aboutsummaryrefslogtreecommitdiffstats
path: root/arch/x86
diff options
context:
space:
mode:
Diffstat (limited to 'arch/x86')
-rw-r--r--arch/x86/Kconfig9
-rw-r--r--arch/x86/Makefile2
-rw-r--r--arch/x86/ia32/ia32entry.S22
-rw-r--r--arch/x86/include/asm/compat.h2
-rw-r--r--arch/x86/include/asm/cpufeature.h4
-rw-r--r--arch/x86/include/asm/hpet.h1
-rw-r--r--arch/x86/include/asm/hw_breakpoint.h2
-rw-r--r--arch/x86/include/asm/iomap.h4
-rw-r--r--arch/x86/include/asm/kvm_emulate.h7
-rw-r--r--arch/x86/include/asm/pci.h6
-rw-r--r--arch/x86/include/asm/pgtable_32.h1
-rw-r--r--arch/x86/include/asm/syscalls.h5
-rw-r--r--arch/x86/include/asm/trampoline.h5
-rw-r--r--arch/x86/include/asm/tsc.h2
-rw-r--r--arch/x86/kernel/apic/io_apic.c2
-rw-r--r--arch/x86/kernel/apic/x2apic_uv_x.c6
-rw-r--r--arch/x86/kernel/cpu/amd.c2
-rw-r--r--arch/x86/kernel/cpu/mcheck/mce_amd.c4
-rw-r--r--arch/x86/kernel/cpu/mcheck/therm_throt.c9
-rw-r--r--arch/x86/kernel/cpu/perf_event.c59
-rw-r--r--arch/x86/kernel/cpu/perf_event_intel.c96
-rw-r--r--arch/x86/kernel/cpu/perf_event_p4.c4
-rw-r--r--arch/x86/kernel/early-quirks.c18
-rw-r--r--arch/x86/kernel/head_32.S8
-rw-r--r--arch/x86/kernel/hpet.c31
-rw-r--r--arch/x86/kernel/hw_breakpoint.c40
-rw-r--r--arch/x86/kernel/i387.c1
-rw-r--r--arch/x86/kernel/kgdb.c2
-rw-r--r--arch/x86/kernel/kprobes.c25
-rw-r--r--arch/x86/kernel/process.c5
-rw-r--r--arch/x86/kernel/setup.c2
-rw-r--r--arch/x86/kernel/smpboot.c51
-rw-r--r--arch/x86/kernel/sys_i386_32.c4
-rw-r--r--arch/x86/kernel/trampoline.c17
-rw-r--r--arch/x86/kernel/tsc.c38
-rw-r--r--arch/x86/kvm/emulate.c9
-rw-r--r--arch/x86/kvm/i8254.c3
-rw-r--r--arch/x86/kvm/i8259.c3
-rw-r--r--arch/x86/kvm/irq.h2
-rw-r--r--arch/x86/kvm/x86.c4
-rw-r--r--arch/x86/lguest/boot.c13
-rw-r--r--arch/x86/mm/iomap_32.c6
-rw-r--r--arch/x86/oprofile/nmi_int.c26
-rw-r--r--arch/x86/power/cpu.c2
-rw-r--r--arch/x86/xen/platform-pci-unplug.c18
45 files changed, 401 insertions, 181 deletions
diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
index a84fc34c8f77..cea0cd9a316f 100644
--- a/arch/x86/Kconfig
+++ b/arch/x86/Kconfig
@@ -245,6 +245,11 @@ config ARCH_HWEIGHT_CFLAGS
245 245
246config KTIME_SCALAR 246config KTIME_SCALAR
247 def_bool X86_32 247 def_bool X86_32
248
249config ARCH_CPU_PROBE_RELEASE
250 def_bool y
251 depends on HOTPLUG_CPU
252
248source "init/Kconfig" 253source "init/Kconfig"
249source "kernel/Kconfig.freezer" 254source "kernel/Kconfig.freezer"
250 255
@@ -749,11 +754,11 @@ config IOMMU_API
749 def_bool (AMD_IOMMU || DMAR) 754 def_bool (AMD_IOMMU || DMAR)
750 755
751config MAXSMP 756config MAXSMP
752 bool "Configure Maximum number of SMP Processors and NUMA Nodes" 757 bool "Enable Maximum number of SMP Processors and NUMA Nodes"
753 depends on X86_64 && SMP && DEBUG_KERNEL && EXPERIMENTAL 758 depends on X86_64 && SMP && DEBUG_KERNEL && EXPERIMENTAL
754 select CPUMASK_OFFSTACK 759 select CPUMASK_OFFSTACK
755 ---help--- 760 ---help---
756 Configure maximum number of CPUS and NUMA Nodes for this architecture. 761 Enable maximum number of CPUS and NUMA Nodes for this architecture.
757 If unsure, say N. 762 If unsure, say N.
758 763
759config NR_CPUS 764config NR_CPUS
diff --git a/arch/x86/Makefile b/arch/x86/Makefile
index 8aa1b59b9074..e8c8881351b3 100644
--- a/arch/x86/Makefile
+++ b/arch/x86/Makefile
@@ -74,7 +74,7 @@ endif
74 74
75ifdef CONFIG_CC_STACKPROTECTOR 75ifdef CONFIG_CC_STACKPROTECTOR
76 cc_has_sp := $(srctree)/scripts/gcc-x86_$(BITS)-has-stack-protector.sh 76 cc_has_sp := $(srctree)/scripts/gcc-x86_$(BITS)-has-stack-protector.sh
77 ifeq ($(shell $(CONFIG_SHELL) $(cc_has_sp) $(CC) $(biarch)),y) 77 ifeq ($(shell $(CONFIG_SHELL) $(cc_has_sp) $(CC) $(KBUILD_CPPFLAGS) $(biarch)),y)
78 stackp-y := -fstack-protector 78 stackp-y := -fstack-protector
79 KBUILD_CFLAGS += $(stackp-y) 79 KBUILD_CFLAGS += $(stackp-y)
80 else 80 else
diff --git a/arch/x86/ia32/ia32entry.S b/arch/x86/ia32/ia32entry.S
index b86feabed69b..518bb99c3394 100644
--- a/arch/x86/ia32/ia32entry.S
+++ b/arch/x86/ia32/ia32entry.S
@@ -50,7 +50,12 @@
50 /* 50 /*
51 * Reload arg registers from stack in case ptrace changed them. 51 * Reload arg registers from stack in case ptrace changed them.
52 * We don't reload %eax because syscall_trace_enter() returned 52 * We don't reload %eax because syscall_trace_enter() returned
53 * the value it wants us to use in the table lookup. 53 * the %rax value we should see. Instead, we just truncate that
54 * value to 32 bits again as we did on entry from user mode.
55 * If it's a new value set by user_regset during entry tracing,
56 * this matches the normal truncation of the user-mode value.
57 * If it's -1 to make us punt the syscall, then (u32)-1 is still
58 * an appropriately invalid value.
54 */ 59 */
55 .macro LOAD_ARGS32 offset, _r9=0 60 .macro LOAD_ARGS32 offset, _r9=0
56 .if \_r9 61 .if \_r9
@@ -60,6 +65,7 @@
60 movl \offset+48(%rsp),%edx 65 movl \offset+48(%rsp),%edx
61 movl \offset+56(%rsp),%esi 66 movl \offset+56(%rsp),%esi
62 movl \offset+64(%rsp),%edi 67 movl \offset+64(%rsp),%edi
68 movl %eax,%eax /* zero extension */
63 .endm 69 .endm
64 70
65 .macro CFI_STARTPROC32 simple 71 .macro CFI_STARTPROC32 simple
@@ -153,7 +159,7 @@ ENTRY(ia32_sysenter_target)
153 testl $_TIF_WORK_SYSCALL_ENTRY,TI_flags(%r10) 159 testl $_TIF_WORK_SYSCALL_ENTRY,TI_flags(%r10)
154 CFI_REMEMBER_STATE 160 CFI_REMEMBER_STATE
155 jnz sysenter_tracesys 161 jnz sysenter_tracesys
156 cmpl $(IA32_NR_syscalls-1),%eax 162 cmpq $(IA32_NR_syscalls-1),%rax
157 ja ia32_badsys 163 ja ia32_badsys
158sysenter_do_call: 164sysenter_do_call:
159 IA32_ARG_FIXUP 165 IA32_ARG_FIXUP
@@ -195,7 +201,7 @@ sysexit_from_sys_call:
195 movl $AUDIT_ARCH_I386,%edi /* 1st arg: audit arch */ 201 movl $AUDIT_ARCH_I386,%edi /* 1st arg: audit arch */
196 call audit_syscall_entry 202 call audit_syscall_entry
197 movl RAX-ARGOFFSET(%rsp),%eax /* reload syscall number */ 203 movl RAX-ARGOFFSET(%rsp),%eax /* reload syscall number */
198 cmpl $(IA32_NR_syscalls-1),%eax 204 cmpq $(IA32_NR_syscalls-1),%rax
199 ja ia32_badsys 205 ja ia32_badsys
200 movl %ebx,%edi /* reload 1st syscall arg */ 206 movl %ebx,%edi /* reload 1st syscall arg */
201 movl RCX-ARGOFFSET(%rsp),%esi /* reload 2nd syscall arg */ 207 movl RCX-ARGOFFSET(%rsp),%esi /* reload 2nd syscall arg */
@@ -248,7 +254,7 @@ sysenter_tracesys:
248 call syscall_trace_enter 254 call syscall_trace_enter
249 LOAD_ARGS32 ARGOFFSET /* reload args from stack in case ptrace changed it */ 255 LOAD_ARGS32 ARGOFFSET /* reload args from stack in case ptrace changed it */
250 RESTORE_REST 256 RESTORE_REST
251 cmpl $(IA32_NR_syscalls-1),%eax 257 cmpq $(IA32_NR_syscalls-1),%rax
252 ja int_ret_from_sys_call /* sysenter_tracesys has set RAX(%rsp) */ 258 ja int_ret_from_sys_call /* sysenter_tracesys has set RAX(%rsp) */
253 jmp sysenter_do_call 259 jmp sysenter_do_call
254 CFI_ENDPROC 260 CFI_ENDPROC
@@ -314,7 +320,7 @@ ENTRY(ia32_cstar_target)
314 testl $_TIF_WORK_SYSCALL_ENTRY,TI_flags(%r10) 320 testl $_TIF_WORK_SYSCALL_ENTRY,TI_flags(%r10)
315 CFI_REMEMBER_STATE 321 CFI_REMEMBER_STATE
316 jnz cstar_tracesys 322 jnz cstar_tracesys
317 cmpl $IA32_NR_syscalls-1,%eax 323 cmpq $IA32_NR_syscalls-1,%rax
318 ja ia32_badsys 324 ja ia32_badsys
319cstar_do_call: 325cstar_do_call:
320 IA32_ARG_FIXUP 1 326 IA32_ARG_FIXUP 1
@@ -367,7 +373,7 @@ cstar_tracesys:
367 LOAD_ARGS32 ARGOFFSET, 1 /* reload args from stack in case ptrace changed it */ 373 LOAD_ARGS32 ARGOFFSET, 1 /* reload args from stack in case ptrace changed it */
368 RESTORE_REST 374 RESTORE_REST
369 xchgl %ebp,%r9d 375 xchgl %ebp,%r9d
370 cmpl $(IA32_NR_syscalls-1),%eax 376 cmpq $(IA32_NR_syscalls-1),%rax
371 ja int_ret_from_sys_call /* cstar_tracesys has set RAX(%rsp) */ 377 ja int_ret_from_sys_call /* cstar_tracesys has set RAX(%rsp) */
372 jmp cstar_do_call 378 jmp cstar_do_call
373END(ia32_cstar_target) 379END(ia32_cstar_target)
@@ -425,7 +431,7 @@ ENTRY(ia32_syscall)
425 orl $TS_COMPAT,TI_status(%r10) 431 orl $TS_COMPAT,TI_status(%r10)
426 testl $_TIF_WORK_SYSCALL_ENTRY,TI_flags(%r10) 432 testl $_TIF_WORK_SYSCALL_ENTRY,TI_flags(%r10)
427 jnz ia32_tracesys 433 jnz ia32_tracesys
428 cmpl $(IA32_NR_syscalls-1),%eax 434 cmpq $(IA32_NR_syscalls-1),%rax
429 ja ia32_badsys 435 ja ia32_badsys
430ia32_do_call: 436ia32_do_call:
431 IA32_ARG_FIXUP 437 IA32_ARG_FIXUP
@@ -444,7 +450,7 @@ ia32_tracesys:
444 call syscall_trace_enter 450 call syscall_trace_enter
445 LOAD_ARGS32 ARGOFFSET /* reload args from stack in case ptrace changed it */ 451 LOAD_ARGS32 ARGOFFSET /* reload args from stack in case ptrace changed it */
446 RESTORE_REST 452 RESTORE_REST
447 cmpl $(IA32_NR_syscalls-1),%eax 453 cmpq $(IA32_NR_syscalls-1),%rax
448 ja int_ret_from_sys_call /* ia32_tracesys has set RAX(%rsp) */ 454 ja int_ret_from_sys_call /* ia32_tracesys has set RAX(%rsp) */
449 jmp ia32_do_call 455 jmp ia32_do_call
450END(ia32_syscall) 456END(ia32_syscall)
diff --git a/arch/x86/include/asm/compat.h b/arch/x86/include/asm/compat.h
index 306160e58b48..1d9cd27c2920 100644
--- a/arch/x86/include/asm/compat.h
+++ b/arch/x86/include/asm/compat.h
@@ -205,7 +205,7 @@ static inline compat_uptr_t ptr_to_compat(void __user *uptr)
205 return (u32)(unsigned long)uptr; 205 return (u32)(unsigned long)uptr;
206} 206}
207 207
208static inline void __user *compat_alloc_user_space(long len) 208static inline void __user *arch_compat_alloc_user_space(long len)
209{ 209{
210 struct pt_regs *regs = task_pt_regs(current); 210 struct pt_regs *regs = task_pt_regs(current);
211 return (void __user *)regs->sp - len; 211 return (void __user *)regs->sp - len;
diff --git a/arch/x86/include/asm/cpufeature.h b/arch/x86/include/asm/cpufeature.h
index 781a50b29a49..c6fbb7b430d1 100644
--- a/arch/x86/include/asm/cpufeature.h
+++ b/arch/x86/include/asm/cpufeature.h
@@ -296,6 +296,7 @@ extern const char * const x86_power_flags[32];
296 296
297#endif /* CONFIG_X86_64 */ 297#endif /* CONFIG_X86_64 */
298 298
299#if __GNUC__ >= 4
299/* 300/*
300 * Static testing of CPU features. Used the same as boot_cpu_has(). 301 * Static testing of CPU features. Used the same as boot_cpu_has().
