..2ac0ab71412a
--- /dev/null
+++ b/arch/x86/kernel/check.c
@@ -0,0 +1,161 @@
1#include <linux/module.h>
2#include <linux/sched.h>
3#include <linux/kthread.h>
4#include <linux/workqueue.h>
5#include <asm/e820.h>
6#include <asm/proto.h>
7
8/*
9 * Some BIOSes seem to corrupt the low 64k of memory during events
10 * like suspend/resume and unplugging an HDMI cable. Reserve all
11 * remaining free memory in that area and fill it with a distinct
12 * pattern.
13 */
14#define MAX_SCAN_AREAS 8
15
16static int __read_mostly memory_corruption_check = -1;
17
18static unsigned __read_mostly corruption_check_size = 64*1024;
19static unsigned __read_mostly corruption_check_period = 60; /* seconds */
20
21static struct e820entry scan_areas[MAX_SCAN_AREAS];
22static int num_scan_areas;
23
24
25static __init int set_corruption_check(char *arg)
26{
27 char *end;
28
29 memory_corruption_check = simple_strtol(arg, &end, 10);
30
31 return (*end == 0) ? 0 : -EINVAL;
32}
33early_param("memory_corruption_check", set_corruption_check);
34
35static __init int set_corruption_check_period(char *arg)
36{
37 char *end;
38
39 corruption_check_period = simple_strtoul(arg, &end, 10);
40
41 return (*end == 0) ? 0 : -EINVAL;
42}
43early_param("memory_corruption_check_period", set_corruption_check_period);
44
45static __init int set_corruption_check_size(char *arg)
46{
47 char *end;
48 unsigned size;
49
50 size = memparse(arg, &end);
51
52 if (*end == '\0')
53 corruption_check_size = size;
54
55 return (size == corruption_check_size) ? 0 : -EINVAL;
56}
57early_param("memory_corruption_check_size", set_corruption_check_size);
58
59
60void __init setup_bios_corruption_check(void)
61{
62 u64 addr = PAGE_SIZE; /* assume first page is reserved anyway */
63
64 if (memory_corruption_check == -1) {
65 memory_corruption_check =
66#ifdef CONFIG_X86_BOOTPARAM_MEMORY_CORRUPTION_CHECK
67 1
68#else
69 0
70#endif
71 ;
72 }
73
74 if (corruption_check_size == 0)
75 memory_corruption_check = 0;
76
77 if (!memory_corruption_check)
78 return;
79
80 corruption_check_size = round_up(corruption_check_size, PAGE_SIZE);
81
82 while (addr < corruption_check_size && num_scan_areas < MAX_SCAN_AREAS) {
83 u64 size;
84 addr = find_e820_area_size(addr, &size, PAGE_SIZE);
85
86 if (addr == 0)
87 break;
88
89 if ((addr + size) > corruption_check_size)
90 size = corruption_check_size - addr;
91
92 if (size == 0)
93 break;
94
95 e820_update_range(addr, size, E820_RAM, E820_RESERVED);
96 scan_areas[num_scan_areas].addr = addr;
97 scan_areas[num_scan_areas].size = size;
98 num_scan_areas++;
99
100 /* Assume we've already mapped this early memory */
101 memset(__va(addr), 0, size);
102
103 addr += size;
104 }
105
106 printk(KERN_INFO "Scanning %d areas for low memory corruption\n",
107 num_scan_areas);
108 update_e820();
109}
110
111
112void check_for_bios_corruption(void)
113{
114 int i;
115 int corruption = 0;
116
117 if (!memory_corruption_check)
118 return;
119
120 for (i = 0; i < num_scan_areas; i++) {
121 unsigned long *addr = __va(scan_areas[i].addr);
122 unsigned long size = scan_areas[i].size;
123
124 for (; size; addr++, size -= sizeof(unsigned long)) {
125 if (!*addr)
126 continue;
127 printk(KERN_ERR "Corrupted low memory at %p (%lx phys) = %08lx\n",
128 addr, __pa(addr), *addr);
129 corruption = 1;
130 *addr = 0;
131 }
132 }
133
134 WARN_ONCE(corruption, KERN_ERR "Memory corruption detected in low memory\n");
135}
136
137static void check_corruption(struct work_struct *dummy);
138static DECLARE_DELAYED_WORK(bios_check_work, check_corruption);
139
140static void check_corruption(struct work_struct *dummy)
141{
142 check_for_bios_corruption();
143 schedule_delayed_work(&bios_check_work,
144 round_jiffies_relative(corruption_check_period*HZ));
145}
146
147static int start_periodic_check_for_corruption(void)
148{
149 if (!memory_corruption_check || corruption_check_period == 0)
150 return 0;
151
152 printk(KERN_INFO "Scanning for low memory corruption every %d seconds\n",
153 corruption_check_period);
154
155 /* First time we run the checks right away */
156 schedule_delayed_work(&bios_check_work, 0);
157 return 0;
158}
159
160module_init(start_periodic_check_for_corruption);
161
diff --git a/arch/x86/kernel/cpu/Makefile b/arch/x86/kernel/cpu/Makefile
index 82ec6075c057..a5c04e88777e 100644
--- a/arch/x86/kernel/cpu/Makefile
+++ b/arch/x86/kernel/cpu/Makefile
@@ -4,6 +4,7 @@
4 4
5obj-y := intel_cacheinfo.o addon_cpuid_features.o 5obj-y := intel_cacheinfo.o addon_cpuid_features.o
6obj-y += proc.o capflags.o powerflags.o common.o 6obj-y += proc.o capflags.o powerflags.o common.o
7obj-y += vmware.o hypervisor.o
7 8
8obj-$(CONFIG_X86_32) += bugs.o cmpxchg.o 9obj-$(CONFIG_X86_32) += bugs.o cmpxchg.o
9obj-$(CONFIG_X86_64) += bugs_64.o 10obj-$(CONFIG_X86_64) += bugs_64.o
diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c
index b9c9ea0217a9..42e0853030cb 100644
--- a/arch/x86/kernel/cpu/common.c
+++ b/arch/x86/kernel/cpu/common.c
@@ -36,6 +36,7 @@
36#include <asm/proto.h> 36#include <asm/proto.h>
37#include <asm/sections.h> 37#include <asm/sections.h>
38#include <asm/setup.h> 38#include <asm/setup.h>
39#include <asm/hypervisor.h>
39 40
40#include "cpu.h" 41#include "cpu.h"
41 42
@@ -703,6 +704,7 @@ static void __cpuinit identify_cpu(struct cpuinfo_x86 *c)
703 detect_ht(c); 704 detect_ht(c);
704#endif 705#endif
705 706
707 init_hypervisor(c);
706 /* 708 /*
707 * On SMP, boot_cpu_data holds the common feature set between 709 * On SMP, boot_cpu_data holds the common feature set between
708 * all CPUs; so make sure that we indicate which features are 710 * all CPUs; so make sure that we indicate which features are
@@ -862,7 +864,7 @@ EXPORT_SYMBOL(_cpu_pda);
862 864
863struct desc_ptr idt_descr = { 256 * 16 - 1, (unsigned long) idt_table }; 865struct desc_ptr idt_descr = { 256 * 16 - 1, (unsigned long) idt_table };
864 866
865char boot_cpu_stack[IRQSTACKSIZE] __page_aligned_bss; 867static char boot_cpu_stack[IRQSTACKSIZE] __page_aligned_bss;
866 868
867void __cpuinit pda_init(int cpu) 869void __cpuinit pda_init(int cpu)
868{ 870{
@@ -903,8 +905,8 @@ void __cpuinit pda_init(int cpu)
903 } 905 }
904} 906}
905 907
906char boot_exception_stacks[(N_EXCEPTION_STACKS - 1) * EXCEPTION_STKSZ + 908static char boot_exception_stacks[(N_EXCEPTION_STACKS - 1) * EXCEPTION_STKSZ +
907 DEBUG_STKSZ] __page_aligned_bss; 909 DEBUG_STKSZ] __page_aligned_bss;
908 910
909extern asmlinkage void ignore_sysret(void); 911extern asmlinkage void ignore_sysret(void);
910 912
diff --git a/arch/x86/kernel/cpu/cpufreq/powernow-k8.c b/arch/x86/kernel/cpu/cpufreq/powernow-k8.c
index d3dcd58b87cd..7f05f44b97e9 100644
--- a/arch/x86/kernel/cpu/cpufreq/powernow-k8.c
+++ b/arch/x86/kernel/cpu/cpufreq/powernow-k8.c
@@ -115,9 +115,20 @@ static int query_current_values_with_pending_wait(struct powernow_k8_data *data)
115 u32 i = 0; 115 u32 i = 0;
116 116
117 if (cpu_family == CPU_HW_PSTATE) { 117 if (cpu_family == CPU_HW_PSTATE) {
118 rdmsr(MSR_PSTATE_STATUS, lo, hi); 118 if (data->currpstate == HW_PSTATE_INVALID) {
119 i = lo & HW_PSTATE_MASK; 119 /* read (initial) hw pstate if not yet set */
120 data->currpstate = i; 120 rdmsr(MSR_PSTATE_STATUS, lo, hi);
121 i = lo & HW_PSTATE_MASK;
122
123 /*
124 * a workaround for family 11h erratum 311 might cause
125 * an "out-of-range Pstate if the core is in Pstate-0
126 */
127 if (i >= data->numps)
128 data->currpstate = HW_PSTATE_0;
129 else
130 data->currpstate = i;
131 }
121 return 0; 132 return 0;
122 } 133 }
123 do { 134 do {
@@ -1121,6 +1132,7 @@ static int __cpuinit powernowk8_cpu_init(struct cpufreq_policy *pol)
1121 } 1132 }
1122 1133
1123 data->cpu = pol->cpu; 1134 data->cpu = pol->cpu;
1135 data->currpstate = HW_PSTATE_INVALID;
1124 1136
1125 if (powernow_k8_cpu_init_acpi(data)) { 1137 if (powernow_k8_cpu_init_acpi(data)) {
1126 /* 1138 /*
diff --git a/arch/x86/kernel/cpu/cpufreq/powernow-k8.h b/arch/x86/kernel/cpu/cpufreq/powernow-k8.h
index ab48cfed4d96..65cfb5d7f77f 100644
--- a/arch/x86/kernel/cpu/cpufreq/powernow-k8.h
+++ b/arch/x86/kernel/cpu/cpufreq/powernow-k8.h
@@ -5,6 +5,19 @@
5 * http://www.gnu.org/licenses/gpl.html 5 * http://www.gnu.org/licenses/gpl.html
6 */ 6 */
7 7
8
9enum pstate {
10 HW_PSTATE_INVALID = 0xff,
11 HW_PSTATE_0 = 0,
12 HW_PSTATE_1 = 1,
13 HW_PSTATE_2 = 2,
14 HW_PSTATE_3 = 3,
15 HW_PSTATE_4 = 4,
16 HW_PSTATE_5 = 5,
17 HW_PSTATE_6 = 6,
18 HW_PSTATE_7 = 7,
19};
20
8struct powernow_k8_data { 21struct powernow_k8_data {
9 unsigned int cpu; 22 unsigned int cpu;
10 23
@@ -23,7 +36,9 @@ struct powernow_k8_data {
23 u32 exttype; /* extended interface = 1 */ 36 u32 exttype; /* extended interface = 1 */
24 37
25 /* keep track of the current fid / vid or pstate */ 38 /* keep track of the current fid / vid or pstate */
26 u32 currvid, currfid, currpstate; 39 u32 currvid;
40 u32 currfid;
41 enum pstate currpstate;
27 42
28 /* the powernow_table includes all frequency and vid/fid pairings: 43 /* the powernow_table includes all frequency and vid/fid pairings:
29 * fid are the lower 8 bits of the index, vid are the upper 8 bits. 44 * fid are the lower 8 bits of the index, vid are the upper 8 bits.
diff --git a/arch/x86/kernel/cpu/hypervisor.c b/arch/x86/kernel/cpu/hypervisor.c
new file mode 100644
index 000000000000..fb5b86af0b01
--- /dev/null
+++ b/arch/x86/kernel/cpu/hypervisor.c
@@ -0,0 +1,58 @@
1/*
2 * Common hypervisor code
3 *
4 * Copyright (C) 2008, VMware, Inc.
5 * Author : Alok N Kataria <akataria@vmware.com>
6 *
7 * This program is free software; you can redistribute it and/or modify
8 * it under the terms of the GNU General Public License as published by
9 * the Free Software Foundation; either version 2 of the License, or
10 * (at your option) any later version.
11 *
12 * This program is distributed in the hope that it will be useful, but
13 * WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or
15 * NON INFRINGEMENT. See the GNU General Public License for more
16 * details.
17 *
18 * You should have received a copy of the GNU General Public License
19 * along with this program; if not, write to the Free Software
20 * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
21 *
22 */
23
24#include <asm/processor.h>
25#include <asm/vmware.h>
26#include <asm/hypervisor.h>
27
28static inline void __cpuinit
29detect_hypervisor_vendor(struct cpuinfo_x86 *c)
30{
31 if (vmware_platform()) {
32 c->x86_hyper_vendor = X86_HYPER_VENDOR_VMWARE;
33 } else {
34 c->x86_hyper_vendor = X86_HYPER_VENDOR_NONE;
35 }
36}
37
38unsigned long get_hypervisor_tsc_freq(void)
39{
40 if (boot_cpu_data.x86_hyper_vendor == X86_HYPER_VENDOR_VMWARE)
41 return vmware_get_tsc_khz();
42 return 0;
43}
44
45static inline void __cpuinit
46hypervisor_set_feature_bits(struct cpuinfo_x86 *c)
47{
48 if (boot_cpu_data.x86_hyper_vendor == X86_HYPER_VENDOR_VMWARE) {
49 vmware_set_feature_bits(c);
50 return;
51 }
52}
53
54void __cpuinit init_hypervisor(struct cpuinfo_x86 *c)
55{
56 detect_hypervisor_vendor(c);
57 hypervisor_set_feature_bits(c);
58}
diff --git a/arch/x86/kernel/cpu/intel_cacheinfo.c b/arch/x86/kernel/cpu/intel_cacheinfo.c
index 3f46afbb1cf1..68b5d8681cbb 100644
--- a/arch/x86/kernel/cpu/intel_cacheinfo.c
+++ b/arch/x86/kernel/cpu/intel_cacheinfo.c
@@ -644,20 +644,17 @@ static inline ssize_t show_shared_cpu_list(struct _cpuid4_info *leaf, char *buf)
644 return show_shared_cpu_map_func(leaf, 1, buf); 644 return show_shared_cpu_map_func(leaf, 1, buf);
645} 645}
646 646
647static ssize_t show_type(struct _cpuid4_info *this_leaf, char *buf) { 647static ssize_t show_type(struct _cpuid4_info *this_leaf, char *buf)
648 switch(this_leaf->eax.split.type) { 648{
649 case CACHE_TYPE_DATA: 649 switch (this_leaf->eax.split.type) {
650 case CACHE_TYPE_DATA:
650 return sprintf(buf, "Data\n"); 651 return sprintf(buf, "Data\n");
651 break; 652 case CACHE_TYPE_INST:
652 case CACHE_TYPE_INST:
653 return sprintf(buf, "Instruction\n"); 653 return sprintf(buf, "Instruction\n");
654 break; 654 case CACHE_TYPE_UNIFIED:
655 case CACHE_TYPE_UNIFIED:
656 return sprintf(buf, "Unified\n"); 655 return sprintf(buf, "Unified\n");
657 break; 656 default:
658 default:
659 return sprintf(buf, "Unknown\n"); 657 return sprintf(buf, "Unknown\n");
660 break;
661 } 658 }
662} 659}
663 660
diff --git a/arch/x86/kernel/cpu/mcheck/mce_64.c b/arch/x86/kernel/cpu/mcheck/mce_64.c
index 4b031a4ac856..1c838032fd37 100644
--- a/arch/x86/kernel/cpu/mcheck/mce_64.c
+++ b/arch/x86/kernel/cpu/mcheck/mce_64.c
@@ -510,12 +510,9 @@ static void __cpuinit mce_cpu_features(struct cpuinfo_x86 *c)
510 */ 510 */
511void __cpuinit mcheck_init(struct cpuinfo_x86 *c) 511void __cpuinit mcheck_init(struct cpuinfo_x86 *c)
512{ 512{
513 static cpumask_t mce_cpus = CPU_MASK_NONE;
514
515 mce_cpu_quirks(c); 513 mce_cpu_quirks(c);
516 514
517 if (mce_dont_init || 515 if (mce_dont_init ||
518 cpu_test_and_set(smp_processor_id(), mce_cpus) ||
519 !mce_available(c)) 516 !mce_available(c))
520 return; 517 return;
521 518
diff --git a/arch/x86/kernel/cpu/mtrr/main.c b/arch/x86/kernel/cpu/mtrr/main.c
index c78c04821ea1..1159e269e596 100644
--- a/arch/x86/kernel/cpu/mtrr/main.c
+++ b/arch/x86/kernel/cpu/mtrr/main.c
@@ -803,6 +803,7 @@ x86_get_mtrr_mem_range(struct res_range *range, int nr_range,
803} 803}
804 804
805static struct res_range __initdata range[RANGE_NUM]; 805static struct res_range __initdata range[RANGE_NUM];
806static int __initdata nr_range;
806 807
807#ifdef CONFIG_MTRR_SANITIZER 808#ifdef CONFIG_MTRR_SANITIZER
808 809
@@ -1206,39 +1207,43 @@ struct mtrr_cleanup_result {
1206#define PSHIFT (PAGE_SHIFT - 10) 1207#define PSHIFT (PAGE_SHIFT - 10)
1207 1208
1208static struct mtrr_cleanup_result __initdata result[NUM_RESULT]; 1209static struct mtrr_cleanup_result __initdata result[NUM_RESULT];
1209static struct res_range __initdata range_new[RANGE_NUM];
1210static unsigned long __initdata min_loss_pfn[RANGE_NUM]; 1210static unsigned long __initdata min_loss_pfn[RANGE_NUM];
1211 1211
1212static int __init mtrr_cleanup(unsigned address_bits) 1212static void __init print_out_mtrr_range_state(void)
1213{ 1213{
1214 unsigned long extra_remove_base, extra_remove_size;
1215 unsigned long base, size, def, dummy;
1216 mtrr_type type;
1217 int nr_range, nr_range_new;
1218 u64 chunk_size, gran_size;
1219 unsigned long range_sums, range_sums_new;
1220 int index_good;
1221 int num_reg_good;
1222 int i; 1214 int i;
1215 char start_factor = 'K', size_factor = 'K';
1216 unsigned long start_base, size_base;
1217 mtrr_type type;
1223 1218
1224 /* extra one for all 0 */ 1219 for (i = 0; i < num_var_ranges; i++) {
1225 int num[MTRR_NUM_TYPES + 1];
1226 1220
1227 if (!is_cpu(INTEL) || enable_mtrr_cleanup < 1) 1221 size_base = range_state[i].size_pfn << (PAGE_SHIFT - 10);
1228 return 0; 1222 if (!size_base)
1229 rdmsr(MTRRdefType_MSR, def, dummy); 1223 continue;
1230 def &= 0xff;
1231 if (def != MTRR_TYPE_UNCACHABLE)
1232 return 0;
1233 1224
1234 /* get it and store it aside */ 1225 size_base = to_size_factor(size_base, &size_factor),
1235 memset(range_state, 0, sizeof(range_state)); 1226 start_base = range_state[i].base_pfn << (PAGE_SHIFT - 10);
1236 for (i = 0; i < num_var_ranges; i++) { 1227 start_base = to_size_factor(start_base, &start_factor),
1237 mtrr_if->get(i, &base, &size, &type); 1228 type = range_state[i].type;
1238 range_state[i].base_pfn = base; 1229
1239 range_state[i].size_pfn = size; 1230 printk(KERN_DEBUG "reg %d, base: %ld%cB, range: %ld%cB, type %s\n",
1240 range_state[i].type = type; 1231 i, start_base, start_factor,
1232 size_base, size_factor,
1233 (type == MTRR_TYPE_UNCACHABLE) ? "UC" :
1234 ((type == MTRR_TYPE_WRPROT) ? "WP" :
1235 ((type == MTRR_TYPE_WRBACK) ? "WB" : "Other"))
1236 );
1241 } 1237 }
1238}
1239
1240static int __init mtrr_need_cleanup(void)
1241{
1242 int i;
1243 mtrr_type type;
1244 unsigned long size;
1245 /* extra one for all 0 */
1246 int num[MTRR_NUM_TYPES + 1];
1242 1247
1243 /* check entries number */ 1248 /* check entries number */
1244 memset(num, 0, sizeof(num)); 1249 memset(num, 0, sizeof(num));
@@ -1263,29 +1268,133 @@ static int __init mtrr_cleanup(unsigned address_bits)
1263 num_var_ranges - num[MTRR_NUM_TYPES]) 1268 num_var_ranges - num[MTRR_NUM_TYPES])
1264 return 0; 1269 return 0;
1265 1270
1266 /* print original var MTRRs at first, for debugging: */ 1271 return 1;
1267 printk(KERN_DEBUG "original variable MTRRs\n"); 1272}
1268 for (i = 0; i < num_var_ranges; i++) {
1269 char start_factor = 'K', size_factor = 'K';
1270 unsigned long start_base, size_base;
1271 1273
1272 size_base = range_state[i].size_pfn << (PAGE_SHIFT - 10); 1274static unsigned long __initdata range_sums;
1273 if (!size_base) 1275static void __init mtrr_calc_range_state(u64 chunk_size, u64 gran_size,
1274 continue; 1276 unsigned long extra_remove_base,
1277 unsigned long extra_remove_size,
1278 int i)
1279{
1280 int num_reg;
1281 static struct res_range range_new[RANGE_NUM];
1282 static int nr_range_new;
1283 unsigned long range_sums_new;
1284
1285 /* convert ranges to var ranges state */
1286 num_reg = x86_setup_var_mtrrs(range, nr_range,
1287 chunk_size, gran_size);
1288
1289 /* we got new setting in range_state, check it */
1290 memset(range_new, 0, sizeof(range_new));
1291 nr_range_new = x86_get_mtrr_mem_range(range_new, 0,
1292 extra_remove_base, extra_remove_size);
1293 range_sums_new = sum_ranges(range_new, nr_range_new);
1294
1295 result[i].chunk_sizek = chunk_size >> 10;
1296 result[i].gran_sizek = gran_size >> 10;
1297 result[i].num_reg = num_reg;
1298 if (range_sums < range_sums_new) {
1299 result[i].lose_cover_sizek =
1300 (range_sums_new - range_sums) << PSHIFT;
1301 result[i].bad = 1;
1302 } else
1303 result[i].lose_cover_sizek =
1304 (range_sums - range_sums_new) << PSHIFT;
1275 1305
1276 size_base = to_size_factor(size_base, &size_factor), 1306 /* double check it */
1277 start_base = range_state[i].base_pfn << (PAGE_SHIFT - 10); 1307 if (!result[i].bad && !result[i].lose_cover_sizek) {
1278 start_base = to_size_factor(start_base, &start_factor), 1308 if (nr_range_new != nr_range ||
1279 type = range_state[i].type; 1309 memcmp(range, range_new, sizeof(range)))
1310 result[i].bad = 1;
1311 }
1280 1312
1281 printk(KERN_DEBUG "reg %d, base: %ld%cB, range: %ld%cB, type %s\n", 1313 if (!result[i].bad && (range_sums - range_sums_new <
1282 i, start_base, start_factor, 1314 min_loss_pfn[num_reg])) {
1283 size_base, size_factor, 1315 min_loss_pfn[num_reg] =
1284 (type == MTRR_TYPE_UNCACHABLE) ? "UC" : 1316 range_sums - range_sums_new;
1285 ((type == MTRR_TYPE_WRPROT) ? "WP" :
1286 ((type == MTRR_TYPE_WRBACK) ? "WB" : "Other"))
1287 );
1288 } 1317 }
1318}
1319
1320static void __init mtrr_print_out_one_result(int i)
1321{
1322 char gran_factor, chunk_factor, lose_factor;
1323 unsigned long gran_base, chunk_base, lose_base;
1324
1325 gran_base = to_size_factor(result[i].gran_sizek, &gran_factor),
1326 chunk_base = to_size_factor(result[i].chunk_sizek, &chunk_factor),
1327 lose_base = to_size_factor(result[i].lose_cover_sizek, &lose_factor),
1328 printk(KERN_INFO "%sgran_size: %ld%c \tchunk_size: %ld%c \t",
1329 result[i].bad ? "*BAD*" : " ",
1330 gran_base, gran_factor, chunk_base, chunk_factor);
1331 printk(KERN_CONT "num_reg: %d \tlose cover RAM: %s%ld%c\n",
1332 result[i].num_reg, result[i].bad ? "-" : "",
1333 lose_base, lose_factor);
1334}
1335
1336static int __init mtrr_search_optimal_index(void)
1337{
1338 int i;
1339 int num_reg_good;
1340 int index_good;
1341
1342 if (nr_mtrr_spare_reg >= num_var_ranges)
1343 nr_mtrr_spare_reg = num_var_ranges - 1;
1344 num_reg_good = -1;
1345 for (i = num_var_ranges - nr_mtrr_spare_reg; i > 0; i--) {
1346 if (!min_loss_pfn[i])
1347 num_reg_good = i;
1348 }
1349
1350 index_good = -1;
1351 if (num_reg_good != -1) {
1352 for (i = 0; i < NUM_RESULT; i++) {
1353 if (!result[i].bad &&
1354 result[i].num_reg == num_reg_good &&
1355 !result[i].lose_cover_sizek) {
1356 index_good = i;
1357 break;
1358 }
1359 }
1360 }
1361
1362 return index_good;
1363}
1364
1365
1366static int __init mtrr_cleanup(unsigned address_bits)
1367{
1368 unsigned long extra_remove_base, extra_remove_size;
1369 unsigned long base, size, def, dummy;
1370 mtrr_type type;
1371 u64 chunk_size, gran_size;
1372 int index_good;
1373 int i;
1374
1375 if (!is_cpu(INTEL) || enable_mtrr_cleanup < 1)
1376 return 0;
1377 rdmsr(MTRRdefType_MSR, def, dummy);
1378 def &= 0xff;
1379 if (def != MTRR_TYPE_UNCACHABLE)
1380 return 0;
1381
1382 /* get it and store it aside */
1383 memset(range_state, 0, sizeof(range_state));
1384 for (i = 0; i < num_var_ranges; i++) {
1385 mtrr_if->get(i, &base, &size, &type);
1386 range_state[i].base_pfn = base;
1387 range_state[i].size_pfn = size;
1388 range_state[i].type = type;
1389 }
1390
1391 /* check if we need handle it and can handle it */
1392 if (!mtrr_need_cleanup())
1393 return 0;
1394
1395 /* print original var MTRRs at first, for debugging: */
1396 printk(KERN_DEBUG "original variable MTRRs\n");
1397 print_out_mtrr_range_state();
1289 1398
1290 memset(range, 0, sizeof(range)); 1399 memset(range, 0, sizeof(range));
1291 extra_remove_size = 0; 1400 extra_remove_size = 0;
@@ -1309,176 +1418,64 @@ static int __init mtrr_cleanup(unsigned address_bits)
1309 range_sums >> (20 - PAGE_SHIFT)); 1418 range_sums >> (20 - PAGE_SHIFT));
1310 1419
1311 if (mtrr_chunk_size && mtrr_gran_size) { 1420 if (mtrr_chunk_size && mtrr_gran_size) {
1312 int num_reg; 1421 i = 0;
1313 char gran_factor, chunk_factor, lose_factor; 1422 mtrr_calc_range_state(mtrr_chunk_size, mtrr_gran_size,
1314 unsigned long gran_base, chunk_base, lose_base; 1423 extra_remove_base, extra_remove_size, i);
1315
1316 debug_print++;
1317 /* convert ranges to var ranges state */
1318 num_reg = x86_setup_var_mtrrs(range, nr_range, mtrr_chunk_size,
1319 mtrr_gran_size);
1320 1424
1321 /* we got new setting in range_state, check it */ 1425 mtrr_print_out_one_result(i);
1322 memset(range_new, 0, sizeof(range_new));
1323 nr_range_new = x86_get_mtrr_mem_range(range_new, 0,
1324 extra_remove_base,
1325 extra_remove_size);
1326 range_sums_new = sum_ranges(range_new, nr_range_new);
1327 1426
1328 i = 0;
1329 result[i].chunk_sizek = mtrr_chunk_size >> 10;
1330 result[i].gran_sizek = mtrr_gran_size >> 10;
1331 result[i].num_reg = num_reg;
1332 if (range_sums < range_sums_new) {
1333 result[i].lose_cover_sizek =
1334 (range_sums_new - range_sums) << PSHIFT;
1335 result[i].bad = 1;
1336 } else
1337 result[i].lose_cover_sizek =
1338 (range_sums - range_sums_new) << PSHIFT;
1339
1340 gran_base = to_size_factor(result[i].gran_sizek, &gran_factor),
1341 chunk_base = to_size_factor(result[i].chunk_sizek, &chunk_factor),
1342 lose_base = to_size_factor(result[i].lose_cover_sizek, &lose_factor),
1343 printk(KERN_INFO "%sgran_size: %ld%c \tchunk_size: %ld%c \t",
1344 result[i].bad?"*BAD*":" ",
1345 gran_base, gran_factor, chunk_base, chunk_factor);
1346 printk(KERN_CONT "num_reg: %d \tlose cover RAM: %s%ld%c\n",
1347 result[i].num_reg, result[i].bad?"-":"",
1348 lose_base, lose_factor);
1349 if (!result[i].bad) { 1427 if (!result[i].bad) {
1350 set_var_mtrr_all(address_bits); 1428 set_var_mtrr_all(address_bits);
1351 return 1; 1429 return 1;
1352 } 1430 }
1353 printk(KERN_INFO "invalid mtrr_gran_size or mtrr_chunk_size, " 1431 printk(KERN_INFO "invalid mtrr_gran_size or mtrr_chunk_size, "
1354 "will find optimal one\n"); 1432 "will find optimal one\n");
1355 debug_print--;
1356 memset(result, 0, sizeof(result[0]));
1357 } 1433 }
1358 1434
1359 i = 0; 1435 i = 0;
1360 memset(min_loss_pfn, 0xff, sizeof(min_loss_pfn)); 1436 memset(min_loss_pfn, 0xff, sizeof(min_loss_pfn));
1361 memset(result, 0, sizeof(result)); 1437 memset(result, 0, sizeof(result));
1362 for (gran_size = (1ULL<<16); gran_size < (1ULL<<32); gran_size <<= 1) { 1438 for (gran_size = (1ULL<<16); gran_size < (1ULL<<32); gran_size <<= 1) {
1363 char gran_factor;
1364 unsigned long gran_base;
1365
1366 if (debug_print)
1367 gran_base = to_size_factor(gran_size >> 10, &gran_factor);
1368 1439
1369 for (chunk_size = gran_size; chunk_size < (1ULL<<32); 1440 for (chunk_size = gran_size; chunk_size < (1ULL<<32);
1370 chunk_size <<= 1) { 1441 chunk_size <<= 1) {
1371 int num_reg;
1372 1442
1373 if (debug_print) {
1374 char chunk_factor;
1375 unsigned long chunk_base;
1376
1377 chunk_base = to_size_factor(chunk_size>>10, &chunk_factor),
1378 printk(KERN_INFO "\n");
1379 printk(KERN_INFO "gran_size: %ld%c chunk_size: %ld%c \n",
1380 gran_base, gran_factor, chunk_base, chunk_factor);
1381 }
1382 if (i >= NUM_RESULT) 1443 if (i >= NUM_RESULT)
1383 continue; 1444 continue;
1384 1445
1385 /* convert ranges to var ranges state */ 1446 mtrr_calc_range_state(chunk_size, gran_size,
1386 num_reg = x86_setup_var_mtrrs(range, nr_range, 1447 extra_remove_base, extra_remove_size, i);
1387 chunk_size, gran_size); 1448 if (debug_print) {
1388 1449 mtrr_print_out_one_result(i);
1389 /* we got new setting in range_state, check it */ 1450 printk(KERN_INFO "\n");
1390 memset(range_new, 0, sizeof(range_new));
1391 nr_range_new = x86_get_mtrr_mem_range(range_new, 0,
1392 extra_remove_base, extra_remove_size);
1393 range_sums_new = sum_ranges(range_new, nr_range_new);
1394
1395 result[i].chunk_sizek = chunk_size >> 10;
1396 result[i].gran_sizek = gran_size >> 10;
1397 result[i].num_reg = num_reg;
1398 if (range_sums < range_sums_new) {
1399 result[i].lose_cover_sizek =
1400 (range_sums_new - range_sums) << PSHIFT;
1401 result[i].bad = 1;
1402 } else
1403 result[i].lose_cover_sizek =
1404 (range_sums - range_sums_new) << PSHIFT;
1405
1406 /* double check it */
1407 if (!result[i].bad && !result[i].lose_cover_sizek) {
1408 if (nr_range_new != nr_range ||
1409 memcmp(range, range_new, sizeof(range)))
1410 result[i].bad = 1;
1411 } 1451 }
1412 1452
1413 if (!result[i].bad && (range_sums - range_sums_new <
1414 min_loss_pfn[num_reg])) {
1415 min_loss_pfn[num_reg] =
1416 range_sums - range_sums_new;
1417 }
1418 i++; 1453 i++;
1419 } 1454 }
1420 } 1455 }
1421 1456
1422 /* print out all */
1423 for (i = 0; i < NUM_RESULT; i++) {
1424 char gran_factor, chunk_factor, lose_factor;
1425 unsigned long gran_base, chunk_base, lose_base;
1426
1427 gran_base = to_size_factor(result[i].gran_sizek, &gran_factor),
1428 chunk_base = to_size_factor(result[i].chunk_sizek, &chunk_factor),
1429 lose_base = to_size_factor(result[i].lose_cover_sizek, &lose_factor),
1430 printk(KERN_INFO "%sgran_size: %ld%c \tchunk_size: %ld%c \t",
1431 result[i].bad?"*BAD*":" ",
1432 gran_base, gran_factor, chunk_base, chunk_factor);
1433 printk(KERN_CONT "num_reg: %d \tlose cover RAM: %s%ld%c\n",
1434 result[i].num_reg, result[i].bad?"-":"",
1435 lose_base, lose_factor);
1436 }
1437
1438 /* try to find the optimal index */ 1457 /* try to find the optimal index */
1439 if (nr_mtrr_spare_reg >= num_var_ranges) 1458 index_good = mtrr_search_optimal_index();
1440 nr_mtrr_spare_reg = num_var_ranges - 1;
1441 num_reg_good = -1;
1442 for (i = num_var_ranges - nr_mtrr_spare_reg; i > 0; i--) {
1443 if (!min_loss_pfn[i])
1444 num_reg_good = i;
1445 }
1446
1447 index_good = -1;
1448 if (num_reg_good != -1) {
1449 for (i = 0; i < NUM_RESULT; i++) {
1450 if (!result[i].bad &&
1451 result[i].num_reg == num_reg_good &&
1452 !result[i].lose_cover_sizek) {
1453 index_good = i;
1454 break;
1455 }
1456 }
1457 }
1458 1459
1459 if (index_good != -1) { 1460 if (index_good != -1) {
1460 char gran_factor, chunk_factor, lose_factor;
1461 unsigned long gran_base, chunk_base, lose_base;
1462
1463 printk(KERN_INFO "Found optimal setting for mtrr clean up\n"); 1461 printk(KERN_INFO "Found optimal setting for mtrr clean up\n");
1464 i = index_good; 1462 i = index_good;
1465 gran_base = to_size_factor(result[i].gran_sizek, &gran_factor), 1463 mtrr_print_out_one_result(i);
1466 chunk_base = to_size_factor(result[i].chunk_sizek, &chunk_factor), 1464
1467 lose_base = to_size_factor(result[i].lose_cover_sizek, &lose_factor),
1468 printk(KERN_INFO "gran_size: %ld%c \tchunk_size: %ld%c \t",
1469 gran_base, gran_factor, chunk_base, chunk_factor);
1470 printk(KERN_CONT "num_reg: %d \tlose RAM: %ld%c\n",
1471 result[i].num_reg, lose_base, lose_factor);
1472 /* convert ranges to var ranges state */ 1465 /* convert ranges to var ranges state */
1473 chunk_size = result[i].chunk_sizek; 1466 chunk_size = result[i].chunk_sizek;
1474 chunk_size <<= 10; 1467 chunk_size <<= 10;
1475 gran_size = result[i].gran_sizek; 1468 gran_size = result[i].gran_sizek;
1476 gran_size <<= 10; 1469 gran_size <<= 10;
1477 debug_print++;
1478 x86_setup_var_mtrrs(range, nr_range, chunk_size, gran_size); 1470 x86_setup_var_mtrrs(range, nr_range, chunk_size, gran_size);
1479 debug_print--;
1480 set_var_mtrr_all(address_bits); 1471 set_var_mtrr_all(address_bits);
1472 printk(KERN_DEBUG "New variable MTRRs\n");
1473 print_out_mtrr_range_state();
1481 return 1; 1474 return 1;
1475 } else {
1476 /* print out all */
1477 for (i = 0; i < NUM_RESULT; i++)
1478 mtrr_print_out_one_result(i);
1482 } 1479 }
1483 1480
1484 printk(KERN_INFO "mtrr_cleanup: can not find optimal value\n"); 1481 printk(KERN_INFO "mtrr_cleanup: can not find optimal value\n");
@@ -1562,7 +1559,6 @@ int __init mtrr_trim_uncached_memory(unsigned long end_pfn)
1562{ 1559{
1563 unsigned long i, base, size, highest_pfn = 0, def, dummy; 1560 unsigned long i, base, size, highest_pfn = 0, def, dummy;
1564 mtrr_type type; 1561 mtrr_type type;
1565 int nr_range;
1566 u64 total_trim_size; 1562 u64 total_trim_size;
1567 1563
1568 /* extra one for all 0 */ 1564 /* extra one for all 0 */
diff --git a/arch/x86/kernel/cpu/vmware.c b/arch/x86/kernel/cpu/vmware.c
new file mode 100644
index 000000000000..284c399e3234
--- /dev/null
+++ b/arch/x86/kernel/cpu/vmware.c
@@ -0,0 +1,112 @@
1/*
2 * VMware Detection code.
3 *
4 * Copyright (C) 2008, VMware, Inc.
5 * Author : Alok N Kataria <akataria@vmware.com>
6 *
7 * This program is free software; you can redistribute it and/or modify
8 * it under the terms of the GNU General Public License as published by
9 * the Free Software Foundation; either version 2 of the License, or
10 * (at your option) any later version.
11 *
12 * This program is distributed in the hope that it will be useful, but
13 * WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or
15 * NON INFRINGEMENT. See the GNU General Public License for more
16 * details.
17 *
18 * You should have received a copy of the GNU General Public License
19 * along with this program; if not, write to the Free Software
20 * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
21 *
22 */
23
24#include <linux/dmi.h>
25#include <asm/div64.h>
26#include <asm/vmware.h>
27
28#define CPUID_VMWARE_INFO_LEAF 0x40000000
29#define VMWARE_HYPERVISOR_MAGIC 0x564D5868
30#define VMWARE_HYPERVISOR_PORT 0x5658
31
32#define VMWARE_PORT_CMD_GETVERSION 10
33#define VMWARE_PORT_CMD_GETHZ 45
34
35#define VMWARE_PORT(cmd, eax, ebx, ecx, edx) \
36 __asm__("inl (%%dx)" : \
37 "=a"(eax), "=c"(ecx), "=d"(edx), "=b"(ebx) : \
38 "0"(VMWARE_HYPERVISOR_MAGIC), \
39 "1"(VMWARE_PORT_CMD_##cmd), \
40 "2"(VMWARE_HYPERVISOR_PORT), "3"(UINT_MAX) : \
41 "memory");
42
43static inline int __vmware_platform(void)
44{
45 uint32_t eax, ebx, ecx, edx;
46 VMWARE_PORT(GETVERSION, eax, ebx, ecx, edx);
47 return eax != (uint32_t)-1 && ebx == VMWARE_HYPERVISOR_MAGIC;
48}
49
50static unsigned long __vmware_get_tsc_khz(void)
51{
52 uint64_t tsc_hz;
53 uint32_t eax, ebx, ecx, edx;
54
55 VMWARE_PORT(GETHZ, eax, ebx, ecx, edx);
56
57 if (ebx == UINT_MAX)
58 return 0;
59 tsc_hz = eax | (((uint64_t)ebx) << 32);
60 do_div(tsc_hz, 1000);
61 BUG_ON(tsc_hz >> 32);
62 return tsc_hz;
63}
64
65/*
66 * While checking the dmi string infomation, just checking the product
67 * serial key should be enough, as this will always have a VMware
68 * specific string when running under VMware hypervisor.
69 */
70int vmware_platform(void)
71{
72 if (cpu_has_hypervisor) {
73 unsigned int eax, ebx, ecx, edx;
74 char hyper_vendor_id[13];
75
76 cpuid(CPUID_VMWARE_INFO_LEAF, &eax, &ebx, &ecx, &edx);
77 memcpy(hyper_vendor_id + 0, &ebx, 4);
78 memcpy(hyper_vendor_id + 4, &ecx, 4);
79 memcpy(hyper_vendor_id + 8, &edx, 4);
80 hyper_vendor_id[12] = '\0';
81 if (!strcmp(hyper_vendor_id, "VMwareVMware"))
82 return 1;
83 } else if (dmi_available && dmi_name_in_serial("VMware") &&
84 __vmware_platform())
85 return 1;
86
87 return 0;
88}
89
90unsigned long vmware_get_tsc_khz(void)
91{
92 BUG_ON(!vmware_platform());
93 return __vmware_get_tsc_khz();
94}
95
96/*
97 * VMware hypervisor takes care of exporting a reliable TSC to the guest.
98 * Still, due to timing difference when running on virtual cpus, the TSC can
99 * be marked as unstable in some cases. For example, the TSC sync check at
100 * bootup can fail due to a marginal offset between vcpus' TSCs (though the
101 * TSCs do not drift from each other). Also, the ACPI PM timer clocksource
102 * is not suitable as a watchdog when running on a hypervisor because the
103 * kernel may miss a wrap of the counter if the vcpu is descheduled for a
104 * long time. To skip these checks at runtime we set these capability bits,
105 * so that the kernel could just trust the hypervisor with providing a
106 * reliable virtual TSC that is suitable for timekeeping.
107 */
108void __cpuinit vmware_set_feature_bits(struct cpuinfo_x86 *c)
109{
110 set_cpu_cap(c, X86_FEATURE_CONSTANT_TSC);
111 set_cpu_cap(c, X86_FEATURE_TSC_RELIABLE);
112}
diff --git a/arch/x86/kernel/crash.c b/arch/x86/kernel/crash.c
index 268553817909..d84a852e4cd7 100644
--- a/arch/x86/kernel/crash.c
+++ b/arch/x86/kernel/crash.c
@@ -29,34 +29,17 @@
29 29
30#include <mach_ipi.h> 30#include <mach_ipi.h>
31 31
32/* This keeps a track of which one is crashing cpu. */
33static int crashing_cpu;
34 32
35#if defined(CONFIG_SMP) && defined(CONFIG_X86_LOCAL_APIC) 33#if defined(CONFIG_SMP) && defined(CONFIG_X86_LOCAL_APIC)
36static atomic_t waiting_for_crash_ipi;
37 34
38static int crash_nmi_callback(struct notifier_block *self, 35static void kdump_nmi_callback(int cpu, struct die_args *args)
39 unsigned long val, void *data)
40{ 36{
41 struct pt_regs *regs; 37 struct pt_regs *regs;
42#ifdef CONFIG_X86_32 38#ifdef CONFIG_X86_32
43 struct pt_regs fixed_regs; 39 struct pt_regs fixed_regs;
44#endif 40#endif
45 int cpu;
46 41
47 if (val != DIE_NMI_IPI) 42 regs = args->regs;
48 return NOTIFY_OK;
49
50 regs = ((struct die_args *)data)->regs;
51 cpu = raw_smp_processor_id();
52
53 /* Don't do anything if this handler is invoked on crashing cpu.
54 * Otherwise, system will completely hang. Crashing cpu can get
55 * an NMI if system was initially booted with nmi_watchdog parameter.
56 */
57 if (cpu == crashing_cpu)
58 return NOTIFY_STOP;
59 local_irq_disable();
60 43
61#ifdef CONFIG_X86_32 44#ifdef CONFIG_X86_32
62 if (!user_mode_vm(regs)) { 45 if (!user_mode_vm(regs)) {
@@ -65,54 +48,19 @@ static int crash_nmi_callback(struct notifier_block *self,
65 } 48 }
66#endif 49#endif
67 crash_save_cpu(regs, cpu); 50 crash_save_cpu(regs, cpu);
68 disable_local_APIC();
69 atomic_dec(&waiting_for_crash_ipi);
70 /* Assume hlt works */
71 halt();
72 for (;;)
73 cpu_relax();
74
75 return 1;
76}
77 51
78static void smp_send_nmi_allbutself(void) 52 disable_local_APIC();
79{
80 cpumask_t mask = cpu_online_map;
81 cpu_clear(safe_smp_processor_id(), mask);
82 if (!cpus_empty(mask))
83 send_IPI_mask(mask, NMI_VECTOR);
84} 53}
85 54
86static struct notifier_block crash_nmi_nb = { 55static void kdump_nmi_shootdown_cpus(void)
87 .notifier_call = crash_nmi_callback,
88};
89
90static void nmi_shootdown_cpus(void)
91{ 56{
92 unsigned long msecs; 57 nmi_shootdown_cpus(kdump_nmi_callback);
93
94 atomic_set(&waiting_for_crash_ipi, num_online_cpus() - 1);
95 /* Would it be better to replace the trap vector here? */
96 if (register_die_notifier(&crash_nmi_nb))
97 return; /* return what? */
98 /* Ensure the new callback function is set before sending
99 * out the NMI
100 */
101 wmb();
102 58
103 smp_send_nmi_allbutself();
104
105 msecs = 1000; /* Wait at most a second for the other cpus to stop */
106 while ((atomic_read(&waiting_for_crash_ipi) > 0) && msecs) {
107 mdelay(1);
108 msecs--;
109 }
110
111 /* Leave the nmi callback set */
112 disable_local_APIC(); 59 disable_local_APIC();
113} 60}
61
114#else 62#else
115static void nmi_shootdown_cpus(void) 63static void kdump_nmi_shootdown_cpus(void)
116{ 64{
117 /* There are no cpus to shootdown */ 65 /* There are no cpus to shootdown */
118} 66}
@@ -131,9 +79,7 @@ void native_machine_crash_shutdown(struct pt_regs *regs)
131 /* The kernel is broken so disable interrupts */ 79 /* The kernel is broken so disable interrupts */
132 local_irq_disable(); 80 local_irq_disable();
133 81
134 /* Make a note of crashing cpu. Will be used in NMI callback.*/ 82 kdump_nmi_shootdown_cpus();
135 crashing_cpu = safe_smp_processor_id();
136 nmi_shootdown_cpus();
137 lapic_shutdown(); 83 lapic_shutdown();
138#if defined(CONFIG_X86_IO_APIC) 84#if defined(CONFIG_X86_IO_APIC)
139 disable_IO_APIC(); 85 disable_IO_APIC();
diff --git a/arch/x86/kernel/ds.c b/arch/x86/kernel/ds.c
index c570252905a1..d6938d9351cf 100644
--- a/arch/x86/kernel/ds.c
+++ b/arch/x86/kernel/ds.c
@@ -21,8 +21,6 @@
21 */ 21 */
22 22
23 23
24#ifdef CONFIG_X86_DS
25
26#include <asm/ds.h> 24#include <asm/ds.h>
27 25
28#include <linux/errno.h> 26#include <linux/errno.h>
@@ -211,14 +209,15 @@ static DEFINE_PER_CPU(struct ds_context *, system_context);
211static inline struct ds_context *ds_get_context(struct task_struct *task) 209static inline struct ds_context *ds_get_context(struct task_struct *task)
212{ 210{
213 struct ds_context *context; 211 struct ds_context *context;
212 unsigned long irq;
214 213
215 spin_lock(&ds_lock); 214 spin_lock_irqsave(&ds_lock, irq);
216 215
217 context = (task ? task->thread.ds_ctx : this_system_context); 216 context = (task ? task->thread.ds_ctx : this_system_context);
218 if (context) 217 if (context)
219 context->count++; 218 context->count++;
220 219
221 spin_unlock(&ds_lock); 220 spin_unlock_irqrestore(&ds_lock, irq);
222 221
223 return context; 222 return context;
224} 223}
@@ -226,18 +225,16 @@ static inline struct ds_context *ds_get_context(struct task_struct *task)
226/* 225/*
227 * Same as ds_get_context, but allocates the context and it's DS 226 * Same as ds_get_context, but allocates the context and it's DS
228 * structure, if necessary; returns NULL; if out of memory. 227 * structure, if necessary; returns NULL; if out of memory.
