aboutsummaryrefslogtreecommitdiffstats
path: root/arch/x86
diff options
context:
space:
mode:
authorJiri Kosina <jkosina@suse.cz>2010-04-22 20:08:44 -0400
committerJiri Kosina <jkosina@suse.cz>2010-04-22 20:08:44 -0400
commit6c9468e9eb1252eaefd94ce7f06e1be9b0b641b1 (patch)
tree797676a336b050bfa1ef879377c07e541b9075d6 /arch/x86
parent4cb3ca7cd7e2cae8d1daf5345ec99a1e8502cf3f (diff)
parentc81eddb0e3728661d1585fbc564449c94165cc36 (diff)
Merge branch 'master' into for-next
Diffstat (limited to 'arch/x86')
-rw-r--r--arch/x86/Kconfig8
-rw-r--r--arch/x86/crypto/fpu.c1
-rw-r--r--arch/x86/ia32/ia32_aout.c1
-rw-r--r--arch/x86/ia32/ia32entry.S2
-rw-r--r--arch/x86/ia32/sys_ia32.c1
-rw-r--r--arch/x86/include/asm/amd_iommu_types.h3
-rw-r--r--arch/x86/include/asm/fixmap.h6
-rw-r--r--arch/x86/include/asm/hw_breakpoint.h1
-rw-r--r--arch/x86/include/asm/hw_irq.h1
-rw-r--r--arch/x86/include/asm/lguest_hcall.h29
-rw-r--r--arch/x86/include/asm/msr-index.h2
-rw-r--r--arch/x86/include/asm/perf_event.h16
-rw-r--r--arch/x86/include/asm/pgtable_32.h1
-rw-r--r--arch/x86/kernel/acpi/boot.c101
-rw-r--r--arch/x86/kernel/alternative.c1
-rw-r--r--arch/x86/kernel/amd_iommu.c22
-rw-r--r--arch/x86/kernel/amd_iommu_init.c50
-rw-r--r--arch/x86/kernel/apb_timer.c1
-rw-r--r--arch/x86/kernel/aperture_64.c16
-rw-r--r--arch/x86/kernel/apic/apic.c2
-rw-r--r--arch/x86/kernel/apic/es7000_32.c1
-rw-r--r--arch/x86/kernel/apic/io_apic.c9
-rw-r--r--arch/x86/kernel/apic/nmi.c1
-rw-r--r--arch/x86/kernel/apic/x2apic_uv_x.c5
-rw-r--r--arch/x86/kernel/bootflag.c1
-rw-r--r--arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c1
-rw-r--r--arch/x86/kernel/cpu/cpufreq/elanfreq.c1
-rw-r--r--arch/x86/kernel/cpu/cpufreq/gx-suspmod.c1
-rw-r--r--arch/x86/kernel/cpu/cpufreq/longrun.c1
-rw-r--r--arch/x86/kernel/cpu/cpufreq/p4-clockmod.c1
-rw-r--r--arch/x86/kernel/cpu/cpufreq/pcc-cpufreq.c1
-rw-r--r--arch/x86/kernel/cpu/cpufreq/powernow-k6.c1
-rw-r--r--arch/x86/kernel/cpu/cpufreq/speedstep-centrino.c1
-rw-r--r--arch/x86/kernel/cpu/cpufreq/speedstep-ich.c1
-rw-r--r--arch/x86/kernel/cpu/cpufreq/speedstep-lib.c1
-rw-r--r--arch/x86/kernel/cpu/cpufreq/speedstep-smi.c1
-rw-r--r--arch/x86/kernel/cpu/intel.c3
-rw-r--r--arch/x86/kernel/cpu/mcheck/mce-inject.c1
-rw-r--r--arch/x86/kernel/cpu/mcheck/mce.c16
-rw-r--r--arch/x86/kernel/cpu/mcheck/mce_amd.c1
-rw-r--r--arch/x86/kernel/cpu/mcheck/mce_intel.c5
-rw-r--r--arch/x86/kernel/cpu/mtrr/generic.c1
-rw-r--r--arch/x86/kernel/cpu/mtrr/if.c1
-rw-r--r--arch/x86/kernel/cpu/perf_event.c267
-rw-r--r--arch/x86/kernel/cpu/perf_event_amd.c142
-rw-r--r--arch/x86/kernel/cpu/perf_event_intel.c95
-rw-r--r--arch/x86/kernel/cpu/perf_event_p6.c18
-rw-r--r--arch/x86/kernel/cpu/perfctr-watchdog.c2
-rw-r--r--arch/x86/kernel/cpuid.c1
-rw-r--r--arch/x86/kernel/crash.c6
-rw-r--r--arch/x86/kernel/crash_dump_32.c1
-rw-r--r--arch/x86/kernel/dumpstack.h24
-rw-r--r--arch/x86/kernel/dumpstack_64.c14
-rw-r--r--arch/x86/kernel/e820.c24
-rw-r--r--arch/x86/kernel/head32.c4
-rw-r--r--arch/x86/kernel/head64.c3
-rw-r--r--arch/x86/kernel/hpet.c10
-rw-r--r--arch/x86/kernel/hw_breakpoint.c12
-rw-r--r--arch/x86/kernel/i387.c1
-rw-r--r--arch/x86/kernel/i8259.c1
-rw-r--r--arch/x86/kernel/irqinit.c23
-rw-r--r--arch/x86/kernel/k8.c16
-rw-r--r--arch/x86/kernel/kdebugfs.c1
-rw-r--r--arch/x86/kernel/kgdb.c2
-rw-r--r--arch/x86/kernel/ldt.c1
-rw-r--r--arch/x86/kernel/machine_kexec_64.c1
-rw-r--r--arch/x86/kernel/mca_32.c1
-rw-r--r--arch/x86/kernel/module.c1
-rw-r--r--arch/x86/kernel/mpparse.c4
-rw-r--r--arch/x86/kernel/msr.c1
-rw-r--r--arch/x86/kernel/pci-dma.c1
-rw-r--r--arch/x86/kernel/pci-gart_64.c6
-rw-r--r--arch/x86/kernel/pci-nommu.c1
-rw-r--r--arch/x86/kernel/process.c34
-rw-r--r--arch/x86/kernel/ptrace.c1
-rw-r--r--arch/x86/kernel/setup.c25
-rw-r--r--arch/x86/kernel/smp.c1
-rw-r--r--arch/x86/kernel/smpboot.c7
-rw-r--r--arch/x86/kernel/tlb_uv.c1
-rw-r--r--arch/x86/kernel/uv_irq.c1
-rw-r--r--arch/x86/kernel/uv_time.c1
-rw-r--r--arch/x86/kernel/vmi_32.c1
-rw-r--r--arch/x86/kernel/vmlinux.lds.S2
-rw-r--r--arch/x86/kvm/i8254.c1
-rw-r--r--arch/x86/kvm/i8259.c1
-rw-r--r--arch/x86/kvm/lapic.c1
-rw-r--r--arch/x86/kvm/mmu.c12
-rw-r--r--arch/x86/kvm/svm.c26
-rw-r--r--arch/x86/kvm/vmx.c25
-rw-r--r--arch/x86/kvm/x86.c49
-rw-r--r--arch/x86/lguest/boot.c61
-rw-r--r--arch/x86/lguest/i386_head.S2
-rw-r--r--arch/x86/mm/hugetlbpage.c1
-rw-r--r--arch/x86/mm/init.c33
-rw-r--r--arch/x86/mm/init_32.c2
-rw-r--r--arch/x86/mm/init_64.c1
-rw-r--r--arch/x86/mm/kmmio.c1
-rw-r--r--arch/x86/mm/mmio-mod.c1
-rw-r--r--arch/x86/mm/pageattr.c27
-rw-r--r--arch/x86/mm/pat.c2
-rw-r--r--arch/x86/mm/pgtable.c1
-rw-r--r--arch/x86/mm/pgtable_32.c1
-rw-r--r--arch/x86/oprofile/op_model_amd.c23
-rw-r--r--arch/x86/oprofile/op_model_ppro.c6
-rw-r--r--arch/x86/pci/acpi.c23
-rw-r--r--arch/x86/pci/common.c1
-rw-r--r--arch/x86/pci/i386.c5
-rw-r--r--arch/x86/pci/irq.c1
-rw-r--r--arch/x86/pci/mmconfig-shared.c1
-rw-r--r--arch/x86/pci/pcbios.c1
-rw-r--r--arch/x86/power/hibernate_32.c1
-rw-r--r--arch/x86/power/hibernate_64.c1
-rw-r--r--arch/x86/power/hibernate_asm_32.S15
-rw-r--r--arch/x86/vdso/vma.c1
-rw-r--r--arch/x86/xen/debugfs.c1
-rw-r--r--arch/x86/xen/enlighten.c1
-rw-r--r--arch/x86/xen/mmu.c1
-rw-r--r--arch/x86/xen/smp.c1
-rw-r--r--arch/x86/xen/spinlock.c1
-rw-r--r--arch/x86/xen/time.c1
120 files changed, 838 insertions, 574 deletions
diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
index 93936de67796..9458685902bd 100644
--- a/arch/x86/Kconfig
+++ b/arch/x86/Kconfig
@@ -662,7 +662,7 @@ config GART_IOMMU
662 bool "GART IOMMU support" if EMBEDDED 662 bool "GART IOMMU support" if EMBEDDED
663 default y 663 default y
664 select SWIOTLB 664 select SWIOTLB
665 depends on X86_64 && PCI 665 depends on X86_64 && PCI && K8_NB
666 ---help--- 666 ---help---
667 Support for full DMA access of devices with 32bit memory access only 667 Support for full DMA access of devices with 32bit memory access only
668 on systems with more than 3GB. This is usually needed for USB, 668 on systems with more than 3GB. This is usually needed for USB,
@@ -1216,8 +1216,8 @@ config NUMA_EMU
1216 1216
1217config NODES_SHIFT 1217config NODES_SHIFT
1218 int "Maximum NUMA Nodes (as a power of 2)" if !MAXSMP 1218 int "Maximum NUMA Nodes (as a power of 2)" if !MAXSMP
1219 range 1 9 1219 range 1 10
1220 default "9" if MAXSMP 1220 default "10" if MAXSMP
1221 default "6" if X86_64 1221 default "6" if X86_64
1222 default "4" if X86_NUMAQ 1222 default "4" if X86_NUMAQ
1223 default "3" 1223 default "3"
@@ -2061,7 +2061,7 @@ endif # X86_32
2061 2061
2062config K8_NB 2062config K8_NB
2063 def_bool y 2063 def_bool y
2064 depends on AGP_AMD64 || (X86_64 && (GART_IOMMU || (PCI && NUMA))) 2064 depends on CPU_SUP_AMD && PCI
2065 2065
2066source "drivers/pcmcia/Kconfig" 2066source "drivers/pcmcia/Kconfig"
2067 2067
diff --git a/arch/x86/crypto/fpu.c b/arch/x86/crypto/fpu.c
index daef6cd2b45d..1a8f8649c035 100644
--- a/arch/x86/crypto/fpu.c
+++ b/arch/x86/crypto/fpu.c
@@ -16,6 +16,7 @@
16#include <linux/init.h> 16#include <linux/init.h>
17#include <linux/kernel.h> 17#include <linux/kernel.h>
18#include <linux/module.h> 18#include <linux/module.h>
19#include <linux/slab.h>
19#include <asm/i387.h> 20#include <asm/i387.h>
20 21
21struct crypto_fpu_ctx { 22struct crypto_fpu_ctx {
diff --git a/arch/x86/ia32/ia32_aout.c b/arch/x86/ia32/ia32_aout.c
index 280c019cfad8..0350311906ae 100644
--- a/arch/x86/ia32/ia32_aout.c
+++ b/arch/x86/ia32/ia32_aout.c
@@ -21,7 +21,6 @@
21#include <linux/fcntl.h> 21#include <linux/fcntl.h>
22#include <linux/ptrace.h> 22#include <linux/ptrace.h>
23#include <linux/user.h> 23#include <linux/user.h>
24#include <linux/slab.h>
25#include <linux/binfmts.h> 24#include <linux/binfmts.h>
26#include <linux/personality.h> 25#include <linux/personality.h>
27#include <linux/init.h> 26#include <linux/init.h>
diff --git a/arch/x86/ia32/ia32entry.S b/arch/x86/ia32/ia32entry.S
index 59b4556a5b92..e790bc1fbfa3 100644
--- a/arch/x86/ia32/ia32entry.S
+++ b/arch/x86/ia32/ia32entry.S
@@ -626,7 +626,7 @@ ia32_sys_call_table:
626 .quad stub32_sigreturn 626 .quad stub32_sigreturn
627 .quad stub32_clone /* 120 */ 627 .quad stub32_clone /* 120 */
628 .quad sys_setdomainname 628 .quad sys_setdomainname
629 .quad sys_uname 629 .quad sys_newuname
630 .quad sys_modify_ldt 630 .quad sys_modify_ldt
631 .quad compat_sys_adjtimex 631 .quad compat_sys_adjtimex
632 .quad sys32_mprotect /* 125 */ 632 .quad sys32_mprotect /* 125 */
diff --git a/arch/x86/ia32/sys_ia32.c b/arch/x86/ia32/sys_ia32.c
index 74c35431b7d8..626be156d88d 100644
--- a/arch/x86/ia32/sys_ia32.c
+++ b/arch/x86/ia32/sys_ia32.c
@@ -40,6 +40,7 @@
40#include <linux/ptrace.h> 40#include <linux/ptrace.h>
41#include <linux/highuid.h> 41#include <linux/highuid.h>
42#include <linux/sysctl.h> 42#include <linux/sysctl.h>
43#include <linux/slab.h>
43#include <asm/mman.h> 44#include <asm/mman.h>
44#include <asm/types.h> 45#include <asm/types.h>
45#include <asm/uaccess.h> 46#include <asm/uaccess.h>
diff --git a/arch/x86/include/asm/amd_iommu_types.h b/arch/x86/include/asm/amd_iommu_types.h
index ba19ad4c47d0..86a0ff0aeac7 100644
--- a/arch/x86/include/asm/amd_iommu_types.h
+++ b/arch/x86/include/asm/amd_iommu_types.h
@@ -21,6 +21,7 @@
21#define _ASM_X86_AMD_IOMMU_TYPES_H 21#define _ASM_X86_AMD_IOMMU_TYPES_H
22 22
23#include <linux/types.h> 23#include <linux/types.h>
24#include <linux/mutex.h>
24#include <linux/list.h> 25#include <linux/list.h>
25#include <linux/spinlock.h> 26#include <linux/spinlock.h>
26 27
@@ -140,6 +141,7 @@
140 141
141/* constants to configure the command buffer */ 142/* constants to configure the command buffer */
142#define CMD_BUFFER_SIZE 8192 143#define CMD_BUFFER_SIZE 8192
144#define CMD_BUFFER_UNINITIALIZED 1
143#define CMD_BUFFER_ENTRIES 512 145#define CMD_BUFFER_ENTRIES 512
144#define MMIO_CMD_SIZE_SHIFT 56 146#define MMIO_CMD_SIZE_SHIFT 56
145#define MMIO_CMD_SIZE_512 (0x9ULL << MMIO_CMD_SIZE_SHIFT) 147#define MMIO_CMD_SIZE_512 (0x9ULL << MMIO_CMD_SIZE_SHIFT)
@@ -237,6 +239,7 @@ struct protection_domain {
237 struct list_head list; /* for list of all protection domains */ 239 struct list_head list; /* for list of all protection domains */
238 struct list_head dev_list; /* List of all devices in this domain */ 240 struct list_head dev_list; /* List of all devices in this domain */
239 spinlock_t lock; /* mostly used to lock the page table*/ 241 spinlock_t lock; /* mostly used to lock the page table*/
242 struct mutex api_lock; /* protect page tables in the iommu-api path */
240 u16 id; /* the domain id written to the device table */ 243 u16 id; /* the domain id written to the device table */
241 int mode; /* paging mode (0-6 levels) */ 244 int mode; /* paging mode (0-6 levels) */
242 u64 *pt_root; /* page table root pointer */ 245 u64 *pt_root; /* page table root pointer */
diff --git a/arch/x86/include/asm/fixmap.h b/arch/x86/include/asm/fixmap.h
index 635f03bb4995..d07b44f7d1dc 100644
--- a/arch/x86/include/asm/fixmap.h
+++ b/arch/x86/include/asm/fixmap.h
@@ -82,6 +82,9 @@ enum fixed_addresses {
82#endif 82#endif
83 FIX_DBGP_BASE, 83 FIX_DBGP_BASE,
84 FIX_EARLYCON_MEM_BASE, 84 FIX_EARLYCON_MEM_BASE,
85#ifdef CONFIG_PROVIDE_OHCI1394_DMA_INIT
86 FIX_OHCI1394_BASE,
87#endif
85#ifdef CONFIG_X86_LOCAL_APIC 88#ifdef CONFIG_X86_LOCAL_APIC
86 FIX_APIC_BASE, /* local (CPU) APIC) -- required for SMP or not */ 89 FIX_APIC_BASE, /* local (CPU) APIC) -- required for SMP or not */
87#endif 90#endif
@@ -132,9 +135,6 @@ enum fixed_addresses {
132 (__end_of_permanent_fixed_addresses & (TOTAL_FIX_BTMAPS - 1)) 135 (__end_of_permanent_fixed_addresses & (TOTAL_FIX_BTMAPS - 1))
133 : __end_of_permanent_fixed_addresses, 136 : __end_of_permanent_fixed_addresses,
134 FIX_BTMAP_BEGIN = FIX_BTMAP_END + TOTAL_FIX_BTMAPS - 1, 137 FIX_BTMAP_BEGIN = FIX_BTMAP_END + TOTAL_FIX_BTMAPS - 1,
135#ifdef CONFIG_PROVIDE_OHCI1394_DMA_INIT
136 FIX_OHCI1394_BASE,
137#endif
138#ifdef CONFIG_X86_32 138#ifdef CONFIG_X86_32
139 FIX_WP_TEST, 139 FIX_WP_TEST,
140#endif 140#endif
diff --git a/arch/x86/include/asm/hw_breakpoint.h b/arch/x86/include/asm/hw_breakpoint.h
index 0675a7c4c20e..2a1bd8f4f23a 100644
--- a/arch/x86/include/asm/hw_breakpoint.h
+++ b/arch/x86/include/asm/hw_breakpoint.h
@@ -10,7 +10,6 @@
10 * (display/resolving) 10 * (display/resolving)
11 */ 11 */
12struct arch_hw_breakpoint { 12struct arch_hw_breakpoint {
13 char *name; /* Contains name of the symbol to set bkpt */
14 unsigned long address; 13 unsigned long address;
15 u8 len; 14 u8 len;
16 u8 type; 15 u8 type;
diff --git a/arch/x86/include/asm/hw_irq.h b/arch/x86/include/asm/hw_irq.h
index a929c9ede33d..46c0fe05f230 100644
--- a/arch/x86/include/asm/hw_irq.h
+++ b/arch/x86/include/asm/hw_irq.h
@@ -133,6 +133,7 @@ extern void (*__initconst interrupt[NR_VECTORS-FIRST_EXTERNAL_VECTOR])(void);
133 133
134typedef int vector_irq_t[NR_VECTORS]; 134typedef int vector_irq_t[NR_VECTORS];
135DECLARE_PER_CPU(vector_irq_t, vector_irq); 135DECLARE_PER_CPU(vector_irq_t, vector_irq);
136extern void setup_vector_irq(int cpu);
136 137
137#ifdef CONFIG_X86_IO_APIC 138#ifdef CONFIG_X86_IO_APIC
138extern void lock_vector_lock(void); 139extern void lock_vector_lock(void);
diff --git a/arch/x86/include/asm/lguest_hcall.h b/arch/x86/include/asm/lguest_hcall.h
index ba0eed8aa1a6..b60f2924c413 100644
--- a/arch/x86/include/asm/lguest_hcall.h
+++ b/arch/x86/include/asm/lguest_hcall.h
@@ -28,22 +28,39 @@
28 28
29#ifndef __ASSEMBLY__ 29#ifndef __ASSEMBLY__
30#include <asm/hw_irq.h> 30#include <asm/hw_irq.h>
31#include <asm/kvm_para.h>
32 31
33/*G:030 32/*G:030
34 * But first, how does our Guest contact the Host to ask for privileged 33 * But first, how does our Guest contact the Host to ask for privileged
35 * operations? There are two ways: the direct way is to make a "hypercall", 34 * operations? There are two ways: the direct way is to make a "hypercall",
36 * to make requests of the Host Itself. 35 * to make requests of the Host Itself.
37 * 36 *
38 * We use the KVM hypercall mechanism, though completely different hypercall 37 * Our hypercall mechanism uses the highest unused trap code (traps 32 and
39 * numbers. Seventeen hypercalls are available: the hypercall number is put in 38 * above are used by real hardware interrupts). Seventeen hypercalls are
40 * the %eax register, and the arguments (when required) are placed in %ebx, 39 * available: the hypercall number is put in the %eax register, and the
41 * %ecx, %edx and %esi. If a return value makes sense, it's returned in %eax. 40 * arguments (when required) are placed in %ebx, %ecx, %edx and %esi.
41 * If a return value makes sense, it's returned in %eax.
42 * 42 *
43 * Grossly invalid calls result in Sudden Death at the hands of the vengeful 43 * Grossly invalid calls result in Sudden Death at the hands of the vengeful
44 * Host, rather than returning failure. This reflects Winston Churchill's 44 * Host, rather than returning failure. This reflects Winston Churchill's
45 * definition of a gentleman: "someone who is only rude intentionally". 45 * definition of a gentleman: "someone who is only rude intentionally".
46:*/ 46 */
47static inline unsigned long
48hcall(unsigned long call,
49 unsigned long arg1, unsigned long arg2, unsigned long arg3,
50 unsigned long arg4)
51{
52 /* "int" is the Intel instruction to trigger a trap. */
53 asm volatile("int $" __stringify(LGUEST_TRAP_ENTRY)
54 /* The call in %eax (aka "a") might be overwritten */
55 : "=a"(call)
56 /* The arguments are in %eax, %ebx, %ecx, %edx & %esi */
57 : "a"(call), "b"(arg1), "c"(arg2), "d"(arg3), "S"(arg4)
58 /* "memory" means this might write somewhere in memory.
59 * This isn't true for all calls, but it's safe to tell
60 * gcc that it might happen so it doesn't get clever. */
61 : "memory");
62 return call;
63}
47 64
48/* Can't use our min() macro here: needs to be a constant */ 65/* Can't use our min() macro here: needs to be a constant */
49#define LGUEST_IRQS (NR_IRQS < 32 ? NR_IRQS: 32) 66#define LGUEST_IRQS (NR_IRQS < 32 ? NR_IRQS: 32)
diff --git a/arch/x86/include/asm/msr-index.h b/arch/x86/include/asm/msr-index.h
index 1cd58cdbc03f..4604e6a54d36 100644
--- a/arch/x86/include/asm/msr-index.h
+++ b/arch/x86/include/asm/msr-index.h
@@ -105,6 +105,8 @@
105#define MSR_AMD64_PATCH_LEVEL 0x0000008b 105#define MSR_AMD64_PATCH_LEVEL 0x0000008b
106#define MSR_AMD64_NB_CFG 0xc001001f 106#define MSR_AMD64_NB_CFG 0xc001001f
107#define MSR_AMD64_PATCH_LOADER 0xc0010020 107#define MSR_AMD64_PATCH_LOADER 0xc0010020
108#define MSR_AMD64_OSVW_ID_LENGTH 0xc0010140
109#define MSR_AMD64_OSVW_STATUS 0xc0010141
108#define MSR_AMD64_IBSFETCHCTL 0xc0011030 110#define MSR_AMD64_IBSFETCHCTL 0xc0011030
109#define MSR_AMD64_IBSFETCHLINAD 0xc0011031 111#define MSR_AMD64_IBSFETCHLINAD 0xc0011031
110#define MSR_AMD64_IBSFETCHPHYSAD 0xc0011032 112#define MSR_AMD64_IBSFETCHPHYSAD 0xc0011032
diff --git a/arch/x86/include/asm/perf_event.h b/arch/x86/include/asm/perf_event.h
index befd172c82ad..db6109a885a7 100644
--- a/arch/x86/include/asm/perf_event.h
+++ b/arch/x86/include/asm/perf_event.h
@@ -18,7 +18,7 @@
18#define MSR_ARCH_PERFMON_EVENTSEL0 0x186 18#define MSR_ARCH_PERFMON_EVENTSEL0 0x186
19#define MSR_ARCH_PERFMON_EVENTSEL1 0x187 19#define MSR_ARCH_PERFMON_EVENTSEL1 0x187
20 20
21#define ARCH_PERFMON_EVENTSEL0_ENABLE (1 << 22) 21#define ARCH_PERFMON_EVENTSEL_ENABLE (1 << 22)
22#define ARCH_PERFMON_EVENTSEL_ANY (1 << 21) 22#define ARCH_PERFMON_EVENTSEL_ANY (1 << 21)
23#define ARCH_PERFMON_EVENTSEL_INT (1 << 20) 23#define ARCH_PERFMON_EVENTSEL_INT (1 << 20)
24#define ARCH_PERFMON_EVENTSEL_OS (1 << 17) 24#define ARCH_PERFMON_EVENTSEL_OS (1 << 17)
@@ -50,7 +50,7 @@
50 INTEL_ARCH_INV_MASK| \ 50 INTEL_ARCH_INV_MASK| \
51 INTEL_ARCH_EDGE_MASK|\ 51 INTEL_ARCH_EDGE_MASK|\
52 INTEL_ARCH_UNIT_MASK|\ 52 INTEL_ARCH_UNIT_MASK|\
53 INTEL_ARCH_EVTSEL_MASK) 53 INTEL_ARCH_EVENT_MASK)
54 54
55#define ARCH_PERFMON_UNHALTED_CORE_CYCLES_SEL 0x3c 55#define ARCH_PERFMON_UNHALTED_CORE_CYCLES_SEL 0x3c
56#define ARCH_PERFMON_UNHALTED_CORE_CYCLES_UMASK (0x00 << 8) 56#define ARCH_PERFMON_UNHALTED_CORE_CYCLES_UMASK (0x00 << 8)
@@ -117,6 +117,18 @@ union cpuid10_edx {
117 */ 117 */
118#define X86_PMC_IDX_FIXED_BTS (X86_PMC_IDX_FIXED + 16) 118#define X86_PMC_IDX_FIXED_BTS (X86_PMC_IDX_FIXED + 16)
119 119
120/* IbsFetchCtl bits/masks */
121#define IBS_FETCH_RAND_EN (1ULL<<57)
122#define IBS_FETCH_VAL (1ULL<<49)
123#define IBS_FETCH_ENABLE (1ULL<<48)
124#define IBS_FETCH_CNT 0xFFFF0000ULL
125#define IBS_FETCH_MAX_CNT 0x0000FFFFULL
126
127/* IbsOpCtl bits */
128#define IBS_OP_CNT_CTL (1ULL<<19)
129#define IBS_OP_VAL (1ULL<<18)
130#define IBS_OP_ENABLE (1ULL<<17)
131#define IBS_OP_MAX_CNT 0x0000FFFFULL
120 132
121#ifdef CONFIG_PERF_EVENTS 133#ifdef CONFIG_PERF_EVENTS
122extern void init_hw_perf_events(void); 134extern void init_hw_perf_events(void);
diff --git a/arch/x86/include/asm/pgtable_32.h b/arch/x86/include/asm/pgtable_32.h
index 47339a1ac7b6..2984a25ff383 100644
--- a/arch/x86/include/asm/pgtable_32.h
+++ b/arch/x86/include/asm/pgtable_32.h
@@ -19,7 +19,6 @@
19#include <asm/paravirt.h> 19#include <asm/paravirt.h>
20 20
21#include <linux/bitops.h> 21#include <linux/bitops.h>
22#include <linux/slab.h>
23#include <linux/list.h> 22#include <linux/list.h>
24#include <linux/spinlock.h> 23#include <linux/spinlock.h>
25 24
diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c
index a54d714545ff..cd40aba6aa95 100644
--- a/arch/x86/kernel/acpi/boot.c
+++ b/arch/x86/kernel/acpi/boot.c
@@ -31,6 +31,7 @@
31#include <linux/module.h> 31#include <linux/module.h>
32#include <linux/dmi.h> 32#include <linux/dmi.h>
33#include <linux/irq.h> 33#include <linux/irq.h>
34#include <linux/slab.h>
34#include <linux/bootmem.h> 35#include <linux/bootmem.h>
35#include <linux/ioport.h> 36#include <linux/ioport.h>
36#include <linux/pci.h> 37#include <linux/pci.h>
@@ -490,6 +491,7 @@ int acpi_register_gsi(struct device *dev, u32 gsi, int trigger, int polarity)
490 * ACPI based hotplug support for CPU 491 * ACPI based hotplug support for CPU
491 */ 492 */
492#ifdef CONFIG_ACPI_HOTPLUG_CPU 493#ifdef CONFIG_ACPI_HOTPLUG_CPU
494#include <acpi/processor.h>
493 495
494static void acpi_map_cpu2node(acpi_handle handle, int cpu, int physid) 496static void acpi_map_cpu2node(acpi_handle handle, int cpu, int physid)
495{ 497{
@@ -567,6 +569,8 @@ static int __cpuinit _acpi_map_lsapic(acpi_handle handle, int *pcpu)
567 goto free_new_map; 569 goto free_new_map;
568 } 570 }
569 571
572 acpi_processor_set_pdc(handle);
573
570 cpu = cpumask_first(new_map); 574 cpu = cpumask_first(new_map);
571 acpi_map_cpu2node(handle, cpu, physid); 575 acpi_map_cpu2node(handle, cpu, physid);
572 576
@@ -1293,23 +1297,6 @@ static int __init dmi_disable_acpi(const struct dmi_system_id *d)
1293} 1297}
1294 1298
1295/* 1299/*
1296 * Limit ACPI to CPU enumeration for HT
1297 */
1298static int __init force_acpi_ht(const struct dmi_system_id *d)
1299{
1300 if (!acpi_force) {
1301 printk(KERN_NOTICE "%s detected: force use of acpi=ht\n",
1302 d->ident);
1303 disable_acpi();
1304 acpi_ht = 1;
1305 } else {
1306 printk(KERN_NOTICE
1307 "Warning: acpi=force overrules DMI blacklist: acpi=ht\n");
1308 }
1309 return 0;
1310}
1311
1312/*
1313 * Force ignoring BIOS IRQ0 pin2 override 1300 * Force ignoring BIOS IRQ0 pin2 override
1314 */ 1301 */
1315static int __init dmi_ignore_irq0_timer_override(const struct dmi_system_id *d) 1302static int __init dmi_ignore_irq0_timer_override(const struct dmi_system_id *d)
@@ -1345,82 +1332,6 @@ static struct dmi_system_id __initdata acpi_dmi_table[] = {
1345 }, 1332 },
1346 1333
1347 /* 1334 /*
1348 * Boxes that need acpi=ht
1349 */
1350 {
1351 .callback = force_acpi_ht,
1352 .ident = "FSC Primergy T850",
1353 .matches = {
1354 DMI_MATCH(DMI_SYS_VENDOR, "FUJITSU SIEMENS"),
1355 DMI_MATCH(DMI_PRODUCT_NAME, "PRIMERGY T850"),
1356 },
1357 },
1358 {
1359 .callback = force_acpi_ht,
1360 .ident = "HP VISUALIZE NT Workstation",
1361 .matches = {
1362 DMI_MATCH(DMI_BOARD_VENDOR, "Hewlett-Packard"),
1363 DMI_MATCH(DMI_PRODUCT_NAME, "HP VISUALIZE NT Workstation"),
1364 },
1365 },
1366 {
1367 .callback = force_acpi_ht,
1368 .ident = "Compaq Workstation W8000",
1369 .matches = {
1370 DMI_MATCH(DMI_SYS_VENDOR, "Compaq"),
1371 DMI_MATCH(DMI_PRODUCT_NAME, "Workstation W8000"),
1372 },
1373 },
1374 {
1375 .callback = force_acpi_ht,
1376 .ident = "ASUS CUR-DLS",
1377 .matches = {
1378 DMI_MATCH(DMI_BOARD_VENDOR, "ASUSTeK Computer INC."),
1379 DMI_MATCH(DMI_BOARD_NAME, "CUR-DLS"),
1380 },
1381 },
1382 {
1383 .callback = force_acpi_ht,
1384 .ident = "ABIT i440BX-W83977",
1385 .matches = {
1386 DMI_MATCH(DMI_BOARD_VENDOR, "ABIT <http://www.abit.com>"),
1387 DMI_MATCH(DMI_BOARD_NAME, "i440BX-W83977 (BP6)"),
1388 },
1389 },
1390 {
1391 .callback = force_acpi_ht,
1392 .ident = "IBM Bladecenter",
1393 .matches = {
1394 DMI_MATCH(DMI_BOARD_VENDOR, "IBM"),
1395 DMI_MATCH(DMI_BOARD_NAME, "IBM eServer BladeCenter HS20"),
1396 },
1397 },
1398 {
1399 .callback = force_acpi_ht,
1400 .ident = "IBM eServer xSeries 360",
1401 .matches = {
1402 DMI_MATCH(DMI_BOARD_VENDOR, "IBM"),
1403 DMI_MATCH(DMI_BOARD_NAME, "eServer xSeries 360"),
1404 },
1405 },
1406 {
1407 .callback = force_acpi_ht,
1408 .ident = "IBM eserver xSeries 330",
1409 .matches = {
1410 DMI_MATCH(DMI_BOARD_VENDOR, "IBM"),
1411 DMI_MATCH(DMI_BOARD_NAME, "eserver xSeries 330"),
1412 },
1413 },
1414 {
1415 .callback = force_acpi_ht,
1416 .ident = "IBM eserver xSeries 440",
1417 .matches = {
1418 DMI_MATCH(DMI_BOARD_VENDOR, "IBM"),
1419 DMI_MATCH(DMI_PRODUCT_NAME, "eserver xSeries 440"),
1420 },
1421 },
1422
1423 /*
1424 * Boxes that need ACPI PCI IRQ routing disabled 1335 * Boxes that need ACPI PCI IRQ routing disabled
1425 */ 1336 */
1426 { 1337 {
@@ -1652,8 +1563,10 @@ static int __init parse_acpi(char *arg)
1652 } 1563 }
1653 /* Limit ACPI just to boot-time to enable HT */ 1564 /* Limit ACPI just to boot-time to enable HT */
1654 else if (strcmp(arg, "ht") == 0) { 1565 else if (strcmp(arg, "ht") == 0) {
1655 if (!acpi_force) 1566 if (!acpi_force) {
1567 printk(KERN_WARNING "acpi=ht will be removed in Linux-2.6.35\n");
1656 disable_acpi(); 1568 disable_acpi();
1569 }
1657 acpi_ht = 1; 1570 acpi_ht = 1;
1658 } 1571 }
1659 /* acpi=rsdt use RSDT instead of XSDT */ 1572 /* acpi=rsdt use RSDT instead of XSDT */
diff --git a/arch/x86/kernel/alternative.c b/arch/x86/kernel/alternative.c
index 3a4bf35c179b..1a160d5d44d0 100644
--- a/arch/x86/kernel/alternative.c
+++ b/arch/x86/kernel/alternative.c
@@ -8,6 +8,7 @@
8#include <linux/vmalloc.h> 8#include <linux/vmalloc.h>
9#include <linux/memory.h> 9#include <linux/memory.h>
10#include <linux/stop_machine.h> 10#include <linux/stop_machine.h>
11#include <linux/slab.h>
11#include <asm/alternative.h> 12#include <asm/alternative.h>
12#include <asm/sections.h> 13#include <asm/sections.h>
13#include <asm/pgtable.h> 14#include <asm/pgtable.h>
diff --git a/arch/x86/kernel/amd_iommu.c b/arch/x86/kernel/amd_iommu.c
index adb0ba025702..f854d89b7edf 100644
--- a/arch/x86/kernel/amd_iommu.c
+++ b/arch/x86/kernel/amd_iommu.c
@@ -18,8 +18,8 @@
18 */ 18 */
19 19
20#include <linux/pci.h> 20#include <linux/pci.h>
21#include <linux/gfp.h>
22#include <linux/bitmap.h> 21#include <linux/bitmap.h>
22#include <linux/slab.h>
23#include <linux/debugfs.h> 23#include <linux/debugfs.h>
24#include <linux/scatterlist.h> 24#include <linux/scatterlist.h>
25#include <linux/dma-mapping.h> 25#include <linux/dma-mapping.h>
@@ -118,7 +118,7 @@ static bool check_device(struct device *dev)
118 return false; 118 return false;
119 119