301 * These are only valid after alternatives have run, but will statically 302 * These are only valid after alternatives have run, but will statically
@@ -304,7 +305,7 @@ extern const char * const x86_power_flags[32];
304 */ 305 */
305static __always_inline __pure bool __static_cpu_has(u16 bit) 306static __always_inline __pure bool __static_cpu_has(u16 bit)
306{ 307{
307#if __GNUC__ > 4 || (__GNUC__ == 4 && __GNUC_MINOR__ >= 5) 308#if __GNUC__ > 4 || __GNUC_MINOR__ >= 5
308 asm goto("1: jmp %l[t_no]\n" 309 asm goto("1: jmp %l[t_no]\n"
309 "2:\n" 310 "2:\n"
310 ".section .altinstructions,\"a\"\n" 311 ".section .altinstructions,\"a\"\n"
@@ -345,7 +346,6 @@ static __always_inline __pure bool __static_cpu_has(u16 bit)
345#endif 346#endif
346} 347}
347 348
348#if __GNUC__ >= 4
349#define static_cpu_has(bit) \ 349#define static_cpu_has(bit) \
350( \ 350( \
351 __builtin_constant_p(boot_cpu_has(bit)) ? \ 351 __builtin_constant_p(boot_cpu_has(bit)) ? \
diff --git a/arch/x86/include/asm/hpet.h b/arch/x86/include/asm/hpet.h
index 004e6e25e913..1d5c08a1bdfd 100644
--- a/arch/x86/include/asm/hpet.h
+++ b/arch/x86/include/asm/hpet.h
@@ -68,7 +68,6 @@ extern unsigned long force_hpet_address;
68extern u8 hpet_blockid; 68extern u8 hpet_blockid;
69extern int hpet_force_user; 69extern int hpet_force_user;
70extern u8 hpet_msi_disable; 70extern u8 hpet_msi_disable;
71extern u8 hpet_readback_cmp;
72extern int is_hpet_enabled(void); 71extern int is_hpet_enabled(void);
73extern int hpet_enable(void); 72extern int hpet_enable(void);
74extern void hpet_disable(void); 73extern void hpet_disable(void);
diff --git a/arch/x86/include/asm/hw_breakpoint.h b/arch/x86/include/asm/hw_breakpoint.h
index 528a11e8d3e3..824ca07860d0 100644
--- a/arch/x86/include/asm/hw_breakpoint.h
+++ b/arch/x86/include/asm/hw_breakpoint.h
@@ -20,7 +20,7 @@ struct arch_hw_breakpoint {
20#include <linux/list.h> 20#include <linux/list.h>
21 21
22/* Available HW breakpoint length encodings */ 22/* Available HW breakpoint length encodings */
23#define X86_BREAKPOINT_LEN_X 0x00 23#define X86_BREAKPOINT_LEN_X 0x40
24#define X86_BREAKPOINT_LEN_1 0x40 24#define X86_BREAKPOINT_LEN_1 0x40
25#define X86_BREAKPOINT_LEN_2 0x44 25#define X86_BREAKPOINT_LEN_2 0x44
26#define X86_BREAKPOINT_LEN_4 0x4c 26#define X86_BREAKPOINT_LEN_4 0x4c
diff --git a/arch/x86/include/asm/iomap.h b/arch/x86/include/asm/iomap.h
index f35eb45d6576..c4191b3b7056 100644
--- a/arch/x86/include/asm/iomap.h
+++ b/arch/x86/include/asm/iomap.h
@@ -26,11 +26,11 @@
26#include <asm/pgtable.h> 26#include <asm/pgtable.h>
27#include <asm/tlbflush.h> 27#include <asm/tlbflush.h>
28 28
29void * 29void __iomem *
30iomap_atomic_prot_pfn(unsigned long pfn, enum km_type type, pgprot_t prot); 30iomap_atomic_prot_pfn(unsigned long pfn, enum km_type type, pgprot_t prot);
31 31
32void 32void
33iounmap_atomic(void *kvaddr, enum km_type type); 33iounmap_atomic(void __iomem *kvaddr, enum km_type type);
34 34
35int 35int
36iomap_create_wc(resource_size_t base, unsigned long size, pgprot_t *prot); 36iomap_create_wc(resource_size_t base, unsigned long size, pgprot_t *prot);
diff --git a/arch/x86/include/asm/kvm_emulate.h b/arch/x86/include/asm/kvm_emulate.h
index 51cfd730ac5d..1f99ecfc48e1 100644
--- a/arch/x86/include/asm/kvm_emulate.h
+++ b/arch/x86/include/asm/kvm_emulate.h
@@ -152,9 +152,14 @@ struct x86_emulate_ops {
152struct operand { 152struct operand {
153 enum { OP_REG, OP_MEM, OP_IMM, OP_NONE } type; 153 enum { OP_REG, OP_MEM, OP_IMM, OP_NONE } type;
154 unsigned int bytes; 154 unsigned int bytes;
155 unsigned long orig_val, *ptr; 155 union {
156 unsigned long orig_val;
157 u64 orig_val64;
158 };
159 unsigned long *ptr;
156 union { 160 union {
157 unsigned long val; 161 unsigned long val;
162 u64 val64;
158 char valptr[sizeof(unsigned long) + 2]; 163 char valptr[sizeof(unsigned long) + 2];
159 }; 164 };
160}; 165};
diff --git a/arch/x86/include/asm/pci.h b/arch/x86/include/asm/pci.h
index 404a880ea325..d395540ff894 100644
--- a/arch/x86/include/asm/pci.h
+++ b/arch/x86/include/asm/pci.h
@@ -27,6 +27,9 @@ extern struct pci_bus *pci_scan_bus_on_node(int busno, struct pci_ops *ops,
27 int node); 27 int node);
28extern struct pci_bus *pci_scan_bus_with_sysdata(int busno); 28extern struct pci_bus *pci_scan_bus_with_sysdata(int busno);
29 29
30#ifdef CONFIG_PCI
31
32#ifdef CONFIG_PCI_DOMAINS
30static inline int pci_domain_nr(struct pci_bus *bus) 33static inline int pci_domain_nr(struct pci_bus *bus)
31{ 34{
32 struct pci_sysdata *sd = bus->sysdata; 35 struct pci_sysdata *sd = bus->sysdata;
@@ -37,13 +40,12 @@ static inline int pci_proc_domain(struct pci_bus *bus)
37{ 40{
38 return pci_domain_nr(bus); 41 return pci_domain_nr(bus);
39} 42}
40 43#endif
41 44
42/* Can be used to override the logic in pci_scan_bus for skipping 45/* Can be used to override the logic in pci_scan_bus for skipping
43 already-configured bus numbers - to be used for buggy BIOSes 46 already-configured bus numbers - to be used for buggy BIOSes
44 or architectures with incomplete PCI setup by the loader */ 47 or architectures with incomplete PCI setup by the loader */
45 48
46#ifdef CONFIG_PCI
47extern unsigned int pcibios_assign_all_busses(void); 49extern unsigned int pcibios_assign_all_busses(void);
48extern int pci_legacy_init(void); 50extern int pci_legacy_init(void);
49# ifdef CONFIG_ACPI 51# ifdef CONFIG_ACPI
diff --git a/arch/x86/include/asm/pgtable_32.h b/arch/x86/include/asm/pgtable_32.h
index 2984a25ff383..f686f49e8b7b 100644
--- a/arch/x86/include/asm/pgtable_32.h
+++ b/arch/x86/include/asm/pgtable_32.h
@@ -26,6 +26,7 @@ struct mm_struct;
26struct vm_area_struct; 26struct vm_area_struct;
27 27
28extern pgd_t swapper_pg_dir[1024]; 28extern pgd_t swapper_pg_dir[1024];
29extern pgd_t trampoline_pg_dir[1024];
29 30
30static inline void pgtable_cache_init(void) { } 31static inline void pgtable_cache_init(void) { }
31static inline void check_pgt_cache(void) { } 32static inline void check_pgt_cache(void) { }
diff --git a/arch/x86/include/asm/syscalls.h b/arch/x86/include/asm/syscalls.h
index feb2ff9bfc2d..f1d8b441fc77 100644
--- a/arch/x86/include/asm/syscalls.h
+++ b/arch/x86/include/asm/syscalls.h
@@ -23,8 +23,9 @@ long sys_iopl(unsigned int, struct pt_regs *);
23/* kernel/process.c */ 23/* kernel/process.c */
24int sys_fork(struct pt_regs *); 24int sys_fork(struct pt_regs *);
25int sys_vfork(struct pt_regs *); 25int sys_vfork(struct pt_regs *);
26long sys_execve(const char __user *, char __user * __user *, 26long sys_execve(const char __user *,
27 char __user * __user *, struct pt_regs *); 27 const char __user *const __user *,
28 const char __user *const __user *, struct pt_regs *);
28long sys_clone(unsigned long, unsigned long, void __user *, 29long sys_clone(unsigned long, unsigned long, void __user *,
29 void __user *, struct pt_regs *); 30 void __user *, struct pt_regs *);
30 31
diff --git a/arch/x86/include/asm/trampoline.h b/arch/x86/include/asm/trampoline.h
index cb507bb05d79..4dde797c0578 100644
--- a/arch/x86/include/asm/trampoline.h
+++ b/arch/x86/include/asm/trampoline.h
@@ -13,14 +13,17 @@ extern unsigned char *trampoline_base;
13 13
14extern unsigned long init_rsp; 14extern unsigned long init_rsp;
15extern unsigned long initial_code; 15extern unsigned long initial_code;
16extern unsigned long initial_page_table;
16extern unsigned long initial_gs; 17extern unsigned long initial_gs;
17 18
18#define TRAMPOLINE_SIZE roundup(trampoline_end - trampoline_data, PAGE_SIZE) 19#define TRAMPOLINE_SIZE roundup(trampoline_end - trampoline_data, PAGE_SIZE)
19 20
20extern unsigned long setup_trampoline(void); 21extern unsigned long setup_trampoline(void);
22extern void __init setup_trampoline_page_table(void);
21extern void __init reserve_trampoline_memory(void); 23extern void __init reserve_trampoline_memory(void);
22#else 24#else
23static inline void reserve_trampoline_memory(void) {}; 25static inline void setup_trampoline_page_table(void) {}
26static inline void reserve_trampoline_memory(void) {}
24#endif /* CONFIG_X86_TRAMPOLINE */ 27#endif /* CONFIG_X86_TRAMPOLINE */
25 28
26#endif /* __ASSEMBLY__ */ 29#endif /* __ASSEMBLY__ */
diff --git a/arch/x86/include/asm/tsc.h b/arch/x86/include/asm/tsc.h
index c0427295e8f5..1ca132fc0d03 100644
--- a/arch/x86/include/asm/tsc.h
+++ b/arch/x86/include/asm/tsc.h