229 *
230 * pre: requires ds_lock to be held
231 */ 228 */
232static inline struct ds_context *ds_alloc_context(struct task_struct *task) 229static inline struct ds_context *ds_alloc_context(struct task_struct *task)
233{ 230{
234 struct ds_context **p_context = 231 struct ds_context **p_context =
235 (task ? &task->thread.ds_ctx : &this_system_context); 232 (task ? &task->thread.ds_ctx : &this_system_context);
236 struct ds_context *context = *p_context; 233 struct ds_context *context = *p_context;
234 unsigned long irq;
237 235
238 if (!context) { 236 if (!context) {
239 context = kzalloc(sizeof(*context), GFP_KERNEL); 237 context = kzalloc(sizeof(*context), GFP_KERNEL);
240
241 if (!context) 238 if (!context)
242 return NULL; 239 return NULL;
243 240
@@ -247,18 +244,27 @@ static inline struct ds_context *ds_alloc_context(struct task_struct *task)
247 return NULL; 244 return NULL;
248 } 245 }
249 246
250 *p_context = context; 247 spin_lock_irqsave(&ds_lock, irq);
251 248
252 context->this = p_context; 249 if (*p_context) {
253 context->task = task; 250 kfree(context->ds);
251 kfree(context);
252
253 context = *p_context;
254 } else {
255 *p_context = context;
254 256
255 if (task) 257 context->this = p_context;
256 set_tsk_thread_flag(task, TIF_DS_AREA_MSR); 258 context->task = task;
257 259
258 if (!task || (task == current)) 260 if (task)
259 wrmsr(MSR_IA32_DS_AREA, (unsigned long)context->ds, 0); 261 set_tsk_thread_flag(task, TIF_DS_AREA_MSR);
260 262
261 get_tracer(task); 263 if (!task || (task == current))
264 wrmsrl(MSR_IA32_DS_AREA,
265 (unsigned long)context->ds);
266 }
267 spin_unlock_irqrestore(&ds_lock, irq);
262 } 268 }
263 269
264 context->count++; 270 context->count++;
@@ -272,10 +278,12 @@ static inline struct ds_context *ds_alloc_context(struct task_struct *task)
272 */ 278 */
273static inline void ds_put_context(struct ds_context *context) 279static inline void ds_put_context(struct ds_context *context)
274{ 280{
281 unsigned long irq;
282
275 if (!context) 283 if (!context)
276 return; 284 return;
277 285
278 spin_lock(&ds_lock); 286 spin_lock_irqsave(&ds_lock, irq);
279 287
280 if (--context->count) 288 if (--context->count)
281 goto out; 289 goto out;
@@ -297,7 +305,7 @@ static inline void ds_put_context(struct ds_context *context)
297 kfree(context->ds); 305 kfree(context->ds);
298 kfree(context); 306 kfree(context);
299 out: 307 out:
300 spin_unlock(&ds_lock); 308 spin_unlock_irqrestore(&ds_lock, irq);
301} 309}
302 310
303 311
@@ -368,6 +376,7 @@ static int ds_request(struct task_struct *task, void *base, size_t size,
368 struct ds_context *context; 376 struct ds_context *context;
369 unsigned long buffer, adj; 377 unsigned long buffer, adj;
370 const unsigned long alignment = (1 << 3); 378 const unsigned long alignment = (1 << 3);
379 unsigned long irq;
371 int error = 0; 380 int error = 0;
372 381
373 if (!ds_cfg.sizeof_ds) 382 if (!ds_cfg.sizeof_ds)
@@ -382,25 +391,27 @@ static int ds_request(struct task_struct *task, void *base, size_t size,
382 return -EOPNOTSUPP; 391 return -EOPNOTSUPP;
383 392
384 393
385 spin_lock(&ds_lock);
386
387 if (!check_tracer(task))
388 return -EPERM;
389
390 error = -ENOMEM;
391 context = ds_alloc_context(task); 394 context = ds_alloc_context(task);
392 if (!context) 395 if (!context)
396 return -ENOMEM;
397
398 spin_lock_irqsave(&ds_lock, irq);
399
400 error = -EPERM;
401 if (!check_tracer(task))
393 goto out_unlock; 402 goto out_unlock;
394 403
404 get_tracer(task);
405
395 error = -EALREADY; 406 error = -EALREADY;
396 if (context->owner[qual] == current) 407 if (context->owner[qual] == current)
397 goto out_unlock; 408 goto out_put_tracer;
398 error = -EPERM; 409 error = -EPERM;
399 if (context->owner[qual] != NULL) 410 if (context->owner[qual] != NULL)
400 goto out_unlock; 411 goto out_put_tracer;
401 context->owner[qual] = current; 412 context->owner[qual] = current;
402 413
403 spin_unlock(&ds_lock); 414 spin_unlock_irqrestore(&ds_lock, irq);
404 415
405 416
406 error = -ENOMEM; 417 error = -ENOMEM;
@@ -448,10 +459,17 @@ static int ds_request(struct task_struct *task, void *base, size_t size,
448 out_release: 459 out_release:
449 context->owner[qual] = NULL; 460 context->owner[qual] = NULL;
450 ds_put_context(context); 461 ds_put_context(context);
462 put_tracer(task);
463 return error;
464
465 out_put_tracer:
466 spin_unlock_irqrestore(&ds_lock, irq);
467 ds_put_context(context);
468 put_tracer(task);
451 return error; 469 return error;
452 470
453 out_unlock: 471 out_unlock:
454 spin_unlock(&ds_lock); 472 spin_unlock_irqrestore(&ds_lock, irq);
455 ds_put_context(context); 473 ds_put_context(context);
456 return error; 474 return error;
457} 475}
@@ -801,13 +819,21 @@ static const struct ds_configuration ds_cfg_var = {
801 .sizeof_ds = sizeof(long) * 12, 819 .sizeof_ds = sizeof(long) * 12,
802 .sizeof_field = sizeof(long), 820 .sizeof_field = sizeof(long),
803 .sizeof_rec[ds_bts] = sizeof(long) * 3, 821 .sizeof_rec[ds_bts] = sizeof(long) * 3,
822#ifdef __i386__
804 .sizeof_rec[ds_pebs] = sizeof(long) * 10 823 .sizeof_rec[ds_pebs] = sizeof(long) * 10
824#else
825 .sizeof_rec[ds_pebs] = sizeof(long) * 18
826#endif
805}; 827};
806static const struct ds_configuration ds_cfg_64 = { 828static const struct ds_configuration ds_cfg_64 = {
807 .sizeof_ds = 8 * 12, 829 .sizeof_ds = 8 * 12,
808 .sizeof_field = 8, 830 .sizeof_field = 8,
809 .sizeof_rec[ds_bts] = 8 * 3, 831 .sizeof_rec[ds_bts] = 8 * 3,
832#ifdef __i386__
810 .sizeof_rec[ds_pebs] = 8 * 10 833 .sizeof_rec[ds_pebs] = 8 * 10
834#else
835 .sizeof_rec[ds_pebs] = 8 * 18
836#endif
811}; 837};
812 838
813static inline void 839static inline void
@@ -860,4 +886,3 @@ void ds_free(struct ds_context *context)
860 while (leftovers--) 886 while (leftovers--)
861 ds_put_context(context); 887 ds_put_context(context);
862} 888}
863#endif /* CONFIG_X86_DS */
diff --git a/arch/x86/kernel/dumpstack.c b/arch/x86/kernel/dumpstack.c
new file mode 100644
index 000000000000..5962176dfabb
--- /dev/null
+++ b/arch/x86/kernel/dumpstack.c
@@ -0,0 +1,319 @@
1/*
2 * Copyright (C) 1991, 1992 Linus Torvalds
3 * Copyright (C) 2000, 2001, 2002 Andi Kleen, SuSE Labs
4 */
5#include <linux/kallsyms.h>
6#include <linux/kprobes.h>
7#include <linux/uaccess.h>
8#include <linux/utsname.h>
9#include <linux/hardirq.h>
10#include <linux/kdebug.h>
11#include <linux/module.h>
12#include <linux/ptrace.h>
13#include <linux/kexec.h>
14#include <linux/bug.h>
15#include <linux/nmi.h>
16#include <linux/sysfs.h>
17
18#include <asm/stacktrace.h>
19
20#include "dumpstack.h"
21
22int panic_on_unrecovered_nmi;
23unsigned int code_bytes = 64;
24int kstack_depth_to_print = 3 * STACKSLOTS_PER_LINE;
25static int die_counter;
26
27void printk_address(unsigned long address, int reliable)
28{
29 printk(" [<%p>] %s%pS\n", (void *) address,
30 reliable ? "" : "? ", (void *) address);
31}
32
33/*
34 * x86-64 can have up to three kernel stacks:
35 * process stack
36 * interrupt stack
37 * severe exception (double fault, nmi, stack fault, debug, mce) hardware stack
38 */
39
40static inline int valid_stack_ptr(struct thread_info *tinfo,
41 void *p, unsigned int size, void *end)
42{
43 void *t = tinfo;
44 if (end) {
45 if (p < end && p >= (end-THREAD_SIZE))
46 return 1;
47 else
48 return 0;
49 }
50 return p > t && p < t + THREAD_SIZE - size;
51}
52
53unsigned long
54print_context_stack(struct thread_info *tinfo,
55 unsigned long *stack, unsigned long bp,
56 const struct stacktrace_ops *ops, void *data,
57 unsigned long *end)
58{
59 struct stack_frame *frame = (struct stack_frame *)bp;
60
61 while (valid_stack_ptr(tinfo, stack, sizeof(*stack), end)) {
62 unsigned long addr;
63
64 addr = *stack;
65 if (__kernel_text_address(addr)) {
66 if ((unsigned long) stack == bp + sizeof(long)) {
67 ops->address(data, addr, 1);
68 frame = frame->next_frame;
69 bp = (unsigned long) frame;
70 } else {
71 ops->address(data, addr, bp == 0);
72 }
73 }
74 stack++;
75 }
76 return bp;
77}
78
79
80static void
81print_trace_warning_symbol(void *data, char *msg, unsigned long symbol)
82{
83 printk(data);
84 print_symbol(msg, symbol);
85 printk("\n");
86}
87
88static void print_trace_warning(void *data, char *msg)
89{
90 printk("%s%s\n", (char *)data, msg);
91}
92
93static int print_trace_stack(void *data, char *name)
94{
95 printk("%s <%s> ", (char *)data, name);
96 return 0;
97}
98
99/*
100 * Print one address/symbol entries per line.
101 */
102static void print_trace_address(void *data, unsigned long addr, int reliable)
103{
104 touch_nmi_watchdog();
105 printk(data);
106 printk_address(addr, reliable);
107}
108
109static const struct stacktrace_ops print_trace_ops = {
110 .warning = print_trace_warning,
111 .warning_symbol = print_trace_warning_symbol,
112 .stack = print_trace_stack,
113 .address = print_trace_address,
114};
115
116void
117show_trace_log_lvl(struct task_struct *task, struct pt_regs *regs,
118 unsigned long *stack, unsigned long bp, char *log_lvl)
119{
120 printk("%sCall Trace:\n", log_lvl);
121 dump_trace(task, regs, stack, bp, &print_trace_ops, log_lvl);
122}
123
124void show_trace(struct task_struct *task, struct pt_regs *regs,
125 unsigned long *stack, unsigned long bp)
126{
127 show_trace_log_lvl(task, regs, stack, bp, "");
128}
129
130void show_stack(struct task_struct *task, unsigned long *sp)
131{
132 show_stack_log_lvl(task, NULL, sp, 0, "");
133}
134
135/*
136 * The architecture-independent dump_stack generator
137 */
138void dump_stack(void)
139{
140 unsigned long bp = 0;
141 unsigned long stack;
142
143#ifdef CONFIG_FRAME_POINTER
144 if (!bp)
145 get_bp(bp);
146#endif
147
148 printk("Pid: %d, comm: %.20s %s %s %.*s\n",
149 current->pid, current->comm, print_tainted(),
150 init_utsname()->release,
151 (int)strcspn(init_utsname()->version, " "),
152 init_utsname()->version);
153 show_trace(NULL, NULL, &stack, bp);
154}
155EXPORT_SYMBOL(dump_stack);
156
157static raw_spinlock_t die_lock = __RAW_SPIN_LOCK_UNLOCKED;
158static int die_owner = -1;
159static unsigned int die_nest_count;
160
161unsigned __kprobes long oops_begin(void)
162{
163 int cpu;
164 unsigned long flags;
165
166 oops_enter();
167
168 /* racy, but better than risking deadlock. */
169 raw_local_irq_save(flags);
170 cpu = smp_processor_id();
171 if (!__raw_spin_trylock(&die_lock)) {
172 if (cpu == die_owner)
173 /* nested oops. should stop eventually */;
174 else
175 __raw_spin_lock(&die_lock);
176 }
177 die_nest_count++;
178 die_owner = cpu;
179 console_verbose();
180 bust_spinlocks(1);
181 return flags;
182}
183
184void __kprobes oops_end(unsigned long flags, struct pt_regs *regs, int signr)
185{
186 if (regs && kexec_should_crash(current))
187 crash_kexec(regs);
188
189 bust_spinlocks(0);
190 die_owner = -1;
191 add_taint(TAINT_DIE);
192 die_nest_count--;
193 if (!die_nest_count)
194 /* Nest count reaches zero, release the lock. */
195 __raw_spin_unlock(&die_lock);
196 raw_local_irq_restore(flags);
197 oops_exit();
198
199 if (!signr)
200 return;
201 if (in_interrupt())
202 panic("Fatal exception in interrupt");
203 if (panic_on_oops)
204 panic("Fatal exception");
205 do_exit(signr);
206}
207
208int __kprobes __die(const char *str, struct pt_regs *regs, long err)
209{
210#ifdef CONFIG_X86_32
211 unsigned short ss;
212 unsigned long sp;
213#endif
214 printk(KERN_EMERG "%s: %04lx [#%d] ", str, err & 0xffff, ++die_counter);
215#ifdef CONFIG_PREEMPT
216 printk("PREEMPT ");
217#endif
218#ifdef CONFIG_SMP
219 printk("SMP ");
220#endif
221#ifdef CONFIG_DEBUG_PAGEALLOC
222 printk("DEBUG_PAGEALLOC");
223#endif
224 printk("\n");
225 sysfs_printk_last_file();
226 if (notify_die(DIE_OOPS, str, regs, err,
227 current->thread.trap_no, SIGSEGV) == NOTIFY_STOP)
228 return 1;
229
230 show_registers(regs);
231#ifdef CONFIG_X86_32
232 sp = (unsigned long) (&regs->sp);
233 savesegment(ss, ss);
234 if (user_mode(regs)) {
235 sp = regs->sp;
236 ss = regs->ss & 0xffff;
237 }
238 printk(KERN_EMERG "EIP: [<%08lx>] ", regs->ip);
239 print_symbol("%s", regs->ip);
240 printk(" SS:ESP %04x:%08lx\n", ss, sp);
241#else
242 /* Executive summary in case the oops scrolled away */
243 printk(KERN_ALERT "RIP ");
244 printk_address(regs->ip, 1);
245 printk(" RSP <%016lx>\n", regs->sp);
246#endif
247 return 0;
248}
249
250/*
251 * This is gone through when something in the kernel has done something bad
252 * and is about to be terminated:
253 */
254void die(const char *str, struct pt_regs *regs, long err)
255{
256 unsigned long flags = oops_begin();
257 int sig = SIGSEGV;
258
259 if (!user_mode_vm(regs))
260 report_bug(regs->ip, regs);
261
262 if (__die(str, regs, err))
263 sig = 0;
264 oops_end(flags, regs, sig);
265}
266
267void notrace __kprobes
268die_nmi(char *str, struct pt_regs *regs, int do_panic)
269{
270 unsigned long flags;
271
272 if (notify_die(DIE_NMIWATCHDOG, str, regs, 0, 2, SIGINT) == NOTIFY_STOP)
273 return;
274
275 /*
276 * We are in trouble anyway, lets at least try
277 * to get a message out.
278 */
279 flags = oops_begin();
280 printk(KERN_EMERG "%s", str);
281 printk(" on CPU%d, ip %08lx, registers:\n",
282 smp_processor_id(), regs->ip);
283 show_registers(regs);
284 oops_end(flags, regs, 0);
285 if (do_panic || panic_on_oops)
286 panic("Non maskable interrupt");
287 nmi_exit();
288 local_irq_enable();
289 do_exit(SIGBUS);
290}
291
292static int __init oops_setup(char *s)
293{
294 if (!s)
295 return -EINVAL;
296 if (!strcmp(s, "panic"))
297 panic_on_oops = 1;
298 return 0;
299}
300early_param("oops", oops_setup);
301
302static int __init kstack_setup(char *s)
303{
304 if (!s)
305 return -EINVAL;
306 kstack_depth_to_print = simple_strtoul(s, NULL, 0);
307 return 0;
308}
309early_param("kstack", kstack_setup);
310
311static int __init code_bytes_setup(char *s)
312{
313 code_bytes = simple_strtoul(s, NULL, 0);
314 if (code_bytes > 8192)
315 code_bytes = 8192;
316
317 return 1;
318}
319__setup("code_bytes=", code_bytes_setup);
diff --git a/arch/x86/kernel/dumpstack.h b/arch/x86/kernel/dumpstack.h
new file mode 100644
index 000000000000..3119a801c32b
--- /dev/null
+++ b/arch/x86/kernel/dumpstack.h
@@ -0,0 +1,39 @@
1/*
2 * Copyright (C) 1991, 1992 Linus Torvalds
3 * Copyright (C) 2000, 2001, 2002 Andi Kleen, SuSE Labs
4 */
5
6#ifndef DUMPSTACK_H
7#define DUMPSTACK_H
8
9#ifdef CONFIG_X86_32
10#define STACKSLOTS_PER_LINE 8
11#define get_bp(bp) asm("movl %%ebp, %0" : "=r" (bp) :)
12#else
13#define STACKSLOTS_PER_LINE 4
14#define get_bp(bp) asm("movq %%rbp, %0" : "=r" (bp) :)
15#endif
16
17extern unsigned long
18print_context_stack(struct thread_info *tinfo,
19 unsigned long *stack, unsigned long bp,
20 const struct stacktrace_ops *ops, void *data,
21 unsigned long *end);
22
23extern void
24show_trace_log_lvl(struct task_struct *task, struct pt_regs *regs,
25 unsigned long *stack, unsigned long bp, char *log_lvl);
26
27extern void
28show_stack_log_lvl(struct task_struct *task, struct pt_regs *regs,
29 unsigned long *sp, unsigned long bp, char *log_lvl);
30
31extern unsigned int code_bytes;
32extern int kstack_depth_to_print;
33
34/* The form of the top of the frame on the stack */
35struct stack_frame {
36 struct stack_frame *next_frame;
37 unsigned long return_address;
38};
39#endif
diff --git a/arch/x86/kernel/dumpstack_32.c b/arch/x86/kernel/dumpstack_32.c
index b3614752197b..7b031b106ec8 100644
--- a/arch/x86/kernel/dumpstack_32.c
+++ b/arch/x86/kernel/dumpstack_32.c
@@ -17,64 +17,7 @@
17 17
18#include <asm/stacktrace.h> 18#include <asm/stacktrace.h>
19 19
20#define STACKSLOTS_PER_LINE 8 20#include "dumpstack.h"
21#define get_bp(bp) asm("movl %%ebp, %0" : "=r" (bp) :)
22
23int panic_on_unrecovered_nmi;
24int kstack_depth_to_print = 3 * STACKSLOTS_PER_LINE;
25static unsigned int code_bytes = 64;
26static int die_counter;
27
28void printk_address(unsigned long address, int reliable)
29{
30 printk(" [<%p>] %s%pS\n", (void *) address,
31 reliable ? "" : "? ", (void *) address);
32}
33
34static inline int valid_stack_ptr(struct thread_info *tinfo,
35 void *p, unsigned int size, void *end)
36{
37 void *t = tinfo;
38 if (end) {
39 if (p < end && p >= (end-THREAD_SIZE))
40 return 1;
41 else
42 return 0;
43 }
44 return p > t && p < t + THREAD_SIZE - size;
45}
46
47/* The form of the top of the frame on the stack */
48struct stack_frame {
49 struct stack_frame *next_frame;
50 unsigned long return_address;
51};
52
53static inline unsigned long
54print_context_stack(struct thread_info *tinfo,
55 unsigned long *stack, unsigned long bp,
56 const struct stacktrace_ops *ops, void *data,
57 unsigned long *end)
58{
59 struct stack_frame *frame = (struct stack_frame *)bp;
60
61 while (valid_stack_ptr(tinfo, stack, sizeof(*stack), end)) {
62 unsigned long addr;
63
64 addr = *stack;
65 if (__kernel_text_address(addr)) {
66 if ((unsigned long) stack == bp + sizeof(long)) {
67 ops->address(data, addr, 1);
68 frame = frame->next_frame;
69 bp = (unsigned long) frame;
70 } else {
71 ops->address(data, addr, bp == 0);
72 }
73 }
74 stack++;
75 }
76 return bp;
77}
78 21
79void dump_trace(struct task_struct *task, struct pt_regs *regs, 22void dump_trace(struct task_struct *task, struct pt_regs *regs,
80 unsigned long *stack, unsigned long bp, 23 unsigned long *stack, unsigned long bp,
@@ -119,57 +62,7 @@ void dump_trace(struct task_struct *task, struct pt_regs *regs,
119} 62}
120EXPORT_SYMBOL(dump_trace); 63EXPORT_SYMBOL(dump_trace);
121 64
122static void 65void
123print_trace_warning_symbol(void *data, char *msg, unsigned long symbol)
124{
125 printk(data);
126 print_symbol(msg, symbol);
127 printk("\n");
128}
129
130static void print_trace_warning(void *data, char *msg)
131{
132 printk("%s%s\n", (char *)data, msg);
133}
134
135static int print_trace_stack(void *data, char *name)
136{
137 printk("%s <%s> ", (char *)data, name);
138 return 0;
139}
140
141/*
142 * Print one address/symbol entries per line.
143 */
144static void print_trace_address(void *data, unsigned long addr, int reliable)
145{
146 touch_nmi_watchdog();
147 printk(data);
148 printk_address(addr, reliable);
149}
150
151static const struct stacktrace_ops print_trace_ops = {
152 .warning = print_trace_warning,
153 .warning_symbol = print_trace_warning_symbol,
154 .stack = print_trace_stack,
155 .address = print_trace_address,
156};
157
158static void
159show_trace_log_lvl(struct task_struct *task, struct pt_regs *regs,
160 unsigned long *stack, unsigned long bp, char *log_lvl)
161{
162 printk("%sCall Trace:\n", log_lvl);
163 dump_trace(task, regs, stack, bp, &print_trace_ops, log_lvl);
164}
165
166void show_trace(struct task_struct *task, struct pt_regs *regs,
167 unsigned long *stack, unsigned long bp)
168{
169 show_trace_log_lvl(task, regs, stack, bp, "");
170}
171
172static void
173show_stack_log_lvl(struct task_struct *task, struct pt_regs *regs, 66show_stack_log_lvl(struct task_struct *task, struct pt_regs *regs,
174 unsigned long *sp, unsigned long bp, char *log_lvl) 67 unsigned long *sp, unsigned long bp, char *log_lvl)
175{ 68{
@@ -196,33 +89,6 @@ show_stack_log_lvl(struct task_struct *task, struct pt_regs *regs,
196 show_trace_log_lvl(task, regs, sp, bp, log_lvl); 89 show_trace_log_lvl(task, regs, sp, bp, log_lvl);
197} 90}
198 91
199void show_stack(struct task_struct *task, unsigned long *sp)
200{
201 show_stack_log_lvl(task, NULL, sp, 0, "");
202}
203
204/*
205 * The architecture-independent dump_stack generator
206 */
207void dump_stack(void)
208{
209 unsigned long bp = 0;
210 unsigned long stack;
211
212#ifdef CONFIG_FRAME_POINTER
213 if (!bp)
214 get_bp(bp);
215#endif
216
217 printk("Pid: %d, comm: %.20s %s %s %.*s\n",
218 current->pid, current->comm, print_tainted(),
219 init_utsname()->release,
220 (int)strcspn(init_utsname()->version, " "),
221 init_utsname()->version);
222 show_trace(NULL, NULL, &stack, bp);
223}
224
225EXPORT_SYMBOL(dump_stack);
226 92
227void show_registers(struct pt_regs *regs) 93void show_registers(struct pt_regs *regs)
228{ 94{
@@ -283,167 +149,3 @@ int is_valid_bugaddr(unsigned long ip)
283 return ud2 == 0x0b0f; 149 return ud2 == 0x0b0f;
284} 150}
285 151
286static raw_spinlock_t die_lock = __RAW_SPIN_LOCK_UNLOCKED;
287static int die_owner = -1;
288static unsigned int die_nest_count;
289
290unsigned __kprobes long oops_begin(void)
291{
292 unsigned long flags;
293
294 oops_enter();
295
296 if (die_owner != raw_smp_processor_id()) {
297 console_verbose();
298 raw_local_irq_save(flags);
299 __raw_spin_lock(&die_lock);
300 die_owner = smp_processor_id();
301 die_nest_count = 0;
302 bust_spinlocks(1);
303 } else {
304 raw_local_irq_save(flags);
305 }
306 die_nest_count++;
307 return flags;
308}
309
310void __kprobes oops_end(unsigned long flags, struct pt_regs *regs, int signr)
311{
312 bust_spinlocks(0);
313 die_owner = -1;
314 add_taint(TAINT_DIE);
315 __raw_spin_unlock(&die_lock);
316 raw_local_irq_restore(flags);
317
318 if (!regs)
319 return;
320
321 if (kexec_should_crash(current))
322 crash_kexec(regs);
323 if (in_interrupt())
324 panic("Fatal exception in interrupt");
325 if (panic_on_oops)
326 panic("Fatal exception");
327 oops_exit();
328 do_exit(signr);
329}
330
331int __kprobes __die(const char *str, struct pt_regs *regs, long err)
332{
333 unsigned short ss;
334 unsigned long sp;
335
336 printk(KERN_EMERG "%s: %04lx [#%d] ", str, err & 0xffff, ++die_counter);
337#ifdef CONFIG_PREEMPT
338 printk("PREEMPT ");
339#endif
340#ifdef CONFIG_SMP
341 printk("SMP ");
342#endif
343#ifdef CONFIG_DEBUG_PAGEALLOC
344 printk("DEBUG_PAGEALLOC");
345#endif
346 printk("\n");
347 sysfs_printk_last_file();
348 if (notify_die(DIE_OOPS, str, regs, err,
349 current->thread.trap_no, SIGSEGV) == NOTIFY_STOP)
350 return 1;
351
352 show_registers(regs);
353 /* Executive summary in case the oops scrolled away */
354 sp = (unsigned long) (&regs->sp);
355 savesegment(ss, ss);
356 if (user_mode(regs)) {
357 sp = regs->sp;
358 ss = regs->ss & 0xffff;
359 }
360 printk(KERN_EMERG "EIP: [<%08lx>] ", regs->ip);
361 print_symbol("%s", regs->ip);
362 printk(" SS:ESP %04x:%08lx\n", ss, sp);
363 return 0;
364}
365
366/*
367 * This is gone through when something in the kernel has done something bad
368 * and is about to be terminated:
369 */
370void die(const char *str, struct pt_regs *regs, long err)
371{
372 unsigned long flags = oops_begin();
373
374 if (die_nest_count < 3) {
375 report_bug(regs->ip, regs);
376
377 if (__die(str, regs, err))
378 regs = NULL;
379 } else {
380 printk(KERN_EMERG "Recursive die() failure, output suppressed\n");
381 }
382
383 oops_end(flags, regs, SIGSEGV);
384}
385
386static DEFINE_SPINLOCK(nmi_print_lock);
387
388void notrace __kprobes
389die_nmi(char *str, struct pt_regs *regs, int do_panic)
390{
391 if (notify_die(DIE_NMIWATCHDOG, str, regs, 0, 2, SIGINT) == NOTIFY_STOP)
392 return;
393
394 spin_lock(&nmi_print_lock);
395 /*
396 * We are in trouble anyway, lets at least try
397 * to get a message out:
398 */
399 bust_spinlocks(1);
400 printk(KERN_EMERG "%s", str);
401 printk(" on CPU%d, ip %08lx, registers:\n",
402 smp_processor_id(), regs->ip);
403 show_registers(regs);
404 if (do_panic)
405 panic("Non maskable interrupt");
406 console_silent();
407 spin_unlock(&nmi_print_lock);
408
409 /*
410 * If we are in kernel we are probably nested up pretty bad
411 * and might aswell get out now while we still can:
412 */
413 if (!user_mode_vm(regs)) {
414 current->thread.trap_no = 2;
415 crash_kexec(regs);
416 }
417
418 bust_spinlocks(0);
419 do_exit(SIGSEGV);
420}
421
422static int __init oops_setup(char *s)
423{
424 if (!s)
425 return -EINVAL;
426 if (!strcmp(s, "panic"))
427 panic_on_oops = 1;
428 return 0;
429}
430early_param("oops", oops_setup);
431
432static int __init kstack_setup(char *s)
433{
434 if (!s)
435 return -EINVAL;
436 kstack_depth_to_print = simple_strtoul(s, NULL, 0);
437 return 0;
438}
439early_param("kstack", kstack_setup);
440
441static int __init code_bytes_setup(char *s)
442{
443 code_bytes = simple_strtoul(s, NULL, 0);
444 if (code_bytes > 8192)
445 code_bytes = 8192;
446
447 return 1;
448}
449__setup("code_bytes=", code_bytes_setup);
diff --git a/arch/x86/kernel/dumpstack_64.c b/arch/x86/kernel/dumpstack_64.c
index 96a5db7da8a7..33ff10287a5d 100644
--- a/arch/x86/kernel/dumpstack_64.c
+++ b/arch/x86/kernel/dumpstack_64.c
@@ -17,19 +17,7 @@
17 17
18#include <asm/stacktrace.h> 18#include <asm/stacktrace.h>
19 19
20#define STACKSLOTS_PER_LINE 4 20#include "dumpstack.h"
21#define get_bp(bp) asm("movq %%rbp, %0" : "=r" (bp) :)
22
23int panic_on_unrecovered_nmi;
24int kstack_depth_to_print = 3 * STACKSLOTS_PER_LINE;
25static unsigned int code_bytes = 64;
26static int die_counter;
27
28void printk_address(unsigned long address, int reliable)
29{
30 printk(" [<%p>] %s%pS\n", (void *) address,
31 reliable ? "" : "? ", (void *) address);
32}
33 21
34static unsigned long *in_exception_stack(unsigned cpu, unsigned long stack, 22static unsigned long *in_exception_stack(unsigned cpu, unsigned long stack,
35 unsigned *usedp, char **idp) 23 unsigned *usedp, char **idp)
@@ -113,51 +101,6 @@ static unsigned long *in_exception_stack(unsigned cpu, unsigned long stack,
113 * severe exception (double fault, nmi, stack fault, debug, mce) hardware stack 101 * severe exception (double fault, nmi, stack fault, debug, mce) hardware stack
114 */ 102 */
115 103
116static inline int valid_stack_ptr(struct thread_info *tinfo,
117 void *p, unsigned int size, void *end)
118{
119 void *t = tinfo;
120 if (end) {
121 if (p < end && p >= (end-THREAD_SIZE))
122 return 1;
123 else
124 return 0;
125 }
126 return p > t && p < t + THREAD_SIZE - size;
127}
128
129/* The form of the top of the frame on the stack */
130struct stack_frame {
131 struct stack_frame *next_frame;
132 unsigned long return_address;
133};
134
135static inline unsigned long
136print_context_stack(struct thread_info *tinfo,
137 unsigned long *stack, unsigned long bp,
138 const struct stacktrace_ops *ops, void *data,
139 unsigned long *end)
140{
141 struct stack_frame *frame = (struct stack_frame *)bp;
142
143 while (valid_stack_ptr(tinfo, stack, sizeof(*stack), end)) {
144 unsigned long addr;
145
146 addr = *stack;
147 if (__kernel_text_address(addr)) {
148 if ((unsigned long) stack == bp + sizeof(long)) {
149 ops->address(data, addr, 1);
150 frame = frame->next_frame;
151 bp = (unsigned long) frame;
152 } else {
153 ops->address(data, addr, bp == 0);
154 }
155 }
156 stack++;
157 }
158 return bp;
159}
160
161void dump_trace(struct task_struct *task, struct pt_regs *regs, 104void dump_trace(struct task_struct *task, struct pt_regs *regs,
162 unsigned long *stack, unsigned long bp, 105 unsigned long *stack, unsigned long bp,
163 const struct stacktrace_ops *ops, void *data) 106 const struct stacktrace_ops *ops, void *data)
@@ -248,57 +191,7 @@ void dump_trace(struct task_struct *task, struct pt_regs *regs,
248} 191}
249EXPORT_SYMBOL(dump_trace); 192EXPORT_SYMBOL(dump_trace);
250 193
251static void 194void
252print_trace_warning_symbol(void *data, char *msg, unsigned long symbol)
253{
254 printk(data);
255 print_symbol(msg, symbol);
256 printk("\n");
257}
258
259static void print_trace_warning(void *data, char *msg)
260{
261 printk("%s%s\n", (char *)data, msg);
262}
263
264static int print_trace_stack(void *data, char *name)
265{
266 printk("%s <%s> ", (char *)data, name);
267 return 0;
268}
269
270/*
271 * Print one address/symbol entries per line.
272 */
273static void print_trace_address(void *data, unsigned long addr, int reliable)
274{
275 touch_nmi_watchdog();
276 printk(data);
277 printk_address(addr, reliable);
278}
279
280static const struct stacktrace_ops print_trace_ops = {
281 .warning = print_trace_warning,
282 .warning_symbol = print_trace_warning_symbol,
283 .stack = print_trace_stack,
284 .address = print_trace_address,
285};
286
287static void
288show_trace_log_lvl(struct task_struct *task, struct pt_regs *regs,
289 unsigned long *stack, unsigned long bp, char *log_lvl)
290{
291 printk("%sCall Trace:\n", log_lvl);
292 dump_trace(task, regs, stack, bp, &print_trace_ops, log_lvl);
293}
294
295void show_trace(struct task_struct *task, struct pt_regs *regs,
296 unsigned long *stack, unsigned long bp)
297{
298 show_trace_log_lvl(task, regs, stack, bp, "");
299}
300
301static void
302show_stack_log_lvl(struct task_struct *task, struct pt_regs *regs, 195show_stack_log_lvl(struct task_struct *task, struct pt_regs *regs,
303 unsigned long *sp, unsigned long bp, char *log_lvl) 196 unsigned long *sp, unsigned long bp, char *log_lvl)
304{ 197{
@@ -342,33 +235,6 @@ show_stack_log_lvl(struct task_struct *task, struct pt_regs *regs,
342 show_trace_log_lvl(task, regs, sp, bp, log_lvl); 235 show_trace_log_lvl(task, regs, sp, bp, log_lvl);
343} 236}
344 237
345void show_stack(struct task_struct *task, unsigned long *sp)
346{
347 show_stack_log_lvl(task, NULL, sp, 0, "");
348}
349
350/*
351 * The architecture-independent dump_stack generator
352 */
353void dump_stack(void)
354{
355 unsigned long bp = 0;
356 unsigned long stack;
357
358#ifdef CONFIG_FRAME_POINTER
359 if (!bp)
360 get_bp(bp);
361#endif
362
363 printk("Pid: %d, comm: %.20s %s %s %.*s\n",
364 current->pid, current->comm, print_tainted(),
365 init_utsname()->release,
366 (int)strcspn(init_utsname()->version, " "),
367 init_utsname()->version);
368 show_trace(NULL, NULL, &stack, bp);
369}
370EXPORT_SYMBOL(dump_stack);
371
372void show_registers(struct pt_regs *regs) 238void show_registers(struct pt_regs *regs)
373{ 239{
374 int i; 240 int i;
@@ -429,147 +295,3 @@ int is_valid_bugaddr(unsigned long ip)
429 return ud2 == 0x0b0f; 295 return ud2 == 0x0b0f;
430} 296}
431 297
432static raw_spinlock_t die_lock = __RAW_SPIN_LOCK_UNLOCKED;
433static int die_owner = -1;
434static unsigned int die_nest_count;
435
436unsigned __kprobes long oops_begin(void)
437{
438 int cpu;
439 unsigned long flags;
440
441 oops_enter();
442
443 /* racy, but better than risking deadlock. */
444 raw_local_irq_save(flags);
445 cpu = smp_processor_id();
446 if (!__raw_spin_trylock(&die_lock)) {
447 if (cpu == die_owner)
448 /* nested oops. should stop eventually */;
449 else
450 __raw_spin_lock(&die_lock);
451 }
452 die_nest_count++;
453 die_owner = cpu;
454 console_verbose();
455 bust_spinlocks(1);
456 return flags;
457}
458
459void __kprobes oops_end(unsigned long flags, struct pt_regs *regs, int signr)
460{
461 die_owner = -1;
462 bust_spinlocks(0);
463 die_nest_count--;
464 if (!die_nest_count)
465 /* Nest count reaches zero, release the lock. */
466 __raw_spin_unlock(&die_lock);
467 raw_local_irq_restore(flags);
468 if (!regs) {
469 oops_exit();
470 return;
471 }
472 if (in_interrupt())
473 panic("Fatal exception in interrupt");
474 if (panic_on_oops)
475 panic("Fatal exception");
476 oops_exit();
477 do_exit(signr);
478}
479
480int __kprobes __die(const char *str, struct pt_regs *regs, long err)
481{
482 printk(KERN_EMERG "%s: %04lx [#%d] ", str, err & 0xffff, ++die_counter);
483#ifdef CONFIG_PREEMPT
484 printk("PREEMPT ");
485#endif
486#ifdef CONFIG_SMP
487 printk("SMP ");
488#endif
489#ifdef CONFIG_DEBUG_PAGEALLOC
490 printk("DEBUG_PAGEALLOC");
491#endif
492 printk("\n");
493 sysfs_printk_last_file();
494 if (notify_die(DIE_OOPS, str, regs, err,
495 current->thread.trap_no, SIGSEGV) == NOTIFY_STOP)
496 return 1;
497
498 show_registers(regs);
499 add_taint(TAINT_DIE);
500 /* Executive summary in case the oops scrolled away */
501 printk(KERN_ALERT "RIP ");
502 printk_address(regs->ip, 1);
503 printk(" RSP <%016lx>\n", regs->sp);
504 if (kexec_should_crash(current))
505 crash_kexec(regs);
506 return 0;
507}
508
509void die(const char *str, struct pt_regs *regs, long err)
510{
511 unsigned long flags = oops_begin();
512
513 if (!user_mode(regs))
514 report_bug(regs->ip, regs);
515
516 if (__die(str, regs, err))
517 regs = NULL;
518 oops_end(flags, regs, SIGSEGV);
519}
520
521notrace __kprobes void
522die_nmi(char *str, struct pt_regs *regs, int do_panic)
523{
524 unsigned long flags;
525
526 if (notify_die(DIE_NMIWATCHDOG, str, regs, 0, 2, SIGINT) == NOTIFY_STOP)
527 return;
528
529 flags = oops_begin();
530 /*
531 * We are in trouble anyway, lets at least try
532 * to get a message out.
533 */
534 printk(KERN_EMERG "%s", str);
535 printk(" on CPU%d, ip %08lx, registers:\n",
536 smp_processor_id(), regs->ip);
537 show_registers(regs);
538 if (kexec_should_crash(current))
539 crash_kexec(regs);
540 if (do_panic || panic_on_oops)
541 panic("Non maskable interrupt");
542 oops_end(flags, NULL, SIGBUS);
543 nmi_exit();
544 local_irq_enable();
545 do_exit(SIGBUS);
546}
547
548static int __init oops_setup(char *s)
549{
550 if (!s)
551 return -EINVAL;
552 if (!strcmp(s, "panic"))
553 panic_on_oops = 1;
554 return 0;
555}
556early_param("oops", oops_setup);
557
558static int __init kstack_setup(char *s)
559{
560 if (!s)
561 return -EINVAL;
562 kstack_depth_to_print = simple_strtoul(s, NULL, 0);
563 return 0;
564}
565early_param("kstack", kstack_setup);
566
567static int __init code_bytes_setup(char *s)
568{
569 code_bytes = simple_strtoul(s, NULL, 0);
570 if (code_bytes > 8192)
571 code_bytes = 8192;
572
573 return 1;
574}
575__setup("code_bytes=", code_bytes_setup);
diff --git a/arch/x86/kernel/e820.c b/arch/x86/kernel/e820.c
index 7aafeb5263ef..65a13943e098 100644
--- a/arch/x86/kernel/e820.c
+++ b/arch/x86/kernel/e820.c
@@ -677,22 +677,6 @@ struct early_res {
677}; 677};
678static struct early_res early_res[MAX_EARLY_RES] __initdata = { 678static struct early_res early_res[MAX_EARLY_RES] __initdata = {
679 { 0, PAGE_SIZE, "BIOS data page" }, /* BIOS data page */ 679 { 0, PAGE_SIZE, "BIOS data page" }, /* BIOS data page */
680#if defined(CONFIG_X86_64) && defined(CONFIG_X86_TRAMPOLINE)
681 { TRAMPOLINE_BASE, TRAMPOLINE_BASE + 2 * PAGE_SIZE, "TRAMPOLINE" },
682#endif
683#if defined(CONFIG_X86_32) && defined(CONFIG_SMP)
684 /*
685 * But first pinch a few for the stack/trampoline stuff
686 * FIXME: Don't need the extra page at 4K, but need to fix
687 * trampoline before removing it. (see the GDT stuff)
688 */
689 { PAGE_SIZE, PAGE_SIZE + PAGE_SIZE, "EX TRAMPOLINE" },
690 /*
691 * Has to be in very low memory so we can execute
692 * real-mode AP code.
693 */
694 { TRAMPOLINE_BASE, TRAMPOLINE_BASE + PAGE_SIZE, "TRAMPOLINE" },
695#endif
696 {} 680 {}
697}; 681};
698 682
diff --git a/arch/x86/kernel/early-quirks.c b/arch/x86/kernel/early-quirks.c
index 3ce029ffaa55..1b894b72c0f5 100644
--- a/arch/x86/kernel/early-quirks.c
+++ b/arch/x86/kernel/early-quirks.c
@@ -188,20 +188,6 @@ static void __init ati_bugs_contd(int num, int slot, int func)
188} 188}
189#endif 189#endif
190 190
191#ifdef CONFIG_DMAR
192static void __init intel_g33_dmar(int num, int slot, int func)
193{
194 struct acpi_table_header *dmar_tbl;
195 acpi_status status;
196
197 status = acpi_get_table(ACPI_SIG_DMAR, 0, &dmar_tbl);
198 if (ACPI_SUCCESS(status)) {
199 printk(KERN_INFO "BIOS BUG: DMAR advertised on Intel G31/G33 chipset -- ignoring\n");
200 dmar_disabled = 1;
201 }
202}
203#endif
204
205#define QFLAG_APPLY_ONCE 0x1 191#define QFLAG_APPLY_ONCE 0x1
206#define QFLAG_APPLIED 0x2 192#define QFLAG_APPLIED 0x2
207#define QFLAG_DONE (QFLAG_APPLY_ONCE|QFLAG_APPLIED) 193#define QFLAG_DONE (QFLAG_APPLY_ONCE|QFLAG_APPLIED)
@@ -225,10 +211,6 @@ static struct chipset early_qrk[] __initdata = {
225 PCI_CLASS_SERIAL_SMBUS, PCI_ANY_ID, 0, ati_bugs }, 211 PCI_CLASS_SERIAL_SMBUS, PCI_ANY_ID, 0, ati_bugs },
226 { PCI_VENDOR_ID_ATI, PCI_DEVICE_ID_ATI_SBX00_SMBUS, 212 { PCI_VENDOR_ID_ATI, PCI_DEVICE_ID_ATI_SBX00_SMBUS,
227 PCI_CLASS_SERIAL_SMBUS, PCI_ANY_ID, 0, ati_bugs_contd }, 213 PCI_CLASS_SERIAL_SMBUS, PCI_ANY_ID, 0, ati_bugs_contd },
228#ifdef CONFIG_DMAR
229 { PCI_VENDOR_ID_INTEL, 0x29c0,
230 PCI_CLASS_BRIDGE_HOST, PCI_ANY_ID, 0, intel_g33_dmar },
231#endif
232 {} 214 {}
233}; 215};
234 216
diff --git a/arch/x86/kernel/early_printk.c b/arch/x86/kernel/early_printk.c
index 34ad997d3834..23b138e31e9c 100644
--- a/arch/x86/kernel/early_printk.c
+++ b/arch/x86/kernel/early_printk.c
@@ -875,49 +875,6 @@ static struct console early_dbgp_console = {
875}; 875};
876#endif 876#endif
877 877
878/* Console interface to a host file on AMD's SimNow! */
879
880static int simnow_fd;
881
882enum {
883 MAGIC1 = 0xBACCD00A,
884 MAGIC2 = 0xCA110000,
885 XOPEN = 5,
886 XWRITE = 4,
887};
888
889static noinline long simnow(long cmd, long a, long b, long c)
890{
891 long ret;
892
893 asm volatile("cpuid" :
894 "=a" (ret) :
895 "b" (a), "c" (b), "d" (c), "0" (MAGIC1), "D" (cmd + MAGIC2));
896 return ret;
897}
898
899static void __init simnow_init(char *str)
900{
901 char *fn = "klog";
902
903 if (*str == '=')
904 fn = ++str;
905 /* error ignored */
906 simnow_fd = simnow(XOPEN, (unsigned long)fn, O_WRONLY|O_APPEND|O_CREAT, 0644);
907}
908
909static void simnow_write(struct console *con, const char *s, unsigned n)
910{
911 simnow(XWRITE, simnow_fd, (unsigned long)s, n);
912}
913
914static struct console simnow_console = {
915 .name = "simnow",
916 .write = simnow_write,
917 .flags = CON_PRINTBUFFER,
918 .index = -1,
919};
920
921/* Direct interface for emergencies */ 878/* Direct interface for emergencies */
922static struct console *early_console = &early_vga_console; 879static struct console *early_console = &early_vga_console;
923static int __initdata early_console_initialized; 880static int __initdata early_console_initialized;
@@ -960,10 +917,6 @@ static int __init setup_early_printk(char *buf)
960 max_ypos = boot_params.screen_info.orig_video_lines; 917 max_ypos = boot_params.screen_info.orig_video_lines;
961 current_ypos = boot_params.screen_info.orig_y; 918 current_ypos = boot_params.screen_info.orig_y;
962 early_console = &early_vga_console; 919 early_console = &early_vga_console;
963 } else if (!strncmp(buf, "simnow", 6)) {
964 simnow_init(buf + 6);
965 early_console = &simnow_console;
966 keep_early = 1;
967#ifdef CONFIG_EARLY_PRINTK_DBGP 920#ifdef CONFIG_EARLY_PRINTK_DBGP
968 } else if (!strncmp(buf, "dbgp", 4)) { 921 } else if (!strncmp(buf, "dbgp", 4)) {
969 if (early_dbgp_init(buf+4) < 0) 922 if (early_dbgp_init(buf+4) < 0)
diff --git a/arch/x86/kernel/entry_32.S b/arch/x86/kernel/entry_32.S
index 28b597ef9ca1..f6402c4ba10d 100644
--- a/arch/x86/kernel/entry_32.S
+++ b/arch/x86/kernel/entry_32.S
@@ -1051,6 +1051,7 @@ ENTRY(kernel_thread_helper)
1051 push %eax 1051 push %eax
1052 CFI_ADJUST_CFA_OFFSET 4 1052 CFI_ADJUST_CFA_OFFSET 4
1053 call do_exit 1053 call do_exit
1054 ud2 # padding for call trace
1054 CFI_ENDPROC 1055 CFI_ENDPROC
1055ENDPROC(kernel_thread_helper) 1056ENDPROC(kernel_thread_helper)
1056 1057
diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S
index b86f332c96a6..42571baaca32 100644
--- a/arch/x86/kernel/entry_64.S
+++ b/arch/x86/kernel/entry_64.S
@@ -11,15 +11,15 @@
11 * 11 *
12 * NOTE: This code handles signal-recognition, which happens every time 12 * NOTE: This code handles signal-recognition, which happens every time
13 * after an interrupt and after each system call. 13 * after an interrupt and after each system call.
14 * 14 *
15 * Normal syscalls and interrupts don't save a full stack frame, this is 15 * Normal syscalls and interrupts don't save a full stack frame, this is
16 * only done for syscall tracing, signals or fork/exec et.al. 16 * only done for syscall tracing, signals or fork/exec et.al.
17 * 17 *
18 * A note on terminology: 18 * A note on terminology:
19 * - top of stack: Architecture defined interrupt frame from SS to RIP 19 * - top of stack: Architecture defined interrupt frame from SS to RIP
20 * at the top of the kernel process stack. 20 * at the top of the kernel process stack.
21 * - partial stack frame: partially saved registers upto R11. 21 * - partial stack frame: partially saved registers upto R11.
22 * - full stack frame: Like partial stack frame, but all register saved. 22 * - full stack frame: Like partial stack frame, but all register saved.
23 * 23 *
24 * Some macro usage: 24 * Some macro usage:
25 * - CFI macros are used to generate dwarf2 unwind information for better 25 * - CFI macros are used to generate dwarf2 unwind information for better
@@ -142,7 +142,7 @@ END(mcount)
142 142
143#ifndef CONFIG_PREEMPT 143#ifndef CONFIG_PREEMPT
144#define retint_kernel retint_restore_args 144#define retint_kernel retint_restore_args
145#endif 145#endif
146 146
147#ifdef CONFIG_PARAVIRT 147#ifdef CONFIG_PARAVIRT
148ENTRY(native_usergs_sysret64) 148ENTRY(native_usergs_sysret64)
@@ -161,14 +161,14 @@ ENTRY(native_usergs_sysret64)
161.endm 161.endm
162 162
163/* 163/*
164 * C code is not supposed to know about undefined top of stack. Every time 164 * C code is not supposed to know about undefined top of stack. Every time
165 * a C function with an pt_regs argument is called from the SYSCALL based 165 * a C function with an pt_regs argument is called from the SYSCALL based
166 * fast path FIXUP_TOP_OF_STACK is needed. 166 * fast path FIXUP_TOP_OF_STACK is needed.
167 * RESTORE_TOP_OF_STACK syncs the syscall state after any possible ptregs 167 * RESTORE_TOP_OF_STACK syncs the syscall state after any possible ptregs
168 * manipulation. 168 * manipulation.
169 */ 169 */
170 170
171 /* %rsp:at FRAMEEND */ 171 /* %rsp:at FRAMEEND */
172 .macro FIXUP_TOP_OF_STACK tmp 172 .macro FIXUP_TOP_OF_STACK tmp
173 movq %gs:pda_oldrsp,\tmp 173 movq %gs:pda_oldrsp,\tmp
174 movq \tmp,RSP(%rsp) 174 movq \tmp,RSP(%rsp)
@@ -244,8 +244,8 @@ ENTRY(native_usergs_sysret64)
244 .endm 244 .endm
245/* 245/*
246 * A newly forked process directly context switches into this. 246 * A newly forked process directly context switches into this.
247 */ 247 */
248/* rdi: prev */ 248/* rdi: prev */
249ENTRY(ret_from_fork) 249ENTRY(ret_from_fork)
250 CFI_DEFAULT_STACK 250 CFI_DEFAULT_STACK
251 push kernel_eflags(%rip) 251 push kernel_eflags(%rip)
@@ -255,8 +255,9 @@ ENTRY(ret_from_fork)
255 call schedule_tail 255 call schedule_tail
256 GET_THREAD_INFO(%rcx) 256 GET_THREAD_INFO(%rcx)
257 testl $(_TIF_SYSCALL_TRACE|_TIF_SYSCALL_AUDIT),TI_flags(%rcx) 257 testl $(_TIF_SYSCALL_TRACE|_TIF_SYSCALL_AUDIT),TI_flags(%rcx)
258 CFI_REMEMBER_STATE
258 jnz rff_trace 259 jnz rff_trace
259rff_action: 260rff_action:
260 RESTORE_REST 261 RESTORE_REST
261 testl $3,CS-ARGOFFSET(%rsp) # from kernel_thread? 262 testl $3,CS-ARGOFFSET(%rsp) # from kernel_thread?