120 /* No device or no PCI device */ 120 /* No device or no PCI device */
121 if (!dev || dev->bus != &pci_bus_type) 121 if (dev->bus != &pci_bus_type)
122 return false; 122 return false;
123 123
124 devid = get_device_id(dev); 124 devid = get_device_id(dev);
@@ -392,6 +392,7 @@ static int __iommu_queue_command(struct amd_iommu *iommu, struct iommu_cmd *cmd)
392 u32 tail, head; 392 u32 tail, head;
393 u8 *target; 393 u8 *target;
394 394
395 WARN_ON(iommu->cmd_buf_size & CMD_BUFFER_UNINITIALIZED);
395 tail = readl(iommu->mmio_base + MMIO_CMD_TAIL_OFFSET); 396 tail = readl(iommu->mmio_base + MMIO_CMD_TAIL_OFFSET);
396 target = iommu->cmd_buf + tail; 397 target = iommu->cmd_buf + tail;
397 memcpy_toio(target, cmd, sizeof(*cmd)); 398 memcpy_toio(target, cmd, sizeof(*cmd));
@@ -2186,7 +2187,7 @@ static void prealloc_protection_domains(void)
2186 struct dma_ops_domain *dma_dom; 2187 struct dma_ops_domain *dma_dom;
2187 u16 devid; 2188 u16 devid;
2188 2189
2189 while ((dev = pci_get_device(PCI_ANY_ID, PCI_ANY_ID, dev)) != NULL) { 2190 for_each_pci_dev(dev) {
2190 2191
2191 /* Do we handle this device? */ 2192 /* Do we handle this device? */
2192 if (!check_device(&dev->dev)) 2193 if (!check_device(&dev->dev))
@@ -2298,7 +2299,7 @@ static void cleanup_domain(struct protection_domain *domain)
2298 list_for_each_entry_safe(dev_data, next, &domain->dev_list, list) { 2299 list_for_each_entry_safe(dev_data, next, &domain->dev_list, list) {
2299 struct device *dev = dev_data->dev; 2300 struct device *dev = dev_data->dev;
2300 2301
2301 do_detach(dev); 2302 __detach_device(dev);
2302 atomic_set(&dev_data->bind, 0); 2303 atomic_set(&dev_data->bind, 0);
2303 } 2304 }
2304 2305
@@ -2327,6 +2328,7 @@ static struct protection_domain *protection_domain_alloc(void)
2327 return NULL; 2328 return NULL;
2328 2329
2329 spin_lock_init(&domain->lock); 2330 spin_lock_init(&domain->lock);
2331 mutex_init(&domain->api_lock);
2330 domain->id = domain_id_alloc(); 2332 domain->id = domain_id_alloc();
2331 if (!domain->id) 2333 if (!domain->id)
2332 goto out_err; 2334 goto out_err;
@@ -2379,9 +2381,7 @@ static void amd_iommu_domain_destroy(struct iommu_domain *dom)
2379 2381
2380 free_pagetable(domain); 2382 free_pagetable(domain);
2381 2383
2382 domain_id_free(domain->id); 2384 protection_domain_free(domain);
2383
2384 kfree(domain);
2385 2385
2386 dom->priv = NULL; 2386 dom->priv = NULL;
2387} 2387}
@@ -2456,6 +2456,8 @@ static int amd_iommu_map_range(struct iommu_domain *dom,
2456 iova &= PAGE_MASK; 2456 iova &= PAGE_MASK;
2457 paddr &= PAGE_MASK; 2457 paddr &= PAGE_MASK;
2458 2458
2459 mutex_lock(&domain->api_lock);
2460
2459 for (i = 0; i < npages; ++i) { 2461 for (i = 0; i < npages; ++i) {
2460 ret = iommu_map_page(domain, iova, paddr, prot, PM_MAP_4k); 2462 ret = iommu_map_page(domain, iova, paddr, prot, PM_MAP_4k);
2461 if (ret) 2463 if (ret)
@@ -2465,6 +2467,8 @@ static int amd_iommu_map_range(struct iommu_domain *dom,
2465 paddr += PAGE_SIZE; 2467 paddr += PAGE_SIZE;
2466 } 2468 }
2467 2469
2470 mutex_unlock(&domain->api_lock);
2471
2468 return 0; 2472 return 0;
2469} 2473}
2470 2474
@@ -2477,12 +2481,16 @@ static void amd_iommu_unmap_range(struct iommu_domain *dom,
2477 2481
2478 iova &= PAGE_MASK; 2482 iova &= PAGE_MASK;
2479 2483
2484 mutex_lock(&domain->api_lock);
2485
2480 for (i = 0; i < npages; ++i) { 2486 for (i = 0; i < npages; ++i) {
2481 iommu_unmap_page(domain, iova, PM_MAP_4k); 2487 iommu_unmap_page(domain, iova, PM_MAP_4k);
2482 iova += PAGE_SIZE; 2488 iova += PAGE_SIZE;
2483 } 2489 }
2484 2490
2485 iommu_flush_tlb_pde(domain); 2491 iommu_flush_tlb_pde(domain);
2492
2493 mutex_unlock(&domain->api_lock);
2486} 2494}
2487 2495
2488static phys_addr_t amd_iommu_iova_to_phys(struct iommu_domain *dom, 2496static phys_addr_t amd_iommu_iova_to_phys(struct iommu_domain *dom,
diff --git a/arch/x86/kernel/amd_iommu_init.c b/arch/x86/kernel/amd_iommu_init.c
index 9dc91b431470..6360abf993d4 100644
--- a/arch/x86/kernel/amd_iommu_init.c
+++ b/arch/x86/kernel/amd_iommu_init.c
@@ -19,8 +19,8 @@
19 19
20#include <linux/pci.h> 20#include <linux/pci.h>
21#include <linux/acpi.h> 21#include <linux/acpi.h>
22#include <linux/gfp.h>
23#include <linux/list.h> 22#include <linux/list.h>
23#include <linux/slab.h>
24#include <linux/sysdev.h> 24#include <linux/sysdev.h>
25#include <linux/interrupt.h> 25#include <linux/interrupt.h>
26#include <linux/msi.h> 26#include <linux/msi.h>
@@ -138,9 +138,9 @@ int amd_iommus_present;
138bool amd_iommu_np_cache __read_mostly; 138bool amd_iommu_np_cache __read_mostly;
139 139
140/* 140/*
141 * Set to true if ACPI table parsing and hardware intialization went properly 141 * The ACPI table parsing functions set this variable on an error
142 */ 142 */
143static bool amd_iommu_initialized; 143static int __initdata amd_iommu_init_err;
144 144
145/* 145/*
146 * List of protection domains - used during resume 146 * List of protection domains - used during resume
@@ -391,9 +391,11 @@ static int __init find_last_devid_acpi(struct acpi_table_header *table)
391 */ 391 */
392 for (i = 0; i < table->length; ++i) 392 for (i = 0; i < table->length; ++i)
393 checksum += p[i]; 393 checksum += p[i];
394 if (checksum != 0) 394 if (checksum != 0) {
395 /* ACPI table corrupt */ 395 /* ACPI table corrupt */
396 return -ENODEV; 396 amd_iommu_init_err = -ENODEV;
397 return 0;
398 }
397 399
398 p += IVRS_HEADER_LENGTH; 400 p += IVRS_HEADER_LENGTH;
399 401
@@ -436,7 +438,7 @@ static u8 * __init alloc_command_buffer(struct amd_iommu *iommu)
436 if (cmd_buf == NULL) 438 if (cmd_buf == NULL)
437 return NULL; 439 return NULL;
438 440
439 iommu->cmd_buf_size = CMD_BUFFER_SIZE; 441 iommu->cmd_buf_size = CMD_BUFFER_SIZE | CMD_BUFFER_UNINITIALIZED;
440 442
441 return cmd_buf; 443 return cmd_buf;
442} 444}
@@ -472,12 +474,13 @@ static void iommu_enable_command_buffer(struct amd_iommu *iommu)
472 &entry, sizeof(entry)); 474 &entry, sizeof(entry));
473 475
474 amd_iommu_reset_cmd_buffer(iommu); 476 amd_iommu_reset_cmd_buffer(iommu);
477 iommu->cmd_buf_size &= ~(CMD_BUFFER_UNINITIALIZED);
475} 478}
476 479
477static void __init free_command_buffer(struct amd_iommu *iommu) 480static void __init free_command_buffer(struct amd_iommu *iommu)
478{ 481{
479 free_pages((unsigned long)iommu->cmd_buf, 482 free_pages((unsigned long)iommu->cmd_buf,
480 get_order(iommu->cmd_buf_size)); 483 get_order(iommu->cmd_buf_size & ~(CMD_BUFFER_UNINITIALIZED)));
481} 484}
482 485
483/* allocates the memory where the IOMMU will log its events to */ 486/* allocates the memory where the IOMMU will log its events to */
@@ -920,11 +923,16 @@ static int __init init_iommu_all(struct acpi_table_header *table)
920 h->mmio_phys); 923 h->mmio_phys);
921 924
922 iommu = kzalloc(sizeof(struct amd_iommu), GFP_KERNEL); 925 iommu = kzalloc(sizeof(struct amd_iommu), GFP_KERNEL);
923 if (iommu == NULL) 926 if (iommu == NULL) {
924 return -ENOMEM; 927 amd_iommu_init_err = -ENOMEM;
928 return 0;
929 }
930
925 ret = init_iommu_one(iommu, h); 931 ret = init_iommu_one(iommu, h);
926 if (ret) 932 if (ret) {
927 return ret; 933 amd_iommu_init_err = ret;
934 return 0;
935 }
928 break; 936 break;
929 default: 937 default:
930 break; 938 break;
@@ -934,8 +942,6 @@ static int __init init_iommu_all(struct acpi_table_header *table)
934 } 942 }
935 WARN_ON(p != end); 943 WARN_ON(p != end);
936 944
937 amd_iommu_initialized = true;
938
939 return 0; 945 return 0;
940} 946}
941 947
@@ -1211,6 +1217,10 @@ static int __init amd_iommu_init(void)
1211 if (acpi_table_parse("IVRS", find_last_devid_acpi) != 0) 1217 if (acpi_table_parse("IVRS", find_last_devid_acpi) != 0)
1212 return -ENODEV; 1218 return -ENODEV;
1213 1219
1220 ret = amd_iommu_init_err;
1221 if (ret)
1222 goto out;
1223
1214 dev_table_size = tbl_size(DEV_TABLE_ENTRY_SIZE); 1224 dev_table_size = tbl_size(DEV_TABLE_ENTRY_SIZE);
1215 alias_table_size = tbl_size(ALIAS_TABLE_ENTRY_SIZE); 1225 alias_table_size = tbl_size(ALIAS_TABLE_ENTRY_SIZE);
1216 rlookup_table_size = tbl_size(RLOOKUP_TABLE_ENTRY_SIZE); 1226 rlookup_table_size = tbl_size(RLOOKUP_TABLE_ENTRY_SIZE);
@@ -1270,12 +1280,19 @@ static int __init amd_iommu_init(void)
1270 if (acpi_table_parse("IVRS", init_iommu_all) != 0) 1280 if (acpi_table_parse("IVRS", init_iommu_all) != 0)
1271 goto free; 1281 goto free;
1272 1282
1273 if (!amd_iommu_initialized) 1283 if (amd_iommu_init_err) {
1284 ret = amd_iommu_init_err;
1274 goto free; 1285 goto free;
1286 }
1275 1287
1276 if (acpi_table_parse("IVRS", init_memory_definitions) != 0) 1288 if (acpi_table_parse("IVRS", init_memory_definitions) != 0)
1277 goto free; 1289 goto free;
1278 1290
1291 if (amd_iommu_init_err) {
1292 ret = amd_iommu_init_err;
1293 goto free;
1294 }
1295
1279 ret = sysdev_class_register(&amd_iommu_sysdev_class); 1296 ret = sysdev_class_register(&amd_iommu_sysdev_class);
1280 if (ret) 1297 if (ret)
1281 goto free; 1298 goto free;
@@ -1288,6 +1305,8 @@ static int __init amd_iommu_init(void)
1288 if (ret) 1305 if (ret)
1289 goto free; 1306 goto free;
1290 1307
1308 enable_iommus();
1309
1291 if (iommu_pass_through) 1310 if (iommu_pass_through)
1292 ret = amd_iommu_init_passthrough(); 1311 ret = amd_iommu_init_passthrough();
1293 else 1312 else
@@ -1300,8 +1319,6 @@ static int __init amd_iommu_init(void)
1300 1319
1301 amd_iommu_init_notifier(); 1320 amd_iommu_init_notifier();
1302 1321
1303 enable_iommus();
1304
1305 if (iommu_pass_through) 1322 if (iommu_pass_through)
1306 goto out; 1323 goto out;
1307 1324
@@ -1315,6 +1332,7 @@ out:
1315 return ret; 1332 return ret;
1316 1333
1317free: 1334free:
1335 disable_iommus();
1318 1336
1319 amd_iommu_uninit_devices(); 1337 amd_iommu_uninit_devices();
1320 1338
diff --git a/arch/x86/kernel/apb_timer.c b/arch/x86/kernel/apb_timer.c
index 4b7099526d2c..ff469e470059 100644
--- a/arch/x86/kernel/apb_timer.c
+++ b/arch/x86/kernel/apb_timer.c
@@ -33,6 +33,7 @@
33#include <linux/errno.h> 33#include <linux/errno.h>
34#include <linux/init.h> 34#include <linux/init.h>
35#include <linux/sysdev.h> 35#include <linux/sysdev.h>
36#include <linux/slab.h>
36#include <linux/pm.h> 37#include <linux/pm.h>
37#include <linux/pci.h> 38#include <linux/pci.h>
38#include <linux/sfi.h> 39#include <linux/sfi.h>
diff --git a/arch/x86/kernel/aperture_64.c b/arch/x86/kernel/aperture_64.c
index f147a95fd84a..b5d8b0bcf235 100644
--- a/arch/x86/kernel/aperture_64.c
+++ b/arch/x86/kernel/aperture_64.c
@@ -31,7 +31,6 @@
31#include <asm/x86_init.h> 31#include <asm/x86_init.h>
32 32
33int gart_iommu_aperture; 33int gart_iommu_aperture;
34EXPORT_SYMBOL_GPL(gart_iommu_aperture);
35int gart_iommu_aperture_disabled __initdata; 34int gart_iommu_aperture_disabled __initdata;
36int gart_iommu_aperture_allowed __initdata; 35int gart_iommu_aperture_allowed __initdata;
37 36
@@ -394,6 +393,7 @@ void __init gart_iommu_hole_init(void)
394 for (i = 0; i < ARRAY_SIZE(bus_dev_ranges); i++) { 393 for (i = 0; i < ARRAY_SIZE(bus_dev_ranges); i++) {
395 int bus; 394 int bus;
396 int dev_base, dev_limit; 395 int dev_base, dev_limit;
396 u32 ctl;
397 397
398 bus = bus_dev_ranges[i].bus; 398 bus = bus_dev_ranges[i].bus;
399 dev_base = bus_dev_ranges[i].dev_base; 399 dev_base = bus_dev_ranges[i].dev_base;
@@ -407,7 +407,19 @@ void __init gart_iommu_hole_init(void)
407 gart_iommu_aperture = 1; 407 gart_iommu_aperture = 1;
408 x86_init.iommu.iommu_init = gart_iommu_init; 408 x86_init.iommu.iommu_init = gart_iommu_init;
409 409
410 aper_order = (read_pci_config(bus, slot, 3, AMD64_GARTAPERTURECTL) >> 1) & 7; 410 ctl = read_pci_config(bus, slot, 3,
411 AMD64_GARTAPERTURECTL);
412
413 /*
414 * Before we do anything else disable the GART. It may
415 * still be enabled if we boot into a crash-kernel here.
416 * Reconfiguring the GART while it is enabled could have
417 * unknown side-effects.
418 */
419 ctl &= ~GARTEN;
420 write_pci_config(bus, slot, 3, AMD64_GARTAPERTURECTL, ctl);
421
422 aper_order = (ctl >> 1) & 7;
411 aper_size = (32 * 1024 * 1024) << aper_order; 423 aper_size = (32 * 1024 * 1024) << aper_order;
412 aper_base = read_pci_config(bus, slot, 3, AMD64_GARTAPERTUREBASE) & 0x7fff; 424 aper_base = read_pci_config(bus, slot, 3, AMD64_GARTAPERTUREBASE) & 0x7fff;
413 aper_base <<= 25; 425 aper_base <<= 25;
diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c
index 00187f1fcfb7..e5a4a1e01618 100644
--- a/arch/x86/kernel/apic/apic.c
+++ b/arch/x86/kernel/apic/apic.c
@@ -1640,8 +1640,10 @@ int __init APIC_init_uniprocessor(void)
1640 } 1640 }
1641#endif 1641#endif
1642 1642
1643#ifndef CONFIG_SMP
1643 enable_IR_x2apic(); 1644 enable_IR_x2apic();
1644 default_setup_apic_routing(); 1645 default_setup_apic_routing();
1646#endif
1645 1647
1646 verify_local_APIC(); 1648 verify_local_APIC();
1647 connect_bsp_APIC(); 1649 connect_bsp_APIC();
diff --git a/arch/x86/kernel/apic/es7000_32.c b/arch/x86/kernel/apic/es7000_32.c
index dd2b5f264643..03ba1b895f5e 100644
--- a/arch/x86/kernel/apic/es7000_32.c
+++ b/arch/x86/kernel/apic/es7000_32.c
@@ -42,6 +42,7 @@
42#include <linux/errno.h> 42#include <linux/errno.h>
43#include <linux/acpi.h> 43#include <linux/acpi.h>
44#include <linux/init.h> 44#include <linux/init.h>
45#include <linux/gfp.h>
45#include <linux/nmi.h> 46#include <linux/nmi.h>
46#include <linux/smp.h> 47#include <linux/smp.h>
47#include <linux/io.h> 48#include <linux/io.h>
diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c
index e4e0ddcb1546..127b8718abfb 100644
--- a/arch/x86/kernel/apic/io_apic.c
+++ b/arch/x86/kernel/apic/io_apic.c
@@ -36,6 +36,7 @@
36#include <linux/freezer.h> 36#include <linux/freezer.h>
37#include <linux/kthread.h> 37#include <linux/kthread.h>
38#include <linux/jiffies.h> /* time_after() */ 38#include <linux/jiffies.h> /* time_after() */
39#include <linux/slab.h>
39#ifdef CONFIG_ACPI 40#ifdef CONFIG_ACPI
40#include <acpi/acpi_bus.h> 41#include <acpi/acpi_bus.h>
41#endif 42#endif
@@ -1268,6 +1269,14 @@ void __setup_vector_irq(int cpu)
1268 /* Mark the inuse vectors */ 1269 /* Mark the inuse vectors */
1269 for_each_irq_desc(irq, desc) { 1270 for_each_irq_desc(irq, desc) {
1270 cfg = desc->chip_data; 1271 cfg = desc->chip_data;
1272
1273 /*
1274 * If it is a legacy IRQ handled by the legacy PIC, this cpu
1275 * will be part of the irq_cfg's domain.
1276 */
1277 if (irq < legacy_pic->nr_legacy_irqs && !IO_APIC_IRQ(irq))
1278 cpumask_set_cpu(cpu, cfg->domain);
1279
1271 if (!cpumask_test_cpu(cpu, cfg->domain)) 1280 if (!cpumask_test_cpu(cpu, cfg->domain))
1272 continue; 1281 continue;
1273 vector = cfg->vector; 1282 vector = cfg->vector;
diff --git a/arch/x86/kernel/apic/nmi.c b/arch/x86/kernel/apic/nmi.c
index 8aa65adbd25d..1edaf15c0b8e 100644
--- a/arch/x86/kernel/apic/nmi.c
+++ b/arch/x86/kernel/apic/nmi.c
@@ -18,6 +18,7 @@
18#include <linux/delay.h> 18#include <linux/delay.h>
19#include <linux/interrupt.h> 19#include <linux/interrupt.h>
20#include <linux/module.h> 20#include <linux/module.h>
21#include <linux/slab.h>
21#include <linux/sysdev.h> 22#include <linux/sysdev.h>
22#include <linux/sysctl.h> 23#include <linux/sysctl.h>
23#include <linux/percpu.h> 24#include <linux/percpu.h>
diff --git a/arch/x86/kernel/apic/x2apic_uv_x.c b/arch/x86/kernel/apic/x2apic_uv_x.c
index 3740c8a4eae7..c085d52dbaf2 100644
--- a/arch/x86/kernel/apic/x2apic_uv_x.c
+++ b/arch/x86/kernel/apic/x2apic_uv_x.c
@@ -17,6 +17,7 @@
17#include <linux/ctype.h> 17#include <linux/ctype.h>
18#include <linux/sched.h> 18#include <linux/sched.h>
19#include <linux/timer.h> 19#include <linux/timer.h>
20#include <linux/slab.h>
20#include <linux/cpu.h> 21#include <linux/cpu.h>
21#include <linux/init.h> 22#include <linux/init.h>
22#include <linux/io.h> 23#include <linux/io.h>
@@ -120,11 +121,9 @@ EXPORT_SYMBOL_GPL(uv_possible_blades);
120unsigned long sn_rtc_cycles_per_second; 121unsigned long sn_rtc_cycles_per_second;
121EXPORT_SYMBOL(sn_rtc_cycles_per_second); 122EXPORT_SYMBOL(sn_rtc_cycles_per_second);
122 123
123/* Start with all IRQs pointing to boot CPU. IRQ balancing will shift them. */
124
125static const struct cpumask *uv_target_cpus(void) 124static const struct cpumask *uv_target_cpus(void)
126{ 125{
127 return cpumask_of(0); 126 return cpu_online_mask;
128} 127}
129 128
130static void uv_vector_allocation_domain(int cpu, struct cpumask *retmask) 129static void uv_vector_allocation_domain(int cpu, struct cpumask *retmask)
diff --git a/arch/x86/kernel/bootflag.c b/arch/x86/kernel/bootflag.c
index 30f25a75fe28..5de7f4c56971 100644
--- a/arch/x86/kernel/bootflag.c
+++ b/arch/x86/kernel/bootflag.c
@@ -5,7 +5,6 @@
5#include <linux/kernel.h> 5#include <linux/kernel.h>
6#include <linux/init.h> 6#include <linux/init.h>
7#include <linux/string.h> 7#include <linux/string.h>
8#include <linux/slab.h>
9#include <linux/spinlock.h> 8#include <linux/spinlock.h>
10#include <linux/acpi.h> 9#include <linux/acpi.h>
11#include <asm/io.h> 10#include <asm/io.h>
diff --git a/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c b/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c
index 1b1920fa7c80..459168083b77 100644
--- a/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c
+++ b/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c
@@ -33,6 +33,7 @@
33#include <linux/cpufreq.h> 33#include <linux/cpufreq.h>
34#include <linux/compiler.h> 34#include <linux/compiler.h>
35#include <linux/dmi.h> 35#include <linux/dmi.h>
36#include <linux/slab.h>
36#include <trace/events/power.h> 37#include <trace/events/power.h>
37 38
38#include <linux/acpi.h> 39#include <linux/acpi.h>
diff --git a/arch/x86/kernel/cpu/cpufreq/elanfreq.c b/arch/x86/kernel/cpu/cpufreq/elanfreq.c
index 006b278b0d5d..c587db472a75 100644
--- a/arch/x86/kernel/cpu/cpufreq/elanfreq.c
+++ b/arch/x86/kernel/cpu/cpufreq/elanfreq.c
@@ -20,7 +20,6 @@
20#include <linux/module.h> 20#include <linux/module.h>
21#include <linux/init.h> 21#include <linux/init.h>
22 22
23#include <linux/slab.h>
24#include <linux/delay.h> 23#include <linux/delay.h>
25#include <linux/cpufreq.h> 24#include <linux/cpufreq.h>
26 25
diff --git a/arch/x86/kernel/cpu/cpufreq/gx-suspmod.c b/arch/x86/kernel/cpu/cpufreq/gx-suspmod.c
index ac27ec2264d5..16e3483be9e3 100644
--- a/arch/x86/kernel/cpu/cpufreq/gx-suspmod.c
+++ b/arch/x86/kernel/cpu/cpufreq/gx-suspmod.c
@@ -80,6 +80,7 @@
80#include <linux/cpufreq.h> 80#include <linux/cpufreq.h>
81#include <linux/pci.h> 81#include <linux/pci.h>
82#include <linux/errno.h> 82#include <linux/errno.h>
83#include <linux/slab.h>
83 84
84#include <asm/processor-cyrix.h> 85#include <asm/processor-cyrix.h>
85 86
diff --git a/arch/x86/kernel/cpu/cpufreq/longrun.c b/arch/x86/kernel/cpu/cpufreq/longrun.c
index da5f70fcb766..e7b559d74c52 100644
--- a/arch/x86/kernel/cpu/cpufreq/longrun.c
+++ b/arch/x86/kernel/cpu/cpufreq/longrun.c
@@ -9,7 +9,6 @@
9#include <linux/kernel.h> 9#include <linux/kernel.h>
10#include <linux/module.h> 10#include <linux/module.h>
11#include <linux/init.h> 11#include <linux/init.h>
12#include <linux/slab.h>
13#include <linux/cpufreq.h> 12#include <linux/cpufreq.h>
14#include <linux/timex.h> 13#include <linux/timex.h>
15 14
diff --git a/arch/x86/kernel/cpu/cpufreq/p4-clockmod.c b/arch/x86/kernel/cpu/cpufreq/p4-clockmod.c
index 869615193720..7b8a8ba67b07 100644
--- a/arch/x86/kernel/cpu/cpufreq/p4-clockmod.c
+++ b/arch/x86/kernel/cpu/cpufreq/p4-clockmod.c
@@ -25,7 +25,6 @@
25#include <linux/init.h> 25#include <linux/init.h>
26#include <linux/smp.h> 26#include <linux/smp.h>
27#include <linux/cpufreq.h> 27#include <linux/cpufreq.h>
28#include <linux/slab.h>
29#include <linux/cpumask.h> 28#include <linux/cpumask.h>
30#include <linux/timex.h> 29#include <linux/timex.h>
31 30
diff --git a/arch/x86/kernel/cpu/cpufreq/pcc-cpufreq.c b/arch/x86/kernel/cpu/cpufreq/pcc-cpufreq.c
index ff36d2979a90..ce7cde713e71 100644
--- a/arch/x86/kernel/cpu/cpufreq/pcc-cpufreq.c
+++ b/arch/x86/kernel/cpu/cpufreq/pcc-cpufreq.c
@@ -30,6 +30,7 @@
30#include <linux/sched.h> 30#include <linux/sched.h>
31#include <linux/cpufreq.h> 31#include <linux/cpufreq.h>
32#include <linux/compiler.h> 32#include <linux/compiler.h>
33#include <linux/slab.h>
33 34
34#include <linux/acpi.h> 35#include <linux/acpi.h>
35#include <linux/io.h> 36#include <linux/io.h>
diff --git a/arch/x86/kernel/cpu/cpufreq/powernow-k6.c b/arch/x86/kernel/cpu/cpufreq/powernow-k6.c
index cb01dac267d3..b3379d6a5c57 100644
--- a/arch/x86/kernel/cpu/cpufreq/powernow-k6.c
+++ b/arch/x86/kernel/cpu/cpufreq/powernow-k6.c
@@ -13,7 +13,6 @@
13#include <linux/init.h> 13#include <linux/init.h>
14#include <linux/cpufreq.h> 14#include <linux/cpufreq.h>
15#include <linux/ioport.h> 15#include <linux/ioport.h>
16#include <linux/slab.h>
17#include <linux/timex.h> 16#include <linux/timex.h>
18#include <linux/io.h> 17#include <linux/io.h>
19 18
diff --git a/arch/x86/kernel/cpu/cpufreq/speedstep-centrino.c b/arch/x86/kernel/cpu/cpufreq/speedstep-centrino.c
index 8d672ef162ce..9b1ff37de46a 100644
--- a/arch/x86/kernel/cpu/cpufreq/speedstep-centrino.c
+++ b/arch/x86/kernel/cpu/cpufreq/speedstep-centrino.c
@@ -20,6 +20,7 @@
20#include <linux/sched.h> /* current */ 20#include <linux/sched.h> /* current */
21#include <linux/delay.h> 21#include <linux/delay.h>
22#include <linux/compiler.h> 22#include <linux/compiler.h>
23#include <linux/gfp.h>
23 24
24#include <asm/msr.h> 25#include <asm/msr.h>
25#include <asm/processor.h> 26#include <asm/processor.h>
diff --git a/arch/x86/kernel/cpu/cpufreq/speedstep-ich.c b/arch/x86/kernel/cpu/cpufreq/speedstep-ich.c
index 2ce8e0b5cc54..561758e95180 100644
--- a/arch/x86/kernel/cpu/cpufreq/speedstep-ich.c
+++ b/arch/x86/kernel/cpu/cpufreq/speedstep-ich.c
@@ -23,7 +23,6 @@
23#include <linux/init.h> 23#include <linux/init.h>
24#include <linux/cpufreq.h> 24#include <linux/cpufreq.h>
25#include <linux/pci.h> 25#include <linux/pci.h>
26#include <linux/slab.h>
27#include <linux/sched.h> 26#include <linux/sched.h>
28 27
29#include "speedstep-lib.h" 28#include "speedstep-lib.h"
diff --git a/arch/x86/kernel/cpu/cpufreq/speedstep-lib.c b/arch/x86/kernel/cpu/cpufreq/speedstep-lib.c
index ad0083abfa23..a94ec6be69fa 100644
--- a/arch/x86/kernel/cpu/cpufreq/speedstep-lib.c
+++ b/arch/x86/kernel/cpu/cpufreq/speedstep-lib.c
@@ -13,7 +13,6 @@
13#include <linux/moduleparam.h> 13#include <linux/moduleparam.h>
14#include <linux/init.h> 14#include <linux/init.h>
15#include <linux/cpufreq.h> 15#include <linux/cpufreq.h>
16#include <linux/slab.h>
17 16
18#include <asm/msr.h> 17#include <asm/msr.h>
19#include <asm/tsc.h> 18#include <asm/tsc.h>
diff --git a/arch/x86/kernel/cpu/cpufreq/speedstep-smi.c b/arch/x86/kernel/cpu/cpufreq/speedstep-smi.c
index 04d73c114e49..8abd869baabf 100644
--- a/arch/x86/kernel/cpu/cpufreq/speedstep-smi.c
+++ b/arch/x86/kernel/cpu/cpufreq/speedstep-smi.c
@@ -17,7 +17,6 @@
17#include <linux/moduleparam.h> 17#include <linux/moduleparam.h>
18#include <linux/init.h> 18#include <linux/init.h>
19#include <linux/cpufreq.h> 19#include <linux/cpufreq.h>
20#include <linux/slab.h>
21#include <linux/delay.h> 20#include <linux/delay.h>
22#include <linux/io.h> 21#include <linux/io.h>
23#include <asm/ist.h> 22#include <asm/ist.h>
diff --git a/arch/x86/kernel/cpu/intel.c b/arch/x86/kernel/cpu/intel.c
index 879666f4d871..7e1cca13af35 100644
--- a/arch/x86/kernel/cpu/intel.c
+++ b/arch/x86/kernel/cpu/intel.c
@@ -70,7 +70,8 @@ static void __cpuinit early_init_intel(struct cpuinfo_x86 *c)
70 if (c->x86_power & (1 << 8)) { 70 if (c->x86_power & (1 << 8)) {
71 set_cpu_cap(c, X86_FEATURE_CONSTANT_TSC); 71 set_cpu_cap(c, X86_FEATURE_CONSTANT_TSC);
72 set_cpu_cap(c, X86_FEATURE_NONSTOP_TSC); 72 set_cpu_cap(c, X86_FEATURE_NONSTOP_TSC);
73 sched_clock_stable = 1; 73 if (!check_tsc_unstable())
74 sched_clock_stable = 1;
74 } 75 }
75 76
76 /* 77 /*
diff --git a/arch/x86/kernel/cpu/mcheck/mce-inject.c b/arch/x86/kernel/cpu/mcheck/mce-inject.c
index 73734baa50f2..e7dbde7bfedb 100644
--- a/arch/x86/kernel/cpu/mcheck/mce-inject.c
+++ b/arch/x86/kernel/cpu/mcheck/mce-inject.c
@@ -22,6 +22,7 @@
22#include <linux/kdebug.h> 22#include <linux/kdebug.h>
23#include <linux/cpu.h> 23#include <linux/cpu.h>
24#include <linux/sched.h> 24#include <linux/sched.h>
25#include <linux/gfp.h>
25#include <asm/mce.h> 26#include <asm/mce.h>
26#include <asm/apic.h> 27#include <asm/apic.h>
27 28
diff --git a/arch/x86/kernel/cpu/mcheck/mce.c b/arch/x86/kernel/cpu/mcheck/mce.c
index 28cba46bf32c..8a6f0afa767e 100644
--- a/arch/x86/kernel/cpu/mcheck/mce.c
+++ b/arch/x86/kernel/cpu/mcheck/mce.c
@@ -26,6 +26,7 @@
26#include <linux/sched.h> 26#include <linux/sched.h>
27#include <linux/sysfs.h> 27#include <linux/sysfs.h>
28#include <linux/types.h> 28#include <linux/types.h>
29#include <linux/slab.h>
29#include <linux/init.h> 30#include <linux/init.h>
30#include <linux/kmod.h> 31#include <linux/kmod.h>
31#include <linux/poll.h> 32#include <linux/poll.h>
@@ -46,6 +47,13 @@
46 47
47#include "mce-internal.h" 48#include "mce-internal.h"
48 49
50static DEFINE_MUTEX(mce_read_mutex);
51
52#define rcu_dereference_check_mce(p) \
53 rcu_dereference_check((p), \
54 rcu_read_lock_sched_held() || \
55 lockdep_is_held(&mce_read_mutex))
56
49#define CREATE_TRACE_POINTS 57#define CREATE_TRACE_POINTS
50#include <trace/events/mce.h> 58#include <trace/events/mce.h>
51 59
@@ -158,7 +166,7 @@ void mce_log(struct mce *mce)
158 mce->finished = 0; 166 mce->finished = 0;
159 wmb(); 167 wmb();
160 for (;;) { 168 for (;;) {
161 entry = rcu_dereference(mcelog.next); 169 entry = rcu_dereference_check_mce(mcelog.next);
162 for (;;) { 170 for (;;) {
163 /* 171 /*
164 * When the buffer fills up discard new entries. 172 * When the buffer fills up discard new entries.