@@ -59,5 +59,7 @@ extern void check_tsc_sync_source(int cpu);
59extern void check_tsc_sync_target(void); 59extern void check_tsc_sync_target(void);
60 60
61extern int notsc_setup(char *); 61extern int notsc_setup(char *);
62extern void save_sched_clock_state(void);
63extern void restore_sched_clock_state(void);
62 64
63#endif /* _ASM_X86_TSC_H */ 65#endif /* _ASM_X86_TSC_H */
diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c
index 4dc0084ec1b1..f1efebaf5510 100644
--- a/arch/x86/kernel/apic/io_apic.c
+++ b/arch/x86/kernel/apic/io_apic.c
@@ -1728,6 +1728,8 @@ __apicdebuginit(void) print_IO_APIC(void)
1728 struct irq_pin_list *entry; 1728 struct irq_pin_list *entry;
1729 1729
1730 cfg = desc->chip_data; 1730 cfg = desc->chip_data;
1731 if (!cfg)
1732 continue;
1731 entry = cfg->irq_2_pin; 1733 entry = cfg->irq_2_pin;
1732 if (!entry) 1734 if (!entry)
1733 continue; 1735 continue;
diff --git a/arch/x86/kernel/apic/x2apic_uv_x.c b/arch/x86/kernel/apic/x2apic_uv_x.c
index 7b598b84c902..f744f54cb248 100644
--- a/arch/x86/kernel/apic/x2apic_uv_x.c
+++ b/arch/x86/kernel/apic/x2apic_uv_x.c
@@ -698,9 +698,11 @@ void __init uv_system_init(void)
698 for (j = 0; j < 64; j++) { 698 for (j = 0; j < 64; j++) {
699 if (!test_bit(j, &present)) 699 if (!test_bit(j, &present))
700 continue; 700 continue;
701 uv_blade_info[blade].pnode = (i * 64 + j); 701 pnode = (i * 64 + j);
702 uv_blade_info[blade].pnode = pnode;
702 uv_blade_info[blade].nr_possible_cpus = 0; 703 uv_blade_info[blade].nr_possible_cpus = 0;
703 uv_blade_info[blade].nr_online_cpus = 0; 704 uv_blade_info[blade].nr_online_cpus = 0;
705 max_pnode = max(pnode, max_pnode);
704 blade++; 706 blade++;
705 } 707 }
706 } 708 }
@@ -738,7 +740,6 @@ void __init uv_system_init(void)
738 uv_cpu_hub_info(cpu)->scir.offset = uv_scir_offset(apicid); 740 uv_cpu_hub_info(cpu)->scir.offset = uv_scir_offset(apicid);
739 uv_node_to_blade[nid] = blade; 741 uv_node_to_blade[nid] = blade;
740 uv_cpu_to_blade[cpu] = blade; 742 uv_cpu_to_blade[cpu] = blade;
741 max_pnode = max(pnode, max_pnode);
742 } 743 }
743 744
744 /* Add blade/pnode info for nodes without cpus */ 745 /* Add blade/pnode info for nodes without cpus */
@@ -750,7 +751,6 @@ void __init uv_system_init(void)
750 pnode = (paddr >> m_val) & pnode_mask; 751 pnode = (paddr >> m_val) & pnode_mask;
751 blade = boot_pnode_to_blade(pnode); 752 blade = boot_pnode_to_blade(pnode);
752 uv_node_to_blade[nid] = blade; 753 uv_node_to_blade[nid] = blade;
753 max_pnode = max(pnode, max_pnode);
754 } 754 }
755 755
756 map_gru_high(max_pnode); 756 map_gru_high(max_pnode);
diff --git a/arch/x86/kernel/cpu/amd.c b/arch/x86/kernel/cpu/amd.c
index 60a57b13082d..ba5f62f45f01 100644
--- a/arch/x86/kernel/cpu/amd.c
+++ b/arch/x86/kernel/cpu/amd.c
@@ -669,7 +669,7 @@ bool cpu_has_amd_erratum(const int *erratum)
669 } 669 }
670 670
671 /* OSVW unavailable or ID unknown, match family-model-stepping range */ 671 /* OSVW unavailable or ID unknown, match family-model-stepping range */
672 ms = (cpu->x86_model << 8) | cpu->x86_mask; 672 ms = (cpu->x86_model << 4) | cpu->x86_mask;
673 while ((range = *erratum++)) 673 while ((range = *erratum++))
674 if ((cpu->x86 == AMD_MODEL_RANGE_FAMILY(range)) && 674 if ((cpu->x86 == AMD_MODEL_RANGE_FAMILY(range)) &&
675 (ms >= AMD_MODEL_RANGE_START(range)) && 675 (ms >= AMD_MODEL_RANGE_START(range)) &&
diff --git a/arch/x86/kernel/cpu/mcheck/mce_amd.c b/arch/x86/kernel/cpu/mcheck/mce_amd.c
index 224392d8fe8c..5e975298fa81 100644
--- a/arch/x86/kernel/cpu/mcheck/mce_amd.c
+++ b/arch/x86/kernel/cpu/mcheck/mce_amd.c
@@ -530,7 +530,7 @@ static __cpuinit int threshold_create_bank(unsigned int cpu, unsigned int bank)
530 err = -ENOMEM; 530 err = -ENOMEM;
531 goto out; 531 goto out;
532 } 532 }
533 if (!alloc_cpumask_var(&b->cpus, GFP_KERNEL)) { 533 if (!zalloc_cpumask_var(&b->cpus, GFP_KERNEL)) {
534 kfree(b); 534 kfree(b);
535 err = -ENOMEM; 535 err = -ENOMEM;
536 goto out; 536 goto out;
@@ -543,7 +543,7 @@ static __cpuinit int threshold_create_bank(unsigned int cpu, unsigned int bank)
543#ifndef CONFIG_SMP 543#ifndef CONFIG_SMP
544 cpumask_setall(b->cpus); 544 cpumask_setall(b->cpus);
545#else 545#else
546 cpumask_copy(b->cpus, c->llc_shared_map); 546 cpumask_set_cpu(cpu, b->cpus);
547#endif 547#endif
548 548
549 per_cpu(threshold_banks, cpu)[bank] = b; 549 per_cpu(threshold_banks, cpu)[bank] = b;
diff --git a/arch/x86/kernel/cpu/mcheck/therm_throt.c b/arch/x86/kernel/cpu/mcheck/therm_throt.c
index c2a8b26d4fea..d9368eeda309 100644
--- a/arch/x86/kernel/cpu/mcheck/therm_throt.c
+++ b/arch/x86/kernel/cpu/mcheck/therm_throt.c
@@ -202,10 +202,11 @@ static int therm_throt_process(bool new_event, int event, int level)
202 202
203#ifdef CONFIG_SYSFS 203#ifdef CONFIG_SYSFS
204/* Add/Remove thermal_throttle interface for CPU device: */ 204/* Add/Remove thermal_throttle interface for CPU device: */
205static __cpuinit int thermal_throttle_add_dev(struct sys_device *sys_dev) 205static __cpuinit int thermal_throttle_add_dev(struct sys_device *sys_dev,
206 unsigned int cpu)
206{ 207{
207 int err; 208 int err;
208 struct cpuinfo_x86 *c = &cpu_data(smp_processor_id()); 209 struct cpuinfo_x86 *c = &cpu_data(cpu);
209 210
210 err = sysfs_create_group(&sys_dev->kobj, &thermal_attr_group); 211 err = sysfs_create_group(&sys_dev->kobj, &thermal_attr_group);
211 if (err) 212 if (err)
@@ -251,7 +252,7 @@ thermal_throttle_cpu_callback(struct notifier_block *nfb,
251 case CPU_UP_PREPARE: 252 case CPU_UP_PREPARE:
252 case CPU_UP_PREPARE_FROZEN: 253 case CPU_UP_PREPARE_FROZEN:
253 mutex_lock(&therm_cpu_lock); 254 mutex_lock(&therm_cpu_lock);
254 err = thermal_throttle_add_dev(sys_dev); 255 err = thermal_throttle_add_dev(sys_dev, cpu);
255 mutex_unlock(&therm_cpu_lock); 256 mutex_unlock(&therm_cpu_lock);
256 WARN_ON(err); 257 WARN_ON(err);
257 break; 258 break;
@@ -287,7 +288,7 @@ static __init int thermal_throttle_init_device(void)
287#endif 288#endif
288 /* connect live CPUs to sysfs */ 289 /* connect live CPUs to sysfs */
289 for_each_online_cpu(cpu) { 290 for_each_online_cpu(cpu) {
290 err = thermal_throttle_add_dev(get_cpu_sysdev(cpu)); 291 err = thermal_throttle_add_dev(get_cpu_sysdev(cpu), cpu);
291 WARN_ON(err); 292 WARN_ON(err);
292 } 293 }
293#ifdef CONFIG_HOTPLUG_CPU 294#ifdef CONFIG_HOTPLUG_CPU
diff --git a/arch/x86/kernel/cpu/perf_event.c b/arch/x86/kernel/cpu/perf_event.c
index f2da20fda02d..3efdf2870a35 100644
--- a/arch/x86/kernel/cpu/perf_event.c
+++ b/arch/x86/kernel/cpu/perf_event.c
@@ -1154,7 +1154,7 @@ static int x86_pmu_handle_irq(struct pt_regs *regs)
1154 /* 1154 /*
1155 * event overflow 1155 * event overflow
1156 */ 1156 */
1157 handled = 1; 1157 handled++;
1158 data.period = event->hw.last_period; 1158 data.period = event->hw.last_period;
1159 1159
1160 if (!x86_perf_event_set_period(event)) 1160 if (!x86_perf_event_set_period(event))
@@ -1200,12 +1200,20 @@ void perf_events_lapic_init(void)
1200 apic_write(APIC_LVTPC, APIC_DM_NMI); 1200 apic_write(APIC_LVTPC, APIC_DM_NMI);
1201} 1201}
1202 1202
1203struct pmu_nmi_state {
1204 unsigned int marked;
1205 int handled;
1206};
1207
1208static DEFINE_PER_CPU(struct pmu_nmi_state, pmu_nmi);
1209
1203static int __kprobes 1210static int __kprobes
1204perf_event_nmi_handler(struct notifier_block *self, 1211perf_event_nmi_handler(struct notifier_block *self,
1205 unsigned long cmd, void *__args) 1212 unsigned long cmd, void *__args)
1206{ 1213{
1207 struct die_args *args = __args; 1214 struct die_args *args = __args;
1208 struct pt_regs *regs; 1215 unsigned int this_nmi;
1216 int handled;
1209 1217
1210 if (!atomic_read(&active_events)) 1218 if (!atomic_read(&active_events))
1211 return NOTIFY_DONE; 1219 return NOTIFY_DONE;
@@ -1214,22 +1222,47 @@ perf_event_nmi_handler(struct notifier_block *self,
1214 case DIE_NMI: 1222 case DIE_NMI:
1215 case DIE_NMI_IPI: 1223 case DIE_NMI_IPI:
1216 break; 1224 break;
1217 1225 case DIE_NMIUNKNOWN:
1226 this_nmi = percpu_read(irq_stat.__nmi_count);
1227 if (this_nmi != __get_cpu_var(pmu_nmi).marked)
1228 /* let the kernel handle the unknown nmi */
1229 return NOTIFY_DONE;
1230 /*
1231 * This one is a PMU back-to-back nmi. Two events
1232 * trigger 'simultaneously' raising two back-to-back
1233 * NMIs. If the first NMI handles both, the latter
1234 * will be empty and daze the CPU. So, we drop it to
1235 * avoid false-positive 'unknown nmi' messages.
