aboutsummaryrefslogtreecommitdiffstats
path: root/arch/x86
diff options
context:
space:
mode:
Diffstat (limited to 'arch/x86')
-rw-r--r--arch/x86/Kconfig16
-rw-r--r--arch/x86/boot/video-vga.c33
-rw-r--r--arch/x86/include/asm/apic.h3
-rw-r--r--arch/x86/include/asm/cacheflush.h5
-rw-r--r--[-rwxr-xr-x]arch/x86/include/asm/cpu_debug.h0
-rw-r--r--arch/x86/include/asm/cpufeature.h1
-rw-r--r--arch/x86/include/asm/cpumask.h18
-rw-r--r--arch/x86/include/asm/dma-mapping.h6
-rw-r--r--arch/x86/include/asm/fixmap.h6
-rw-r--r--arch/x86/include/asm/ftrace.h7
-rw-r--r--arch/x86/include/asm/io.h6
-rw-r--r--arch/x86/include/asm/io_apic.h11
-rw-r--r--arch/x86/include/asm/paravirt.h4
-rw-r--r--arch/x86/include/asm/pci.h5
-rw-r--r--arch/x86/include/asm/processor.h9
-rw-r--r--arch/x86/include/asm/ptrace-abi.h3
-rw-r--r--arch/x86/include/asm/required-features.h2
-rw-r--r--arch/x86/include/asm/sigcontext.h6
-rw-r--r--arch/x86/include/asm/smp.h13
-rw-r--r--arch/x86/include/asm/thread_info.h9
-rw-r--r--arch/x86/include/asm/topology.h87
-rw-r--r--arch/x86/include/asm/xen/page.h3
-rw-r--r--arch/x86/include/asm/xsave.h3
-rw-r--r--arch/x86/kernel/Makefile3
-rw-r--r--arch/x86/kernel/acpi/boot.c65
-rw-r--r--arch/x86/kernel/alternative.c29
-rw-r--r--arch/x86/kernel/amd_iommu.c7
-rw-r--r--arch/x86/kernel/apic/apic.c76
-rw-r--r--arch/x86/kernel/apic/apic_flat_64.c4
-rw-r--r--arch/x86/kernel/apic/bigsmp_32.c16
-rw-r--r--arch/x86/kernel/apic/es7000_32.c17
-rw-r--r--arch/x86/kernel/apic/io_apic.c143
-rw-r--r--arch/x86/kernel/apic/nmi.c11
-rw-r--r--arch/x86/kernel/apic/numaq_32.c11
-rw-r--r--arch/x86/kernel/apic/probe_32.c3
-rw-r--r--arch/x86/kernel/apic/summit_32.c21
-rw-r--r--arch/x86/kernel/apm_32.c248
-rw-r--r--arch/x86/kernel/cpu/addon_cpuid_features.c1
-rw-r--r--arch/x86/kernel/cpu/common.c13
-rw-r--r--[-rwxr-xr-x]arch/x86/kernel/cpu/cpu_debug.c0
-rw-r--r--arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c97
-rw-r--r--arch/x86/kernel/cpu/cpufreq/longhaul.c7
-rw-r--r--arch/x86/kernel/cpu/cpufreq/p4-clockmod.c2
-rw-r--r--arch/x86/kernel/cpu/cpufreq/powernow-k8.c13
-rw-r--r--arch/x86/kernel/cpu/cpufreq/speedstep-ich.c2
-rw-r--r--arch/x86/kernel/cpu/intel_cacheinfo.c24
-rw-r--r--arch/x86/kernel/cpu/mcheck/mce_64.c10
-rw-r--r--arch/x86/kernel/cpu/mcheck/mce_amd_64.c6
-rw-r--r--arch/x86/kernel/cpu/mcheck/mce_intel_64.c2
-rw-r--r--arch/x86/kernel/cpu/mtrr/generic.c3
-rw-r--r--arch/x86/kernel/cpu/proc.c6
-rw-r--r--arch/x86/kernel/dumpstack.c6
-rw-r--r--arch/x86/kernel/ftrace.c195
-rw-r--r--arch/x86/kernel/irq.c2
-rw-r--r--arch/x86/kernel/kprobes.c17
-rw-r--r--arch/x86/kernel/microcode_amd.c43
-rw-r--r--arch/x86/kernel/microcode_core.c127
-rw-r--r--arch/x86/kernel/microcode_intel.c83
-rw-r--r--arch/x86/kernel/mpparse.c7
-rw-r--r--arch/x86/kernel/pci-dma.c8
-rw-r--r--arch/x86/kernel/pci-nommu.c2
-rw-r--r--arch/x86/kernel/process.c25