@@ -1485,8 +1493,6 @@ static void collect_tscs(void *data)
1485 rdtscll(cpu_tsc[smp_processor_id()]); 1493 rdtscll(cpu_tsc[smp_processor_id()]);
1486} 1494}
1487 1495
1488static DEFINE_MUTEX(mce_read_mutex);
1489
1490static ssize_t mce_read(struct file *filp, char __user *ubuf, size_t usize, 1496static ssize_t mce_read(struct file *filp, char __user *ubuf, size_t usize,
1491 loff_t *off) 1497 loff_t *off)
1492{ 1498{
@@ -1500,7 +1506,7 @@ static ssize_t mce_read(struct file *filp, char __user *ubuf, size_t usize,
1500 return -ENOMEM; 1506 return -ENOMEM;
1501 1507
1502 mutex_lock(&mce_read_mutex); 1508 mutex_lock(&mce_read_mutex);
1503 next = rcu_dereference(mcelog.next); 1509 next = rcu_dereference_check_mce(mcelog.next);
1504 1510
1505 /* Only supports full reads right now */ 1511 /* Only supports full reads right now */
1506 if (*off != 0 || usize < MCE_LOG_LEN*sizeof(struct mce)) { 1512 if (*off != 0 || usize < MCE_LOG_LEN*sizeof(struct mce)) {
@@ -1565,7 +1571,7 @@ timeout:
1565static unsigned int mce_poll(struct file *file, poll_table *wait) 1571static unsigned int mce_poll(struct file *file, poll_table *wait)
1566{ 1572{
1567 poll_wait(file, &mce_wait, wait); 1573 poll_wait(file, &mce_wait, wait);
1568 if (rcu_dereference(mcelog.next)) 1574 if (rcu_dereference_check_mce(mcelog.next))
1569 return POLLIN | POLLRDNORM; 1575 return POLLIN | POLLRDNORM;
1570 return 0; 1576 return 0;
1571} 1577}
diff --git a/arch/x86/kernel/cpu/mcheck/mce_amd.c b/arch/x86/kernel/cpu/mcheck/mce_amd.c
index cda932ca3ade..224392d8fe8c 100644
--- a/arch/x86/kernel/cpu/mcheck/mce_amd.c
+++ b/arch/x86/kernel/cpu/mcheck/mce_amd.c
@@ -21,6 +21,7 @@
21#include <linux/errno.h> 21#include <linux/errno.h>
22#include <linux/sched.h> 22#include <linux/sched.h>
23#include <linux/sysfs.h> 23#include <linux/sysfs.h>
24#include <linux/slab.h>
24#include <linux/init.h> 25#include <linux/init.h>
25#include <linux/cpu.h> 26#include <linux/cpu.h>
26#include <linux/smp.h> 27#include <linux/smp.h>
diff --git a/arch/x86/kernel/cpu/mcheck/mce_intel.c b/arch/x86/kernel/cpu/mcheck/mce_intel.c
index 7c785634af2b..62b48e40920a 100644
--- a/arch/x86/kernel/cpu/mcheck/mce_intel.c
+++ b/arch/x86/kernel/cpu/mcheck/mce_intel.c
@@ -5,6 +5,7 @@
5 * Author: Andi Kleen 5 * Author: Andi Kleen
6 */ 6 */
7 7
8#include <linux/gfp.h>
8#include <linux/init.h> 9#include <linux/init.h>
9#include <linux/interrupt.h> 10#include <linux/interrupt.h>
10#include <linux/percpu.h> 11#include <linux/percpu.h>
@@ -95,7 +96,7 @@ static void cmci_discover(int banks, int boot)
95 96
96 /* Already owned by someone else? */ 97 /* Already owned by someone else? */
97 if (val & CMCI_EN) { 98 if (val & CMCI_EN) {
98 if (test_and_clear_bit(i, owned) || boot) 99 if (test_and_clear_bit(i, owned) && !boot)
99 print_update("SHD", &hdr, i); 100 print_update("SHD", &hdr, i);
100 __clear_bit(i, __get_cpu_var(mce_poll_banks)); 101 __clear_bit(i, __get_cpu_var(mce_poll_banks));
101 continue; 102 continue;
@@ -107,7 +108,7 @@ static void cmci_discover(int banks, int boot)
107 108
108 /* Did the enable bit stick? -- the bank supports CMCI */ 109 /* Did the enable bit stick? -- the bank supports CMCI */
109 if (val & CMCI_EN) { 110 if (val & CMCI_EN) {
110 if (!test_and_set_bit(i, owned) || boot) 111 if (!test_and_set_bit(i, owned) && !boot)
111 print_update("CMCI", &hdr, i); 112 print_update("CMCI", &hdr, i);
112 __clear_bit(i, __get_cpu_var(mce_poll_banks)); 113 __clear_bit(i, __get_cpu_var(mce_poll_banks));
113 } else { 114 } else {
diff --git a/arch/x86/kernel/cpu/mtrr/generic.c b/arch/x86/kernel/cpu/mtrr/generic.c
index 9aa5dc76ff4a..fd31a441c61c 100644
--- a/arch/x86/kernel/cpu/mtrr/generic.c
+++ b/arch/x86/kernel/cpu/mtrr/generic.c
@@ -6,7 +6,6 @@
6 6
7#include <linux/module.h> 7#include <linux/module.h>
8#include <linux/init.h> 8#include <linux/init.h>
9#include <linux/slab.h>
10#include <linux/io.h> 9#include <linux/io.h>
11#include <linux/mm.h> 10#include <linux/mm.h>
12 11
diff --git a/arch/x86/kernel/cpu/mtrr/if.c b/arch/x86/kernel/cpu/mtrr/if.c
index e006e56f699c..79289632cb27 100644
--- a/arch/x86/kernel/cpu/mtrr/if.c
+++ b/arch/x86/kernel/cpu/mtrr/if.c
@@ -5,6 +5,7 @@
5#include <linux/module.h> 5#include <linux/module.h>
6#include <linux/ctype.h> 6#include <linux/ctype.h>
7#include <linux/string.h> 7#include <linux/string.h>
8#include <linux/slab.h>
8#include <linux/init.h> 9#include <linux/init.h>
9 10
10#define LINE_SIZE 80 11#define LINE_SIZE 80
diff --git a/arch/x86/kernel/cpu/perf_event.c b/arch/x86/kernel/cpu/perf_event.c
index b1fbdeecf6c9..db5bdc8addf8 100644
--- a/arch/x86/kernel/cpu/perf_event.c
+++ b/arch/x86/kernel/cpu/perf_event.c
@@ -21,6 +21,7 @@
21#include <linux/kdebug.h> 21#include <linux/kdebug.h>
22#include <linux/sched.h> 22#include <linux/sched.h>
23#include <linux/uaccess.h> 23#include <linux/uaccess.h>
24#include <linux/slab.h>
24#include <linux/highmem.h> 25#include <linux/highmem.h>
25#include <linux/cpu.h> 26#include <linux/cpu.h>
26#include <linux/bitops.h> 27#include <linux/bitops.h>
@@ -28,6 +29,7 @@
28#include <asm/apic.h> 29#include <asm/apic.h>
29#include <asm/stacktrace.h> 30#include <asm/stacktrace.h>
30#include <asm/nmi.h> 31#include <asm/nmi.h>
32#include <asm/compat.h>
31 33
32static u64 perf_event_mask __read_mostly; 34static u64 perf_event_mask __read_mostly;
33 35
@@ -73,10 +75,10 @@ struct debug_store {
73struct event_constraint { 75struct event_constraint {
74 union { 76 union {
75 unsigned long idxmsk[BITS_TO_LONGS(X86_PMC_IDX_MAX)]; 77 unsigned long idxmsk[BITS_TO_LONGS(X86_PMC_IDX_MAX)];
76 u64 idxmsk64[1]; 78 u64 idxmsk64;
77 }; 79 };
78 int code; 80 u64 code;
79 int cmask; 81 u64 cmask;
80 int weight; 82 int weight;
81}; 83};
82 84
@@ -103,7 +105,7 @@ struct cpu_hw_events {
103}; 105};
104 106
105#define __EVENT_CONSTRAINT(c, n, m, w) {\ 107#define __EVENT_CONSTRAINT(c, n, m, w) {\
106 { .idxmsk64[0] = (n) }, \ 108 { .idxmsk64 = (n) }, \
107 .code = (c), \ 109 .code = (c), \
108 .cmask = (m), \ 110 .cmask = (m), \
109 .weight = (w), \ 111 .weight = (w), \
@@ -116,7 +118,7 @@ struct cpu_hw_events {
116 EVENT_CONSTRAINT(c, n, INTEL_ARCH_EVTSEL_MASK) 118 EVENT_CONSTRAINT(c, n, INTEL_ARCH_EVTSEL_MASK)
117 119
118#define FIXED_EVENT_CONSTRAINT(c, n) \ 120#define FIXED_EVENT_CONSTRAINT(c, n) \
119 EVENT_CONSTRAINT(c, n, INTEL_ARCH_FIXED_MASK) 121 EVENT_CONSTRAINT(c, (1ULL << (32+n)), INTEL_ARCH_FIXED_MASK)
120 122
121#define EVENT_CONSTRAINT_END \ 123#define EVENT_CONSTRAINT_END \
122 EVENT_CONSTRAINT(0, 0, 0) 124 EVENT_CONSTRAINT(0, 0, 0)
@@ -133,8 +135,8 @@ struct x86_pmu {
133 int (*handle_irq)(struct pt_regs *); 135 int (*handle_irq)(struct pt_regs *);
134 void (*disable_all)(void); 136 void (*disable_all)(void);
135 void (*enable_all)(void); 137 void (*enable_all)(void);
136 void (*enable)(struct hw_perf_event *, int); 138 void (*enable)(struct perf_event *);
137 void (*disable)(struct hw_perf_event *, int); 139 void (*disable)(struct perf_event *);
138 unsigned eventsel; 140 unsigned eventsel;
139 unsigned perfctr; 141 unsigned perfctr;
140 u64 (*event_map)(int); 142 u64 (*event_map)(int);
@@ -157,6 +159,11 @@ struct x86_pmu {
157 void (*put_event_constraints)(struct cpu_hw_events *cpuc, 159 void (*put_event_constraints)(struct cpu_hw_events *cpuc,
158 struct perf_event *event); 160 struct perf_event *event);
159 struct event_constraint *event_constraints; 161 struct event_constraint *event_constraints;
162
163 int (*cpu_prepare)(int cpu);
164 void (*cpu_starting)(int cpu);
165 void (*cpu_dying)(int cpu);
166 void (*cpu_dead)(int cpu);
160}; 167};
161 168
162static struct x86_pmu x86_pmu __read_mostly; 169static struct x86_pmu x86_pmu __read_mostly;
@@ -165,8 +172,7 @@ static DEFINE_PER_CPU(struct cpu_hw_events, cpu_hw_events) = {
165 .enabled = 1, 172 .enabled = 1,
166}; 173};
167 174
168static int x86_perf_event_set_period(struct perf_event *event, 175static int x86_perf_event_set_period(struct perf_event *event);
169 struct hw_perf_event *hwc, int idx);
170 176
171/* 177/*
172 * Generalized hw caching related hw_event table, filled 178 * Generalized hw caching related hw_event table, filled
@@ -189,11 +195,12 @@ static u64 __read_mostly hw_cache_event_ids
189 * Returns the delta events processed. 195 * Returns the delta events processed.
190 */ 196 */
191static u64 197static u64
192x86_perf_event_update(struct perf_event *event, 198x86_perf_event_update(struct perf_event *event)
193 struct hw_perf_event *hwc, int idx)
194{ 199{
200 struct hw_perf_event *hwc = &event->hw;
195 int shift = 64 - x86_pmu.event_bits; 201 int shift = 64 - x86_pmu.event_bits;
196 u64 prev_raw_count, new_raw_count; 202 u64 prev_raw_count, new_raw_count;
203 int idx = hwc->idx;
197 s64 delta; 204 s64 delta;
198 205
199 if (idx == X86_PMC_IDX_FIXED_BTS) 206 if (idx == X86_PMC_IDX_FIXED_BTS)
@@ -293,7 +300,7 @@ static inline bool bts_available(void)
293 return x86_pmu.enable_bts != NULL; 300 return x86_pmu.enable_bts != NULL;
294} 301}
295 302
296static inline void init_debug_store_on_cpu(int cpu) 303static void init_debug_store_on_cpu(int cpu)
297{ 304{
298 struct debug_store *ds = per_cpu(cpu_hw_events, cpu).ds; 305 struct debug_store *ds = per_cpu(cpu_hw_events, cpu).ds;
299 306
@@ -305,7 +312,7 @@ static inline void init_debug_store_on_cpu(int cpu)
305 (u32)((u64)(unsigned long)ds >> 32)); 312 (u32)((u64)(unsigned long)ds >> 32));
306} 313}
307 314
308static inline void fini_debug_store_on_cpu(int cpu) 315static void fini_debug_store_on_cpu(int cpu)
309{ 316{
310 if (!per_cpu(cpu_hw_events, cpu).ds) 317 if (!per_cpu(cpu_hw_events, cpu).ds)
311 return; 318 return;
@@ -503,6 +510,9 @@ static int __hw_perf_event_init(struct perf_event *event)
503 */ 510 */
504 if (attr->type == PERF_TYPE_RAW) { 511 if (attr->type == PERF_TYPE_RAW) {
505 hwc->config |= x86_pmu.raw_event(attr->config); 512 hwc->config |= x86_pmu.raw_event(attr->config);
513 if ((hwc->config & ARCH_PERFMON_EVENTSEL_ANY) &&
514 perf_paranoid_cpu() && !capable(CAP_SYS_ADMIN))
515 return -EACCES;
506 return 0; 516 return 0;
507 } 517 }
508 518
@@ -553,9 +563,9 @@ static void x86_pmu_disable_all(void)
553 if (!test_bit(idx, cpuc->active_mask)) 563 if (!test_bit(idx, cpuc->active_mask))
554 continue; 564 continue;
555 rdmsrl(x86_pmu.eventsel + idx, val); 565 rdmsrl(x86_pmu.eventsel + idx, val);
556 if (!(val & ARCH_PERFMON_EVENTSEL0_ENABLE)) 566 if (!(val & ARCH_PERFMON_EVENTSEL_ENABLE))
557 continue; 567 continue;
558 val &= ~ARCH_PERFMON_EVENTSEL0_ENABLE; 568 val &= ~ARCH_PERFMON_EVENTSEL_ENABLE;
559 wrmsrl(x86_pmu.eventsel + idx, val); 569 wrmsrl(x86_pmu.eventsel + idx, val);
560 } 570 }
561} 571}
@@ -590,7 +600,7 @@ static void x86_pmu_enable_all(void)
590 continue; 600 continue;
591 601
592 val = event->hw.config; 602 val = event->hw.config;
593 val |= ARCH_PERFMON_EVENTSEL0_ENABLE; 603 val |= ARCH_PERFMON_EVENTSEL_ENABLE;
594 wrmsrl(x86_pmu.eventsel + idx, val); 604 wrmsrl(x86_pmu.eventsel + idx, val);
595 } 605 }
596} 606}
@@ -612,8 +622,8 @@ static int x86_schedule_events(struct cpu_hw_events *cpuc, int n, int *assign)
612 bitmap_zero(used_mask, X86_PMC_IDX_MAX); 622 bitmap_zero(used_mask, X86_PMC_IDX_MAX);
613 623
614 for (i = 0; i < n; i++) { 624 for (i = 0; i < n; i++) {
615 constraints[i] = 625 c = x86_pmu.get_event_constraints(cpuc, cpuc->event_list[i]);
616 x86_pmu.get_event_constraints(cpuc, cpuc->event_list[i]); 626 constraints[i] = c;
617 } 627 }
618 628
619 /* 629 /*
@@ -635,7 +645,7 @@ static int x86_schedule_events(struct cpu_hw_events *cpuc, int n, int *assign)
635 if (test_bit(hwc->idx, used_mask)) 645 if (test_bit(hwc->idx, used_mask))
636 break; 646 break;
637 647
638 set_bit(hwc->idx, used_mask); 648 __set_bit(hwc->idx, used_mask);
639 if (assign) 649 if (assign)
640 assign[i] = hwc->idx; 650 assign[i] = hwc->idx;
641 } 651 }
@@ -684,7 +694,7 @@ static int x86_schedule_events(struct cpu_hw_events *cpuc, int n, int *assign)
684 if (j == X86_PMC_IDX_MAX) 694 if (j == X86_PMC_IDX_MAX)
685 break; 695 break;
686 696
687 set_bit(j, used_mask); 697 __set_bit(j, used_mask);
688 698
689 if (assign) 699 if (assign)
690 assign[i] = j; 700 assign[i] = j;
@@ -777,6 +787,7 @@ static inline int match_prev_assignment(struct hw_perf_event *hwc,
777 hwc->last_tag == cpuc->tags[i]; 787 hwc->last_tag == cpuc->tags[i];
778} 788}
779 789
790static int x86_pmu_start(struct perf_event *event);
780static void x86_pmu_stop(struct perf_event *event); 791static void x86_pmu_stop(struct perf_event *event);
781 792
782void hw_perf_enable(void) 793void hw_perf_enable(void)
@@ -793,6 +804,7 @@ void hw_perf_enable(void)
793 return; 804 return;
794 805
795 if (cpuc->n_added) { 806 if (cpuc->n_added) {
807 int n_running = cpuc->n_events - cpuc->n_added;
796 /* 808 /*
797 * apply assignment obtained either from 809 * apply assignment obtained either from
798 * hw_perf_group_sched_in() or x86_pmu_enable() 810 * hw_perf_group_sched_in() or x86_pmu_enable()
@@ -800,8 +812,7 @@ void hw_perf_enable(void)
800 * step1: save events moving to new counters 812 * step1: save events moving to new counters
801 * step2: reprogram moved events into new counters 813 * step2: reprogram moved events into new counters
802 */ 814 */
803 for (i = 0; i < cpuc->n_events; i++) { 815 for (i = 0; i < n_running; i++) {
804
805 event = cpuc->event_list[i]; 816 event = cpuc->event_list[i];
806 hwc = &event->hw; 817 hwc = &event->hw;
807 818
@@ -816,29 +827,18 @@ void hw_perf_enable(void)
816 continue; 827 continue;
817 828
818 x86_pmu_stop(event); 829 x86_pmu_stop(event);
819
820 hwc->idx = -1;
821 } 830 }
822 831
823 for (i = 0; i < cpuc->n_events; i++) { 832 for (i = 0; i < cpuc->n_events; i++) {
824
825 event = cpuc->event_list[i]; 833 event = cpuc->event_list[i];
826 hwc = &event->hw; 834 hwc = &event->hw;
827 835
828 if (hwc->idx == -1) { 836 if (!match_prev_assignment(hwc, cpuc, i))
829 x86_assign_hw_event(event, cpuc, i); 837 x86_assign_hw_event(event, cpuc, i);
830 x86_perf_event_set_period(event, hwc, hwc->idx); 838 else if (i < n_running)
831 } 839 continue;
832 /*
833 * need to mark as active because x86_pmu_disable()
834 * clear active_mask and events[] yet it preserves
835 * idx
836 */
837 set_bit(hwc->idx, cpuc->active_mask);
838 cpuc->events[hwc->idx] = event;
839 840
840 x86_pmu.enable(hwc, hwc->idx); 841 x86_pmu_start(event);
841 perf_event_update_userpage(event);
842 } 842 }
843 cpuc->n_added = 0; 843 cpuc->n_added = 0;
844 perf_events_lapic_init(); 844 perf_events_lapic_init();
@@ -850,15 +850,16 @@ void hw_perf_enable(void)
850 x86_pmu.enable_all(); 850 x86_pmu.enable_all();
851} 851}
852 852
853static inline void __x86_pmu_enable_event(struct hw_perf_event *hwc, int idx) 853static inline void __x86_pmu_enable_event(struct hw_perf_event *hwc)
854{ 854{
855 (void)checking_wrmsrl(hwc->config_base + idx, 855 (void)checking_wrmsrl(hwc->config_base + hwc->idx,
856 hwc->config | ARCH_PERFMON_EVENTSEL0_ENABLE); 856 hwc->config | ARCH_PERFMON_EVENTSEL_ENABLE);
857} 857}
858 858
859static inline void x86_pmu_disable_event(struct hw_perf_event *hwc, int idx) 859static inline void x86_pmu_disable_event(struct perf_event *event)
860{ 860{
861 (void)checking_wrmsrl(hwc->config_base + idx, hwc->config); 861 struct hw_perf_event *hwc = &event->hw;
862 (void)checking_wrmsrl(hwc->config_base + hwc->idx, hwc->config);
862} 863}
863 864
864static DEFINE_PER_CPU(u64 [X86_PMC_IDX_MAX], pmc_prev_left); 865static DEFINE_PER_CPU(u64 [X86_PMC_IDX_MAX], pmc_prev_left);
@@ -868,12 +869,12 @@ static DEFINE_PER_CPU(u64 [X86_PMC_IDX_MAX], pmc_prev_left);
868 * To be called with the event disabled in hw: 869 * To be called with the event disabled in hw:
869 */ 870 */
870static int 871static int
871x86_perf_event_set_period(struct perf_event *event, 872x86_perf_event_set_period(struct perf_event *event)
872 struct hw_perf_event *hwc, int idx)
873{ 873{
874 struct hw_perf_event *hwc = &event->hw;
874 s64 left = atomic64_read(&hwc->period_left); 875 s64 left = atomic64_read(&hwc->period_left);
875 s64 period = hwc->sample_period; 876 s64 period = hwc->sample_period;
876 int err, ret = 0; 877 int err, ret = 0, idx = hwc->idx;
877 878
878 if (idx == X86_PMC_IDX_FIXED_BTS) 879 if (idx == X86_PMC_IDX_FIXED_BTS)
879 return 0; 880 return 0;
@@ -919,11 +920,11 @@ x86_perf_event_set_period(struct perf_event *event,
919 return ret; 920 return ret;
920} 921}
921 922
922static void x86_pmu_enable_event(struct hw_perf_event *hwc, int idx) 923static void x86_pmu_enable_event(struct perf_event *event)
923{ 924{
924 struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); 925 struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
925 if (cpuc->enabled) 926 if (cpuc->enabled)
926 __x86_pmu_enable_event(hwc, idx); 927 __x86_pmu_enable_event(&event->hw);
927} 928}
928 929
929/* 930/*
@@ -959,34 +960,32 @@ static int x86_pmu_enable(struct perf_event *event)
959 memcpy(cpuc->assign, assign, n*sizeof(int)); 960 memcpy(cpuc->assign, assign, n*sizeof(int));
960 961
961 cpuc->n_events = n; 962 cpuc->n_events = n;
962 cpuc->n_added = n - n0; 963 cpuc->n_added += n - n0;
963 964
964 return 0; 965 return 0;
965} 966}
966 967
967static int x86_pmu_start(struct perf_event *event) 968static int x86_pmu_start(struct perf_event *event)
968{ 969{
969 struct hw_perf_event *hwc = &event->hw; 970 struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
971 int idx = event->hw.idx;
970 972
971 if (hwc->idx == -1) 973 if (idx == -1)
972 return -EAGAIN; 974 return -EAGAIN;
973 975
974 x86_perf_event_set_period(event, hwc, hwc->idx); 976 x86_perf_event_set_period(event);
975 x86_pmu.enable(hwc, hwc->idx); 977 cpuc->events[idx] = event;
978 __set_bit(idx, cpuc->active_mask);
979 x86_pmu.enable(event);
980 perf_event_update_userpage(event);
976 981
977 return 0; 982 return 0;
978} 983}
979 984
980static void x86_pmu_unthrottle(struct perf_event *event) 985static void x86_pmu_unthrottle(struct perf_event *event)
981{ 986{
982 struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); 987 int ret = x86_pmu_start(event);
983 struct hw_perf_event *hwc = &event->hw; 988 WARN_ON_ONCE(ret);
984
985 if (WARN_ON_ONCE(hwc->idx >= X86_PMC_IDX_MAX ||
986 cpuc->events[hwc->idx] != event))
987 return;
988
989 x86_pmu.enable(hwc, hwc->idx);
990} 989}
991 990
992void perf_event_print_debug(void) 991void perf_event_print_debug(void)
@@ -1046,18 +1045,16 @@ static void x86_pmu_stop(struct perf_event *event)
1046 struct hw_perf_event *hwc = &event->hw; 1045 struct hw_perf_event *hwc = &event->hw;
1047 int idx = hwc->idx; 1046 int idx = hwc->idx;
1048 1047
1049 /* 1048 if (!__test_and_clear_bit(idx, cpuc->active_mask))
1050 * Must be done before we disable, otherwise the nmi handler 1049 return;
1051 * could reenable again: 1050
1052 */ 1051 x86_pmu.disable(event);
1053 clear_bit(idx, cpuc->active_mask);
1054 x86_pmu.disable(hwc, idx);
1055 1052
1056 /* 1053 /*
1057 * Drain the remaining delta count out of a event 1054 * Drain the remaining delta count out of a event
1058 * that we are disabling: 1055 * that we are disabling:
1059 */ 1056 */
1060 x86_perf_event_update(event, hwc, idx); 1057 x86_perf_event_update(event);
1061 1058
1062 cpuc->events[idx] = NULL; 1059 cpuc->events[idx] = NULL;
1063} 1060}
@@ -1094,8 +1091,7 @@ static int x86_pmu_handle_irq(struct pt_regs *regs)
1094 int idx, handled = 0; 1091 int idx, handled = 0;
1095 u64 val; 1092 u64 val;
1096 1093
1097 data.addr = 0; 1094 perf_sample_data_init(&data, 0);
1098 data.raw = NULL;
1099 1095
1100 cpuc = &__get_cpu_var(cpu_hw_events); 1096 cpuc = &__get_cpu_var(cpu_hw_events);
1101 1097
@@ -1106,7 +1102,7 @@ static int x86_pmu_handle_irq(struct pt_regs *regs)
1106 event = cpuc->events[idx]; 1102 event = cpuc->events[idx];
1107 hwc = &event->hw; 1103 hwc = &event->hw;
1108 1104
1109 val = x86_perf_event_update(event, hwc, idx); 1105 val = x86_perf_event_update(event);
1110 if (val & (1ULL << (x86_pmu.event_bits - 1))) 1106 if (val & (1ULL << (x86_pmu.event_bits - 1)))
1111 continue; 1107 continue;
1112 1108
@@ -1116,11 +1112,11 @@ static int x86_pmu_handle_irq(struct pt_regs *regs)
1116 handled = 1; 1112 handled = 1;
1117 data.period = event->hw.last_period; 1113 data.period = event->hw.last_period;
1118 1114
1119 if (!x86_perf_event_set_period(event, hwc, idx)) 1115 if (!x86_perf_event_set_period(event))
1120 continue; 1116 continue;
1121 1117
1122 if (perf_event_overflow(event, 1, &data, regs)) 1118 if (perf_event_overflow(event, 1, &data, regs))
1123 x86_pmu.disable(hwc, idx); 1119 x86_pmu_stop(event);
1124 } 1120 }
1125 1121
1126 if (handled) 1122 if (handled)
@@ -1307,7 +1303,7 @@ int hw_perf_group_sched_in(struct perf_event *leader,
1307 memcpy(cpuc->assign, assign, n0*sizeof(int)); 1303 memcpy(cpuc->assign, assign, n0*sizeof(int));
1308 1304
1309 cpuc->n_events = n0; 1305 cpuc->n_events = n0;
1310 cpuc->n_added = n1; 1306 cpuc->n_added += n1;
1311 ctx->nr_active += n1; 1307 ctx->nr_active += n1;
1312 1308
1313 /* 1309 /*
@@ -1335,6 +1331,41 @@ undo:
1335#include "perf_event_p6.c" 1331#include "perf_event_p6.c"
1336#include "perf_event_intel.c" 1332#include "perf_event_intel.c"
1337 1333
1334static int __cpuinit
1335x86_pmu_notifier(struct notifier_block *self, unsigned long action, void *hcpu)
1336{
1337 unsigned int cpu = (long)hcpu;
1338 int ret = NOTIFY_OK;
1339
1340 switch (action & ~CPU_TASKS_FROZEN) {
1341 case CPU_UP_PREPARE:
1342 if (x86_pmu.cpu_prepare)
1343 ret = x86_pmu.cpu_prepare(cpu);
1344 break;
1345
1346 case CPU_STARTING:
1347 if (x86_pmu.cpu_starting)
1348 x86_pmu.cpu_starting(cpu);
1349 break;
1350
1351 case CPU_DYING:
1352 if (x86_pmu.cpu_dying)
1353 x86_pmu.cpu_dying(cpu);
1354 break;
1355
1356 case CPU_UP_CANCELED:
1357 case CPU_DEAD:
1358 if (x86_pmu.cpu_dead)
1359 x86_pmu.cpu_dead(cpu);
1360 break;
1361
1362 default:
1363 break;
1364 }
1365
1366 return ret;
1367}
1368
1338static void __init pmu_check_apic(void) 1369static void __init pmu_check_apic(void)
1339{ 1370{
1340 if (cpu_has_apic) 1371 if (cpu_has_apic)
@@ -1347,6 +1378,7 @@ static void __init pmu_check_apic(void)
1347 1378
1348void __init init_hw_perf_events(void) 1379void __init init_hw_perf_events(void)
1349{ 1380{
1381 struct event_constraint *c;
1350 int err; 1382 int err;
1351 1383
1352 pr_info("Performance Events: "); 1384 pr_info("Performance Events: ");
@@ -1395,6 +1427,16 @@ void __init init_hw_perf_events(void)
1395 __EVENT_CONSTRAINT(0, (1ULL << x86_pmu.num_events) - 1, 1427 __EVENT_CONSTRAINT(0, (1ULL << x86_pmu.num_events) - 1,
1396 0, x86_pmu.num_events); 1428 0, x86_pmu.num_events);