1236 */
1237 return NOTIFY_STOP;
1218 default: 1238 default:
1219 return NOTIFY_DONE; 1239 return NOTIFY_DONE;
1220 } 1240 }
1221 1241
1222 regs = args->regs;
1223
1224 apic_write(APIC_LVTPC, APIC_DM_NMI); 1242 apic_write(APIC_LVTPC, APIC_DM_NMI);
1225 /* 1243
1226 * Can't rely on the handled return value to say it was our NMI, two 1244 handled = x86_pmu.handle_irq(args->regs);
1227 * events could trigger 'simultaneously' raising two back-to-back NMIs. 1245 if (!handled)
1228 * 1246 return NOTIFY_DONE;
1229 * If the first NMI handles both, the latter will be empty and daze 1247
1230 * the CPU. 1248 this_nmi = percpu_read(irq_stat.__nmi_count);
1231 */ 1249 if ((handled > 1) ||
1232 x86_pmu.handle_irq(regs); 1250 /* the next nmi could be a back-to-back nmi */
1251 ((__get_cpu_var(pmu_nmi).marked == this_nmi) &&
1252 (__get_cpu_var(pmu_nmi).handled > 1))) {
1253 /*
1254 * We could have two subsequent back-to-back nmis: The
1255 * first handles more than one counter, the 2nd
1256 * handles only one counter and the 3rd handles no
1257 * counter.
1258 *
1259 * This is the 2nd nmi because the previous was
1260 * handling more than one counter. We will mark the
1261 * next (3rd) and then drop it if unhandled.
1262 */
1263 __get_cpu_var(pmu_nmi).marked = this_nmi + 1;
1264 __get_cpu_var(pmu_nmi).handled = handled;
1265 }
1233 1266
1234 return NOTIFY_STOP; 1267 return NOTIFY_STOP;
1235} 1268}
diff --git a/arch/x86/kernel/cpu/perf_event_intel.c b/arch/x86/kernel/cpu/perf_event_intel.c
index 214ac860ebe0..ee05c90012d2 100644
--- a/arch/x86/kernel/cpu/perf_event_intel.c
+++ b/arch/x86/kernel/cpu/perf_event_intel.c
@@ -491,33 +491,78 @@ static void intel_pmu_enable_all(int added)
491 * Intel Errata AAP53 (model 30) 491 * Intel Errata AAP53 (model 30)
492 * Intel Errata BD53 (model 44) 492 * Intel Errata BD53 (model 44)
493 * 493 *
494 * These chips need to be 'reset' when adding counters by programming 494 * The official story:
495 * the magic three (non counting) events 0x4300D2, 0x4300B1 and 0x4300B5 495 * These chips need to be 'reset' when adding counters by programming the
496 * either in sequence on the same PMC or on different PMCs. 496 * magic three (non-counting) events 0x4300B5, 0x4300D2, and 0x4300B1 either
497 * in sequence on the same PMC or on different PMCs.
498 *
499 * In practise it appears some of these events do in fact count, and
500 * we need to programm all 4 events.
497 */ 501 */
498static void intel_pmu_nhm_enable_all(int added) 502static void intel_pmu_nhm_workaround(void)
499{ 503{
500 if (added) { 504 struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
501 struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); 505 static const unsigned long nhm_magic[4] = {
502 int i; 506 0x4300B5,
507 0x4300D2,
508 0x4300B1,
509 0x4300B1
510 };
511 struct perf_event *event;
512 int i;
513
514 /*
515 * The Errata requires below steps:
516 * 1) Clear MSR_IA32_PEBS_ENABLE and MSR_CORE_PERF_GLOBAL_CTRL;
517 * 2) Configure 4 PERFEVTSELx with the magic events and clear
518 * the corresponding PMCx;
519 * 3) set bit0~bit3 of MSR_CORE_PERF_GLOBAL_CTRL;
520 * 4) Clear MSR_CORE_PERF_GLOBAL_CTRL;
521 * 5) Clear 4 pairs of ERFEVTSELx and PMCx;
522 */
503 523
504 wrmsrl(MSR_ARCH_PERFMON_EVENTSEL0 + 0, 0x4300D2); 524 /*
505 wrmsrl(MSR_ARCH_PERFMON_EVENTSEL0 + 1, 0x4300B1); 525 * The real steps we choose are a little different from above.
506 wrmsrl(MSR_ARCH_PERFMON_EVENTSEL0 + 2, 0x4300B5); 526 * A) To reduce MSR operations, we don't run step 1) as they
527 * are already cleared before this function is called;
528 * B) Call x86_perf_event_update to save PMCx before configuring
529 * PERFEVTSELx with magic number;
530 * C) With step 5), we do clear only when the PERFEVTSELx is
531 * not used currently.
532 * D) Call x86_perf_event_set_period to restore PMCx;
533 */
534
535 /* We always operate 4 pairs of PERF Counters */
536 for (i = 0; i < 4; i++) {
537 event = cpuc->events[i];
538 if (event)
539 x86_perf_event_update(event);
540 }
507 541
508 wrmsrl(MSR_CORE_PERF_GLOBAL_CTRL, 0x3); 542 for (i = 0; i < 4; i++) {
509 wrmsrl(MSR_CORE_PERF_GLOBAL_CTRL, 0x0); 543 wrmsrl(MSR_ARCH_PERFMON_EVENTSEL0 + i, nhm_magic[i]);
544 wrmsrl(MSR_ARCH_PERFMON_PERFCTR0 + i, 0x0);
545 }
510 546
511 for (i = 0; i < 3; i++) { 547 wrmsrl(MSR_CORE_PERF_GLOBAL_CTRL, 0xf);
512 struct perf_event *event = cpuc->events[i]; 548 wrmsrl(MSR_CORE_PERF_GLOBAL_CTRL, 0x0);
513 549
514 if (!event) 550 for (i = 0; i < 4; i++) {
515 continue; 551 event = cpuc->events[i];
516 552
553 if (event) {
554 x86_perf_event_set_period(event);
517 __x86_pmu_enable_event(&event->hw, 555 __x86_pmu_enable_event(&event->hw,
518 ARCH_PERFMON_EVENTSEL_ENABLE); 556 ARCH_PERFMON_EVENTSEL_ENABLE);
519 } 557 } else
558 wrmsrl(MSR_ARCH_PERFMON_EVENTSEL0 + i, 0x0);
520 } 559 }
560}
561
562static void intel_pmu_nhm_enable_all(int added)
563{
564 if (added)
565 intel_pmu_nhm_workaround();
521 intel_pmu_enable_all(added); 566 intel_pmu_enable_all(added);
522} 567}
523 568
@@ -667,7 +712,8 @@ static int intel_pmu_handle_irq(struct pt_regs *regs)
667 struct perf_sample_data data; 712 struct perf_sample_data data;
668 struct cpu_hw_events *cpuc; 713 struct cpu_hw_events *cpuc;
669 int bit, loops; 714 int bit, loops;
670 u64 ack, status; 715 u64 status;
716 int handled = 0;
671 717
672 perf_sample_data_init(&data, 0); 718 perf_sample_data_init(&data, 0);
673 719
@@ -683,6 +729,7 @@ static int intel_pmu_handle_irq(struct pt_regs *regs)
683 729
684 loops = 0; 730 loops = 0;
685again: 731again:
732 intel_pmu_ack_status(status);
686 if (++loops > 100) { 733 if (++loops > 100) {
687 WARN_ONCE(1, "perfevents: irq loop stuck!\n"); 734 WARN_ONCE(1, "perfevents: irq loop stuck!\n");
688 perf_event_print_debug(); 735 perf_event_print_debug();
@@ -691,19 +738,22 @@ again:
691 } 738 }
692 739
693 inc_irq_stat(apic_perf_irqs); 740 inc_irq_stat(apic_perf_irqs);
694 ack = status;
695 741
696 intel_pmu_lbr_read(); 742 intel_pmu_lbr_read();
697 743
698 /* 744 /*
699 * PEBS overflow sets bit 62 in the global status register 745 * PEBS overflow sets bit 62 in the global status register
700 */ 746 */
701 if (__test_and_clear_bit(62, (unsigned long *)&status)) 747 if (__test_and_clear_bit(62, (unsigned long *)&status)) {
748 handled++;
702 x86_pmu.drain_pebs(regs); 749 x86_pmu.drain_pebs(regs);
750 }
703 751
704 for_each_set_bit(bit, (unsigned long *)&status, X86_PMC_IDX_MAX) { 752 for_each_set_bit(bit, (unsigned long *)&status, X86_PMC_IDX_MAX) {
705 struct perf_event *event = cpuc->events[bit]; 753 struct perf_event *event = cpuc->events[bit];
706 754
755 handled++;
756
707 if (!test_bit(bit, cpuc->active_mask)) 757 if (!test_bit(bit, cpuc->active_mask))
708 continue; 758 continue;
709 759
@@ -716,8 +766,6 @@ again:
716 x86_pmu_stop(event); 766 x86_pmu_stop(event);
717 } 767 }
718 768
719 intel_pmu_ack_status(ack);
720
721 /* 769 /*
722 * Repeat if there is more work to be done: 770 * Repeat if there is more work to be done:
723 */ 771 */
@@ -727,7 +775,7 @@ again:
727 775
728done: 776done:
729 intel_pmu_enable_all(0); 777 intel_pmu_enable_all(0);
730 return 1; 778 return handled;
731} 779}
732 780
733static struct event_constraint * 781static struct event_constraint *
diff --git a/arch/x86/kernel/cpu/perf_event_p4.c b/arch/x86/kernel/cpu/perf_event_p4.c
index febb12cea795..b560db3305be 100644
--- a/arch/x86/kernel/cpu/perf_event_p4.c
+++ b/arch/x86/kernel/cpu/perf_event_p4.c
@@ -497,6 +497,8 @@ static int p4_hw_config(struct perf_event *event)
497 event->hw.config |= event->attr.config & 497 event->hw.config |= event->attr.config &
498 (p4_config_pack_escr(P4_ESCR_MASK_HT) | 498 (p4_config_pack_escr(P4_ESCR_MASK_HT) |
499 p4_config_pack_cccr(P4_CCCR_MASK_HT | P4_CCCR_RESERVED)); 499 p4_config_pack_cccr(P4_CCCR_MASK_HT | P4_CCCR_RESERVED));
500
501 event->hw.config &= ~P4_CCCR_FORCE_OVF;
500 } 502 }
501 503
502 rc = x86_setup_perfctr(event); 504 rc = x86_setup_perfctr(event);
@@ -690,7 +692,7 @@ static int p4_pmu_handle_irq(struct pt_regs *regs)
690 inc_irq_stat(apic_perf_irqs); 692 inc_irq_stat(apic_perf_irqs);
691 } 693 }
692 694
693 return handled > 0; 695 return handled;
694} 696}
695 697
696/* 698/*
diff --git a/arch/x86/kernel/early-quirks.c b/arch/x86/kernel/early-quirks.c
index e5cc7e82e60d..ebdb85cf2686 100644
--- a/arch/x86/kernel/early-quirks.c
+++ b/arch/x86/kernel/early-quirks.c
@@ -18,7 +18,6 @@
18#include <asm/apic.h> 18#include <asm/apic.h>
19#include <asm/iommu.h> 19#include <asm/iommu.h>
20#include <asm/gart.h> 20#include <asm/gart.h>
21#include <asm/hpet.h>
22 21
23static void __init fix_hypertransport_config(int num, int slot, int func) 22static void __init fix_hypertransport_config(int num, int slot, int func)
24{ 23{
@@ -192,21 +191,6 @@ static void __init ati_bugs_contd(int num, int slot, int func)
192} 191}
193#endif 192#endif
194 193
195/*
196 * Force the read back of the CMP register in hpet_next_event()
197 * to work around the problem that the CMP register write seems to be
198 * delayed. See hpet_next_event() for details.