262 je int_ret_from_sys_call 263 je int_ret_from_sys_call
@@ -264,10 +265,11 @@ rff_action:
264 jnz int_ret_from_sys_call 265 jnz int_ret_from_sys_call
265 RESTORE_TOP_OF_STACK %rdi,ARGOFFSET 266 RESTORE_TOP_OF_STACK %rdi,ARGOFFSET
266 jmp ret_from_sys_call 267 jmp ret_from_sys_call
268 CFI_RESTORE_STATE
267rff_trace: 269rff_trace:
268 movq %rsp,%rdi 270 movq %rsp,%rdi
269 call syscall_trace_leave 271 call syscall_trace_leave
270 GET_THREAD_INFO(%rcx) 272 GET_THREAD_INFO(%rcx)
271 jmp rff_action 273 jmp rff_action
272 CFI_ENDPROC 274 CFI_ENDPROC
273END(ret_from_fork) 275END(ret_from_fork)
@@ -278,20 +280,20 @@ END(ret_from_fork)
278 * SYSCALL does not save anything on the stack and does not change the 280 * SYSCALL does not save anything on the stack and does not change the
279 * stack pointer. 281 * stack pointer.
280 */ 282 */
281 283
282/* 284/*
283 * Register setup: 285 * Register setup:
284 * rax system call number 286 * rax system call number
285 * rdi arg0 287 * rdi arg0
286 * rcx return address for syscall/sysret, C arg3 288 * rcx return address for syscall/sysret, C arg3
287 * rsi arg1 289 * rsi arg1
288 * rdx arg2 290 * rdx arg2
289 * r10 arg3 (--> moved to rcx for C) 291 * r10 arg3 (--> moved to rcx for C)
290 * r8 arg4 292 * r8 arg4
291 * r9 arg5 293 * r9 arg5
292 * r11 eflags for syscall/sysret, temporary for C 294 * r11 eflags for syscall/sysret, temporary for C
293 * r12-r15,rbp,rbx saved by C code, not touched. 295 * r12-r15,rbp,rbx saved by C code, not touched.
294 * 296 *
295 * Interrupts are off on entry. 297 * Interrupts are off on entry.
296 * Only called from user space. 298 * Only called from user space.
297 * 299 *
@@ -301,7 +303,7 @@ END(ret_from_fork)
301 * When user can change the frames always force IRET. That is because 303 * When user can change the frames always force IRET. That is because
302 * it deals with uncanonical addresses better. SYSRET has trouble 304 * it deals with uncanonical addresses better. SYSRET has trouble
303 * with them due to bugs in both AMD and Intel CPUs. 305 * with them due to bugs in both AMD and Intel CPUs.
304 */ 306 */
305 307
306ENTRY(system_call) 308ENTRY(system_call)
307 CFI_STARTPROC simple 309 CFI_STARTPROC simple
@@ -317,7 +319,7 @@ ENTRY(system_call)
317 */ 319 */
318ENTRY(system_call_after_swapgs) 320ENTRY(system_call_after_swapgs)
319 321
320 movq %rsp,%gs:pda_oldrsp 322 movq %rsp,%gs:pda_oldrsp
321 movq %gs:pda_kernelstack,%rsp 323 movq %gs:pda_kernelstack,%rsp
322 /* 324 /*
323 * No need to follow this irqs off/on section - it's straight 325 * No need to follow this irqs off/on section - it's straight
@@ -325,7 +327,7 @@ ENTRY(system_call_after_swapgs)
325 */ 327 */
326 ENABLE_INTERRUPTS(CLBR_NONE) 328 ENABLE_INTERRUPTS(CLBR_NONE)
327 SAVE_ARGS 8,1 329 SAVE_ARGS 8,1
328 movq %rax,ORIG_RAX-ARGOFFSET(%rsp) 330 movq %rax,ORIG_RAX-ARGOFFSET(%rsp)
329 movq %rcx,RIP-ARGOFFSET(%rsp) 331 movq %rcx,RIP-ARGOFFSET(%rsp)
330 CFI_REL_OFFSET rip,RIP-ARGOFFSET 332 CFI_REL_OFFSET rip,RIP-ARGOFFSET
331 GET_THREAD_INFO(%rcx) 333 GET_THREAD_INFO(%rcx)
@@ -339,19 +341,19 @@ system_call_fastpath:
339 movq %rax,RAX-ARGOFFSET(%rsp) 341 movq %rax,RAX-ARGOFFSET(%rsp)
340/* 342/*
341 * Syscall return path ending with SYSRET (fast path) 343 * Syscall return path ending with SYSRET (fast path)
342 * Has incomplete stack frame and undefined top of stack. 344 * Has incomplete stack frame and undefined top of stack.
343 */ 345 */
344ret_from_sys_call: 346ret_from_sys_call:
345 movl $_TIF_ALLWORK_MASK,%edi 347 movl $_TIF_ALLWORK_MASK,%edi
346 /* edi: flagmask */ 348 /* edi: flagmask */
347sysret_check: 349sysret_check:
348 LOCKDEP_SYS_EXIT 350 LOCKDEP_SYS_EXIT
349 GET_THREAD_INFO(%rcx) 351 GET_THREAD_INFO(%rcx)
350 DISABLE_INTERRUPTS(CLBR_NONE) 352 DISABLE_INTERRUPTS(CLBR_NONE)
351 TRACE_IRQS_OFF 353 TRACE_IRQS_OFF
352 movl TI_flags(%rcx),%edx 354 movl TI_flags(%rcx),%edx
353 andl %edi,%edx 355 andl %edi,%edx
354 jnz sysret_careful 356 jnz sysret_careful
355 CFI_REMEMBER_STATE 357 CFI_REMEMBER_STATE
356 /* 358 /*
357 * sysretq will re-enable interrupts: 359 * sysretq will re-enable interrupts:
@@ -366,7 +368,7 @@ sysret_check:
366 368
367 CFI_RESTORE_STATE 369 CFI_RESTORE_STATE
368 /* Handle reschedules */ 370 /* Handle reschedules */
369 /* edx: work, edi: workmask */ 371 /* edx: work, edi: workmask */
370sysret_careful: 372sysret_careful:
371 bt $TIF_NEED_RESCHED,%edx 373 bt $TIF_NEED_RESCHED,%edx
372 jnc sysret_signal 374 jnc sysret_signal
@@ -379,7 +381,7 @@ sysret_careful:
379 CFI_ADJUST_CFA_OFFSET -8 381 CFI_ADJUST_CFA_OFFSET -8
380 jmp sysret_check 382 jmp sysret_check
381 383
382 /* Handle a signal */ 384 /* Handle a signal */
383sysret_signal: 385sysret_signal:
384 TRACE_IRQS_ON 386 TRACE_IRQS_ON
385 ENABLE_INTERRUPTS(CLBR_NONE) 387 ENABLE_INTERRUPTS(CLBR_NONE)
@@ -398,7 +400,7 @@ sysret_signal:
398 DISABLE_INTERRUPTS(CLBR_NONE) 400 DISABLE_INTERRUPTS(CLBR_NONE)
399 TRACE_IRQS_OFF 401 TRACE_IRQS_OFF
400 jmp int_with_check 402 jmp int_with_check
401 403
402badsys: 404badsys:
403 movq $-ENOSYS,RAX-ARGOFFSET(%rsp) 405 movq $-ENOSYS,RAX-ARGOFFSET(%rsp)
404 jmp ret_from_sys_call 406 jmp ret_from_sys_call
@@ -437,7 +439,7 @@ sysret_audit:
437#endif /* CONFIG_AUDITSYSCALL */ 439#endif /* CONFIG_AUDITSYSCALL */
438 440
439 /* Do syscall tracing */ 441 /* Do syscall tracing */
440tracesys: 442tracesys:
441#ifdef CONFIG_AUDITSYSCALL 443#ifdef CONFIG_AUDITSYSCALL
442 testl $(_TIF_WORK_SYSCALL_ENTRY & ~_TIF_SYSCALL_AUDIT),TI_flags(%rcx) 444 testl $(_TIF_WORK_SYSCALL_ENTRY & ~_TIF_SYSCALL_AUDIT),TI_flags(%rcx)
443 jz auditsys 445 jz auditsys
@@ -460,8 +462,8 @@ tracesys:
460 call *sys_call_table(,%rax,8) 462 call *sys_call_table(,%rax,8)
461 movq %rax,RAX-ARGOFFSET(%rsp) 463 movq %rax,RAX-ARGOFFSET(%rsp)
462 /* Use IRET because user could have changed frame */ 464 /* Use IRET because user could have changed frame */
463 465
464/* 466/*
465 * Syscall return path ending with IRET. 467 * Syscall return path ending with IRET.
466 * Has correct top of stack, but partial stack frame. 468 * Has correct top of stack, but partial stack frame.
467 */ 469 */
@@ -505,18 +507,18 @@ int_very_careful:
505 TRACE_IRQS_ON 507 TRACE_IRQS_ON
506 ENABLE_INTERRUPTS(CLBR_NONE) 508 ENABLE_INTERRUPTS(CLBR_NONE)
507 SAVE_REST 509 SAVE_REST
508 /* Check for syscall exit trace */ 510 /* Check for syscall exit trace */
509 testl $_TIF_WORK_SYSCALL_EXIT,%edx 511 testl $_TIF_WORK_SYSCALL_EXIT,%edx
510 jz int_signal 512 jz int_signal
511 pushq %rdi 513 pushq %rdi
512 CFI_ADJUST_CFA_OFFSET 8 514 CFI_ADJUST_CFA_OFFSET 8
513 leaq 8(%rsp),%rdi # &ptregs -> arg1 515 leaq 8(%rsp),%rdi # &ptregs -> arg1
514 call syscall_trace_leave 516 call syscall_trace_leave
515 popq %rdi 517 popq %rdi
516 CFI_ADJUST_CFA_OFFSET -8 518 CFI_ADJUST_CFA_OFFSET -8
517 andl $~(_TIF_WORK_SYSCALL_EXIT|_TIF_SYSCALL_EMU),%edi 519 andl $~(_TIF_WORK_SYSCALL_EXIT|_TIF_SYSCALL_EMU),%edi
518 jmp int_restore_rest 520 jmp int_restore_rest
519 521
520int_signal: 522int_signal:
521 testl $_TIF_DO_NOTIFY_MASK,%edx 523 testl $_TIF_DO_NOTIFY_MASK,%edx
522 jz 1f 524 jz 1f
@@ -531,11 +533,11 @@ int_restore_rest:
531 jmp int_with_check 533 jmp int_with_check
532 CFI_ENDPROC 534 CFI_ENDPROC
533END(system_call) 535END(system_call)
534 536
535/* 537/*
536 * Certain special system calls that need to save a complete full stack frame. 538 * Certain special system calls that need to save a complete full stack frame.
537 */ 539 */
538 540
539 .macro PTREGSCALL label,func,arg 541 .macro PTREGSCALL label,func,arg
540 .globl \label 542 .globl \label
541\label: 543\label:
@@ -572,7 +574,7 @@ ENTRY(ptregscall_common)
572 ret 574 ret
573 CFI_ENDPROC 575 CFI_ENDPROC
574END(ptregscall_common) 576END(ptregscall_common)
575 577
576ENTRY(stub_execve) 578ENTRY(stub_execve)
577 CFI_STARTPROC 579 CFI_STARTPROC
578 popq %r11 580 popq %r11
@@ -588,11 +590,11 @@ ENTRY(stub_execve)
588 jmp int_ret_from_sys_call 590 jmp int_ret_from_sys_call
589 CFI_ENDPROC 591 CFI_ENDPROC
590END(stub_execve) 592END(stub_execve)
591 593
592/* 594/*
593 * sigreturn is special because it needs to restore all registers on return. 595 * sigreturn is special because it needs to restore all registers on return.
594 * This cannot be done with SYSRET, so use the IRET return path instead. 596 * This cannot be done with SYSRET, so use the IRET return path instead.
595 */ 597 */
596ENTRY(stub_rt_sigreturn) 598ENTRY(stub_rt_sigreturn)
597 CFI_STARTPROC 599 CFI_STARTPROC
598 addq $8, %rsp 600 addq $8, %rsp
@@ -685,12 +687,12 @@ exit_intr:
685 GET_THREAD_INFO(%rcx) 687 GET_THREAD_INFO(%rcx)
686 testl $3,CS-ARGOFFSET(%rsp) 688 testl $3,CS-ARGOFFSET(%rsp)
687 je retint_kernel 689 je retint_kernel
688 690
689 /* Interrupt came from user space */ 691 /* Interrupt came from user space */
690 /* 692 /*
691 * Has a correct top of stack, but a partial stack frame 693 * Has a correct top of stack, but a partial stack frame
692 * %rcx: thread info. Interrupts off. 694 * %rcx: thread info. Interrupts off.
693 */ 695 */
694retint_with_reschedule: 696retint_with_reschedule:
695 movl $_TIF_WORK_MASK,%edi 697 movl $_TIF_WORK_MASK,%edi
696retint_check: 698retint_check:
@@ -763,20 +765,20 @@ retint_careful:
763 pushq %rdi 765 pushq %rdi
764 CFI_ADJUST_CFA_OFFSET 8 766 CFI_ADJUST_CFA_OFFSET 8
765 call schedule 767 call schedule
766 popq %rdi 768 popq %rdi
767 CFI_ADJUST_CFA_OFFSET -8 769 CFI_ADJUST_CFA_OFFSET -8
768 GET_THREAD_INFO(%rcx) 770 GET_THREAD_INFO(%rcx)
769 DISABLE_INTERRUPTS(CLBR_NONE) 771 DISABLE_INTERRUPTS(CLBR_NONE)
770 TRACE_IRQS_OFF 772 TRACE_IRQS_OFF
771 jmp retint_check 773 jmp retint_check
772 774
773retint_signal: 775retint_signal:
774 testl $_TIF_DO_NOTIFY_MASK,%edx 776 testl $_TIF_DO_NOTIFY_MASK,%edx
775 jz retint_swapgs 777 jz retint_swapgs
776 TRACE_IRQS_ON 778 TRACE_IRQS_ON
777 ENABLE_INTERRUPTS(CLBR_NONE) 779 ENABLE_INTERRUPTS(CLBR_NONE)
778 SAVE_REST 780 SAVE_REST
779 movq $-1,ORIG_RAX(%rsp) 781 movq $-1,ORIG_RAX(%rsp)
780 xorl %esi,%esi # oldset 782 xorl %esi,%esi # oldset
781 movq %rsp,%rdi # &pt_regs 783 movq %rsp,%rdi # &pt_regs
782 call do_notify_resume 784 call do_notify_resume
@@ -798,14 +800,14 @@ ENTRY(retint_kernel)
798 jnc retint_restore_args 800 jnc retint_restore_args
799 call preempt_schedule_irq 801 call preempt_schedule_irq
800 jmp exit_intr 802 jmp exit_intr
801#endif 803#endif
802 804
803 CFI_ENDPROC 805 CFI_ENDPROC
804END(common_interrupt) 806END(common_interrupt)
805 807
806/* 808/*
807 * APIC interrupts. 809 * APIC interrupts.
808 */ 810 */
809 .macro apicinterrupt num,func 811 .macro apicinterrupt num,func
810 INTR_FRAME 812 INTR_FRAME
811 pushq $~(\num) 813 pushq $~(\num)
@@ -823,14 +825,14 @@ ENTRY(threshold_interrupt)
823 apicinterrupt THRESHOLD_APIC_VECTOR,mce_threshold_interrupt 825 apicinterrupt THRESHOLD_APIC_VECTOR,mce_threshold_interrupt
824END(threshold_interrupt) 826END(threshold_interrupt)
825 827
826#ifdef CONFIG_SMP 828#ifdef CONFIG_SMP
827ENTRY(reschedule_interrupt) 829ENTRY(reschedule_interrupt)
828 apicinterrupt RESCHEDULE_VECTOR,smp_reschedule_interrupt 830 apicinterrupt RESCHEDULE_VECTOR,smp_reschedule_interrupt
829END(reschedule_interrupt) 831END(reschedule_interrupt)
830 832
831 .macro INVALIDATE_ENTRY num 833 .macro INVALIDATE_ENTRY num
832ENTRY(invalidate_interrupt\num) 834ENTRY(invalidate_interrupt\num)
833 apicinterrupt INVALIDATE_TLB_VECTOR_START+\num,smp_invalidate_interrupt 835 apicinterrupt INVALIDATE_TLB_VECTOR_START+\num,smp_invalidate_interrupt
834END(invalidate_interrupt\num) 836END(invalidate_interrupt\num)
835 .endm 837 .endm
836 838
@@ -869,22 +871,22 @@ END(error_interrupt)
869ENTRY(spurious_interrupt) 871ENTRY(spurious_interrupt)
870 apicinterrupt SPURIOUS_APIC_VECTOR,smp_spurious_interrupt 872 apicinterrupt SPURIOUS_APIC_VECTOR,smp_spurious_interrupt
871END(spurious_interrupt) 873END(spurious_interrupt)
872 874
873/* 875/*
874 * Exception entry points. 876 * Exception entry points.
875 */ 877 */
876 .macro zeroentry sym 878 .macro zeroentry sym
877 INTR_FRAME 879 INTR_FRAME
878 PARAVIRT_ADJUST_EXCEPTION_FRAME 880 PARAVIRT_ADJUST_EXCEPTION_FRAME
879 pushq $0 /* push error code/oldrax */ 881 pushq $0 /* push error code/oldrax */
880 CFI_ADJUST_CFA_OFFSET 8 882 CFI_ADJUST_CFA_OFFSET 8
881 pushq %rax /* push real oldrax to the rdi slot */ 883 pushq %rax /* push real oldrax to the rdi slot */
882 CFI_ADJUST_CFA_OFFSET 8 884 CFI_ADJUST_CFA_OFFSET 8
883 CFI_REL_OFFSET rax,0 885 CFI_REL_OFFSET rax,0
884 leaq \sym(%rip),%rax 886 leaq \sym(%rip),%rax
885 jmp error_entry 887 jmp error_entry
886 CFI_ENDPROC 888 CFI_ENDPROC
887 .endm 889 .endm
888 890
889 .macro errorentry sym 891 .macro errorentry sym
890 XCPT_FRAME 892 XCPT_FRAME
@@ -998,13 +1000,13 @@ paranoid_schedule\trace:
998 1000
999/* 1001/*
1000 * Exception entry point. This expects an error code/orig_rax on the stack 1002 * Exception entry point. This expects an error code/orig_rax on the stack
1001 * and the exception handler in %rax. 1003 * and the exception handler in %rax.
1002 */ 1004 */
1003KPROBE_ENTRY(error_entry) 1005KPROBE_ENTRY(error_entry)
1004 _frame RDI 1006 _frame RDI
1005 CFI_REL_OFFSET rax,0 1007 CFI_REL_OFFSET rax,0
1006 /* rdi slot contains rax, oldrax contains error code */ 1008 /* rdi slot contains rax, oldrax contains error code */
1007 cld 1009 cld
1008 subq $14*8,%rsp 1010 subq $14*8,%rsp
1009 CFI_ADJUST_CFA_OFFSET (14*8) 1011 CFI_ADJUST_CFA_OFFSET (14*8)
1010 movq %rsi,13*8(%rsp) 1012 movq %rsi,13*8(%rsp)
@@ -1015,7 +1017,7 @@ KPROBE_ENTRY(error_entry)
1015 CFI_REL_OFFSET rdx,RDX 1017 CFI_REL_OFFSET rdx,RDX
1016 movq %rcx,11*8(%rsp) 1018 movq %rcx,11*8(%rsp)
1017 CFI_REL_OFFSET rcx,RCX 1019 CFI_REL_OFFSET rcx,RCX
1018 movq %rsi,10*8(%rsp) /* store rax */ 1020 movq %rsi,10*8(%rsp) /* store rax */
1019 CFI_REL_OFFSET rax,RAX 1021 CFI_REL_OFFSET rax,RAX
1020 movq %r8, 9*8(%rsp) 1022 movq %r8, 9*8(%rsp)
1021 CFI_REL_OFFSET r8,R8 1023 CFI_REL_OFFSET r8,R8
@@ -1025,29 +1027,29 @@ KPROBE_ENTRY(error_entry)
1025 CFI_REL_OFFSET r10,R10 1027 CFI_REL_OFFSET r10,R10
1026 movq %r11,6*8(%rsp) 1028 movq %r11,6*8(%rsp)
1027 CFI_REL_OFFSET r11,R11 1029 CFI_REL_OFFSET r11,R11
1028 movq %rbx,5*8(%rsp) 1030 movq %rbx,5*8(%rsp)
1029 CFI_REL_OFFSET rbx,RBX 1031 CFI_REL_OFFSET rbx,RBX
1030 movq %rbp,4*8(%rsp) 1032 movq %rbp,4*8(%rsp)
1031 CFI_REL_OFFSET rbp,RBP 1033 CFI_REL_OFFSET rbp,RBP
1032 movq %r12,3*8(%rsp) 1034 movq %r12,3*8(%rsp)
1033 CFI_REL_OFFSET r12,R12 1035 CFI_REL_OFFSET r12,R12
1034 movq %r13,2*8(%rsp) 1036 movq %r13,2*8(%rsp)
1035 CFI_REL_OFFSET r13,R13 1037 CFI_REL_OFFSET r13,R13
1036 movq %r14,1*8(%rsp) 1038 movq %r14,1*8(%rsp)
1037 CFI_REL_OFFSET r14,R14 1039 CFI_REL_OFFSET r14,R14
1038 movq %r15,(%rsp) 1040 movq %r15,(%rsp)
1039 CFI_REL_OFFSET r15,R15 1041 CFI_REL_OFFSET r15,R15
1040 xorl %ebx,%ebx 1042 xorl %ebx,%ebx
1041 testl $3,CS(%rsp) 1043 testl $3,CS(%rsp)
1042 je error_kernelspace 1044 je error_kernelspace
1043error_swapgs: 1045error_swapgs:
1044 SWAPGS 1046 SWAPGS
1045error_sti: 1047error_sti:
1046 TRACE_IRQS_OFF 1048 TRACE_IRQS_OFF
1047 movq %rdi,RDI(%rsp) 1049 movq %rdi,RDI(%rsp)
1048 CFI_REL_OFFSET rdi,RDI 1050 CFI_REL_OFFSET rdi,RDI
1049 movq %rsp,%rdi 1051 movq %rsp,%rdi
1050 movq ORIG_RAX(%rsp),%rsi /* get error code */ 1052 movq ORIG_RAX(%rsp),%rsi /* get error code */
1051 movq $-1,ORIG_RAX(%rsp) 1053 movq $-1,ORIG_RAX(%rsp)
1052 call *%rax 1054 call *%rax
1053 /* ebx: no swapgs flag (1: don't need swapgs, 0: need it) */ 1055 /* ebx: no swapgs flag (1: don't need swapgs, 0: need it) */
@@ -1056,7 +1058,7 @@ error_exit:
1056 RESTORE_REST 1058 RESTORE_REST
1057 DISABLE_INTERRUPTS(CLBR_NONE) 1059 DISABLE_INTERRUPTS(CLBR_NONE)
1058 TRACE_IRQS_OFF 1060 TRACE_IRQS_OFF
1059 GET_THREAD_INFO(%rcx) 1061 GET_THREAD_INFO(%rcx)
1060 testl %eax,%eax 1062 testl %eax,%eax
1061 jne retint_kernel 1063 jne retint_kernel
1062 LOCKDEP_SYS_EXIT_IRQ 1064 LOCKDEP_SYS_EXIT_IRQ
@@ -1072,7 +1074,7 @@ error_kernelspace:
1072 /* There are two places in the kernel that can potentially fault with 1074 /* There are two places in the kernel that can potentially fault with
1073 usergs. Handle them here. The exception handlers after 1075 usergs. Handle them here. The exception handlers after
1074 iret run with kernel gs again, so don't set the user space flag. 1076 iret run with kernel gs again, so don't set the user space flag.
1075 B stepping K8s sometimes report an truncated RIP for IRET 1077 B stepping K8s sometimes report an truncated RIP for IRET
1076 exceptions returning to compat mode. Check for these here too. */ 1078 exceptions returning to compat mode. Check for these here too. */
1077 leaq irq_return(%rip),%rcx 1079 leaq irq_return(%rip),%rcx
1078 cmpq %rcx,RIP(%rsp) 1080 cmpq %rcx,RIP(%rsp)
@@ -1084,17 +1086,17 @@ error_kernelspace:
1084 je error_swapgs 1086 je error_swapgs
1085 jmp error_sti 1087 jmp error_sti
1086KPROBE_END(error_entry) 1088KPROBE_END(error_entry)
1087 1089
1088 /* Reload gs selector with exception handling */ 1090 /* Reload gs selector with exception handling */
1089 /* edi: new selector */ 1091 /* edi: new selector */
1090ENTRY(native_load_gs_index) 1092ENTRY(native_load_gs_index)
1091 CFI_STARTPROC 1093 CFI_STARTPROC
1092 pushf 1094 pushf
1093 CFI_ADJUST_CFA_OFFSET 8 1095 CFI_ADJUST_CFA_OFFSET 8
1094 DISABLE_INTERRUPTS(CLBR_ANY | ~(CLBR_RDI)) 1096 DISABLE_INTERRUPTS(CLBR_ANY | ~(CLBR_RDI))
1095 SWAPGS 1097 SWAPGS
1096gs_change: 1098gs_change:
1097 movl %edi,%gs 1099 movl %edi,%gs
10982: mfence /* workaround */ 11002: mfence /* workaround */
1099 SWAPGS 1101 SWAPGS
1100 popf 1102 popf
@@ -1102,20 +1104,20 @@ gs_change:
1102 ret 1104 ret
1103 CFI_ENDPROC 1105 CFI_ENDPROC
1104ENDPROC(native_load_gs_index) 1106ENDPROC(native_load_gs_index)
1105 1107
1106 .section __ex_table,"a" 1108 .section __ex_table,"a"
1107 .align 8 1109 .align 8
1108 .quad gs_change,bad_gs 1110 .quad gs_change,bad_gs
1109 .previous 1111 .previous
1110 .section .fixup,"ax" 1112 .section .fixup,"ax"
1111 /* running with kernelgs */ 1113 /* running with kernelgs */
1112bad_gs: 1114bad_gs:
1113 SWAPGS /* switch back to user gs */ 1115 SWAPGS /* switch back to user gs */
1114 xorl %eax,%eax 1116 xorl %eax,%eax
1115 movl %eax,%gs 1117 movl %eax,%gs
1116 jmp 2b 1118 jmp 2b
1117 .previous 1119 .previous
1118 1120
1119/* 1121/*
1120 * Create a kernel thread. 1122 * Create a kernel thread.
1121 * 1123 *
@@ -1138,7 +1140,7 @@ ENTRY(kernel_thread)
1138 1140
1139 xorl %r8d,%r8d 1141 xorl %r8d,%r8d
1140 xorl %r9d,%r9d 1142 xorl %r9d,%r9d
1141 1143
1142 # clone now 1144 # clone now
1143 call do_fork 1145 call do_fork
1144 movq %rax,RAX(%rsp) 1146 movq %rax,RAX(%rsp)
@@ -1149,14 +1151,14 @@ ENTRY(kernel_thread)
1149 * so internally to the x86_64 port you can rely on kernel_thread() 1151 * so internally to the x86_64 port you can rely on kernel_thread()
1150 * not to reschedule the child before returning, this avoids the need 1152 * not to reschedule the child before returning, this avoids the need
1151 * of hacks for example to fork off the per-CPU idle tasks. 1153 * of hacks for example to fork off the per-CPU idle tasks.
1152 * [Hopefully no generic code relies on the reschedule -AK] 1154 * [Hopefully no generic code relies on the reschedule -AK]
1153 */ 1155 */
1154 RESTORE_ALL 1156 RESTORE_ALL
1155 UNFAKE_STACK_FRAME 1157 UNFAKE_STACK_FRAME
1156 ret 1158 ret
1157 CFI_ENDPROC 1159 CFI_ENDPROC
1158ENDPROC(kernel_thread) 1160ENDPROC(kernel_thread)
1159 1161
1160child_rip: 1162child_rip:
1161 pushq $0 # fake return address 1163 pushq $0 # fake return address
1162 CFI_STARTPROC 1164 CFI_STARTPROC
@@ -1170,6 +1172,7 @@ child_rip:
1170 # exit 1172 # exit
1171 mov %eax, %edi 1173 mov %eax, %edi
1172 call do_exit 1174 call do_exit
1175 ud2 # padding for call trace
1173 CFI_ENDPROC 1176 CFI_ENDPROC
1174ENDPROC(child_rip) 1177ENDPROC(child_rip)
1175 1178
@@ -1191,10 +1194,10 @@ ENDPROC(child_rip)
1191ENTRY(kernel_execve) 1194ENTRY(kernel_execve)
1192 CFI_STARTPROC 1195 CFI_STARTPROC
1193 FAKE_STACK_FRAME $0 1196 FAKE_STACK_FRAME $0
1194 SAVE_ALL 1197 SAVE_ALL
1195 movq %rsp,%rcx 1198 movq %rsp,%rcx
1196 call sys_execve 1199 call sys_execve
1197 movq %rax, RAX(%rsp) 1200 movq %rax, RAX(%rsp)
1198 RESTORE_REST 1201 RESTORE_REST
1199 testq %rax,%rax 1202 testq %rax,%rax
1200 je int_ret_from_sys_call 1203 je int_ret_from_sys_call
@@ -1213,7 +1216,7 @@ ENTRY(coprocessor_error)
1213END(coprocessor_error) 1216END(coprocessor_error)
1214 1217
1215ENTRY(simd_coprocessor_error) 1218ENTRY(simd_coprocessor_error)
1216 zeroentry do_simd_coprocessor_error 1219 zeroentry do_simd_coprocessor_error
1217END(simd_coprocessor_error) 1220END(simd_coprocessor_error)
1218 1221
1219ENTRY(device_not_available) 1222ENTRY(device_not_available)
@@ -1225,12 +1228,12 @@ KPROBE_ENTRY(debug)
1225 INTR_FRAME 1228 INTR_FRAME
1226 PARAVIRT_ADJUST_EXCEPTION_FRAME 1229 PARAVIRT_ADJUST_EXCEPTION_FRAME
1227 pushq $0 1230 pushq $0
1228 CFI_ADJUST_CFA_OFFSET 8 1231 CFI_ADJUST_CFA_OFFSET 8
1229 paranoidentry do_debug, DEBUG_STACK 1232 paranoidentry do_debug, DEBUG_STACK
1230 paranoidexit 1233 paranoidexit
1231KPROBE_END(debug) 1234KPROBE_END(debug)
1232 1235
1233 /* runs on exception stack */ 1236 /* runs on exception stack */
1234KPROBE_ENTRY(nmi) 1237KPROBE_ENTRY(nmi)
1235 INTR_FRAME 1238 INTR_FRAME
1236 PARAVIRT_ADJUST_EXCEPTION_FRAME 1239 PARAVIRT_ADJUST_EXCEPTION_FRAME
@@ -1264,7 +1267,7 @@ ENTRY(bounds)
1264END(bounds) 1267END(bounds)
1265 1268
1266ENTRY(invalid_op) 1269ENTRY(invalid_op)
1267 zeroentry do_invalid_op 1270 zeroentry do_invalid_op
1268END(invalid_op) 1271END(invalid_op)
1269 1272
1270ENTRY(coprocessor_segment_overrun) 1273ENTRY(coprocessor_segment_overrun)
@@ -1319,7 +1322,7 @@ ENTRY(machine_check)
1319 INTR_FRAME 1322 INTR_FRAME
1320 PARAVIRT_ADJUST_EXCEPTION_FRAME 1323 PARAVIRT_ADJUST_EXCEPTION_FRAME
1321 pushq $0 1324 pushq $0
1322 CFI_ADJUST_CFA_OFFSET 8 1325 CFI_ADJUST_CFA_OFFSET 8
1323 paranoidentry do_machine_check 1326 paranoidentry do_machine_check
1324 jmp paranoid_exit1 1327 jmp paranoid_exit1
1325 CFI_ENDPROC 1328 CFI_ENDPROC
diff --git a/arch/x86/kernel/es7000_32.c b/arch/x86/kernel/es7000_32.c
index f454c78fcef6..53699c931ad4 100644
--- a/arch/x86/kernel/es7000_32.c
+++ b/arch/x86/kernel/es7000_32.c
@@ -38,8 +38,11 @@
38#include <asm/io.h> 38#include <asm/io.h>
39#include <asm/nmi.h> 39#include <asm/nmi.h>
40#include <asm/smp.h> 40#include <asm/smp.h>
41#include <asm/atomic.h>
41#include <asm/apicdef.h> 42#include <asm/apicdef.h>
42#include <mach_mpparse.h> 43#include <mach_mpparse.h>
44#include <asm/genapic.h>
45#include <asm/setup.h>
43 46
44/* 47/*
45 * ES7000 chipsets 48 * ES7000 chipsets
@@ -161,6 +164,43 @@ es7000_rename_gsi(int ioapic, int gsi)
161 return gsi; 164 return gsi;
162} 165}
163 166
167static int wakeup_secondary_cpu_via_mip(int cpu, unsigned long eip)
168{
169 unsigned long vect = 0, psaival = 0;
170
171 if (psai == NULL)
172 return -1;
173
174 vect = ((unsigned long)__pa(eip)/0x1000) << 16;
175 psaival = (0x1000000 | vect | cpu);
176
177 while (*psai & 0x1000000)
178 ;
179
180 *psai = psaival;
181
182 return 0;
183}
184
185static void noop_wait_for_deassert(atomic_t *deassert_not_used)
186{
187}
188
189static int __init es7000_update_genapic(void)
190{
191 genapic->wakeup_cpu = wakeup_secondary_cpu_via_mip;
192
193 /* MPENTIUMIII */
194 if (boot_cpu_data.x86 == 6 &&
195 (boot_cpu_data.x86_model >= 7 || boot_cpu_data.x86_model <= 11)) {
196 es7000_update_genapic_to_cluster();
197 genapic->wait_for_init_deassert = noop_wait_for_deassert;
198 genapic->wakeup_cpu = wakeup_secondary_cpu_via_mip;
199 }
200
201 return 0;
202}
203
164void __init 204void __init
165setup_unisys(void) 205setup_unisys(void)
166{ 206{
@@ -176,6 +216,8 @@ setup_unisys(void)
176 else 216 else
177 es7000_plat = ES7000_CLASSIC; 217 es7000_plat = ES7000_CLASSIC;
178 ioapic_renumber_irq = es7000_rename_gsi; 218 ioapic_renumber_irq = es7000_rename_gsi;
219
220 x86_quirks->update_genapic = es7000_update_genapic;
179} 221}
180 222
181/* 223/*
@@ -250,31 +292,24 @@ int __init find_unisys_acpi_oem_table(unsigned long *oem_addr)
250{ 292{
251 struct acpi_table_header *header = NULL; 293 struct acpi_table_header *header = NULL;
252 int i = 0; 294 int i = 0;
253 acpi_size tbl_size;
254 295
255 while (ACPI_SUCCESS(acpi_get_table_with_size("OEM1", i++, &header, &tbl_size))) { 296 while (ACPI_SUCCESS(acpi_get_table("OEM1", i++, &header))) {
256 if (!memcmp((char *) &header->oem_id, "UNISYS", 6)) { 297 if (!memcmp((char *) &header->oem_id, "UNISYS", 6)) {
257 struct oem_table *t = (struct oem_table *)header; 298 struct oem_table *t = (struct oem_table *)header;
258 299
259 oem_addrX = t->OEMTableAddr; 300 oem_addrX = t->OEMTableAddr;
260 oem_size = t->OEMTableSize; 301 oem_size = t->OEMTableSize;
261 early_acpi_os_unmap_memory(header, tbl_size);
262 302
263 *oem_addr = (unsigned long)__acpi_map_table(oem_addrX, 303 *oem_addr = (unsigned long)__acpi_map_table(oem_addrX,
264 oem_size); 304 oem_size);
265 return 0; 305 return 0;
266 } 306 }
267 early_acpi_os_unmap_memory(header, tbl_size);
268 } 307 }
269 return -1; 308 return -1;
270} 309}
271 310
272void __init unmap_unisys_acpi_oem_table(unsigned long oem_addr) 311void __init unmap_unisys_acpi_oem_table(unsigned long oem_addr)
273{ 312{
274 if (!oem_addr)
275 return;
276
277 __acpi_unmap_table((char *)oem_addr, oem_size);
278} 313}
279#endif 314#endif
280 315
@@ -324,26 +359,6 @@ es7000_mip_write(struct mip_reg *mip_reg)
324 return status; 359 return status;
325} 360}
326 361
327int
328es7000_start_cpu(int cpu, unsigned long eip)
329{
330 unsigned long vect = 0, psaival = 0;
331
332 if (psai == NULL)
333 return -1;
334
335 vect = ((unsigned long)__pa(eip)/0x1000) << 16;
336 psaival = (0x1000000 | vect | cpu);
337
338 while (*psai & 0x1000000)
339 ;
340
341 *psai = psaival;
342
343 return 0;
344
345}
346
347void __init 362void __init
348es7000_sw_apic(void) 363es7000_sw_apic(void)
349{ 364{
diff --git a/arch/x86/kernel/genapic_64.c b/arch/x86/kernel/genapic_64.c
index 6c9bfc9e1e95..2bced78b0b8e 100644
--- a/arch/x86/kernel/genapic_64.c
+++ b/arch/x86/kernel/genapic_64.c
@@ -21,6 +21,7 @@
21#include <asm/smp.h> 21#include <asm/smp.h>
22#include <asm/ipi.h> 22#include <asm/ipi.h>
23#include <asm/genapic.h> 23#include <asm/genapic.h>
24#include <asm/setup.h>
24 25
25extern struct genapic apic_flat; 26extern struct genapic apic_flat;
26extern struct genapic apic_physflat; 27extern struct genapic apic_physflat;
@@ -53,6 +54,9 @@ void __init setup_apic_routing(void)
53 genapic = &apic_physflat; 54 genapic = &apic_physflat;
54 printk(KERN_INFO "Setting APIC routing to %s\n", genapic->name); 55 printk(KERN_INFO "Setting APIC routing to %s\n", genapic->name);
55 } 56 }
57
58 if (x86_quirks->update_genapic)
59 x86_quirks->update_genapic();
56} 60}
57 61
58/* Same for both flat and physical. */ 62/* Same for both flat and physical. */
diff --git a/arch/x86/kernel/genx2apic_uv_x.c b/arch/x86/kernel/genx2apic_uv_x.c
index 2c7dbdb98278..dece17289731 100644
--- a/arch/x86/kernel/genx2apic_uv_x.c
+++ b/arch/x86/kernel/genx2apic_uv_x.c
@@ -10,6 +10,7 @@
10 10
11#include <linux/kernel.h> 11#include <linux/kernel.h>
12#include <linux/threads.h> 12#include <linux/threads.h>
13#include <linux/cpu.h>
13#include <linux/cpumask.h> 14#include <linux/cpumask.h>
14#include <linux/string.h> 15#include <linux/string.h>
15#include <linux/ctype.h> 16#include <linux/ctype.h>
@@ -17,6 +18,9 @@
17#include <linux/sched.h> 18#include <linux/sched.h>
18#include <linux/module.h> 19#include <linux/module.h>
19#include <linux/hardirq.h> 20#include <linux/hardirq.h>
21#include <linux/timer.h>
22#include <linux/proc_fs.h>
23#include <asm/current.h>
20#include <asm/smp.h> 24#include <asm/smp.h>
21#include <asm/ipi.h> 25#include <asm/ipi.h>
22#include <asm/genapic.h> 26#include <asm/genapic.h>
@@ -356,6 +360,103 @@ static __init void uv_rtc_init(void)
356} 360}
357 361
358/* 362/*
363 * percpu heartbeat timer
364 */
365static void uv_heartbeat(unsigned long ignored)
366{
367 struct timer_list *timer = &uv_hub_info->scir.timer;
368 unsigned char bits = uv_hub_info->scir.state;
369
370 /* flip heartbeat bit */
371 bits ^= SCIR_CPU_HEARTBEAT;
372
373 /* is this cpu idle? */
374 if (idle_cpu(raw_smp_processor_id()))
375 bits &= ~SCIR_CPU_ACTIVITY;
376 else
377 bits |= SCIR_CPU_ACTIVITY;
378
379 /* update system controller interface reg */
380 uv_set_scir_bits(bits);
381
382 /* enable next timer period */
383 mod_timer(timer, jiffies + SCIR_CPU_HB_INTERVAL);
384}
385
386static void __cpuinit uv_heartbeat_enable(int cpu)
387{
388 if (!uv_cpu_hub_info(cpu)->scir.enabled) {
389 struct timer_list *timer = &uv_cpu_hub_info(cpu)->scir.timer;
390
391 uv_set_cpu_scir_bits(cpu, SCIR_CPU_HEARTBEAT|SCIR_CPU_ACTIVITY);
392 setup_timer(timer, uv_heartbeat, cpu);
393 timer->expires = jiffies + SCIR_CPU_HB_INTERVAL;
394 add_timer_on(timer, cpu);
395 uv_cpu_hub_info(cpu)->scir.enabled = 1;
396 }
397
398 /* check boot cpu */
399 if (!uv_cpu_hub_info(0)->scir.enabled)
400 uv_heartbeat_enable(0);
401}
402
403#ifdef CONFIG_HOTPLUG_CPU
404static void __cpuinit uv_heartbeat_disable(int cpu)
405{
406 if (uv_cpu_hub_info(cpu)->scir.enabled) {
407 uv_cpu_hub_info(cpu)->scir.enabled = 0;
408 del_timer(&uv_cpu_hub_info(cpu)->scir.timer);
409 }
410 uv_set_cpu_scir_bits(cpu, 0xff);
411}
412
413/*
414 * cpu hotplug notifier
415 */
416static __cpuinit int uv_scir_cpu_notify(struct notifier_block *self,
417 unsigned long action, void *hcpu)
418{
419 long cpu = (long)hcpu;
420
421 switch (action) {
422 case CPU_ONLINE:
423 uv_heartbeat_enable(cpu);
424 break;
425 case CPU_DOWN_PREPARE:
426 uv_heartbeat_disable(cpu);
427 break;
428 default:
429 break;
430 }
431 return NOTIFY_OK;
432}
433
434static __init void uv_scir_register_cpu_notifier(void)
435{
436 hotcpu_notifier(uv_scir_cpu_notify, 0);
437}
438
439#else /* !CONFIG_HOTPLUG_CPU */
440
441static __init void uv_scir_register_cpu_notifier(void)
442{
443}
444
445static __init int uv_init_heartbeat(void)
446{
447 int cpu;
448
449 if (is_uv_system())
450 for_each_online_cpu(cpu)
451 uv_heartbeat_enable(cpu);
452 return 0;
453}
454
455late_initcall(uv_init_heartbeat);
456
457#endif /* !CONFIG_HOTPLUG_CPU */
458
459/*
359 * Called on each cpu to initialize the per_cpu UV data area. 460 * Called on each cpu to initialize the per_cpu UV data area.
360 * ZZZ hotplug not supported yet 461 * ZZZ hotplug not supported yet
361 */ 462 */
@@ -428,7 +529,7 @@ void __init uv_system_init(void)
428 529
429 uv_bios_init(); 530 uv_bios_init();
430 uv_bios_get_sn_info(0, &uv_type, &sn_partition_id, 531 uv_bios_get_sn_info(0, &uv_type, &sn_partition_id,
431 &uv_coherency_id, &uv_region_size); 532 &sn_coherency_id, &sn_region_size);
432 uv_rtc_init(); 533 uv_rtc_init();
433 534
434 for_each_present_cpu(cpu) { 535 for_each_present_cpu(cpu) {
@@ -439,8 +540,7 @@ void __init uv_system_init(void)
439 uv_blade_info[blade].nr_possible_cpus++; 540 uv_blade_info[blade].nr_possible_cpus++;
440 541
441 uv_cpu_hub_info(cpu)->lowmem_remap_base = lowmem_redir_base; 542 uv_cpu_hub_info(cpu)->lowmem_remap_base = lowmem_redir_base;
442 uv_cpu_hub_info(cpu)->lowmem_remap_top = 543 uv_cpu_hub_info(cpu)->lowmem_remap_top = lowmem_redir_size;
443 lowmem_redir_base + lowmem_redir_size;
444 uv_cpu_hub_info(cpu)->m_val = m_val; 544 uv_cpu_hub_info(cpu)->m_val = m_val;
445 uv_cpu_hub_info(cpu)->n_val = m_val; 545 uv_cpu_hub_info(cpu)->n_val = m_val;
446 uv_cpu_hub_info(cpu)->numa_blade_id = blade; 546 uv_cpu_hub_info(cpu)->numa_blade_id = blade;
@@ -450,7 +550,8 @@ void __init uv_system_init(void)
450 uv_cpu_hub_info(cpu)->gpa_mask = (1 << (m_val + n_val)) - 1; 550 uv_cpu_hub_info(cpu)->gpa_mask = (1 << (m_val + n_val)) - 1;
451 uv_cpu_hub_info(cpu)->gnode_upper = gnode_upper; 551 uv_cpu_hub_info(cpu)->gnode_upper = gnode_upper;
452 uv_cpu_hub_info(cpu)->global_mmr_base = mmr_base; 552 uv_cpu_hub_info(cpu)->global_mmr_base = mmr_base;
453 uv_cpu_hub_info(cpu)->coherency_domain_number = uv_coherency_id; 553 uv_cpu_hub_info(cpu)->coherency_domain_number = sn_coherency_id;
554 uv_cpu_hub_info(cpu)->scir.offset = SCIR_LOCAL_MMR_BASE + lcpu;
454 uv_node_to_blade[nid] = blade; 555 uv_node_to_blade[nid] = blade;
455 uv_cpu_to_blade[cpu] = blade; 556 uv_cpu_to_blade[cpu] = blade;
456 max_pnode = max(pnode, max_pnode); 557 max_pnode = max(pnode, max_pnode);
@@ -467,4 +568,6 @@ void __init uv_system_init(void)
467 map_mmioh_high(max_pnode); 568 map_mmioh_high(max_pnode);
468 569
469 uv_cpu_init(); 570 uv_cpu_init();
571 uv_scir_register_cpu_notifier();
572 proc_mkdir("sgi_uv", NULL);
470} 573}
diff --git a/arch/x86/kernel/head.c b/arch/x86/kernel/head.c
index 1dcb0f13897e..3e66bd364a9d 100644
--- a/arch/x86/kernel/head.c
+++ b/arch/x86/kernel/head.c
@@ -35,7 +35,6 @@ void __init reserve_ebda_region(void)
35 35
36 /* start of EBDA area */ 36 /* start of EBDA area */
37 ebda_addr = get_bios_ebda(); 37 ebda_addr = get_bios_ebda();
38 printk(KERN_INFO "BIOS EBDA/lowmem at: %08x/%08x\n", ebda_addr, lowmem);
39 38
40 /* Fixup: bios puts an EBDA in the top 64K segment */ 39 /* Fixup: bios puts an EBDA in the top 64K segment */
41 /* of conventional memory, but does not adjust lowmem. */ 40 /* of conventional memory, but does not adjust lowmem. */
diff --git a/arch/x86/kernel/head32.c b/arch/x86/kernel/head32.c
index fa1d25dd83e3..ac108d1fe182 100644
--- a/arch/x86/kernel/head32.c
+++ b/arch/x86/kernel/head32.c
@@ -12,9 +12,12 @@
12#include <asm/sections.h> 12#include <asm/sections.h>
13#include <asm/e820.h> 13#include <asm/e820.h>
14#include <asm/bios_ebda.h> 14#include <asm/bios_ebda.h>
15#include <asm/trampoline.h>
15 16
16void __init i386_start_kernel(void) 17void __init i386_start_kernel(void)
17{ 18{
19 reserve_trampoline_memory();
20
18 reserve_early(__pa_symbol(&_text), __pa_symbol(&_end), "TEXT DATA BSS"); 21 reserve_early(__pa_symbol(&_text), __pa_symbol(&_end), "TEXT DATA BSS");
19 22
20#ifdef CONFIG_BLK_DEV_INITRD 23#ifdef CONFIG_BLK_DEV_INITRD
diff --git a/arch/x86/kernel/head64.c b/arch/x86/kernel/head64.c
index d16084f90649..388e05a5fc17 100644
--- a/arch/x86/kernel/head64.c
+++ b/arch/x86/kernel/head64.c
@@ -24,6 +24,7 @@
24#include <asm/kdebug.h> 24#include <asm/kdebug.h>
25#include <asm/e820.h> 25#include <asm/e820.h>
26#include <asm/bios_ebda.h> 26#include <asm/bios_ebda.h>
27#include <asm/trampoline.h>
27 28
28/* boot cpu pda */ 29/* boot cpu pda */
29static struct x8664_pda _boot_cpu_pda __read_mostly; 30static struct x8664_pda _boot_cpu_pda __read_mostly;
@@ -120,6 +121,8 @@ void __init x86_64_start_reservations(char *real_mode_data)
120{ 121{
121 copy_bootdata(__va(real_mode_data)); 122 copy_bootdata(__va(real_mode_data));
122 123
124 reserve_trampoline_memory();
125
123 reserve_early(__pa_symbol(&_text), __pa_symbol(&_end), "TEXT DATA BSS"); 126 reserve_early(__pa_symbol(&_text), __pa_symbol(&_end), "TEXT DATA BSS");
124 127
125#ifdef CONFIG_BLK_DEV_INITRD 128#ifdef CONFIG_BLK_DEV_INITRD
diff --git a/arch/x86/kernel/hpet.c b/arch/x86/kernel/hpet.c
index 77017e834cf7..3f0a3edf0a57 100644
--- a/arch/x86/kernel/hpet.c
+++ b/arch/x86/kernel/hpet.c
@@ -33,7 +33,9 @@
33 * HPET address is set in acpi/boot.c, when an ACPI entry exists 33 * HPET address is set in acpi/boot.c, when an ACPI entry exists
34 */ 34 */
35unsigned long hpet_address; 35unsigned long hpet_address;
36unsigned long hpet_num_timers; 36#ifdef CONFIG_PCI_MSI
37static unsigned long hpet_num_timers;
38#endif
37static void __iomem *hpet_virt_address; 39static void __iomem *hpet_virt_address;
38 40
39struct hpet_dev { 41struct hpet_dev {
@@ -322,7 +324,7 @@ static int hpet_next_event(unsigned long delta,
322 * what we wrote hit the chip before we compare it to the 324 * what we wrote hit the chip before we compare it to the
323 * counter. 325 * counter.