1397 1429
1430 if (x86_pmu.event_constraints) {
1431 for_each_event_constraint(c, x86_pmu.event_constraints) {
1432 if (c->cmask != INTEL_ARCH_FIXED_MASK)
1433 continue;
1434
1435 c->idxmsk64 |= (1ULL << x86_pmu.num_events) - 1;
1436 c->weight += x86_pmu.num_events;
1437 }
1438 }
1439
1398 pr_info("... version: %d\n", x86_pmu.version); 1440 pr_info("... version: %d\n", x86_pmu.version);
1399 pr_info("... bit width: %d\n", x86_pmu.event_bits); 1441 pr_info("... bit width: %d\n", x86_pmu.event_bits);
1400 pr_info("... generic registers: %d\n", x86_pmu.num_events); 1442 pr_info("... generic registers: %d\n", x86_pmu.num_events);
@@ -1402,11 +1444,13 @@ void __init init_hw_perf_events(void)
1402 pr_info("... max period: %016Lx\n", x86_pmu.max_period); 1444 pr_info("... max period: %016Lx\n", x86_pmu.max_period);
1403 pr_info("... fixed-purpose events: %d\n", x86_pmu.num_events_fixed); 1445 pr_info("... fixed-purpose events: %d\n", x86_pmu.num_events_fixed);
1404 pr_info("... event mask: %016Lx\n", perf_event_mask); 1446 pr_info("... event mask: %016Lx\n", perf_event_mask);
1447
1448 perf_cpu_notifier(x86_pmu_notifier);
1405} 1449}
1406 1450
1407static inline void x86_pmu_read(struct perf_event *event) 1451static inline void x86_pmu_read(struct perf_event *event)
1408{ 1452{
1409 x86_perf_event_update(event, &event->hw, event->hw.idx); 1453 x86_perf_event_update(event);
1410} 1454}
1411 1455
1412static const struct pmu pmu = { 1456static const struct pmu pmu = {
@@ -1588,14 +1632,42 @@ copy_from_user_nmi(void *to, const void __user *from, unsigned long n)
1588 return len; 1632 return len;
1589} 1633}
1590 1634
1591static int copy_stack_frame(const void __user *fp, struct stack_frame *frame) 1635#ifdef CONFIG_COMPAT
1636static inline int
1637perf_callchain_user32(struct pt_regs *regs, struct perf_callchain_entry *entry)
1592{ 1638{
1593 unsigned long bytes; 1639 /* 32-bit process in 64-bit kernel. */
1640 struct stack_frame_ia32 frame;
1641 const void __user *fp;
1594 1642
1595 bytes = copy_from_user_nmi(frame, fp, sizeof(*frame)); 1643 if (!test_thread_flag(TIF_IA32))
1644 return 0;
1596 1645
1597 return bytes == sizeof(*frame); 1646 fp = compat_ptr(regs->bp);
1647 while (entry->nr < PERF_MAX_STACK_DEPTH) {
1648 unsigned long bytes;
1649 frame.next_frame = 0;
1650 frame.return_address = 0;
1651
1652 bytes = copy_from_user_nmi(&frame, fp, sizeof(frame));
1653 if (bytes != sizeof(frame))
1654 break;
1655
1656 if (fp < compat_ptr(regs->sp))
1657 break;
1658
1659 callchain_store(entry, frame.return_address);
1660 fp = compat_ptr(frame.next_frame);
1661 }
1662 return 1;
1663}
1664#else
1665static inline int
1666perf_callchain_user32(struct pt_regs *regs, struct perf_callchain_entry *entry)
1667{
1668 return 0;
1598} 1669}
1670#endif
1599 1671
1600static void 1672static void
1601perf_callchain_user(struct pt_regs *regs, struct perf_callchain_entry *entry) 1673perf_callchain_user(struct pt_regs *regs, struct perf_callchain_entry *entry)
@@ -1611,11 +1683,16 @@ perf_callchain_user(struct pt_regs *regs, struct perf_callchain_entry *entry)
1611 callchain_store(entry, PERF_CONTEXT_USER); 1683 callchain_store(entry, PERF_CONTEXT_USER);
1612 callchain_store(entry, regs->ip); 1684 callchain_store(entry, regs->ip);
1613 1685
1686 if (perf_callchain_user32(regs, entry))
1687 return;
1688
1614 while (entry->nr < PERF_MAX_STACK_DEPTH) { 1689 while (entry->nr < PERF_MAX_STACK_DEPTH) {
1690 unsigned long bytes;
1615 frame.next_frame = NULL; 1691 frame.next_frame = NULL;
1616 frame.return_address = 0; 1692 frame.return_address = 0;
1617 1693
1618 if (!copy_stack_frame(fp, &frame)) 1694 bytes = copy_from_user_nmi(&frame, fp, sizeof(frame));
1695 if (bytes != sizeof(frame))
1619 break; 1696 break;
1620 1697
1621 if ((unsigned long)fp < regs->sp) 1698 if ((unsigned long)fp < regs->sp)
@@ -1662,28 +1739,14 @@ struct perf_callchain_entry *perf_callchain(struct pt_regs *regs)
1662 return entry; 1739 return entry;
1663} 1740}
1664 1741
1665void hw_perf_event_setup_online(int cpu) 1742void perf_arch_fetch_caller_regs(struct pt_regs *regs, unsigned long ip, int skip)
1666{
1667 init_debug_store_on_cpu(cpu);
1668
1669 switch (boot_cpu_data.x86_vendor) {
1670 case X86_VENDOR_AMD:
1671 amd_pmu_cpu_online(cpu);
1672 break;
1673 default:
1674 return;
1675 }
1676}
1677
1678void hw_perf_event_setup_offline(int cpu)
1679{ 1743{
1680 init_debug_store_on_cpu(cpu); 1744 regs->ip = ip;
1681 1745 /*
1682 switch (boot_cpu_data.x86_vendor) { 1746 * perf_arch_fetch_caller_regs adds another call, we need to increment
1683 case X86_VENDOR_AMD: 1747 * the skip level
1684 amd_pmu_cpu_offline(cpu); 1748 */
1685 break; 1749 regs->bp = rewind_frame_pointer(skip + 1);
1686 default: 1750 regs->cs = __KERNEL_CS;
1687 return; 1751 local_save_flags(regs->flags);
1688 }
1689} 1752}
diff --git a/arch/x86/kernel/cpu/perf_event_amd.c b/arch/x86/kernel/cpu/perf_event_amd.c
index 8f3dbfda3c4f..db6f7d4056e1 100644
--- a/arch/x86/kernel/cpu/perf_event_amd.c
+++ b/arch/x86/kernel/cpu/perf_event_amd.c
@@ -137,6 +137,13 @@ static inline int amd_is_nb_event(struct hw_perf_event *hwc)
137 return (hwc->config & 0xe0) == 0xe0; 137 return (hwc->config & 0xe0) == 0xe0;
138} 138}
139 139
140static inline int amd_has_nb(struct cpu_hw_events *cpuc)
141{
142 struct amd_nb *nb = cpuc->amd_nb;
143
144 return nb && nb->nb_id != -1;
145}
146
140static void amd_put_event_constraints(struct cpu_hw_events *cpuc, 147static void amd_put_event_constraints(struct cpu_hw_events *cpuc,
141 struct perf_event *event) 148 struct perf_event *event)
142{ 149{
@@ -147,7 +154,7 @@ static void amd_put_event_constraints(struct cpu_hw_events *cpuc,
147 /* 154 /*
148 * only care about NB events 155 * only care about NB events
149 */ 156 */
150 if (!(nb && amd_is_nb_event(hwc))) 157 if (!(amd_has_nb(cpuc) && amd_is_nb_event(hwc)))
151 return; 158 return;
152 159
153 /* 160 /*
@@ -214,7 +221,7 @@ amd_get_event_constraints(struct cpu_hw_events *cpuc, struct perf_event *event)
214 /* 221 /*
215 * if not NB event or no NB, then no constraints 222 * if not NB event or no NB, then no constraints
216 */ 223 */
217 if (!(nb && amd_is_nb_event(hwc))) 224 if (!(amd_has_nb(cpuc) && amd_is_nb_event(hwc)))
218 return &unconstrained; 225 return &unconstrained;
219 226
220 /* 227 /*
@@ -271,28 +278,6 @@ done:
271 return &emptyconstraint; 278 return &emptyconstraint;
272} 279}
273 280
274static __initconst struct x86_pmu amd_pmu = {
275 .name = "AMD",
276 .handle_irq = x86_pmu_handle_irq,
277 .disable_all = x86_pmu_disable_all,
278 .enable_all = x86_pmu_enable_all,
279 .enable = x86_pmu_enable_event,
280 .disable = x86_pmu_disable_event,
281 .eventsel = MSR_K7_EVNTSEL0,
282 .perfctr = MSR_K7_PERFCTR0,
283 .event_map = amd_pmu_event_map,
284 .raw_event = amd_pmu_raw_event,
285 .max_events = ARRAY_SIZE(amd_perfmon_event_map),
286 .num_events = 4,
287 .event_bits = 48,
288 .event_mask = (1ULL << 48) - 1,
289 .apic = 1,
290 /* use highest bit to detect overflow */
291 .max_period = (1ULL << 47) - 1,
292 .get_event_constraints = amd_get_event_constraints,
293 .put_event_constraints = amd_put_event_constraints
294};
295
296static struct amd_nb *amd_alloc_nb(int cpu, int nb_id) 281static struct amd_nb *amd_alloc_nb(int cpu, int nb_id)
297{ 282{
298 struct amd_nb *nb; 283 struct amd_nb *nb;
@@ -309,57 +294,61 @@ static struct amd_nb *amd_alloc_nb(int cpu, int nb_id)
309 * initialize all possible NB constraints 294 * initialize all possible NB constraints
310 */ 295 */
311 for (i = 0; i < x86_pmu.num_events; i++) { 296 for (i = 0; i < x86_pmu.num_events; i++) {
312 set_bit(i, nb->event_constraints[i].idxmsk); 297 __set_bit(i, nb->event_constraints[i].idxmsk);
313 nb->event_constraints[i].weight = 1; 298 nb->event_constraints[i].weight = 1;
314 } 299 }
315 return nb; 300 return nb;
316} 301}
317 302
318static void amd_pmu_cpu_online(int cpu) 303static int amd_pmu_cpu_prepare(int cpu)
319{ 304{
320 struct cpu_hw_events *cpu1, *cpu2; 305 struct cpu_hw_events *cpuc = &per_cpu(cpu_hw_events, cpu);
321 struct amd_nb *nb = NULL; 306
307 WARN_ON_ONCE(cpuc->amd_nb);
308
309 if (boot_cpu_data.x86_max_cores < 2)
310 return NOTIFY_OK;
311
312 cpuc->amd_nb = amd_alloc_nb(cpu, -1);
313 if (!cpuc->amd_nb)
314 return NOTIFY_BAD;
315
316 return NOTIFY_OK;
317}
318
319static void amd_pmu_cpu_starting(int cpu)
320{
321 struct cpu_hw_events *cpuc = &per_cpu(cpu_hw_events, cpu);
322 struct amd_nb *nb;
322 int i, nb_id; 323 int i, nb_id;
323 324
324 if (boot_cpu_data.x86_max_cores < 2) 325 if (boot_cpu_data.x86_max_cores < 2)
325 return; 326 return;
326 327
327 /*
328 * function may be called too early in the
329 * boot process, in which case nb_id is bogus
330 */
331 nb_id = amd_get_nb_id(cpu); 328 nb_id = amd_get_nb_id(cpu);
332 if (nb_id == BAD_APICID) 329 WARN_ON_ONCE(nb_id == BAD_APICID);
333 return;
334
335 cpu1 = &per_cpu(cpu_hw_events, cpu);
336 cpu1->amd_nb = NULL;
337 330
338 raw_spin_lock(&amd_nb_lock); 331 raw_spin_lock(&amd_nb_lock);
339 332
340 for_each_online_cpu(i) { 333 for_each_online_cpu(i) {
341 cpu2 = &per_cpu(cpu_hw_events, i); 334 nb = per_cpu(cpu_hw_events, i).amd_nb;
342 nb = cpu2->amd_nb; 335 if (WARN_ON_ONCE(!nb))
343 if (!nb)
344 continue; 336 continue;
345 if (nb->nb_id == nb_id)
346 goto found;
347 }
348 337
349 nb = amd_alloc_nb(cpu, nb_id); 338 if (nb->nb_id == nb_id) {
350 if (!nb) { 339 kfree(cpuc->amd_nb);
351 pr_err("perf_events: failed NB allocation for CPU%d\n", cpu); 340 cpuc->amd_nb = nb;
352 raw_spin_unlock(&amd_nb_lock); 341 break;
353 return; 342 }
354 } 343 }
355found: 344
356 nb->refcnt++; 345 cpuc->amd_nb->nb_id = nb_id;
357 cpu1->amd_nb = nb; 346 cpuc->amd_nb->refcnt++;
358 347
359 raw_spin_unlock(&amd_nb_lock); 348 raw_spin_unlock(&amd_nb_lock);
360} 349}
361 350
362static void amd_pmu_cpu_offline(int cpu) 351static void amd_pmu_cpu_dead(int cpu)
363{ 352{
364 struct cpu_hw_events *cpuhw; 353 struct cpu_hw_events *cpuhw;
365 354
@@ -370,14 +359,44 @@ static void amd_pmu_cpu_offline(int cpu)
370 359
371 raw_spin_lock(&amd_nb_lock); 360 raw_spin_lock(&amd_nb_lock);
372 361
373 if (--cpuhw->amd_nb->refcnt == 0) 362 if (cpuhw->amd_nb) {
374 kfree(cpuhw->amd_nb); 363 struct amd_nb *nb = cpuhw->amd_nb;
364
365 if (nb->nb_id == -1 || --nb->refcnt == 0)
366 kfree(nb);
375 367
376 cpuhw->amd_nb = NULL; 368 cpuhw->amd_nb = NULL;
369 }
377 370
378 raw_spin_unlock(&amd_nb_lock); 371 raw_spin_unlock(&amd_nb_lock);
379} 372}
380 373
374static __initconst struct x86_pmu amd_pmu = {
375 .name = "AMD",
376 .handle_irq = x86_pmu_handle_irq,
377 .disable_all = x86_pmu_disable_all,
378 .enable_all = x86_pmu_enable_all,
379 .enable = x86_pmu_enable_event,
380 .disable = x86_pmu_disable_event,
381 .eventsel = MSR_K7_EVNTSEL0,
382 .perfctr = MSR_K7_PERFCTR0,
383 .event_map = amd_pmu_event_map,
384 .raw_event = amd_pmu_raw_event,
385 .max_events = ARRAY_SIZE(amd_perfmon_event_map),
386 .num_events = 4,
387 .event_bits = 48,
388 .event_mask = (1ULL << 48) - 1,
389 .apic = 1,
390 /* use highest bit to detect overflow */
391 .max_period = (1ULL << 47) - 1,
392 .get_event_constraints = amd_get_event_constraints,
393 .put_event_constraints = amd_put_event_constraints,
394
395 .cpu_prepare = amd_pmu_cpu_prepare,
396 .cpu_starting = amd_pmu_cpu_starting,
397 .cpu_dead = amd_pmu_cpu_dead,
398};
399
381static __init int amd_pmu_init(void) 400static __init int amd_pmu_init(void)
382{ 401{
383 /* Performance-monitoring supported from K7 and later: */ 402 /* Performance-monitoring supported from K7 and later: */
@@ -390,11 +409,6 @@ static __init int amd_pmu_init(void)
390 memcpy(hw_cache_event_ids, amd_hw_cache_event_ids, 409 memcpy(hw_cache_event_ids, amd_hw_cache_event_ids,
391 sizeof(hw_cache_event_ids)); 410 sizeof(hw_cache_event_ids));
392 411
393 /*
394 * explicitly initialize the boot cpu, other cpus will get
395 * the cpu hotplug callbacks from smp_init()
396 */
397 amd_pmu_cpu_online(smp_processor_id());
398 return 0; 412 return 0;
399} 413}
400 414
@@ -405,12 +419,4 @@ static int amd_pmu_init(void)
405 return 0; 419 return 0;
406} 420}
407 421
408static void amd_pmu_cpu_online(int cpu)
409{
410}
411
412static void amd_pmu_cpu_offline(int cpu)
413{
414}
415
416#endif 422#endif
diff --git a/arch/x86/kernel/cpu/perf_event_intel.c b/arch/x86/kernel/cpu/perf_event_intel.c
index 977e7544738c..9c794ac87837 100644
--- a/arch/x86/kernel/cpu/perf_event_intel.c
+++ b/arch/x86/kernel/cpu/perf_event_intel.c
@@ -1,7 +1,7 @@
1#ifdef CONFIG_CPU_SUP_INTEL 1#ifdef CONFIG_CPU_SUP_INTEL
2 2
3/* 3/*
4 * Intel PerfMon v3. Used on Core2 and later. 4 * Intel PerfMon, used on Core and later.
5 */ 5 */
6static const u64 intel_perfmon_event_map[] = 6static const u64 intel_perfmon_event_map[] =
7{ 7{
@@ -27,8 +27,14 @@ static struct event_constraint intel_core_event_constraints[] =
27 27
28static struct event_constraint intel_core2_event_constraints[] = 28static struct event_constraint intel_core2_event_constraints[] =
29{ 29{
30 FIXED_EVENT_CONSTRAINT(0xc0, (0x3|(1ULL<<32))), /* INSTRUCTIONS_RETIRED */ 30 FIXED_EVENT_CONSTRAINT(0x00c0, 0), /* INST_RETIRED.ANY */
31 FIXED_EVENT_CONSTRAINT(0x3c, (0x3|(1ULL<<33))), /* UNHALTED_CORE_CYCLES */ 31 FIXED_EVENT_CONSTRAINT(0x003c, 1), /* CPU_CLK_UNHALTED.CORE */
32 /*
33 * Core2 has Fixed Counter 2 listed as CPU_CLK_UNHALTED.REF and event
34 * 0x013c as CPU_CLK_UNHALTED.BUS and specifies there is a fixed
35 * ratio between these counters.
36 */
37 /* FIXED_EVENT_CONSTRAINT(0x013c, 2), CPU_CLK_UNHALTED.REF */
32 INTEL_EVENT_CONSTRAINT(0x10, 0x1), /* FP_COMP_OPS_EXE */ 38 INTEL_EVENT_CONSTRAINT(0x10, 0x1), /* FP_COMP_OPS_EXE */
33 INTEL_EVENT_CONSTRAINT(0x11, 0x2), /* FP_ASSIST */ 39 INTEL_EVENT_CONSTRAINT(0x11, 0x2), /* FP_ASSIST */
34 INTEL_EVENT_CONSTRAINT(0x12, 0x2), /* MUL */ 40 INTEL_EVENT_CONSTRAINT(0x12, 0x2), /* MUL */
@@ -37,14 +43,16 @@ static struct event_constraint intel_core2_event_constraints[] =
37 INTEL_EVENT_CONSTRAINT(0x18, 0x1), /* IDLE_DURING_DIV */ 43 INTEL_EVENT_CONSTRAINT(0x18, 0x1), /* IDLE_DURING_DIV */
38 INTEL_EVENT_CONSTRAINT(0x19, 0x2), /* DELAYED_BYPASS */ 44 INTEL_EVENT_CONSTRAINT(0x19, 0x2), /* DELAYED_BYPASS */
39 INTEL_EVENT_CONSTRAINT(0xa1, 0x1), /* RS_UOPS_DISPATCH_CYCLES */ 45 INTEL_EVENT_CONSTRAINT(0xa1, 0x1), /* RS_UOPS_DISPATCH_CYCLES */
46 INTEL_EVENT_CONSTRAINT(0xc9, 0x1), /* ITLB_MISS_RETIRED (T30-9) */
40 INTEL_EVENT_CONSTRAINT(0xcb, 0x1), /* MEM_LOAD_RETIRED */ 47 INTEL_EVENT_CONSTRAINT(0xcb, 0x1), /* MEM_LOAD_RETIRED */
41 EVENT_CONSTRAINT_END 48 EVENT_CONSTRAINT_END
42}; 49};
43 50
44static struct event_constraint intel_nehalem_event_constraints[] = 51static struct event_constraint intel_nehalem_event_constraints[] =
45{ 52{
46 FIXED_EVENT_CONSTRAINT(0xc0, (0xf|(1ULL<<32))), /* INSTRUCTIONS_RETIRED */ 53 FIXED_EVENT_CONSTRAINT(0x00c0, 0), /* INST_RETIRED.ANY */
47 FIXED_EVENT_CONSTRAINT(0x3c, (0xf|(1ULL<<33))), /* UNHALTED_CORE_CYCLES */ 54 FIXED_EVENT_CONSTRAINT(0x003c, 1), /* CPU_CLK_UNHALTED.CORE */
55 /* FIXED_EVENT_CONSTRAINT(0x013c, 2), CPU_CLK_UNHALTED.REF */
48 INTEL_EVENT_CONSTRAINT(0x40, 0x3), /* L1D_CACHE_LD */ 56 INTEL_EVENT_CONSTRAINT(0x40, 0x3), /* L1D_CACHE_LD */
49 INTEL_EVENT_CONSTRAINT(0x41, 0x3), /* L1D_CACHE_ST */ 57 INTEL_EVENT_CONSTRAINT(0x41, 0x3), /* L1D_CACHE_ST */
50 INTEL_EVENT_CONSTRAINT(0x42, 0x3), /* L1D_CACHE_LOCK */ 58 INTEL_EVENT_CONSTRAINT(0x42, 0x3), /* L1D_CACHE_LOCK */
@@ -58,8 +66,9 @@ static struct event_constraint intel_nehalem_event_constraints[] =
58 66
59static struct event_constraint intel_westmere_event_constraints[] = 67static struct event_constraint intel_westmere_event_constraints[] =
60{ 68{
61 FIXED_EVENT_CONSTRAINT(0xc0, (0xf|(1ULL<<32))), /* INSTRUCTIONS_RETIRED */ 69 FIXED_EVENT_CONSTRAINT(0x00c0, 0), /* INST_RETIRED.ANY */
62 FIXED_EVENT_CONSTRAINT(0x3c, (0xf|(1ULL<<33))), /* UNHALTED_CORE_CYCLES */ 70 FIXED_EVENT_CONSTRAINT(0x003c, 1), /* CPU_CLK_UNHALTED.CORE */
71 /* FIXED_EVENT_CONSTRAINT(0x013c, 2), CPU_CLK_UNHALTED.REF */
63 INTEL_EVENT_CONSTRAINT(0x51, 0x3), /* L1D */ 72 INTEL_EVENT_CONSTRAINT(0x51, 0x3), /* L1D */
64 INTEL_EVENT_CONSTRAINT(0x60, 0x1), /* OFFCORE_REQUESTS_OUTSTANDING */ 73 INTEL_EVENT_CONSTRAINT(0x60, 0x1), /* OFFCORE_REQUESTS_OUTSTANDING */
65 INTEL_EVENT_CONSTRAINT(0x63, 0x3), /* CACHE_LOCK_CYCLES */ 74 INTEL_EVENT_CONSTRAINT(0x63, 0x3), /* CACHE_LOCK_CYCLES */
@@ -68,8 +77,9 @@ static struct event_constraint intel_westmere_event_constraints[] =
68 77
69static struct event_constraint intel_gen_event_constraints[] = 78static struct event_constraint intel_gen_event_constraints[] =
70{ 79{
71 FIXED_EVENT_CONSTRAINT(0xc0, (0x3|(1ULL<<32))), /* INSTRUCTIONS_RETIRED */ 80 FIXED_EVENT_CONSTRAINT(0x00c0, 0), /* INST_RETIRED.ANY */
72 FIXED_EVENT_CONSTRAINT(0x3c, (0x3|(1ULL<<33))), /* UNHALTED_CORE_CYCLES */ 81 FIXED_EVENT_CONSTRAINT(0x003c, 1), /* CPU_CLK_UNHALTED.CORE */
82 /* FIXED_EVENT_CONSTRAINT(0x013c, 2), CPU_CLK_UNHALTED.REF */
73 EVENT_CONSTRAINT_END 83 EVENT_CONSTRAINT_END
74}; 84};
75 85
@@ -538,9 +548,9 @@ static inline void intel_pmu_ack_status(u64 ack)
538} 548}
539 549
540static inline void 550static inline void
541intel_pmu_disable_fixed(struct hw_perf_event *hwc, int __idx) 551intel_pmu_disable_fixed(struct hw_perf_event *hwc)
542{ 552{
543 int idx = __idx - X86_PMC_IDX_FIXED; 553 int idx = hwc->idx - X86_PMC_IDX_FIXED;
544 u64 ctrl_val, mask; 554 u64 ctrl_val, mask;
545 555
546 mask = 0xfULL << (idx * 4); 556 mask = 0xfULL << (idx * 4);
@@ -580,10 +590,9 @@ static void intel_pmu_drain_bts_buffer(void)
580 590
581 ds->bts_index = ds->bts_buffer_base; 591 ds->bts_index = ds->bts_buffer_base;
582 592
593 perf_sample_data_init(&data, 0);
583 594
584 data.period = event->hw.last_period; 595 data.period = event->hw.last_period;
585 data.addr = 0;
586 data.raw = NULL;
587 regs.ip = 0; 596 regs.ip = 0;
588 597
589 /* 598 /*
@@ -612,26 +621,28 @@ static void intel_pmu_drain_bts_buffer(void)
612} 621}
613 622
614static inline void 623static inline void
615intel_pmu_disable_event(struct hw_perf_event *hwc, int idx) 624intel_pmu_disable_event(struct perf_event *event)
616{ 625{
617 if (unlikely(idx == X86_PMC_IDX_FIXED_BTS)) { 626 struct hw_perf_event *hwc = &event->hw;
627
628 if (unlikely(hwc->idx == X86_PMC_IDX_FIXED_BTS)) {
618 intel_pmu_disable_bts(); 629 intel_pmu_disable_bts();
619 intel_pmu_drain_bts_buffer(); 630 intel_pmu_drain_bts_buffer();
620 return; 631 return;
621 } 632 }
622 633
623 if (unlikely(hwc->config_base == MSR_ARCH_PERFMON_FIXED_CTR_CTRL)) { 634 if (unlikely(hwc->config_base == MSR_ARCH_PERFMON_FIXED_CTR_CTRL)) {
624 intel_pmu_disable_fixed(hwc, idx); 635 intel_pmu_disable_fixed(hwc);
625 return; 636 return;
626 } 637 }
627 638
628 x86_pmu_disable_event(hwc, idx); 639 x86_pmu_disable_event(event);
629} 640}
630 641
631static inline void 642static inline void
632intel_pmu_enable_fixed(struct hw_perf_event *hwc, int __idx) 643intel_pmu_enable_fixed(struct hw_perf_event *hwc)
633{ 644{
634 int idx = __idx - X86_PMC_IDX_FIXED; 645 int idx = hwc->idx - X86_PMC_IDX_FIXED;
635 u64 ctrl_val, bits, mask; 646 u64 ctrl_val, bits, mask;
636 int err; 647 int err;
637 648
@@ -661,9 +672,11 @@ intel_pmu_enable_fixed(struct hw_perf_event *hwc, int __idx)
661 err = checking_wrmsrl(hwc->config_base, ctrl_val); 672 err = checking_wrmsrl(hwc->config_base, ctrl_val);
662} 673}
663 674
664static void intel_pmu_enable_event(struct hw_perf_event *hwc, int idx) 675static void intel_pmu_enable_event(struct perf_event *event)
665{ 676{
666 if (unlikely(idx == X86_PMC_IDX_FIXED_BTS)) { 677 struct hw_perf_event *hwc = &event->hw;
678
679 if (unlikely(hwc->idx == X86_PMC_IDX_FIXED_BTS)) {
667 if (!__get_cpu_var(cpu_hw_events).enabled) 680 if (!__get_cpu_var(cpu_hw_events).enabled)
668 return; 681 return;
669 682
@@ -672,11 +685,11 @@ static void intel_pmu_enable_event(struct hw_perf_event *hwc, int idx)
672 } 685 }
673 686
674 if (unlikely(hwc->config_base == MSR_ARCH_PERFMON_FIXED_CTR_CTRL)) { 687 if (unlikely(hwc->config_base == MSR_ARCH_PERFMON_FIXED_CTR_CTRL)) {
675 intel_pmu_enable_fixed(hwc, idx); 688 intel_pmu_enable_fixed(hwc);
676 return; 689 return;
677 } 690 }
678 691
679 __x86_pmu_enable_event(hwc, idx); 692 __x86_pmu_enable_event(hwc);
680} 693}
681 694
682/* 695/*
@@ -685,14 +698,8 @@ static void intel_pmu_enable_event(struct hw_perf_event *hwc, int idx)
685 */ 698 */
686static int intel_pmu_save_and_restart(struct perf_event *event) 699static int intel_pmu_save_and_restart(struct perf_event *event)
687{ 700{
688 struct hw_perf_event *hwc = &event->hw; 701 x86_perf_event_update(event);
689 int idx = hwc->idx; 702 return x86_perf_event_set_period(event);
690 int ret;
691
692 x86_perf_event_update(event, hwc, idx);
693 ret = x86_perf_event_set_period(event, hwc, idx);
694
695 return ret;
696} 703}
697 704
698static void intel_pmu_reset(void) 705static void intel_pmu_reset(void)
@@ -732,16 +739,15 @@ static int intel_pmu_handle_irq(struct pt_regs *regs)
732 int bit, loops; 739 int bit, loops;
733 u64 ack, status; 740 u64 ack, status;
734 741
735 data.addr = 0; 742 perf_sample_data_init(&data, 0);
736 data.raw = NULL;
737 743
738 cpuc = &__get_cpu_var(cpu_hw_events); 744 cpuc = &__get_cpu_var(cpu_hw_events);
739 745
740 perf_disable(); 746 intel_pmu_disable_all();
741 intel_pmu_drain_bts_buffer(); 747 intel_pmu_drain_bts_buffer();
742 status = intel_pmu_get_status(); 748 status = intel_pmu_get_status();
743 if (!status) { 749 if (!status) {
744 perf_enable(); 750 intel_pmu_enable_all();
745 return 0; 751 return 0;
746 } 752 }
747 753
@@ -751,8 +757,7 @@ again:
751 WARN_ONCE(1, "perfevents: irq loop stuck!\n"); 757 WARN_ONCE(1, "perfevents: irq loop stuck!\n");
752 perf_event_print_debug(); 758 perf_event_print_debug();
753 intel_pmu_reset(); 759 intel_pmu_reset();
754 perf_enable(); 760 goto done;
755 return 1;
756 } 761 }
757 762
758 inc_irq_stat(apic_perf_irqs); 763 inc_irq_stat(apic_perf_irqs);
@@ -760,7 +765,6 @@ again:
760 for_each_set_bit(bit, (unsigned long *)&status, X86_PMC_IDX_MAX) { 765 for_each_set_bit(bit, (unsigned long *)&status, X86_PMC_IDX_MAX) {
761 struct perf_event *event = cpuc->events[bit]; 766 struct perf_event *event = cpuc->events[bit];
762 767
763 clear_bit(bit, (unsigned long *) &status);
764 if (!test_bit(bit, cpuc->active_mask)) 768 if (!test_bit(bit, cpuc->active_mask))
765 continue; 769 continue;
766 770
@@ -770,7 +774,7 @@ again:
770 data.period = event->hw.last_period; 774 data.period = event->hw.last_period;
771 775
772 if (perf_event_overflow(event, 1, &data, regs)) 776 if (perf_event_overflow(event, 1, &data, regs))