199 *
200 * We do this on all SMBUS incarnations for now until we have more
201 * information about the affected chipsets.
202 */
203static void __init ati_hpet_bugs(int num, int slot, int func)
204{
205#ifdef CONFIG_HPET_TIMER
206 hpet_readback_cmp = 1;
207#endif
208}
209
210#define QFLAG_APPLY_ONCE 0x1 194#define QFLAG_APPLY_ONCE 0x1
211#define QFLAG_APPLIED 0x2 195#define QFLAG_APPLIED 0x2
212#define QFLAG_DONE (QFLAG_APPLY_ONCE|QFLAG_APPLIED) 196#define QFLAG_DONE (QFLAG_APPLY_ONCE|QFLAG_APPLIED)
@@ -236,8 +220,6 @@ static struct chipset early_qrk[] __initdata = {
236 PCI_CLASS_SERIAL_SMBUS, PCI_ANY_ID, 0, ati_bugs }, 220 PCI_CLASS_SERIAL_SMBUS, PCI_ANY_ID, 0, ati_bugs },
237 { PCI_VENDOR_ID_ATI, PCI_DEVICE_ID_ATI_SBX00_SMBUS, 221 { PCI_VENDOR_ID_ATI, PCI_DEVICE_ID_ATI_SBX00_SMBUS,
238 PCI_CLASS_SERIAL_SMBUS, PCI_ANY_ID, 0, ati_bugs_contd }, 222 PCI_CLASS_SERIAL_SMBUS, PCI_ANY_ID, 0, ati_bugs_contd },
239 { PCI_VENDOR_ID_ATI, PCI_ANY_ID,
240 PCI_CLASS_SERIAL_SMBUS, PCI_ANY_ID, 0, ati_hpet_bugs },
241 {} 223 {}
242}; 224};
243 225
diff --git a/arch/x86/kernel/head_32.S b/arch/x86/kernel/head_32.S
index ff4c453e13f3..fa8c1b8e09fb 100644
--- a/arch/x86/kernel/head_32.S
+++ b/arch/x86/kernel/head_32.S
@@ -334,7 +334,7 @@ ENTRY(startup_32_smp)
334/* 334/*
335 * Enable paging 335 * Enable paging
336 */ 336 */
337 movl $pa(swapper_pg_dir),%eax 337 movl pa(initial_page_table), %eax
338 movl %eax,%cr3 /* set the page table pointer.. */ 338 movl %eax,%cr3 /* set the page table pointer.. */
339 movl %cr0,%eax 339 movl %cr0,%eax
340 orl $X86_CR0_PG,%eax 340 orl $X86_CR0_PG,%eax
@@ -614,6 +614,8 @@ ignore_int:
614.align 4 614.align 4
615ENTRY(initial_code) 615ENTRY(initial_code)
616 .long i386_start_kernel 616 .long i386_start_kernel
617ENTRY(initial_page_table)
618 .long pa(swapper_pg_dir)
617 619
618/* 620/*
619 * BSS section 621 * BSS section
@@ -629,6 +631,10 @@ ENTRY(swapper_pg_dir)
629#endif 631#endif
630swapper_pg_fixmap: 632swapper_pg_fixmap:
631 .fill 1024,4,0 633 .fill 1024,4,0
634#ifdef CONFIG_X86_TRAMPOLINE
635ENTRY(trampoline_pg_dir)
636 .fill 1024,4,0
637#endif
632ENTRY(empty_zero_page) 638ENTRY(empty_zero_page)
633 .fill 4096,1,0 639 .fill 4096,1,0
634 640
diff --git a/arch/x86/kernel/hpet.c b/arch/x86/kernel/hpet.c
index 351f9c0fea1f..410fdb3f1939 100644
--- a/arch/x86/kernel/hpet.c
+++ b/arch/x86/kernel/hpet.c
@@ -35,7 +35,6 @@
35unsigned long hpet_address; 35unsigned long hpet_address;
36u8 hpet_blockid; /* OS timer block num */ 36u8 hpet_blockid; /* OS timer block num */
37u8 hpet_msi_disable; 37u8 hpet_msi_disable;
38u8 hpet_readback_cmp;
39 38
40#ifdef CONFIG_PCI_MSI 39#ifdef CONFIG_PCI_MSI
41static unsigned long hpet_num_timers; 40static unsigned long hpet_num_timers;
@@ -395,23 +394,27 @@ static int hpet_next_event(unsigned long delta,
395 * at that point and we would wait for the next hpet interrupt 394 * at that point and we would wait for the next hpet interrupt
396 * forever. We found out that reading the CMP register back 395 * forever. We found out that reading the CMP register back
397 * forces the transfer so we can rely on the comparison with 396 * forces the transfer so we can rely on the comparison with
398 * the counter register below. 397 * the counter register below. If the read back from the
398 * compare register does not match the value we programmed
399 * then we might have a real hardware problem. We can not do
400 * much about it here, but at least alert the user/admin with
401 * a prominent warning.
399 * 402 *
400 * That works fine on those ATI chipsets, but on newer Intel 403 * An erratum on some chipsets (ICH9,..), results in
401 * chipsets (ICH9...) this triggers due to an erratum: Reading 404 * comparator read immediately following a write returning old
402 * the comparator immediately following a write is returning 405 * value. Workaround for this is to read this value second
403 * the old value. 406 * time, when first read returns old value.
404 * 407 *
405 * We restrict the read back to the affected ATI chipsets (set 408 * In fact the write to the comparator register is delayed up
406 * by quirks) and also run it with hpet=verbose for debugging 409 * to two HPET cycles so the workaround we tried to restrict
407 * purposes. 410 * the readback to those known to be borked ATI chipsets
411 * failed miserably. So we give up on optimizations forever
412 * and penalize all HPET incarnations unconditionally.
408 */ 413 */
409 if (hpet_readback_cmp || hpet_verbose) { 414 if (unlikely((u32)hpet_readl(HPET_Tn_CMP(timer)) != cnt)) {
410 u32 cmp = hpet_readl(HPET_Tn_CMP(timer)); 415 if (hpet_readl(HPET_Tn_CMP(timer)) != cnt)
411
412 if (cmp != cnt)
413 printk_once(KERN_WARNING 416 printk_once(KERN_WARNING
414 "hpet: compare register read back failed.\n"); 417 "hpet: compare register read back failed.\n");
415 } 418 }
416 419
417 return (s32)(hpet_readl(HPET_COUNTER) - cnt) >= 0 ? -ETIME : 0; 420 return (s32)(hpet_readl(HPET_COUNTER) - cnt) >= 0 ? -ETIME : 0;
diff --git a/arch/x86/kernel/hw_breakpoint.c b/arch/x86/kernel/hw_breakpoint.c
index a474ec37c32f..ff15c9dcc25d 100644
--- a/arch/x86/kernel/hw_breakpoint.c
+++ b/arch/x86/kernel/hw_breakpoint.c
@@ -206,11 +206,27 @@ int arch_check_bp_in_kernelspace(struct perf_event *bp)
206int arch_bp_generic_fields(int x86_len, int x86_type, 206int arch_bp_generic_fields(int x86_len, int x86_type,
207 int *gen_len, int *gen_type) 207 int *gen_len, int *gen_type)
208{ 208{
209 /* Len */ 209 /* Type */
210 switch (x86_len) { 210 switch (x86_type) {
211 case X86_BREAKPOINT_LEN_X: 211 case X86_BREAKPOINT_EXECUTE:
212 if (x86_len != X86_BREAKPOINT_LEN_X)
213 return -EINVAL;
214
215 *gen_type = HW_BREAKPOINT_X;
212 *gen_len = sizeof(long); 216 *gen_len = sizeof(long);
217 return 0;
218 case X86_BREAKPOINT_WRITE:
219 *gen_type = HW_BREAKPOINT_W;
213 break; 220 break;
221 case X86_BREAKPOINT_RW:
222 *gen_type = HW_BREAKPOINT_W | HW_BREAKPOINT_R;
223 break;
224 default:
225 return -EINVAL;
226 }
227
228 /* Len */
229 switch (x86_len) {
214 case X86_BREAKPOINT_LEN_1: 230 case X86_BREAKPOINT_LEN_1:
215 *gen_len = HW_BREAKPOINT_LEN_1; 231 *gen_len = HW_BREAKPOINT_LEN_1;
216 break; 232 break;
@@ -229,21 +245,6 @@ int arch_bp_generic_fields(int x86_len, int x86_type,
229 return -EINVAL; 245 return -EINVAL;
230 } 246 }
231 247
232 /* Type */
233 switch (x86_type) {
234 case X86_BREAKPOINT_EXECUTE:
235 *gen_type = HW_BREAKPOINT_X;
236 break;
237 case X86_BREAKPOINT_WRITE:
238 *gen_type = HW_BREAKPOINT_W;
239 break;
240 case X86_BREAKPOINT_RW:
241 *gen_type = HW_BREAKPOINT_W | HW_BREAKPOINT_R;
242 break;
243 default:
244 return -EINVAL;
245 }
246
247 return 0; 248 return 0;
248} 249}
249 250
@@ -316,9 +317,6 @@ int arch_validate_hwbkpt_settings(struct perf_event *bp)
316 ret = -EINVAL; 317 ret = -EINVAL;
317 318
318 switch (info->len) { 319 switch (info->len) {
319 case X86_BREAKPOINT_LEN_X:
320 align = sizeof(long) -1;
321 break;
322 case X86_BREAKPOINT_LEN_1: 320 case X86_BREAKPOINT_LEN_1:
323 align = 0; 321 align = 0;
324 break; 322 break;
diff --git a/arch/x86/kernel/i387.c b/arch/x86/kernel/i387.c
index 1f11f5ce668f..a46cb3522c0c 100644
--- a/arch/x86/kernel/i387.c
+++ b/arch/x86/kernel/i387.c
@@ -40,6 +40,7 @@
40 40
41static unsigned int mxcsr_feature_mask __read_mostly = 0xffffffffu; 41static unsigned int mxcsr_feature_mask __read_mostly = 0xffffffffu;
42unsigned int xstate_size; 42unsigned int xstate_size;
43EXPORT_SYMBOL_GPL(xstate_size);
43unsigned int sig_xstate_ia32_size = sizeof(struct _fpstate_ia32); 44unsigned int sig_xstate_ia32_size = sizeof(struct _fpstate_ia32);
44static struct i387_fxsave_struct fx_scratch __cpuinitdata; 45static struct i387_fxsave_struct fx_scratch __cpuinitdata;
45 46
diff --git a/arch/x86/kernel/kgdb.c b/arch/x86/kernel/kgdb.c
index ef10940e1af0..852b81967a37 100644
--- a/arch/x86/kernel/kgdb.c
+++ b/arch/x86/kernel/kgdb.c
@@ -194,7 +194,7 @@ static struct hw_breakpoint {
194 unsigned long addr; 194 unsigned long addr;
195 int len; 195 int len;
196 int type; 196 int type;
197 struct perf_event **pev; 197 struct perf_event * __percpu *pev;
198} breakinfo[HBP_NUM]; 198} breakinfo[HBP_NUM];
199 199
200static unsigned long early_dr7; 200static unsigned long early_dr7;
diff --git a/arch/x86/kernel/kprobes.c b/arch/x86/kernel/kprobes.c
index 1bfb6cf4dd55..770ebfb349e9 100644
--- a/arch/x86/kernel/kprobes.c