324 */ 326 */
325 WARN_ON((u32)hpet_readl(HPET_T0_CMP) != cnt); 327 WARN_ON_ONCE((u32)hpet_readl(HPET_Tn_CMP(timer)) != cnt);
326 328
327 return (s32)((u32)hpet_readl(HPET_COUNTER) - cnt) >= 0 ? -ETIME : 0; 329 return (s32)((u32)hpet_readl(HPET_COUNTER) - cnt) >= 0 ? -ETIME : 0;
328} 330}
@@ -445,7 +447,7 @@ static int hpet_setup_irq(struct hpet_dev *dev)
445{ 447{
446 448
447 if (request_irq(dev->irq, hpet_interrupt_handler, 449 if (request_irq(dev->irq, hpet_interrupt_handler,
448 IRQF_SHARED|IRQF_NOBALANCING, dev->name, dev)) 450 IRQF_DISABLED|IRQF_NOBALANCING, dev->name, dev))
449 return -1; 451 return -1;
450 452
451 disable_irq(dev->irq); 453 disable_irq(dev->irq);
diff --git a/arch/x86/kernel/i387.c b/arch/x86/kernel/i387.c
index 1f20608d4ca8..b0f61f0dcd0a 100644
--- a/arch/x86/kernel/i387.c
+++ b/arch/x86/kernel/i387.c
@@ -58,7 +58,7 @@ void __cpuinit mxcsr_feature_mask_init(void)
58 stts(); 58 stts();
59} 59}
60 60
61void __init init_thread_xstate(void) 61void __cpuinit init_thread_xstate(void)
62{ 62{
63 if (!HAVE_HWFP) { 63 if (!HAVE_HWFP) {
64 xstate_size = sizeof(struct i387_soft_struct); 64 xstate_size = sizeof(struct i387_soft_struct);
diff --git a/arch/x86/kernel/init_task.c b/arch/x86/kernel/init_task.c
index a4f93b4120c1..d39918076bb4 100644
--- a/arch/x86/kernel/init_task.c
+++ b/arch/x86/kernel/init_task.c
@@ -14,7 +14,6 @@ static struct fs_struct init_fs = INIT_FS;
14static struct signal_struct init_signals = INIT_SIGNALS(init_signals); 14static struct signal_struct init_signals = INIT_SIGNALS(init_signals);
15static struct sighand_struct init_sighand = INIT_SIGHAND(init_sighand); 15static struct sighand_struct init_sighand = INIT_SIGHAND(init_sighand);
16struct mm_struct init_mm = INIT_MM(init_mm); 16struct mm_struct init_mm = INIT_MM(init_mm);
17EXPORT_UNUSED_SYMBOL(init_mm); /* will be removed in 2.6.26 */
18 17
19/* 18/*
20 * Initial thread structure. 19 * Initial thread structure.
diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c
index 7a3f2028e2eb..679e7bbbbcd6 100644
--- a/arch/x86/kernel/io_apic.c
+++ b/arch/x86/kernel/io_apic.c
@@ -1140,6 +1140,20 @@ static void __clear_irq_vector(int irq)
1140 1140
1141 cfg->vector = 0; 1141 cfg->vector = 0;
1142 cpus_clear(cfg->domain); 1142 cpus_clear(cfg->domain);
1143
1144 if (likely(!cfg->move_in_progress))
1145 return;
1146 cpus_and(mask, cfg->old_domain, cpu_online_map);
1147 for_each_cpu_mask_nr(cpu, mask) {
1148 for (vector = FIRST_EXTERNAL_VECTOR; vector < NR_VECTORS;
1149 vector++) {
1150 if (per_cpu(vector_irq, cpu)[vector] != irq)
1151 continue;
1152 per_cpu(vector_irq, cpu)[vector] = -1;
1153 break;
1154 }
1155 }
1156 cfg->move_in_progress = 0;
1143} 1157}
1144 1158
1145void __setup_vector_irq(int cpu) 1159void __setup_vector_irq(int cpu)
@@ -2202,10 +2216,9 @@ static void set_ir_ioapic_affinity_irq(unsigned int irq, cpumask_t mask)
2202asmlinkage void smp_irq_move_cleanup_interrupt(void) 2216asmlinkage void smp_irq_move_cleanup_interrupt(void)
2203{ 2217{
2204 unsigned vector, me; 2218 unsigned vector, me;
2219
2205 ack_APIC_irq(); 2220 ack_APIC_irq();
2206#ifdef CONFIG_X86_64
2207 exit_idle(); 2221 exit_idle();
2208#endif
2209 irq_enter(); 2222 irq_enter();
2210 2223
2211 me = smp_processor_id(); 2224 me = smp_processor_id();
@@ -3594,27 +3607,7 @@ int __init io_apic_get_redir_entries (int ioapic)
3594 3607
3595int __init probe_nr_irqs(void) 3608int __init probe_nr_irqs(void)
3596{ 3609{
3597 int idx; 3610 return NR_IRQS;
3598 int nr = 0;
3599#ifndef CONFIG_XEN
3600 int nr_min = 32;
3601#else
3602 int nr_min = NR_IRQS;
3603#endif
3604
3605 for (idx = 0; idx < nr_ioapics; idx++)
3606 nr += io_apic_get_redir_entries(idx) + 1;
3607
3608 /* double it for hotplug and msi and nmi */
3609 nr <<= 1;
3610
3611 /* something wrong ? */
3612 if (nr < nr_min)
3613 nr = nr_min;
3614 if (WARN_ON(nr > NR_IRQS))
3615 nr = NR_IRQS;
3616
3617 return nr;
3618} 3611}
3619 3612
3620/* -------------------------------------------------------------------------- 3613/* --------------------------------------------------------------------------
@@ -3761,7 +3754,9 @@ int acpi_get_override_irq(int bus_irq, int *trigger, int *polarity)
3761void __init setup_ioapic_dest(void) 3754void __init setup_ioapic_dest(void)
3762{ 3755{
3763 int pin, ioapic, irq, irq_entry; 3756 int pin, ioapic, irq, irq_entry;
3757 struct irq_desc *desc;
3764 struct irq_cfg *cfg; 3758 struct irq_cfg *cfg;
3759 cpumask_t mask;
3765 3760
3766 if (skip_ioapic_setup == 1) 3761 if (skip_ioapic_setup == 1)
3767 return; 3762 return;
@@ -3778,16 +3773,30 @@ void __init setup_ioapic_dest(void)
3778 * cpu is online. 3773 * cpu is online.
3779 */ 3774 */
3780 cfg = irq_cfg(irq); 3775 cfg = irq_cfg(irq);
3781 if (!cfg->vector) 3776 if (!cfg->vector) {
3782 setup_IO_APIC_irq(ioapic, pin, irq, 3777 setup_IO_APIC_irq(ioapic, pin, irq,
3783 irq_trigger(irq_entry), 3778 irq_trigger(irq_entry),
3784 irq_polarity(irq_entry)); 3779 irq_polarity(irq_entry));
3780 continue;
3781
3782 }
3783
3784 /*
3785 * Honour affinities which have been set in early boot
3786 */
3787 desc = irq_to_desc(irq);
3788 if (desc->status &
3789 (IRQ_NO_BALANCING | IRQ_AFFINITY_SET))
3790 mask = desc->affinity;
3791 else
3792 mask = TARGET_CPUS;
3793
3785#ifdef CONFIG_INTR_REMAP 3794#ifdef CONFIG_INTR_REMAP
3786 else if (intr_remapping_enabled) 3795 if (intr_remapping_enabled)
3787 set_ir_ioapic_affinity_irq(irq, TARGET_CPUS); 3796 set_ir_ioapic_affinity_irq(irq, mask);
3788#endif
3789 else 3797 else
3790 set_ioapic_affinity_irq(irq, TARGET_CPUS); 3798#endif
3799 set_ioapic_affinity_irq(irq, mask);
3791 } 3800 }
3792 3801
3793 } 3802 }
diff --git a/arch/x86/kernel/irq_64.c b/arch/x86/kernel/irq_64.c
index 60eb84eb77a0..1d3d0e71b044 100644
--- a/arch/x86/kernel/irq_64.c
+++ b/arch/x86/kernel/irq_64.c
@@ -18,7 +18,6 @@
18#include <asm/idle.h> 18#include <asm/idle.h>
19#include <asm/smp.h> 19#include <asm/smp.h>
20 20
21#ifdef CONFIG_DEBUG_STACKOVERFLOW
22/* 21/*
23 * Probabilistic stack overflow check: 22 * Probabilistic stack overflow check:
24 * 23 *
@@ -28,19 +27,18 @@
28 */ 27 */
29static inline void stack_overflow_check(struct pt_regs *regs) 28static inline void stack_overflow_check(struct pt_regs *regs)
30{ 29{
30#ifdef CONFIG_DEBUG_STACKOVERFLOW
31 u64 curbase = (u64)task_stack_page(current); 31 u64 curbase = (u64)task_stack_page(current);
32 static unsigned long warned = -60*HZ; 32
33 33 WARN_ONCE(regs->sp >= curbase &&
34 if (regs->sp >= curbase && regs->sp <= curbase + THREAD_SIZE && 34 regs->sp <= curbase + THREAD_SIZE &&
35 regs->sp < curbase + sizeof(struct thread_info) + 128 && 35 regs->sp < curbase + sizeof(struct thread_info) +
36 time_after(jiffies, warned + 60*HZ)) { 36 sizeof(struct pt_regs) + 128,
37 printk("do_IRQ: %s near stack overflow (cur:%Lx,sp:%lx)\n", 37
38 current->comm, curbase, regs->sp); 38 "do_IRQ: %s near stack overflow (cur:%Lx,sp:%lx)\n",
39 show_stack(NULL,NULL); 39 current->comm, curbase, regs->sp);
40 warned = jiffies;
41 }
42}
43#endif 40#endif
41}
44 42
45/* 43/*
46 * do_IRQ handles all normal device IRQ's (the special 44 * do_IRQ handles all normal device IRQ's (the special
@@ -60,9 +58,7 @@ asmlinkage unsigned int do_IRQ(struct pt_regs *regs)
60 irq_enter(); 58 irq_enter();
61 irq = __get_cpu_var(vector_irq)[vector]; 59 irq = __get_cpu_var(vector_irq)[vector];
62 60
63#ifdef CONFIG_DEBUG_STACKOVERFLOW
64 stack_overflow_check(regs); 61 stack_overflow_check(regs);
65#endif
66 62
67 desc = irq_to_desc(irq); 63 desc = irq_to_desc(irq);
68 if (likely(desc)) 64 if (likely(desc))
diff --git a/arch/x86/kernel/kvmclock.c b/arch/x86/kernel/kvmclock.c
index 774ac4991568..e169ae9b6a62 100644
--- a/arch/x86/kernel/kvmclock.c
+++ b/arch/x86/kernel/kvmclock.c
@@ -128,7 +128,7 @@ static int kvm_register_clock(char *txt)
128} 128}
129 129
130#ifdef CONFIG_X86_LOCAL_APIC 130#ifdef CONFIG_X86_LOCAL_APIC
131static void kvm_setup_secondary_clock(void) 131static void __cpuinit kvm_setup_secondary_clock(void)
132{ 132{
133 /* 133 /*
134 * Now that the first cpu already had this clocksource initialized, 134 * Now that the first cpu already had this clocksource initialized,
diff --git a/arch/x86/kernel/machine_kexec_32.c b/arch/x86/kernel/machine_kexec_32.c
index 7a385746509a..37f420018a41 100644
--- a/arch/x86/kernel/machine_kexec_32.c
+++ b/arch/x86/kernel/machine_kexec_32.c
@@ -13,6 +13,7 @@
13#include <linux/numa.h> 13#include <linux/numa.h>
14#include <linux/ftrace.h> 14#include <linux/ftrace.h>
15#include <linux/suspend.h> 15#include <linux/suspend.h>
16#include <linux/gfp.h>
16 17
17#include <asm/pgtable.h> 18#include <asm/pgtable.h>
18#include <asm/pgalloc.h> 19#include <asm/pgalloc.h>
@@ -25,15 +26,6 @@
25#include <asm/system.h> 26#include <asm/system.h>
26#include <asm/cacheflush.h> 27#include <asm/cacheflush.h>
27 28
28#define PAGE_ALIGNED __attribute__ ((__aligned__(PAGE_SIZE)))
29static u32 kexec_pgd[1024] PAGE_ALIGNED;
30#ifdef CONFIG_X86_PAE
31static u32 kexec_pmd0[1024] PAGE_ALIGNED;
32static u32 kexec_pmd1[1024] PAGE_ALIGNED;
33#endif
34static u32 kexec_pte0[1024] PAGE_ALIGNED;
35static u32 kexec_pte1[1024] PAGE_ALIGNED;
36
37static void set_idt(void *newidt, __u16 limit) 29static void set_idt(void *newidt, __u16 limit)
38{ 30{
39 struct desc_ptr curidt; 31 struct desc_ptr curidt;
@@ -76,6 +68,76 @@ static void load_segments(void)
76#undef __STR 68#undef __STR
77} 69}
78 70
71static void machine_kexec_free_page_tables(struct kimage *image)
72{
73 free_page((unsigned long)image->arch.pgd);
74#ifdef CONFIG_X86_PAE
75 free_page((unsigned long)image->arch.pmd0);
76 free_page((unsigned long)image->arch.pmd1);
77#endif
78 free_page((unsigned long)image->arch.pte0);
79 free_page((unsigned long)image->arch.pte1);
80}
81
82static int machine_kexec_alloc_page_tables(struct kimage *image)
83{
84 image->arch.pgd = (pgd_t *)get_zeroed_page(GFP_KERNEL);
85#ifdef CONFIG_X86_PAE
86 image->arch.pmd0 = (pmd_t *)get_zeroed_page(GFP_KERNEL);
87 image->arch.pmd1 = (pmd_t *)get_zeroed_page(GFP_KERNEL);
88#endif
89 image->arch.pte0 = (pte_t *)get_zeroed_page(GFP_KERNEL);
90 image->arch.pte1 = (pte_t *)get_zeroed_page(GFP_KERNEL);
91 if (!image->arch.pgd ||
92#ifdef CONFIG_X86_PAE
93 !image->arch.pmd0 || !image->arch.pmd1 ||
94#endif
95 !image->arch.pte0 || !image->arch.pte1) {
96 machine_kexec_free_page_tables(image);
97 return -ENOMEM;
98 }
99 return 0;
100}
101
102static void machine_kexec_page_table_set_one(
103 pgd_t *pgd, pmd_t *pmd, pte_t *pte,
104 unsigned long vaddr, unsigned long paddr)
105{
106 pud_t *pud;
107
108 pgd += pgd_index(vaddr);
109#ifdef CONFIG_X86_PAE
110 if (!(pgd_val(*pgd) & _PAGE_PRESENT))
111 set_pgd(pgd, __pgd(__pa(pmd) | _PAGE_PRESENT));
112#endif
113 pud = pud_offset(pgd, vaddr);
114 pmd = pmd_offset(pud, vaddr);
115 if (!(pmd_val(*pmd) & _PAGE_PRESENT))
116 set_pmd(pmd, __pmd(__pa(pte) | _PAGE_TABLE));
117 pte = pte_offset_kernel(pmd, vaddr);
118 set_pte(pte, pfn_pte(paddr >> PAGE_SHIFT, PAGE_KERNEL_EXEC));
119}
120
121static void machine_kexec_prepare_page_tables(struct kimage *image)
122{
123 void *control_page;
124 pmd_t *pmd = 0;
125
126 control_page = page_address(image->control_code_page);
127#ifdef CONFIG_X86_PAE
128 pmd = image->arch.pmd0;
129#endif
130 machine_kexec_page_table_set_one(
131 image->arch.pgd, pmd, image->arch.pte0,
132 (unsigned long)control_page, __pa(control_page));
133#ifdef CONFIG_X86_PAE
134 pmd = image->arch.pmd1;
135#endif
136 machine_kexec_page_table_set_one(
137 image->arch.pgd, pmd, image->arch.pte1,
138 __pa(control_page), __pa(control_page));
139}
140
79/* 141/*
80 * A architecture hook called to validate the 142 * A architecture hook called to validate the
81 * proposed image and prepare the control pages 143 * proposed image and prepare the control pages
@@ -87,12 +149,20 @@ static void load_segments(void)
87 * reboot code buffer to allow us to avoid allocations 149 * reboot code buffer to allow us to avoid allocations
88 * later. 150 * later.
89 * 151 *
90 * Make control page executable. 152 * - Make control page executable.
153 * - Allocate page tables
154 * - Setup page tables
91 */ 155 */
92int machine_kexec_prepare(struct kimage *image) 156int machine_kexec_prepare(struct kimage *image)
93{ 157{
158 int error;
159
94 if (nx_enabled) 160 if (nx_enabled)
95 set_pages_x(image->control_code_page, 1); 161 set_pages_x(image->control_code_page, 1);
162 error = machine_kexec_alloc_page_tables(image);
163 if (error)
164 return error;
165 machine_kexec_prepare_page_tables(image);
96 return 0; 166 return 0;
97} 167}
98 168
@@ -104,6 +174,7 @@ void machine_kexec_cleanup(struct kimage *image)
104{ 174{
105 if (nx_enabled) 175 if (nx_enabled)
106 set_pages_nx(image->control_code_page, 1); 176 set_pages_nx(image->control_code_page, 1);
177 machine_kexec_free_page_tables(image);
107} 178}
108 179
109/* 180/*
@@ -150,18 +221,7 @@ void machine_kexec(struct kimage *image)
150 relocate_kernel_ptr = control_page; 221 relocate_kernel_ptr = control_page;
151 page_list[PA_CONTROL_PAGE] = __pa(control_page); 222 page_list[PA_CONTROL_PAGE] = __pa(control_page);
152 page_list[VA_CONTROL_PAGE] = (unsigned long)control_page; 223 page_list[VA_CONTROL_PAGE] = (unsigned long)control_page;
153 page_list[PA_PGD] = __pa(kexec_pgd); 224 page_list[PA_PGD] = __pa(image->arch.pgd);
154 page_list[VA_PGD] = (unsigned long)kexec_pgd;
155#ifdef CONFIG_X86_PAE
156 page_list[PA_PMD_0] = __pa(kexec_pmd0);
157 page_list[VA_PMD_0] = (unsigned long)kexec_pmd0;
158 page_list[PA_PMD_1] = __pa(kexec_pmd1);
159 page_list[VA_PMD_1] = (unsigned long)kexec_pmd1;
160#endif
161 page_list[PA_PTE_0] = __pa(kexec_pte0);
162 page_list[VA_PTE_0] = (unsigned long)kexec_pte0;
163 page_list[PA_PTE_1] = __pa(kexec_pte1);
164 page_list[VA_PTE_1] = (unsigned long)kexec_pte1;
165 225
166 if (image->type == KEXEC_TYPE_DEFAULT) 226 if (image->type == KEXEC_TYPE_DEFAULT)
167 page_list[PA_SWAP_PAGE] = (page_to_pfn(image->swap_page) 227 page_list[PA_SWAP_PAGE] = (page_to_pfn(image->swap_page)
diff --git a/arch/x86/kernel/microcode_amd.c b/arch/x86/kernel/microcode_amd.c
index 5f8e5d75a254..c25fdb382292 100644
--- a/arch/x86/kernel/microcode_amd.c
+++ b/arch/x86/kernel/microcode_amd.c
@@ -10,7 +10,7 @@
10 * This driver allows to upgrade microcode on AMD 10 * This driver allows to upgrade microcode on AMD
11 * family 0x10 and 0x11 processors. 11 * family 0x10 and 0x11 processors.
12 * 12 *
13 * Licensed unter the terms of the GNU General Public 13 * Licensed under the terms of the GNU General Public
14 * License version 2. See file COPYING for details. 14 * License version 2. See file COPYING for details.
15*/ 15*/
16 16
@@ -32,9 +32,9 @@
32#include <linux/platform_device.h> 32#include <linux/platform_device.h>
33#include <linux/pci.h> 33#include <linux/pci.h>
34#include <linux/pci_ids.h> 34#include <linux/pci_ids.h>
35#include <linux/uaccess.h>
35 36
36#include <asm/msr.h> 37#include <asm/msr.h>
37#include <asm/uaccess.h>
38#include <asm/processor.h> 38#include <asm/processor.h>
39#include <asm/microcode.h> 39#include <asm/microcode.h>
40 40
@@ -47,43 +47,38 @@ MODULE_LICENSE("GPL v2");
47#define UCODE_UCODE_TYPE 0x00000001 47#define UCODE_UCODE_TYPE 0x00000001
48 48
49struct equiv_cpu_entry { 49struct equiv_cpu_entry {
50 unsigned int installed_cpu; 50 u32 installed_cpu;
51 unsigned int fixed_errata_mask; 51 u32 fixed_errata_mask;
52 unsigned int fixed_errata_compare; 52 u32 fixed_errata_compare;
53 unsigned int equiv_cpu; 53 u16 equiv_cpu;
54}; 54 u16 res;
55} __attribute__((packed));
55 56
56struct microcode_header_amd { 57struct microcode_header_amd {
57 unsigned int data_code; 58 u32 data_code;
58 unsigned int patch_id; 59 u32 patch_id;
59 unsigned char mc_patch_data_id[2]; 60 u16 mc_patch_data_id;
60 unsigned char mc_patch_data_len; 61 u8 mc_patch_data_len;
61 unsigned char init_flag; 62 u8 init_flag;
62 unsigned int mc_patch_data_checksum; 63 u32 mc_patch_data_checksum;
63 unsigned int nb_dev_id; 64 u32 nb_dev_id;
64 unsigned int sb_dev_id; 65 u32 sb_dev_id;
65 unsigned char processor_rev_id[2]; 66 u16 processor_rev_id;
66 unsigned char nb_rev_id; 67 u8 nb_rev_id;
67 unsigned char sb_rev_id; 68 u8 sb_rev_id;
68 unsigned char bios_api_rev; 69 u8 bios_api_rev;
69 unsigned char reserved1[3]; 70 u8 reserved1[3];
70 unsigned int match_reg[8]; 71 u32 match_reg[8];
71}; 72} __attribute__((packed));
72 73
73struct microcode_amd { 74struct microcode_amd {
74 struct microcode_header_amd hdr; 75 struct microcode_header_amd hdr;
75 unsigned int mpb[0]; 76 unsigned int mpb[0];
76}; 77};
77 78
78#define UCODE_MAX_SIZE (2048) 79#define UCODE_MAX_SIZE 2048
79#define DEFAULT_UCODE_DATASIZE (896) 80#define UCODE_CONTAINER_SECTION_HDR 8
80#define MC_HEADER_SIZE (sizeof(struct microcode_header_amd)) 81#define UCODE_CONTAINER_HEADER_SIZE 12
81#define DEFAULT_UCODE_TOTALSIZE (DEFAULT_UCODE_DATASIZE + MC_HEADER_SIZE)
82#define DWSIZE (sizeof(u32))
83/* For now we support a fixed ucode total size only */
84#define get_totalsize(mc) \
85 ((((struct microcode_amd *)mc)->hdr.mc_patch_data_len * 28) \
86 + MC_HEADER_SIZE)
87 82
88/* serialize access to the physical write */ 83/* serialize access to the physical write */
89static DEFINE_SPINLOCK(microcode_update_lock); 84static DEFINE_SPINLOCK(microcode_update_lock);
@@ -93,31 +88,24 @@ static struct equiv_cpu_entry *equiv_cpu_table;
93static int collect_cpu_info_amd(int cpu, struct cpu_signature *csig) 88static int collect_cpu_info_amd(int cpu, struct cpu_signature *csig)
94{ 89{
95 struct cpuinfo_x86 *c = &cpu_data(cpu); 90 struct cpuinfo_x86 *c = &cpu_data(cpu);
91 u32 dummy;
96 92
97 memset(csig, 0, sizeof(*csig)); 93 memset(csig, 0, sizeof(*csig));
98
99 if (c->x86_vendor != X86_VENDOR_AMD || c->x86 < 0x10) { 94 if (c->x86_vendor != X86_VENDOR_AMD || c->x86 < 0x10) {
100 printk(KERN_ERR "microcode: CPU%d not a capable AMD processor\n", 95 printk(KERN_WARNING "microcode: CPU%d: AMD CPU family 0x%x not "
101 cpu); 96 "supported\n", cpu, c->x86);
102 return -1; 97 return -1;
103 } 98 }
104 99 rdmsr(MSR_AMD64_PATCH_LEVEL, csig->rev, dummy);
105 asm volatile("movl %1, %%ecx; rdmsr" 100 printk(KERN_INFO "microcode: CPU%d: patch_level=0x%x\n", cpu, csig->rev);
106 : "=a" (csig->rev)
107 : "i" (0x0000008B) : "ecx");
108
109 printk(KERN_INFO "microcode: collect_cpu_info_amd : patch_id=0x%x\n",
110 csig->rev);
111
112 return 0; 101 return 0;
113} 102}
114 103
115static int get_matching_microcode(int cpu, void *mc, int rev) 104static int get_matching_microcode(int cpu, void *mc, int rev)
116{ 105{
117 struct microcode_header_amd *mc_header = mc; 106 struct microcode_header_amd *mc_header = mc;
118 struct pci_dev *nb_pci_dev, *sb_pci_dev;
119 unsigned int current_cpu_id; 107 unsigned int current_cpu_id;
120 unsigned int equiv_cpu_id = 0x00; 108 u16 equiv_cpu_id = 0;
121 unsigned int i = 0; 109 unsigned int i = 0;
122 110
123 BUG_ON(equiv_cpu_table == NULL); 111 BUG_ON(equiv_cpu_table == NULL);
@@ -132,57 +120,25 @@ static int get_matching_microcode(int cpu, void *mc, int rev)
132 } 120 }
133 121
134 if (!equiv_cpu_id) { 122 if (!equiv_cpu_id) {
135 printk(KERN_ERR "microcode: CPU%d cpu_id " 123 printk(KERN_WARNING "microcode: CPU%d: cpu revision "
136 "not found in equivalent cpu table \n", cpu); 124 "not listed in equivalent cpu table\n", cpu);
137 return 0; 125 return 0;
138 } 126 }
139 127
140 if ((mc_header->processor_rev_id[0]) != (equiv_cpu_id & 0xff)) { 128 if (mc_header->processor_rev_id != equiv_cpu_id) {
141 printk(KERN_ERR 129 printk(KERN_ERR "microcode: CPU%d: patch mismatch "
142 "microcode: CPU%d patch does not match " 130 "(processor_rev_id: %x, equiv_cpu_id: %x)\n",
143 "(patch is %x, cpu extended is %x) \n", 131 cpu, mc_header->processor_rev_id, equiv_cpu_id);
144 cpu, mc_header->processor_rev_id[0],
145 (equiv_cpu_id & 0xff));
146 return 0; 132 return 0;
147 } 133 }
148 134
149 if ((mc_header->processor_rev_id[1]) != ((equiv_cpu_id >> 16) & 0xff)) { 135 /* ucode might be chipset specific -- currently we don't support this */
150 printk(KERN_ERR "microcode: CPU%d patch does not match " 136 if (mc_header->nb_dev_id || mc_header->sb_dev_id) {
151 "(patch is %x, cpu base id is %x) \n", 137 printk(KERN_ERR "microcode: CPU%d: loading of chipset "
152 cpu, mc_header->processor_rev_id[1], 138 "specific code not yet supported\n", cpu);
153 ((equiv_cpu_id >> 16) & 0xff));
154
155 return 0; 139 return 0;
156 } 140 }
157 141
158 /* ucode may be northbridge specific */
159 if (mc_header->nb_dev_id) {
160 nb_pci_dev = pci_get_device(PCI_VENDOR_ID_AMD,
161 (mc_header->nb_dev_id & 0xff),
162 NULL);
163 if ((!nb_pci_dev) ||
164 (mc_header->nb_rev_id != nb_pci_dev->revision)) {
165 printk(KERN_ERR "microcode: CPU%d NB mismatch \n", cpu);
166 pci_dev_put(nb_pci_dev);
167 return 0;
168 }
169 pci_dev_put(nb_pci_dev);
170 }
171
172 /* ucode may be southbridge specific */
173 if (mc_header->sb_dev_id) {
174 sb_pci_dev = pci_get_device(PCI_VENDOR_ID_AMD,
175 (mc_header->sb_dev_id & 0xff),
176 NULL);
177 if ((!sb_pci_dev) ||
178 (mc_header->sb_rev_id != sb_pci_dev->revision)) {
179 printk(KERN_ERR "microcode: CPU%d SB mismatch \n", cpu);
180 pci_dev_put(sb_pci_dev);
181 return 0;
182 }
183 pci_dev_put(sb_pci_dev);
184 }
185
186 if (mc_header->patch_id <= rev) 142 if (mc_header->patch_id <= rev)
187 return 0; 143 return 0;
188 144
@@ -192,12 +148,10 @@ static int get_matching_microcode(int cpu, void *mc, int rev)
192static void apply_microcode_amd(int cpu) 148static void apply_microcode_amd(int cpu)
193{ 149{
194 unsigned long flags; 150 unsigned long flags;
195 unsigned int eax, edx; 151 u32 rev, dummy;
196 unsigned int rev;
197 int cpu_num = raw_smp_processor_id(); 152 int cpu_num = raw_smp_processor_id();
198 struct ucode_cpu_info *uci = ucode_cpu_info + cpu_num; 153 struct ucode_cpu_info *uci = ucode_cpu_info + cpu_num;
199 struct microcode_amd *mc_amd = uci->mc; 154 struct microcode_amd *mc_amd = uci->mc;
200 unsigned long addr;
201 155
202 /* We should bind the task to the CPU */ 156 /* We should bind the task to the CPU */
203 BUG_ON(cpu_num != cpu); 157 BUG_ON(cpu_num != cpu);
@@ -206,42 +160,34 @@ static void apply_microcode_amd(int cpu)
206 return; 160 return;
207 161
208 spin_lock_irqsave(&microcode_update_lock, flags); 162 spin_lock_irqsave(&microcode_update_lock, flags);
209 163 wrmsrl(MSR_AMD64_PATCH_LOADER, (u64)(long)&mc_amd->hdr.data_code);
210 addr = (unsigned long)&mc_amd->hdr.data_code;
211 edx = (unsigned int)(((unsigned long)upper_32_bits(addr)));
212 eax = (unsigned int)(((unsigned long)lower_32_bits(addr)));
213
214 asm volatile("movl %0, %%ecx; wrmsr" :
215 : "i" (0xc0010020), "a" (eax), "d" (edx) : "ecx");
216
217 /* get patch id after patching */ 164 /* get patch id after patching */
218 asm volatile("movl %1, %%ecx; rdmsr" 165 rdmsr(MSR_AMD64_PATCH_LEVEL, rev, dummy);
219 : "=a" (rev)
220 : "i" (0x0000008B) : "ecx");
221
222 spin_unlock_irqrestore(&microcode_update_lock, flags); 166 spin_unlock_irqrestore(&microcode_update_lock, flags);
223 167
224 /* check current patch id and patch's id for match */ 168 /* check current patch id and patch's id for match */
225 if (rev != mc_amd->hdr.patch_id) { 169 if (rev != mc_amd->hdr.patch_id) {
226 printk(KERN_ERR "microcode: CPU%d update from revision " 170 printk(KERN_ERR "microcode: CPU%d: update failed "
227 "0x%x to 0x%x failed\n", cpu_num, 171 "(for patch_level=0x%x)\n", cpu, mc_amd->hdr.patch_id);
228 mc_amd->hdr.patch_id, rev);
229 return; 172 return;
230 } 173 }
231 174
232 printk(KERN_INFO "microcode: CPU%d updated from revision " 175 printk(KERN_INFO "microcode: CPU%d: updated (new patch_level=0x%x)\n",
233 "0x%x to 0x%x \n", 176 cpu, rev);
234 cpu_num, uci->cpu_sig.rev, mc_amd->hdr.patch_id);
235 177
236 uci->cpu_sig.rev = rev; 178 uci->cpu_sig.rev = rev;
237} 179}
238 180
239static void * get_next_ucode(u8 *buf, unsigned int size, 181static int get_ucode_data(void *to, const u8 *from, size_t n)
240 int (*get_ucode_data)(void *, const void *, size_t), 182{
241 unsigned int *mc_size) 183 memcpy(to, from, n);
184 return 0;
185}
186
187static void *get_next_ucode(const u8 *buf, unsigned int size,
188 unsigned int *mc_size)
242{ 189{
243 unsigned int total_size; 190 unsigned int total_size;
244#define UCODE_CONTAINER_SECTION_HDR 8
245 u8 section_hdr[UCODE_CONTAINER_SECTION_HDR]; 191 u8 section_hdr[UCODE_CONTAINER_SECTION_HDR];
246 void *mc; 192 void *mc;
247 193
@@ -249,39 +195,37 @@ static void * get_next_ucode(u8 *buf, unsigned int size,
249 return NULL; 195 return NULL;
250 196
251 if (section_hdr[0] != UCODE_UCODE_TYPE) { 197 if (section_hdr[0] != UCODE_UCODE_TYPE) {
252 printk(KERN_ERR "microcode: error! " 198 printk(KERN_ERR "microcode: error: invalid type field in "
253 "Wrong microcode payload type field\n"); 199 "container file section header\n");
254 return NULL; 200 return NULL;
255 } 201 }
256 202
257 total_size = (unsigned long) (section_hdr[4] + (section_hdr[5] << 8)); 203 total_size = (unsigned long) (section_hdr[4] + (section_hdr[5] << 8));
258 204
259 printk(KERN_INFO "microcode: size %u, total_size %u\n", 205 printk(KERN_DEBUG "microcode: size %u, total_size %u\n",
260 size, total_size); 206 size, total_size);
261 207
262 if (total_size > size || total_size > UCODE_MAX_SIZE) { 208 if (total_size > size || total_size > UCODE_MAX_SIZE) {
263 printk(KERN_ERR "microcode: error! Bad data in microcode data file\n"); 209 printk(KERN_ERR "microcode: error: size mismatch\n");
264 return NULL; 210 return NULL;
265 } 211 }
266 212
267 mc = vmalloc(UCODE_MAX_SIZE); 213 mc = vmalloc(UCODE_MAX_SIZE);
268 if (mc) { 214 if (mc) {
269 memset(mc, 0, UCODE_MAX_SIZE); 215 memset(mc, 0, UCODE_MAX_SIZE);
270 if (get_ucode_data(mc, buf + UCODE_CONTAINER_SECTION_HDR, total_size)) { 216 if (get_ucode_data(mc, buf + UCODE_CONTAINER_SECTION_HDR,
217 total_size)) {
271 vfree(mc); 218 vfree(mc);
272 mc = NULL; 219 mc = NULL;
273 } else 220 } else
274 *mc_size = total_size + UCODE_CONTAINER_SECTION_HDR; 221 *mc_size = total_size + UCODE_CONTAINER_SECTION_HDR;
275 } 222 }
276#undef UCODE_CONTAINER_SECTION_HDR
277 return mc; 223 return mc;
278} 224}
279 225
280 226
281static int install_equiv_cpu_table(u8 *buf, 227static int install_equiv_cpu_table(const u8 *buf)
282 int (*get_ucode_data)(void *, const void *, size_t))
283{ 228{
284#define UCODE_CONTAINER_HEADER_SIZE 12
285 u8 *container_hdr[UCODE_CONTAINER_HEADER_SIZE]; 229 u8 *container_hdr[UCODE_CONTAINER_HEADER_SIZE];
286 unsigned int *buf_pos = (unsigned int *)container_hdr; 230 unsigned int *buf_pos = (unsigned int *)container_hdr;
287 unsigned long size; 231 unsigned long size;
@@ -292,14 +236,15 @@ static int install_equiv_cpu_table(u8 *buf,
292 size = buf_pos[2]; 236 size = buf_pos[2];
293 237
294 if (buf_pos[1] != UCODE_EQUIV_CPU_TABLE_TYPE || !size) { 238 if (buf_pos[1] != UCODE_EQUIV_CPU_TABLE_TYPE || !size) {
295 printk(KERN_ERR "microcode: error! " 239 printk(KERN_ERR "microcode: error: invalid type field in "
296 "Wrong microcode equivalnet cpu table\n"); 240 "container file section header\n");
297 return 0; 241 return 0;
298 } 242 }
299 243
300 equiv_cpu_table = (struct equiv_cpu_entry *) vmalloc(size); 244 equiv_cpu_table = (struct equiv_cpu_entry *) vmalloc(size);
301 if (!equiv_cpu_table) { 245 if (!equiv_cpu_table) {
302 printk(KERN_ERR "microcode: error, can't allocate memory for equiv CPU table\n"); 246 printk(KERN_ERR "microcode: failed to allocate "
247 "equivalent CPU table\n");
303 return 0; 248 return 0;
304 } 249 }
305 250
@@ -310,7 +255,6 @@ static int install_equiv_cpu_table(u8 *buf,
310 } 255 }
311 256
312 return size + UCODE_CONTAINER_HEADER_SIZE; /* add header length */ 257 return size + UCODE_CONTAINER_HEADER_SIZE; /* add header length */
313#undef UCODE_CONTAINER_HEADER_SIZE
314} 258}
315 259
316static void free_equiv_cpu_table(void) 260static void free_equiv_cpu_table(void)
@@ -321,18 +265,20 @@ static void free_equiv_cpu_table(void)
321 } 265 }
322} 266}
323 267
324static int generic_load_microcode(int cpu, void *data, size_t size, 268static int generic_load_microcode(int cpu, const u8 *data, size_t size)
325 int (*get_ucode_data)(void *, const void *, size_t))
326{ 269{
327 struct ucode_cpu_info *uci = ucode_cpu_info + cpu; 270 struct ucode_cpu_info *uci = ucode_cpu_info + cpu;
328 u8 *ucode_ptr = data, *new_mc = NULL, *mc; 271 const u8 *ucode_ptr = data;
272 void *new_mc = NULL;
273 void *mc;
329 int new_rev = uci->cpu_sig.rev; 274 int new_rev = uci->cpu_sig.rev;
330 unsigned int leftover; 275 unsigned int leftover;
331 unsigned long offset; 276 unsigned long offset;
332 277
333 offset = install_equiv_cpu_table(ucode_ptr, get_ucode_data); 278 offset = install_equiv_cpu_table(ucode_ptr);
334 if (!offset) { 279 if (!offset) {
335 printk(KERN_ERR "microcode: installing equivalent cpu table failed\n"); 280 printk(KERN_ERR "microcode: failed to create "
281 "equivalent cpu table\n");
336 return -EINVAL; 282 return -EINVAL;
337 } 283 }
338 284
@@ -343,7 +289,7 @@ static int generic_load_microcode(int cpu, void *data, size_t size,
343 unsigned int uninitialized_var(mc_size); 289 unsigned int uninitialized_var(mc_size);
344 struct microcode_header_amd *mc_header; 290 struct microcode_header_amd *mc_header;
345 291
346 mc = get_next_ucode(ucode_ptr, leftover, get_ucode_data, &mc_size); 292 mc = get_next_ucode(ucode_ptr, leftover, &mc_size);
347 if (!mc) 293 if (!mc)
348 break; 294 break;
349 295
@@ -353,7 +299,7 @@ static int generic_load_microcode(int cpu, void *data, size_t size,
353 vfree(new_mc); 299 vfree(new_mc);
354 new_rev = mc_header->patch_id; 300 new_rev = mc_header->patch_id;
355 new_mc = mc; 301 new_mc = mc;
356 } else 302 } else
357 vfree(mc); 303 vfree(mc);
358 304
359 ucode_ptr += mc_size; 305 ucode_ptr += mc_size;
@@ -365,9 +311,9 @@ static int generic_load_microcode(int cpu, void *data, size_t size,
365 if (uci->mc) 311 if (uci->mc)
366 vfree(uci->mc); 312 vfree(uci->mc);
367 uci->mc = new_mc; 313 uci->mc = new_mc;
368 pr_debug("microcode: CPU%d found a matching microcode update with" 314 pr_debug("microcode: CPU%d found a matching microcode "
369 " version 0x%x (current=0x%x)\n", 315 "update with version 0x%x (current=0x%x)\n",
370 cpu, new_rev, uci->cpu_sig.rev); 316 cpu, new_rev, uci->cpu_sig.rev);
371 } else 317 } else
372 vfree(new_mc); 318 vfree(new_mc);
373 } 319 }
@@ -377,12 +323,6 @@ static int generic_load_microcode(int cpu, void *data, size_t size,
377 return (int)leftover; 323 return (int)leftover;
378} 324}
379 325
380static int get_ucode_fw(void *to, const void *from, size_t n)
381{
382 memcpy(to, from, n);
383 return 0;
384}
385
386static int request_microcode_fw(int cpu, struct device *device) 326static int request_microcode_fw(int cpu, struct device *device)
387{ 327{
388 const char *fw_name = "amd-ucode/microcode_amd.bin"; 328 const char *fw_name = "amd-ucode/microcode_amd.bin";
@@ -394,12 +334,11 @@ static int request_microcode_fw(int cpu, struct device *device)
394 334
395 ret = request_firmware(&firmware, fw_name, device); 335 ret = request_firmware(&firmware, fw_name, device);
396 if (ret) { 336 if (ret) {
397 printk(KERN_ERR "microcode: ucode data file %s load failed\n", fw_name); 337 printk(KERN_ERR "microcode: failed to load file %s\n", fw_name);
398 return ret; 338 return ret;
399 } 339 }
400 340
401 ret = generic_load_microcode(cpu, (void*)firmware->data, firmware->size, 341 ret = generic_load_microcode(cpu, firmware->data, firmware->size);
402 &get_ucode_fw);
403 342
404 release_firmware(firmware); 343 release_firmware(firmware);
405 344
@@ -408,8 +347,8 @@ static int request_microcode_fw(int cpu, struct device *device)
408 347
409static int request_microcode_user(int cpu, const void __user *buf, size_t size) 348static int request_microcode_user(int cpu, const void __user *buf, size_t size)
410{ 349{
411 printk(KERN_WARNING "microcode: AMD microcode update via /dev/cpu/microcode" 350 printk(KERN_INFO "microcode: AMD microcode update via "
412 "is not supported\n"); 351 "/dev/cpu/microcode not supported\n");
413 return -1; 352 return -1;
414} 353}
415 354
@@ -433,3 +372,4 @@ struct microcode_ops * __init init_amd_microcode(void)
433{ 372{
434 return &microcode_amd_ops; 373 return &microcode_amd_ops;
435} 374}
375
diff --git a/arch/x86/kernel/microcode_core.c b/arch/x86/kernel/microcode_core.c
index 82fb2809ce32..c9b721ba968c 100644
--- a/arch/x86/kernel/microcode_core.c
+++ b/arch/x86/kernel/microcode_core.c
@@ -99,7 +99,7 @@ MODULE_LICENSE("GPL");
99 99
100#define MICROCODE_VERSION "2.00" 100#define MICROCODE_VERSION "2.00"
101 101
102struct microcode_ops *microcode_ops; 102static struct microcode_ops *microcode_ops;
103 103
104/* no concurrent ->write()s are allowed on /dev/cpu/microcode */ 104/* no concurrent ->write()s are allowed on /dev/cpu/microcode */
105static DEFINE_MUTEX(microcode_mutex); 105static DEFINE_MUTEX(microcode_mutex);
@@ -203,7 +203,7 @@ MODULE_ALIAS_MISCDEV(MICROCODE_MINOR);
203#endif 203#endif
204 204
205/* fake device for request_firmware */ 205/* fake device for request_firmware */
206struct platform_device *microcode_pdev; 206static struct platform_device *microcode_pdev;
207 207
208static ssize_t reload_store(struct sys_device *dev, 208static ssize_t reload_store(struct sys_device *dev,
209 struct sysdev_attribute *attr, 209 struct sysdev_attribute *attr,
@@ -272,13 +272,18 @@ static struct attribute_group mc_attr_group = {
272 .name = "microcode", 272 .name = "microcode",
273}; 273};
274 274
275static void microcode_fini_cpu(int cpu) 275static void __microcode_fini_cpu(int cpu)
276{ 276{
277 struct ucode_cpu_info *uci = ucode_cpu_info + cpu; 277 struct ucode_cpu_info *uci = ucode_cpu_info + cpu;
278 278
279 mutex_lock(&microcode_mutex);
280 microcode_ops->microcode_fini_cpu(cpu); 279 microcode_ops->microcode_fini_cpu(cpu);
281 uci->valid = 0; 280 uci->valid = 0;
281}
282
283static void microcode_fini_cpu(int cpu)
284{
285 mutex_lock(&microcode_mutex);
286 __microcode_fini_cpu(cpu);
282 mutex_unlock(&microcode_mutex); 287 mutex_unlock(&microcode_mutex);
283} 288}
284 289
@@ -306,12 +311,16 @@ static int microcode_resume_cpu(int cpu)
306 * to this cpu (a bit of paranoia): 311 * to this cpu (a bit of paranoia):
307 */ 312 */
308 if (microcode_ops->collect_cpu_info(cpu, &nsig)) { 313 if (microcode_ops->collect_cpu_info(cpu, &nsig)) {
309 microcode_fini_cpu(cpu); 314 __microcode_fini_cpu(cpu);
315 printk(KERN_ERR "failed to collect_cpu_info for resuming cpu #%d\n",
316 cpu);
310 return -1; 317 return -1;
311 } 318 }
312 319
313 if (memcmp(&nsig, &uci->cpu_sig, sizeof(nsig))) { 320 if ((nsig.sig != uci->cpu_sig.sig) || (nsig.pf != uci->cpu_sig.pf)) {
314 microcode_fini_cpu(cpu); 321 __microcode_fini_cpu(cpu);
322 printk(KERN_ERR "cached ucode doesn't match the resuming cpu #%d\n",
323 cpu);
315 /* Should we look for a new ucode here? */ 324 /* Should we look for a new ucode here? */
316 return 1; 325 return 1;
317 } 326 }
@@ -319,7 +328,7 @@ static int microcode_resume_cpu(int cpu)
319 return 0; 328 return 0;
320} 329}
321 330
322void microcode_update_cpu(int cpu) 331static void microcode_update_cpu(int cpu)
323{ 332{
324 struct ucode_cpu_info *uci = ucode_cpu_info + cpu; 333 struct ucode_cpu_info *uci = ucode_cpu_info + cpu;
325 int err = 0; 334 int err = 0;
diff --git a/arch/x86/kernel/microcode_intel.c b/arch/x86/kernel/microcode_intel.c
index 622dc4a21784..b7f4c929e615 100644
--- a/arch/x86/kernel/microcode_intel.c
+++ b/arch/x86/kernel/microcode_intel.c
@@ -155,6 +155,7 @@ static DEFINE_SPINLOCK(microcode_update_lock);
155static int collect_cpu_info(int cpu_num, struct cpu_signature *csig) 155static int collect_cpu_info(int cpu_num, struct cpu_signature *csig)
156{ 156{
157 struct cpuinfo_x86 *c = &cpu_data(cpu_num); 157 struct cpuinfo_x86 *c = &cpu_data(cpu_num);
158 unsigned long flags;
158 unsigned int val[2]; 159 unsigned int val[2];
159 160
160 memset(csig, 0, sizeof(*csig)); 161 memset(csig, 0, sizeof(*csig));
@@ -174,11 +175,16 @@ static int collect_cpu_info(int cpu_num, struct cpu_signature *csig)
174 csig->pf = 1 << ((val[1] >> 18) & 7); 175 csig->pf = 1 << ((val[1] >> 18) & 7);
175 } 176 }
176 177
178 /* serialize access to the physical write to MSR 0x79 */
179 spin_lock_irqsave(&microcode_update_lock, flags);
180
177 wrmsr(MSR_IA32_UCODE_REV, 0, 0); 181 wrmsr(MSR_IA32_UCODE_REV, 0, 0);
178 /* see notes above for revision 1.07. Apparent chip bug */ 182 /* see notes above for revision 1.07. Apparent chip bug */
179 sync_core(); 183 sync_core();
180 /* get the current revision from MSR 0x8B */ 184 /* get the current revision from MSR 0x8B */
181 rdmsr(MSR_IA32_UCODE_REV, val[0], csig->rev); 185 rdmsr(MSR_IA32_UCODE_REV, val[0], csig->rev);
186 spin_unlock_irqrestore(&microcode_update_lock, flags);
187
182 pr_debug("microcode: collect_cpu_info : sig=0x%x, pf=0x%x, rev=0x%x\n", 188 pr_debug("microcode: collect_cpu_info : sig=0x%x, pf=0x%x, rev=0x%x\n",
183 csig->sig, csig->pf, csig->rev); 189 csig->sig, csig->pf, csig->rev);
184 190
@@ -465,7 +471,7 @@ static void microcode_fini_cpu(int cpu)
465 uci->mc = NULL; 471 uci->mc = NULL;
466} 472}
467 473
468struct microcode_ops microcode_intel_ops = { 474static struct microcode_ops microcode_intel_ops = {
469 .request_microcode_user = request_microcode_user, 475 .request_microcode_user = request_microcode_user,
470 .request_microcode_fw = request_microcode_fw, 476 .request_microcode_fw = request_microcode_fw,
471 .collect_cpu_info = collect_cpu_info, 477 .collect_cpu_info = collect_cpu_info,
diff --git a/arch/x86/kernel/mpparse.c b/arch/x86/kernel/mpparse.c
index f98f4e1dba09..45e3b69808ba 100644
--- a/arch/x86/kernel/mpparse.c
+++ b/arch/x86/kernel/mpparse.c
@@ -586,23 +586,23 @@ static void __init __get_smp_config(unsigned int early)
586{ 586{