773 intel_pmu_disable_event(&event->hw, bit); 777 x86_pmu_stop(event);
774 } 778 }
775 779
776 intel_pmu_ack_status(ack); 780 intel_pmu_ack_status(ack);
@@ -782,8 +786,8 @@ again:
782 if (status) 786 if (status)
783 goto again; 787 goto again;
784 788
785 perf_enable(); 789done:
786 790 intel_pmu_enable_all();
787 return 1; 791 return 1;
788} 792}
789 793
@@ -862,7 +866,10 @@ static __initconst struct x86_pmu intel_pmu = {
862 .max_period = (1ULL << 31) - 1, 866 .max_period = (1ULL << 31) - 1,
863 .enable_bts = intel_pmu_enable_bts, 867 .enable_bts = intel_pmu_enable_bts,
864 .disable_bts = intel_pmu_disable_bts, 868 .disable_bts = intel_pmu_disable_bts,
865 .get_event_constraints = intel_get_event_constraints 869 .get_event_constraints = intel_get_event_constraints,
870
871 .cpu_starting = init_debug_store_on_cpu,
872 .cpu_dying = fini_debug_store_on_cpu,
866}; 873};
867 874
868static __init int intel_pmu_init(void) 875static __init int intel_pmu_init(void)
@@ -929,13 +936,14 @@ static __init int intel_pmu_init(void)
929 936
930 case 26: /* 45 nm nehalem, "Bloomfield" */ 937 case 26: /* 45 nm nehalem, "Bloomfield" */
931 case 30: /* 45 nm nehalem, "Lynnfield" */ 938 case 30: /* 45 nm nehalem, "Lynnfield" */
939 case 46: /* 45 nm nehalem-ex, "Beckton" */
932 memcpy(hw_cache_event_ids, nehalem_hw_cache_event_ids, 940 memcpy(hw_cache_event_ids, nehalem_hw_cache_event_ids,
933 sizeof(hw_cache_event_ids)); 941 sizeof(hw_cache_event_ids));
934 942
935 x86_pmu.event_constraints = intel_nehalem_event_constraints; 943 x86_pmu.event_constraints = intel_nehalem_event_constraints;
936 pr_cont("Nehalem/Corei7 events, "); 944 pr_cont("Nehalem/Corei7 events, ");
937 break; 945 break;
938 case 28: 946 case 28: /* Atom */
939 memcpy(hw_cache_event_ids, atom_hw_cache_event_ids, 947 memcpy(hw_cache_event_ids, atom_hw_cache_event_ids,
940 sizeof(hw_cache_event_ids)); 948 sizeof(hw_cache_event_ids));
941 949
@@ -951,6 +959,7 @@ static __init int intel_pmu_init(void)
951 x86_pmu.event_constraints = intel_westmere_event_constraints; 959 x86_pmu.event_constraints = intel_westmere_event_constraints;
952 pr_cont("Westmere events, "); 960 pr_cont("Westmere events, ");
953 break; 961 break;
962
954 default: 963 default:
955 /* 964 /*
956 * default constraints for v2 and up 965 * default constraints for v2 and up
diff --git a/arch/x86/kernel/cpu/perf_event_p6.c b/arch/x86/kernel/cpu/perf_event_p6.c
index 1ca5ba078afd..a330485d14da 100644
--- a/arch/x86/kernel/cpu/perf_event_p6.c
+++ b/arch/x86/kernel/cpu/perf_event_p6.c
@@ -62,7 +62,7 @@ static void p6_pmu_disable_all(void)
62 62
63 /* p6 only has one enable register */ 63 /* p6 only has one enable register */
64 rdmsrl(MSR_P6_EVNTSEL0, val); 64 rdmsrl(MSR_P6_EVNTSEL0, val);
65 val &= ~ARCH_PERFMON_EVENTSEL0_ENABLE; 65 val &= ~ARCH_PERFMON_EVENTSEL_ENABLE;
66 wrmsrl(MSR_P6_EVNTSEL0, val); 66 wrmsrl(MSR_P6_EVNTSEL0, val);
67} 67}
68 68
@@ -72,32 +72,34 @@ static void p6_pmu_enable_all(void)
72 72
73 /* p6 only has one enable register */ 73 /* p6 only has one enable register */
74 rdmsrl(MSR_P6_EVNTSEL0, val); 74 rdmsrl(MSR_P6_EVNTSEL0, val);
75 val |= ARCH_PERFMON_EVENTSEL0_ENABLE; 75 val |= ARCH_PERFMON_EVENTSEL_ENABLE;
76 wrmsrl(MSR_P6_EVNTSEL0, val); 76 wrmsrl(MSR_P6_EVNTSEL0, val);
77} 77}
78 78
79static inline void 79static inline void
80p6_pmu_disable_event(struct hw_perf_event *hwc, int idx) 80p6_pmu_disable_event(struct perf_event *event)
81{ 81{
82 struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); 82 struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
83 struct hw_perf_event *hwc = &event->hw;
83 u64 val = P6_NOP_EVENT; 84 u64 val = P6_NOP_EVENT;
84 85
85 if (cpuc->enabled) 86 if (cpuc->enabled)
86 val |= ARCH_PERFMON_EVENTSEL0_ENABLE; 87 val |= ARCH_PERFMON_EVENTSEL_ENABLE;
87 88
88 (void)checking_wrmsrl(hwc->config_base + idx, val); 89 (void)checking_wrmsrl(hwc->config_base + hwc->idx, val);
89} 90}
90 91
91static void p6_pmu_enable_event(struct hw_perf_event *hwc, int idx) 92static void p6_pmu_enable_event(struct perf_event *event)
92{ 93{
93 struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); 94 struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
95 struct hw_perf_event *hwc = &event->hw;
94 u64 val; 96 u64 val;
95 97
96 val = hwc->config; 98 val = hwc->config;
97 if (cpuc->enabled) 99 if (cpuc->enabled)
98 val |= ARCH_PERFMON_EVENTSEL0_ENABLE; 100 val |= ARCH_PERFMON_EVENTSEL_ENABLE;
99 101
100 (void)checking_wrmsrl(hwc->config_base + idx, val); 102 (void)checking_wrmsrl(hwc->config_base + hwc->idx, val);
101} 103}
102 104
103static __initconst struct x86_pmu p6_pmu = { 105static __initconst struct x86_pmu p6_pmu = {
diff --git a/arch/x86/kernel/cpu/perfctr-watchdog.c b/arch/x86/kernel/cpu/perfctr-watchdog.c
index 74f4e85a5727..fb329e9f8494 100644
--- a/arch/x86/kernel/cpu/perfctr-watchdog.c
+++ b/arch/x86/kernel/cpu/perfctr-watchdog.c
@@ -680,7 +680,7 @@ static int setup_intel_arch_watchdog(unsigned nmi_hz)
680 cpu_nmi_set_wd_enabled(); 680 cpu_nmi_set_wd_enabled();
681 681
682 apic_write(APIC_LVTPC, APIC_DM_NMI); 682 apic_write(APIC_LVTPC, APIC_DM_NMI);
683 evntsel |= ARCH_PERFMON_EVENTSEL0_ENABLE; 683 evntsel |= ARCH_PERFMON_EVENTSEL_ENABLE;
684 wrmsr(evntsel_msr, evntsel, 0); 684 wrmsr(evntsel_msr, evntsel, 0);
685 intel_arch_wd_ops.checkbit = 1ULL << (eax.split.bit_width - 1); 685 intel_arch_wd_ops.checkbit = 1ULL << (eax.split.bit_width - 1);
686 return 1; 686 return 1;
diff --git a/arch/x86/kernel/cpuid.c b/arch/x86/kernel/cpuid.c
index 83e5e628de73..8b862d5900fe 100644
--- a/arch/x86/kernel/cpuid.c
+++ b/arch/x86/kernel/cpuid.c
@@ -40,6 +40,7 @@
40#include <linux/cpu.h> 40#include <linux/cpu.h>
41#include <linux/notifier.h> 41#include <linux/notifier.h>
42#include <linux/uaccess.h> 42#include <linux/uaccess.h>
43#include <linux/gfp.h>
43 44
44#include <asm/processor.h> 45#include <asm/processor.h>
45#include <asm/msr.h> 46#include <asm/msr.h>
diff --git a/arch/x86/kernel/crash.c b/arch/x86/kernel/crash.c
index a4849c10a77e..ebd4c51d096a 100644
--- a/arch/x86/kernel/crash.c
+++ b/arch/x86/kernel/crash.c
@@ -27,7 +27,6 @@
27#include <asm/cpu.h> 27#include <asm/cpu.h>
28#include <asm/reboot.h> 28#include <asm/reboot.h>
29#include <asm/virtext.h> 29#include <asm/virtext.h>
30#include <asm/x86_init.h>
31 30
32#if defined(CONFIG_SMP) && defined(CONFIG_X86_LOCAL_APIC) 31#if defined(CONFIG_SMP) && defined(CONFIG_X86_LOCAL_APIC)
33 32
@@ -103,10 +102,5 @@ void native_machine_crash_shutdown(struct pt_regs *regs)
103#ifdef CONFIG_HPET_TIMER 102#ifdef CONFIG_HPET_TIMER
104 hpet_disable(); 103 hpet_disable();
105#endif 104#endif
106
107#ifdef CONFIG_X86_64
108 x86_platform.iommu_shutdown();
109#endif
110
111 crash_save_cpu(regs, safe_smp_processor_id()); 105 crash_save_cpu(regs, safe_smp_processor_id());
112} 106}
diff --git a/arch/x86/kernel/crash_dump_32.c b/arch/x86/kernel/crash_dump_32.c
index cd97ce18c29d..67414550c3cc 100644
--- a/arch/x86/kernel/crash_dump_32.c
+++ b/arch/x86/kernel/crash_dump_32.c
@@ -5,6 +5,7 @@
5 * Copyright (C) IBM Corporation, 2004. All rights reserved 5 * Copyright (C) IBM Corporation, 2004. All rights reserved
6 */ 6 */
7 7
8#include <linux/slab.h>
8#include <linux/errno.h> 9#include <linux/errno.h>
9#include <linux/highmem.h> 10#include <linux/highmem.h>
10#include <linux/crash_dump.h> 11#include <linux/crash_dump.h>
diff --git a/arch/x86/kernel/dumpstack.h b/arch/x86/kernel/dumpstack.h
index 4fd1420faffa..e1a93be4fd44 100644
--- a/arch/x86/kernel/dumpstack.h
+++ b/arch/x86/kernel/dumpstack.h
@@ -14,6 +14,8 @@
14#define get_bp(bp) asm("movq %%rbp, %0" : "=r" (bp) :) 14#define get_bp(bp) asm("movq %%rbp, %0" : "=r" (bp) :)
15#endif 15#endif
16 16
17#include <linux/uaccess.h>
18
17extern void 19extern void
18show_trace_log_lvl(struct task_struct *task, struct pt_regs *regs, 20show_trace_log_lvl(struct task_struct *task, struct pt_regs *regs,
19 unsigned long *stack, unsigned long bp, char *log_lvl); 21 unsigned long *stack, unsigned long bp, char *log_lvl);
@@ -29,4 +31,26 @@ struct stack_frame {
29 struct stack_frame *next_frame; 31 struct stack_frame *next_frame;
30 unsigned long return_address; 32 unsigned long return_address;
31}; 33};
34
35struct stack_frame_ia32 {
36 u32 next_frame;
37 u32 return_address;
38};
39
40static inline unsigned long rewind_frame_pointer(int n)
41{
42 struct stack_frame *frame;
43
44 get_bp(frame);
45
46#ifdef CONFIG_FRAME_POINTER
47 while (n--) {
48 if (probe_kernel_address(&frame->next_frame, frame))
49 break;
50 }
32#endif 51#endif
52
53 return (unsigned long)frame;
54}
55
56#endif /* DUMPSTACK_H */
diff --git a/arch/x86/kernel/dumpstack_64.c b/arch/x86/kernel/dumpstack_64.c
index dce99abb4496..272c9f1f05f3 100644
--- a/arch/x86/kernel/dumpstack_64.c
+++ b/arch/x86/kernel/dumpstack_64.c
@@ -120,9 +120,15 @@ fixup_bp_irq_link(unsigned long bp, unsigned long *stack,
120{ 120{
121#ifdef CONFIG_FRAME_POINTER 121#ifdef CONFIG_FRAME_POINTER
122 struct stack_frame *frame = (struct stack_frame *)bp; 122 struct stack_frame *frame = (struct stack_frame *)bp;
123 unsigned long next;
123 124
124 if (!in_irq_stack(stack, irq_stack, irq_stack_end)) 125 if (!in_irq_stack(stack, irq_stack, irq_stack_end)) {
125 return (unsigned long)frame->next_frame; 126 if (!probe_kernel_address(&frame->next_frame, next))
127 return next;
128 else
129 WARN_ONCE(1, "Perf: bad frame pointer = %p in "
130 "callchain\n", &frame->next_frame);
131 }
126#endif 132#endif
127 return bp; 133 return bp;
128} 134}
@@ -202,7 +208,7 @@ void dump_trace(struct task_struct *task, struct pt_regs *regs,
202 if (in_irq_stack(stack, irq_stack, irq_stack_end)) { 208 if (in_irq_stack(stack, irq_stack, irq_stack_end)) {
203 if (ops->stack(data, "IRQ") < 0) 209 if (ops->stack(data, "IRQ") < 0)
204 break; 210 break;
205 bp = print_context_stack(tinfo, stack, bp, 211 bp = ops->walk_stack(tinfo, stack, bp,
206 ops, data, irq_stack_end, &graph); 212 ops, data, irq_stack_end, &graph);
207 /* 213 /*
208 * We link to the next stack (which would be 214 * We link to the next stack (which would be
@@ -223,7 +229,7 @@ void dump_trace(struct task_struct *task, struct pt_regs *regs,
223 /* 229 /*
224 * This handles the process stack: 230 * This handles the process stack:
225 */ 231 */
226 bp = print_context_stack(tinfo, stack, bp, ops, data, NULL, &graph); 232 bp = ops->walk_stack(tinfo, stack, bp, ops, data, NULL, &graph);
227 put_cpu(); 233 put_cpu();
228} 234}
229EXPORT_SYMBOL(dump_trace); 235EXPORT_SYMBOL(dump_trace);
diff --git a/arch/x86/kernel/e820.c b/arch/x86/kernel/e820.c
index 740b440fbd73..7bca3c6a02fb 100644
--- a/arch/x86/kernel/e820.c
+++ b/arch/x86/kernel/e820.c
@@ -519,29 +519,45 @@ u64 __init e820_remove_range(u64 start, u64 size, unsigned old_type,
519 printk(KERN_DEBUG "e820 remove range: %016Lx - %016Lx ", 519 printk(KERN_DEBUG "e820 remove range: %016Lx - %016Lx ",
520 (unsigned long long) start, 520 (unsigned long long) start,
521 (unsigned long long) end); 521 (unsigned long long) end);
522 e820_print_type(old_type); 522 if (checktype)
523 e820_print_type(old_type);
523 printk(KERN_CONT "\n"); 524 printk(KERN_CONT "\n");
524 525
525 for (i = 0; i < e820.nr_map; i++) { 526 for (i = 0; i < e820.nr_map; i++) {
526 struct e820entry *ei = &e820.map[i]; 527 struct e820entry *ei = &e820.map[i];
527 u64 final_start, final_end; 528 u64 final_start, final_end;
529 u64 ei_end;
528 530
529 if (checktype && ei->type != old_type) 531 if (checktype && ei->type != old_type)
530 continue; 532 continue;
533
534 ei_end = ei->addr + ei->size;
531 /* totally covered? */ 535 /* totally covered? */
532 if (ei->addr >= start && 536 if (ei->addr >= start && ei_end <= end) {
533 (ei->addr + ei->size) <= (start + size)) {
534 real_removed_size += ei->size; 537 real_removed_size += ei->size;
535 memset(ei, 0, sizeof(struct e820entry)); 538 memset(ei, 0, sizeof(struct e820entry));
536 continue; 539 continue;
537 } 540 }
541
542 /* new range is totally covered? */
543 if (ei->addr < start && ei_end > end) {
544 e820_add_region(end, ei_end - end, ei->type);
545 ei->size = start - ei->addr;
546 real_removed_size += size;
547 continue;
548 }
549
538 /* partially covered */ 550 /* partially covered */
539 final_start = max(start, ei->addr); 551 final_start = max(start, ei->addr);
540 final_end = min(start + size, ei->addr + ei->size); 552 final_end = min(end, ei_end);
541 if (final_start >= final_end) 553 if (final_start >= final_end)
542 continue; 554 continue;
543 real_removed_size += final_end - final_start; 555 real_removed_size += final_end - final_start;
544 556
557 /*
558 * left range could be head or tail, so need to update
559 * size at first.
560 */
545 ei->size -= final_end - final_start; 561 ei->size -= final_end - final_start;
546 if (ei->addr < final_start) 562 if (ei->addr < final_start)
547 continue; 563 continue;
diff --git a/arch/x86/kernel/head32.c b/arch/x86/kernel/head32.c
index adedeef1dedc..b2e246037392 100644
--- a/arch/x86/kernel/head32.c
+++ b/arch/x86/kernel/head32.c
@@ -7,6 +7,7 @@
7 7
8#include <linux/init.h> 8#include <linux/init.h>
9#include <linux/start_kernel.h> 9#include <linux/start_kernel.h>
10#include <linux/mm.h>
10 11
11#include <asm/setup.h> 12#include <asm/setup.h>
12#include <asm/sections.h> 13#include <asm/sections.h>
@@ -44,9 +45,10 @@ void __init i386_start_kernel(void)
44#ifdef CONFIG_BLK_DEV_INITRD 45#ifdef CONFIG_BLK_DEV_INITRD
45 /* Reserve INITRD */ 46 /* Reserve INITRD */
46 if (boot_params.hdr.type_of_loader && boot_params.hdr.ramdisk_image) { 47 if (boot_params.hdr.type_of_loader && boot_params.hdr.ramdisk_image) {
48 /* Assume only end is not page aligned */
47 u64 ramdisk_image = boot_params.hdr.ramdisk_image; 49 u64 ramdisk_image = boot_params.hdr.ramdisk_image;
48 u64 ramdisk_size = boot_params.hdr.ramdisk_size; 50 u64 ramdisk_size = boot_params.hdr.ramdisk_size;
49 u64 ramdisk_end = ramdisk_image + ramdisk_size; 51 u64 ramdisk_end = PAGE_ALIGN(ramdisk_image + ramdisk_size);
50 reserve_early(ramdisk_image, ramdisk_end, "RAMDISK"); 52 reserve_early(ramdisk_image, ramdisk_end, "RAMDISK");
51 } 53 }
52#endif 54#endif
diff --git a/arch/x86/kernel/head64.c b/arch/x86/kernel/head64.c
index b5a9896ca1e7..7147143fd614 100644
--- a/arch/x86/kernel/head64.c
+++ b/arch/x86/kernel/head64.c
@@ -103,9 +103,10 @@ void __init x86_64_start_reservations(char *real_mode_data)
103#ifdef CONFIG_BLK_DEV_INITRD 103#ifdef CONFIG_BLK_DEV_INITRD
104 /* Reserve INITRD */ 104 /* Reserve INITRD */
105 if (boot_params.hdr.type_of_loader && boot_params.hdr.ramdisk_image) { 105 if (boot_params.hdr.type_of_loader && boot_params.hdr.ramdisk_image) {
106 /* Assume only end is not page aligned */
106 unsigned long ramdisk_image = boot_params.hdr.ramdisk_image; 107 unsigned long ramdisk_image = boot_params.hdr.ramdisk_image;
107 unsigned long ramdisk_size = boot_params.hdr.ramdisk_size; 108 unsigned long ramdisk_size = boot_params.hdr.ramdisk_size;
108 unsigned long ramdisk_end = ramdisk_image + ramdisk_size; 109 unsigned long ramdisk_end = PAGE_ALIGN(ramdisk_image + ramdisk_size);
109 reserve_early(ramdisk_image, ramdisk_end, "RAMDISK"); 110 reserve_early(ramdisk_image, ramdisk_end, "RAMDISK");
110 } 111 }
111#endif 112#endif
diff --git a/arch/x86/kernel/hpet.c b/arch/x86/kernel/hpet.c
index ee4fa1bfcb33..23b4ecdffa9b 100644
--- a/arch/x86/kernel/hpet.c
+++ b/arch/x86/kernel/hpet.c
@@ -4,6 +4,7 @@
4#include <linux/sysdev.h> 4#include <linux/sysdev.h>
5#include <linux/delay.h> 5#include <linux/delay.h>
6#include <linux/errno.h> 6#include <linux/errno.h>
7#include <linux/slab.h>
7#include <linux/hpet.h> 8#include <linux/hpet.h>
8#include <linux/init.h> 9#include <linux/init.h>
9#include <linux/cpu.h> 10#include <linux/cpu.h>
@@ -399,9 +400,15 @@ static int hpet_next_event(unsigned long delta,
399 * then we might have a real hardware problem. We can not do 400 * 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 * much about it here, but at least alert the user/admin with
401 * a prominent warning. 402 * a prominent warning.
403 * An erratum on some chipsets (ICH9,..), results in comparator read
404 * immediately following a write returning old value. Workaround
405 * for this is to read this value second time, when first
406 * read returns old value.
402 */ 407 */
403 WARN_ONCE(hpet_readl(HPET_Tn_CMP(timer)) != cnt, 408 if (unlikely((u32)hpet_readl(HPET_Tn_CMP(timer)) != cnt)) {
409 WARN_ONCE(hpet_readl(HPET_Tn_CMP(timer)) != cnt,
404 KERN_WARNING "hpet: compare register read back failed.\n"); 410 KERN_WARNING "hpet: compare register read back failed.\n");
411 }
405 412
406 return (s32)(hpet_readl(HPET_COUNTER) - cnt) >= 0 ? -ETIME : 0; 413 return (s32)(hpet_readl(HPET_COUNTER) - cnt) >= 0 ? -ETIME : 0;
407} 414}
@@ -1143,6 +1150,7 @@ int hpet_set_periodic_freq(unsigned long freq)
1143 do_div(clc, freq); 1150 do_div(clc, freq);
1144 clc >>= hpet_clockevent.shift; 1151 clc >>= hpet_clockevent.shift;
1145 hpet_pie_delta = clc; 1152 hpet_pie_delta = clc;
1153 hpet_pie_limit = 0;
1146 } 1154 }
1147 return 1; 1155 return 1;
1148} 1156}
diff --git a/arch/x86/kernel/hw_breakpoint.c b/arch/x86/kernel/hw_breakpoint.c
index dca2802c666f..d6cc065f519f 100644
--- a/arch/x86/kernel/hw_breakpoint.c
+++ b/arch/x86/kernel/hw_breakpoint.c
@@ -344,13 +344,6 @@ int arch_validate_hwbkpt_settings(struct perf_event *bp,
344 } 344 }
345 345
346 /* 346 /*
347 * For kernel-addresses, either the address or symbol name can be
348 * specified.
349 */
350 if (info->name)
351 info->address = (unsigned long)
352 kallsyms_lookup_name(info->name);
353 /*
354 * Check that the low-order bits of the address are appropriate 347 * Check that the low-order bits of the address are appropriate
355 * for the alignment implied by len. 348 * for the alignment implied by len.
356 */ 349 */
@@ -535,8 +528,3 @@ void hw_breakpoint_pmu_read(struct perf_event *bp)
535{ 528{
536 /* TODO */ 529 /* TODO */
537} 530}
538
539void hw_breakpoint_pmu_unthrottle(struct perf_event *bp)
540{
541 /* TODO */
542}
diff --git a/arch/x86/kernel/i387.c b/arch/x86/kernel/i387.c
index c01a2b846d47..54c31c285488 100644
--- a/arch/x86/kernel/i387.c
+++ b/arch/x86/kernel/i387.c
@@ -8,6 +8,7 @@
8#include <linux/module.h> 8#include <linux/module.h>
9#include <linux/regset.h> 9#include <linux/regset.h>
10#include <linux/sched.h> 10#include <linux/sched.h>
11#include <linux/slab.h>
11 12
12#include <asm/sigcontext.h> 13#include <asm/sigcontext.h>
13#include <asm/processor.h> 14#include <asm/processor.h>
diff --git a/arch/x86/kernel/i8259.c b/arch/x86/kernel/i8259.c
index fb725ee15f55..7c9f02c130f3 100644
--- a/arch/x86/kernel/i8259.c
+++ b/arch/x86/kernel/i8259.c
@@ -5,7 +5,6 @@
5#include <linux/ioport.h> 5#include <linux/ioport.h>
6#include <linux/interrupt.h> 6#include <linux/interrupt.h>
7#include <linux/timex.h> 7#include <linux/timex.h>
8#include <linux/slab.h>
9#include <linux/random.h> 8#include <linux/random.h>
10#include <linux/init.h> 9#include <linux/init.h>
11#include <linux/kernel_stat.h> 10#include <linux/kernel_stat.h>
diff --git a/arch/x86/kernel/irqinit.c b/arch/x86/kernel/irqinit.c
index ef257fc2921b..0ed2d300cd46 100644
--- a/arch/x86/kernel/irqinit.c
+++ b/arch/x86/kernel/irqinit.c
@@ -5,7 +5,6 @@
5#include <linux/ioport.h> 5#include <linux/ioport.h>
6#include <linux/interrupt.h> 6#include <linux/interrupt.h>
7#include <linux/timex.h> 7#include <linux/timex.h>
8#include <linux/slab.h>
9#include <linux/random.h> 8#include <linux/random.h>
10#include <linux/kprobes.h> 9#include <linux/kprobes.h>
11#include <linux/init.h> 10#include <linux/init.h>
@@ -141,6 +140,28 @@ void __init init_IRQ(void)
141 x86_init.irqs.intr_init(); 140 x86_init.irqs.intr_init();
142} 141}
143 142
143/*
144 * Setup the vector to irq mappings.
145 */
146void setup_vector_irq(int cpu)
147{
148#ifndef CONFIG_X86_IO_APIC
149 int irq;
150
151 /*
152 * On most of the platforms, legacy PIC delivers the interrupts on the
153 * boot cpu. But there are certain platforms where PIC interrupts are
154 * delivered to multiple cpu's. If the legacy IRQ is handled by the
155 * legacy PIC, for the new cpu that is coming online, setup the static
156 * legacy vector to irq mapping:
157 */
158 for (irq = 0; irq < legacy_pic->nr_legacy_irqs; irq++)
159 per_cpu(vector_irq, cpu)[IRQ0_VECTOR + irq] = irq;
160#endif
161
162 __setup_vector_irq(cpu);
163}
164
144static void __init smp_intr_init(void) 165static void __init smp_intr_init(void)
145{ 166{
146#ifdef CONFIG_SMP 167#ifdef CONFIG_SMP
diff --git a/arch/x86/kernel/k8.c b/arch/x86/kernel/k8.c
index cbc4332a77b2..0f7bc20cfcde 100644
--- a/arch/x86/kernel/k8.c
+++ b/arch/x86/kernel/k8.c
@@ -2,8 +2,8 @@
2 * Shared support code for AMD K8 northbridges and derivates. 2 * Shared support code for AMD K8 northbridges and derivates.
3 * Copyright 2006 Andi Kleen, SUSE Labs. Subject to GPLv2. 3 * Copyright 2006 Andi Kleen, SUSE Labs. Subject to GPLv2.
4 */ 4 */
5#include <linux/gfp.h>
6#include <linux/types.h> 5#include <linux/types.h>
6#include <linux/slab.h>
7#include <linux/init.h> 7#include <linux/init.h>
8#include <linux/errno.h> 8#include <linux/errno.h>
9#include <linux/module.h> 9#include <linux/module.h>
@@ -121,3 +121,17 @@ void k8_flush_garts(void)
121} 121}
122EXPORT_SYMBOL_GPL(k8_flush_garts); 122EXPORT_SYMBOL_GPL(k8_flush_garts);
123 123
124static __init int init_k8_nbs(void)
125{
126 int err = 0;
127
128 err = cache_k8_northbridges();
129
130 if (err < 0)
131 printk(KERN_NOTICE "K8 NB: Cannot enumerate AMD northbridges.\n");
132
133 return err;
134}
135
136/* This has to go after the PCI subsystem */
137fs_initcall(init_k8_nbs);
diff --git a/arch/x86/kernel/kdebugfs.c b/arch/x86/kernel/kdebugfs.c
index e444357375ce..8afd9f321f10 100644
--- a/arch/x86/kernel/kdebugfs.c
+++ b/arch/x86/kernel/kdebugfs.c
@@ -9,6 +9,7 @@
9#include <linux/debugfs.h> 9#include <linux/debugfs.h>
10#include <linux/uaccess.h> 10#include <linux/uaccess.h>
11#include <linux/module.h> 11#include <linux/module.h>
12#include <linux/slab.h>
12#include <linux/init.h> 13#include <linux/init.h>
13#include <linux/stat.h> 14#include <linux/stat.h>
14#include <linux/io.h> 15#include <linux/io.h>
diff --git a/arch/x86/kernel/kgdb.c b/arch/x86/kernel/kgdb.c
index bfba6019d762..b2258ca91003 100644
--- a/arch/x86/kernel/kgdb.c
+++ b/arch/x86/kernel/kgdb.c
@@ -618,8 +618,8 @@ int kgdb_arch_init(void)
618 * portion of kgdb because this operation requires mutexs to 618 * portion of kgdb because this operation requires mutexs to
619 * complete. 619 * complete.