+++ b/arch/x86/kernel/kprobes.c
@@ -709,6 +709,7 @@ static __used __kprobes void *trampoline_handler(struct pt_regs *regs)
709 struct hlist_node *node, *tmp; 709 struct hlist_node *node, *tmp;
710 unsigned long flags, orig_ret_address = 0; 710 unsigned long flags, orig_ret_address = 0;
711 unsigned long trampoline_address = (unsigned long)&kretprobe_trampoline; 711 unsigned long trampoline_address = (unsigned long)&kretprobe_trampoline;
712 kprobe_opcode_t *correct_ret_addr = NULL;
712 713
713 INIT_HLIST_HEAD(&empty_rp); 714 INIT_HLIST_HEAD(&empty_rp);
714 kretprobe_hash_lock(current, &head, &flags); 715 kretprobe_hash_lock(current, &head, &flags);
@@ -740,14 +741,34 @@ static __used __kprobes void *trampoline_handler(struct pt_regs *regs)
740 /* another task is sharing our hash bucket */ 741 /* another task is sharing our hash bucket */
741 continue; 742 continue;
742 743
744 orig_ret_address = (unsigned long)ri->ret_addr;
745
746 if (orig_ret_address != trampoline_address)
747 /*
748 * This is the real return address. Any other
749 * instances associated with this task are for
750 * other calls deeper on the call stack
751 */
752 break;
753 }
754
755 kretprobe_assert(ri, orig_ret_address, trampoline_address);
756
757 correct_ret_addr = ri->ret_addr;
758 hlist_for_each_entry_safe(ri, node, tmp, head, hlist) {
759 if (ri->task != current)
760 /* another task is sharing our hash bucket */
761 continue;
762
763 orig_ret_address = (unsigned long)ri->ret_addr;
743 if (ri->rp && ri->rp->handler) { 764 if (ri->rp && ri->rp->handler) {
744 __get_cpu_var(current_kprobe) = &ri->rp->kp; 765 __get_cpu_var(current_kprobe) = &ri->rp->kp;
745 get_kprobe_ctlblk()->kprobe_status = KPROBE_HIT_ACTIVE; 766 get_kprobe_ctlblk()->kprobe_status = KPROBE_HIT_ACTIVE;
767 ri->ret_addr = correct_ret_addr;
746 ri->rp->handler(ri, regs); 768 ri->rp->handler(ri, regs);
747 __get_cpu_var(current_kprobe) = NULL; 769 __get_cpu_var(current_kprobe) = NULL;
748 } 770 }
749 771
750 orig_ret_address = (unsigned long)ri->ret_addr;
751 recycle_rp_inst(ri, &empty_rp); 772 recycle_rp_inst(ri, &empty_rp);
752 773
753 if (orig_ret_address != trampoline_address) 774 if (orig_ret_address != trampoline_address)
@@ -759,8 +780,6 @@ static __used __kprobes void *trampoline_handler(struct pt_regs *regs)
759 break; 780 break;
760 } 781 }
761 782
762 kretprobe_assert(ri, orig_ret_address, trampoline_address);
763
764 kretprobe_hash_unlock(current, &flags); 783 kretprobe_hash_unlock(current, &flags);
765 784
766 hlist_for_each_entry_safe(ri, node, tmp, &empty_rp, hlist) { 785 hlist_for_each_entry_safe(ri, node, tmp, &empty_rp, hlist) {
diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c
index 64ecaf0af9af..57d1868a86aa 100644
--- a/arch/x86/kernel/process.c
+++ b/arch/x86/kernel/process.c
@@ -301,8 +301,9 @@ EXPORT_SYMBOL(kernel_thread);
301/* 301/*
302 * sys_execve() executes a new program. 302 * sys_execve() executes a new program.
303 */ 303 */
304long sys_execve(const char __user *name, char __user * __user *argv, 304long sys_execve(const char __user *name,
305 char __user * __user *envp, struct pt_regs *regs) 305 const char __user *const __user *argv,
306 const char __user *const __user *envp, struct pt_regs *regs)
306{ 307{
307 long error; 308 long error;
308 char *filename; 309 char *filename;
diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c
index b008e7883207..c3a4fbb2b996 100644
--- a/arch/x86/kernel/setup.c
+++ b/arch/x86/kernel/setup.c
@@ -1014,6 +1014,8 @@ void __init setup_arch(char **cmdline_p)
1014 paging_init(); 1014 paging_init();
1015 x86_init.paging.pagetable_setup_done(swapper_pg_dir); 1015 x86_init.paging.pagetable_setup_done(swapper_pg_dir);
1016 1016
1017 setup_trampoline_page_table();
1018
1017 tboot_probe(); 1019 tboot_probe();
1018 1020
1019#ifdef CONFIG_X86_64 1021#ifdef CONFIG_X86_64
diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c
index a5e928b0cb5f..8b3bfc4dd708 100644
--- a/arch/x86/kernel/smpboot.c
+++ b/arch/x86/kernel/smpboot.c
@@ -73,7 +73,6 @@
73 73
74#ifdef CONFIG_X86_32 74#ifdef CONFIG_X86_32
75u8 apicid_2_node[MAX_APICID]; 75u8 apicid_2_node[MAX_APICID];
76static int low_mappings;
77#endif 76#endif
78 77
79/* State of each CPU */ 78/* State of each CPU */
@@ -91,6 +90,25 @@ DEFINE_PER_CPU(int, cpu_state) = { 0 };
91static DEFINE_PER_CPU(struct task_struct *, idle_thread_array); 90static DEFINE_PER_CPU(struct task_struct *, idle_thread_array);
92#define get_idle_for_cpu(x) (per_cpu(idle_thread_array, x)) 91#define get_idle_for_cpu(x) (per_cpu(idle_thread_array, x))
93#define set_idle_for_cpu(x, p) (per_cpu(idle_thread_array, x) = (p)) 92#define set_idle_for_cpu(x, p) (per_cpu(idle_thread_array, x) = (p))
93
94/*
95 * We need this for trampoline_base protection from concurrent accesses when
96 * off- and onlining cores wildly.
97 */
98static DEFINE_MUTEX(x86_cpu_hotplug_driver_mutex);
99
100void cpu_hotplug_driver_lock()
101{
102 mutex_lock(&x86_cpu_hotplug_driver_mutex);
103}
104
105void cpu_hotplug_driver_unlock()
106{
107 mutex_unlock(&x86_cpu_hotplug_driver_mutex);
108}
109
110ssize_t arch_cpu_probe(const char *buf, size_t count) { return -1; }
111ssize_t arch_cpu_release(const char *buf, size_t count) { return -1; }
94#else 112#else
95static struct task_struct *idle_thread_array[NR_CPUS] __cpuinitdata ; 113static struct task_struct *idle_thread_array[NR_CPUS] __cpuinitdata ;
96#define get_idle_for_cpu(x) (idle_thread_array[(x)]) 114#define get_idle_for_cpu(x) (idle_thread_array[(x)])
@@ -281,6 +299,18 @@ notrace static void __cpuinit start_secondary(void *unused)
281 * fragile that we want to limit the things done here to the 299 * fragile that we want to limit the things done here to the
282 * most necessary things. 300 * most necessary things.
283 */ 301 */
302
303#ifdef CONFIG_X86_32
304 /*
305 * Switch away from the trampoline page-table
306 *
307 * Do this before cpu_init() because it needs to access per-cpu
308 * data which may not be mapped in the trampoline page-table.
309 */
310 load_cr3(swapper_pg_dir);
311 __flush_tlb_all();
312#endif
313
284 vmi_bringup(); 314 vmi_bringup();
285 cpu_init(); 315 cpu_init();
286 preempt_disable(); 316 preempt_disable();
@@ -299,12 +329,6 @@ notrace static void __cpuinit start_secondary(void *unused)
299 legacy_pic->chip->unmask(0); 329 legacy_pic->chip->unmask(0);
300 } 330 }
301 331
302#ifdef CONFIG_X86_32
303 while (low_mappings)
304 cpu_relax();
305 __flush_tlb_all();
306#endif
307
308 /* This must be done before setting cpu_online_mask */ 332 /* This must be done before setting cpu_online_mask */
309 set_cpu_sibling_map(raw_smp_processor_id()); 333 set_cpu_sibling_map(raw_smp_processor_id());
310 wmb(); 334 wmb();
@@ -750,6 +774,7 @@ do_rest:
750#ifdef CONFIG_X86_32 774#ifdef CONFIG_X86_32
751 /* Stack for startup_32 can be just as for start_secondary onwards */ 775 /* Stack for startup_32 can be just as for start_secondary onwards */
752 irq_ctx_init(cpu); 776 irq_ctx_init(cpu);
777 initial_page_table = __pa(&trampoline_pg_dir);
753#else 778#else
754 clear_tsk_thread_flag(c_idle.idle, TIF_FORK); 779 clear_tsk_thread_flag(c_idle.idle, TIF_FORK);
755 initial_gs = per_cpu_offset(cpu); 780 initial_gs = per_cpu_offset(cpu);
@@ -897,20 +922,8 @@ int __cpuinit native_cpu_up(unsigned int cpu)
897 922
898 per_cpu(cpu_state, cpu) = CPU_UP_PREPARE; 923 per_cpu(cpu_state, cpu) = CPU_UP_PREPARE;
899 924
900#ifdef CONFIG_X86_32
901 /* init low mem mapping */
902 clone_pgd_range(swapper_pg_dir, swapper_pg_dir + KERNEL_PGD_BOUNDARY,
903 min_t(unsigned long, KERNEL_PGD_PTRS, KERNEL_PGD_BOUNDARY));
904 flush_tlb_all();
905 low_mappings = 1;
906
907 err = do_boot_cpu(apicid, cpu); 925 err = do_boot_cpu(apicid, cpu);
908 926
909 zap_low_mappings(false);
910 low_mappings = 0;
911#else
912 err = do_boot_cpu(apicid, cpu);
913#endif
914 if (err) { 927 if (err) {
915 pr_debug("do_boot_cpu failed %d\n", err); 928 pr_debug("do_boot_cpu failed %d\n", err);
916 return -EIO; 929 return -EIO;
diff --git a/arch/x86/kernel/sys_i386_32.c b/arch/x86/kernel/sys_i386_32.c
index 196552bb412c..d5e06624e34a 100644
--- a/arch/x86/kernel/sys_i386_32.c
+++ b/arch/x86/kernel/sys_i386_32.c
@@ -28,7 +28,9 @@
28 * Do a system call from kernel instead of calling sys_execve so we 28 * Do a system call from kernel instead of calling sys_execve so we
29 * end up with proper pt_regs. 29 * end up with proper pt_regs.