587 struct intel_mp_floating *mpf = mpf_found; 587 struct intel_mp_floating *mpf = mpf_found;
588 588
589 if (x86_quirks->mach_get_smp_config) { 589 if (!mpf)
590 if (x86_quirks->mach_get_smp_config(early)) 590 return;
591 return; 591
592 }
593 if (acpi_lapic && early) 592 if (acpi_lapic && early)
594 return; 593 return;
594
595 /* 595 /*
596 * ACPI supports both logical (e.g. Hyper-Threading) and physical 596 * MPS doesn't support hyperthreading, aka only have
597 * processors, where MPS only supports physical. 597 * thread 0 apic id in MPS table
598 */ 598 */
599 if (acpi_lapic && acpi_ioapic) { 599 if (acpi_lapic && acpi_ioapic)
600 printk(KERN_INFO "Using ACPI (MADT) for SMP configuration "
601 "information\n");
602 return; 600 return;
603 } else if (acpi_lapic) 601
604 printk(KERN_INFO "Using ACPI for processor (LAPIC) " 602 if (x86_quirks->mach_get_smp_config) {
605 "configuration information\n"); 603 if (x86_quirks->mach_get_smp_config(early))
604 return;
605 }
606 606
607 printk(KERN_INFO "Intel MultiProcessor Specification v1.%d\n", 607 printk(KERN_INFO "Intel MultiProcessor Specification v1.%d\n",
608 mpf->mpf_specification); 608 mpf->mpf_specification);
diff --git a/arch/x86/kernel/nmi.c b/arch/x86/kernel/nmi.c
index 2c97f07f1c2c..8bd1bf9622a7 100644
--- a/arch/x86/kernel/nmi.c
+++ b/arch/x86/kernel/nmi.c
@@ -131,6 +131,11 @@ static void report_broken_nmi(int cpu, int *prev_nmi_count)
131 atomic_dec(&nmi_active); 131 atomic_dec(&nmi_active);
132} 132}
133 133
134static void __acpi_nmi_disable(void *__unused)
135{
136 apic_write(APIC_LVT0, APIC_DM_NMI | APIC_LVT_MASKED);
137}
138
134int __init check_nmi_watchdog(void) 139int __init check_nmi_watchdog(void)
135{ 140{
136 unsigned int *prev_nmi_count; 141 unsigned int *prev_nmi_count;
@@ -179,8 +184,12 @@ int __init check_nmi_watchdog(void)
179 kfree(prev_nmi_count); 184 kfree(prev_nmi_count);
180 return 0; 185 return 0;
181error: 186error:
182 if (nmi_watchdog == NMI_IO_APIC && !timer_through_8259) 187 if (nmi_watchdog == NMI_IO_APIC) {
183 disable_8259A_irq(0); 188 if (!timer_through_8259)
189 disable_8259A_irq(0);
190 on_each_cpu(__acpi_nmi_disable, NULL, 1);
191 }
192
184#ifdef CONFIG_X86_32 193#ifdef CONFIG_X86_32
185 timer_ack = 0; 194 timer_ack = 0;
186#endif 195#endif
@@ -199,12 +208,17 @@ static int __init setup_nmi_watchdog(char *str)
199 ++str; 208 ++str;
200 } 209 }
201 210
202 get_option(&str, &nmi); 211 if (!strncmp(str, "lapic", 5))
203 212 nmi_watchdog = NMI_LOCAL_APIC;
204 if (nmi >= NMI_INVALID) 213 else if (!strncmp(str, "ioapic", 6))
205 return 0; 214 nmi_watchdog = NMI_IO_APIC;
215 else {
216 get_option(&str, &nmi);
217 if (nmi >= NMI_INVALID)
218 return 0;
219 nmi_watchdog = nmi;
220 }
206 221
207 nmi_watchdog = nmi;
208 return 1; 222 return 1;
209} 223}
210__setup("nmi_watchdog=", setup_nmi_watchdog); 224__setup("nmi_watchdog=", setup_nmi_watchdog);
@@ -285,11 +299,6 @@ void acpi_nmi_enable(void)
285 on_each_cpu(__acpi_nmi_enable, NULL, 1); 299 on_each_cpu(__acpi_nmi_enable, NULL, 1);
286} 300}
287 301
288static void __acpi_nmi_disable(void *__unused)
289{
290 apic_write(APIC_LVT0, APIC_DM_NMI | APIC_LVT_MASKED);
291}
292
293/* 302/*
294 * Disable timer based NMIs on all CPUs: 303 * Disable timer based NMIs on all CPUs:
295 */ 304 */
@@ -340,6 +349,8 @@ void stop_apic_nmi_watchdog(void *unused)
340 return; 349 return;
341 if (nmi_watchdog == NMI_LOCAL_APIC) 350 if (nmi_watchdog == NMI_LOCAL_APIC)
342 lapic_watchdog_stop(); 351 lapic_watchdog_stop();
352 else
353 __acpi_nmi_disable(NULL);
343 __get_cpu_var(wd_enabled) = 0; 354 __get_cpu_var(wd_enabled) = 0;
344 atomic_dec(&nmi_active); 355 atomic_dec(&nmi_active);
345} 356}
@@ -465,6 +476,24 @@ nmi_watchdog_tick(struct pt_regs *regs, unsigned reason)
465 476
466#ifdef CONFIG_SYSCTL 477#ifdef CONFIG_SYSCTL
467 478
479static void enable_ioapic_nmi_watchdog_single(void *unused)
480{
481 __get_cpu_var(wd_enabled) = 1;
482 atomic_inc(&nmi_active);
483 __acpi_nmi_enable(NULL);
484}
485
486static void enable_ioapic_nmi_watchdog(void)
487{
488 on_each_cpu(enable_ioapic_nmi_watchdog_single, NULL, 1);
489 touch_nmi_watchdog();
490}
491
492static void disable_ioapic_nmi_watchdog(void)
493{
494 on_each_cpu(stop_apic_nmi_watchdog, NULL, 1);
495}
496
468static int __init setup_unknown_nmi_panic(char *str) 497static int __init setup_unknown_nmi_panic(char *str)
469{ 498{
470 unknown_nmi_panic = 1; 499 unknown_nmi_panic = 1;
@@ -507,6 +536,11 @@ int proc_nmi_enabled(struct ctl_table *table, int write, struct file *file,
507 enable_lapic_nmi_watchdog(); 536 enable_lapic_nmi_watchdog();
508 else 537 else
509 disable_lapic_nmi_watchdog(); 538 disable_lapic_nmi_watchdog();
539 } else if (nmi_watchdog == NMI_IO_APIC) {
540 if (nmi_watchdog_enabled)
541 enable_ioapic_nmi_watchdog();
542 else
543 disable_ioapic_nmi_watchdog();
510 } else { 544 } else {
511 printk(KERN_WARNING 545 printk(KERN_WARNING
512 "NMI watchdog doesn't know what hardware to touch\n"); 546 "NMI watchdog doesn't know what hardware to touch\n");
diff --git a/arch/x86/kernel/numaq_32.c b/arch/x86/kernel/numaq_32.c
index 4caff39078e0..0deea37a53cf 100644
--- a/arch/x86/kernel/numaq_32.c
+++ b/arch/x86/kernel/numaq_32.c
@@ -31,7 +31,7 @@
31#include <asm/numaq.h> 31#include <asm/numaq.h>
32#include <asm/topology.h> 32#include <asm/topology.h>
33#include <asm/processor.h> 33#include <asm/processor.h>
34#include <asm/mpspec.h> 34#include <asm/genapic.h>
35#include <asm/e820.h> 35#include <asm/e820.h>
36#include <asm/setup.h> 36#include <asm/setup.h>
37 37
@@ -235,6 +235,13 @@ static int __init numaq_setup_ioapic_ids(void)
235 return 1; 235 return 1;
236} 236}
237 237
238static int __init numaq_update_genapic(void)
239{
240 genapic->wakeup_cpu = wakeup_secondary_cpu_via_nmi;
241
242 return 0;
243}
244
238static struct x86_quirks numaq_x86_quirks __initdata = { 245static struct x86_quirks numaq_x86_quirks __initdata = {
239 .arch_pre_time_init = numaq_pre_time_init, 246 .arch_pre_time_init = numaq_pre_time_init,
240 .arch_time_init = NULL, 247 .arch_time_init = NULL,
@@ -250,6 +257,7 @@ static struct x86_quirks numaq_x86_quirks __initdata = {
250 .mpc_oem_pci_bus = mpc_oem_pci_bus, 257 .mpc_oem_pci_bus = mpc_oem_pci_bus,
251 .smp_read_mpc_oem = smp_read_mpc_oem, 258 .smp_read_mpc_oem = smp_read_mpc_oem,
252 .setup_ioapic_ids = numaq_setup_ioapic_ids, 259 .setup_ioapic_ids = numaq_setup_ioapic_ids,
260 .update_genapic = numaq_update_genapic,
253}; 261};
254 262
255void numaq_mps_oem_check(struct mp_config_table *mpc, char *oem, 263void numaq_mps_oem_check(struct mp_config_table *mpc, char *oem,
diff --git a/arch/x86/kernel/paravirt-spinlocks.c b/arch/x86/kernel/paravirt-spinlocks.c
index 0e9f1982b1dd..95777b0faa73 100644
--- a/arch/x86/kernel/paravirt-spinlocks.c
+++ b/arch/x86/kernel/paravirt-spinlocks.c
@@ -7,7 +7,8 @@
7 7
8#include <asm/paravirt.h> 8#include <asm/paravirt.h>
9 9
10static void default_spin_lock_flags(struct raw_spinlock *lock, unsigned long flags) 10static inline void
11default_spin_lock_flags(raw_spinlock_t *lock, unsigned long flags)
11{ 12{
12 __raw_spin_lock(lock); 13 __raw_spin_lock(lock);
13} 14}
diff --git a/arch/x86/kernel/pci-calgary_64.c b/arch/x86/kernel/pci-calgary_64.c
index e1e731d78f38..d28bbdc35e4e 100644
--- a/arch/x86/kernel/pci-calgary_64.c
+++ b/arch/x86/kernel/pci-calgary_64.c
@@ -1567,7 +1567,7 @@ static int __init calgary_parse_options(char *p)
1567 ++p; 1567 ++p;
1568 if (*p == '\0') 1568 if (*p == '\0')
1569 break; 1569 break;
1570 bridge = simple_strtol(p, &endp, 0); 1570 bridge = simple_strtoul(p, &endp, 0);
1571 if (p == endp) 1571 if (p == endp)
1572 break; 1572 break;
1573 1573
diff --git a/arch/x86/kernel/pci-dma.c b/arch/x86/kernel/pci-dma.c
index 192624820217..dc572994703d 100644
--- a/arch/x86/kernel/pci-dma.c
+++ b/arch/x86/kernel/pci-dma.c
@@ -300,8 +300,8 @@ fs_initcall(pci_iommu_init);
300static __devinit void via_no_dac(struct pci_dev *dev) 300static __devinit void via_no_dac(struct pci_dev *dev)
301{ 301{
302 if ((dev->class >> 8) == PCI_CLASS_BRIDGE_PCI && forbid_dac == 0) { 302 if ((dev->class >> 8) == PCI_CLASS_BRIDGE_PCI && forbid_dac == 0) {
303 printk(KERN_INFO "PCI: VIA PCI bridge detected." 303 printk(KERN_INFO
304 "Disabling DAC.\n"); 304 "PCI: VIA PCI bridge detected. Disabling DAC.\n");
305 forbid_dac = 1; 305 forbid_dac = 1;
306 } 306 }
307} 307}
diff --git a/arch/x86/kernel/pci-gart_64.c b/arch/x86/kernel/pci-gart_64.c
index a42b02b4df68..a35eaa379ff6 100644
--- a/arch/x86/kernel/pci-gart_64.c
+++ b/arch/x86/kernel/pci-gart_64.c
@@ -123,6 +123,8 @@ static void free_iommu(unsigned long offset, int size)
123 123
124 spin_lock_irqsave(&iommu_bitmap_lock, flags); 124 spin_lock_irqsave(&iommu_bitmap_lock, flags);
125 iommu_area_free(iommu_gart_bitmap, offset, size); 125 iommu_area_free(iommu_gart_bitmap, offset, size);
126 if (offset >= next_bit)
127 next_bit = offset + size;
126 spin_unlock_irqrestore(&iommu_bitmap_lock, flags); 128 spin_unlock_irqrestore(&iommu_bitmap_lock, flags);
127} 129}
128 130
@@ -743,10 +745,8 @@ void __init gart_iommu_init(void)
743 unsigned long scratch; 745 unsigned long scratch;
744 long i; 746 long i;
745 747
746 if (cache_k8_northbridges() < 0 || num_k8_northbridges == 0) { 748 if (cache_k8_northbridges() < 0 || num_k8_northbridges == 0)
747 printk(KERN_INFO "PCI-GART: No AMD GART found.\n");
748 return; 749 return;
749 }
750 750
751#ifndef CONFIG_AGP_AMD64 751#ifndef CONFIG_AGP_AMD64
752 no_agp = 1; 752 no_agp = 1;
diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c
index c622772744d8..9bfc3f719d47 100644
--- a/arch/x86/kernel/process.c
+++ b/arch/x86/kernel/process.c
@@ -1,6 +1,7 @@
1#include <linux/errno.h> 1#include <linux/errno.h>
2#include <linux/kernel.h> 2#include <linux/kernel.h>
3#include <linux/mm.h> 3#include <linux/mm.h>
4#include <asm/idle.h>
4#include <linux/smp.h> 5#include <linux/smp.h>
5#include <linux/slab.h> 6#include <linux/slab.h>
6#include <linux/sched.h> 7#include <linux/sched.h>
@@ -8,6 +9,7 @@
8#include <linux/pm.h> 9#include <linux/pm.h>
9#include <linux/clockchips.h> 10#include <linux/clockchips.h>
10#include <asm/system.h> 11#include <asm/system.h>
12#include <asm/apic.h>
11 13
12unsigned long idle_halt; 14unsigned long idle_halt;
13EXPORT_SYMBOL(idle_halt); 15EXPORT_SYMBOL(idle_halt);
@@ -122,6 +124,21 @@ void default_idle(void)
122EXPORT_SYMBOL(default_idle); 124EXPORT_SYMBOL(default_idle);
123#endif 125#endif
124 126
127void stop_this_cpu(void *dummy)
128{
129 local_irq_disable();
130 /*
131 * Remove this CPU:
132 */
133 cpu_clear(smp_processor_id(), cpu_online_map);
134 disable_local_APIC();
135
136 for (;;) {
137 if (hlt_works(smp_processor_id()))
138 halt();
139 }
140}
141
125static void do_nothing(void *unused) 142static void do_nothing(void *unused)
126{ 143{
127} 144}
diff --git a/arch/x86/kernel/reboot.c b/arch/x86/kernel/reboot.c
index 724adfc63cb9..61f718df6eec 100644
--- a/arch/x86/kernel/reboot.c
+++ b/arch/x86/kernel/reboot.c
@@ -21,6 +21,9 @@
21# include <asm/iommu.h> 21# include <asm/iommu.h>
22#endif 22#endif
23 23
24#include <mach_ipi.h>
25
26
24/* 27/*
25 * Power off function, if any 28 * Power off function, if any
26 */ 29 */
@@ -36,7 +39,10 @@ int reboot_force;
36static int reboot_cpu = -1; 39static int reboot_cpu = -1;
37#endif 40#endif
38 41
39/* reboot=b[ios] | s[mp] | t[riple] | k[bd] | e[fi] [, [w]arm | [c]old] 42/* This is set by the PCI code if either type 1 or type 2 PCI is detected */
43bool port_cf9_safe = false;
44
45/* reboot=b[ios] | s[mp] | t[riple] | k[bd] | e[fi] [, [w]arm | [c]old] | p[ci]
40 warm Don't set the cold reboot flag 46 warm Don't set the cold reboot flag
41 cold Set the cold reboot flag 47 cold Set the cold reboot flag
42 bios Reboot by jumping through the BIOS (only for X86_32) 48 bios Reboot by jumping through the BIOS (only for X86_32)
@@ -45,6 +51,7 @@ static int reboot_cpu = -1;
45 kbd Use the keyboard controller. cold reset (default) 51 kbd Use the keyboard controller. cold reset (default)
46 acpi Use the RESET_REG in the FADT 52 acpi Use the RESET_REG in the FADT
47 efi Use efi reset_system runtime service 53 efi Use efi reset_system runtime service
54 pci Use the so-called "PCI reset register", CF9
48 force Avoid anything that could hang. 55 force Avoid anything that could hang.
49 */ 56 */
50static int __init reboot_setup(char *str) 57static int __init reboot_setup(char *str)
@@ -79,6 +86,7 @@ static int __init reboot_setup(char *str)
79 case 'k': 86 case 'k':
80 case 't': 87 case 't':
81 case 'e': 88 case 'e':
89 case 'p':
82 reboot_type = *str; 90 reboot_type = *str;
83 break; 91 break;
84 92
@@ -169,6 +177,15 @@ static struct dmi_system_id __initdata reboot_dmi_table[] = {
169 DMI_MATCH(DMI_BOARD_NAME, "0KW626"), 177 DMI_MATCH(DMI_BOARD_NAME, "0KW626"),
170 }, 178 },
171 }, 179 },
180 { /* Handle problems with rebooting on Dell Optiplex 330 with 0KP561 */
181 .callback = set_bios_reboot,
182 .ident = "Dell OptiPlex 330",
183 .matches = {
184 DMI_MATCH(DMI_SYS_VENDOR, "Dell Inc."),
185 DMI_MATCH(DMI_PRODUCT_NAME, "OptiPlex 330"),
186 DMI_MATCH(DMI_BOARD_NAME, "0KP561"),
187 },
188 },
172 { /* Handle problems with rebooting on Dell 2400's */ 189 { /* Handle problems with rebooting on Dell 2400's */
173 .callback = set_bios_reboot, 190 .callback = set_bios_reboot,
174 .ident = "Dell PowerEdge 2400", 191 .ident = "Dell PowerEdge 2400",
@@ -395,12 +412,27 @@ static void native_machine_emergency_restart(void)
395 reboot_type = BOOT_KBD; 412 reboot_type = BOOT_KBD;
396 break; 413 break;
397 414
398
399 case BOOT_EFI: 415 case BOOT_EFI:
400 if (efi_enabled) 416 if (efi_enabled)
401 efi.reset_system(reboot_mode ? EFI_RESET_WARM : EFI_RESET_COLD, 417 efi.reset_system(reboot_mode ?
418 EFI_RESET_WARM :
419 EFI_RESET_COLD,
402 EFI_SUCCESS, 0, NULL); 420 EFI_SUCCESS, 0, NULL);
421 reboot_type = BOOT_KBD;
422 break;
403 423
424 case BOOT_CF9:
425 port_cf9_safe = true;
426 /* fall through */
427
428 case BOOT_CF9_COND:
429 if (port_cf9_safe) {
430 u8 cf9 = inb(0xcf9) & ~6;
431 outb(cf9|2, 0xcf9); /* Request hard reset */
432 udelay(50);
433 outb(cf9|6, 0xcf9); /* Actually do the reset */
434 udelay(50);
435 }
404 reboot_type = BOOT_KBD; 436 reboot_type = BOOT_KBD;
405 break; 437 break;
406 } 438 }
@@ -461,6 +493,11 @@ static void native_machine_restart(char *__unused)
461 493
462static void native_machine_halt(void) 494static void native_machine_halt(void)
463{ 495{
496 /* stop other cpus and apics */
497 machine_shutdown();
498
499 /* stop this cpu */
500 stop_this_cpu(NULL);
464} 501}
465 502
466static void native_machine_power_off(void) 503static void native_machine_power_off(void)
@@ -514,3 +551,95 @@ void machine_crash_shutdown(struct pt_regs *regs)
514 machine_ops.crash_shutdown(regs); 551 machine_ops.crash_shutdown(regs);
515} 552}
516#endif 553#endif
554
555
556#if defined(CONFIG_SMP)
557
558/* This keeps a track of which one is crashing cpu. */
559static int crashing_cpu;
560static nmi_shootdown_cb shootdown_callback;
561
562static atomic_t waiting_for_crash_ipi;
563
564static int crash_nmi_callback(struct notifier_block *self,
565 unsigned long val, void *data)
566{
567 int cpu;
568
569 if (val != DIE_NMI_IPI)
570 return NOTIFY_OK;
571
572 cpu = raw_smp_processor_id();
573
574 /* Don't do anything if this handler is invoked on crashing cpu.
575 * Otherwise, system will completely hang. Crashing cpu can get
576 * an NMI if system was initially booted with nmi_watchdog parameter.
577 */
578 if (cpu == crashing_cpu)
579 return NOTIFY_STOP;
580 local_irq_disable();
581
582 shootdown_callback(cpu, (struct die_args *)data);
583
584 atomic_dec(&waiting_for_crash_ipi);
585 /* Assume hlt works */
586 halt();
587 for (;;)
588 cpu_relax();
589
590 return 1;
591}
592
593static void smp_send_nmi_allbutself(void)
594{
595 cpumask_t mask = cpu_online_map;
596 cpu_clear(safe_smp_processor_id(), mask);
597 if (!cpus_empty(mask))
598 send_IPI_mask(mask, NMI_VECTOR);
599}
600
601static struct notifier_block crash_nmi_nb = {
602 .notifier_call = crash_nmi_callback,
603};
604
605/* Halt all other CPUs, calling the specified function on each of them
606 *
607 * This function can be used to halt all other CPUs on crash
608 * or emergency reboot time. The function passed as parameter
609 * will be called inside a NMI handler on all CPUs.
610 */
611void nmi_shootdown_cpus(nmi_shootdown_cb callback)
612{
613 unsigned long msecs;
614 local_irq_disable();
615
616 /* Make a note of crashing cpu. Will be used in NMI callback.*/
617 crashing_cpu = safe_smp_processor_id();
618
619 shootdown_callback = callback;
620
621 atomic_set(&waiting_for_crash_ipi, num_online_cpus() - 1);
622 /* Would it be better to replace the trap vector here? */
623 if (register_die_notifier(&crash_nmi_nb))
624 return; /* return what? */
625 /* Ensure the new callback function is set before sending
626 * out the NMI
627 */
628 wmb();
629
630 smp_send_nmi_allbutself();
631
632 msecs = 1000; /* Wait at most a second for the other cpus to stop */
633 while ((atomic_read(&waiting_for_crash_ipi) > 0) && msecs) {
634 mdelay(1);
635 msecs--;
636 }
637
638 /* Leave the nmi callback set */
639}
640#else /* !CONFIG_SMP */
641void nmi_shootdown_cpus(nmi_shootdown_cb callback)
642{
643 /* No other CPUs to shoot down */
644}
645#endif
diff --git a/arch/x86/kernel/relocate_kernel_32.S b/arch/x86/kernel/relocate_kernel_32.S
index 6f50664b2ba5..a160f3119725 100644
--- a/arch/x86/kernel/relocate_kernel_32.S
+++ b/arch/x86/kernel/relocate_kernel_32.S
@@ -10,15 +10,12 @@
10#include <asm/page.h> 10#include <asm/page.h>
11#include <asm/kexec.h> 11#include <asm/kexec.h>
12#include <asm/processor-flags.h> 12#include <asm/processor-flags.h>
13#include <asm/pgtable.h>
14 13
15/* 14/*
16 * Must be relocatable PIC code callable as a C function 15 * Must be relocatable PIC code callable as a C function
17 */ 16 */
18 17
19#define PTR(x) (x << 2) 18#define PTR(x) (x << 2)
20#define PAGE_ATTR (_PAGE_PRESENT | _PAGE_RW | _PAGE_ACCESSED | _PAGE_DIRTY)
21#define PAE_PGD_ATTR (_PAGE_PRESENT)
22 19
23/* control_page + KEXEC_CONTROL_CODE_MAX_SIZE 20/* control_page + KEXEC_CONTROL_CODE_MAX_SIZE
24 * ~ control_page + PAGE_SIZE are used as data storage and stack for 21 * ~ control_page + PAGE_SIZE are used as data storage and stack for
@@ -39,7 +36,6 @@
39#define CP_PA_BACKUP_PAGES_MAP DATA(0x1c) 36#define CP_PA_BACKUP_PAGES_MAP DATA(0x1c)
40 37
41 .text 38 .text
42 .align PAGE_SIZE
43 .globl relocate_kernel 39 .globl relocate_kernel
44relocate_kernel: 40relocate_kernel:
45 /* Save the CPU context, used for jumping back */ 41 /* Save the CPU context, used for jumping back */
@@ -60,117 +56,6 @@ relocate_kernel:
60 movl %cr4, %eax 56 movl %cr4, %eax
61 movl %eax, CR4(%edi) 57 movl %eax, CR4(%edi)
62 58
63#ifdef CONFIG_X86_PAE
64 /* map the control page at its virtual address */
65
66 movl PTR(VA_PGD)(%ebp), %edi
67 movl PTR(VA_CONTROL_PAGE)(%ebp), %eax
68 andl $0xc0000000, %eax
69 shrl $27, %eax
70 addl %edi, %eax
71
72 movl PTR(PA_PMD_0)(%ebp), %edx
73 orl $PAE_PGD_ATTR, %edx
74 movl %edx, (%eax)
75
76 movl PTR(VA_PMD_0)(%ebp), %edi
77 movl PTR(VA_CONTROL_PAGE)(%ebp), %eax
78 andl $0x3fe00000, %eax
79 shrl $18, %eax
80 addl %edi, %eax
81
82 movl PTR(PA_PTE_0)(%ebp), %edx
83 orl $PAGE_ATTR, %edx
84 movl %edx, (%eax)
85
86 movl PTR(VA_PTE_0)(%ebp), %edi
87 movl PTR(VA_CONTROL_PAGE)(%ebp), %eax
88 andl $0x001ff000, %eax
89 shrl $9, %eax
90 addl %edi, %eax
91
92 movl PTR(PA_CONTROL_PAGE)(%ebp), %edx
93 orl $PAGE_ATTR, %edx
94 movl %edx, (%eax)
95
96 /* identity map the control page at its physical address */
97
98 movl PTR(VA_PGD)(%ebp), %edi
99 movl PTR(PA_CONTROL_PAGE)(%ebp), %eax
100 andl $0xc0000000, %eax
101 shrl $27, %eax
102 addl %edi, %eax
103
104 movl PTR(PA_PMD_1)(%ebp), %edx
105 orl $PAE_PGD_ATTR, %edx
106 movl %edx, (%eax)
107
108 movl PTR(VA_PMD_1)(%ebp), %edi
109 movl PTR(PA_CONTROL_PAGE)(%ebp), %eax
110 andl $0x3fe00000, %eax
111 shrl $18, %eax
112 addl %edi, %eax
113
114 movl PTR(PA_PTE_1)(%ebp), %edx
115 orl $PAGE_ATTR, %edx
116 movl %edx, (%eax)
117
118 movl PTR(VA_PTE_1)(%ebp), %edi
119 movl PTR(PA_CONTROL_PAGE)(%ebp), %eax
120 andl $0x001ff000, %eax
121 shrl $9, %eax
122 addl %edi, %eax
123
124 movl PTR(PA_CONTROL_PAGE)(%ebp), %edx
125 orl $PAGE_ATTR, %edx
126 movl %edx, (%eax)
127#else
128 /* map the control page at its virtual address */
129
130 movl PTR(VA_PGD)(%ebp), %edi
131 movl PTR(VA_CONTROL_PAGE)(%ebp), %eax
132 andl $0xffc00000, %eax
133 shrl $20, %eax
134 addl %edi, %eax
135
136 movl PTR(PA_PTE_0)(%ebp), %edx
137 orl $PAGE_ATTR, %edx
138 movl %edx, (%eax)
139
140 movl PTR(VA_PTE_0)(%ebp), %edi
141 movl PTR(VA_CONTROL_PAGE)(%ebp), %eax
142 andl $0x003ff000, %eax
143 shrl $10, %eax
144 addl %edi, %eax
145
146 movl PTR(PA_CONTROL_PAGE)(%ebp), %edx
147 orl $PAGE_ATTR, %edx
148 movl %edx, (%eax)
149
150 /* identity map the control page at its physical address */
151
152 movl PTR(VA_PGD)(%ebp), %edi
153 movl PTR(PA_CONTROL_PAGE)(%ebp), %eax
154 andl $0xffc00000, %eax
155 shrl $20, %eax
156 addl %edi, %eax
157
158 movl PTR(PA_PTE_1)(%ebp), %edx
159 orl $PAGE_ATTR, %edx
160 movl %edx, (%eax)
161
162 movl PTR(VA_PTE_1)(%ebp), %edi
163 movl PTR(PA_CONTROL_PAGE)(%ebp), %eax
164 andl $0x003ff000, %eax
165 shrl $10, %eax
166 addl %edi, %eax
167
168 movl PTR(PA_CONTROL_PAGE)(%ebp), %edx
169 orl $PAGE_ATTR, %edx
170 movl %edx, (%eax)
171#endif
172
173relocate_new_kernel:
174 /* read the arguments and say goodbye to the stack */ 59 /* read the arguments and say goodbye to the stack */
175 movl 20+4(%esp), %ebx /* page_list */ 60 movl 20+4(%esp), %ebx /* page_list */
176 movl 20+8(%esp), %ebp /* list of pages */ 61 movl 20+8(%esp), %ebp /* list of pages */
diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c
index 0fa6790c1dd3..a31223828597 100644
--- a/arch/x86/kernel/setup.c
+++ b/arch/x86/kernel/setup.c
@@ -98,6 +98,7 @@
98 98
99#include <mach_apic.h> 99#include <mach_apic.h>
100#include <asm/paravirt.h> 100#include <asm/paravirt.h>
101#include <asm/hypervisor.h>
101 102
102#include <asm/percpu.h> 103#include <asm/percpu.h>
103#include <asm/topology.h> 104#include <asm/topology.h>
@@ -448,6 +449,7 @@ static void __init reserve_early_setup_data(void)
448 * @size: Size of the crashkernel memory to reserve. 449 * @size: Size of the crashkernel memory to reserve.
449 * Returns the base address on success, and -1ULL on failure. 450 * Returns the base address on success, and -1ULL on failure.
450 */ 451 */
452static
451unsigned long long __init find_and_reserve_crashkernel(unsigned long long size) 453unsigned long long __init find_and_reserve_crashkernel(unsigned long long size)
452{ 454{
453 const unsigned long long alignment = 16<<20; /* 16M */ 455 const unsigned long long alignment = 16<<20; /* 16M */
@@ -583,161 +585,24 @@ static int __init setup_elfcorehdr(char *arg)
583early_param("elfcorehdr", setup_elfcorehdr); 585early_param("elfcorehdr", setup_elfcorehdr);
584#endif 586#endif
585 587
586static struct x86_quirks default_x86_quirks __initdata; 588static int __init default_update_genapic(void)
587
588struct x86_quirks *x86_quirks __initdata = &default_x86_quirks;
589
590/*
591 * Some BIOSes seem to corrupt the low 64k of memory during events
592 * like suspend/resume and unplugging an HDMI cable. Reserve all
593 * remaining free memory in that area and fill it with a distinct
594 * pattern.
595 */
596#ifdef CONFIG_X86_CHECK_BIOS_CORRUPTION
597#define MAX_SCAN_AREAS 8
598
599static int __read_mostly memory_corruption_check = -1;
600
601static unsigned __read_mostly corruption_check_size = 64*1024;
602static unsigned __read_mostly corruption_check_period = 60; /* seconds */
603
604static struct e820entry scan_areas[MAX_SCAN_AREAS];
605static int num_scan_areas;
606
607
608static int set_corruption_check(char *arg)
609{ 589{
610 char *end; 590#ifdef CONFIG_X86_SMP
611 591# if defined(CONFIG_X86_GENERICARCH) || defined(CONFIG_X86_64)
612 memory_corruption_check = simple_strtol(arg, &end, 10); 592 genapic->wakeup_cpu = wakeup_secondary_cpu_via_init;
613 593# endif
614 return (*end == 0) ? 0 : -EINVAL;
615}
616early_param("memory_corruption_check", set_corruption_check);
617
618static int set_corruption_check_period(char *arg)
619{
620 char *end;
621
622 corruption_check_period = simple_strtoul(arg, &end, 10);
623
624 return (*end == 0) ? 0 : -EINVAL;
625}
626early_param("memory_corruption_check_period", set_corruption_check_period);
627
628static int set_corruption_check_size(char *arg)
629{
630 char *end;
631 unsigned size;
632
633 size = memparse(arg, &end);
634
635 if (*end == '\0')
636 corruption_check_size = size;
637
638 return (size == corruption_check_size) ? 0 : -EINVAL;
639}
640early_param("memory_corruption_check_size", set_corruption_check_size);
641
642
643static void __init setup_bios_corruption_check(void)
644{
645 u64 addr = PAGE_SIZE; /* assume first page is reserved anyway */
646
647 if (memory_corruption_check == -1) {
648 memory_corruption_check =
649#ifdef CONFIG_X86_BOOTPARAM_MEMORY_CORRUPTION_CHECK
650 1
651#else
652 0
653#endif 594#endif
654 ;
655 }
656
657 if (corruption_check_size == 0)
658 memory_corruption_check = 0;
659
660 if (!memory_corruption_check)
661 return;
662
663 corruption_check_size = round_up(corruption_check_size, PAGE_SIZE);
664
665 while(addr < corruption_check_size && num_scan_areas < MAX_SCAN_AREAS) {
666 u64 size;
667 addr = find_e820_area_size(addr, &size, PAGE_SIZE);
668
669 if (addr == 0)
670 break;
671 595
672 if ((addr + size) > corruption_check_size) 596 return 0;
673 size = corruption_check_size - addr;
674
675 if (size == 0)
676 break;
677
678 e820_update_range(addr, size, E820_RAM, E820_RESERVED);
679 scan_areas[num_scan_areas].addr = addr;
680 scan_areas[num_scan_areas].size = size;
681 num_scan_areas++;
682
683 /* Assume we've already mapped this early memory */
684 memset(__va(addr), 0, size);
685
686 addr += size;
687 }
688
689 printk(KERN_INFO "Scanning %d areas for low memory corruption\n",
690 num_scan_areas);
691 update_e820();
692}
693
694static struct timer_list periodic_check_timer;
695
696void check_for_bios_corruption(void)
697{
698 int i;
699 int corruption = 0;
700
701 if (!memory_corruption_check)
702 return;
703
704 for(i = 0; i < num_scan_areas; i++) {
705 unsigned long *addr = __va(scan_areas[i].addr);
706 unsigned long size = scan_areas[i].size;
707
708 for(; size; addr++, size -= sizeof(unsigned long)) {
709 if (!*addr)
710 continue;
711 printk(KERN_ERR "Corrupted low memory at %p (%lx phys) = %08lx\n",
712 addr, __pa(addr), *addr);
713 corruption = 1;
714 *addr = 0;
715 }
716 }
717
718 WARN(corruption, KERN_ERR "Memory corruption detected in low memory\n");
719}
720
721static void periodic_check_for_corruption(unsigned long data)
722{
723 check_for_bios_corruption();
724 mod_timer(&periodic_check_timer, round_jiffies(jiffies + corruption_check_period*HZ));
725} 597}
726 598
727void start_periodic_check_for_corruption(void) 599static struct x86_quirks default_x86_quirks __initdata = {
728{ 600 .update_genapic = default_update_genapic,
729 if (!memory_corruption_check || corruption_check_period == 0) 601};
730 return;
731
732 printk(KERN_INFO "Scanning for low memory corruption every %d seconds\n",
733 corruption_check_period);
734 602
735 init_timer(&periodic_check_timer); 603struct x86_quirks *x86_quirks __initdata = &default_x86_quirks;
736 periodic_check_timer.function = &periodic_check_for_corruption;
737 periodic_check_for_corruption(0);
738}
739#endif
740 604
605#ifdef CONFIG_X86_RESERVE_LOW_64K
741static int __init dmi_low_memory_corruption(const struct dmi_system_id *d) 606static int __init dmi_low_memory_corruption(const struct dmi_system_id *d)
742{ 607{
743 printk(KERN_NOTICE 608 printk(KERN_NOTICE
@@ -749,6 +614,7 @@ static int __init dmi_low_memory_corruption(const struct dmi_system_id *d)
749 614
750 return 0; 615 return 0;
751} 616}
617#endif
752 618
753/* List of systems that have known low memory corruption BIOS problems */ 619/* List of systems that have known low memory corruption BIOS problems */
754static struct dmi_system_id __initdata bad_bios_dmi_table[] = { 620static struct dmi_system_id __initdata bad_bios_dmi_table[] = {
@@ -764,7 +630,7 @@ static struct dmi_system_id __initdata bad_bios_dmi_table[] = {
764 .callback = dmi_low_memory_corruption, 630 .callback = dmi_low_memory_corruption,
765 .ident = "Phoenix BIOS", 631 .ident = "Phoenix BIOS",
766 .matches = { 632 .matches = {
767 DMI_MATCH(DMI_BIOS_VENDOR, "Phoenix Technologies, LTD"), 633 DMI_MATCH(DMI_BIOS_VENDOR, "Phoenix Technologies"),
768 }, 634 },
769 }, 635 },
770#endif 636#endif
@@ -794,6 +660,9 @@ void __init setup_arch(char **cmdline_p)
794 printk(KERN_INFO "Command line: %s\n", boot_command_line); 660 printk(KERN_INFO "Command line: %s\n", boot_command_line);
795#endif 661#endif
796 662
663 /* VMI may relocate the fixmap; do this before touching ioremap area */
664 vmi_init();
665
797 early_cpu_init(); 666 early_cpu_init();
798 early_ioremap_init(); 667 early_ioremap_init();
799 668
@@ -880,13 +749,8 @@ void __init setup_arch(char **cmdline_p)
880 check_efer(); 749 check_efer();
881#endif 750#endif
882 751
883#if defined(CONFIG_VMI) && defined(CONFIG_X86_32) 752 /* Must be before kernel pagetables are setup */
884 /* 753 vmi_activate();
885 * Must be before kernel pagetables are setup
886 * or fixmap area is touched.
887 */
888 vmi_init();
889#endif
890 754
891 /* after early param, so could get panic from serial */ 755 /* after early param, so could get panic from serial */
892 reserve_early_setup_data(); 756 reserve_early_setup_data();
@@ -909,6 +773,12 @@ void __init setup_arch(char **cmdline_p)
909 773
910 dmi_check_system(bad_bios_dmi_table); 774 dmi_check_system(bad_bios_dmi_table);
911 775
776 /*
777 * VMware detection requires dmi to be available, so this
778 * needs to be done after dmi_scan_machine, for the BP.
779 */
780 init_hypervisor(&boot_cpu_data);
781
912#ifdef CONFIG_X86_32 782#ifdef CONFIG_X86_32
913 probe_roms(); 783 probe_roms();
914#endif 784#endif
diff --git a/arch/x86/kernel/sigframe.h b/arch/x86/kernel/sigframe.h
deleted file mode 100644
index cc673aa55ce4..000000000000
--- a/arch/x86/kernel/sigframe.h
+++ /dev/null
@@ -1,42 +0,0 @@
1#ifdef CONFIG_X86_32
2struct sigframe {
3 char __user *pretcode;
4 int sig;
5 struct sigcontext sc;
6 /*
7 * fpstate is unused. fpstate is moved/allocated after
8 * retcode[] below. This movement allows to have the FP state and the
9 * future state extensions (xsave) stay together.
10 * And at the same time retaining the unused fpstate, prevents changing
11 * the offset of extramask[] in the sigframe and thus prevent any
12 * legacy application accessing/modifying it.
13 */
14 struct _fpstate fpstate_unused;
15 unsigned long extramask[_NSIG_WORDS-1];
16 char retcode[8];
17 /* fp state follows here */
18};
19
20struct rt_sigframe {
21 char __user *pretcode;
22 int sig;
23 struct siginfo __user *pinfo;
24 void __user *puc;
25 struct siginfo info;
26 struct ucontext uc;
27 char retcode[8];
28 /* fp state follows here */
29};
30#else
31struct rt_sigframe {
32 char __user *pretcode;
33 struct ucontext uc;
34 struct siginfo info;
35 /* fp state follows here */
36};
37
38int ia32_setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info,
39 sigset_t *set, struct pt_regs *regs);
40int ia32_setup_frame(int sig, struct k_sigaction *ka,
41 sigset_t *set, struct pt_regs *regs);
42#endif
diff --git a/arch/x86/kernel/signal_32.c b/arch/x86/kernel/signal.c
index d6dd057d0f22..89bb7668041d 100644
--- a/arch/x86/kernel/signal_32.c
+++ b/arch/x86/kernel/signal.c
@@ -1,36 +1,41 @@
1/* 1/*
2 * Copyright (C) 1991, 1992 Linus Torvalds 2 * Copyright (C) 1991, 1992 Linus Torvalds
3 * Copyright (C) 2000, 2001, 2002 Andi Kleen SuSE Labs
3 * 4 *
4 * 1997-11-28 Modified for POSIX.1b signals by Richard Henderson 5 * 1997-11-28 Modified for POSIX.1b signals by Richard Henderson
5 * 2000-06-20 Pentium III FXSR, SSE support by Gareth Hughes 6 * 2000-06-20 Pentium III FXSR, SSE support by Gareth Hughes
7 * 2000-2002 x86-64 support by Andi Kleen
6 */ 8 */
7#include <linux/list.h>
8 9
9#include <linux/personality.h> 10#include <linux/sched.h>
10#include <linux/binfmts.h> 11#include <linux/mm.h>
11#include <linux/suspend.h> 12#include <linux/smp.h>
12#include <linux/kernel.h> 13#include <linux/kernel.h>
13#include <linux/ptrace.h>
14#include <linux/signal.h> 14#include <linux/signal.h>
15#include <linux/stddef.h>
16#include <linux/unistd.h>
17#include <linux/errno.h> 15#include <linux/errno.h>
18#include <linux/sched.h>
19#include <linux/wait.h> 16#include <linux/wait.h>
17#include <linux/ptrace.h>
20#include <linux/tracehook.h> 18#include <linux/tracehook.h>
21#include <linux/elf.h> 19#include <linux/unistd.h>
22#include <linux/smp.h> 20#include <linux/stddef.h>
23#include <linux/mm.h> 21#include <linux/personality.h>
22#include <linux/uaccess.h>
24 23
25#include <asm/processor.h> 24#include <asm/processor.h>
26#include <asm/ucontext.h> 25#include <asm/ucontext.h>
27#include <asm/uaccess.h>
28#include <asm/i387.h> 26#include <asm/i387.h>
29#include <asm/vdso.h> 27#include <asm/vdso.h>
28
29#ifdef CONFIG_X86_64
30#include <asm/proto.h>
31#include <asm/ia32_unistd.h>
32#include <asm/mce.h>
33#endif /* CONFIG_X86_64 */
34
30#include <asm/syscall.h> 35#include <asm/syscall.h>
31#include <asm/syscalls.h> 36#include <asm/syscalls.h>
32 37
33#include "sigframe.h" 38#include <asm/sigframe.h>
34 39
35#define _BLOCKABLE (~(sigmask(SIGKILL) | sigmask(SIGSTOP))) 40#define _BLOCKABLE (~(sigmask(SIGKILL) | sigmask(SIGSTOP)))
36 41
@@ -45,74 +50,6 @@
45# define FIX_EFLAGS __FIX_EFLAGS 50# define FIX_EFLAGS __FIX_EFLAGS
46#endif 51#endif
47 52
48/*
49 * Atomically swap in the new signal mask, and wait for a signal.
50 */
51asmlinkage int
52sys_sigsuspend(int history0, int history1, old_sigset_t mask)
53{
54 mask &= _BLOCKABLE;
55 spin_lock_irq(&current->sighand->siglock);
56 current->saved_sigmask = current->blocked;
57 siginitset(&current->blocked, mask);
58 recalc_sigpending();
59 spin_unlock_irq(&current->sighand->siglock);
60
61 current->state = TASK_INTERRUPTIBLE;
62 schedule();
63 set_restore_sigmask();
64
65 return -ERESTARTNOHAND;
66}
67
68asmlinkage int
69sys_sigaction(int sig, const struct old_sigaction __user *act,
70 struct old_sigaction __user *oact)
71{
72 struct k_sigaction new_ka, old_ka;
73 int ret;
74
75 if (act) {
76 old_sigset_t mask;
77
78 if (!access_ok(VERIFY_READ, act, sizeof(*act)) ||
79 __get_user(new_ka.sa.sa_handler, &act->sa_handler) ||
80 __get_user(new_ka.sa.sa_restorer, &act->sa_restorer))
81 return -EFAULT;
82
83 __get_user(new_ka.sa.sa_flags, &act->sa_flags);
84 __get_user(mask, &act->sa_mask);
85 siginitset(&new_ka.sa.sa_mask, mask);
86 }
87
88 ret = do_sigaction(sig, act ? &new_ka : NULL, oact ? &old_ka : NULL);
89
90 if (!ret && oact) {
91 if (!access_ok(VERIFY_WRITE, oact, sizeof(*oact)) ||
92 __put_user(old_ka.sa.sa_handler, &oact->sa_handler) ||
93 __put_user(old_ka.sa.sa_restorer, &oact->sa_restorer))
94 return -EFAULT;
95
96 __put_user(old_ka.sa.sa_flags, &oact->sa_flags);
97 __put_user(old_ka.sa.sa_mask.sig[0], &oact->sa_mask);
98 }
99
100 return ret;
101}
102
103asmlinkage int sys_sigaltstack(unsigned long bx)
104{
105 /*
106 * This is needed to make gcc realize it doesn't own the
107 * "struct pt_regs"
108 */
109 struct pt_regs *regs = (struct pt_regs *)&bx;
110 const stack_t __user *uss = (const stack_t __user *)bx;
111 stack_t __user *uoss = (stack_t __user *)regs->cx;
112
113 return do_sigaltstack(uss, uoss, regs->sp);
114}
115
116#define COPY(x) { \ 53#define COPY(x) { \
117 err |= __get_user(regs->x, &sc->x); \ 54 err |= __get_user(regs->x, &sc->x); \
118} 55}
@@ -123,7 +60,7 @@ asmlinkage int sys_sigaltstack(unsigned long bx)
123 regs->seg = tmp; \ 60 regs->seg = tmp; \
124} 61}
125 62
126#define COPY_SEG_STRICT(seg) { \ 63#define COPY_SEG_CPL3(seg) { \
127 unsigned short tmp; \ 64 unsigned short tmp; \
128 err |= __get_user(tmp, &sc->seg); \ 65 err |= __get_user(tmp, &sc->seg); \
129 regs->seg = tmp | 3; \ 66 regs->seg = tmp | 3; \
@@ -135,9 +72,6 @@ asmlinkage int sys_sigaltstack(unsigned long bx)
135 loadsegment(seg, tmp); \ 72 loadsegment(seg, tmp); \
136} 73}
137 74
138/*
139 * Do a signal return; undo the signal stack.
140 */
141static int 75static int
142restore_sigcontext(struct pt_regs *regs, struct sigcontext __user *sc, 76restore_sigcontext(struct pt_regs *regs, struct sigcontext __user *sc,
143 unsigned long *pax) 77 unsigned long *pax)
@@ -149,14 +83,36 @@ restore_sigcontext(struct pt_regs *regs, struct sigcontext __user *sc,
149 /* Always make any pending restarted system calls return -EINTR */ 83 /* Always make any pending restarted system calls return -EINTR */
150 current_thread_info()->restart_block.fn = do_no_restart_syscall; 84 current_thread_info()->restart_block.fn = do_no_restart_syscall;
151 85
86#ifdef CONFIG_X86_32
152 GET_SEG(gs); 87 GET_SEG(gs);
153 COPY_SEG(fs); 88 COPY_SEG(fs);
154 COPY_SEG(es); 89 COPY_SEG(es);
155 COPY_SEG(ds); 90 COPY_SEG(ds);
91#endif /* CONFIG_X86_32 */
92
156 COPY(di); COPY(si); COPY(bp); COPY(sp); COPY(bx); 93 COPY(di); COPY(si); COPY(bp); COPY(sp); COPY(bx);
157 COPY(dx); COPY(cx); COPY(ip); 94 COPY(dx); COPY(cx); COPY(ip);
158 COPY_SEG_STRICT(cs); 95
159 COPY_SEG_STRICT(ss); 96#ifdef CONFIG_X86_64
97 COPY(r8);
98 COPY(r9);
99 COPY(r10);
100 COPY(r11);
101 COPY(r12);
102 COPY(r13);
103 COPY(r14);
104 COPY(r15);
105#endif /* CONFIG_X86_64 */
106
107#ifdef CONFIG_X86_32
108 COPY_SEG_CPL3(cs);
109 COPY_SEG_CPL3(ss);
110#else /* !CONFIG_X86_32 */
111 /* Kernel saves and restores only the CS segment register on signals,
112 * which is the bare minimum needed to allow mixed 32/64-bit code.