620 */ 620 */
621 hw_breakpoint_init(&attr);
621 attr.bp_addr = (unsigned long)kgdb_arch_init; 622 attr.bp_addr = (unsigned long)kgdb_arch_init;
622 attr.type = PERF_TYPE_BREAKPOINT;
623 attr.bp_len = HW_BREAKPOINT_LEN_1; 623 attr.bp_len = HW_BREAKPOINT_LEN_1;
624 attr.bp_type = HW_BREAKPOINT_W; 624 attr.bp_type = HW_BREAKPOINT_W;
625 attr.disabled = 1; 625 attr.disabled = 1;
diff --git a/arch/x86/kernel/ldt.c b/arch/x86/kernel/ldt.c
index ec6ef60cbd17..ea697263b373 100644
--- a/arch/x86/kernel/ldt.c
+++ b/arch/x86/kernel/ldt.c
@@ -7,6 +7,7 @@
7 */ 7 */
8 8
9#include <linux/errno.h> 9#include <linux/errno.h>
10#include <linux/gfp.h>
10#include <linux/sched.h> 11#include <linux/sched.h>
11#include <linux/string.h> 12#include <linux/string.h>
12#include <linux/mm.h> 13#include <linux/mm.h>
diff --git a/arch/x86/kernel/machine_kexec_64.c b/arch/x86/kernel/machine_kexec_64.c
index 4a8bb82248ae..035c8c529181 100644
--- a/arch/x86/kernel/machine_kexec_64.c
+++ b/arch/x86/kernel/machine_kexec_64.c
@@ -9,6 +9,7 @@
9#include <linux/mm.h> 9#include <linux/mm.h>
10#include <linux/kexec.h> 10#include <linux/kexec.h>
11#include <linux/string.h> 11#include <linux/string.h>
12#include <linux/gfp.h>
12#include <linux/reboot.h> 13#include <linux/reboot.h>
13#include <linux/numa.h> 14#include <linux/numa.h>
14#include <linux/ftrace.h> 15#include <linux/ftrace.h>
diff --git a/arch/x86/kernel/mca_32.c b/arch/x86/kernel/mca_32.c
index 845d80ce1ef1..63eaf6596233 100644
--- a/arch/x86/kernel/mca_32.c
+++ b/arch/x86/kernel/mca_32.c
@@ -42,6 +42,7 @@
42#include <linux/kernel.h> 42#include <linux/kernel.h>
43#include <linux/mca.h> 43#include <linux/mca.h>
44#include <linux/kprobes.h> 44#include <linux/kprobes.h>
45#include <linux/slab.h>
45#include <asm/system.h> 46#include <asm/system.h>
46#include <asm/io.h> 47#include <asm/io.h>
47#include <linux/proc_fs.h> 48#include <linux/proc_fs.h>
diff --git a/arch/x86/kernel/module.c b/arch/x86/kernel/module.c
index 89f386f044e4..e0bc186d7501 100644
--- a/arch/x86/kernel/module.c
+++ b/arch/x86/kernel/module.c
@@ -23,6 +23,7 @@
23#include <linux/kernel.h> 23#include <linux/kernel.h>
24#include <linux/bug.h> 24#include <linux/bug.h>
25#include <linux/mm.h> 25#include <linux/mm.h>
26#include <linux/gfp.h>
26 27
27#include <asm/system.h> 28#include <asm/system.h>
28#include <asm/page.h> 29#include <asm/page.h>
diff --git a/arch/x86/kernel/mpparse.c b/arch/x86/kernel/mpparse.c
index a2c1edd2d3ac..e81030f71a8f 100644
--- a/arch/x86/kernel/mpparse.c
+++ b/arch/x86/kernel/mpparse.c
@@ -664,7 +664,7 @@ static void __init smp_reserve_memory(struct mpf_intel *mpf)
664{ 664{
665 unsigned long size = get_mpc_size(mpf->physptr); 665 unsigned long size = get_mpc_size(mpf->physptr);
666 666
667 reserve_early(mpf->physptr, mpf->physptr+size, "MP-table mpc"); 667 reserve_early_overlap_ok(mpf->physptr, mpf->physptr+size, "MP-table mpc");
668} 668}
669 669
670static int __init smp_scan_config(unsigned long base, unsigned long length) 670static int __init smp_scan_config(unsigned long base, unsigned long length)
@@ -693,7 +693,7 @@ static int __init smp_scan_config(unsigned long base, unsigned long length)
693 mpf, (u64)virt_to_phys(mpf)); 693 mpf, (u64)virt_to_phys(mpf));
694 694
695 mem = virt_to_phys(mpf); 695 mem = virt_to_phys(mpf);
696 reserve_early(mem, mem + sizeof(*mpf), "MP-table mpf"); 696 reserve_early_overlap_ok(mem, mem + sizeof(*mpf), "MP-table mpf");
697 if (mpf->physptr) 697 if (mpf->physptr)
698 smp_reserve_memory(mpf); 698 smp_reserve_memory(mpf);
699 699
diff --git a/arch/x86/kernel/msr.c b/arch/x86/kernel/msr.c
index 206735ac8cbd..4d4468e9f47c 100644
--- a/arch/x86/kernel/msr.c
+++ b/arch/x86/kernel/msr.c
@@ -37,6 +37,7 @@
37#include <linux/cpu.h> 37#include <linux/cpu.h>
38#include <linux/notifier.h> 38#include <linux/notifier.h>
39#include <linux/uaccess.h> 39#include <linux/uaccess.h>
40#include <linux/gfp.h>
40 41
41#include <asm/processor.h> 42#include <asm/processor.h>
42#include <asm/msr.h> 43#include <asm/msr.h>
diff --git a/arch/x86/kernel/pci-dma.c b/arch/x86/kernel/pci-dma.c
index a4ac764a6880..4b7e3d8b01dd 100644
--- a/arch/x86/kernel/pci-dma.c
+++ b/arch/x86/kernel/pci-dma.c
@@ -2,6 +2,7 @@
2#include <linux/dma-debug.h> 2#include <linux/dma-debug.h>
3#include <linux/dmar.h> 3#include <linux/dmar.h>
4#include <linux/bootmem.h> 4#include <linux/bootmem.h>
5#include <linux/gfp.h>
5#include <linux/pci.h> 6#include <linux/pci.h>
6#include <linux/kmemleak.h> 7#include <linux/kmemleak.h>
7 8
diff --git a/arch/x86/kernel/pci-gart_64.c b/arch/x86/kernel/pci-gart_64.c
index 34de53b46f87..0f7f130caa67 100644
--- a/arch/x86/kernel/pci-gart_64.c
+++ b/arch/x86/kernel/pci-gart_64.c
@@ -29,6 +29,7 @@
29#include <linux/iommu-helper.h> 29#include <linux/iommu-helper.h>
30#include <linux/sysdev.h> 30#include <linux/sysdev.h>
31#include <linux/io.h> 31#include <linux/io.h>
32#include <linux/gfp.h>
32#include <asm/atomic.h> 33#include <asm/atomic.h>
33#include <asm/mtrr.h> 34#include <asm/mtrr.h>
34#include <asm/pgtable.h> 35#include <asm/pgtable.h>
@@ -564,6 +565,9 @@ static void enable_gart_translations(void)
564 565
565 enable_gart_translation(dev, __pa(agp_gatt_table)); 566 enable_gart_translation(dev, __pa(agp_gatt_table));
566 } 567 }
568
569 /* Flush the GART-TLB to remove stale entries */
570 k8_flush_garts();
567} 571}
568 572
569/* 573/*
@@ -735,7 +739,7 @@ int __init gart_iommu_init(void)
735 unsigned long scratch; 739 unsigned long scratch;
736 long i; 740 long i;
737 741
738 if (cache_k8_northbridges() < 0 || num_k8_northbridges == 0) 742 if (num_k8_northbridges == 0)
739 return 0; 743 return 0;
740 744
741#ifndef CONFIG_AGP_AMD64 745#ifndef CONFIG_AGP_AMD64
diff --git a/arch/x86/kernel/pci-nommu.c b/arch/x86/kernel/pci-nommu.c
index 22be12b60a8f..3af4af810c07 100644
--- a/arch/x86/kernel/pci-nommu.c
+++ b/arch/x86/kernel/pci-nommu.c
@@ -4,6 +4,7 @@
4#include <linux/scatterlist.h> 4#include <linux/scatterlist.h>
5#include <linux/string.h> 5#include <linux/string.h>
6#include <linux/init.h> 6#include <linux/init.h>
7#include <linux/gfp.h>
7#include <linux/pci.h> 8#include <linux/pci.h>
8#include <linux/mm.h> 9#include <linux/mm.h>
9 10
diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c
index 02d678065d7d..28ad9f4d8b94 100644
--- a/arch/x86/kernel/process.c
+++ b/arch/x86/kernel/process.c
@@ -526,21 +526,37 @@ static int __cpuinit mwait_usable(const struct cpuinfo_x86 *c)
526} 526}
527 527
528/* 528/*
529 * Check for AMD CPUs, which have potentially C1E support 529 * Check for AMD CPUs, where APIC timer interrupt does not wake up CPU from C1e.
530 * For more information see
531 * - Erratum #400 for NPT family 0xf and family 0x10 CPUs
532 * - Erratum #365 for family 0x11 (not affected because C1e not in use)
530 */ 533 */
531static int __cpuinit check_c1e_idle(const struct cpuinfo_x86 *c) 534static int __cpuinit check_c1e_idle(const struct cpuinfo_x86 *c)
532{ 535{
536 u64 val;
533 if (c->x86_vendor != X86_VENDOR_AMD) 537 if (c->x86_vendor != X86_VENDOR_AMD)
534 return 0; 538 goto no_c1e_idle;
535
536 if (c->x86 < 0x0F)
537 return 0;
538 539
539 /* Family 0x0f models < rev F do not have C1E */ 540 /* Family 0x0f models < rev F do not have C1E */
540 if (c->x86 == 0x0f && c->x86_model < 0x40) 541 if (c->x86 == 0x0F && c->x86_model >= 0x40)
541 return 0; 542 return 1;
542 543
543 return 1; 544 if (c->x86 == 0x10) {
545 /*
546 * check OSVW bit for CPUs that are not affected
547 * by erratum #400
548 */
549 rdmsrl(MSR_AMD64_OSVW_ID_LENGTH, val);
550 if (val >= 2) {
551 rdmsrl(MSR_AMD64_OSVW_STATUS, val);
552 if (!(val & BIT(1)))
553 goto no_c1e_idle;
554 }
555 return 1;
556 }
557
558no_c1e_idle:
559 return 0;
544} 560}
545 561
546static cpumask_var_t c1e_mask; 562static cpumask_var_t c1e_mask;
@@ -607,7 +623,7 @@ void __cpuinit select_idle_routine(const struct cpuinfo_x86 *c)
607{ 623{
608#ifdef CONFIG_SMP 624#ifdef CONFIG_SMP
609 if (pm_idle == poll_idle && smp_num_siblings > 1) { 625 if (pm_idle == poll_idle && smp_num_siblings > 1) {
610 printk(KERN_WARNING "WARNING: polling idle and HT enabled," 626 printk_once(KERN_WARNING "WARNING: polling idle and HT enabled,"
611 " performance may degrade.\n"); 627 " performance may degrade.\n");
612 } 628 }
613#endif 629#endif
diff --git a/arch/x86/kernel/ptrace.c b/arch/x86/kernel/ptrace.c
index a503b1fd04e5..2e9b55027b7e 100644
--- a/arch/x86/kernel/ptrace.c
+++ b/arch/x86/kernel/ptrace.c
@@ -12,6 +12,7 @@
12#include <linux/mm.h> 12#include <linux/mm.h>
13#include <linux/smp.h> 13#include <linux/smp.h>
14#include <linux/errno.h> 14#include <linux/errno.h>
15#include <linux/slab.h>
15#include <linux/ptrace.h> 16#include <linux/ptrace.h>
16#include <linux/regset.h> 17#include <linux/regset.h>
17#include <linux/tracehook.h> 18#include <linux/tracehook.h>
diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c
index 5d7ba1a449bd..c4851eff57b3 100644
--- a/arch/x86/kernel/setup.c
+++ b/arch/x86/kernel/setup.c
@@ -55,7 +55,6 @@
55#include <linux/stddef.h> 55#include <linux/stddef.h>
56#include <linux/unistd.h> 56#include <linux/unistd.h>
57#include <linux/ptrace.h> 57#include <linux/ptrace.h>
58#include <linux/slab.h>
59#include <linux/user.h> 58#include <linux/user.h>
60#include <linux/delay.h> 59#include <linux/delay.h>
61 60
@@ -314,16 +313,17 @@ static void __init reserve_brk(void)
314#define MAX_MAP_CHUNK (NR_FIX_BTMAPS << PAGE_SHIFT) 313#define MAX_MAP_CHUNK (NR_FIX_BTMAPS << PAGE_SHIFT)
315static void __init relocate_initrd(void) 314static void __init relocate_initrd(void)
316{ 315{
317 316 /* Assume only end is not page aligned */
318 u64 ramdisk_image = boot_params.hdr.ramdisk_image; 317 u64 ramdisk_image = boot_params.hdr.ramdisk_image;
319 u64 ramdisk_size = boot_params.hdr.ramdisk_size; 318 u64 ramdisk_size = boot_params.hdr.ramdisk_size;
319 u64 area_size = PAGE_ALIGN(ramdisk_size);
320 u64 end_of_lowmem = max_low_pfn_mapped << PAGE_SHIFT; 320 u64 end_of_lowmem = max_low_pfn_mapped << PAGE_SHIFT;
321 u64 ramdisk_here; 321 u64 ramdisk_here;
322 unsigned long slop, clen, mapaddr; 322 unsigned long slop, clen, mapaddr;
323 char *p, *q; 323 char *p, *q;
324 324
325 /* We need to move the initrd down into lowmem */ 325 /* We need to move the initrd down into lowmem */
326 ramdisk_here = find_e820_area(0, end_of_lowmem, ramdisk_size, 326 ramdisk_here = find_e820_area(0, end_of_lowmem, area_size,
327 PAGE_SIZE); 327 PAGE_SIZE);
328 328
329 if (ramdisk_here == -1ULL) 329 if (ramdisk_here == -1ULL)
@@ -332,7 +332,7 @@ static void __init relocate_initrd(void)
332 332
333 /* Note: this includes all the lowmem currently occupied by 333 /* Note: this includes all the lowmem currently occupied by
334 the initrd, we rely on that fact to keep the data intact. */ 334 the initrd, we rely on that fact to keep the data intact. */
335 reserve_early(ramdisk_here, ramdisk_here + ramdisk_size, 335 reserve_early(ramdisk_here, ramdisk_here + area_size,
336 "NEW RAMDISK"); 336 "NEW RAMDISK");
337 initrd_start = ramdisk_here + PAGE_OFFSET; 337 initrd_start = ramdisk_here + PAGE_OFFSET;
338 initrd_end = initrd_start + ramdisk_size; 338 initrd_end = initrd_start + ramdisk_size;
@@ -376,9 +376,10 @@ static void __init relocate_initrd(void)
376 376
377static void __init reserve_initrd(void) 377static void __init reserve_initrd(void)
378{ 378{
379 /* Assume only end is not page aligned */
379 u64 ramdisk_image = boot_params.hdr.ramdisk_image; 380 u64 ramdisk_image = boot_params.hdr.ramdisk_image;
380 u64 ramdisk_size = boot_params.hdr.ramdisk_size; 381 u64 ramdisk_size = boot_params.hdr.ramdisk_size;
381 u64 ramdisk_end = ramdisk_image + ramdisk_size; 382 u64 ramdisk_end = PAGE_ALIGN(ramdisk_image + ramdisk_size);
382 u64 end_of_lowmem = max_low_pfn_mapped << PAGE_SHIFT; 383 u64 end_of_lowmem = max_low_pfn_mapped << PAGE_SHIFT;
383 384
384 if (!boot_params.hdr.type_of_loader || 385 if (!boot_params.hdr.type_of_loader ||
@@ -606,6 +607,16 @@ static int __init setup_elfcorehdr(char *arg)
606early_param("elfcorehdr", setup_elfcorehdr); 607early_param("elfcorehdr", setup_elfcorehdr);
607#endif 608#endif
608 609
610static __init void reserve_ibft_region(void)
611{
612 unsigned long addr, size = 0;
613
614 addr = find_ibft_region(&size);
615
616 if (size)
617 reserve_early_overlap_ok(addr, addr + size, "ibft");
618}
619
609#ifdef CONFIG_X86_RESERVE_LOW_64K 620#ifdef CONFIG_X86_RESERVE_LOW_64K
610static int __init dmi_low_memory_corruption(const struct dmi_system_id *d) 621static int __init dmi_low_memory_corruption(const struct dmi_system_id *d)
611{ 622{
@@ -908,6 +919,8 @@ void __init setup_arch(char **cmdline_p)
908 */ 919 */
909 find_smp_config(); 920 find_smp_config();
910 921
922 reserve_ibft_region();
923
911 reserve_trampoline_memory(); 924 reserve_trampoline_memory();
912 925
913#ifdef CONFIG_ACPI_SLEEP 926#ifdef CONFIG_ACPI_SLEEP
@@ -975,8 +988,6 @@ void __init setup_arch(char **cmdline_p)
975 988
976 dma32_reserve_bootmem(); 989 dma32_reserve_bootmem();
977 990
978 reserve_ibft_region();
979
980#ifdef CONFIG_KVM_CLOCK 991#ifdef CONFIG_KVM_CLOCK
981 kvmclock_init(); 992 kvmclock_init();
982#endif 993#endif
diff --git a/arch/x86/kernel/smp.c b/arch/x86/kernel/smp.c
index ec1de97600e7..d801210945d6 100644
--- a/arch/x86/kernel/smp.c
+++ b/arch/x86/kernel/smp.c
@@ -21,6 +21,7 @@
21#include <linux/cache.h> 21#include <linux/cache.h>
22#include <linux/interrupt.h> 22#include <linux/interrupt.h>
23#include <linux/cpu.h> 23#include <linux/cpu.h>
24#include <linux/gfp.h>
24 25
25#include <asm/mtrr.h> 26#include <asm/mtrr.h>
26#include <asm/tlbflush.h> 27#include <asm/tlbflush.h>
diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c
index a02e80c3c54b..763d815e27a0 100644
--- a/arch/x86/kernel/smpboot.c
+++ b/arch/x86/kernel/smpboot.c
@@ -49,6 +49,7 @@
49#include <linux/nmi.h> 49#include <linux/nmi.h>
50#include <linux/tboot.h> 50#include <linux/tboot.h>
51#include <linux/stackprotector.h> 51#include <linux/stackprotector.h>
52#include <linux/gfp.h>
52 53
53#include <asm/acpi.h> 54#include <asm/acpi.h>
54#include <asm/desc.h> 55#include <asm/desc.h>
@@ -242,12 +243,10 @@ static void __cpuinit smp_callin(void)
242 end_local_APIC_setup(); 243 end_local_APIC_setup();
243 map_cpu_to_logical_apicid(); 244 map_cpu_to_logical_apicid();
244 245
245 notify_cpu_starting(cpuid);
246
247 /* 246 /*
248 * Need to setup vector mappings before we enable interrupts. 247 * Need to setup vector mappings before we enable interrupts.
249 */ 248 */
250 __setup_vector_irq(smp_processor_id()); 249 setup_vector_irq(smp_processor_id());
251 /* 250 /*
252 * Get our bogomips. 251 * Get our bogomips.
253 * 252 *
@@ -264,6 +263,8 @@ static void __cpuinit smp_callin(void)
264 */ 263 */
265 smp_store_cpu_info(cpuid); 264 smp_store_cpu_info(cpuid);
266 265
266 notify_cpu_starting(cpuid);
267
267 /* 268 /*
268 * Allow the master to continue. 269 * Allow the master to continue.
269 */ 270 */
diff --git a/arch/x86/kernel/tlb_uv.c b/arch/x86/kernel/tlb_uv.c
index 364d015efebc..17b03dd3a6b5 100644
--- a/arch/x86/kernel/tlb_uv.c
+++ b/arch/x86/kernel/tlb_uv.c
@@ -9,6 +9,7 @@
9#include <linux/seq_file.h> 9#include <linux/seq_file.h>
10#include <linux/proc_fs.h> 10#include <linux/proc_fs.h>
11#include <linux/kernel.h> 11#include <linux/kernel.h>
12#include <linux/slab.h>
12 13
13#include <asm/mmu_context.h> 14#include <asm/mmu_context.h>
14#include <asm/uv/uv.h> 15#include <asm/uv/uv.h>
diff --git a/arch/x86/kernel/uv_irq.c b/arch/x86/kernel/uv_irq.c
index ece73d8e3240..1d40336b030a 100644
--- a/arch/x86/kernel/uv_irq.c
+++ b/arch/x86/kernel/uv_irq.c
@@ -10,6 +10,7 @@
10 10
11#include <linux/module.h> 11#include <linux/module.h>
12#include <linux/rbtree.h> 12#include <linux/rbtree.h>
13#include <linux/slab.h>
13#include <linux/irq.h> 14#include <linux/irq.h>
14 15
15#include <asm/apic.h> 16#include <asm/apic.h>
diff --git a/arch/x86/kernel/uv_time.c b/arch/x86/kernel/uv_time.c
index 2b75ef638dbc..56e421bc379b 100644
--- a/arch/x86/kernel/uv_time.c
+++ b/arch/x86/kernel/uv_time.c
@@ -19,6 +19,7 @@
19 * Copyright (c) Dimitri Sivanich 19 * Copyright (c) Dimitri Sivanich
20 */ 20 */
21#include <linux/clockchips.h> 21#include <linux/clockchips.h>
22#include <linux/slab.h>
22 23
23#include <asm/uv/uv_mmrs.h> 24#include <asm/uv/uv_mmrs.h>
24#include <asm/uv/uv_hub.h> 25#include <asm/uv/uv_hub.h>
diff --git a/arch/x86/kernel/vmi_32.c b/arch/x86/kernel/vmi_32.c
index 7dd599deca4a..ce9fbacb7526 100644
--- a/arch/x86/kernel/vmi_32.c
+++ b/arch/x86/kernel/vmi_32.c
@@ -28,6 +28,7 @@
28#include <linux/mm.h> 28#include <linux/mm.h>
29#include <linux/highmem.h> 29#include <linux/highmem.h>
30#include <linux/sched.h> 30#include <linux/sched.h>
31#include <linux/gfp.h>
31#include <asm/vmi.h> 32#include <asm/vmi.h>
32#include <asm/io.h> 33#include <asm/io.h>
33#include <asm/fixmap.h> 34#include <asm/fixmap.h>
diff --git a/arch/x86/kernel/vmlinux.lds.S b/arch/x86/kernel/vmlinux.lds.S
index 44879df55696..2cc249718c46 100644
--- a/arch/x86/kernel/vmlinux.lds.S
+++ b/arch/x86/kernel/vmlinux.lds.S
@@ -291,8 +291,8 @@ SECTIONS
291 .smp_locks : AT(ADDR(.smp_locks) - LOAD_OFFSET) { 291 .smp_locks : AT(ADDR(.smp_locks) - LOAD_OFFSET) {
292 __smp_locks = .; 292 __smp_locks = .;
293 *(.smp_locks) 293 *(.smp_locks)
294 __smp_locks_end = .;
295 . = ALIGN(PAGE_SIZE); 294 . = ALIGN(PAGE_SIZE);
295 __smp_locks_end = .;
296 } 296 }
297 297
298#ifdef CONFIG_X86_64 298#ifdef CONFIG_X86_64
diff --git a/arch/x86/kvm/i8254.c b/arch/x86/kvm/i8254.c
index 294698b6daff..0150affad25d 100644
--- a/arch/x86/kvm/i8254.c
+++ b/arch/x86/kvm/i8254.c
@@ -32,6 +32,7 @@
32#define pr_fmt(fmt) "pit: " fmt 32#define pr_fmt(fmt) "pit: " fmt
33 33
34#include <linux/kvm_host.h> 34#include <linux/kvm_host.h>
35#include <linux/slab.h>
35 36
36#include "irq.h" 37#include "irq.h"
37#include "i8254.h" 38#include "i8254.h"
diff --git a/arch/x86/kvm/i8259.c b/arch/x86/kvm/i8259.c
index 07771da85de5..a790fa128a9f 100644
--- a/arch/x86/kvm/i8259.c
+++ b/arch/x86/kvm/i8259.c
@@ -26,6 +26,7 @@
26 * Port from Qemu. 26 * Port from Qemu.
27 */ 27 */
28#include <linux/mm.h> 28#include <linux/mm.h>
29#include <linux/slab.h>
29#include <linux/bitops.h> 30#include <linux/bitops.h>
30#include "irq.h" 31#include "irq.h"
31 32
diff --git a/arch/x86/kvm/lapic.c b/arch/x86/kvm/lapic.c
index 4b224f90087b..1eb7a4ae0c9c 100644
--- a/arch/x86/kvm/lapic.c
+++ b/arch/x86/kvm/lapic.c
@@ -26,6 +26,7 @@
26#include <linux/io.h> 26#include <linux/io.h>
27#include <linux/module.h> 27#include <linux/module.h>
28#include <linux/math64.h> 28#include <linux/math64.h>
29#include <linux/slab.h>
29#include <asm/processor.h> 30#include <asm/processor.h>
30#include <asm/msr.h> 31#include <asm/msr.h>
31#include <asm/page.h> 32#include <asm/page.h>
diff --git a/arch/x86/kvm/mmu.c b/arch/x86/kvm/mmu.c
index 741373e8ca77..19a8906bcaa2 100644
--- a/arch/x86/kvm/mmu.c
+++ b/arch/x86/kvm/mmu.c
@@ -31,6 +31,7 @@
31#include <linux/hugetlb.h> 31#include <linux/hugetlb.h>
32#include <linux/compiler.h> 32#include <linux/compiler.h>
33#include <linux/srcu.h> 33#include <linux/srcu.h>
34#include <linux/slab.h>
34 35
35#include <asm/page.h> 36#include <asm/page.h>
36#include <asm/cmpxchg.h> 37#include <asm/cmpxchg.h>
@@ -1489,8 +1490,8 @@ static int mmu_zap_unsync_children(struct kvm *kvm,
1489 for_each_sp(pages, sp, parents, i) { 1490 for_each_sp(pages, sp, parents, i) {
1490 kvm_mmu_zap_page(kvm, sp); 1491 kvm_mmu_zap_page(kvm, sp);
1491 mmu_pages_clear_parents(&parents); 1492 mmu_pages_clear_parents(&parents);
1493 zapped++;
1492 } 1494 }
1493 zapped += pages.nr;
1494 kvm_mmu_pages_init(parent, &parents, &pages); 1495 kvm_mmu_pages_init(parent, &parents, &pages);
1495 } 1496 }
1496 1497
@@ -1541,14 +1542,16 @@ void kvm_mmu_change_mmu_pages(struct kvm *kvm, unsigned int kvm_nr_mmu_pages)
1541 */ 1542 */
1542 1543
1543 if (used_pages > kvm_nr_mmu_pages) { 1544 if (used_pages > kvm_nr_mmu_pages) {
1544 while (used_pages > kvm_nr_mmu_pages) { 1545 while (used_pages > kvm_nr_mmu_pages &&
1546 !list_empty(&kvm->arch.active_mmu_pages)) {
1545 struct kvm_mmu_page *page; 1547 struct kvm_mmu_page *page;
1546 1548
1547 page = container_of(kvm->arch.active_mmu_pages.prev, 1549 page = container_of(kvm->arch.active_mmu_pages.prev,
1548 struct kvm_mmu_page, link); 1550 struct kvm_mmu_page, link);
1549 kvm_mmu_zap_page(kvm, page); 1551 used_pages -= kvm_mmu_zap_page(kvm, page);
1550 used_pages--; 1552 used_pages--;
1551 } 1553 }
1554 kvm_nr_mmu_pages = used_pages;
1552 kvm->arch.n_free_mmu_pages = 0; 1555 kvm->arch.n_free_mmu_pages = 0;
1553 } 1556 }
1554 else 1557 else
@@ -1595,7 +1598,8 @@ static void mmu_unshadow(struct kvm *kvm, gfn_t gfn)
1595 && !sp->role.invalid) { 1598 && !sp->role.invalid) {
1596 pgprintk("%s: zap %lx %x\n", 1599 pgprintk("%s: zap %lx %x\n",
1597 __func__, gfn, sp->role.word); 1600 __func__, gfn, sp->role.word);
1598 kvm_mmu_zap_page(kvm, sp); 1601 if (kvm_mmu_zap_page(kvm, sp))
1602 nn = bucket->first;
1599 } 1603 }
1600 } 1604 }
1601} 1605}
diff --git a/arch/x86/kvm/svm.c b/arch/x86/kvm/svm.c
index 52f78dd03010..2ba58206812a 100644
--- a/arch/x86/kvm/svm.c
+++ b/arch/x86/kvm/svm.c
@@ -26,6 +26,7 @@
26#include <linux/highmem.h> 26#include <linux/highmem.h>
27#include <linux/sched.h> 27#include <linux/sched.h>
28#include <linux/ftrace_event.h> 28#include <linux/ftrace_event.h>
29#include <linux/slab.h>
29 30
30#include <asm/desc.h> 31#include <asm/desc.h>
31 32
@@ -705,29 +706,28 @@ static struct kvm_vcpu *svm_create_vcpu(struct kvm *kvm, unsigned int id)
705 if (err) 706 if (err)
706 goto free_svm; 707 goto free_svm;
707 708
709 err = -ENOMEM;
708 page = alloc_page(GFP_KERNEL); 710 page = alloc_page(GFP_KERNEL);
709 if (!page) { 711 if (!page)
710 err = -ENOMEM;
711 goto uninit; 712 goto uninit;
712 }
713 713
714 err = -ENOMEM;
715 msrpm_pages = alloc_pages(GFP_KERNEL, MSRPM_ALLOC_ORDER); 714 msrpm_pages = alloc_pages(GFP_KERNEL, MSRPM_ALLOC_ORDER);
716 if (!msrpm_pages) 715 if (!msrpm_pages)
717 goto uninit; 716 goto free_page1;
718 717
719 nested_msrpm_pages = alloc_pages(GFP_KERNEL, MSRPM_ALLOC_ORDER); 718 nested_msrpm_pages = alloc_pages(GFP_KERNEL, MSRPM_ALLOC_ORDER);
720 if (!nested_msrpm_pages) 719 if (!nested_msrpm_pages)
721 goto uninit; 720 goto free_page2;
722
723 svm->msrpm = page_address(msrpm_pages);
724 svm_vcpu_init_msrpm(svm->msrpm);
725 721
726 hsave_page = alloc_page(GFP_KERNEL); 722 hsave_page = alloc_page(GFP_KERNEL);
727 if (!hsave_page) 723 if (!hsave_page)
728 goto uninit; 724 goto free_page3;
725
729 svm->nested.hsave = page_address(hsave_page); 726 svm->nested.hsave = page_address(hsave_page);
730 727
728 svm->msrpm = page_address(msrpm_pages);
729 svm_vcpu_init_msrpm(svm->msrpm);
730
731 svm->nested.msrpm = page_address(nested_msrpm_pages); 731 svm->nested.msrpm = page_address(nested_msrpm_pages);
732 732
733 svm->vmcb = page_address(page); 733 svm->vmcb = page_address(page);
@@ -743,6 +743,12 @@ static struct kvm_vcpu *svm_create_vcpu(struct kvm *kvm, unsigned int id)
743 743
744 return &svm->vcpu; 744 return &svm->vcpu;
745 745
746free_page3:
747 __free_pages(nested_msrpm_pages, MSRPM_ALLOC_ORDER);
748free_page2:
749 __free_pages(msrpm_pages, MSRPM_ALLOC_ORDER);
750free_page1:
751 __free_page(page);
746uninit: 752uninit:
747 kvm_vcpu_uninit(&svm->vcpu); 753 kvm_vcpu_uninit(&svm->vcpu);
748free_svm: 754free_svm:
diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
index 14873b9f8430..bc933cfb4e66 100644
--- a/arch/x86/kvm/vmx.c
+++ b/arch/x86/kvm/vmx.c
@@ -26,6 +26,7 @@
26#include <linux/sched.h> 26#include <linux/sched.h>
27#include <linux/moduleparam.h> 27#include <linux/moduleparam.h>
28#include <linux/ftrace_event.h> 28#include <linux/ftrace_event.h>
29#include <linux/slab.h>
29#include "kvm_cache_regs.h" 30#include "kvm_cache_regs.h"
30#include "x86.h" 31#include "x86.h"
31 32
@@ -76,6 +77,8 @@ module_param(emulate_invalid_guest_state, bool, S_IRUGO);
76#define KVM_PMODE_VM_CR4_ALWAYS_ON (X86_CR4_PAE | X86_CR4_VMXE) 77#define KVM_PMODE_VM_CR4_ALWAYS_ON (X86_CR4_PAE | X86_CR4_VMXE)
77#define KVM_RMODE_VM_CR4_ALWAYS_ON (X86_CR4_VME | X86_CR4_PAE | X86_CR4_VMXE) 78#define KVM_RMODE_VM_CR4_ALWAYS_ON (X86_CR4_VME | X86_CR4_PAE | X86_CR4_VMXE)
78 79
80#define RMODE_GUEST_OWNED_EFLAGS_BITS (~(X86_EFLAGS_IOPL | X86_EFLAGS_VM))
81
79/* 82/*
80 * These 2 parameters are used to config the controls for Pause-Loop Exiting: 83 * These 2 parameters are used to config the controls for Pause-Loop Exiting:
81 * ple_gap: upper bound on the amount of time between two successive 84 * ple_gap: upper bound on the amount of time between two successive
@@ -130,7 +133,7 @@ struct vcpu_vmx {
130 } host_state; 133 } host_state;
131 struct { 134 struct {
132 int vm86_active; 135 int vm86_active;
133 u8 save_iopl; 136 ulong save_rflags;
134 struct kvm_save_segment { 137 struct kvm_save_segment {
135 u16 selector; 138 u16 selector;
136 unsigned long base; 139 unsigned long base;
@@ -817,18 +820,23 @@ static void vmx_fpu_deactivate(struct kvm_vcpu *vcpu)
817 820
818static unsigned long vmx_get_rflags(struct kvm_vcpu *vcpu) 821static unsigned long vmx_get_rflags(struct kvm_vcpu *vcpu)
819{ 822{
820 unsigned long rflags; 823 unsigned long rflags, save_rflags;
821 824
822 rflags = vmcs_readl(GUEST_RFLAGS); 825 rflags = vmcs_readl(GUEST_RFLAGS);
823 if (to_vmx(vcpu)->rmode.vm86_active) 826 if (to_vmx(vcpu)->rmode.vm86_active) {
824 rflags &= ~(unsigned long)(X86_EFLAGS_IOPL | X86_EFLAGS_VM); 827 rflags &= RMODE_GUEST_OWNED_EFLAGS_BITS;
828 save_rflags = to_vmx(vcpu)->rmode.save_rflags;
829 rflags |= save_rflags & ~RMODE_GUEST_OWNED_EFLAGS_BITS;
830 }
825 return rflags; 831 return rflags;
826} 832}
827 833
828static void vmx_set_rflags(struct kvm_vcpu *vcpu, unsigned long rflags) 834static void vmx_set_rflags(struct kvm_vcpu *vcpu, unsigned long rflags)
829{ 835{
830 if (to_vmx(vcpu)->rmode.vm86_active) 836 if (to_vmx(vcpu)->rmode.vm86_active) {
837 to_vmx(vcpu)->rmode.save_rflags = rflags;
831 rflags |= X86_EFLAGS_IOPL | X86_EFLAGS_VM; 838 rflags |= X86_EFLAGS_IOPL | X86_EFLAGS_VM;
839 }
832 vmcs_writel(GUEST_RFLAGS, rflags); 840 vmcs_writel(GUEST_RFLAGS, rflags);
833} 841}
834 842
@@ -1482,8 +1490,8 @@ static void enter_pmode(struct kvm_vcpu *vcpu)
1482 vmcs_write32(GUEST_TR_AR_BYTES, vmx->rmode.tr.ar); 1490 vmcs_write32(GUEST_TR_AR_BYTES, vmx->rmode.tr.ar);
1483 1491
1484 flags = vmcs_readl(GUEST_RFLAGS); 1492 flags = vmcs_readl(GUEST_RFLAGS);
1485 flags &= ~(X86_EFLAGS_IOPL | X86_EFLAGS_VM); 1493 flags &= RMODE_GUEST_OWNED_EFLAGS_BITS;
1486 flags |= (vmx->rmode.save_iopl << IOPL_SHIFT); 1494 flags |= vmx->rmode.save_rflags & ~RMODE_GUEST_OWNED_EFLAGS_BITS;
1487 vmcs_writel(GUEST_RFLAGS, flags); 1495 vmcs_writel(GUEST_RFLAGS, flags);
1488 1496
1489 vmcs_writel(GUEST_CR4, (vmcs_readl(GUEST_CR4) & ~X86_CR4_VME) | 1497 vmcs_writel(GUEST_CR4, (vmcs_readl(GUEST_CR4) & ~X86_CR4_VME) |
@@ -1556,8 +1564,7 @@ static void enter_rmode(struct kvm_vcpu *vcpu)
1556 vmcs_write32(GUEST_TR_AR_BYTES, 0x008b); 1564 vmcs_write32(GUEST_TR_AR_BYTES, 0x008b);
1557 1565
1558 flags = vmcs_readl(GUEST_RFLAGS); 1566 flags = vmcs_readl(GUEST_RFLAGS);
1559 vmx->rmode.save_iopl 1567 vmx->rmode.save_rflags = flags;
1560 = (flags & X86_EFLAGS_IOPL) >> IOPL_SHIFT;
1561 1568
1562 flags |= X86_EFLAGS_IOPL | X86_EFLAGS_VM; 1569 flags |= X86_EFLAGS_IOPL | X86_EFLAGS_VM;
1563 1570
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index e46282a56565..3c4ca98ad27f 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -39,6 +39,7 @@
39#include <linux/cpufreq.h> 39#include <linux/cpufreq.h>