30 */ 30 */
31int kernel_execve(const char *filename, char *const argv[], char *const envp[]) 31int kernel_execve(const char *filename,
32 const char *const argv[],
33 const char *const envp[])
32{ 34{
33 long __res; 35 long __res;
34 asm volatile ("push %%ebx ; movl %2,%%ebx ; int $0x80 ; pop %%ebx" 36 asm volatile ("push %%ebx ; movl %2,%%ebx ; int $0x80 ; pop %%ebx"
diff --git a/arch/x86/kernel/trampoline.c b/arch/x86/kernel/trampoline.c
index c652ef62742d..e2a595257390 100644
--- a/arch/x86/kernel/trampoline.c
+++ b/arch/x86/kernel/trampoline.c
@@ -1,6 +1,7 @@
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/pgtable.h>
4#include <asm/e820.h> 5#include <asm/e820.h>
5 6
6#if defined(CONFIG_X86_64) && defined(CONFIG_ACPI_SLEEP) 7#if defined(CONFIG_X86_64) && defined(CONFIG_ACPI_SLEEP)
@@ -37,3 +38,19 @@ unsigned long __trampinit setup_trampoline(void)
37 memcpy(trampoline_base, trampoline_data, TRAMPOLINE_SIZE); 38 memcpy(trampoline_base, trampoline_data, TRAMPOLINE_SIZE);
38 return virt_to_phys(trampoline_base); 39 return virt_to_phys(trampoline_base);
39} 40}
41
42void __init setup_trampoline_page_table(void)
43{
44#ifdef CONFIG_X86_32
45 /* Copy kernel address range */
46 clone_pgd_range(trampoline_pg_dir + KERNEL_PGD_BOUNDARY,
47 swapper_pg_dir + KERNEL_PGD_BOUNDARY,
48 KERNEL_PGD_PTRS);
49
50 /* Initialize low mappings */
51 clone_pgd_range(trampoline_pg_dir,
52 swapper_pg_dir + KERNEL_PGD_BOUNDARY,
53 min_t(unsigned long, KERNEL_PGD_PTRS,
54 KERNEL_PGD_BOUNDARY));
55#endif
56}
diff --git a/arch/x86/kernel/tsc.c b/arch/x86/kernel/tsc.c
index ce8e50239332..26a863a9c2a8 100644
--- a/arch/x86/kernel/tsc.c
+++ b/arch/x86/kernel/tsc.c
@@ -626,6 +626,44 @@ static void set_cyc2ns_scale(unsigned long cpu_khz, int cpu)
626 local_irq_restore(flags); 626 local_irq_restore(flags);
627} 627}
628 628
629static unsigned long long cyc2ns_suspend;
630
631void save_sched_clock_state(void)
632{
633 if (!sched_clock_stable)
634 return;
635
636 cyc2ns_suspend = sched_clock();
637}
638
639/*
640 * Even on processors with invariant TSC, TSC gets reset in some the
641 * ACPI system sleep states. And in some systems BIOS seem to reinit TSC to
642 * arbitrary value (still sync'd across cpu's) during resume from such sleep
643 * states. To cope up with this, recompute the cyc2ns_offset for each cpu so
644 * that sched_clock() continues from the point where it was left off during
645 * suspend.
646 */
647void restore_sched_clock_state(void)
648{
649 unsigned long long offset;
650 unsigned long flags;
651 int cpu;
652
653 if (!sched_clock_stable)
654 return;
655
656 local_irq_save(flags);
657
658 __get_cpu_var(cyc2ns_offset) = 0;
659 offset = cyc2ns_suspend - sched_clock();
660
661 for_each_possible_cpu(cpu)
662 per_cpu(cyc2ns_offset, cpu) = offset;
663
664 local_irq_restore(flags);
665}
666
629#ifdef CONFIG_CPU_FREQ 667#ifdef CONFIG_CPU_FREQ
630 668
631/* Frequency scaling support. Adjust the TSC based timer when the cpu frequency 669/* Frequency scaling support. Adjust the TSC based timer when the cpu frequency
diff --git a/arch/x86/kvm/emulate.c b/arch/x86/kvm/emulate.c
index b38bd8b92aa6..66ca98aafdd6 100644
--- a/arch/x86/kvm/emulate.c
+++ b/arch/x86/kvm/emulate.c
@@ -1870,17 +1870,16 @@ static inline int emulate_grp9(struct x86_emulate_ctxt *ctxt,
1870 struct x86_emulate_ops *ops) 1870 struct x86_emulate_ops *ops)
1871{ 1871{
1872 struct decode_cache *c = &ctxt->decode; 1872 struct decode_cache *c = &ctxt->decode;
1873 u64 old = c->dst.orig_val; 1873 u64 old = c->dst.orig_val64;
1874 1874
1875 if (((u32) (old >> 0) != (u32) c->regs[VCPU_REGS_RAX]) || 1875 if (((u32) (old >> 0) != (u32) c->regs[VCPU_REGS_RAX]) ||
1876 ((u32) (old >> 32) != (u32) c->regs[VCPU_REGS_RDX])) { 1876 ((u32) (old >> 32) != (u32) c->regs[VCPU_REGS_RDX])) {
1877
1878 c->regs[VCPU_REGS_RAX] = (u32) (old >> 0); 1877 c->regs[VCPU_REGS_RAX] = (u32) (old >> 0);
1879 c->regs[VCPU_REGS_RDX] = (u32) (old >> 32); 1878 c->regs[VCPU_REGS_RDX] = (u32) (old >> 32);
1880 ctxt->eflags &= ~EFLG_ZF; 1879 ctxt->eflags &= ~EFLG_ZF;
1881 } else { 1880 } else {
1882 c->dst.val = ((u64)c->regs[VCPU_REGS_RCX] << 32) | 1881 c->dst.val64 = ((u64)c->regs[VCPU_REGS_RCX] << 32) |
1883 (u32) c->regs[VCPU_REGS_RBX]; 1882 (u32) c->regs[VCPU_REGS_RBX];
1884 1883
1885 ctxt->eflags |= EFLG_ZF; 1884 ctxt->eflags |= EFLG_ZF;
1886 } 1885 }
@@ -2616,7 +2615,7 @@ x86_emulate_insn(struct x86_emulate_ctxt *ctxt, struct x86_emulate_ops *ops)
2616 c->src.valptr, c->src.bytes); 2615 c->src.valptr, c->src.bytes);
2617 if (rc != X86EMUL_CONTINUE) 2616 if (rc != X86EMUL_CONTINUE)
2618 goto done; 2617 goto done;
2619 c->src.orig_val = c->src.val; 2618 c->src.orig_val64 = c->src.val64;
2620 } 2619 }
2621 2620
2622 if (c->src2.type == OP_MEM) { 2621 if (c->src2.type == OP_MEM) {
diff --git a/arch/x86/kvm/i8254.c b/arch/x86/kvm/i8254.c
index 0fd6378981f4..ddeb2314b522 100644
--- a/arch/x86/kvm/i8254.c
+++ b/arch/x86/kvm/i8254.c
@@ -697,6 +697,7 @@ struct kvm_pit *kvm_create_pit(struct kvm *kvm, u32 flags)
697 pit->wq = create_singlethread_workqueue("kvm-pit-wq"); 697 pit->wq = create_singlethread_workqueue("kvm-pit-wq");
698 if (!pit->wq) { 698 if (!pit->wq) {
699 mutex_unlock(&pit->pit_state.lock); 699 mutex_unlock(&pit->pit_state.lock);
700 kvm_free_irq_source_id(kvm, pit->irq_source_id);
700 kfree(pit); 701 kfree(pit);
701 return NULL; 702 return NULL;
702 } 703 }
@@ -742,7 +743,7 @@ fail:
742 kvm_unregister_irq_mask_notifier(kvm, 0, &pit->mask_notifier); 743 kvm_unregister_irq_mask_notifier(kvm, 0, &pit->mask_notifier);
743 kvm_unregister_irq_ack_notifier(kvm, &pit_state->irq_ack_notifier); 744 kvm_unregister_irq_ack_notifier(kvm, &pit_state->irq_ack_notifier);
744 kvm_free_irq_source_id(kvm, pit->irq_source_id); 745 kvm_free_irq_source_id(kvm, pit->irq_source_id);
745 746 destroy_workqueue(pit->wq);
746 kfree(pit); 747 kfree(pit);
747 return NULL; 748 return NULL;
748} 749}
diff --git a/arch/x86/kvm/i8259.c b/arch/x86/kvm/i8259.c
index 8d10c063d7f2..4b7b73ce2098 100644
--- a/arch/x86/kvm/i8259.c
+++ b/arch/x86/kvm/i8259.c
@@ -64,6 +64,9 @@ static void pic_unlock(struct kvm_pic *s)
64 if (!found) 64 if (!found)
65 found = s->kvm->bsp_vcpu; 65 found = s->kvm->bsp_vcpu;
66 66
67 if (!found)
68 return;
69
67 kvm_vcpu_kick(found); 70 kvm_vcpu_kick(found);
68 } 71 }
69} 72}
diff --git a/arch/x86/kvm/irq.h b/arch/x86/kvm/irq.h
index ffed06871c5c..63c314502993 100644
--- a/arch/x86/kvm/irq.h
+++ b/arch/x86/kvm/irq.h
@@ -43,7 +43,6 @@ struct kvm_kpic_state {
43 u8 irr; /* interrupt request register */ 43 u8 irr; /* interrupt request register */
44 u8 imr; /* interrupt mask register */ 44 u8 imr; /* interrupt mask register */
45 u8 isr; /* interrupt service register */ 45 u8 isr; /* interrupt service register */
46 u8 isr_ack; /* interrupt ack detection */
47 u8 priority_add; /* highest irq priority */ 46 u8 priority_add; /* highest irq priority */
48 u8 irq_base; 47 u8 irq_base;
49 u8 read_reg_select; 48 u8 read_reg_select;
@@ -56,6 +55,7 @@ struct kvm_kpic_state {
56 u8 init4; /* true if 4 byte init */ 55 u8 init4; /* true if 4 byte init */
57 u8 elcr; /* PIIX edge/trigger selection */ 56 u8 elcr; /* PIIX edge/trigger selection */
58 u8 elcr_mask; 57 u8 elcr_mask;
58 u8 isr_ack; /* interrupt ack detection */
59 struct kvm_pic *pics_state; 59 struct kvm_pic *pics_state;
60}; 60};
61 61
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 25f19078b321..3a09c625d526 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -2387,7 +2387,7 @@ static void kvm_vcpu_ioctl_x86_get_xsave(struct kvm_vcpu *vcpu,
2387 if (cpu_has_xsave) 2387 if (cpu_has_xsave)
2388 memcpy(guest_xsave->region, 2388 memcpy(guest_xsave->region,
2389 &vcpu->arch.guest_fpu.state->xsave, 2389 &vcpu->arch.guest_fpu.state->xsave,
2390 sizeof(struct xsave_struct)); 2390 xstate_size);
2391 else { 2391 else {
2392 memcpy(guest_xsave->region, 2392 memcpy(guest_xsave->region,
2393 &vcpu->arch.guest_fpu.state->fxsave, 2393 &vcpu->arch.guest_fpu.state->fxsave,
@@ -2405,7 +2405,7 @@ static int kvm_vcpu_ioctl_x86_set_xsave(struct kvm_vcpu *vcpu,
2405 2405
2406 if (cpu_has_xsave) 2406 if (cpu_has_xsave)
2407 memcpy(&vcpu->arch.guest_fpu.state->xsave, 2407 memcpy(&vcpu->arch.guest_fpu.state->xsave,
2408 guest_xsave->region, sizeof(struct xsave_struct)); 2408 guest_xsave->region, xstate_size);
2409 else { 2409 else {
2410 if (xstate_bv & ~XSTATE_FPSSE) 2410 if (xstate_bv & ~XSTATE_FPSSE)
2411 return -EINVAL; 2411 return -EINVAL;
diff --git a/arch/x86/lguest/boot.c b/arch/x86/lguest/boot.c
index 9257510b4836..9d5f55848455 100644
--- a/arch/x86/lguest/boot.c
+++ b/arch/x86/lguest/boot.c
@@ -324,9 +324,8 @@ static void lguest_load_gdt(const struct desc_ptr *desc)
324} 324}
325 325
326/* 326/*
327 * For a single GDT entry which changes, we do the lazy thing: alter our GDT, 327 * For a single GDT entry which changes, we simply change our copy and
328 * then tell the Host to reload the entire thing. This operation is so rare 328 * then tell the host about it.