113 * App's signal handler can save/restore other segments if needed. */
114 COPY_SEG_CPL3(cs);
115#endif /* CONFIG_X86_32 */
160 116
161 err |= __get_user(tmpflags, &sc->flags); 117 err |= __get_user(tmpflags, &sc->flags);
162 regs->flags = (regs->flags & ~FIX_EFLAGS) | (tmpflags & FIX_EFLAGS); 118 regs->flags = (regs->flags & ~FIX_EFLAGS) | (tmpflags & FIX_EFLAGS);
@@ -169,102 +125,24 @@ restore_sigcontext(struct pt_regs *regs, struct sigcontext __user *sc,
169 return err; 125 return err;
170} 126}
171 127
172asmlinkage unsigned long sys_sigreturn(unsigned long __unused)
173{
174 struct sigframe __user *frame;
175 struct pt_regs *regs;
176 unsigned long ax;
177 sigset_t set;
178
179 regs = (struct pt_regs *) &__unused;
180 frame = (struct sigframe __user *)(regs->sp - 8);
181
182 if (!access_ok(VERIFY_READ, frame, sizeof(*frame)))
183 goto badframe;
184 if (__get_user(set.sig[0], &frame->sc.oldmask) || (_NSIG_WORDS > 1
185 && __copy_from_user(&set.sig[1], &frame->extramask,
186 sizeof(frame->extramask))))
187 goto badframe;
188
189 sigdelsetmask(&set, ~_BLOCKABLE);
190 spin_lock_irq(&current->sighand->siglock);
191 current->blocked = set;
192 recalc_sigpending();
193 spin_unlock_irq(&current->sighand->siglock);
194
195 if (restore_sigcontext(regs, &frame->sc, &ax))
196 goto badframe;
197 return ax;
198
199badframe:
200 if (show_unhandled_signals && printk_ratelimit()) {
201 printk("%s%s[%d] bad frame in sigreturn frame:"
202 "%p ip:%lx sp:%lx oeax:%lx",
203 task_pid_nr(current) > 1 ? KERN_INFO : KERN_EMERG,
204 current->comm, task_pid_nr(current), frame, regs->ip,
205 regs->sp, regs->orig_ax);
206 print_vma_addr(" in ", regs->ip);
207 printk(KERN_CONT "\n");
208 }
209
210 force_sig(SIGSEGV, current);
211
212 return 0;
213}
214
215static long do_rt_sigreturn(struct pt_regs *regs)
216{
217 struct rt_sigframe __user *frame;
218 unsigned long ax;
219 sigset_t set;
220
221 frame = (struct rt_sigframe __user *)(regs->sp - sizeof(long));
222 if (!access_ok(VERIFY_READ, frame, sizeof(*frame)))
223 goto badframe;
224 if (__copy_from_user(&set, &frame->uc.uc_sigmask, sizeof(set)))
225 goto badframe;
226
227 sigdelsetmask(&set, ~_BLOCKABLE);
228 spin_lock_irq(&current->sighand->siglock);
229 current->blocked = set;
230 recalc_sigpending();
231 spin_unlock_irq(&current->sighand->siglock);
232
233 if (restore_sigcontext(regs, &frame->uc.uc_mcontext, &ax))
234 goto badframe;
235
236 if (do_sigaltstack(&frame->uc.uc_stack, NULL, regs->sp) == -EFAULT)
237 goto badframe;
238
239 return ax;
240
241badframe:
242 signal_fault(regs, frame, "rt_sigreturn");
243 return 0;
244}
245
246asmlinkage int sys_rt_sigreturn(unsigned long __unused)
247{
248 struct pt_regs *regs = (struct pt_regs *)&__unused;
249
250 return do_rt_sigreturn(regs);
251}
252
253/*
254 * Set up a signal frame.
255 */
256static int 128static int
257setup_sigcontext(struct sigcontext __user *sc, void __user *fpstate, 129setup_sigcontext(struct sigcontext __user *sc, void __user *fpstate,
258 struct pt_regs *regs, unsigned long mask) 130 struct pt_regs *regs, unsigned long mask)
259{ 131{
260 int tmp, err = 0; 132 int err = 0;
261 133
262 err |= __put_user(regs->fs, (unsigned int __user *)&sc->fs); 134#ifdef CONFIG_X86_32
263 savesegment(gs, tmp); 135 {
264 err |= __put_user(tmp, (unsigned int __user *)&sc->gs); 136 unsigned int tmp;
265 137
138 savesegment(gs, tmp);
139 err |= __put_user(tmp, (unsigned int __user *)&sc->gs);
140 }
141 err |= __put_user(regs->fs, (unsigned int __user *)&sc->fs);
266 err |= __put_user(regs->es, (unsigned int __user *)&sc->es); 142 err |= __put_user(regs->es, (unsigned int __user *)&sc->es);
267 err |= __put_user(regs->ds, (unsigned int __user *)&sc->ds); 143 err |= __put_user(regs->ds, (unsigned int __user *)&sc->ds);
144#endif /* CONFIG_X86_32 */
145
268 err |= __put_user(regs->di, &sc->di); 146 err |= __put_user(regs->di, &sc->di);
269 err |= __put_user(regs->si, &sc->si); 147 err |= __put_user(regs->si, &sc->si);
270 err |= __put_user(regs->bp, &sc->bp); 148 err |= __put_user(regs->bp, &sc->bp);
@@ -273,19 +151,33 @@ setup_sigcontext(struct sigcontext __user *sc, void __user *fpstate,
273 err |= __put_user(regs->dx, &sc->dx); 151 err |= __put_user(regs->dx, &sc->dx);
274 err |= __put_user(regs->cx, &sc->cx); 152 err |= __put_user(regs->cx, &sc->cx);
275 err |= __put_user(regs->ax, &sc->ax); 153 err |= __put_user(regs->ax, &sc->ax);
154#ifdef CONFIG_X86_64
155 err |= __put_user(regs->r8, &sc->r8);
156 err |= __put_user(regs->r9, &sc->r9);
157 err |= __put_user(regs->r10, &sc->r10);
158 err |= __put_user(regs->r11, &sc->r11);
159 err |= __put_user(regs->r12, &sc->r12);
160 err |= __put_user(regs->r13, &sc->r13);
161 err |= __put_user(regs->r14, &sc->r14);
162 err |= __put_user(regs->r15, &sc->r15);
163#endif /* CONFIG_X86_64 */
164
276 err |= __put_user(current->thread.trap_no, &sc->trapno); 165 err |= __put_user(current->thread.trap_no, &sc->trapno);
277 err |= __put_user(current->thread.error_code, &sc->err); 166 err |= __put_user(current->thread.error_code, &sc->err);
278 err |= __put_user(regs->ip, &sc->ip); 167 err |= __put_user(regs->ip, &sc->ip);
168#ifdef CONFIG_X86_32
279 err |= __put_user(regs->cs, (unsigned int __user *)&sc->cs); 169 err |= __put_user(regs->cs, (unsigned int __user *)&sc->cs);
280 err |= __put_user(regs->flags, &sc->flags); 170 err |= __put_user(regs->flags, &sc->flags);
281 err |= __put_user(regs->sp, &sc->sp_at_signal); 171 err |= __put_user(regs->sp, &sc->sp_at_signal);
282 err |= __put_user(regs->ss, (unsigned int __user *)&sc->ss); 172 err |= __put_user(regs->ss, (unsigned int __user *)&sc->ss);
173#else /* !CONFIG_X86_32 */
174 err |= __put_user(regs->flags, &sc->flags);
175 err |= __put_user(regs->cs, &sc->cs);
176 err |= __put_user(0, &sc->gs);
177 err |= __put_user(0, &sc->fs);
178#endif /* CONFIG_X86_32 */
283 179
284 tmp = save_i387_xstate(fpstate); 180 err |= __put_user(fpstate, &sc->fpstate);
285 if (tmp < 0)
286 err = 1;
287 else
288 err |= __put_user(tmp ? fpstate : NULL, &sc->fpstate);
289 181
290 /* non-iBCS2 extensions.. */ 182 /* non-iBCS2 extensions.. */
291 err |= __put_user(mask, &sc->oldmask); 183 err |= __put_user(mask, &sc->oldmask);
@@ -295,6 +187,32 @@ setup_sigcontext(struct sigcontext __user *sc, void __user *fpstate,
295} 187}
296 188
297/* 189/*
190 * Set up a signal frame.
191 */
192#ifdef CONFIG_X86_32
193static const struct {
194 u16 poplmovl;
195 u32 val;
196 u16 int80;
197} __attribute__((packed)) retcode = {
198 0xb858, /* popl %eax; movl $..., %eax */
199 __NR_sigreturn,
200 0x80cd, /* int $0x80 */
201};
202
203static const struct {
204 u8 movl;
205 u32 val;
206 u16 int80;
207 u8 pad;
208} __attribute__((packed)) rt_retcode = {
209 0xb8, /* movl $..., %eax */
210 __NR_rt_sigreturn,
211 0x80cd, /* int $0x80 */
212 0
213};
214
215/*
298 * Determine which stack to use.. 216 * Determine which stack to use..
299 */ 217 */
300static inline void __user * 218static inline void __user *
@@ -328,6 +246,8 @@ get_sigframe(struct k_sigaction *ka, struct pt_regs *regs, size_t frame_size,
328 if (used_math()) { 246 if (used_math()) {
329 sp = sp - sig_xstate_size; 247 sp = sp - sig_xstate_size;
330 *fpstate = (struct _fpstate *) sp; 248 *fpstate = (struct _fpstate *) sp;
249 if (save_i387_xstate(*fpstate) < 0)
250 return (void __user *)-1L;
331 } 251 }
332 252
333 sp -= frame_size; 253 sp -= frame_size;
@@ -383,9 +303,7 @@ __setup_frame(int sig, struct k_sigaction *ka, sigset_t *set,
383 * reasons and because gdb uses it as a signature to notice 303 * reasons and because gdb uses it as a signature to notice
384 * signal handler stack frames. 304 * signal handler stack frames.
385 */ 305 */
386 err |= __put_user(0xb858, (short __user *)(frame->retcode+0)); 306 err |= __put_user(*((u64 *)&retcode), (u64 *)frame->retcode);
387 err |= __put_user(__NR_sigreturn, (int __user *)(frame->retcode+2));
388 err |= __put_user(0x80cd, (short __user *)(frame->retcode+6));
389 307
390 if (err) 308 if (err)
391 return -EFAULT; 309 return -EFAULT;
@@ -454,9 +372,7 @@ static int __setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info,
454 * reasons and because gdb uses it as a signature to notice 372 * reasons and because gdb uses it as a signature to notice
455 * signal handler stack frames. 373 * signal handler stack frames.
456 */ 374 */
457 err |= __put_user(0xb8, (char __user *)(frame->retcode+0)); 375 err |= __put_user(*((u64 *)&rt_retcode), (u64 *)frame->retcode);
458 err |= __put_user(__NR_rt_sigreturn, (int __user *)(frame->retcode+1));
459 err |= __put_user(0x80cd, (short __user *)(frame->retcode+5));
460 376
461 if (err) 377 if (err)
462 return -EFAULT; 378 return -EFAULT;
@@ -475,23 +391,293 @@ static int __setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info,
475 391
476 return 0; 392 return 0;
477} 393}
394#else /* !CONFIG_X86_32 */
395/*
396 * Determine which stack to use..
397 */
398static void __user *
399get_stack(struct k_sigaction *ka, unsigned long sp, unsigned long size)
400{
401 /* Default to using normal stack - redzone*/
402 sp -= 128;
403
404 /* This is the X/Open sanctioned signal stack switching. */
405 if (ka->sa.sa_flags & SA_ONSTACK) {
406 if (sas_ss_flags(sp) == 0)
407 sp = current->sas_ss_sp + current->sas_ss_size;
408 }
409
410 return (void __user *)round_down(sp - size, 64);
411}
412
413static int __setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info,
414 sigset_t *set, struct pt_regs *regs)
415{
416 struct rt_sigframe __user *frame;
417 void __user *fp = NULL;
418 int err = 0;
419 struct task_struct *me = current;
420
421 if (used_math()) {
422 fp = get_stack(ka, regs->sp, sig_xstate_size);
423 frame = (void __user *)round_down(
424 (unsigned long)fp - sizeof(struct rt_sigframe), 16) - 8;
425
426 if (save_i387_xstate(fp) < 0)
427 return -EFAULT;
428 } else
429 frame = get_stack(ka, regs->sp, sizeof(struct rt_sigframe)) - 8;
430
431 if (!access_ok(VERIFY_WRITE, frame, sizeof(*frame)))
432 return -EFAULT;
433
434 if (ka->sa.sa_flags & SA_SIGINFO) {
435 if (copy_siginfo_to_user(&frame->info, info))
436 return -EFAULT;
437 }
438
439 /* Create the ucontext. */
440 if (cpu_has_xsave)
441 err |= __put_user(UC_FP_XSTATE, &frame->uc.uc_flags);
442 else
443 err |= __put_user(0, &frame->uc.uc_flags);
444 err |= __put_user(0, &frame->uc.uc_link);
445 err |= __put_user(me->sas_ss_sp, &frame->uc.uc_stack.ss_sp);
446 err |= __put_user(sas_ss_flags(regs->sp),
447 &frame->uc.uc_stack.ss_flags);
448 err |= __put_user(me->sas_ss_size, &frame->uc.uc_stack.ss_size);
449 err |= setup_sigcontext(&frame->uc.uc_mcontext, fp, regs, set->sig[0]);
450 err |= __copy_to_user(&frame->uc.uc_sigmask, set, sizeof(*set));
451
452 /* Set up to return from userspace. If provided, use a stub
453 already in userspace. */
454 /* x86-64 should always use SA_RESTORER. */
455 if (ka->sa.sa_flags & SA_RESTORER) {
456 err |= __put_user(ka->sa.sa_restorer, &frame->pretcode);
457 } else {
458 /* could use a vstub here */
459 return -EFAULT;
460 }
461
462 if (err)
463 return -EFAULT;
464
465 /* Set up registers for signal handler */
466 regs->di = sig;
467 /* In case the signal handler was declared without prototypes */
468 regs->ax = 0;
469
470 /* This also works for non SA_SIGINFO handlers because they expect the
471 next argument after the signal number on the stack. */
472 regs->si = (unsigned long)&frame->info;
473 regs->dx = (unsigned long)&frame->uc;
474 regs->ip = (unsigned long) ka->sa.sa_handler;
475
476 regs->sp = (unsigned long)frame;
477
478 /* Set up the CS register to run signal handlers in 64-bit mode,
479 even if the handler happens to be interrupting 32-bit code. */
480 regs->cs = __USER_CS;
481
482 return 0;
483}
484#endif /* CONFIG_X86_32 */
485
486#ifdef CONFIG_X86_32
487/*
488 * Atomically swap in the new signal mask, and wait for a signal.
489 */
490asmlinkage int
491sys_sigsuspend(int history0, int history1, old_sigset_t mask)
492{
493 mask &= _BLOCKABLE;
494 spin_lock_irq(&current->sighand->siglock);
495 current->saved_sigmask = current->blocked;
496 siginitset(&current->blocked, mask);
497 recalc_sigpending();
498 spin_unlock_irq(&current->sighand->siglock);
499
500 current->state = TASK_INTERRUPTIBLE;
501 schedule();
502 set_restore_sigmask();
503
504 return -ERESTARTNOHAND;
505}
506
507asmlinkage int
508sys_sigaction(int sig, const struct old_sigaction __user *act,
509 struct old_sigaction __user *oact)
510{
511 struct k_sigaction new_ka, old_ka;
512 int ret;
513
514 if (act) {
515 old_sigset_t mask;
516
517 if (!access_ok(VERIFY_READ, act, sizeof(*act)) ||
518 __get_user(new_ka.sa.sa_handler, &act->sa_handler) ||
519 __get_user(new_ka.sa.sa_restorer, &act->sa_restorer))
520 return -EFAULT;
521
522 __get_user(new_ka.sa.sa_flags, &act->sa_flags);
523 __get_user(mask, &act->sa_mask);
524 siginitset(&new_ka.sa.sa_mask, mask);
525 }
526
527 ret = do_sigaction(sig, act ? &new_ka : NULL, oact ? &old_ka : NULL);
528
529 if (!ret && oact) {
530 if (!access_ok(VERIFY_WRITE, oact, sizeof(*oact)) ||
531 __put_user(old_ka.sa.sa_handler, &oact->sa_handler) ||
532 __put_user(old_ka.sa.sa_restorer, &oact->sa_restorer))
533 return -EFAULT;
534
535 __put_user(old_ka.sa.sa_flags, &oact->sa_flags);
536 __put_user(old_ka.sa.sa_mask.sig[0], &oact->sa_mask);
537 }
538
539 return ret;
540}
541#endif /* CONFIG_X86_32 */
542
543#ifdef CONFIG_X86_32
544asmlinkage int sys_sigaltstack(unsigned long bx)
545{
546 /*
547 * This is needed to make gcc realize it doesn't own the
548 * "struct pt_regs"
549 */
550 struct pt_regs *regs = (struct pt_regs *)&bx;
551 const stack_t __user *uss = (const stack_t __user *)bx;
552 stack_t __user *uoss = (stack_t __user *)regs->cx;
553
554 return do_sigaltstack(uss, uoss, regs->sp);
555}
556#else /* !CONFIG_X86_32 */
557asmlinkage long
558sys_sigaltstack(const stack_t __user *uss, stack_t __user *uoss,
559 struct pt_regs *regs)
560{
561 return do_sigaltstack(uss, uoss, regs->sp);
562}
563#endif /* CONFIG_X86_32 */
564
565/*
566 * Do a signal return; undo the signal stack.
567 */
568#ifdef CONFIG_X86_32
569asmlinkage unsigned long sys_sigreturn(unsigned long __unused)
570{
571 struct sigframe __user *frame;
572 struct pt_regs *regs;
573 unsigned long ax;
574 sigset_t set;
575
576 regs = (struct pt_regs *) &__unused;
577 frame = (struct sigframe __user *)(regs->sp - 8);
578
579 if (!access_ok(VERIFY_READ, frame, sizeof(*frame)))
580 goto badframe;
581 if (__get_user(set.sig[0], &frame->sc.oldmask) || (_NSIG_WORDS > 1
582 && __copy_from_user(&set.sig[1], &frame->extramask,
583 sizeof(frame->extramask))))
584 goto badframe;
585
586 sigdelsetmask(&set, ~_BLOCKABLE);
587 spin_lock_irq(&current->sighand->siglock);
588 current->blocked = set;
589 recalc_sigpending();
590 spin_unlock_irq(&current->sighand->siglock);
591
592 if (restore_sigcontext(regs, &frame->sc, &ax))
593 goto badframe;
594 return ax;
595
596badframe:
597 signal_fault(regs, frame, "sigreturn");
598
599 return 0;
600}
601#endif /* CONFIG_X86_32 */
602
603static long do_rt_sigreturn(struct pt_regs *regs)
604{
605 struct rt_sigframe __user *frame;
606 unsigned long ax;
607 sigset_t set;
608
609 frame = (struct rt_sigframe __user *)(regs->sp - sizeof(long));
610 if (!access_ok(VERIFY_READ, frame, sizeof(*frame)))
611 goto badframe;
612 if (__copy_from_user(&set, &frame->uc.uc_sigmask, sizeof(set)))
613 goto badframe;
614
615 sigdelsetmask(&set, ~_BLOCKABLE);
616 spin_lock_irq(&current->sighand->siglock);
617 current->blocked = set;
618 recalc_sigpending();
619 spin_unlock_irq(&current->sighand->siglock);
620
621 if (restore_sigcontext(regs, &frame->uc.uc_mcontext, &ax))
622 goto badframe;
623
624 if (do_sigaltstack(&frame->uc.uc_stack, NULL, regs->sp) == -EFAULT)
625 goto badframe;
626
627 return ax;
628
629badframe:
630 signal_fault(regs, frame, "rt_sigreturn");
631 return 0;
632}
633
634#ifdef CONFIG_X86_32
635asmlinkage int sys_rt_sigreturn(struct pt_regs regs)
636{
637 return do_rt_sigreturn(&regs);
638}
639#else /* !CONFIG_X86_32 */
640asmlinkage long sys_rt_sigreturn(struct pt_regs *regs)
641{
642 return do_rt_sigreturn(regs);
643}
644#endif /* CONFIG_X86_32 */
478 645
479/* 646/*
480 * OK, we're invoking a handler: 647 * OK, we're invoking a handler:
481 */ 648 */
482static int signr_convert(int sig) 649static int signr_convert(int sig)
483{ 650{
651#ifdef CONFIG_X86_32
484 struct thread_info *info = current_thread_info(); 652 struct thread_info *info = current_thread_info();
485 653
486 if (info->exec_domain && info->exec_domain->signal_invmap && sig < 32) 654 if (info->exec_domain && info->exec_domain->signal_invmap && sig < 32)
487 return info->exec_domain->signal_invmap[sig]; 655 return info->exec_domain->signal_invmap[sig];
656#endif /* CONFIG_X86_32 */
488 return sig; 657 return sig;
489} 658}
490 659
660#ifdef CONFIG_X86_32
661
491#define is_ia32 1 662#define is_ia32 1
492#define ia32_setup_frame __setup_frame 663#define ia32_setup_frame __setup_frame
493#define ia32_setup_rt_frame __setup_rt_frame 664#define ia32_setup_rt_frame __setup_rt_frame
494 665
666#else /* !CONFIG_X86_32 */
667
668#ifdef CONFIG_IA32_EMULATION
669#define is_ia32 test_thread_flag(TIF_IA32)
670#else /* !CONFIG_IA32_EMULATION */
671#define is_ia32 0
672#endif /* CONFIG_IA32_EMULATION */
673
674int ia32_setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info,
675 sigset_t *set, struct pt_regs *regs);
676int ia32_setup_frame(int sig, struct k_sigaction *ka,
677 sigset_t *set, struct pt_regs *regs);
678
679#endif /* CONFIG_X86_32 */
680
495static int 681static int
496setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info, 682setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info,
497 sigset_t *set, struct pt_regs *regs) 683 sigset_t *set, struct pt_regs *regs)
@@ -592,7 +778,13 @@ handle_signal(unsigned long sig, siginfo_t *info, struct k_sigaction *ka,
592 return 0; 778 return 0;
593} 779}
594 780
781#ifdef CONFIG_X86_32
595#define NR_restart_syscall __NR_restart_syscall 782#define NR_restart_syscall __NR_restart_syscall
783#else /* !CONFIG_X86_32 */
784#define NR_restart_syscall \
785 test_thread_flag(TIF_IA32) ? __NR_ia32_restart_syscall : __NR_restart_syscall
786#endif /* CONFIG_X86_32 */
787
596/* 788/*
597 * Note that 'init' is a special process: it doesn't get signals it doesn't 789 * Note that 'init' is a special process: it doesn't get signals it doesn't
598 * want to handle. Thus you cannot kill init even with a SIGKILL even by 790 * want to handle. Thus you cannot kill init even with a SIGKILL even by
@@ -704,8 +896,9 @@ void signal_fault(struct pt_regs *regs, void __user *frame, char *where)
704 struct task_struct *me = current; 896 struct task_struct *me = current;
705 897
706 if (show_unhandled_signals && printk_ratelimit()) { 898 if (show_unhandled_signals && printk_ratelimit()) {
707 printk(KERN_INFO 899 printk("%s"
708 "%s[%d] bad frame in %s frame:%p ip:%lx sp:%lx orax:%lx", 900 "%s[%d] bad frame in %s frame:%p ip:%lx sp:%lx orax:%lx",
901 task_pid_nr(current) > 1 ? KERN_INFO : KERN_EMERG,
709 me->comm, me->pid, where, frame, 902 me->comm, me->pid, where, frame,
710 regs->ip, regs->sp, regs->orig_ax); 903 regs->ip, regs->sp, regs->orig_ax);
711 print_vma_addr(" in ", regs->ip); 904 print_vma_addr(" in ", regs->ip);
diff --git a/arch/x86/kernel/signal_64.c b/arch/x86/kernel/signal_64.c
deleted file mode 100644
index a5c9627f4db9..000000000000
--- a/arch/x86/kernel/signal_64.c
+++ /dev/null
@@ -1,516 +0,0 @@
1/*
2 * Copyright (C) 1991, 1992 Linus Torvalds
3 * Copyright (C) 2000, 2001, 2002 Andi Kleen SuSE Labs
4 *
5 * 1997-11-28 Modified for POSIX.1b signals by Richard Henderson
6 * 2000-06-20 Pentium III FXSR, SSE support by Gareth Hughes
7 * 2000-2002 x86-64 support by Andi Kleen
8 */
9
10#include <linux/sched.h>
11#include <linux/mm.h>
12#include <linux/smp.h>
13#include <linux/kernel.h>
14#include <linux/signal.h>
15#include <linux/errno.h>
16#include <linux/wait.h>
17#include <linux/ptrace.h>
18#include <linux/tracehook.h>
19#include <linux/unistd.h>
20#include <linux/stddef.h>
21#include <linux/personality.h>
22#include <linux/compiler.h>
23#include <linux/uaccess.h>
24
25#include <asm/processor.h>
26#include <asm/ucontext.h>
27#include <asm/i387.h>
28#include <asm/proto.h>
29#include <asm/ia32_unistd.h>
30#include <asm/mce.h>
31#include <asm/syscall.h>
32#include <asm/syscalls.h>
33#include "sigframe.h"
34
35#define _BLOCKABLE (~(sigmask(SIGKILL) | sigmask(SIGSTOP)))
36
37#define __FIX_EFLAGS (X86_EFLAGS_AC | X86_EFLAGS_OF | \
38 X86_EFLAGS_DF | X86_EFLAGS_TF | X86_EFLAGS_SF | \
39 X86_EFLAGS_ZF | X86_EFLAGS_AF | X86_EFLAGS_PF | \
40 X86_EFLAGS_CF)
41
42#ifdef CONFIG_X86_32
43# define FIX_EFLAGS (__FIX_EFLAGS | X86_EFLAGS_RF)
44#else
45# define FIX_EFLAGS __FIX_EFLAGS
46#endif
47
48asmlinkage long
49sys_sigaltstack(const stack_t __user *uss, stack_t __user *uoss,
50 struct pt_regs *regs)
51{
52 return do_sigaltstack(uss, uoss, regs->sp);
53}
54
55#define COPY(x) { \
56 err |= __get_user(regs->x, &sc->x); \
57}
58
59#define COPY_SEG_STRICT(seg) { \
60 unsigned short tmp; \
61 err |= __get_user(tmp, &sc->seg); \
62 regs->seg = tmp | 3; \
63}
64
65/*
66 * Do a signal return; undo the signal stack.
67 */
68static int
69restore_sigcontext(struct pt_regs *regs, struct sigcontext __user *sc,
70 unsigned long *pax)
71{
72 void __user *buf;
73 unsigned int tmpflags;
74 unsigned int err = 0;
75
76 /* Always make any pending restarted system calls return -EINTR */
77 current_thread_info()->restart_block.fn = do_no_restart_syscall;
78
79 COPY(di); COPY(si); COPY(bp); COPY(sp); COPY(bx);
80 COPY(dx); COPY(cx); COPY(ip);
81 COPY(r8);
82 COPY(r9);
83 COPY(r10);
84 COPY(r11);
85 COPY(r12);
86 COPY(r13);
87 COPY(r14);
88 COPY(r15);
89
90 /* Kernel saves and restores only the CS segment register on signals,
91 * which is the bare minimum needed to allow mixed 32/64-bit code.
92 * App's signal handler can save/restore other segments if needed. */
93 COPY_SEG_STRICT(cs);
94
95 err |= __get_user(tmpflags, &sc->flags);
96 regs->flags = (regs->flags & ~FIX_EFLAGS) | (tmpflags & FIX_EFLAGS);
97 regs->orig_ax = -1; /* disable syscall checks */
98
99 err |= __get_user(buf, &sc->fpstate);
100 err |= restore_i387_xstate(buf);
101
102 err |= __get_user(*pax, &sc->ax);
103 return err;
104}
105
106static long do_rt_sigreturn(struct pt_regs *regs)
107{
108 struct rt_sigframe __user *frame;
109 unsigned long ax;
110 sigset_t set;
111
112 frame = (struct rt_sigframe __user *)(regs->sp - sizeof(long));
113 if (!access_ok(VERIFY_READ, frame, sizeof(*frame)))
114 goto badframe;
115 if (__copy_from_user(&set, &frame->uc.uc_sigmask, sizeof(set)))
116 goto badframe;
117
118 sigdelsetmask(&set, ~_BLOCKABLE);
119 spin_lock_irq(&current->sighand->siglock);
120 current->blocked = set;
121 recalc_sigpending();
122 spin_unlock_irq(&current->sighand->siglock);
123
124 if (restore_sigcontext(regs, &frame->uc.uc_mcontext, &ax))
125 goto badframe;
126
127 if (do_sigaltstack(&frame->uc.uc_stack, NULL, regs->sp) == -EFAULT)
128 goto badframe;
129
130 return ax;
131
132badframe:
133 signal_fault(regs, frame, "rt_sigreturn");
134 return 0;
135}
136
137asmlinkage long sys_rt_sigreturn(struct pt_regs *regs)
138{
139 return do_rt_sigreturn(regs);
140}
141
142/*
143 * Set up a signal frame.
144 */
145
146static inline int
147setup_sigcontext(struct sigcontext __user *sc, struct pt_regs *regs,
148 unsigned long mask, struct task_struct *me)
149{
150 int err = 0;
151
152 err |= __put_user(regs->cs, &sc->cs);
153 err |= __put_user(0, &sc->gs);
154 err |= __put_user(0, &sc->fs);
155
156 err |= __put_user(regs->di, &sc->di);
157 err |= __put_user(regs->si, &sc->si);
158 err |= __put_user(regs->bp, &sc->bp);
159 err |= __put_user(regs->sp, &sc->sp);
160 err |= __put_user(regs->bx, &sc->bx);
161 err |= __put_user(regs->dx, &sc->dx);
162 err |= __put_user(regs->cx, &sc->cx);
163 err |= __put_user(regs->ax, &sc->ax);
164 err |= __put_user(regs->r8, &sc->r8);
165 err |= __put_user(regs->r9, &sc->r9);
166 err |= __put_user(regs->r10, &sc->r10);
167 err |= __put_user(regs->r11, &sc->r11);
168 err |= __put_user(regs->r12, &sc->r12);
169 err |= __put_user(regs->r13, &sc->r13);
170 err |= __put_user(regs->r14, &sc->r14);
171 err |= __put_user(regs->r15, &sc->r15);
172 err |= __put_user(me->thread.trap_no, &sc->trapno);
173 err |= __put_user(me->thread.error_code, &sc->err);
174 err |= __put_user(regs->ip, &sc->ip);
175 err |= __put_user(regs->flags, &sc->flags);
176 err |= __put_user(mask, &sc->oldmask);
177 err |= __put_user(me->thread.cr2, &sc->cr2);
178
179 return err;
180}
181
182/*
183 * Determine which stack to use..
184 */
185
186static void __user *
187get_stack(struct k_sigaction *ka, struct pt_regs *regs, unsigned long size)
188{
189 unsigned long sp;
190
191 /* Default to using normal stack - redzone*/
192 sp = regs->sp - 128;
193
194 /* This is the X/Open sanctioned signal stack switching. */
195 if (ka->sa.sa_flags & SA_ONSTACK) {
196 if (sas_ss_flags(sp) == 0)
197 sp = current->sas_ss_sp + current->sas_ss_size;
198 }
199
200 return (void __user *)round_down(sp - size, 64);
201}
202
203static int __setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info,
204 sigset_t *set, struct pt_regs *regs)
205{
206 struct rt_sigframe __user *frame;
207 void __user *fp = NULL;
208 int err = 0;
209 struct task_struct *me = current;
210
211 if (used_math()) {
212 fp = get_stack(ka, regs, sig_xstate_size);
213 frame = (void __user *)round_down(
214 (unsigned long)fp - sizeof(struct rt_sigframe), 16) - 8;
215
216 if (save_i387_xstate(fp) < 0)
217 return -EFAULT;
218 } else
219 frame = get_stack(ka, regs, sizeof(struct rt_sigframe)) - 8;
220
221 if (!access_ok(VERIFY_WRITE, frame, sizeof(*frame)))
222 return -EFAULT;
223
224 if (ka->sa.sa_flags & SA_SIGINFO) {
225 if (copy_siginfo_to_user(&frame->info, info))
226 return -EFAULT;
227 }
228
229 /* Create the ucontext. */
230 if (cpu_has_xsave)
231 err |= __put_user(UC_FP_XSTATE, &frame->uc.uc_flags);
232 else
233 err |= __put_user(0, &frame->uc.uc_flags);
234 err |= __put_user(0, &frame->uc.uc_link);
235 err |= __put_user(me->sas_ss_sp, &frame->uc.uc_stack.ss_sp);
236 err |= __put_user(sas_ss_flags(regs->sp),
237 &frame->uc.uc_stack.ss_flags);
238 err |= __put_user(me->sas_ss_size, &frame->uc.uc_stack.ss_size);
239 err |= setup_sigcontext(&frame->uc.uc_mcontext, regs, set->sig[0], me);
240 err |= __put_user(fp, &frame->uc.uc_mcontext.fpstate);
241 if (sizeof(*set) == 16) {
242 __put_user(set->sig[0], &frame->uc.uc_sigmask.sig[0]);
243 __put_user(set->sig[1], &frame->uc.uc_sigmask.sig[1]);
244 } else
245 err |= __copy_to_user(&frame->uc.uc_sigmask, set, sizeof(*set));
246
247 /* Set up to return from userspace. If provided, use a stub
248 already in userspace. */
249 /* x86-64 should always use SA_RESTORER. */
250 if (ka->sa.sa_flags & SA_RESTORER) {
251 err |= __put_user(ka->sa.sa_restorer, &frame->pretcode);
252 } else {
253 /* could use a vstub here */
254 return -EFAULT;
255 }
256
257 if (err)
258 return -EFAULT;
259
260 /* Set up registers for signal handler */
261 regs->di = sig;
262 /* In case the signal handler was declared without prototypes */
263 regs->ax = 0;
264
265 /* This also works for non SA_SIGINFO handlers because they expect the
266 next argument after the signal number on the stack. */
267 regs->si = (unsigned long)&frame->info;
268 regs->dx = (unsigned long)&frame->uc;
269 regs->ip = (unsigned long) ka->sa.sa_handler;
270
271 regs->sp = (unsigned long)frame;
272
273 /* Set up the CS register to run signal handlers in 64-bit mode,
274 even if the handler happens to be interrupting 32-bit code. */
275 regs->cs = __USER_CS;
276
277 return 0;
278}
279
280/*
281 * OK, we're invoking a handler
282 */
283static int signr_convert(int sig)
284{
285 return sig;
286}
287
288#ifdef CONFIG_IA32_EMULATION
289#define is_ia32 test_thread_flag(TIF_IA32)
290#else
291#define is_ia32 0
292#endif
293
294static int
295setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info,
296 sigset_t *set, struct pt_regs *regs)
297{
298 int usig = signr_convert(sig);
299 int ret;
300
301 /* Set up the stack frame */
302 if (is_ia32) {
303 if (ka->sa.sa_flags & SA_SIGINFO)
304 ret = ia32_setup_rt_frame(usig, ka, info, set, regs);
305 else
306 ret = ia32_setup_frame(usig, ka, set, regs);
307 } else
308 ret = __setup_rt_frame(sig, ka, info, set, regs);
309
310 if (ret) {
311 force_sigsegv(sig, current);
312 return -EFAULT;
313 }
314
315 return ret;
316}
317
318static int
319handle_signal(unsigned long sig, siginfo_t *info, struct k_sigaction *ka,
320 sigset_t *oldset, struct pt_regs *regs)
321{
322 int ret;
323
324 /* Are we from a system call? */
325 if (syscall_get_nr(current, regs) >= 0) {
326 /* If so, check system call restarting.. */
327 switch (syscall_get_error(current, regs)) {
328 case -ERESTART_RESTARTBLOCK:
329 case -ERESTARTNOHAND:
330 regs->ax = -EINTR;
331 break;
332
333 case -ERESTARTSYS:
334 if (!(ka->sa.sa_flags & SA_RESTART)) {
335 regs->ax = -EINTR;
336 break;
337 }
338 /* fallthrough */
339 case -ERESTARTNOINTR:
340 regs->ax = regs->orig_ax;
341 regs->ip -= 2;
342 break;
343 }
344 }
345
346 /*
347 * If TF is set due to a debugger (TIF_FORCED_TF), clear the TF
348 * flag so that register information in the sigcontext is correct.
349 */
350 if (unlikely(regs->flags & X86_EFLAGS_TF) &&
351 likely(test_and_clear_thread_flag(TIF_FORCED_TF)))
352 regs->flags &= ~X86_EFLAGS_TF;
353
354 ret = setup_rt_frame(sig, ka, info, oldset, regs);
355
356 if (ret)
357 return ret;
358
359#ifdef CONFIG_X86_64
360 /*
361 * This has nothing to do with segment registers,
362 * despite the name. This magic affects uaccess.h
363 * macros' behavior. Reset it to the normal setting.
364 */
365 set_fs(USER_DS);
366#endif
367
368 /*
369 * Clear the direction flag as per the ABI for function entry.
370 */
371 regs->flags &= ~X86_EFLAGS_DF;
372
373 /*
374 * Clear TF when entering the signal handler, but
375 * notify any tracer that was single-stepping it.
376 * The tracer may want to single-step inside the
377 * handler too.
378 */
379 regs->flags &= ~X86_EFLAGS_TF;
380
381 spin_lock_irq(&current->sighand->siglock);
382 sigorsets(&current->blocked, &current->blocked, &ka->sa.sa_mask);
383 if (!(ka->sa.sa_flags & SA_NODEFER))
384 sigaddset(&current->blocked, sig);
385 recalc_sigpending();
386 spin_unlock_irq(&current->sighand->siglock);
387
388 tracehook_signal_handler(sig, info, ka, regs,
389 test_thread_flag(TIF_SINGLESTEP));
390
391 return 0;
392}
393
394#define NR_restart_syscall \
395 test_thread_flag(TIF_IA32) ? __NR_ia32_restart_syscall : __NR_restart_syscall
396/*
397 * Note that 'init' is a special process: it doesn't get signals it doesn't
398 * want to handle. Thus you cannot kill init even with a SIGKILL even by
399 * mistake.
400 */
401static void do_signal(struct pt_regs *regs)
402{
403 struct k_sigaction ka;
404 siginfo_t info;
405 int signr;
406 sigset_t *oldset;
407
408 /*
409 * We want the common case to go fast, which is why we may in certain
410 * cases get here from kernel mode. Just return without doing anything
411 * if so.
412 * X86_32: vm86 regs switched out by assembly code before reaching
413 * here, so testing against kernel CS suffices.
414 */
415 if (!user_mode(regs))
416 return;
417
418 if (current_thread_info()->status & TS_RESTORE_SIGMASK)
419 oldset = &current->saved_sigmask;
420 else
421 oldset = &current->blocked;
422
423 signr = get_signal_to_deliver(&info, &ka, regs, NULL);
424 if (signr > 0) {
425 /*
426 * Re-enable any watchpoints before delivering the
427 * signal to user space. The processor register will
428 * have been cleared if the watchpoint triggered
429 * inside the kernel.
430 */
431 if (current->thread.debugreg7)
432 set_debugreg(current->thread.debugreg7, 7);
433
434 /* Whee! Actually deliver the signal. */
435 if (handle_signal(signr, &info, &ka, oldset, regs) == 0) {
436 /*
437 * A signal was successfully delivered; the saved
438 * sigmask will have been stored in the signal frame,
439 * and will be restored by sigreturn, so we can simply
440 * clear the TS_RESTORE_SIGMASK flag.
441 */
442 current_thread_info()->status &= ~TS_RESTORE_SIGMASK;
443 }
444 return;
445 }
446
447 /* Did we come from a system call? */
448 if (syscall_get_nr(current, regs) >= 0) {
449 /* Restart the system call - no handlers present */
450 switch (syscall_get_error(current, regs)) {
451 case -ERESTARTNOHAND:
452 case -ERESTARTSYS:
453 case -ERESTARTNOINTR:
454 regs->ax = regs->orig_ax;
455 regs->ip -= 2;
456 break;
457
458 case -ERESTART_RESTARTBLOCK:
459 regs->ax = NR_restart_syscall;
460 regs->ip -= 2;
461 break;
462 }
463 }
464
465 /*
466 * If there's no signal to deliver, we just put the saved sigmask
467 * back.
468 */
469 if (current_thread_info()->status & TS_RESTORE_SIGMASK) {
470 current_thread_info()->status &= ~TS_RESTORE_SIGMASK;
471 sigprocmask(SIG_SETMASK, &current->saved_sigmask, NULL);
472 }
473}
474
475/*
476 * notification of userspace execution resumption
477 * - triggered by the TIF_WORK_MASK flags
478 */
479void
480do_notify_resume(struct pt_regs *regs, void *unused, __u32 thread_info_flags)
481{
482#if defined(CONFIG_X86_64) && defined(CONFIG_X86_MCE)
483 /* notify userspace of pending MCEs */
484 if (thread_info_flags & _TIF_MCE_NOTIFY)
485 mce_notify_user();
486#endif /* CONFIG_X86_64 && CONFIG_X86_MCE */
487
488 /* deal with pending signal delivery */
489 if (thread_info_flags & _TIF_SIGPENDING)
490 do_signal(regs);
491
492 if (thread_info_flags & _TIF_NOTIFY_RESUME) {
493 clear_thread_flag(TIF_NOTIFY_RESUME);
494 tracehook_notify_resume(regs);
495 }
496
497#ifdef CONFIG_X86_32
498 clear_thread_flag(TIF_IRET);
499#endif /* CONFIG_X86_32 */
500}
501
502void signal_fault(struct pt_regs *regs, void __user *frame, char *where)
503{
504 struct task_struct *me = current;
505
506 if (show_unhandled_signals && printk_ratelimit()) {
507 printk(KERN_INFO
508 "%s[%d] bad frame in %s frame:%p ip:%lx sp:%lx orax:%lx",
509 me->comm, me->pid, where, frame,
510 regs->ip, regs->sp, regs->orig_ax);
511 print_vma_addr(" in ", regs->ip);
512 printk(KERN_CONT "\n");
513 }
514
515 force_sig(SIGSEGV, me);
516}
diff --git a/arch/x86/kernel/smp.c b/arch/x86/kernel/smp.c
index 18f9b19f5f8f..3f92b134ab90 100644
--- a/arch/x86/kernel/smp.c
+++ b/arch/x86/kernel/smp.c
@@ -140,19 +140,6 @@ void native_send_call_func_ipi(cpumask_t mask)
140 send_IPI_mask(mask, CALL_FUNCTION_VECTOR); 140 send_IPI_mask(mask, CALL_FUNCTION_VECTOR);
141} 141}
142 142
143static void stop_this_cpu(void *dummy)
144{
145 local_irq_disable();
146 /*
147 * Remove this CPU:
148 */
149 cpu_clear(smp_processor_id(), cpu_online_map);
150 disable_local_APIC();
151 if (hlt_works(smp_processor_id()))
152 for (;;) halt();
153 for (;;);
154}
155
156/* 143/*
157 * this function calls the 'stop' function on all other CPUs in the system. 144 * this function calls the 'stop' function on all other CPUs in the system.
158 */ 145 */
diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c
index 7b1093397319..7a430c4d1551 100644
--- a/arch/x86/kernel/smpboot.c
+++ b/arch/x86/kernel/smpboot.c
@@ -62,6 +62,7 @@
62#include <asm/mtrr.h> 62#include <asm/mtrr.h>
63#include <asm/vmi.h> 63#include <asm/vmi.h>
64#include <asm/genapic.h> 64#include <asm/genapic.h>
65#include <asm/setup.h>
65#include <linux/mc146818rtc.h> 66#include <linux/mc146818rtc.h>
66 67
67#include <mach_apic.h> 68#include <mach_apic.h>
@@ -294,9 +295,7 @@ static void __cpuinit start_secondary(void *unused)
294 * fragile that we want to limit the things done here to the 295 * fragile that we want to limit the things done here to the
295 * most necessary things. 296 * most necessary things.
296 */ 297 */
297#ifdef CONFIG_VMI
298 vmi_bringup(); 298 vmi_bringup();
299#endif
300 cpu_init(); 299 cpu_init();
301 preempt_disable(); 300 preempt_disable();
302 smp_callin(); 301 smp_callin();
@@ -536,7 +535,7 @@ static void impress_friends(void)
536 pr_debug("Before bogocount - setting activated=1.\n"); 535 pr_debug("Before bogocount - setting activated=1.\n");
537} 536}
538 537
539static inline void __inquire_remote_apic(int apicid) 538void __inquire_remote_apic(int apicid)
540{ 539{
541 unsigned i, regs[] = { APIC_ID >> 4, APIC_LVR >> 4, APIC_SPIV >> 4 }; 540 unsigned i, regs[] = { APIC_ID >> 4, APIC_LVR >> 4, APIC_SPIV >> 4 };
542 char *names[] = { "ID", "VERSION", "SPIV" }; 541 char *names[] = { "ID", "VERSION", "SPIV" };
@@ -575,14 +574,13 @@ static inline void __inquire_remote_apic(int apicid)
575 } 574 }
576} 575}
577 576
578#ifdef WAKE_SECONDARY_VIA_NMI
579/* 577/*
580 * Poke the other CPU in the eye via NMI to wake it up. Remember that the normal 578 * Poke the other CPU in the eye via NMI to wake it up. Remember that the normal
581 * INIT, INIT, STARTUP sequence will reset the chip hard for us, and this 579 * INIT, INIT, STARTUP sequence will reset the chip hard for us, and this
582 * won't ... remember to clear down the APIC, etc later. 580 * won't ... remember to clear down the APIC, etc later.
583 */ 581 */
584static int __devinit 582int __devinit
585wakeup_secondary_cpu(int logical_apicid, unsigned long start_eip) 583wakeup_secondary_cpu_via_nmi(int logical_apicid, unsigned long start_eip)
586{ 584{
587 unsigned long send_status, accept_status = 0; 585 unsigned long send_status, accept_status = 0;
588 int maxlvt; 586 int maxlvt;
@@ -599,7 +597,7 @@ wakeup_secondary_cpu(int logical_apicid, unsigned long start_eip)
599 * Give the other CPU some time to accept the IPI. 597 * Give the other CPU some time to accept the IPI.
600 */ 598 */
601 udelay(200); 599 udelay(200);
602 if (APIC_INTEGRATED(apic_version[phys_apicid])) { 600 if (APIC_INTEGRATED(apic_version[boot_cpu_physical_apicid])) {
603 maxlvt = lapic_get_maxlvt(); 601 maxlvt = lapic_get_maxlvt();
604 if (maxlvt > 3) /* Due to the Pentium erratum 3AP. */ 602 if (maxlvt > 3) /* Due to the Pentium erratum 3AP. */
605 apic_write(APIC_ESR, 0); 603 apic_write(APIC_ESR, 0);
@@ -614,11 +612,9 @@ wakeup_secondary_cpu(int logical_apicid, unsigned long start_eip)
614 612
615 return (send_status | accept_status); 613 return (send_status | accept_status);
616} 614}
617#endif /* WAKE_SECONDARY_VIA_NMI */
618 615
619#ifdef WAKE_SECONDARY_VIA_INIT 616int __devinit
620static int __devinit 617wakeup_secondary_cpu_via_init(int phys_apicid, unsigned long start_eip)
621wakeup_secondary_cpu(int phys_apicid, unsigned long start_eip)
622{ 618{
623 unsigned long send_status, accept_status = 0; 619 unsigned long send_status, accept_status = 0;
624 int maxlvt, num_starts, j; 620 int maxlvt, num_starts, j;
@@ -737,7 +733,6 @@ wakeup_secondary_cpu(int phys_apicid, unsigned long start_eip)
737 733
738 return (send_status | accept_status); 734 return (send_status | accept_status);
739} 735}
740#endif /* WAKE_SECONDARY_VIA_INIT */
741 736
742struct create_idle { 737struct create_idle {
743 struct work_struct work; 738 struct work_struct work;
@@ -1086,8 +1081,10 @@ static int __init smp_sanity_check(unsigned max_cpus)
1086#endif 1081#endif
1087 1082
1088 if (!physid_isset(hard_smp_processor_id(), phys_cpu_present_map)) { 1083 if (!physid_isset(hard_smp_processor_id(), phys_cpu_present_map)) {
1089 printk(KERN_WARNING "weird, boot CPU (#%d) not listed" 1084 printk(KERN_WARNING
1090 "by the BIOS.\n", hard_smp_processor_id()); 1085 "weird, boot CPU (#%d) not listed by the BIOS.\n",
1086 hard_smp_processor_id());
1087
1091 physid_set(hard_smp_processor_id(), phys_cpu_present_map); 1088 physid_set(hard_smp_processor_id(), phys_cpu_present_map);
1092 } 1089 }
1093 1090
diff --git a/arch/x86/kernel/time_64.c b/arch/x86/kernel/time_64.c
index cb19d650c216..1b7711b31037 100644
--- a/arch/x86/kernel/time_64.c
+++ b/arch/x86/kernel/time_64.c
@@ -49,7 +49,7 @@ unsigned long profile_pc(struct pt_regs *regs)
49} 49}
50EXPORT_SYMBOL(profile_pc); 50EXPORT_SYMBOL(profile_pc);
51 51
52irqreturn_t timer_interrupt(int irq, void *dev_id) 52static irqreturn_t timer_interrupt(int irq, void *dev_id)
53{ 53{
54 add_pda(irq0_irqs, 1); 54 add_pda(irq0_irqs, 1);
55 55
@@ -80,6 +80,8 @@ unsigned long __init calibrate_cpu(void)
80 break; 80 break;
81 no_ctr_free = (i == 4); 81 no_ctr_free = (i == 4);
82 if (no_ctr_free) { 82 if (no_ctr_free) {
83 WARN(1, KERN_WARNING "Warning: AMD perfctrs busy ... "
84 "cpu_khz value may be incorrect.\n");
83 i = 3; 85 i = 3;
84 rdmsrl(MSR_K7_EVNTSEL3, evntsel3); 86 rdmsrl(MSR_K7_EVNTSEL3, evntsel3);
85 wrmsrl(MSR_K7_EVNTSEL3, 0); 87 wrmsrl(MSR_K7_EVNTSEL3, 0);
diff --git a/arch/x86/kernel/tlb_32.c b/arch/x86/kernel/tlb_32.c
index f4049f3513b6..4290d918b58a 100644
--- a/arch/x86/kernel/tlb_32.c
+++ b/arch/x86/kernel/tlb_32.c
@@ -34,9 +34,8 @@ static DEFINE_SPINLOCK(tlbstate_lock);
34 */ 34 */
35void leave_mm(int cpu) 35void leave_mm(int cpu)
36{ 36{
37 if (per_cpu(cpu_tlbstate, cpu).state == TLBSTATE_OK) 37 BUG_ON(x86_read_percpu(cpu_tlbstate.state) == TLBSTATE_OK);
38 BUG(); 38 cpu_clear(cpu, x86_read_percpu(cpu_tlbstate.active_mm)->cpu_vm_mask);
39 cpu_clear(cpu, per_cpu(cpu_tlbstate, cpu).active_mm->cpu_vm_mask);
40 load_cr3(swapper_pg_dir); 39 load_cr3(swapper_pg_dir);
41} 40}
42EXPORT_SYMBOL_GPL(leave_mm); 41EXPORT_SYMBOL_GPL(leave_mm);
@@ -104,8 +103,8 @@ void smp_invalidate_interrupt(struct pt_regs *regs)
104 * BUG(); 103 * BUG();
105 */ 104 */
106 105
107 if (flush_mm == per_cpu(cpu_tlbstate, cpu).active_mm) { 106 if (flush_mm == x86_read_percpu(cpu_tlbstate.active_mm)) {
108 if (per_cpu(cpu_tlbstate, cpu).state == TLBSTATE_OK) { 107 if (x86_read_percpu(cpu_tlbstate.state) == TLBSTATE_OK) {
109 if (flush_va == TLB_FLUSH_ALL) 108 if (flush_va == TLB_FLUSH_ALL)
110 local_flush_tlb(); 109 local_flush_tlb();
111 else 110 else
@@ -238,7 +237,7 @@ static void do_flush_tlb_all(void *info)
238 unsigned long cpu = smp_processor_id(); 237 unsigned long cpu = smp_processor_id();
239 238
240 __flush_tlb_all(); 239 __flush_tlb_all();
241 if (per_cpu(cpu_tlbstate, cpu).state == TLBSTATE_LAZY) 240 if (x86_read_percpu(cpu_tlbstate.state) == TLBSTATE_LAZY)
242 leave_mm(cpu); 241 leave_mm(cpu);
243} 242}
244 243
diff --git a/arch/x86/kernel/tlb_uv.c b/arch/x86/kernel/tlb_uv.c
index 04431f34fd16..6a00e5faaa74 100644
--- a/arch/x86/kernel/tlb_uv.c
+++ b/arch/x86/kernel/tlb_uv.c
@@ -566,14 +566,10 @@ static int __init uv_ptc_init(void)
566 if (!is_uv_system()) 566 if (!is_uv_system())
567 return 0; 567 return 0;
568 568
569 if (!proc_mkdir("sgi_uv", NULL))
570 return -EINVAL;
571
572 proc_uv_ptc = create_proc_entry(UV_PTC_BASENAME, 0444, NULL); 569 proc_uv_ptc = create_proc_entry(UV_PTC_BASENAME, 0444, NULL);
573 if (!proc_uv_ptc) { 570 if (!proc_uv_ptc) {
574 printk(KERN_ERR "unable to create %s proc entry\n", 571 printk(KERN_ERR "unable to create %s proc entry\n",
575 UV_PTC_BASENAME); 572 UV_PTC_BASENAME);
576 remove_proc_entry("sgi_uv", NULL);
577 return -EINVAL; 573 return -EINVAL;
578 } 574 }
579 proc_uv_ptc->proc_fops = &proc_uv_ptc_operations; 575 proc_uv_ptc->proc_fops = &proc_uv_ptc_operations;
diff --git a/arch/x86/kernel/trampoline.c b/arch/x86/kernel/trampoline.c
index 1106fac6024d..808031a5ba19 100644
--- a/arch/x86/kernel/trampoline.c
+++ b/arch/x86/kernel/trampoline.c
@@ -1,10 +1,26 @@
1#include <linux/io.h> 1#include <linux/io.h>
2 2
3#include <asm/trampoline.h> 3#include <asm/trampoline.h>
4#include <asm/e820.h>
4 5
5/* ready for x86_64 and x86 */ 6/* ready for x86_64 and x86 */
6unsigned char *trampoline_base = __va(TRAMPOLINE_BASE); 7unsigned char *trampoline_base = __va(TRAMPOLINE_BASE);
7 8
9void __init reserve_trampoline_memory(void)
10{
11#ifdef CONFIG_X86_32
12 /*
13 * But first pinch a few for the stack/trampoline stuff
14 * FIXME: Don't need the extra page at 4K, but need to fix
15 * trampoline before removing it. (see the GDT stuff)
16 */
17 reserve_early(PAGE_SIZE, PAGE_SIZE + PAGE_SIZE, "EX TRAMPOLINE");
18#endif
19 /* Has to be in very low memory so we can execute real-mode AP code. */
20 reserve_early(TRAMPOLINE_BASE, TRAMPOLINE_BASE + TRAMPOLINE_SIZE,
21 "TRAMPOLINE");
22}
23
8/* 24/*
9 * Currently trivial. Write the real->protected mode 25 * Currently trivial. Write the real->protected mode
10 * bootstrap into the page concerned. The caller 26 * bootstrap into the page concerned. The caller
@@ -12,7 +28,6 @@ unsigned char *trampoline_base = __va(TRAMPOLINE_BASE);
12 */ 28 */
13unsigned long setup_trampoline(void) 29unsigned long setup_trampoline(void)
14{ 30{
15 memcpy(trampoline_base, trampoline_data, 31 memcpy(trampoline_base, trampoline_data, TRAMPOLINE_SIZE);
16 trampoline_end - trampoline_data);
17 return virt_to_phys(trampoline_base); 32 return virt_to_phys(trampoline_base);
18} 33}
diff --git a/arch/x86/kernel/traps.c b/arch/x86/kernel/traps.c
index 04d242ab0161..c320c29255c2 100644
--- a/arch/x86/kernel/traps.c
+++ b/arch/x86/kernel/traps.c
@@ -664,7 +664,7 @@ void math_error(void __user *ip)
664{ 664{
665 struct task_struct *task; 665 struct task_struct *task;
666 siginfo_t info; 666 siginfo_t info;
667 unsigned short cwd, swd; 667 unsigned short cwd, swd, err;
668 668
669 /* 669 /*
670 * Save the info for the exception handler and clear the error. 670 * Save the info for the exception handler and clear the error.