40#include <linux/user-return-notifier.h> 40#include <linux/user-return-notifier.h>
41#include <linux/srcu.h> 41#include <linux/srcu.h>
42#include <linux/slab.h>
42#include <trace/events/kvm.h> 43#include <trace/events/kvm.h>
43#undef TRACE_INCLUDE_FILE 44#undef TRACE_INCLUDE_FILE
44#define CREATE_TRACE_POINTS 45#define CREATE_TRACE_POINTS
@@ -432,8 +433,6 @@ void kvm_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0)
432 433
433#ifdef CONFIG_X86_64 434#ifdef CONFIG_X86_64
434 if (cr0 & 0xffffffff00000000UL) { 435 if (cr0 & 0xffffffff00000000UL) {
435 printk(KERN_DEBUG "set_cr0: 0x%lx #GP, reserved bits 0x%lx\n",
436 cr0, kvm_read_cr0(vcpu));
437 kvm_inject_gp(vcpu, 0); 436 kvm_inject_gp(vcpu, 0);
438 return; 437 return;
439 } 438 }
@@ -442,14 +441,11 @@ void kvm_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0)
442 cr0 &= ~CR0_RESERVED_BITS; 441 cr0 &= ~CR0_RESERVED_BITS;
443 442
444 if ((cr0 & X86_CR0_NW) && !(cr0 & X86_CR0_CD)) { 443 if ((cr0 & X86_CR0_NW) && !(cr0 & X86_CR0_CD)) {
445 printk(KERN_DEBUG "set_cr0: #GP, CD == 0 && NW == 1\n");
446 kvm_inject_gp(vcpu, 0); 444 kvm_inject_gp(vcpu, 0);
447 return; 445 return;
448 } 446 }
449 447
450 if ((cr0 & X86_CR0_PG) && !(cr0 & X86_CR0_PE)) { 448 if ((cr0 & X86_CR0_PG) && !(cr0 & X86_CR0_PE)) {
451 printk(KERN_DEBUG "set_cr0: #GP, set PG flag "
452 "and a clear PE flag\n");
453 kvm_inject_gp(vcpu, 0); 449 kvm_inject_gp(vcpu, 0);
454 return; 450 return;
455 } 451 }
@@ -460,15 +456,11 @@ void kvm_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0)
460 int cs_db, cs_l; 456 int cs_db, cs_l;
461 457
462 if (!is_pae(vcpu)) { 458 if (!is_pae(vcpu)) {
463 printk(KERN_DEBUG "set_cr0: #GP, start paging "
464 "in long mode while PAE is disabled\n");
465 kvm_inject_gp(vcpu, 0); 459 kvm_inject_gp(vcpu, 0);
466 return; 460 return;
467 } 461 }
468 kvm_x86_ops->get_cs_db_l_bits(vcpu, &cs_db, &cs_l); 462 kvm_x86_ops->get_cs_db_l_bits(vcpu, &cs_db, &cs_l);
469 if (cs_l) { 463 if (cs_l) {
470 printk(KERN_DEBUG "set_cr0: #GP, start paging "
471 "in long mode while CS.L == 1\n");
472 kvm_inject_gp(vcpu, 0); 464 kvm_inject_gp(vcpu, 0);
473 return; 465 return;
474 466
@@ -476,8 +468,6 @@ void kvm_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0)
476 } else 468 } else
477#endif 469#endif
478 if (is_pae(vcpu) && !load_pdptrs(vcpu, vcpu->arch.cr3)) { 470 if (is_pae(vcpu) && !load_pdptrs(vcpu, vcpu->arch.cr3)) {
479 printk(KERN_DEBUG "set_cr0: #GP, pdptrs "
480 "reserved bits\n");
481 kvm_inject_gp(vcpu, 0); 471 kvm_inject_gp(vcpu, 0);
482 return; 472 return;
483 } 473 }
@@ -504,28 +494,23 @@ void kvm_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4)
504 unsigned long pdptr_bits = X86_CR4_PGE | X86_CR4_PSE | X86_CR4_PAE; 494 unsigned long pdptr_bits = X86_CR4_PGE | X86_CR4_PSE | X86_CR4_PAE;
505 495
506 if (cr4 & CR4_RESERVED_BITS) { 496 if (cr4 & CR4_RESERVED_BITS) {
507 printk(KERN_DEBUG "set_cr4: #GP, reserved bits\n");
508 kvm_inject_gp(vcpu, 0); 497 kvm_inject_gp(vcpu, 0);
509 return; 498 return;
510 } 499 }
511 500
512 if (is_long_mode(vcpu)) { 501 if (is_long_mode(vcpu)) {
513 if (!(cr4 & X86_CR4_PAE)) { 502 if (!(cr4 & X86_CR4_PAE)) {
514 printk(KERN_DEBUG "set_cr4: #GP, clearing PAE while "
515 "in long mode\n");
516 kvm_inject_gp(vcpu, 0); 503 kvm_inject_gp(vcpu, 0);
517 return; 504 return;
518 } 505 }
519 } else if (is_paging(vcpu) && (cr4 & X86_CR4_PAE) 506 } else if (is_paging(vcpu) && (cr4 & X86_CR4_PAE)
520 && ((cr4 ^ old_cr4) & pdptr_bits) 507 && ((cr4 ^ old_cr4) & pdptr_bits)
521 && !load_pdptrs(vcpu, vcpu->arch.cr3)) { 508 && !load_pdptrs(vcpu, vcpu->arch.cr3)) {
522 printk(KERN_DEBUG "set_cr4: #GP, pdptrs reserved bits\n");
523 kvm_inject_gp(vcpu, 0); 509 kvm_inject_gp(vcpu, 0);
524 return; 510 return;
525 } 511 }
526 512
527 if (cr4 & X86_CR4_VMXE) { 513 if (cr4 & X86_CR4_VMXE) {
528 printk(KERN_DEBUG "set_cr4: #GP, setting VMXE\n");
529 kvm_inject_gp(vcpu, 0); 514 kvm_inject_gp(vcpu, 0);
530 return; 515 return;
531 } 516 }
@@ -546,21 +531,16 @@ void kvm_set_cr3(struct kvm_vcpu *vcpu, unsigned long cr3)
546 531
547 if (is_long_mode(vcpu)) { 532 if (is_long_mode(vcpu)) {
548 if (cr3 & CR3_L_MODE_RESERVED_BITS) { 533 if (cr3 & CR3_L_MODE_RESERVED_BITS) {
549 printk(KERN_DEBUG "set_cr3: #GP, reserved bits\n");
550 kvm_inject_gp(vcpu, 0); 534 kvm_inject_gp(vcpu, 0);
551 return; 535 return;
552 } 536 }
553 } else { 537 } else {
554 if (is_pae(vcpu)) { 538 if (is_pae(vcpu)) {
555 if (cr3 & CR3_PAE_RESERVED_BITS) { 539 if (cr3 & CR3_PAE_RESERVED_BITS) {
556 printk(KERN_DEBUG
557 "set_cr3: #GP, reserved bits\n");
558 kvm_inject_gp(vcpu, 0); 540 kvm_inject_gp(vcpu, 0);
559 return; 541 return;
560 } 542 }
561 if (is_paging(vcpu) && !load_pdptrs(vcpu, cr3)) { 543 if (is_paging(vcpu) && !load_pdptrs(vcpu, cr3)) {
562 printk(KERN_DEBUG "set_cr3: #GP, pdptrs "
563 "reserved bits\n");
564 kvm_inject_gp(vcpu, 0); 544 kvm_inject_gp(vcpu, 0);
565 return; 545 return;
566 } 546 }
@@ -592,7 +572,6 @@ EXPORT_SYMBOL_GPL(kvm_set_cr3);
592void kvm_set_cr8(struct kvm_vcpu *vcpu, unsigned long cr8) 572void kvm_set_cr8(struct kvm_vcpu *vcpu, unsigned long cr8)
593{ 573{
594 if (cr8 & CR8_RESERVED_BITS) { 574 if (cr8 & CR8_RESERVED_BITS) {
595 printk(KERN_DEBUG "set_cr8: #GP, reserved bits 0x%lx\n", cr8);
596 kvm_inject_gp(vcpu, 0); 575 kvm_inject_gp(vcpu, 0);
597 return; 576 return;
598 } 577 }
@@ -648,15 +627,12 @@ static u32 emulated_msrs[] = {
648static void set_efer(struct kvm_vcpu *vcpu, u64 efer) 627static void set_efer(struct kvm_vcpu *vcpu, u64 efer)
649{ 628{
650 if (efer & efer_reserved_bits) { 629 if (efer & efer_reserved_bits) {
651 printk(KERN_DEBUG "set_efer: 0x%llx #GP, reserved bits\n",
652 efer);
653 kvm_inject_gp(vcpu, 0); 630 kvm_inject_gp(vcpu, 0);
654 return; 631 return;
655 } 632 }
656 633
657 if (is_paging(vcpu) 634 if (is_paging(vcpu)
658 && (vcpu->arch.efer & EFER_LME) != (efer & EFER_LME)) { 635 && (vcpu->arch.efer & EFER_LME) != (efer & EFER_LME)) {
659 printk(KERN_DEBUG "set_efer: #GP, change LME while paging\n");
660 kvm_inject_gp(vcpu, 0); 636 kvm_inject_gp(vcpu, 0);
661 return; 637 return;
662 } 638 }
@@ -666,7 +642,6 @@ static void set_efer(struct kvm_vcpu *vcpu, u64 efer)
666 642
667 feat = kvm_find_cpuid_entry(vcpu, 0x80000001, 0); 643 feat = kvm_find_cpuid_entry(vcpu, 0x80000001, 0);
668 if (!feat || !(feat->edx & bit(X86_FEATURE_FXSR_OPT))) { 644 if (!feat || !(feat->edx & bit(X86_FEATURE_FXSR_OPT))) {
669 printk(KERN_DEBUG "set_efer: #GP, enable FFXSR w/o CPUID capability\n");
670 kvm_inject_gp(vcpu, 0); 645 kvm_inject_gp(vcpu, 0);
671 return; 646 return;
672 } 647 }
@@ -677,7 +652,6 @@ static void set_efer(struct kvm_vcpu *vcpu, u64 efer)
677 652
678 feat = kvm_find_cpuid_entry(vcpu, 0x80000001, 0); 653 feat = kvm_find_cpuid_entry(vcpu, 0x80000001, 0);
679 if (!feat || !(feat->ecx & bit(X86_FEATURE_SVM))) { 654 if (!feat || !(feat->ecx & bit(X86_FEATURE_SVM))) {
680 printk(KERN_DEBUG "set_efer: #GP, enable SVM w/o SVM\n");
681 kvm_inject_gp(vcpu, 0); 655 kvm_inject_gp(vcpu, 0);
682 return; 656 return;
683 } 657 }
@@ -966,9 +940,13 @@ static int set_msr_mce(struct kvm_vcpu *vcpu, u32 msr, u64 data)
966 if (msr >= MSR_IA32_MC0_CTL && 940 if (msr >= MSR_IA32_MC0_CTL &&
967 msr < MSR_IA32_MC0_CTL + 4 * bank_num) { 941 msr < MSR_IA32_MC0_CTL + 4 * bank_num) {
968 u32 offset = msr - MSR_IA32_MC0_CTL; 942 u32 offset = msr - MSR_IA32_MC0_CTL;
969 /* only 0 or all 1s can be written to IA32_MCi_CTL */ 943 /* only 0 or all 1s can be written to IA32_MCi_CTL
944 * some Linux kernels though clear bit 10 in bank 4 to
945 * workaround a BIOS/GART TBL issue on AMD K8s, ignore
946 * this to avoid an uncatched #GP in the guest
947 */
970 if ((offset & 0x3) == 0 && 948 if ((offset & 0x3) == 0 &&
971 data != 0 && data != ~(u64)0) 949 data != 0 && (data | (1 << 10)) != ~(u64)0)
972 return -1; 950 return -1;
973 vcpu->arch.mce_banks[offset] = data; 951 vcpu->arch.mce_banks[offset] = data;
974 break; 952 break;
@@ -2634,8 +2612,9 @@ static int kvm_vm_ioctl_reinject(struct kvm *kvm,
2634int kvm_vm_ioctl_get_dirty_log(struct kvm *kvm, 2612int kvm_vm_ioctl_get_dirty_log(struct kvm *kvm,
2635 struct kvm_dirty_log *log) 2613 struct kvm_dirty_log *log)
2636{ 2614{
2637 int r, n, i; 2615 int r, i;
2638 struct kvm_memory_slot *memslot; 2616 struct kvm_memory_slot *memslot;
2617 unsigned long n;
2639 unsigned long is_dirty = 0; 2618 unsigned long is_dirty = 0;
2640 unsigned long *dirty_bitmap = NULL; 2619 unsigned long *dirty_bitmap = NULL;
2641 2620
@@ -2650,7 +2629,7 @@ int kvm_vm_ioctl_get_dirty_log(struct kvm *kvm,
2650 if (!memslot->dirty_bitmap) 2629 if (!memslot->dirty_bitmap)
2651 goto out; 2630 goto out;
2652 2631
2653 n = ALIGN(memslot->npages, BITS_PER_LONG) / 8; 2632 n = kvm_dirty_bitmap_bytes(memslot);
2654 2633
2655 r = -ENOMEM; 2634 r = -ENOMEM;
2656 dirty_bitmap = vmalloc(n); 2635 dirty_bitmap = vmalloc(n);
@@ -4482,7 +4461,9 @@ int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
4482 kvm_set_cr8(vcpu, kvm_run->cr8); 4461 kvm_set_cr8(vcpu, kvm_run->cr8);
4483 4462
4484 if (vcpu->arch.pio.cur_count) { 4463 if (vcpu->arch.pio.cur_count) {
4464 vcpu->srcu_idx = srcu_read_lock(&vcpu->kvm->srcu);
4485 r = complete_pio(vcpu); 4465 r = complete_pio(vcpu);
4466 srcu_read_unlock(&vcpu->kvm->srcu, vcpu->srcu_idx);
4486 if (r) 4467 if (r)
4487 goto out; 4468 goto out;
4488 } 4469 }
@@ -5145,6 +5126,7 @@ int kvm_task_switch(struct kvm_vcpu *vcpu, u16 tss_selector, int reason)
5145 int ret = 0; 5126 int ret = 0;
5146 u32 old_tss_base = get_segment_base(vcpu, VCPU_SREG_TR); 5127 u32 old_tss_base = get_segment_base(vcpu, VCPU_SREG_TR);
5147 u16 old_tss_sel = get_segment_selector(vcpu, VCPU_SREG_TR); 5128 u16 old_tss_sel = get_segment_selector(vcpu, VCPU_SREG_TR);
5129 u32 desc_limit;
5148 5130
5149 old_tss_base = kvm_mmu_gva_to_gpa_write(vcpu, old_tss_base, NULL); 5131 old_tss_base = kvm_mmu_gva_to_gpa_write(vcpu, old_tss_base, NULL);
5150 5132
@@ -5167,7 +5149,10 @@ int kvm_task_switch(struct kvm_vcpu *vcpu, u16 tss_selector, int reason)
5167 } 5149 }
5168 } 5150 }
5169 5151
5170 if (!nseg_desc.p || get_desc_limit(&nseg_desc) < 0x67) { 5152 desc_limit = get_desc_limit(&nseg_desc);
5153 if (!nseg_desc.p ||
5154 ((desc_limit < 0x67 && (nseg_desc.type & 8)) ||
5155 desc_limit < 0x2b)) {
5171 kvm_queue_exception_e(vcpu, TS_VECTOR, tss_selector & 0xfffc); 5156 kvm_queue_exception_e(vcpu, TS_VECTOR, tss_selector & 0xfffc);
5172 return 1; 5157 return 1;
5173 } 5158 }
diff --git a/arch/x86/lguest/boot.c b/arch/x86/lguest/boot.c
index 7e59dc1d3fc2..2bdf628066bd 100644
--- a/arch/x86/lguest/boot.c
+++ b/arch/x86/lguest/boot.c
@@ -115,7 +115,7 @@ static void async_hcall(unsigned long call, unsigned long arg1,
115 local_irq_save(flags); 115 local_irq_save(flags);
116 if (lguest_data.hcall_status[next_call] != 0xFF) { 116 if (lguest_data.hcall_status[next_call] != 0xFF) {
117 /* Table full, so do normal hcall which will flush table. */ 117 /* Table full, so do normal hcall which will flush table. */
118 kvm_hypercall4(call, arg1, arg2, arg3, arg4); 118 hcall(call, arg1, arg2, arg3, arg4);
119 } else { 119 } else {
120 lguest_data.hcalls[next_call].arg0 = call; 120 lguest_data.hcalls[next_call].arg0 = call;
121 lguest_data.hcalls[next_call].arg1 = arg1; 121 lguest_data.hcalls[next_call].arg1 = arg1;
@@ -145,46 +145,45 @@ static void async_hcall(unsigned long call, unsigned long arg1,
145 * So, when we're in lazy mode, we call async_hcall() to store the call for 145 * So, when we're in lazy mode, we call async_hcall() to store the call for
146 * future processing: 146 * future processing:
147 */ 147 */
148static void lazy_hcall1(unsigned long call, 148static void lazy_hcall1(unsigned long call, unsigned long arg1)
149 unsigned long arg1)
150{ 149{
151 if (paravirt_get_lazy_mode() == PARAVIRT_LAZY_NONE) 150 if (paravirt_get_lazy_mode() == PARAVIRT_LAZY_NONE)
152 kvm_hypercall1(call, arg1); 151 hcall(call, arg1, 0, 0, 0);
153 else 152 else
154 async_hcall(call, arg1, 0, 0, 0); 153 async_hcall(call, arg1, 0, 0, 0);
155} 154}
156 155
157/* You can imagine what lazy_hcall2, 3 and 4 look like. :*/ 156/* You can imagine what lazy_hcall2, 3 and 4 look like. :*/
158static void lazy_hcall2(unsigned long call, 157static void lazy_hcall2(unsigned long call,
159 unsigned long arg1, 158 unsigned long arg1,
160 unsigned long arg2) 159 unsigned long arg2)
161{ 160{
162 if (paravirt_get_lazy_mode() == PARAVIRT_LAZY_NONE) 161 if (paravirt_get_lazy_mode() == PARAVIRT_LAZY_NONE)
163 kvm_hypercall2(call, arg1, arg2); 162 hcall(call, arg1, arg2, 0, 0);
164 else 163 else
165 async_hcall(call, arg1, arg2, 0, 0); 164 async_hcall(call, arg1, arg2, 0, 0);
166} 165}
167 166
168static void lazy_hcall3(unsigned long call, 167static void lazy_hcall3(unsigned long call,
169 unsigned long arg1, 168 unsigned long arg1,
170 unsigned long arg2, 169 unsigned long arg2,
171 unsigned long arg3) 170 unsigned long arg3)
172{ 171{
173 if (paravirt_get_lazy_mode() == PARAVIRT_LAZY_NONE) 172 if (paravirt_get_lazy_mode() == PARAVIRT_LAZY_NONE)
174 kvm_hypercall3(call, arg1, arg2, arg3); 173 hcall(call, arg1, arg2, arg3, 0);
175 else 174 else
176 async_hcall(call, arg1, arg2, arg3, 0); 175 async_hcall(call, arg1, arg2, arg3, 0);
177} 176}
178 177
179#ifdef CONFIG_X86_PAE 178#ifdef CONFIG_X86_PAE
180static void lazy_hcall4(unsigned long call, 179static void lazy_hcall4(unsigned long call,
181 unsigned long arg1, 180 unsigned long arg1,
182 unsigned long arg2, 181 unsigned long arg2,
183 unsigned long arg3, 182 unsigned long arg3,
184 unsigned long arg4) 183 unsigned long arg4)
185{ 184{
186 if (paravirt_get_lazy_mode() == PARAVIRT_LAZY_NONE) 185 if (paravirt_get_lazy_mode() == PARAVIRT_LAZY_NONE)
187 kvm_hypercall4(call, arg1, arg2, arg3, arg4); 186 hcall(call, arg1, arg2, arg3, arg4);
188 else 187 else
189 async_hcall(call, arg1, arg2, arg3, arg4); 188 async_hcall(call, arg1, arg2, arg3, arg4);
190} 189}
@@ -196,13 +195,13 @@ static void lazy_hcall4(unsigned long call,
196:*/ 195:*/
197static void lguest_leave_lazy_mmu_mode(void) 196static void lguest_leave_lazy_mmu_mode(void)
198{ 197{
199 kvm_hypercall0(LHCALL_FLUSH_ASYNC); 198 hcall(LHCALL_FLUSH_ASYNC, 0, 0, 0, 0);
200 paravirt_leave_lazy_mmu(); 199 paravirt_leave_lazy_mmu();
201} 200}
202 201
203static void lguest_end_context_switch(struct task_struct *next) 202static void lguest_end_context_switch(struct task_struct *next)
204{ 203{
205 kvm_hypercall0(LHCALL_FLUSH_ASYNC); 204 hcall(LHCALL_FLUSH_ASYNC, 0, 0, 0, 0);
206 paravirt_end_context_switch(next); 205 paravirt_end_context_switch(next);
207} 206}
208 207
@@ -286,7 +285,7 @@ static void lguest_write_idt_entry(gate_desc *dt,
286 /* Keep the local copy up to date. */ 285 /* Keep the local copy up to date. */
287 native_write_idt_entry(dt, entrynum, g); 286 native_write_idt_entry(dt, entrynum, g);
288 /* Tell Host about this new entry. */ 287 /* Tell Host about this new entry. */
289 kvm_hypercall3(LHCALL_LOAD_IDT_ENTRY, entrynum, desc[0], desc[1]); 288 hcall(LHCALL_LOAD_IDT_ENTRY, entrynum, desc[0], desc[1], 0);
290} 289}
291 290
292/* 291/*
@@ -300,7 +299,7 @@ static void lguest_load_idt(const struct desc_ptr *desc)
300 struct desc_struct *idt = (void *)desc->address; 299 struct desc_struct *idt = (void *)desc->address;
301 300
302 for (i = 0; i < (desc->size+1)/8; i++) 301 for (i = 0; i < (desc->size+1)/8; i++)
303 kvm_hypercall3(LHCALL_LOAD_IDT_ENTRY, i, idt[i].a, idt[i].b); 302 hcall(LHCALL_LOAD_IDT_ENTRY, i, idt[i].a, idt[i].b, 0);
304} 303}
305 304
306/* 305/*
@@ -321,7 +320,7 @@ static void lguest_load_gdt(const struct desc_ptr *desc)
321 struct desc_struct *gdt = (void *)desc->address; 320 struct desc_struct *gdt = (void *)desc->address;
322 321
323 for (i = 0; i < (desc->size+1)/8; i++) 322 for (i = 0; i < (desc->size+1)/8; i++)
324 kvm_hypercall3(LHCALL_LOAD_GDT_ENTRY, i, gdt[i].a, gdt[i].b); 323 hcall(LHCALL_LOAD_GDT_ENTRY, i, gdt[i].a, gdt[i].b, 0);
325} 324}
326 325
327/* 326/*
@@ -334,8 +333,8 @@ static void lguest_write_gdt_entry(struct desc_struct *dt, int entrynum,
334{ 333{
335 native_write_gdt_entry(dt, entrynum, desc, type); 334 native_write_gdt_entry(dt, entrynum, desc, type);
336 /* Tell Host about this new entry. */ 335 /* Tell Host about this new entry. */
337 kvm_hypercall3(LHCALL_LOAD_GDT_ENTRY, entrynum, 336 hcall(LHCALL_LOAD_GDT_ENTRY, entrynum,
338 dt[entrynum].a, dt[entrynum].b); 337 dt[entrynum].a, dt[entrynum].b, 0);
339} 338}
340 339
341/* 340/*
@@ -931,7 +930,7 @@ static int lguest_clockevent_set_next_event(unsigned long delta,
931 } 930 }
932 931
933 /* Please wake us this far in the future. */ 932 /* Please wake us this far in the future. */
934 kvm_hypercall1(LHCALL_SET_CLOCKEVENT, delta); 933 hcall(LHCALL_SET_CLOCKEVENT, delta, 0, 0, 0);
935 return 0; 934 return 0;
936} 935}
937 936
@@ -942,7 +941,7 @@ static void lguest_clockevent_set_mode(enum clock_event_mode mode,
942 case CLOCK_EVT_MODE_UNUSED: 941 case CLOCK_EVT_MODE_UNUSED:
943 case CLOCK_EVT_MODE_SHUTDOWN: 942 case CLOCK_EVT_MODE_SHUTDOWN:
944 /* A 0 argument shuts the clock down. */ 943 /* A 0 argument shuts the clock down. */
945 kvm_hypercall0(LHCALL_SET_CLOCKEVENT); 944 hcall(LHCALL_SET_CLOCKEVENT, 0, 0, 0, 0);
946 break; 945 break;
947 case CLOCK_EVT_MODE_ONESHOT: 946 case CLOCK_EVT_MODE_ONESHOT:
948 /* This is what we expect. */ 947 /* This is what we expect. */
@@ -1100,7 +1099,7 @@ static void set_lguest_basic_apic_ops(void)
1100/* STOP! Until an interrupt comes in. */ 1099/* STOP! Until an interrupt comes in. */
1101static void lguest_safe_halt(void) 1100static void lguest_safe_halt(void)
1102{ 1101{
1103 kvm_hypercall0(LHCALL_HALT); 1102 hcall(LHCALL_HALT, 0, 0, 0, 0);
1104} 1103}
1105 1104
1106/* 1105/*
@@ -1112,8 +1111,8 @@ static void lguest_safe_halt(void)
1112 */ 1111 */
1113static void lguest_power_off(void) 1112static void lguest_power_off(void)
1114{ 1113{
1115 kvm_hypercall2(LHCALL_SHUTDOWN, __pa("Power down"), 1114 hcall(LHCALL_SHUTDOWN, __pa("Power down"),
1116 LGUEST_SHUTDOWN_POWEROFF); 1115 LGUEST_SHUTDOWN_POWEROFF, 0, 0);
1117} 1116}
1118 1117
1119/* 1118/*
@@ -1123,7 +1122,7 @@ static void lguest_power_off(void)
1123 */ 1122 */
1124static int lguest_panic(struct notifier_block *nb, unsigned long l, void *p) 1123static int lguest_panic(struct notifier_block *nb, unsigned long l, void *p)
1125{ 1124{
1126 kvm_hypercall2(LHCALL_SHUTDOWN, __pa(p), LGUEST_SHUTDOWN_POWEROFF); 1125 hcall(LHCALL_SHUTDOWN, __pa(p), LGUEST_SHUTDOWN_POWEROFF, 0, 0);
1127 /* The hcall won't return, but to keep gcc happy, we're "done". */ 1126 /* The hcall won't return, but to keep gcc happy, we're "done". */
1128 return NOTIFY_DONE; 1127 return NOTIFY_DONE;
1129} 1128}
@@ -1162,7 +1161,7 @@ static __init int early_put_chars(u32 vtermno, const char *buf, int count)
1162 len = sizeof(scratch) - 1; 1161 len = sizeof(scratch) - 1;
1163 scratch[len] = '\0'; 1162 scratch[len] = '\0';
1164 memcpy(scratch, buf, len); 1163 memcpy(scratch, buf, len);
1165 kvm_hypercall1(LHCALL_NOTIFY, __pa(scratch)); 1164 hcall(LHCALL_NOTIFY, __pa(scratch), 0, 0, 0);
1166 1165
1167 /* This routine returns the number of bytes actually written. */ 1166 /* This routine returns the number of bytes actually written. */
1168 return len; 1167 return len;
@@ -1174,7 +1173,7 @@ static __init int early_put_chars(u32 vtermno, const char *buf, int count)
1174 */ 1173 */
1175static void lguest_restart(char *reason) 1174static void lguest_restart(char *reason)
1176{ 1175{
1177 kvm_hypercall2(LHCALL_SHUTDOWN, __pa(reason), LGUEST_SHUTDOWN_RESTART); 1176 hcall(LHCALL_SHUTDOWN, __pa(reason), LGUEST_SHUTDOWN_RESTART, 0, 0);
1178} 1177}
1179 1178
1180/*G:050 1179/*G:050
diff --git a/arch/x86/lguest/i386_head.S b/arch/x86/lguest/i386_head.S
index 27eac0faee48..4f420c2f2d55 100644
--- a/arch/x86/lguest/i386_head.S
+++ b/arch/x86/lguest/i386_head.S
@@ -32,7 +32,7 @@ ENTRY(lguest_entry)
32 */ 32 */
33 movl $LHCALL_LGUEST_INIT, %eax 33 movl $LHCALL_LGUEST_INIT, %eax
34 movl $lguest_data - __PAGE_OFFSET, %ebx 34 movl $lguest_data - __PAGE_OFFSET, %ebx
35 .byte 0x0f,0x01,0xc1 /* KVM_HYPERCALL */ 35 int $LGUEST_TRAP_ENTRY
36 36
37 /* Set up the initial stack so we can run C code. */ 37 /* Set up the initial stack so we can run C code. */
38 movl $(init_thread_union+THREAD_SIZE),%esp 38 movl $(init_thread_union+THREAD_SIZE),%esp
diff --git a/arch/x86/mm/hugetlbpage.c b/arch/x86/mm/hugetlbpage.c
index f46c340727b8..069ce7c37c01 100644
--- a/arch/x86/mm/hugetlbpage.c
+++ b/arch/x86/mm/hugetlbpage.c
@@ -9,7 +9,6 @@
9#include <linux/mm.h> 9#include <linux/mm.h>
10#include <linux/hugetlb.h> 10#include <linux/hugetlb.h>
11#include <linux/pagemap.h> 11#include <linux/pagemap.h>
12#include <linux/slab.h>
13#include <linux/err.h> 12#include <linux/err.h>
14#include <linux/sysctl.h> 13#include <linux/sysctl.h>
15#include <asm/mman.h> 14#include <asm/mman.h>
diff --git a/arch/x86/mm/init.c b/arch/x86/mm/init.c
index e71c5cbc8f35..b278535b14aa 100644
--- a/arch/x86/mm/init.c
+++ b/arch/x86/mm/init.c
@@ -1,3 +1,4 @@
1#include <linux/gfp.h>
1#include <linux/initrd.h> 2#include <linux/initrd.h>
2#include <linux/ioport.h> 3#include <linux/ioport.h>
3#include <linux/swap.h> 4#include <linux/swap.h>
@@ -331,11 +332,23 @@ int devmem_is_allowed(unsigned long pagenr)
331 332
332void free_init_pages(char *what, unsigned long begin, unsigned long end) 333void free_init_pages(char *what, unsigned long begin, unsigned long end)
333{ 334{
334 unsigned long addr = begin; 335 unsigned long addr;
336 unsigned long begin_aligned, end_aligned;
335 337
336 if (addr >= end) 338 /* Make sure boundaries are page aligned */
339 begin_aligned = PAGE_ALIGN(begin);
340 end_aligned = end & PAGE_MASK;
341
342 if (WARN_ON(begin_aligned != begin || end_aligned != end)) {
343 begin = begin_aligned;
344 end = end_aligned;
345 }
346
347 if (begin >= end)
337 return; 348 return;
338 349
350 addr = begin;
351
339 /* 352 /*
340 * If debugging page accesses then do not free this memory but 353 * If debugging page accesses then do not free this memory but
341 * mark them not present - any buggy init-section access will 354 * mark them not present - any buggy init-section access will
@@ -343,7 +356,7 @@ void free_init_pages(char *what, unsigned long begin, unsigned long end)
343 */ 356 */
344#ifdef CONFIG_DEBUG_PAGEALLOC 357#ifdef CONFIG_DEBUG_PAGEALLOC
345 printk(KERN_INFO "debug: unmapping init memory %08lx..%08lx\n", 358 printk(KERN_INFO "debug: unmapping init memory %08lx..%08lx\n",
346 begin, PAGE_ALIGN(end)); 359 begin, end);
347 set_memory_np(begin, (end - begin) >> PAGE_SHIFT); 360 set_memory_np(begin, (end - begin) >> PAGE_SHIFT);
348#else 361#else
349 /* 362 /*
@@ -358,8 +371,7 @@ void free_init_pages(char *what, unsigned long begin, unsigned long end)
358 for (; addr < end; addr += PAGE_SIZE) { 371 for (; addr < end; addr += PAGE_SIZE) {
359 ClearPageReserved(virt_to_page(addr)); 372 ClearPageReserved(virt_to_page(addr));
360 init_page_count(virt_to_page(addr)); 373 init_page_count(virt_to_page(addr));
361 memset((void *)(addr & ~(PAGE_SIZE-1)), 374 memset((void *)addr, POISON_FREE_INITMEM, PAGE_SIZE);
362 POISON_FREE_INITMEM, PAGE_SIZE);
363 free_page(addr); 375 free_page(addr);
364 totalram_pages++; 376 totalram_pages++;
365 } 377 }
@@ -376,6 +388,15 @@ void free_initmem(void)
376#ifdef CONFIG_BLK_DEV_INITRD 388#ifdef CONFIG_BLK_DEV_INITRD
377void free_initrd_mem(unsigned long start, unsigned long end) 389void free_initrd_mem(unsigned long start, unsigned long end)
378{ 390{
379 free_init_pages("initrd memory", start, end); 391 /*
392 * end could be not aligned, and We can not align that,
393 * decompresser could be confused by aligned initrd_end
394 * We already reserve the end partial page before in
395 * - i386_start_kernel()
396 * - x86_64_start_kernel()
397 * - relocate_initrd()
398 * So here We can do PAGE_ALIGN() safely to get partial page to be freed
399 */
400 free_init_pages("initrd memory", start, PAGE_ALIGN(end));
380} 401}
381#endif 402#endif
diff --git a/arch/x86/mm/init_32.c b/arch/x86/mm/init_32.c
index 5cb3f0f54f47..bca79091b9d6 100644
--- a/arch/x86/mm/init_32.c
+++ b/arch/x86/mm/init_32.c
@@ -25,11 +25,11 @@
25#include <linux/pfn.h> 25#include <linux/pfn.h>
26#include <linux/poison.h> 26#include <linux/poison.h>
27#include <linux/bootmem.h> 27#include <linux/bootmem.h>
28#include <linux/slab.h>
29#include <linux/proc_fs.h> 28#include <linux/proc_fs.h>
30#include <linux/memory_hotplug.h> 29#include <linux/memory_hotplug.h>
31#include <linux/initrd.h> 30#include <linux/initrd.h>
32#include <linux/cpumask.h> 31#include <linux/cpumask.h>
32#include <linux/gfp.h>
33 33
34#include <asm/asm.h> 34#include <asm/asm.h>
35#include <asm/bios_ebda.h> 35#include <asm/bios_ebda.h>
diff --git a/arch/x86/mm/init_64.c b/arch/x86/mm/init_64.c
index e9b040e1cde5..ee41bba315d1 100644
--- a/arch/x86/mm/init_64.c
+++ b/arch/x86/mm/init_64.c
@@ -29,6 +29,7 @@
29#include <linux/module.h> 29#include <linux/module.h>
30#include <linux/memory_hotplug.h> 30#include <linux/memory_hotplug.h>
31#include <linux/nmi.h> 31#include <linux/nmi.h>
32#include <linux/gfp.h>
32 33
33#include <asm/processor.h> 34#include <asm/processor.h>
34#include <asm/bios_ebda.h> 35#include <asm/bios_ebda.h>
diff --git a/arch/x86/mm/kmmio.c b/arch/x86/mm/kmmio.c
index 536fb6823366..5d0e67fff1a6 100644
--- a/arch/x86/mm/kmmio.c
+++ b/arch/x86/mm/kmmio.c
@@ -21,6 +21,7 @@
21#include <linux/kdebug.h> 21#include <linux/kdebug.h>
22#include <linux/mutex.h> 22#include <linux/mutex.h>
23#include <linux/io.h> 23#include <linux/io.h>
24#include <linux/slab.h>
24#include <asm/cacheflush.h> 25#include <asm/cacheflush.h>
25#include <asm/tlbflush.h> 26#include <asm/tlbflush.h>
26#include <linux/errno.h> 27#include <linux/errno.h>
diff --git a/arch/x86/mm/mmio-mod.c b/arch/x86/mm/mmio-mod.c
index 34a3291ca103..3adff7dcc148 100644
--- a/arch/x86/mm/mmio-mod.c
+++ b/arch/x86/mm/mmio-mod.c
@@ -26,6 +26,7 @@
26 26
27#include <linux/module.h> 27#include <linux/module.h>
28#include <linux/debugfs.h> 28#include <linux/debugfs.h>
29#include <linux/slab.h>
29#include <linux/uaccess.h> 30#include <linux/uaccess.h>
30#include <linux/io.h> 31#include <linux/io.h>
31#include <linux/version.h> 32#include <linux/version.h>
diff --git a/arch/x86/mm/pageattr.c b/arch/x86/mm/pageattr.c
index 1d4eb93d333c..28195c350b97 100644
--- a/arch/x86/mm/pageattr.c
+++ b/arch/x86/mm/pageattr.c
@@ -6,13 +6,13 @@
6#include <linux/bootmem.h> 6#include <linux/bootmem.h>
7#include <linux/module.h> 7#include <linux/module.h>
8#include <linux/sched.h> 8#include <linux/sched.h>
9#include <linux/slab.h>
10#include <linux/mm.h> 9#include <linux/mm.h>
11#include <linux/interrupt.h> 10#include <linux/interrupt.h>
12#include <linux/seq_file.h> 11#include <linux/seq_file.h>
13#include <linux/debugfs.h> 12#include <linux/debugfs.h>
14#include <linux/pfn.h> 13#include <linux/pfn.h>
15#include <linux/percpu.h> 14#include <linux/percpu.h>
15#include <linux/gfp.h>
16 16
17#include <asm/e820.h> 17#include <asm/e820.h>
18#include <asm/processor.h> 18#include <asm/processor.h>
@@ -291,8 +291,29 @@ static inline pgprot_t static_protections(pgprot_t prot, unsigned long address,
291 */ 291 */
292 if (kernel_set_to_readonly && 292 if (kernel_set_to_readonly &&
293 within(address, (unsigned long)_text, 293 within(address, (unsigned long)_text,
294 (unsigned long)__end_rodata_hpage_align)) 294 (unsigned long)__end_rodata_hpage_align)) {
295 pgprot_val(forbidden) |= _PAGE_RW; 295 unsigned int level;
296
297 /*
298 * Don't enforce the !RW mapping for the kernel text mapping,
299 * if the current mapping is already using small page mapping.