-rw-r--r--arch/x86/kernel/ptrace.c8
-rw-r--r--arch/x86/kernel/reboot.c8
-rw-r--r--arch/x86/kernel/smpboot.c70
-rw-r--r--arch/x86/kernel/tlb_uv.c9
-rw-r--r--arch/x86/kernel/xsave.c2
-rw-r--r--arch/x86/kvm/Kconfig3
-rw-r--r--arch/x86/mm/Makefile2
-rw-r--r--arch/x86/mm/gup.c16
-rw-r--r--arch/x86/mm/init_32.c35
-rw-r--r--arch/x86/mm/init_64.c37
-rw-r--r--arch/x86/mm/iomap_32.c2
-rw-r--r--arch/x86/mm/ioremap.c23
-rw-r--r--arch/x86/mm/mmio-mod.c19
-rw-r--r--arch/x86/mm/numa.c67
-rw-r--r--arch/x86/mm/numa_64.c111
-rw-r--r--arch/x86/mm/pat.c2
-rw-r--r--arch/x86/mm/pgtable.c3
-rw-r--r--arch/x86/mm/srat_64.c30
-rw-r--r--arch/x86/oprofile/op_model_p4.c2
-rw-r--r--arch/x86/pci/i386.c46
-rw-r--r--arch/x86/power/Makefile5
-rw-r--r--arch/x86/xen/enlighten.c89
-rw-r--r--arch/x86/xen/mmu.c118
-rw-r--r--arch/x86/xen/mmu.h3
-rw-r--r--arch/x86/xen/smp.c10
-rw-r--r--arch/x86/xen/xen-ops.h2
88 files changed, 1414 insertions, 919 deletions
diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
index 2817ab5a1204..c9086e6307a5 100644
--- a/arch/x86/Kconfig
+++ b/arch/x86/Kconfig
@@ -34,6 +34,8 @@ config X86
34 select HAVE_FUNCTION_TRACER 34 select HAVE_FUNCTION_TRACER
35 select HAVE_FUNCTION_GRAPH_TRACER 35 select HAVE_FUNCTION_GRAPH_TRACER
36 select HAVE_FUNCTION_TRACE_MCOUNT_TEST 36 select HAVE_FUNCTION_TRACE_MCOUNT_TEST
37 select HAVE_FTRACE_NMI_ENTER if DYNAMIC_FTRACE
38 select HAVE_FTRACE_SYSCALLS
37 select HAVE_KVM 39 select HAVE_KVM
38 select HAVE_ARCH_KGDB 40 select HAVE_ARCH_KGDB
39 select HAVE_ARCH_TRACEHOOK 41 select HAVE_ARCH_TRACEHOOK
@@ -250,16 +252,13 @@ config SMP
250 252
251config X86_X2APIC 253config X86_X2APIC
252 bool "Support x2apic" 254 bool "Support x2apic"
253 depends on X86_LOCAL_APIC && X86_64 255 depends on X86_LOCAL_APIC && X86_64 && INTR_REMAP
254 ---help--- 256 ---help---
255 This enables x2apic support on CPUs that have this feature. 257 This enables x2apic support on CPUs that have this feature.
256 258
257 This allows 32-bit apic IDs (so it can support very large systems), 259 This allows 32-bit apic IDs (so it can support very large systems),
258 and accesses the local apic via MSRs not via mmio. 260 and accesses the local apic via MSRs not via mmio.
259 261
260 ( On certain CPU models you may need to enable INTR_REMAP too,
261 to get functional x2apic mode. )
262
263 If you don't know what to do here, say N. 262 If you don't know what to do here, say N.
264 263
265config SPARSE_IRQ 264config SPARSE_IRQ
@@ -1145,7 +1144,7 @@ config NODES_SHIFT
1145 depends on NEED_MULTIPLE_NODES 1144 depends on NEED_MULTIPLE_NODES
1146 ---help--- 1145 ---help---
1147 Specify the maximum number of NUMA Nodes available on the target 1146 Specify the maximum number of NUMA Nodes available on the target
1148 system. Increases memory reserved to accomodate various tables. 1147 system. Increases memory reserved to accommodate various tables.