@@ -675,7 +675,6 @@ void math_error(void __user *ip)
675 task->thread.error_code = 0; 675 task->thread.error_code = 0;
676 info.si_signo = SIGFPE; 676 info.si_signo = SIGFPE;
677 info.si_errno = 0; 677 info.si_errno = 0;
678 info.si_code = __SI_FAULT;
679 info.si_addr = ip; 678 info.si_addr = ip;
680 /* 679 /*
681 * (~cwd & swd) will mask out exceptions that are not set to unmasked 680 * (~cwd & swd) will mask out exceptions that are not set to unmasked
@@ -689,34 +688,31 @@ void math_error(void __user *ip)
689 */ 688 */
690 cwd = get_fpu_cwd(task); 689 cwd = get_fpu_cwd(task);
691 swd = get_fpu_swd(task); 690 swd = get_fpu_swd(task);
692 switch (swd & ~cwd & 0x3f) { 691
693 case 0x000: /* No unmasked exception */ 692 err = swd & ~cwd & 0x3f;
694#ifdef CONFIG_X86_32 693
694#if CONFIG_X86_32
695 if (!err)
695 return; 696 return;
696#endif 697#endif
697 default: /* Multiple exceptions */ 698
698 break; 699 if (err & 0x001) { /* Invalid op */
699 case 0x001: /* Invalid Op */
700 /* 700 /*
701 * swd & 0x240 == 0x040: Stack Underflow 701 * swd & 0x240 == 0x040: Stack Underflow
702 * swd & 0x240 == 0x240: Stack Overflow 702 * swd & 0x240 == 0x240: Stack Overflow
703 * User must clear the SF bit (0x40) if set 703 * User must clear the SF bit (0x40) if set
704 */ 704 */
705 info.si_code = FPE_FLTINV; 705 info.si_code = FPE_FLTINV;
706 break; 706 } else if (err & 0x004) { /* Divide by Zero */
707 case 0x002: /* Denormalize */
708 case 0x010: /* Underflow */
709 info.si_code = FPE_FLTUND;
710 break;
711 case 0x004: /* Zero Divide */
712 info.si_code = FPE_FLTDIV; 707 info.si_code = FPE_FLTDIV;
713 break; 708 } else if (err & 0x008) { /* Overflow */
714 case 0x008: /* Overflow */
715 info.si_code = FPE_FLTOVF; 709 info.si_code = FPE_FLTOVF;
716 break; 710 } else if (err & 0x012) { /* Denormal, Underflow */
717 case 0x020: /* Precision */ 711 info.si_code = FPE_FLTUND;
712 } else if (err & 0x020) { /* Precision */
718 info.si_code = FPE_FLTRES; 713 info.si_code = FPE_FLTRES;
719 break; 714 } else {
715 info.si_code = __SI_FAULT|SI_KERNEL; /* WTF? */
720 } 716 }
721 force_sig_info(SIGFPE, &info, task); 717 force_sig_info(SIGFPE, &info, task);
722} 718}
diff --git a/arch/x86/kernel/tsc.c b/arch/x86/kernel/tsc.c
index 424093b157d3..599e58168631 100644
--- a/arch/x86/kernel/tsc.c
+++ b/arch/x86/kernel/tsc.c
@@ -15,6 +15,7 @@
15#include <asm/vgtod.h> 15#include <asm/vgtod.h>
16#include <asm/time.h> 16#include <asm/time.h>
17#include <asm/delay.h> 17#include <asm/delay.h>
18#include <asm/hypervisor.h>
18 19
19unsigned int cpu_khz; /* TSC clocks / usec, not used here */ 20unsigned int cpu_khz; /* TSC clocks / usec, not used here */
20EXPORT_SYMBOL(cpu_khz); 21EXPORT_SYMBOL(cpu_khz);
@@ -31,6 +32,7 @@ static int tsc_unstable;
31 erroneous rdtsc usage on !cpu_has_tsc processors */ 32 erroneous rdtsc usage on !cpu_has_tsc processors */
32static int tsc_disabled = -1; 33static int tsc_disabled = -1;
33 34
35static int tsc_clocksource_reliable;
34/* 36/*
35 * Scheduler clock - returns current time in nanosec units. 37 * Scheduler clock - returns current time in nanosec units.
36 */ 38 */
@@ -98,6 +100,15 @@ int __init notsc_setup(char *str)
98 100
99__setup("notsc", notsc_setup); 101__setup("notsc", notsc_setup);
100 102
103static int __init tsc_setup(char *str)
104{
105 if (!strcmp(str, "reliable"))
106 tsc_clocksource_reliable = 1;
107 return 1;
108}
109
110__setup("tsc=", tsc_setup);
111
101#define MAX_RETRIES 5 112#define MAX_RETRIES 5
102#define SMI_TRESHOLD 50000 113#define SMI_TRESHOLD 50000
103 114
@@ -352,9 +363,15 @@ unsigned long native_calibrate_tsc(void)
352{ 363{
353 u64 tsc1, tsc2, delta, ref1, ref2; 364 u64 tsc1, tsc2, delta, ref1, ref2;
354 unsigned long tsc_pit_min = ULONG_MAX, tsc_ref_min = ULONG_MAX; 365 unsigned long tsc_pit_min = ULONG_MAX, tsc_ref_min = ULONG_MAX;
355 unsigned long flags, latch, ms, fast_calibrate; 366 unsigned long flags, latch, ms, fast_calibrate, tsc_khz;
356 int hpet = is_hpet_enabled(), i, loopmin; 367 int hpet = is_hpet_enabled(), i, loopmin;
357 368
369 tsc_khz = get_hypervisor_tsc_freq();
370 if (tsc_khz) {
371 printk(KERN_INFO "TSC: Frequency read from the hypervisor\n");
372 return tsc_khz;
373 }
374
358 local_irq_save(flags); 375 local_irq_save(flags);
359 fast_calibrate = quick_pit_calibrate(); 376 fast_calibrate = quick_pit_calibrate();
360 local_irq_restore(flags); 377 local_irq_restore(flags);
@@ -731,24 +748,21 @@ static struct dmi_system_id __initdata bad_tsc_dmi_table[] = {
731 {} 748 {}
732}; 749};
733 750
734/* 751static void __init check_system_tsc_reliable(void)
735 * Geode_LX - the OLPC CPU has a possibly a very reliable TSC 752{
736 */
737#ifdef CONFIG_MGEODE_LX 753#ifdef CONFIG_MGEODE_LX
738/* RTSC counts during suspend */ 754 /* RTSC counts during suspend */
739#define RTSC_SUSP 0x100 755#define RTSC_SUSP 0x100
740
741static void __init check_geode_tsc_reliable(void)
742{
743 unsigned long res_low, res_high; 756 unsigned long res_low, res_high;
744 757
745 rdmsr_safe(MSR_GEODE_BUSCONT_CONF0, &res_low, &res_high); 758 rdmsr_safe(MSR_GEODE_BUSCONT_CONF0, &res_low, &res_high);
759 /* Geode_LX - the OLPC CPU has a possibly a very reliable TSC */
746 if (res_low & RTSC_SUSP) 760 if (res_low & RTSC_SUSP)
747 clocksource_tsc.flags &= ~CLOCK_SOURCE_MUST_VERIFY; 761 tsc_clocksource_reliable = 1;
748}
749#else
750static inline void check_geode_tsc_reliable(void) { }
751#endif 762#endif
763 if (boot_cpu_has(X86_FEATURE_TSC_RELIABLE))
764 tsc_clocksource_reliable = 1;
765}
752 766
753/* 767/*
754 * Make an educated guess if the TSC is trustworthy and synchronized 768 * Make an educated guess if the TSC is trustworthy and synchronized
@@ -783,6 +797,8 @@ static void __init init_tsc_clocksource(void)
783{ 797{
784 clocksource_tsc.mult = clocksource_khz2mult(tsc_khz, 798 clocksource_tsc.mult = clocksource_khz2mult(tsc_khz,
785 clocksource_tsc.shift); 799 clocksource_tsc.shift);
800 if (tsc_clocksource_reliable)
801 clocksource_tsc.flags &= ~CLOCK_SOURCE_MUST_VERIFY;
786 /* lower the rating if we already know its unstable: */ 802 /* lower the rating if we already know its unstable: */
787 if (check_tsc_unstable()) { 803 if (check_tsc_unstable()) {
788 clocksource_tsc.rating = 0; 804 clocksource_tsc.rating = 0;
@@ -843,7 +859,7 @@ void __init tsc_init(void)
843 if (unsynchronized_tsc()) 859 if (unsynchronized_tsc())
844 mark_tsc_unstable("TSCs unsynchronized"); 860 mark_tsc_unstable("TSCs unsynchronized");
845 861
846 check_geode_tsc_reliable(); 862 check_system_tsc_reliable();
847 init_tsc_clocksource(); 863 init_tsc_clocksource();
848} 864}
849 865
diff --git a/arch/x86/kernel/tsc_sync.c b/arch/x86/kernel/tsc_sync.c
index 9ffb01c31c40..bf36328f6ef9 100644
--- a/arch/x86/kernel/tsc_sync.c
+++ b/arch/x86/kernel/tsc_sync.c
@@ -46,7 +46,9 @@ static __cpuinit void check_tsc_warp(void)
46 cycles_t start, now, prev, end; 46 cycles_t start, now, prev, end;
47 int i; 47 int i;
48 48
49 rdtsc_barrier();
49 start = get_cycles(); 50 start = get_cycles();
51 rdtsc_barrier();
50 /* 52 /*
51 * The measurement runs for 20 msecs: 53 * The measurement runs for 20 msecs:
52 */ 54 */
@@ -61,7 +63,9 @@ static __cpuinit void check_tsc_warp(void)
61 */ 63 */
62 __raw_spin_lock(&sync_lock); 64 __raw_spin_lock(&sync_lock);
63 prev = last_tsc; 65 prev = last_tsc;
66 rdtsc_barrier();
64 now = get_cycles(); 67 now = get_cycles();
68 rdtsc_barrier();
65 last_tsc = now; 69 last_tsc = now;
66 __raw_spin_unlock(&sync_lock); 70 __raw_spin_unlock(&sync_lock);
67 71
@@ -108,6 +112,12 @@ void __cpuinit check_tsc_sync_source(int cpu)
108 if (unsynchronized_tsc()) 112 if (unsynchronized_tsc())
109 return; 113 return;
110 114
115 if (boot_cpu_has(X86_FEATURE_TSC_RELIABLE)) {
116 printk(KERN_INFO
117 "Skipping synchronization checks as TSC is reliable.\n");
118 return;
119 }
120
111 printk(KERN_INFO "checking TSC synchronization [CPU#%d -> CPU#%d]:", 121 printk(KERN_INFO "checking TSC synchronization [CPU#%d -> CPU#%d]:",
112 smp_processor_id(), cpu); 122 smp_processor_id(), cpu);
113 123
@@ -161,7 +171,7 @@ void __cpuinit check_tsc_sync_target(void)
161{ 171{
162 int cpus = 2; 172 int cpus = 2;
163 173
164 if (unsynchronized_tsc()) 174 if (unsynchronized_tsc() || boot_cpu_has(X86_FEATURE_TSC_RELIABLE))
165 return; 175 return;
166 176
167 /* 177 /*
diff --git a/arch/x86/kernel/vmi_32.c b/arch/x86/kernel/vmi_32.c
index 8b6c393ab9fd..23206ba16874 100644
--- a/arch/x86/kernel/vmi_32.c
+++ b/arch/x86/kernel/vmi_32.c
@@ -266,109 +266,6 @@ static void vmi_nop(void)
266{ 266{
267} 267}
268 268
269#ifdef CONFIG_DEBUG_PAGE_TYPE
270
271#ifdef CONFIG_X86_PAE
272#define MAX_BOOT_PTS (2048+4+1)
273#else
274#define MAX_BOOT_PTS (1024+1)
275#endif
276
277/*
278 * During boot, mem_map is not yet available in paging_init, so stash
279 * all the boot page allocations here.
280 */
281static struct {
282 u32 pfn;
283 int type;
284} boot_page_allocations[MAX_BOOT_PTS];
285static int num_boot_page_allocations;
286static int boot_allocations_applied;
287
288void vmi_apply_boot_page_allocations(void)
289{
290 int i;
291 BUG_ON(!mem_map);
292 for (i = 0; i < num_boot_page_allocations; i++) {
293 struct page *page = pfn_to_page(boot_page_allocations[i].pfn);
294 page->type = boot_page_allocations[i].type;
295 page->type = boot_page_allocations[i].type &
296 ~(VMI_PAGE_ZEROED | VMI_PAGE_CLONE);
297 }
298 boot_allocations_applied = 1;
299}
300
301static void record_page_type(u32 pfn, int type)
302{
303 BUG_ON(num_boot_page_allocations >= MAX_BOOT_PTS);
304 boot_page_allocations[num_boot_page_allocations].pfn = pfn;
305 boot_page_allocations[num_boot_page_allocations].type = type;
306 num_boot_page_allocations++;
307}
308
309static void check_zeroed_page(u32 pfn, int type, struct page *page)
310{
311 u32 *ptr;
312 int i;
313 int limit = PAGE_SIZE / sizeof(int);
314
315 if (page_address(page))
316 ptr = (u32 *)page_address(page);
317 else
318 ptr = (u32 *)__va(pfn << PAGE_SHIFT);
319 /*
320 * When cloning the root in non-PAE mode, only the userspace
321 * pdes need to be zeroed.
322 */
323 if (type & VMI_PAGE_CLONE)
324 limit = KERNEL_PGD_BOUNDARY;
325 for (i = 0; i < limit; i++)
326 BUG_ON(ptr[i]);
327}
328
329/*
330 * We stash the page type into struct page so we can verify the page
331 * types are used properly.
332 */
333static void vmi_set_page_type(u32 pfn, int type)
334{
335 /* PAE can have multiple roots per page - don't track */
336 if (PTRS_PER_PMD > 1 && (type & VMI_PAGE_PDP))
337 return;
338
339 if (boot_allocations_applied) {
340 struct page *page = pfn_to_page(pfn);
341 if (type != VMI_PAGE_NORMAL)
342 BUG_ON(page->type);
343 else
344 BUG_ON(page->type == VMI_PAGE_NORMAL);
345 page->type = type & ~(VMI_PAGE_ZEROED | VMI_PAGE_CLONE);
346 if (type & VMI_PAGE_ZEROED)
347 check_zeroed_page(pfn, type, page);
348 } else {
349 record_page_type(pfn, type);
350 }
351}
352
353static void vmi_check_page_type(u32 pfn, int type)
354{
355 /* PAE can have multiple roots per page - skip checks */
356 if (PTRS_PER_PMD > 1 && (type & VMI_PAGE_PDP))
357 return;
358
359 type &= ~(VMI_PAGE_ZEROED | VMI_PAGE_CLONE);
360 if (boot_allocations_applied) {
361 struct page *page = pfn_to_page(pfn);
362 BUG_ON((page->type ^ type) & VMI_PAGE_PAE);
363 BUG_ON(type == VMI_PAGE_NORMAL && page->type);
364 BUG_ON((type & page->type) == 0);
365 }
366}
367#else
368#define vmi_set_page_type(p,t) do { } while (0)
369#define vmi_check_page_type(p,t) do { } while (0)
370#endif
371
372#ifdef CONFIG_HIGHPTE 269#ifdef CONFIG_HIGHPTE
373static void *vmi_kmap_atomic_pte(struct page *page, enum km_type type) 270static void *vmi_kmap_atomic_pte(struct page *page, enum km_type type)
374{ 271{
@@ -395,7 +292,6 @@ static void *vmi_kmap_atomic_pte(struct page *page, enum km_type type)
395 292
396static void vmi_allocate_pte(struct mm_struct *mm, unsigned long pfn) 293static void vmi_allocate_pte(struct mm_struct *mm, unsigned long pfn)
397{ 294{
398 vmi_set_page_type(pfn, VMI_PAGE_L1);
399 vmi_ops.allocate_page(pfn, VMI_PAGE_L1, 0, 0, 0); 295 vmi_ops.allocate_page(pfn, VMI_PAGE_L1, 0, 0, 0);
400} 296}
401 297
@@ -406,27 +302,22 @@ static void vmi_allocate_pmd(struct mm_struct *mm, unsigned long pfn)
406 * It is called only for swapper_pg_dir, which already has 302 * It is called only for swapper_pg_dir, which already has
407 * data on it. 303 * data on it.
408 */ 304 */
409 vmi_set_page_type(pfn, VMI_PAGE_L2);
410 vmi_ops.allocate_page(pfn, VMI_PAGE_L2, 0, 0, 0); 305 vmi_ops.allocate_page(pfn, VMI_PAGE_L2, 0, 0, 0);
411} 306}
412 307
413static void vmi_allocate_pmd_clone(unsigned long pfn, unsigned long clonepfn, unsigned long start, unsigned long count) 308static void vmi_allocate_pmd_clone(unsigned long pfn, unsigned long clonepfn, unsigned long start, unsigned long count)
414{ 309{
415 vmi_set_page_type(pfn, VMI_PAGE_L2 | VMI_PAGE_CLONE);
416 vmi_check_page_type(clonepfn, VMI_PAGE_L2);
417 vmi_ops.allocate_page(pfn, VMI_PAGE_L2 | VMI_PAGE_CLONE, clonepfn, start, count); 310 vmi_ops.allocate_page(pfn, VMI_PAGE_L2 | VMI_PAGE_CLONE, clonepfn, start, count);
418} 311}
419 312
420static void vmi_release_pte(unsigned long pfn) 313static void vmi_release_pte(unsigned long pfn)
421{ 314{
422 vmi_ops.release_page(pfn, VMI_PAGE_L1); 315 vmi_ops.release_page(pfn, VMI_PAGE_L1);
423 vmi_set_page_type(pfn, VMI_PAGE_NORMAL);
424} 316}
425 317
426static void vmi_release_pmd(unsigned long pfn) 318static void vmi_release_pmd(unsigned long pfn)
427{ 319{
428 vmi_ops.release_page(pfn, VMI_PAGE_L2); 320 vmi_ops.release_page(pfn, VMI_PAGE_L2);
429 vmi_set_page_type(pfn, VMI_PAGE_NORMAL);
430} 321}
431 322
432/* 323/*
@@ -450,26 +341,22 @@ static void vmi_release_pmd(unsigned long pfn)
450 341
451static void vmi_update_pte(struct mm_struct *mm, unsigned long addr, pte_t *ptep) 342static void vmi_update_pte(struct mm_struct *mm, unsigned long addr, pte_t *ptep)
452{ 343{
453 vmi_check_page_type(__pa(ptep) >> PAGE_SHIFT, VMI_PAGE_PTE);
454 vmi_ops.update_pte(ptep, vmi_flags_addr(mm, addr, VMI_PAGE_PT, 0)); 344 vmi_ops.update_pte(ptep, vmi_flags_addr(mm, addr, VMI_PAGE_PT, 0));
455} 345}
456 346
457static void vmi_update_pte_defer(struct mm_struct *mm, unsigned long addr, pte_t *ptep) 347static void vmi_update_pte_defer(struct mm_struct *mm, unsigned long addr, pte_t *ptep)
458{ 348{
459 vmi_check_page_type(__pa(ptep) >> PAGE_SHIFT, VMI_PAGE_PTE);
460 vmi_ops.update_pte(ptep, vmi_flags_addr_defer(mm, addr, VMI_PAGE_PT, 0)); 349 vmi_ops.update_pte(ptep, vmi_flags_addr_defer(mm, addr, VMI_PAGE_PT, 0));
461} 350}
462 351
463static void vmi_set_pte(pte_t *ptep, pte_t pte) 352static void vmi_set_pte(pte_t *ptep, pte_t pte)
464{ 353{
465 /* XXX because of set_pmd_pte, this can be called on PT or PD layers */ 354 /* XXX because of set_pmd_pte, this can be called on PT or PD layers */
466 vmi_check_page_type(__pa(ptep) >> PAGE_SHIFT, VMI_PAGE_PTE | VMI_PAGE_PD);
467 vmi_ops.set_pte(pte, ptep, VMI_PAGE_PT); 355 vmi_ops.set_pte(pte, ptep, VMI_PAGE_PT);
468} 356}
469 357
470static void vmi_set_pte_at(struct mm_struct *mm, unsigned long addr, pte_t *ptep, pte_t pte) 358static void vmi_set_pte_at(struct mm_struct *mm, unsigned long addr, pte_t *ptep, pte_t pte)
471{ 359{
472 vmi_check_page_type(__pa(ptep) >> PAGE_SHIFT, VMI_PAGE_PTE);
473 vmi_ops.set_pte(pte, ptep, vmi_flags_addr(mm, addr, VMI_PAGE_PT, 0)); 360 vmi_ops.set_pte(pte, ptep, vmi_flags_addr(mm, addr, VMI_PAGE_PT, 0));
474} 361}
475 362
@@ -477,10 +364,8 @@ static void vmi_set_pmd(pmd_t *pmdp, pmd_t pmdval)
477{ 364{
478#ifdef CONFIG_X86_PAE 365#ifdef CONFIG_X86_PAE
479 const pte_t pte = { .pte = pmdval.pmd }; 366 const pte_t pte = { .pte = pmdval.pmd };
480 vmi_check_page_type(__pa(pmdp) >> PAGE_SHIFT, VMI_PAGE_PMD);
481#else 367#else
482 const pte_t pte = { pmdval.pud.pgd.pgd }; 368 const pte_t pte = { pmdval.pud.pgd.pgd };
483 vmi_check_page_type(__pa(pmdp) >> PAGE_SHIFT, VMI_PAGE_PGD);
484#endif 369#endif
485 vmi_ops.set_pte(pte, (pte_t *)pmdp, VMI_PAGE_PD); 370 vmi_ops.set_pte(pte, (pte_t *)pmdp, VMI_PAGE_PD);
486} 371}
@@ -502,7 +387,6 @@ static void vmi_set_pte_atomic(pte_t *ptep, pte_t pteval)
502 387
503static void vmi_set_pte_present(struct mm_struct *mm, unsigned long addr, pte_t *ptep, pte_t pte) 388static void vmi_set_pte_present(struct mm_struct *mm, unsigned long addr, pte_t *ptep, pte_t pte)
504{ 389{
505 vmi_check_page_type(__pa(ptep) >> PAGE_SHIFT, VMI_PAGE_PTE);
506 vmi_ops.set_pte(pte, ptep, vmi_flags_addr_defer(mm, addr, VMI_PAGE_PT, 1)); 390 vmi_ops.set_pte(pte, ptep, vmi_flags_addr_defer(mm, addr, VMI_PAGE_PT, 1));
507} 391}
508 392
@@ -510,21 +394,18 @@ static void vmi_set_pud(pud_t *pudp, pud_t pudval)
510{ 394{
511 /* Um, eww */ 395 /* Um, eww */
512 const pte_t pte = { .pte = pudval.pgd.pgd }; 396 const pte_t pte = { .pte = pudval.pgd.pgd };
513 vmi_check_page_type(__pa(pudp) >> PAGE_SHIFT, VMI_PAGE_PGD);
514 vmi_ops.set_pte(pte, (pte_t *)pudp, VMI_PAGE_PDP); 397 vmi_ops.set_pte(pte, (pte_t *)pudp, VMI_PAGE_PDP);
515} 398}
516 399
517static void vmi_pte_clear(struct mm_struct *mm, unsigned long addr, pte_t *ptep) 400static void vmi_pte_clear(struct mm_struct *mm, unsigned long addr, pte_t *ptep)
518{ 401{
519 const pte_t pte = { .pte = 0 }; 402 const pte_t pte = { .pte = 0 };
520 vmi_check_page_type(__pa(ptep) >> PAGE_SHIFT, VMI_PAGE_PTE);
521 vmi_ops.set_pte(pte, ptep, vmi_flags_addr(mm, addr, VMI_PAGE_PT, 0)); 403 vmi_ops.set_pte(pte, ptep, vmi_flags_addr(mm, addr, VMI_PAGE_PT, 0));
522} 404}
523 405
524static void vmi_pmd_clear(pmd_t *pmd) 406static void vmi_pmd_clear(pmd_t *pmd)
525{ 407{
526 const pte_t pte = { .pte = 0 }; 408 const pte_t pte = { .pte = 0 };
527 vmi_check_page_type(__pa(pmd) >> PAGE_SHIFT, VMI_PAGE_PMD);
528 vmi_ops.set_pte(pte, (pte_t *)pmd, VMI_PAGE_PD); 409 vmi_ops.set_pte(pte, (pte_t *)pmd, VMI_PAGE_PD);
529} 410}
530#endif 411#endif
@@ -960,8 +841,6 @@ static inline int __init activate_vmi(void)
960 841
961void __init vmi_init(void) 842void __init vmi_init(void)
962{ 843{
963 unsigned long flags;
964
965 if (!vmi_rom) 844 if (!vmi_rom)
966 probe_vmi_rom(); 845 probe_vmi_rom();
967 else 846 else
@@ -973,13 +852,21 @@ void __init vmi_init(void)
973 852
974 reserve_top_address(-vmi_rom->virtual_top); 853 reserve_top_address(-vmi_rom->virtual_top);
975 854
976 local_irq_save(flags);
977 activate_vmi();
978
979#ifdef CONFIG_X86_IO_APIC 855#ifdef CONFIG_X86_IO_APIC
980 /* This is virtual hardware; timer routing is wired correctly */ 856 /* This is virtual hardware; timer routing is wired correctly */
981 no_timer_check = 1; 857 no_timer_check = 1;
982#endif 858#endif
859}
860
861void vmi_activate(void)
862{
863 unsigned long flags;
864
865 if (!vmi_rom)
866 return;
867
868 local_irq_save(flags);
869 activate_vmi();
983 local_irq_restore(flags & X86_EFLAGS_IF); 870 local_irq_restore(flags & X86_EFLAGS_IF);
984} 871}
985 872
diff --git a/arch/x86/kernel/vsyscall_64.c b/arch/x86/kernel/vsyscall_64.c
index 0b8b6690a86d..ebf2f12900f5 100644
--- a/arch/x86/kernel/vsyscall_64.c
+++ b/arch/x86/kernel/vsyscall_64.c
@@ -128,7 +128,16 @@ static __always_inline void do_vgettimeofday(struct timeval * tv)
128 gettimeofday(tv,NULL); 128 gettimeofday(tv,NULL);
129 return; 129 return;
130 } 130 }
131
132 /*
133 * Surround the RDTSC by barriers, to make sure it's not
134 * speculated to outside the seqlock critical section and
135 * does not cause time warps:
136 */
137 rdtsc_barrier();
131 now = vread(); 138 now = vread();
139 rdtsc_barrier();
140
132 base = __vsyscall_gtod_data.clock.cycle_last; 141 base = __vsyscall_gtod_data.clock.cycle_last;
133 mask = __vsyscall_gtod_data.clock.mask; 142 mask = __vsyscall_gtod_data.clock.mask;
134 mult = __vsyscall_gtod_data.clock.mult; 143 mult = __vsyscall_gtod_data.clock.mult;
diff --git a/arch/x86/kernel/xsave.c b/arch/x86/kernel/xsave.c
index b13acb75e822..15c3e6999182 100644
--- a/arch/x86/kernel/xsave.c
+++ b/arch/x86/kernel/xsave.c
@@ -310,7 +310,7 @@ static void __init setup_xstate_init(void)
310/* 310/*
311 * Enable and initialize the xsave feature. 311 * Enable and initialize the xsave feature.
312 */ 312 */
313void __init xsave_cntxt_init(void) 313void __ref xsave_cntxt_init(void)
314{ 314{
315 unsigned int eax, ebx, ecx, edx; 315 unsigned int eax, ebx, ecx, edx;
316 316
diff --git a/arch/x86/kvm/Kconfig b/arch/x86/kvm/Kconfig
index ce3251ce5504..b81125f0bdee 100644
--- a/arch/x86/kvm/Kconfig
+++ b/arch/x86/kvm/Kconfig
@@ -20,6 +20,8 @@ if VIRTUALIZATION
20config KVM 20config KVM
21 tristate "Kernel-based Virtual Machine (KVM) support" 21 tristate "Kernel-based Virtual Machine (KVM) support"
22 depends on HAVE_KVM 22 depends on HAVE_KVM
23 # for device assignment:
24 depends on PCI
23 select PREEMPT_NOTIFIERS 25 select PREEMPT_NOTIFIERS
24 select MMU_NOTIFIER 26 select MMU_NOTIFIER
25 select ANON_INODES 27 select ANON_INODES
diff --git a/arch/x86/kvm/i8254.c b/arch/x86/kvm/i8254.c
index 8772dc946823..59ebd37ad79e 100644
--- a/arch/x86/kvm/i8254.c
+++ b/arch/x86/kvm/i8254.c
@@ -548,8 +548,10 @@ struct kvm_pit *kvm_create_pit(struct kvm *kvm)
548 mutex_lock(&kvm->lock); 548 mutex_lock(&kvm->lock);
549 pit->irq_source_id = kvm_request_irq_source_id(kvm); 549 pit->irq_source_id = kvm_request_irq_source_id(kvm);
550 mutex_unlock(&kvm->lock); 550 mutex_unlock(&kvm->lock);
551 if (pit->irq_source_id < 0) 551 if (pit->irq_source_id < 0) {
552 kfree(pit);
552 return NULL; 553 return NULL;
554 }
553 555
554 mutex_init(&pit->pit_state.lock); 556 mutex_init(&pit->pit_state.lock);
555 mutex_lock(&pit->pit_state.lock); 557 mutex_lock(&pit->pit_state.lock);
diff --git a/arch/x86/kvm/mmu.c b/arch/x86/kvm/mmu.c
index 2a5e64881d9b..410ddbc1aa2e 100644
--- a/arch/x86/kvm/mmu.c
+++ b/arch/x86/kvm/mmu.c
@@ -314,7 +314,7 @@ static int mmu_topup_memory_caches(struct kvm_vcpu *vcpu)
314 if (r) 314 if (r)
315 goto out; 315 goto out;
316 r = mmu_topup_memory_cache(&vcpu->arch.mmu_rmap_desc_cache, 316 r = mmu_topup_memory_cache(&vcpu->arch.mmu_rmap_desc_cache,
317 rmap_desc_cache, 1); 317 rmap_desc_cache, 4);
318 if (r) 318 if (r)
319 goto out; 319 goto out;
320 r = mmu_topup_memory_cache_page(&vcpu->arch.mmu_page_cache, 8); 320 r = mmu_topup_memory_cache_page(&vcpu->arch.mmu_page_cache, 8);
@@ -1038,13 +1038,13 @@ static int kvm_sync_page(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp)
1038 } 1038 }
1039 1039
1040 rmap_write_protect(vcpu->kvm, sp->gfn); 1040 rmap_write_protect(vcpu->kvm, sp->gfn);
1041 kvm_unlink_unsync_page(vcpu->kvm, sp);
1041 if (vcpu->arch.mmu.sync_page(vcpu, sp)) { 1042 if (vcpu->arch.mmu.sync_page(vcpu, sp)) {
1042 kvm_mmu_zap_page(vcpu->kvm, sp); 1043 kvm_mmu_zap_page(vcpu->kvm, sp);
1043 return 1; 1044 return 1;
1044 } 1045 }
1045 1046
1046 kvm_mmu_flush_tlb(vcpu); 1047 kvm_mmu_flush_tlb(vcpu);
1047 kvm_unlink_unsync_page(vcpu->kvm, sp);
1048 return 0; 1048 return 0;
1049} 1049}
1050 1050
diff --git a/arch/x86/kvm/paging_tmpl.h b/arch/x86/kvm/paging_tmpl.h
index 613ec9aa674a..84eee43bbe74 100644
--- a/arch/x86/kvm/paging_tmpl.h
+++ b/arch/x86/kvm/paging_tmpl.h
@@ -331,6 +331,7 @@ static int FNAME(shadow_walk_entry)(struct kvm_shadow_walk *_sw,
331 r = kvm_read_guest_atomic(vcpu->kvm, gw->pte_gpa[level - 2], 331 r = kvm_read_guest_atomic(vcpu->kvm, gw->pte_gpa[level - 2],
332 &curr_pte, sizeof(curr_pte)); 332 &curr_pte, sizeof(curr_pte));
333 if (r || curr_pte != gw->ptes[level - 2]) { 333 if (r || curr_pte != gw->ptes[level - 2]) {
334 kvm_mmu_put_page(shadow_page, sptep);
334 kvm_release_pfn_clean(sw->pfn); 335 kvm_release_pfn_clean(sw->pfn);
335 sw->sptep = NULL; 336 sw->sptep = NULL;
336 return 1; 337 return 1;
diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
index 2643b430d83a..a4018b01e1f9 100644
--- a/arch/x86/kvm/vmx.c
+++ b/arch/x86/kvm/vmx.c
@@ -3149,7 +3149,9 @@ static void vmx_intr_assist(struct kvm_vcpu *vcpu)
3149 3149
3150 if (cpu_has_virtual_nmis()) { 3150 if (cpu_has_virtual_nmis()) {
3151 if (vcpu->arch.nmi_pending && !vcpu->arch.nmi_injected) { 3151 if (vcpu->arch.nmi_pending && !vcpu->arch.nmi_injected) {
3152 if (vmx_nmi_enabled(vcpu)) { 3152 if (vcpu->arch.interrupt.pending) {
3153 enable_nmi_window(vcpu);
3154 } else if (vmx_nmi_enabled(vcpu)) {
3153 vcpu->arch.nmi_pending = false; 3155 vcpu->arch.nmi_pending = false;
3154 vcpu->arch.nmi_injected = true; 3156 vcpu->arch.nmi_injected = true;
3155 } else { 3157 } else {
@@ -3564,7 +3566,8 @@ static int __init vmx_init(void)
3564 bypass_guest_pf = 0; 3566 bypass_guest_pf = 0;
3565 kvm_mmu_set_base_ptes(VMX_EPT_READABLE_MASK | 3567 kvm_mmu_set_base_ptes(VMX_EPT_READABLE_MASK |
3566 VMX_EPT_WRITABLE_MASK | 3568 VMX_EPT_WRITABLE_MASK |
3567 VMX_EPT_DEFAULT_MT << VMX_EPT_MT_EPTE_SHIFT); 3569 VMX_EPT_DEFAULT_MT << VMX_EPT_MT_EPTE_SHIFT |
3570 VMX_EPT_IGMT_BIT);
3568 kvm_mmu_set_mask_ptes(0ull, 0ull, 0ull, 0ull, 3571 kvm_mmu_set_mask_ptes(0ull, 0ull, 0ull, 0ull,
3569 VMX_EPT_EXECUTABLE_MASK); 3572 VMX_EPT_EXECUTABLE_MASK);
3570 kvm_enable_tdp(); 3573 kvm_enable_tdp();
diff --git a/arch/x86/kvm/vmx.h b/arch/x86/kvm/vmx.h
index 3e010d21fdd7..ec5edc339da6 100644
--- a/arch/x86/kvm/vmx.h
+++ b/arch/x86/kvm/vmx.h
@@ -352,6 +352,7 @@ enum vmcs_field {
352#define VMX_EPT_READABLE_MASK 0x1ull 352#define VMX_EPT_READABLE_MASK 0x1ull
353#define VMX_EPT_WRITABLE_MASK 0x2ull 353#define VMX_EPT_WRITABLE_MASK 0x2ull
354#define VMX_EPT_EXECUTABLE_MASK 0x4ull 354#define VMX_EPT_EXECUTABLE_MASK 0x4ull
355#define VMX_EPT_IGMT_BIT (1ull << 6)
355 356
356#define VMX_EPT_IDENTITY_PAGETABLE_ADDR 0xfffbc000ul 357#define VMX_EPT_IDENTITY_PAGETABLE_ADDR 0xfffbc000ul
357 358
diff --git a/arch/x86/mach-generic/bigsmp.c b/arch/x86/mach-generic/bigsmp.c
index 3c3b471ea496..3624a364b7f3 100644
--- a/arch/x86/mach-generic/bigsmp.c
+++ b/arch/x86/mach-generic/bigsmp.c
@@ -17,6 +17,7 @@
17#include <asm/bigsmp/apic.h> 17#include <asm/bigsmp/apic.h>
18#include <asm/bigsmp/ipi.h> 18#include <asm/bigsmp/ipi.h>
19#include <asm/mach-default/mach_mpparse.h> 19#include <asm/mach-default/mach_mpparse.h>
20#include <asm/mach-default/mach_wakecpu.h>
20 21
21static int dmi_bigsmp; /* can be set by dmi scanners */ 22static int dmi_bigsmp; /* can be set by dmi scanners */
22 23
diff --git a/arch/x86/mach-generic/default.c b/arch/x86/mach-generic/default.c
index 9e835a11a13a..e63a4a76d8cd 100644
--- a/arch/x86/mach-generic/default.c
+++ b/arch/x86/mach-generic/default.c
@@ -16,6 +16,7 @@
16#include <asm/mach-default/mach_apic.h> 16#include <asm/mach-default/mach_apic.h>
17#include <asm/mach-default/mach_ipi.h> 17#include <asm/mach-default/mach_ipi.h>
18#include <asm/mach-default/mach_mpparse.h> 18#include <asm/mach-default/mach_mpparse.h>
19#include <asm/mach-default/mach_wakecpu.h>
19 20
20/* should be called last. */ 21/* should be called last. */
21static int probe_default(void) 22static int probe_default(void)
diff --git a/arch/x86/mach-generic/es7000.c b/arch/x86/mach-generic/es7000.c
index 28459cab3ddb..7b4e6d0d1690 100644
--- a/arch/x86/mach-generic/es7000.c
+++ b/arch/x86/mach-generic/es7000.c
@@ -16,7 +16,19 @@
16#include <asm/es7000/apic.h> 16#include <asm/es7000/apic.h>
17#include <asm/es7000/ipi.h> 17#include <asm/es7000/ipi.h>
18#include <asm/es7000/mpparse.h> 18#include <asm/es7000/mpparse.h>
19#include <asm/es7000/wakecpu.h> 19#include <asm/mach-default/mach_wakecpu.h>
20
21void __init es7000_update_genapic_to_cluster(void)
22{
23 genapic->target_cpus = target_cpus_cluster;
24 genapic->int_delivery_mode = INT_DELIVERY_MODE_CLUSTER;
25 genapic->int_dest_mode = INT_DEST_MODE_CLUSTER;
26 genapic->no_balance_irq = NO_BALANCE_IRQ_CLUSTER;
27
28 genapic->init_apic_ldr = init_apic_ldr_cluster;
29
30 genapic->cpu_mask_to_apicid = cpu_mask_to_apicid_cluster;
31}
20 32
21static int probe_es7000(void) 33static int probe_es7000(void)
22{ 34{
diff --git a/arch/x86/mach-generic/probe.c b/arch/x86/mach-generic/probe.c
index 5a7e4619e1c4..c346d9d0226f 100644
--- a/arch/x86/mach-generic/probe.c
+++ b/arch/x86/mach-generic/probe.c
@@ -15,6 +15,7 @@
15#include <asm/mpspec.h> 15#include <asm/mpspec.h>
16#include <asm/apicdef.h> 16#include <asm/apicdef.h>
17#include <asm/genapic.h> 17#include <asm/genapic.h>
18#include <asm/setup.h>
18 19
19extern struct genapic apic_numaq; 20extern struct genapic apic_numaq;
20extern struct genapic apic_summit; 21extern struct genapic apic_summit;
@@ -57,6 +58,9 @@ static int __init parse_apic(char *arg)
57 } 58 }
58 } 59 }
59 60
61 if (x86_quirks->update_genapic)
62 x86_quirks->update_genapic();
63
60 /* Parsed again by __setup for debug/verbose */ 64 /* Parsed again by __setup for debug/verbose */
61 return 0; 65 return 0;
62} 66}
@@ -72,12 +76,15 @@ void __init generic_bigsmp_probe(void)
72 * - we find more than 8 CPUs in acpi LAPIC listing with xAPIC support 76 * - we find more than 8 CPUs in acpi LAPIC listing with xAPIC support
73 */ 77 */
74 78
75 if (!cmdline_apic && genapic == &apic_default) 79 if (!cmdline_apic && genapic == &apic_default) {
76 if (apic_bigsmp.probe()) { 80 if (apic_bigsmp.probe()) {
77 genapic = &apic_bigsmp; 81 genapic = &apic_bigsmp;
82 if (x86_quirks->update_genapic)
83 x86_quirks->update_genapic();
78 printk(KERN_INFO "Overriding APIC driver with %s\n", 84 printk(KERN_INFO "Overriding APIC driver with %s\n",
79 genapic->name); 85 genapic->name);
80 } 86 }
87 }
81#endif 88#endif
82} 89}
83 90
@@ -94,6 +101,9 @@ void __init generic_apic_probe(void)
94 /* Not visible without early console */ 101 /* Not visible without early console */
95 if (!apic_probe[i]) 102 if (!apic_probe[i])
96 panic("Didn't find an APIC driver"); 103 panic("Didn't find an APIC driver");
104
105 if (x86_quirks->update_genapic)
106 x86_quirks->update_genapic();
97 } 107 }
98 printk(KERN_INFO "Using APIC driver %s\n", genapic->name); 108 printk(KERN_INFO "Using APIC driver %s\n", genapic->name);
99} 109}
@@ -108,6 +118,8 @@ int __init mps_oem_check(struct mp_config_table *mpc, char *oem,
108 if (apic_probe[i]->mps_oem_check(mpc, oem, productid)) { 118 if (apic_probe[i]->mps_oem_check(mpc, oem, productid)) {
109 if (!cmdline_apic) { 119 if (!cmdline_apic) {
110 genapic = apic_probe[i]; 120 genapic = apic_probe[i];
121 if (x86_quirks->update_genapic)
122 x86_quirks->update_genapic();
111 printk(KERN_INFO "Switched to APIC driver `%s'.\n", 123 printk(KERN_INFO "Switched to APIC driver `%s'.\n",
112 genapic->name); 124 genapic->name);
113 } 125 }
@@ -124,6 +136,8 @@ int __init acpi_madt_oem_check(char *oem_id, char *oem_table_id)
124 if (apic_probe[i]->acpi_madt_oem_check(oem_id, oem_table_id)) { 136 if (apic_probe[i]->acpi_madt_oem_check(oem_id, oem_table_id)) {
125 if (!cmdline_apic) { 137 if (!cmdline_apic) {
126 genapic = apic_probe[i]; 138 genapic = apic_probe[i];
139 if (x86_quirks->update_genapic)
140 x86_quirks->update_genapic();
127 printk(KERN_INFO "Switched to APIC driver `%s'.\n", 141 printk(KERN_INFO "Switched to APIC driver `%s'.\n",
128 genapic->name); 142 genapic->name);
129 } 143 }
diff --git a/arch/x86/mach-generic/summit.c b/arch/x86/mach-generic/summit.c
index 6272b5e69da6..2c6d234e0009 100644
--- a/arch/x86/mach-generic/summit.c
+++ b/arch/x86/mach-generic/summit.c
@@ -16,6 +16,7 @@
16#include <asm/summit/apic.h> 16#include <asm/summit/apic.h>
17#include <asm/summit/ipi.h> 17#include <asm/summit/ipi.h>
18#include <asm/summit/mpparse.h> 18#include <asm/summit/mpparse.h>
19#include <asm/mach-default/mach_wakecpu.h>
19 20
20static int probe_summit(void) 21static int probe_summit(void)
21{ 22{
diff --git a/arch/x86/mach-voyager/voyager_smp.c b/arch/x86/mach-voyager/voyager_smp.c
index 0e331652681e..52145007bd7e 100644
--- a/arch/x86/mach-voyager/voyager_smp.c
+++ b/arch/x86/mach-voyager/voyager_smp.c
@@ -7,6 +7,7 @@
7 * This file provides all the same external entries as smp.c but uses 7 * This file provides all the same external entries as smp.c but uses
8 * the voyager hal to provide the functionality 8 * the voyager hal to provide the functionality
9 */ 9 */
10#include <linux/cpu.h>
10#include <linux/module.h> 11#include <linux/module.h>
11#include <linux/mm.h> 12#include <linux/mm.h>
12#include <linux/kernel_stat.h> 13#include <linux/kernel_stat.h>
@@ -1790,6 +1791,17 @@ void __init smp_setup_processor_id(void)
1790 x86_write_percpu(cpu_number, hard_smp_processor_id()); 1791 x86_write_percpu(cpu_number, hard_smp_processor_id());
1791} 1792}
1792 1793
1794static void voyager_send_call_func(cpumask_t callmask)
1795{
1796 __u32 mask = cpus_addr(callmask)[0] & ~(1 << smp_processor_id());
1797 send_CPI(mask, VIC_CALL_FUNCTION_CPI);
1798}
1799
1800static void voyager_send_call_func_single(int cpu)
1801{
1802 send_CPI(1 << cpu, VIC_CALL_FUNCTION_SINGLE_CPI);
1803}
1804
1793struct smp_ops smp_ops = { 1805struct smp_ops smp_ops = {
1794 .smp_prepare_boot_cpu = voyager_smp_prepare_boot_cpu, 1806 .smp_prepare_boot_cpu = voyager_smp_prepare_boot_cpu,
1795 .smp_prepare_cpus = voyager_smp_prepare_cpus, 1807 .smp_prepare_cpus = voyager_smp_prepare_cpus,
@@ -1799,6 +1811,6 @@ struct smp_ops smp_ops = {
1799 .smp_send_stop = voyager_smp_send_stop, 1811 .smp_send_stop = voyager_smp_send_stop,
1800 .smp_send_reschedule = voyager_smp_send_reschedule, 1812 .smp_send_reschedule = voyager_smp_send_reschedule,
1801 1813
1802 .send_call_func_ipi = native_send_call_func_ipi, 1814 .send_call_func_ipi = voyager_send_call_func,
1803 .send_call_func_single_ipi = native_send_call_func_single_ipi, 1815 .send_call_func_single_ipi = voyager_send_call_func_single,
1804}; 1816};
diff --git a/arch/x86/mm/fault.c b/arch/x86/mm/fault.c
index 31e8730fa246..20ef272c412c 100644
--- a/arch/x86/mm/fault.c
+++ b/arch/x86/mm/fault.c
@@ -413,6 +413,7 @@ static noinline void pgtable_bad(unsigned long address, struct pt_regs *regs,
413 unsigned long error_code) 413 unsigned long error_code)
414{ 414{
415 unsigned long flags = oops_begin(); 415 unsigned long flags = oops_begin();
416 int sig = SIGKILL;
416 struct task_struct *tsk; 417 struct task_struct *tsk;
417 418
418 printk(KERN_ALERT "%s: Corrupted page table at address %lx\n", 419 printk(KERN_ALERT "%s: Corrupted page table at address %lx\n",
@@ -423,8 +424,8 @@ static noinline void pgtable_bad(unsigned long address, struct pt_regs *regs,
423 tsk->thread.trap_no = 14; 424 tsk->thread.trap_no = 14;
424 tsk->thread.error_code = error_code; 425 tsk->thread.error_code = error_code;
425 if (__die("Bad pagetable", regs, error_code)) 426 if (__die("Bad pagetable", regs, error_code))
426 regs = NULL; 427 sig = 0;
427 oops_end(flags, regs, SIGKILL); 428 oops_end(flags, regs, sig);
428} 429}
429#endif 430#endif
430 431
@@ -590,6 +591,7 @@ void __kprobes do_page_fault(struct pt_regs *regs, unsigned long error_code)
590 int fault; 591 int fault;
591#ifdef CONFIG_X86_64 592#ifdef CONFIG_X86_64
592 unsigned long flags; 593 unsigned long flags;
594 int sig;
593#endif 595#endif
594 596
595 tsk = current; 597 tsk = current;
@@ -849,11 +851,12 @@ no_context:
849 bust_spinlocks(0); 851 bust_spinlocks(0);
850 do_exit(SIGKILL); 852 do_exit(SIGKILL);
851#else 853#else
854 sig = SIGKILL;
852 if (__die("Oops", regs, error_code)) 855 if (__die("Oops", regs, error_code))
853 regs = NULL; 856 sig = 0;
854 /* Executive summary in case the body of the oops scrolled away */ 857 /* Executive summary in case the body of the oops scrolled away */
855 printk(KERN_EMERG "CR2: %016lx\n", address); 858 printk(KERN_EMERG "CR2: %016lx\n", address);
856 oops_end(flags, regs, SIGKILL); 859 oops_end(flags, regs, sig);
857#endif 860#endif
858 861
859/* 862/*
diff --git a/arch/x86/mm/init_32.c b/arch/x86/mm/init_32.c
index c483f4242079..800e1d94c1b5 100644
--- a/arch/x86/mm/init_32.c
+++ b/arch/x86/mm/init_32.c
@@ -67,7 +67,7 @@ static unsigned long __meminitdata table_top;
67 67
68static int __initdata after_init_bootmem; 68static int __initdata after_init_bootmem;
69 69
70static __init void *alloc_low_page(unsigned long *phys) 70static __init void *alloc_low_page(void)
71{ 71{
72 unsigned long pfn = table_end++; 72 unsigned long pfn = table_end++;
73 void *adr; 73 void *adr;
@@ -77,7 +77,6 @@ static __init void *alloc_low_page(unsigned long *phys)
77 77
78 adr = __va(pfn * PAGE_SIZE); 78 adr = __va(pfn * PAGE_SIZE);
79 memset(adr, 0, PAGE_SIZE); 79 memset(adr, 0, PAGE_SIZE);
80 *phys = pfn * PAGE_SIZE;
81 return adr; 80 return adr;
82} 81}
83 82
@@ -92,16 +91,17 @@ static pmd_t * __init one_md_table_init(pgd_t *pgd)
92 pmd_t *pmd_table; 91 pmd_t *pmd_table;
93 92
94#ifdef CONFIG_X86_PAE 93#ifdef CONFIG_X86_PAE
95 unsigned long phys;
96 if (!(pgd_val(*pgd) & _PAGE_PRESENT)) { 94 if (!(pgd_val(*pgd) & _PAGE_PRESENT)) {
97 if (after_init_bootmem) 95 if (after_init_bootmem)
98 pmd_table = (pmd_t *)alloc_bootmem_low_pages(PAGE_SIZE); 96 pmd_table = (pmd_t *)alloc_bootmem_low_pages(PAGE_SIZE);
99 else 97 else
100 pmd_table = (pmd_t *)alloc_low_page(&phys); 98 pmd_table = (pmd_t *)alloc_low_page();
101 paravirt_alloc_pmd(&init_mm, __pa(pmd_table) >> PAGE_SHIFT); 99 paravirt_alloc_pmd(&init_mm, __pa(pmd_table) >> PAGE_SHIFT);
102 set_pgd(pgd, __pgd(__pa(pmd_table) | _PAGE_PRESENT)); 100 set_pgd(pgd, __pgd(__pa(pmd_table) | _PAGE_PRESENT));
103 pud = pud_offset(pgd, 0); 101 pud = pud_offset(pgd, 0);
104 BUG_ON(pmd_table != pmd_offset(pud, 0)); 102 BUG_ON(pmd_table != pmd_offset(pud, 0));
103
104 return pmd_table;
105 } 105 }
106#endif 106#endif
107 pud = pud_offset(pgd, 0); 107 pud = pud_offset(pgd, 0);
@@ -126,10 +126,8 @@ static pte_t * __init one_page_table_init(pmd_t *pmd)
126 if (!page_table) 126 if (!page_table)
127 page_table = 127 page_table =
128 (pte_t *)alloc_bootmem_low_pages(PAGE_SIZE); 128 (pte_t *)alloc_bootmem_low_pages(PAGE_SIZE);
129 } else { 129 } else
130 unsigned long phys; 130 page_table = (pte_t *)alloc_low_page();
131 page_table = (pte_t *)alloc_low_page(&phys);
132 }
133 131
134 paravirt_alloc_pte(&init_mm, __pa(page_table) >> PAGE_SHIFT); 132 paravirt_alloc_pte(&init_mm, __pa(page_table) >> PAGE_SHIFT);
135 set_pmd(pmd, __pmd(__pa(page_table) | _PAGE_TABLE)); 133 set_pmd(pmd, __pmd(__pa(page_table) | _PAGE_TABLE));
@@ -969,8 +967,6 @@ void __init mem_init(void)
969 int codesize, reservedpages, datasize, initsize; 967 int codesize, reservedpages, datasize, initsize;
970 int tmp; 968 int tmp;
971 969
972 start_periodic_check_for_corruption();
973
974#ifdef CONFIG_FLATMEM 970#ifdef CONFIG_FLATMEM
975 BUG_ON(!mem_map); 971 BUG_ON(!mem_map);
976#endif 972#endif
@@ -1040,11 +1036,25 @@ void __init mem_init(void)
1040 (unsigned long)&_text, (unsigned long)&_etext, 1036 (unsigned long)&_text, (unsigned long)&_etext,
1041 ((unsigned long)&_etext - (unsigned long)&_text) >> 10); 1037 ((unsigned long)&_etext - (unsigned long)&_text) >> 10);
1042 1038
1039 /*
1040 * Check boundaries twice: Some fundamental inconsistencies can
1041 * be detected at build time already.