329 * that this naive implementation is reasonable.
330 */ 329 */
331static void lguest_write_gdt_entry(struct desc_struct *dt, int entrynum, 330static void lguest_write_gdt_entry(struct desc_struct *dt, int entrynum,
332 const void *desc, int type) 331 const void *desc, int type)
@@ -338,9 +337,13 @@ static void lguest_write_gdt_entry(struct desc_struct *dt, int entrynum,
338} 337}
339 338
340/* 339/*
341 * OK, I lied. There are three "thread local storage" GDT entries which change 340 * There are three "thread local storage" GDT entries which change
342 * on every context switch (these three entries are how glibc implements 341 * on every context switch (these three entries are how glibc implements
343 * __thread variables). So we have a hypercall specifically for this case. 342 * __thread variables). As an optimization, we have a hypercall
343 * specifically for this case.
344 *
345 * Wouldn't it be nicer to have a general LOAD_GDT_ENTRIES hypercall
346 * which took a range of entries?
344 */ 347 */
345static void lguest_load_tls(struct thread_struct *t, unsigned int cpu) 348static void lguest_load_tls(struct thread_struct *t, unsigned int cpu)
346{ 349{
diff --git a/arch/x86/mm/iomap_32.c b/arch/x86/mm/iomap_32.c
index 84e236ce76ba..72fc70cf6184 100644
--- a/arch/x86/mm/iomap_32.c
+++ b/arch/x86/mm/iomap_32.c
@@ -74,7 +74,7 @@ void *kmap_atomic_prot_pfn(unsigned long pfn, enum km_type type, pgprot_t prot)
74/* 74/*
75 * Map 'pfn' using fixed map 'type' and protections 'prot' 75 * Map 'pfn' using fixed map 'type' and protections 'prot'
76 */ 76 */
77void * 77void __iomem *
78iomap_atomic_prot_pfn(unsigned long pfn, enum km_type type, pgprot_t prot) 78iomap_atomic_prot_pfn(unsigned long pfn, enum km_type type, pgprot_t prot)
79{ 79{
80 /* 80 /*
@@ -86,12 +86,12 @@ iomap_atomic_prot_pfn(unsigned long pfn, enum km_type type, pgprot_t prot)
86 if (!pat_enabled && pgprot_val(prot) == pgprot_val(PAGE_KERNEL_WC)) 86 if (!pat_enabled && pgprot_val(prot) == pgprot_val(PAGE_KERNEL_WC))
87 prot = PAGE_KERNEL_UC_MINUS; 87 prot = PAGE_KERNEL_UC_MINUS;
88 88
89 return kmap_atomic_prot_pfn(pfn, type, prot); 89 return (void __force __iomem *) kmap_atomic_prot_pfn(pfn, type, prot);
90} 90}
91EXPORT_SYMBOL_GPL(iomap_atomic_prot_pfn); 91EXPORT_SYMBOL_GPL(iomap_atomic_prot_pfn);
92 92
93void 93void
94iounmap_atomic(void *kvaddr, enum km_type type) 94iounmap_atomic(void __iomem *kvaddr, enum km_type type)
95{ 95{
96 unsigned long vaddr = (unsigned long) kvaddr & PAGE_MASK; 96 unsigned long vaddr = (unsigned long) kvaddr & PAGE_MASK;
97 enum fixed_addresses idx = type + KM_TYPE_NR*smp_processor_id(); 97 enum fixed_addresses idx = type + KM_TYPE_NR*smp_processor_id();
diff --git a/arch/x86/oprofile/nmi_int.c b/arch/x86/oprofile/nmi_int.c
index f6b48f6c5951..009b819f48d0 100644
--- a/arch/x86/oprofile/nmi_int.c
+++ b/arch/x86/oprofile/nmi_int.c
@@ -568,8 +568,13 @@ static int __init init_sysfs(void)
568 int error; 568 int error;
569 569
570 error = sysdev_class_register(&oprofile_sysclass); 570 error = sysdev_class_register(&oprofile_sysclass);
571 if (!error) 571 if (error)
572 error = sysdev_register(&device_oprofile); 572 return error;
573
574 error = sysdev_register(&device_oprofile);
575 if (error)
576 sysdev_class_unregister(&oprofile_sysclass);
577
573 return error; 578 return error;
574} 579}
575 580
@@ -580,8 +585,10 @@ static void exit_sysfs(void)
580} 585}
581 586
582#else 587#else
583#define init_sysfs() do { } while (0) 588
584#define exit_sysfs() do { } while (0) 589static inline int init_sysfs(void) { return 0; }
590static inline void exit_sysfs(void) { }
591
585#endif /* CONFIG_PM */ 592#endif /* CONFIG_PM */
586 593
587static int __init p4_init(char **cpu_type) 594static int __init p4_init(char **cpu_type)
@@ -664,7 +671,9 @@ static int __init ppro_init(char **cpu_type)
664 case 14: 671 case 14:
665 *cpu_type = "i386/core"; 672 *cpu_type = "i386/core";
666 break; 673 break;
667 case 15: case 23: 674 case 0x0f:
675 case 0x16:
676 case 0x17:
668 *cpu_type = "i386/core_2"; 677 *cpu_type = "i386/core_2";
669 break; 678 break;
670 case 0x1a: 679 case 0x1a:
@@ -695,6 +704,8 @@ int __init op_nmi_init(struct oprofile_operations *ops)
695 char *cpu_type = NULL; 704 char *cpu_type = NULL;
696 int ret = 0; 705 int ret = 0;
697 706
707 using_nmi = 0;
708
698 if (!cpu_has_apic) 709 if (!cpu_has_apic)
699 return -ENODEV; 710 return -ENODEV;
700 711
@@ -774,7 +785,10 @@ int __init op_nmi_init(struct oprofile_operations *ops)
774 785
775 mux_init(ops); 786 mux_init(ops);
776 787
777 init_sysfs(); 788 ret = init_sysfs();
789 if (ret)
790 return ret;
791
778 using_nmi = 1; 792 using_nmi = 1;
779 printk(KERN_INFO "oprofile: using NMI interrupt.\n"); 793 printk(KERN_INFO "oprofile: using NMI interrupt.\n");
780 return 0; 794 return 0;
diff --git a/arch/x86/power/cpu.c b/arch/x86/power/cpu.c
index e7e8c5f54956..87bb35e34ef1 100644
--- a/arch/x86/power/cpu.c
+++ b/arch/x86/power/cpu.c
@@ -113,6 +113,7 @@ static void __save_processor_state(struct saved_context *ctxt)
113void save_processor_state(void) 113void save_processor_state(void)
114{ 114{
115 __save_processor_state(&saved_context); 115 __save_processor_state(&saved_context);
116 save_sched_clock_state();
116} 117}
117#ifdef CONFIG_X86_32 118#ifdef CONFIG_X86_32
118EXPORT_SYMBOL(save_processor_state); 119EXPORT_SYMBOL(save_processor_state);
@@ -229,6 +230,7 @@ static void __restore_processor_state(struct saved_context *ctxt)
229void restore_processor_state(void) 230void restore_processor_state(void)
230{ 231{
231 __restore_processor_state(&saved_context); 232 __restore_processor_state(&saved_context);
233 restore_sched_clock_state();
232} 234}
233#ifdef CONFIG_X86_32 235#ifdef CONFIG_X86_32
234EXPORT_SYMBOL(restore_processor_state); 236EXPORT_SYMBOL(restore_processor_state);
diff --git a/arch/x86/xen/platform-pci-unplug.c b/arch/x86/xen/platform-pci-unplug.c
index 554c002a1e1a..0f456386cce5 100644
--- a/arch/x86/xen/platform-pci-unplug.c
+++ b/arch/x86/xen/platform-pci-unplug.c
@@ -72,13 +72,17 @@ void __init xen_unplug_emulated_devices(void)
72{ 72{
73 int r; 73 int r;
74 74
75 /* user explicitly requested no unplug */
76 if (xen_emul_unplug & XEN_UNPLUG_NEVER)
77 return;
75 /* check the version of the xen platform PCI device */ 78 /* check the version of the xen platform PCI device */
76 r = check_platform_magic(); 79 r = check_platform_magic();
77 /* If the version matches enable the Xen platform PCI driver. 80 /* If the version matches enable the Xen platform PCI driver.
78 * Also enable the Xen platform PCI driver if the version is really old 81 * Also enable the Xen platform PCI driver if the host does
79 * and the user told us to ignore it. */ 82 * not support the unplug protocol (XEN_PLATFORM_ERR_MAGIC)
83 * but the user told us that unplugging is unnecessary. */
80 if (r && !(r == XEN_PLATFORM_ERR_MAGIC && 84 if (r && !(r == XEN_PLATFORM_ERR_MAGIC &&
81 (xen_emul_unplug & XEN_UNPLUG_IGNORE))) 85 (xen_emul_unplug & XEN_UNPLUG_UNNECESSARY)))
82 return; 86 return;
83 /* Set the default value of xen_emul_unplug depending on whether or 87 /* Set the default value of xen_emul_unplug depending on whether or
84 * not the Xen PV frontends and the Xen platform PCI driver have 88 * not the Xen PV frontends and the Xen platform PCI driver have
@@ -99,7 +103,7 @@ void __init xen_unplug_emulated_devices(void)
99 } 103 }
100 } 104 }
101 /* Now unplug the emulated devices */ 105 /* Now unplug the emulated devices */
102 if (!(xen_emul_unplug & XEN_UNPLUG_IGNORE)) 106 if (!(xen_emul_unplug & XEN_UNPLUG_UNNECESSARY))
103 outw(xen_emul_unplug, XEN_IOPORT_UNPLUG); 107 outw(xen_emul_unplug, XEN_IOPORT_UNPLUG);
104 xen_platform_pci_unplug = xen_emul_unplug; 108 xen_platform_pci_unplug = xen_emul_unplug;
105} 109}
@@ -125,8 +129,10 @@ static int __init parse_xen_emul_unplug(char *arg)
125 xen_emul_unplug |= XEN_UNPLUG_AUX_IDE_DISKS; 129 xen_emul_unplug |= XEN_UNPLUG_AUX_IDE_DISKS;
126 else if (!strncmp(p, "nics", l)) 130 else if (!strncmp(p, "nics", l))
127 xen_emul_unplug |= XEN_UNPLUG_ALL_NICS; 131 xen_emul_unplug |= XEN_UNPLUG_ALL_NICS;
128 else if (!strncmp(p, "ignore", l)) 132 else if (!strncmp(p, "unnecessary", l))
129 xen_emul_unplug |= XEN_UNPLUG_IGNORE; 133 xen_emul_unplug |= XEN_UNPLUG_UNNECESSARY;
134 else if (!strncmp(p, "never", l))
135 xen_emul_unplug |= XEN_UNPLUG_NEVER;
130 else 136 else
131 printk(KERN_WARNING "unrecognised option '%s' " 137 printk(KERN_WARNING "unrecognised option '%s' "
132 "in parameter 'xen_emul_unplug'\n", p); 138 "in parameter 'xen_emul_unplug'\n", p);