300 * No need to work hard to preserve large page mappings in this
301 * case.
302 *
303 * This also fixes the Linux Xen paravirt guest boot failure
304 * (because of unexpected read-only mappings for kernel identity
305 * mappings). In this paravirt guest case, the kernel text
306 * mapping and the kernel identity mapping share the same
307 * page-table pages. Thus we can't really use different
308 * protections for the kernel text and identity mappings. Also,
309 * these shared mappings are made of small page mappings.
310 * Thus this don't enforce !RW mapping for small page kernel
311 * text mapping logic will help Linux Xen parvirt guest boot
312 * aswell.
313 */
314 if (lookup_address(address, &level) && (level != PG_LEVEL_4K))
315 pgprot_val(forbidden) |= _PAGE_RW;
316 }
296#endif 317#endif
297 318
298 prot = __pgprot(pgprot_val(prot) & ~pgprot_val(forbidden)); 319 prot = __pgprot(pgprot_val(prot) & ~pgprot_val(forbidden));
diff --git a/arch/x86/mm/pat.c b/arch/x86/mm/pat.c
index ae9648eb1c7f..edc8b95afc1a 100644
--- a/arch/x86/mm/pat.c
+++ b/arch/x86/mm/pat.c
@@ -12,7 +12,7 @@
12#include <linux/debugfs.h> 12#include <linux/debugfs.h>
13#include <linux/kernel.h> 13#include <linux/kernel.h>
14#include <linux/module.h> 14#include <linux/module.h>
15#include <linux/gfp.h> 15#include <linux/slab.h>
16#include <linux/mm.h> 16#include <linux/mm.h>
17#include <linux/fs.h> 17#include <linux/fs.h>
18#include <linux/rbtree.h> 18#include <linux/rbtree.h>
diff --git a/arch/x86/mm/pgtable.c b/arch/x86/mm/pgtable.c
index c9ba9deafe83..5c4ee422590e 100644
--- a/arch/x86/mm/pgtable.c
+++ b/arch/x86/mm/pgtable.c
@@ -1,4 +1,5 @@
1#include <linux/mm.h> 1#include <linux/mm.h>
2#include <linux/gfp.h>
2#include <asm/pgalloc.h> 3#include <asm/pgalloc.h>
3#include <asm/pgtable.h> 4#include <asm/pgtable.h>
4#include <asm/tlb.h> 5#include <asm/tlb.h>
diff --git a/arch/x86/mm/pgtable_32.c b/arch/x86/mm/pgtable_32.c
index 46c8834aedc0..1a8faf09afed 100644
--- a/arch/x86/mm/pgtable_32.c
+++ b/arch/x86/mm/pgtable_32.c
@@ -6,7 +6,6 @@
6#include <linux/swap.h> 6#include <linux/swap.h>
7#include <linux/smp.h> 7#include <linux/smp.h>
8#include <linux/highmem.h> 8#include <linux/highmem.h>
9#include <linux/slab.h>
10#include <linux/pagemap.h> 9#include <linux/pagemap.h>
11#include <linux/spinlock.h> 10#include <linux/spinlock.h>
12#include <linux/module.h> 11#include <linux/module.h>
diff --git a/arch/x86/oprofile/op_model_amd.c b/arch/x86/oprofile/op_model_amd.c
index 6a58256dce9f..090cbbec7dbd 100644
--- a/arch/x86/oprofile/op_model_amd.c
+++ b/arch/x86/oprofile/op_model_amd.c
@@ -46,17 +46,6 @@
46 46
47static unsigned long reset_value[NUM_VIRT_COUNTERS]; 47static unsigned long reset_value[NUM_VIRT_COUNTERS];
48 48
49/* IbsFetchCtl bits/masks */
50#define IBS_FETCH_RAND_EN (1ULL<<57)
51#define IBS_FETCH_VAL (1ULL<<49)
52#define IBS_FETCH_ENABLE (1ULL<<48)
53#define IBS_FETCH_CNT_MASK 0xFFFF0000ULL
54
55/* IbsOpCtl bits */
56#define IBS_OP_CNT_CTL (1ULL<<19)
57#define IBS_OP_VAL (1ULL<<18)
58#define IBS_OP_ENABLE (1ULL<<17)
59
60#define IBS_FETCH_SIZE 6 49#define IBS_FETCH_SIZE 6
61#define IBS_OP_SIZE 12 50#define IBS_OP_SIZE 12
62 51
@@ -182,7 +171,7 @@ static void op_amd_setup_ctrs(struct op_x86_model_spec const *model,
182 continue; 171 continue;
183 } 172 }
184 rdmsrl(msrs->controls[i].addr, val); 173 rdmsrl(msrs->controls[i].addr, val);
185 if (val & ARCH_PERFMON_EVENTSEL0_ENABLE) 174 if (val & ARCH_PERFMON_EVENTSEL_ENABLE)
186 op_x86_warn_in_use(i); 175 op_x86_warn_in_use(i);
187 val &= model->reserved; 176 val &= model->reserved;
188 wrmsrl(msrs->controls[i].addr, val); 177 wrmsrl(msrs->controls[i].addr, val);
@@ -290,7 +279,7 @@ op_amd_handle_ibs(struct pt_regs * const regs,
290 oprofile_write_commit(&entry); 279 oprofile_write_commit(&entry);
291 280
292 /* reenable the IRQ */ 281 /* reenable the IRQ */
293 ctl &= ~(IBS_FETCH_VAL | IBS_FETCH_CNT_MASK); 282 ctl &= ~(IBS_FETCH_VAL | IBS_FETCH_CNT);
294 ctl |= IBS_FETCH_ENABLE; 283 ctl |= IBS_FETCH_ENABLE;
295 wrmsrl(MSR_AMD64_IBSFETCHCTL, ctl); 284 wrmsrl(MSR_AMD64_IBSFETCHCTL, ctl);
296 } 285 }
@@ -330,7 +319,7 @@ static inline void op_amd_start_ibs(void)
330 return; 319 return;
331 320
332 if (ibs_config.fetch_enabled) { 321 if (ibs_config.fetch_enabled) {
333 val = (ibs_config.max_cnt_fetch >> 4) & 0xFFFF; 322 val = (ibs_config.max_cnt_fetch >> 4) & IBS_FETCH_MAX_CNT;
334 val |= ibs_config.rand_en ? IBS_FETCH_RAND_EN : 0; 323 val |= ibs_config.rand_en ? IBS_FETCH_RAND_EN : 0;
335 val |= IBS_FETCH_ENABLE; 324 val |= IBS_FETCH_ENABLE;
336 wrmsrl(MSR_AMD64_IBSFETCHCTL, val); 325 wrmsrl(MSR_AMD64_IBSFETCHCTL, val);
@@ -352,7 +341,7 @@ static inline void op_amd_start_ibs(void)
352 * avoid underflows. 341 * avoid underflows.
353 */ 342 */
354 ibs_op_ctl = min(ibs_op_ctl + IBS_RANDOM_MAXCNT_OFFSET, 343 ibs_op_ctl = min(ibs_op_ctl + IBS_RANDOM_MAXCNT_OFFSET,
355 0xFFFFULL); 344 IBS_OP_MAX_CNT);
356 } 345 }
357 if (ibs_caps & IBS_CAPS_OPCNT && ibs_config.dispatched_ops) 346 if (ibs_caps & IBS_CAPS_OPCNT && ibs_config.dispatched_ops)
358 ibs_op_ctl |= IBS_OP_CNT_CTL; 347 ibs_op_ctl |= IBS_OP_CNT_CTL;
@@ -409,7 +398,7 @@ static void op_amd_start(struct op_msrs const * const msrs)
409 if (!reset_value[op_x86_phys_to_virt(i)]) 398 if (!reset_value[op_x86_phys_to_virt(i)])
410 continue; 399 continue;
411 rdmsrl(msrs->controls[i].addr, val); 400 rdmsrl(msrs->controls[i].addr, val);
412 val |= ARCH_PERFMON_EVENTSEL0_ENABLE; 401 val |= ARCH_PERFMON_EVENTSEL_ENABLE;
413 wrmsrl(msrs->controls[i].addr, val); 402 wrmsrl(msrs->controls[i].addr, val);
414 } 403 }
415 404
@@ -429,7 +418,7 @@ static void op_amd_stop(struct op_msrs const * const msrs)
429 if (!reset_value[op_x86_phys_to_virt(i)]) 418 if (!reset_value[op_x86_phys_to_virt(i)])
430 continue; 419 continue;
431 rdmsrl(msrs->controls[i].addr, val); 420 rdmsrl(msrs->controls[i].addr, val);
432 val &= ~ARCH_PERFMON_EVENTSEL0_ENABLE; 421 val &= ~ARCH_PERFMON_EVENTSEL_ENABLE;
433 wrmsrl(msrs->controls[i].addr, val); 422 wrmsrl(msrs->controls[i].addr, val);
434 } 423 }
435 424
diff --git a/arch/x86/oprofile/op_model_ppro.c b/arch/x86/oprofile/op_model_ppro.c
index 5d1727ba409e..2bf90fafa7b5 100644
--- a/arch/x86/oprofile/op_model_ppro.c
+++ b/arch/x86/oprofile/op_model_ppro.c
@@ -88,7 +88,7 @@ static void ppro_setup_ctrs(struct op_x86_model_spec const *model,
88 continue; 88 continue;
89 } 89 }
90 rdmsrl(msrs->controls[i].addr, val); 90 rdmsrl(msrs->controls[i].addr, val);
91 if (val & ARCH_PERFMON_EVENTSEL0_ENABLE) 91 if (val & ARCH_PERFMON_EVENTSEL_ENABLE)
92 op_x86_warn_in_use(i); 92 op_x86_warn_in_use(i);
93 val &= model->reserved; 93 val &= model->reserved;
94 wrmsrl(msrs->controls[i].addr, val); 94 wrmsrl(msrs->controls[i].addr, val);
@@ -166,7 +166,7 @@ static void ppro_start(struct op_msrs const * const msrs)
166 for (i = 0; i < num_counters; ++i) { 166 for (i = 0; i < num_counters; ++i) {
167 if (reset_value[i]) { 167 if (reset_value[i]) {
168 rdmsrl(msrs->controls[i].addr, val); 168 rdmsrl(msrs->controls[i].addr, val);
169 val |= ARCH_PERFMON_EVENTSEL0_ENABLE; 169 val |= ARCH_PERFMON_EVENTSEL_ENABLE;
170 wrmsrl(msrs->controls[i].addr, val); 170 wrmsrl(msrs->controls[i].addr, val);
171 } 171 }
172 } 172 }
@@ -184,7 +184,7 @@ static void ppro_stop(struct op_msrs const * const msrs)
184 if (!reset_value[i]) 184 if (!reset_value[i])
185 continue; 185 continue;
186 rdmsrl(msrs->controls[i].addr, val); 186 rdmsrl(msrs->controls[i].addr, val);
187 val &= ~ARCH_PERFMON_EVENTSEL0_ENABLE; 187 val &= ~ARCH_PERFMON_EVENTSEL_ENABLE;
188 wrmsrl(msrs->controls[i].addr, val); 188 wrmsrl(msrs->controls[i].addr, val);
189 } 189 }
190} 190}
diff --git a/arch/x86/pci/acpi.c b/arch/x86/pci/acpi.c
index 6e22454bfaa6..c7b1ebfb7da7 100644
--- a/arch/x86/pci/acpi.c
+++ b/arch/x86/pci/acpi.c
@@ -3,6 +3,7 @@
3#include <linux/init.h> 3#include <linux/init.h>
4#include <linux/irq.h> 4#include <linux/irq.h>
5#include <linux/dmi.h> 5#include <linux/dmi.h>
6#include <linux/slab.h>
6#include <asm/numa.h> 7#include <asm/numa.h>
7#include <asm/pci_x86.h> 8#include <asm/pci_x86.h>
8 9
@@ -122,8 +123,8 @@ setup_resource(struct acpi_resource *acpi_res, void *data)
122 struct acpi_resource_address64 addr; 123 struct acpi_resource_address64 addr;
123 acpi_status status; 124 acpi_status status;
124 unsigned long flags; 125 unsigned long flags;
125 struct resource *root; 126 struct resource *root, *conflict;
126 u64 start, end; 127 u64 start, end, max_len;
127 128
128 status = resource_to_addr(acpi_res, &addr); 129 status = resource_to_addr(acpi_res, &addr);
129 if (!ACPI_SUCCESS(status)) 130 if (!ACPI_SUCCESS(status))
@@ -140,6 +141,17 @@ setup_resource(struct acpi_resource *acpi_res, void *data)
140 } else 141 } else
141 return AE_OK; 142 return AE_OK;
142 143
144 max_len = addr.maximum - addr.minimum + 1;
145 if (addr.address_length > max_len) {
146 dev_printk(KERN_DEBUG, &info->bridge->dev,
147 "host bridge window length %#llx doesn't fit in "
148 "%#llx-%#llx, trimming\n",
149 (unsigned long long) addr.address_length,
150 (unsigned long long) addr.minimum,
151 (unsigned long long) addr.maximum);
152 addr.address_length = max_len;
153 }
154
143 start = addr.minimum + addr.translation_offset; 155 start = addr.minimum + addr.translation_offset;
144 end = start + addr.address_length - 1; 156 end = start + addr.address_length - 1;
145 157
@@ -157,9 +169,12 @@ setup_resource(struct acpi_resource *acpi_res, void *data)
157 return AE_OK; 169 return AE_OK;
158 } 170 }
159 171
160 if (insert_resource(root, res)) { 172 conflict = insert_resource_conflict(root, res);
173 if (conflict) {
161 dev_err(&info->bridge->dev, 174 dev_err(&info->bridge->dev,
162 "can't allocate host bridge window %pR\n", res); 175 "address space collision: host bridge window %pR "
176 "conflicts with %s %pR\n",
177 res, conflict->name, conflict);
163 } else { 178 } else {
164 pci_bus_add_resource(info->bus, res, 0); 179 pci_bus_add_resource(info->bus, res, 0);
165 info->res_num++; 180 info->res_num++;
diff --git a/arch/x86/pci/common.c b/arch/x86/pci/common.c
index 294e10cb11e1..cf2e93869c48 100644
--- a/arch/x86/pci/common.c
+++ b/arch/x86/pci/common.c
@@ -9,6 +9,7 @@
9#include <linux/ioport.h> 9#include <linux/ioport.h>
10#include <linux/init.h> 10#include <linux/init.h>
11#include <linux/dmi.h> 11#include <linux/dmi.h>
12#include <linux/slab.h>
12 13
13#include <asm/acpi.h> 14#include <asm/acpi.h>
14#include <asm/segment.h> 15#include <asm/segment.h>
diff --git a/arch/x86/pci/i386.c b/arch/x86/pci/i386.c
index dece3eb9c906..46fd43f79103 100644
--- a/arch/x86/pci/i386.c
+++ b/arch/x86/pci/i386.c
@@ -127,9 +127,6 @@ static void __init pcibios_allocate_bus_resources(struct list_head *bus_list)
127 continue; 127 continue;
128 if (!r->start || 128 if (!r->start ||
129 pci_claim_resource(dev, idx) < 0) { 129 pci_claim_resource(dev, idx) < 0) {
130 dev_info(&dev->dev,
131 "can't reserve window %pR\n",
132 r);
133 /* 130 /*
134 * Something is wrong with the region. 131 * Something is wrong with the region.
135 * Invalidate the resource to prevent 132 * Invalidate the resource to prevent
@@ -181,8 +178,6 @@ static void __init pcibios_allocate_resources(int pass)
181 "BAR %d: reserving %pr (d=%d, p=%d)\n", 178 "BAR %d: reserving %pr (d=%d, p=%d)\n",
182 idx, r, disabled, pass); 179 idx, r, disabled, pass);
183 if (pci_claim_resource(dev, idx) < 0) { 180 if (pci_claim_resource(dev, idx) < 0) {
184 dev_info(&dev->dev,
185 "can't reserve %pR\n", r);
186 /* We'll assign a new address later */ 181 /* We'll assign a new address later */
187 r->end -= r->start; 182 r->end -= r->start;
188 r->start = 0; 183 r->start = 0;
diff --git a/arch/x86/pci/irq.c b/arch/x86/pci/irq.c
index 8b107521d24e..5d362b5ba06f 100644
--- a/arch/x86/pci/irq.c
+++ b/arch/x86/pci/irq.c
@@ -8,7 +8,6 @@
8#include <linux/kernel.h> 8#include <linux/kernel.h>
9#include <linux/pci.h> 9#include <linux/pci.h>
10#include <linux/init.h> 10#include <linux/init.h>
11#include <linux/slab.h>
12#include <linux/interrupt.h> 11#include <linux/interrupt.h>
13#include <linux/dmi.h> 12#include <linux/dmi.h>
14#include <linux/io.h> 13#include <linux/io.h>
diff --git a/arch/x86/pci/mmconfig-shared.c b/arch/x86/pci/mmconfig-shared.c
index 8f3f9a50b1e0..39b9ebe8f886 100644
--- a/arch/x86/pci/mmconfig-shared.c
+++ b/arch/x86/pci/mmconfig-shared.c
@@ -16,6 +16,7 @@
16#include <linux/sfi_acpi.h> 16#include <linux/sfi_acpi.h>
17#include <linux/bitmap.h> 17#include <linux/bitmap.h>
18#include <linux/dmi.h> 18#include <linux/dmi.h>
19#include <linux/slab.h>
19#include <asm/e820.h> 20#include <asm/e820.h>
20#include <asm/pci_x86.h> 21#include <asm/pci_x86.h>
21#include <asm/acpi.h> 22#include <asm/acpi.h>
diff --git a/arch/x86/pci/pcbios.c b/arch/x86/pci/pcbios.c
index 1c975cc9839e..59a225c17b84 100644
--- a/arch/x86/pci/pcbios.c
+++ b/arch/x86/pci/pcbios.c
@@ -4,6 +4,7 @@
4 4
5#include <linux/pci.h> 5#include <linux/pci.h>
6#include <linux/init.h> 6#include <linux/init.h>
7#include <linux/slab.h>
7#include <linux/module.h> 8#include <linux/module.h>
8#include <linux/uaccess.h> 9#include <linux/uaccess.h>
9#include <asm/pci_x86.h> 10#include <asm/pci_x86.h>
diff --git a/arch/x86/power/hibernate_32.c b/arch/x86/power/hibernate_32.c
index 81197c62d5b3..3769079874d8 100644
--- a/arch/x86/power/hibernate_32.c
+++ b/arch/x86/power/hibernate_32.c
@@ -6,6 +6,7 @@
6 * Copyright (c) 2006 Rafael J. Wysocki <rjw@sisk.pl> 6 * Copyright (c) 2006 Rafael J. Wysocki <rjw@sisk.pl>
7 */ 7 */
8 8
9#include <linux/gfp.h>
9#include <linux/suspend.h> 10#include <linux/suspend.h>
10#include <linux/bootmem.h> 11#include <linux/bootmem.h>
11 12
diff --git a/arch/x86/power/hibernate_64.c b/arch/x86/power/hibernate_64.c
index 65fdc86e923f..d24f983ba1e5 100644
--- a/arch/x86/power/hibernate_64.c
+++ b/arch/x86/power/hibernate_64.c
@@ -8,6 +8,7 @@
8 * Copyright (c) 2001 Patrick Mochel <mochel@osdl.org> 8 * Copyright (c) 2001 Patrick Mochel <mochel@osdl.org>
9 */ 9 */
10 10
11#include <linux/gfp.h>
11#include <linux/smp.h> 12#include <linux/smp.h>
12#include <linux/suspend.h> 13#include <linux/suspend.h>
13#include <asm/proto.h> 14#include <asm/proto.h>
diff --git a/arch/x86/power/hibernate_asm_32.S b/arch/x86/power/hibernate_asm_32.S
index b641388d8286..ad47daeafa4e 100644
--- a/arch/x86/power/hibernate_asm_32.S
+++ b/arch/x86/power/hibernate_asm_32.S
@@ -27,10 +27,17 @@ ENTRY(swsusp_arch_suspend)
27 ret 27 ret
28 28
29ENTRY(restore_image) 29ENTRY(restore_image)
30 movl mmu_cr4_features, %ecx
30 movl resume_pg_dir, %eax 31 movl resume_pg_dir, %eax
31 subl $__PAGE_OFFSET, %eax 32 subl $__PAGE_OFFSET, %eax
32 movl %eax, %cr3 33 movl %eax, %cr3
33 34
35 jecxz 1f # cr4 Pentium and higher, skip if zero
36 andl $~(X86_CR4_PGE), %ecx
37 movl %ecx, %cr4; # turn off PGE
38 movl %cr3, %eax; # flush TLB
39 movl %eax, %cr3
401:
34 movl restore_pblist, %edx 41 movl restore_pblist, %edx
35 .p2align 4,,7 42 .p2align 4,,7
36 43
@@ -54,16 +61,8 @@ done:
54 movl $swapper_pg_dir, %eax 61 movl $swapper_pg_dir, %eax
55 subl $__PAGE_OFFSET, %eax 62 subl $__PAGE_OFFSET, %eax
56 movl %eax, %cr3 63 movl %eax, %cr3
57 /* Flush TLB, including "global" things (vmalloc) */
58 movl mmu_cr4_features, %ecx 64 movl mmu_cr4_features, %ecx
59 jecxz 1f # cr4 Pentium and higher, skip if zero 65 jecxz 1f # cr4 Pentium and higher, skip if zero
60 movl %ecx, %edx
61 andl $~(X86_CR4_PGE), %edx
62 movl %edx, %cr4; # turn off PGE
631:
64 movl %cr3, %eax; # flush TLB
65 movl %eax, %cr3
66 jecxz 1f # cr4 Pentium and higher, skip if zero
67 movl %ecx, %cr4; # turn PGE back on 66 movl %ecx, %cr4; # turn PGE back on
681: 671:
69 68
diff --git a/arch/x86/vdso/vma.c b/arch/x86/vdso/vma.c
index 21e1aeb9f3ea..ac74869b8140 100644
--- a/arch/x86/vdso/vma.c
+++ b/arch/x86/vdso/vma.c
@@ -6,6 +6,7 @@
6#include <linux/mm.h> 6#include <linux/mm.h>
7#include <linux/err.h> 7#include <linux/err.h>
8#include <linux/sched.h> 8#include <linux/sched.h>
9#include <linux/slab.h>
9#include <linux/init.h> 10#include <linux/init.h>
10#include <linux/random.h> 11#include <linux/random.h>
11#include <linux/elf.h> 12#include <linux/elf.h>
diff --git a/arch/x86/xen/debugfs.c b/arch/x86/xen/debugfs.c
index e133ce25e290..1304bcec8ee5 100644
--- a/arch/x86/xen/debugfs.c
+++ b/arch/x86/xen/debugfs.c
@@ -1,5 +1,6 @@
1#include <linux/init.h> 1#include <linux/init.h>
2#include <linux/debugfs.h> 2#include <linux/debugfs.h>
3#include <linux/slab.h>
3#include <linux/module.h> 4#include <linux/module.h>
4 5
5#include "debugfs.h" 6#include "debugfs.h"
diff --git a/arch/x86/xen/enlighten.c b/arch/x86/xen/enlighten.c
index b607239c1ba8..65d8d79b46a8 100644
--- a/arch/x86/xen/enlighten.c
+++ b/arch/x86/xen/enlighten.c
@@ -28,6 +28,7 @@
28#include <linux/highmem.h> 28#include <linux/highmem.h>
29#include <linux/console.h> 29#include <linux/console.h>
30#include <linux/pci.h> 30#include <linux/pci.h>
31#include <linux/gfp.h>
31 32
32#include <xen/xen.h> 33#include <xen/xen.h>
33#include <xen/interface/xen.h> 34#include <xen/interface/xen.h>
diff --git a/arch/x86/xen/mmu.c b/arch/x86/xen/mmu.c
index f9eb7de74f42..914f04695ce5 100644
--- a/arch/x86/xen/mmu.c
+++ b/arch/x86/xen/mmu.c
@@ -43,6 +43,7 @@
43#include <linux/debugfs.h> 43#include <linux/debugfs.h>
44#include <linux/bug.h> 44#include <linux/bug.h>
45#include <linux/module.h> 45#include <linux/module.h>
46#include <linux/gfp.h>
46 47
47#include <asm/pgtable.h> 48#include <asm/pgtable.h>
48#include <asm/tlbflush.h> 49#include <asm/tlbflush.h>
diff --git a/arch/x86/xen/smp.c b/arch/x86/xen/smp.c
index deafb65ef44e..a29693fd3138 100644
--- a/arch/x86/xen/smp.c
+++ b/arch/x86/xen/smp.c
@@ -14,6 +14,7 @@
14 */ 14 */
15#include <linux/sched.h> 15#include <linux/sched.h>
16#include <linux/err.h> 16#include <linux/err.h>
17#include <linux/slab.h>
17#include <linux/smp.h> 18#include <linux/smp.h>
18 19
19#include <asm/paravirt.h> 20#include <asm/paravirt.h>
diff --git a/arch/x86/xen/spinlock.c b/arch/x86/xen/spinlock.c
index 24ded31b5aec..e0500646585d 100644
--- a/arch/x86/xen/spinlock.c
+++ b/arch/x86/xen/spinlock.c
@@ -6,6 +6,7 @@
6#include <linux/spinlock.h> 6#include <linux/spinlock.h>
7#include <linux/debugfs.h> 7#include <linux/debugfs.h>
8#include <linux/log2.h> 8#include <linux/log2.h>
9#include <linux/gfp.h>
9 10
10#include <asm/paravirt.h> 11#include <asm/paravirt.h>
11 12
diff --git a/arch/x86/xen/time.c b/arch/x86/xen/time.c
index 0d3f07cd1b5f..32764b8880b5 100644
--- a/arch/x86/xen/time.c
+++ b/arch/x86/xen/time.c
@@ -13,6 +13,7 @@
13#include <linux/clockchips.h> 13#include <linux/clockchips.h>
14#include <linux/kernel_stat.h> 14#include <linux/kernel_stat.h>
15#include <linux/math64.h> 15#include <linux/math64.h>
16#include <linux/gfp.h>
16 17
17#include <asm/pvclock.h> 18#include <asm/pvclock.h>
18#include <asm/xen/hypervisor.h> 19#include <asm/xen/hypervisor.h>
generated by cgit v1.2.2 (git 2.25.0) at 2025-03-24 10:13:43 -0400