1042 */
1043#define __FIXADDR_TOP (-PAGE_SIZE)
1044#ifdef CONFIG_HIGHMEM
1045 BUILD_BUG_ON(PKMAP_BASE + LAST_PKMAP*PAGE_SIZE > FIXADDR_START);
1046 BUILD_BUG_ON(VMALLOC_END > PKMAP_BASE);
1047#endif
1048#define high_memory (-128UL << 20)
1049 BUILD_BUG_ON(VMALLOC_START >= VMALLOC_END);
1050#undef high_memory
1051#undef __FIXADDR_TOP
1052
1043#ifdef CONFIG_HIGHMEM 1053#ifdef CONFIG_HIGHMEM
1044 BUG_ON(PKMAP_BASE + LAST_PKMAP*PAGE_SIZE > FIXADDR_START); 1054 BUG_ON(PKMAP_BASE + LAST_PKMAP*PAGE_SIZE > FIXADDR_START);
1045 BUG_ON(VMALLOC_END > PKMAP_BASE); 1055 BUG_ON(VMALLOC_END > PKMAP_BASE);
1046#endif 1056#endif
1047 BUG_ON(VMALLOC_START > VMALLOC_END); 1057 BUG_ON(VMALLOC_START >= VMALLOC_END);
1048 BUG_ON((unsigned long)high_memory > VMALLOC_START); 1058 BUG_ON((unsigned long)high_memory > VMALLOC_START);
1049 1059
1050 if (boot_cpu_data.wp_works_ok < 0) 1060 if (boot_cpu_data.wp_works_ok < 0)
diff --git a/arch/x86/mm/init_64.c b/arch/x86/mm/init_64.c
index 9db01db6e3cd..9f7a0d24d42a 100644
--- a/arch/x86/mm/init_64.c
+++ b/arch/x86/mm/init_64.c
@@ -902,8 +902,6 @@ void __init mem_init(void)
902 long codesize, reservedpages, datasize, initsize; 902 long codesize, reservedpages, datasize, initsize;
903 unsigned long absent_pages; 903 unsigned long absent_pages;
904 904
905 start_periodic_check_for_corruption();
906
907 pci_iommu_alloc(); 905 pci_iommu_alloc();
908 906
909 /* clear_bss() already clear the empty_zero_page */ 907 /* clear_bss() already clear the empty_zero_page */
diff --git a/arch/x86/mm/ioremap.c b/arch/x86/mm/ioremap.c
index d4c4307ff3e0..bd85d42819e1 100644
--- a/arch/x86/mm/ioremap.c
+++ b/arch/x86/mm/ioremap.c
@@ -223,7 +223,8 @@ static void __iomem *__ioremap_caller(resource_size_t phys_addr,
223 * Check if the request spans more than any BAR in the iomem resource 223 * Check if the request spans more than any BAR in the iomem resource
224 * tree. 224 * tree.
225 */ 225 */
226 WARN_ON(iomem_map_sanity_check(phys_addr, size)); 226 WARN_ONCE(iomem_map_sanity_check(phys_addr, size),
227 KERN_INFO "Info: mapping multiple BARs. Your kernel is fine.");
227 228
228 /* 229 /*
229 * Don't allow anybody to remap normal RAM that we're using.. 230 * Don't allow anybody to remap normal RAM that we're using..
diff --git a/arch/x86/mm/numa_32.c b/arch/x86/mm/numa_32.c
index 847c164725f4..8518c678d83f 100644
--- a/arch/x86/mm/numa_32.c
+++ b/arch/x86/mm/numa_32.c
@@ -222,6 +222,41 @@ static void __init remap_numa_kva(void)
222 } 222 }
223} 223}
224 224
225#ifdef CONFIG_HIBERNATION
226/**
227 * resume_map_numa_kva - add KVA mapping to the temporary page tables created
228 * during resume from hibernation
229 * @pgd_base - temporary resume page directory
230 */
231void resume_map_numa_kva(pgd_t *pgd_base)
232{
233 int node;
234
235 for_each_online_node(node) {
236 unsigned long start_va, start_pfn, size, pfn;
237
238 start_va = (unsigned long)node_remap_start_vaddr[node];
239 start_pfn = node_remap_start_pfn[node];
240 size = node_remap_size[node];
241
242 printk(KERN_DEBUG "%s: node %d\n", __FUNCTION__, node);
243
244 for (pfn = 0; pfn < size; pfn += PTRS_PER_PTE) {
245 unsigned long vaddr = start_va + (pfn << PAGE_SHIFT);
246 pgd_t *pgd = pgd_base + pgd_index(vaddr);
247 pud_t *pud = pud_offset(pgd, vaddr);
248 pmd_t *pmd = pmd_offset(pud, vaddr);
249
250 set_pmd(pmd, pfn_pmd(start_pfn + pfn,
251 PAGE_KERNEL_LARGE_EXEC));
252
253 printk(KERN_DEBUG "%s: %08lx -> pfn %08lx\n",
254 __FUNCTION__, vaddr, start_pfn + pfn);
255 }
256 }
257}
258#endif
259
225static unsigned long calculate_numa_remap_pages(void) 260static unsigned long calculate_numa_remap_pages(void)
226{ 261{
227 int nid; 262 int nid;
diff --git a/arch/x86/mm/pat.c b/arch/x86/mm/pat.c
index eb1bf000d12e..541bcc944a5b 100644
--- a/arch/x86/mm/pat.c
+++ b/arch/x86/mm/pat.c
@@ -596,6 +596,242 @@ void unmap_devmem(unsigned long pfn, unsigned long size, pgprot_t vma_prot)
596 free_memtype(addr, addr + size); 596 free_memtype(addr, addr + size);
597} 597}
598 598
599/*
600 * Internal interface to reserve a range of physical memory with prot.
601 * Reserved non RAM regions only and after successful reserve_memtype,
602 * this func also keeps identity mapping (if any) in sync with this new prot.
603 */
604static int reserve_pfn_range(u64 paddr, unsigned long size, pgprot_t vma_prot)
605{
606 int is_ram = 0;
607 int id_sz, ret;
608 unsigned long flags;
609 unsigned long want_flags = (pgprot_val(vma_prot) & _PAGE_CACHE_MASK);
610
611 is_ram = pagerange_is_ram(paddr, paddr + size);
612
613 if (is_ram != 0) {
614 /*
615 * For mapping RAM pages, drivers need to call
616 * set_memory_[uc|wc|wb] directly, for reserve and free, before
617 * setting up the PTE.
618 */
619 WARN_ON_ONCE(1);
620 return 0;
621 }
622
623 ret = reserve_memtype(paddr, paddr + size, want_flags, &flags);
624 if (ret)
625 return ret;
626
627 if (flags != want_flags) {
628 free_memtype(paddr, paddr + size);
629 printk(KERN_ERR
630 "%s:%d map pfn expected mapping type %s for %Lx-%Lx, got %s\n",
631 current->comm, current->pid,
632 cattr_name(want_flags),
633 (unsigned long long)paddr,
634 (unsigned long long)(paddr + size),
635 cattr_name(flags));
636 return -EINVAL;
637 }
638
639 /* Need to keep identity mapping in sync */
640 if (paddr >= __pa(high_memory))
641 return 0;
642
643 id_sz = (__pa(high_memory) < paddr + size) ?
644 __pa(high_memory) - paddr :
645 size;
646
647 if (ioremap_change_attr((unsigned long)__va(paddr), id_sz, flags) < 0) {
648 free_memtype(paddr, paddr + size);
649 printk(KERN_ERR
650 "%s:%d reserve_pfn_range ioremap_change_attr failed %s "
651 "for %Lx-%Lx\n",
652 current->comm, current->pid,
653 cattr_name(flags),
654 (unsigned long long)paddr,
655 (unsigned long long)(paddr + size));
656 return -EINVAL;
657 }
658 return 0;
659}
660
661/*
662 * Internal interface to free a range of physical memory.
663 * Frees non RAM regions only.
664 */
665static void free_pfn_range(u64 paddr, unsigned long size)
666{
667 int is_ram;
668
669 is_ram = pagerange_is_ram(paddr, paddr + size);
670 if (is_ram == 0)
671 free_memtype(paddr, paddr + size);
672}
673
674/*
675 * track_pfn_vma_copy is called when vma that is covering the pfnmap gets
676 * copied through copy_page_range().
677 *
678 * If the vma has a linear pfn mapping for the entire range, we get the prot
679 * from pte and reserve the entire vma range with single reserve_pfn_range call.
680 * Otherwise, we reserve the entire vma range, my ging through the PTEs page
681 * by page to get physical address and protection.
682 */
683int track_pfn_vma_copy(struct vm_area_struct *vma)
684{
685 int retval = 0;
686 unsigned long i, j;
687 u64 paddr;
688 unsigned long prot;
689 unsigned long vma_start = vma->vm_start;
690 unsigned long vma_end = vma->vm_end;
691 unsigned long vma_size = vma_end - vma_start;
692
693 if (!pat_enabled)
694 return 0;
695
696 if (is_linear_pfn_mapping(vma)) {
697 /*
698 * reserve the whole chunk covered by vma. We need the
699 * starting address and protection from pte.
700 */
701 if (follow_phys(vma, vma_start, 0, &prot, &paddr)) {
702 WARN_ON_ONCE(1);
703 return -EINVAL;
704 }
705 return reserve_pfn_range(paddr, vma_size, __pgprot(prot));
706 }
707
708 /* reserve entire vma page by page, using pfn and prot from pte */
709 for (i = 0; i < vma_size; i += PAGE_SIZE) {
710 if (follow_phys(vma, vma_start + i, 0, &prot, &paddr))
711 continue;
712
713 retval = reserve_pfn_range(paddr, PAGE_SIZE, __pgprot(prot));
714 if (retval)
715 goto cleanup_ret;
716 }
717 return 0;
718
719cleanup_ret:
720 /* Reserve error: Cleanup partial reservation and return error */
721 for (j = 0; j < i; j += PAGE_SIZE) {
722 if (follow_phys(vma, vma_start + j, 0, &prot, &paddr))
723 continue;
724
725 free_pfn_range(paddr, PAGE_SIZE);
726 }
727
728 return retval;
729}
730
731/*
732 * track_pfn_vma_new is called when a _new_ pfn mapping is being established
733 * for physical range indicated by pfn and size.
734 *
735 * prot is passed in as a parameter for the new mapping. If the vma has a
736 * linear pfn mapping for the entire range reserve the entire vma range with
737 * single reserve_pfn_range call.
738 * Otherwise, we look t the pfn and size and reserve only the specified range
739 * page by page.
740 *
741 * Note that this function can be called with caller trying to map only a
742 * subrange/page inside the vma.
743 */
744int track_pfn_vma_new(struct vm_area_struct *vma, pgprot_t prot,
745 unsigned long pfn, unsigned long size)
746{
747 int retval = 0;
748 unsigned long i, j;
749 u64 base_paddr;
750 u64 paddr;
751 unsigned long vma_start = vma->vm_start;
752 unsigned long vma_end = vma->vm_end;
753 unsigned long vma_size = vma_end - vma_start;
754
755 if (!pat_enabled)
756 return 0;
757
758 if (is_linear_pfn_mapping(vma)) {
759 /* reserve the whole chunk starting from vm_pgoff */
760 paddr = (u64)vma->vm_pgoff << PAGE_SHIFT;
761 return reserve_pfn_range(paddr, vma_size, prot);
762 }
763
764 /* reserve page by page using pfn and size */
765 base_paddr = (u64)pfn << PAGE_SHIFT;
766 for (i = 0; i < size; i += PAGE_SIZE) {
767 paddr = base_paddr + i;
768 retval = reserve_pfn_range(paddr, PAGE_SIZE, prot);
769 if (retval)
770 goto cleanup_ret;
771 }
772 return 0;
773
774cleanup_ret:
775 /* Reserve error: Cleanup partial reservation and return error */
776 for (j = 0; j < i; j += PAGE_SIZE) {
777 paddr = base_paddr + j;
778 free_pfn_range(paddr, PAGE_SIZE);
779 }
780
781 return retval;
782}
783
784/*
785 * untrack_pfn_vma is called while unmapping a pfnmap for a region.
786 * untrack can be called for a specific region indicated by pfn and size or
787 * can be for the entire vma (in which case size can be zero).
788 */
789void untrack_pfn_vma(struct vm_area_struct *vma, unsigned long pfn,
790 unsigned long size)
791{
792 unsigned long i;
793 u64 paddr;
794 unsigned long prot;
795 unsigned long vma_start = vma->vm_start;
796 unsigned long vma_end = vma->vm_end;
797 unsigned long vma_size = vma_end - vma_start;
798
799 if (!pat_enabled)
800 return;
801
802 if (is_linear_pfn_mapping(vma)) {
803 /* free the whole chunk starting from vm_pgoff */
804 paddr = (u64)vma->vm_pgoff << PAGE_SHIFT;
805 free_pfn_range(paddr, vma_size);
806 return;
807 }
808
809 if (size != 0 && size != vma_size) {
810 /* free page by page, using pfn and size */
811 paddr = (u64)pfn << PAGE_SHIFT;
812 for (i = 0; i < size; i += PAGE_SIZE) {
813 paddr = paddr + i;
814 free_pfn_range(paddr, PAGE_SIZE);
815 }
816 } else {
817 /* free entire vma, page by page, using the pfn from pte */
818 for (i = 0; i < vma_size; i += PAGE_SIZE) {
819 if (follow_phys(vma, vma_start + i, 0, &prot, &paddr))
820 continue;
821
822 free_pfn_range(paddr, PAGE_SIZE);
823 }
824 }
825}
826
827pgprot_t pgprot_writecombine(pgprot_t prot)
828{
829 if (pat_enabled)
830 return __pgprot(pgprot_val(prot) | _PAGE_CACHE_WC);
831 else
832 return pgprot_noncached(prot);
833}
834
599#if defined(CONFIG_DEBUG_FS) && defined(CONFIG_X86_PAT) 835#if defined(CONFIG_DEBUG_FS) && defined(CONFIG_X86_PAT)
600 836
601/* get Nth element of the linked list */ 837/* get Nth element of the linked list */
diff --git a/arch/x86/oprofile/nmi_int.c b/arch/x86/oprofile/nmi_int.c
index 022cd41ea9b4..202864ad49a7 100644
--- a/arch/x86/oprofile/nmi_int.c
+++ b/arch/x86/oprofile/nmi_int.c
@@ -401,14 +401,13 @@ static int __init ppro_init(char **cpu_type)
401 *cpu_type = "i386/pii"; 401 *cpu_type = "i386/pii";
402 break; 402 break;
403 case 6 ... 8: 403 case 6 ... 8:
404 case 10 ... 11:
404 *cpu_type = "i386/piii"; 405 *cpu_type = "i386/piii";
405 break; 406 break;
406 case 9: 407 case 9:
408 case 13:
407 *cpu_type = "i386/p6_mobile"; 409 *cpu_type = "i386/p6_mobile";
408 break; 410 break;
409 case 10 ... 13:
410 *cpu_type = "i386/p6";
411 break;
412 case 14: 411 case 14:
413 *cpu_type = "i386/core"; 412 *cpu_type = "i386/core";
414 break; 413 break;
diff --git a/arch/x86/oprofile/op_model_ppro.c b/arch/x86/oprofile/op_model_ppro.c
index 3f1b81a83e2e..e9f80c744cf3 100644
--- a/arch/x86/oprofile/op_model_ppro.c
+++ b/arch/x86/oprofile/op_model_ppro.c
@@ -69,7 +69,7 @@ static void ppro_setup_ctrs(struct op_msrs const * const msrs)
69 int i; 69 int i;
70 70
71 if (!reset_value) { 71 if (!reset_value) {
72 reset_value = kmalloc(sizeof(unsigned) * num_counters, 72 reset_value = kmalloc(sizeof(reset_value[0]) * num_counters,
73 GFP_ATOMIC); 73 GFP_ATOMIC);
74 if (!reset_value) 74 if (!reset_value)
75 return; 75 return;
@@ -156,6 +156,8 @@ static void ppro_start(struct op_msrs const * const msrs)
156 unsigned int low, high; 156 unsigned int low, high;
157 int i; 157 int i;
158 158
159 if (!reset_value)
160 return;
159 for (i = 0; i < num_counters; ++i) { 161 for (i = 0; i < num_counters; ++i) {
160 if (reset_value[i]) { 162 if (reset_value[i]) {
161 CTRL_READ(low, high, msrs, i); 163 CTRL_READ(low, high, msrs, i);
@@ -171,6 +173,8 @@ static void ppro_stop(struct op_msrs const * const msrs)
171 unsigned int low, high; 173 unsigned int low, high;
172 int i; 174 int i;
173 175
176 if (!reset_value)
177 return;
174 for (i = 0; i < num_counters; ++i) { 178 for (i = 0; i < num_counters; ++i) {
175 if (!reset_value[i]) 179 if (!reset_value[i])
176 continue; 180 continue;
diff --git a/arch/x86/pci/common.c b/arch/x86/pci/common.c
index b67732bbb85a..bb1a01f089e2 100644
--- a/arch/x86/pci/common.c
+++ b/arch/x86/pci/common.c
@@ -23,6 +23,12 @@ unsigned int pci_probe = PCI_PROBE_BIOS | PCI_PROBE_CONF1 | PCI_PROBE_CONF2 |
23unsigned int pci_early_dump_regs; 23unsigned int pci_early_dump_regs;
24static int pci_bf_sort; 24static int pci_bf_sort;
25int pci_routeirq; 25int pci_routeirq;
26int noioapicquirk;
27#ifdef CONFIG_X86_REROUTE_FOR_BROKEN_BOOT_IRQS
28int noioapicreroute = 0;
29#else
30int noioapicreroute = 1;
31#endif
26int pcibios_last_bus = -1; 32int pcibios_last_bus = -1;
27unsigned long pirq_table_addr; 33unsigned long pirq_table_addr;
28struct pci_bus *pci_root_bus; 34struct pci_bus *pci_root_bus;
@@ -519,6 +525,17 @@ char * __devinit pcibios_setup(char *str)
519 } else if (!strcmp(str, "skip_isa_align")) { 525 } else if (!strcmp(str, "skip_isa_align")) {
520 pci_probe |= PCI_CAN_SKIP_ISA_ALIGN; 526 pci_probe |= PCI_CAN_SKIP_ISA_ALIGN;
521 return NULL; 527 return NULL;
528 } else if (!strcmp(str, "noioapicquirk")) {
529 noioapicquirk = 1;
530 return NULL;
531 } else if (!strcmp(str, "ioapicreroute")) {
532 if (noioapicreroute != -1)
533 noioapicreroute = 0;
534 return NULL;
535 } else if (!strcmp(str, "noioapicreroute")) {
536 if (noioapicreroute != -1)
537 noioapicreroute = 1;
538 return NULL;
522 } 539 }
523 return str; 540 return str;
524} 541}
diff --git a/arch/x86/pci/direct.c b/arch/x86/pci/direct.c
index 9915293500fb..9a5af6c8fbe9 100644
--- a/arch/x86/pci/direct.c
+++ b/arch/x86/pci/direct.c
@@ -173,7 +173,7 @@ static int pci_conf2_write(unsigned int seg, unsigned int bus,
173 173
174#undef PCI_CONF2_ADDRESS 174#undef PCI_CONF2_ADDRESS
175 175
176static struct pci_raw_ops pci_direct_conf2 = { 176struct pci_raw_ops pci_direct_conf2 = {
177 .read = pci_conf2_read, 177 .read = pci_conf2_read,
178 .write = pci_conf2_write, 178 .write = pci_conf2_write,
179}; 179};
@@ -289,6 +289,7 @@ int __init pci_direct_probe(void)
289 289
290 if (pci_check_type1()) { 290 if (pci_check_type1()) {
291 raw_pci_ops = &pci_direct_conf1; 291 raw_pci_ops = &pci_direct_conf1;
292 port_cf9_safe = true;
292 return 1; 293 return 1;
293 } 294 }
294 release_resource(region); 295 release_resource(region);
@@ -305,6 +306,7 @@ int __init pci_direct_probe(void)
305 306
306 if (pci_check_type2()) { 307 if (pci_check_type2()) {
307 raw_pci_ops = &pci_direct_conf2; 308 raw_pci_ops = &pci_direct_conf2;
309 port_cf9_safe = true;
308 return 2; 310 return 2;
309 } 311 }
310 312
diff --git a/arch/x86/pci/fixup.c b/arch/x86/pci/fixup.c
index 3c27a809393b..2051dc96b8e9 100644
--- a/arch/x86/pci/fixup.c
+++ b/arch/x86/pci/fixup.c
@@ -496,21 +496,24 @@ DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_SIEMENS, 0x0015,
496 pci_siemens_interrupt_controller); 496 pci_siemens_interrupt_controller);
497 497
498/* 498/*
499 * Regular PCI devices have 256 bytes, but AMD Family 10h Opteron ext config 499 * Regular PCI devices have 256 bytes, but AMD Family 10h/11h CPUs have
500 * have 4096 bytes. Even if the device is capable, that doesn't mean we can 500 * 4096 bytes configuration space for each function of their processor
501 * access it. Maybe we don't have a way to generate extended config space 501 * configuration space.
502 * accesses. So check it
503 */ 502 */
504static void fam10h_pci_cfg_space_size(struct pci_dev *dev) 503static void amd_cpu_pci_cfg_space_size(struct pci_dev *dev)
505{ 504{
506 dev->cfg_size = pci_cfg_space_size_ext(dev); 505 dev->cfg_size = pci_cfg_space_size_ext(dev);
507} 506}
508 507DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_AMD, 0x1200, amd_cpu_pci_cfg_space_size);
509DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_AMD, 0x1200, fam10h_pci_cfg_space_size); 508DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_AMD, 0x1201, amd_cpu_pci_cfg_space_size);
510DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_AMD, 0x1201, fam10h_pci_cfg_space_size); 509DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_AMD, 0x1202, amd_cpu_pci_cfg_space_size);
511DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_AMD, 0x1202, fam10h_pci_cfg_space_size); 510DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_AMD, 0x1203, amd_cpu_pci_cfg_space_size);
512DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_AMD, 0x1203, fam10h_pci_cfg_space_size); 511DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_AMD, 0x1204, amd_cpu_pci_cfg_space_size);
513DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_AMD, 0x1204, fam10h_pci_cfg_space_size); 512DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_AMD, 0x1300, amd_cpu_pci_cfg_space_size);
513DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_AMD, 0x1301, amd_cpu_pci_cfg_space_size);
514DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_AMD, 0x1302, amd_cpu_pci_cfg_space_size);
515DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_AMD, 0x1303, amd_cpu_pci_cfg_space_size);
516DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_AMD, 0x1304, amd_cpu_pci_cfg_space_size);
514 517
515/* 518/*
516 * SB600: Disable BAR1 on device 14.0 to avoid HPET resources from 519 * SB600: Disable BAR1 on device 14.0 to avoid HPET resources from
diff --git a/arch/x86/pci/pci.h b/arch/x86/pci/pci.h
index 15b9cf6be729..1959018aac02 100644
--- a/arch/x86/pci/pci.h
+++ b/arch/x86/pci/pci.h
@@ -96,6 +96,7 @@ extern struct pci_raw_ops *raw_pci_ops;
96extern struct pci_raw_ops *raw_pci_ext_ops; 96extern struct pci_raw_ops *raw_pci_ext_ops;
97 97
98extern struct pci_raw_ops pci_direct_conf1; 98extern struct pci_raw_ops pci_direct_conf1;
99extern bool port_cf9_safe;
99 100
100/* arch_initcall level */ 101/* arch_initcall level */
101extern int pci_direct_probe(void); 102extern int pci_direct_probe(void);
diff --git a/arch/x86/power/hibernate_32.c b/arch/x86/power/hibernate_32.c
index f2b6e3f11bfc..81197c62d5b3 100644
--- a/arch/x86/power/hibernate_32.c
+++ b/arch/x86/power/hibernate_32.c
@@ -12,6 +12,7 @@
12#include <asm/system.h> 12#include <asm/system.h>
13#include <asm/page.h> 13#include <asm/page.h>
14#include <asm/pgtable.h> 14#include <asm/pgtable.h>
15#include <asm/mmzone.h>
15 16
16/* Defined in hibernate_asm_32.S */ 17/* Defined in hibernate_asm_32.S */
17extern int restore_image(void); 18extern int restore_image(void);
@@ -127,6 +128,9 @@ static int resume_physical_mapping_init(pgd_t *pgd_base)
127 } 128 }
128 } 129 }
129 } 130 }
131
132 resume_map_numa_kva(pgd_base);
133
130 return 0; 134 return 0;
131} 135}
132 136
diff --git a/arch/x86/xen/enlighten.c b/arch/x86/xen/enlighten.c
index 5e4686d70f62..bea215230b20 100644
--- a/arch/x86/xen/enlighten.c
+++ b/arch/x86/xen/enlighten.c
@@ -28,6 +28,7 @@
28#include <linux/console.h> 28#include <linux/console.h>
29 29
30#include <xen/interface/xen.h> 30#include <xen/interface/xen.h>
31#include <xen/interface/version.h>
31#include <xen/interface/physdev.h> 32#include <xen/interface/physdev.h>
32#include <xen/interface/vcpu.h> 33#include <xen/interface/vcpu.h>
33#include <xen/features.h> 34#include <xen/features.h>
@@ -793,7 +794,7 @@ static int xen_write_msr_safe(unsigned int msr, unsigned low, unsigned high)
793 794
794 ret = 0; 795 ret = 0;
795 796
796 switch(msr) { 797 switch (msr) {
797#ifdef CONFIG_X86_64 798#ifdef CONFIG_X86_64
798 unsigned which; 799 unsigned which;
799 u64 base; 800 u64 base;
@@ -1453,7 +1454,7 @@ static __init void xen_map_identity_early(pmd_t *pmd, unsigned long max_pfn)
1453 1454
1454 ident_pte = 0; 1455 ident_pte = 0;
1455 pfn = 0; 1456 pfn = 0;
1456 for(pmdidx = 0; pmdidx < PTRS_PER_PMD && pfn < max_pfn; pmdidx++) { 1457 for (pmdidx = 0; pmdidx < PTRS_PER_PMD && pfn < max_pfn; pmdidx++) {
1457 pte_t *pte_page; 1458 pte_t *pte_page;
1458 1459
1459 /* Reuse or allocate a page of ptes */ 1460 /* Reuse or allocate a page of ptes */
@@ -1471,7 +1472,7 @@ static __init void xen_map_identity_early(pmd_t *pmd, unsigned long max_pfn)
1471 } 1472 }
1472 1473
1473 /* Install mappings */ 1474 /* Install mappings */
1474 for(pteidx = 0; pteidx < PTRS_PER_PTE; pteidx++, pfn++) { 1475 for (pteidx = 0; pteidx < PTRS_PER_PTE; pteidx++, pfn++) {
1475 pte_t pte; 1476 pte_t pte;
1476 1477
1477 if (pfn > max_pfn_mapped) 1478 if (pfn > max_pfn_mapped)
@@ -1485,7 +1486,7 @@ static __init void xen_map_identity_early(pmd_t *pmd, unsigned long max_pfn)
1485 } 1486 }
1486 } 1487 }
1487 1488
1488 for(pteidx = 0; pteidx < ident_pte; pteidx += PTRS_PER_PTE) 1489 for (pteidx = 0; pteidx < ident_pte; pteidx += PTRS_PER_PTE)
1489 set_page_prot(&level1_ident_pgt[pteidx], PAGE_KERNEL_RO); 1490 set_page_prot(&level1_ident_pgt[pteidx], PAGE_KERNEL_RO);
1490 1491
1491 set_page_prot(pmd, PAGE_KERNEL_RO); 1492 set_page_prot(pmd, PAGE_KERNEL_RO);
@@ -1499,7 +1500,7 @@ static void convert_pfn_mfn(void *v)
1499 1500
1500 /* All levels are converted the same way, so just treat them 1501 /* All levels are converted the same way, so just treat them
1501 as ptes. */ 1502 as ptes. */
1502 for(i = 0; i < PTRS_PER_PTE; i++) 1503 for (i = 0; i < PTRS_PER_PTE; i++)
1503 pte[i] = xen_make_pte(pte[i].pte); 1504 pte[i] = xen_make_pte(pte[i].pte);
1504} 1505}
1505 1506
@@ -1514,7 +1515,8 @@ static void convert_pfn_mfn(void *v)
1514 * of the physical mapping once some sort of allocator has been set 1515 * of the physical mapping once some sort of allocator has been set
1515 * up. 1516 * up.
1516 */ 1517 */
1517static __init pgd_t *xen_setup_kernel_pagetable(pgd_t *pgd, unsigned long max_pfn) 1518static __init pgd_t *xen_setup_kernel_pagetable(pgd_t *pgd,
1519 unsigned long max_pfn)
1518{ 1520{
1519 pud_t *l3; 1521 pud_t *l3;
1520 pmd_t *l2; 1522 pmd_t *l2;
@@ -1577,7 +1579,8 @@ static __init pgd_t *xen_setup_kernel_pagetable(pgd_t *pgd, unsigned long max_pf
1577#else /* !CONFIG_X86_64 */ 1579#else /* !CONFIG_X86_64 */
1578static pmd_t level2_kernel_pgt[PTRS_PER_PMD] __page_aligned_bss; 1580static pmd_t level2_kernel_pgt[PTRS_PER_PMD] __page_aligned_bss;
1579 1581
1580static __init pgd_t *xen_setup_kernel_pagetable(pgd_t *pgd, unsigned long max_pfn) 1582static __init pgd_t *xen_setup_kernel_pagetable(pgd_t *pgd,
1583 unsigned long max_pfn)
1581{ 1584{
1582 pmd_t *kernel_pmd; 1585 pmd_t *kernel_pmd;
1583 1586
diff --git a/arch/x86/xen/mmu.c b/arch/x86/xen/mmu.c
index 688936044dc9..773d68d3e912 100644
--- a/arch/x86/xen/mmu.c
+++ b/arch/x86/xen/mmu.c
@@ -154,13 +154,13 @@ void xen_setup_mfn_list_list(void)
154{ 154{
155 unsigned pfn, idx; 155 unsigned pfn, idx;
156 156
157 for(pfn = 0; pfn < MAX_DOMAIN_PAGES; pfn += P2M_ENTRIES_PER_PAGE) { 157 for (pfn = 0; pfn < MAX_DOMAIN_PAGES; pfn += P2M_ENTRIES_PER_PAGE) {
158 unsigned topidx = p2m_top_index(pfn); 158 unsigned topidx = p2m_top_index(pfn);
159 159
160 p2m_top_mfn[topidx] = virt_to_mfn(p2m_top[topidx]); 160 p2m_top_mfn[topidx] = virt_to_mfn(p2m_top[topidx]);
161 } 161 }
162 162
163 for(idx = 0; idx < ARRAY_SIZE(p2m_top_mfn_list); idx++) { 163 for (idx = 0; idx < ARRAY_SIZE(p2m_top_mfn_list); idx++) {
164 unsigned topidx = idx * P2M_ENTRIES_PER_PAGE; 164 unsigned topidx = idx * P2M_ENTRIES_PER_PAGE;
165 p2m_top_mfn_list[idx] = virt_to_mfn(&p2m_top_mfn[topidx]); 165 p2m_top_mfn_list[idx] = virt_to_mfn(&p2m_top_mfn[topidx]);
166 } 166 }
@@ -179,7 +179,7 @@ void __init xen_build_dynamic_phys_to_machine(void)
179 unsigned long max_pfn = min(MAX_DOMAIN_PAGES, xen_start_info->nr_pages); 179 unsigned long max_pfn = min(MAX_DOMAIN_PAGES, xen_start_info->nr_pages);
180 unsigned pfn; 180 unsigned pfn;
181 181
182 for(pfn = 0; pfn < max_pfn; pfn += P2M_ENTRIES_PER_PAGE) { 182 for (pfn = 0; pfn < max_pfn; pfn += P2M_ENTRIES_PER_PAGE) {
183 unsigned topidx = p2m_top_index(pfn); 183 unsigned topidx = p2m_top_index(pfn);
184 184
185 p2m_top[topidx] = &mfn_list[pfn]; 185 p2m_top[topidx] = &mfn_list[pfn];
@@ -207,7 +207,7 @@ static void alloc_p2m(unsigned long **pp, unsigned long *mfnp)
207 p = (void *)__get_free_page(GFP_KERNEL | __GFP_NOFAIL); 207 p = (void *)__get_free_page(GFP_KERNEL | __GFP_NOFAIL);
208 BUG_ON(p == NULL); 208 BUG_ON(p == NULL);
209 209
210 for(i = 0; i < P2M_ENTRIES_PER_PAGE; i++) 210 for (i = 0; i < P2M_ENTRIES_PER_PAGE; i++)
211 p[i] = INVALID_P2M_ENTRY; 211 p[i] = INVALID_P2M_ENTRY;
212 212
213 if (cmpxchg(pp, p2m_missing, p) != p2m_missing) 213 if (cmpxchg(pp, p2m_missing, p) != p2m_missing)
@@ -407,7 +407,8 @@ out:
407 preempt_enable(); 407 preempt_enable();
408} 408}
409 409
410pte_t xen_ptep_modify_prot_start(struct mm_struct *mm, unsigned long addr, pte_t *ptep) 410pte_t xen_ptep_modify_prot_start(struct mm_struct *mm,
411 unsigned long addr, pte_t *ptep)
411{ 412{
412 /* Just return the pte as-is. We preserve the bits on commit */ 413 /* Just return the pte as-is. We preserve the bits on commit */
413 return *ptep; 414 return *ptep;
@@ -661,12 +662,11 @@ void xen_set_pgd(pgd_t *ptr, pgd_t val)
661 * For 64-bit, we must skip the Xen hole in the middle of the address 662 * For 64-bit, we must skip the Xen hole in the middle of the address
662 * space, just after the big x86-64 virtual hole. 663 * space, just after the big x86-64 virtual hole.
663 */ 664 */
664static int xen_pgd_walk(struct mm_struct *mm, 665static int __xen_pgd_walk(struct mm_struct *mm, pgd_t *pgd,
665 int (*func)(struct mm_struct *mm, struct page *, 666 int (*func)(struct mm_struct *mm, struct page *,
666 enum pt_level), 667 enum pt_level),
667 unsigned long limit) 668 unsigned long limit)
668{ 669{
669 pgd_t *pgd = mm->pgd;
670 int flush = 0; 670 int flush = 0;
671 unsigned hole_low, hole_high; 671 unsigned hole_low, hole_high;
672 unsigned pgdidx_limit, pudidx_limit, pmdidx_limit; 672 unsigned pgdidx_limit, pudidx_limit, pmdidx_limit;
@@ -753,6 +753,14 @@ out:
753 return flush; 753 return flush;
754} 754}
755 755
756static int xen_pgd_walk(struct mm_struct *mm,
757 int (*func)(struct mm_struct *mm, struct page *,
758 enum pt_level),
759 unsigned long limit)
760{
761 return __xen_pgd_walk(mm, mm->pgd, func, limit);
762}
763
756/* If we're using split pte locks, then take the page's lock and 764/* If we're using split pte locks, then take the page's lock and
757 return a pointer to it. Otherwise return NULL. */ 765 return a pointer to it. Otherwise return NULL. */
758static spinlock_t *xen_pte_lock(struct page *page, struct mm_struct *mm) 766static spinlock_t *xen_pte_lock(struct page *page, struct mm_struct *mm)
@@ -854,7 +862,7 @@ static void __xen_pgd_pin(struct mm_struct *mm, pgd_t *pgd)
854 862
855 xen_mc_batch(); 863 xen_mc_batch();
856 864
857 if (xen_pgd_walk(mm, xen_pin_page, USER_LIMIT)) { 865 if (__xen_pgd_walk(mm, pgd, xen_pin_page, USER_LIMIT)) {
858 /* re-enable interrupts for flushing */ 866 /* re-enable interrupts for flushing */
859 xen_mc_issue(0); 867 xen_mc_issue(0);
860 868
@@ -871,7 +879,8 @@ static void __xen_pgd_pin(struct mm_struct *mm, pgd_t *pgd)
871 879
872 if (user_pgd) { 880 if (user_pgd) {
873 xen_pin_page(mm, virt_to_page(user_pgd), PT_PGD); 881 xen_pin_page(mm, virt_to_page(user_pgd), PT_PGD);
874 xen_do_pin(MMUEXT_PIN_L4_TABLE, PFN_DOWN(__pa(user_pgd))); 882 xen_do_pin(MMUEXT_PIN_L4_TABLE,
883 PFN_DOWN(__pa(user_pgd)));
875 } 884 }
876 } 885 }
877#else /* CONFIG_X86_32 */ 886#else /* CONFIG_X86_32 */
@@ -986,7 +995,8 @@ static void __xen_pgd_unpin(struct mm_struct *mm, pgd_t *pgd)
986 pgd_t *user_pgd = xen_get_user_pgd(pgd); 995 pgd_t *user_pgd = xen_get_user_pgd(pgd);
987 996
988 if (user_pgd) { 997 if (user_pgd) {
989 xen_do_pin(MMUEXT_UNPIN_TABLE, PFN_DOWN(__pa(user_pgd))); 998 xen_do_pin(MMUEXT_UNPIN_TABLE,
999 PFN_DOWN(__pa(user_pgd)));
990 xen_unpin_page(mm, virt_to_page(user_pgd), PT_PGD); 1000 xen_unpin_page(mm, virt_to_page(user_pgd), PT_PGD);
991 } 1001 }
992 } 1002 }
@@ -998,7 +1008,7 @@ static void __xen_pgd_unpin(struct mm_struct *mm, pgd_t *pgd)
998 PT_PMD); 1008 PT_PMD);
999#endif 1009#endif
1000 1010
1001 xen_pgd_walk(mm, xen_unpin_page, USER_LIMIT); 1011 __xen_pgd_walk(mm, pgd, xen_unpin_page, USER_LIMIT);
1002 1012
1003 xen_mc_issue(0); 1013 xen_mc_issue(0);
1004} 1014}
diff --git a/arch/x86/xen/multicalls.c b/arch/x86/xen/multicalls.c
index 8ea8a0d0b0de..c738644b5435 100644
--- a/arch/x86/xen/multicalls.c
+++ b/arch/x86/xen/multicalls.c
@@ -154,7 +154,7 @@ void xen_mc_flush(void)
154 ret, smp_processor_id()); 154 ret, smp_processor_id());
155 dump_stack(); 155 dump_stack();
156 for (i = 0; i < b->mcidx; i++) { 156 for (i = 0; i < b->mcidx; i++) {
157 printk(" call %2d/%d: op=%lu arg=[%lx] result=%ld\n", 157 printk(KERN_DEBUG " call %2d/%d: op=%lu arg=[%lx] result=%ld\n",
158 i+1, b->mcidx, 158 i+1, b->mcidx,
159 b->debug[i].op, 159 b->debug[i].op,
160 b->debug[i].args[0], 160 b->debug[i].args[0],
diff --git a/arch/x86/xen/setup.c b/arch/x86/xen/setup.c
index d67901083888..15c6c68db6a2 100644
--- a/arch/x86/xen/setup.c
+++ b/arch/x86/xen/setup.c
@@ -28,6 +28,9 @@
28/* These are code, but not functions. Defined in entry.S */ 28/* These are code, but not functions. Defined in entry.S */
29extern const char xen_hypervisor_callback[]; 29extern const char xen_hypervisor_callback[];
30extern const char xen_failsafe_callback[]; 30extern const char xen_failsafe_callback[];
31extern void xen_sysenter_target(void);
32extern void xen_syscall_target(void);
33extern void xen_syscall32_target(void);
31 34
32 35
33/** 36/**
@@ -110,7 +113,6 @@ static __cpuinit int register_callback(unsigned type, const void *func)
110 113
111void __cpuinit xen_enable_sysenter(void) 114void __cpuinit xen_enable_sysenter(void)
112{ 115{
113 extern void xen_sysenter_target(void);
114 int ret; 116 int ret;
115 unsigned sysenter_feature; 117 unsigned sysenter_feature;
116 118
@@ -132,8 +134,6 @@ void __cpuinit xen_enable_syscall(void)
132{ 134{
133#ifdef CONFIG_X86_64 135#ifdef CONFIG_X86_64
134 int ret; 136 int ret;
135 extern void xen_syscall_target(void);
136 extern void xen_syscall32_target(void);
137 137
138 ret = register_callback(CALLBACKTYPE_syscall, xen_syscall_target); 138 ret = register_callback(CALLBACKTYPE_syscall, xen_syscall_target);
139 if (ret != 0) { 139 if (ret != 0) {
@@ -160,7 +160,8 @@ void __init xen_arch_setup(void)
160 HYPERVISOR_vm_assist(VMASST_CMD_enable, VMASST_TYPE_writable_pagetables); 160 HYPERVISOR_vm_assist(VMASST_CMD_enable, VMASST_TYPE_writable_pagetables);
161 161
162 if (!xen_feature(XENFEAT_auto_translated_physmap)) 162 if (!xen_feature(XENFEAT_auto_translated_physmap))
163 HYPERVISOR_vm_assist(VMASST_CMD_enable, VMASST_TYPE_pae_extended_cr3); 163 HYPERVISOR_vm_assist(VMASST_CMD_enable,
164 VMASST_TYPE_pae_extended_cr3);
164 165
165 if (register_callback(CALLBACKTYPE_event, xen_hypervisor_callback) || 166 if (register_callback(CALLBACKTYPE_event, xen_hypervisor_callback) ||
166 register_callback(CALLBACKTYPE_failsafe, xen_failsafe_callback)) 167 register_callback(CALLBACKTYPE_failsafe, xen_failsafe_callback))
diff --git a/arch/x86/xen/smp.c b/arch/x86/xen/smp.c
index d77da613b1d2..acd9b6705e02 100644
--- a/arch/x86/xen/smp.c
+++ b/arch/x86/xen/smp.c
@@ -362,7 +362,7 @@ static void xen_cpu_die(unsigned int cpu)
362 alternatives_smp_switch(0); 362 alternatives_smp_switch(0);
363} 363}
364 364
365static void xen_play_dead(void) 365static void __cpuinit xen_play_dead(void) /* used only with CPU_HOTPLUG */
366{ 366{
367 play_dead_common(); 367 play_dead_common();
368 HYPERVISOR_vcpu_op(VCPUOP_down, smp_processor_id(), NULL); 368 HYPERVISOR_vcpu_op(VCPUOP_down, smp_processor_id(), NULL);
diff --git a/arch/x86/xen/xen-ops.h b/arch/x86/xen/xen-ops.h
index d7422dc2a55c..9e1afae8461f 100644
--- a/arch/x86/xen/xen-ops.h
+++ b/arch/x86/xen/xen-ops.h
@@ -49,7 +49,7 @@ bool xen_vcpu_stolen(int vcpu);
49 49
50void xen_mark_init_mm_pinned(void); 50void xen_mark_init_mm_pinned(void);
51 51
52void __init xen_setup_vcpu_info_placement(void); 52void xen_setup_vcpu_info_placement(void);
53 53
54#ifdef CONFIG_SMP 54#ifdef CONFIG_SMP
55void xen_smp_init(void); 55void xen_smp_init(void);