1149 1148
1150config HAVE_ARCH_BOOTMEM 1149config HAVE_ARCH_BOOTMEM
1151 def_bool y 1150 def_bool y
@@ -1323,7 +1322,7 @@ config MTRR_SANITIZER
1323 add writeback entries. 1322 add writeback entries.
1324 1323
1325 Can be disabled with disable_mtrr_cleanup on the kernel command line. 1324 Can be disabled with disable_mtrr_cleanup on the kernel command line.
1326 The largest mtrr entry size for a continous block can be set with 1325 The largest mtrr entry size for a continuous block can be set with
1327 mtrr_chunk_size. 1326 mtrr_chunk_size.
1328 1327
1329 If unsure, say Y. 1328 If unsure, say Y.
@@ -1838,8 +1837,8 @@ config PCI_MMCONFIG
1838 1837
1839config DMAR 1838config DMAR
1840 bool "Support for DMA Remapping Devices (EXPERIMENTAL)" 1839 bool "Support for DMA Remapping Devices (EXPERIMENTAL)"
1841 depends on X86_64 && PCI_MSI && ACPI && EXPERIMENTAL 1840 depends on PCI_MSI && ACPI && EXPERIMENTAL
1842 ---help--- 1841 help
1843 DMA remapping (DMAR) devices support enables independent address 1842 DMA remapping (DMAR) devices support enables independent address
1844 translations for Direct Memory Access (DMA) from devices. 1843 translations for Direct Memory Access (DMA) from devices.
1845 These DMA remapping devices are reported via ACPI tables 1844 These DMA remapping devices are reported via ACPI tables
@@ -1880,7 +1879,6 @@ config DMAR_FLOPPY_WA
1880config INTR_REMAP 1879config INTR_REMAP
1881 bool "Support for Interrupt Remapping (EXPERIMENTAL)" 1880 bool "Support for Interrupt Remapping (EXPERIMENTAL)"
1882 depends on X86_64 && X86_IO_APIC && PCI_MSI && ACPI && EXPERIMENTAL 1881 depends on X86_64 && X86_IO_APIC && PCI_MSI && ACPI && EXPERIMENTAL
1883 select X86_X2APIC
1884 ---help--- 1882 ---help---
1885 Supports Interrupt remapping for IO-APIC and MSI devices. 1883 Supports Interrupt remapping for IO-APIC and MSI devices.
1886 To use x2apic mode in the CPU's which support x2APIC enhancements or 1884 To use x2apic mode in the CPU's which support x2APIC enhancements or
diff --git a/arch/x86/boot/video-vga.c b/arch/x86/boot/video-vga.c
index 95d86ce0421c..9e0587a37768 100644
--- a/arch/x86/boot/video-vga.c
+++ b/arch/x86/boot/video-vga.c
@@ -129,22 +129,18 @@ u16 vga_crtc(void)
129 return (inb(0x3cc) & 1) ? 0x3d4 : 0x3b4; 129 return (inb(0x3cc) & 1) ? 0x3d4 : 0x3b4;
130} 130}
131 131
132static void vga_set_480_scanlines(int lines) 132static void vga_set_480_scanlines(void)
133{ 133{
134 u16 crtc; /* CRTC base address */ 134 u16 crtc; /* CRTC base address */
135 u8 csel; /* CRTC miscellaneous output register */ 135 u8 csel; /* CRTC miscellaneous output register */
136 u8 ovfw; /* CRTC overflow register */
137 int end = lines-1;
138 136
139 crtc = vga_crtc(); 137 crtc = vga_crtc();
140 138
141 ovfw = 0x3c | ((end >> (8-1)) & 0x02) | ((end >> (9-6)) & 0x40);
142
143 out_idx(0x0c, crtc, 0x11); /* Vertical sync end, unlock CR0-7 */ 139 out_idx(0x0c, crtc, 0x11); /* Vertical sync end, unlock CR0-7 */
144 out_idx(0x0b, crtc, 0x06); /* Vertical total */ 140 out_idx(0x0b, crtc, 0x06); /* Vertical total */
145 out_idx(ovfw, crtc, 0x07); /* Vertical overflow */ 141 out_idx(0x3e, crtc, 0x07); /* Vertical overflow */
146 out_idx(0xea, crtc, 0x10); /* Vertical sync start */ 142 out_idx(0xea, crtc, 0x10); /* Vertical sync start */
147 out_idx(end, crtc, 0x12); /* Vertical display end */ 143 out_idx(0xdf, crtc, 0x12); /* Vertical display end */
148 out_idx(0xe7, crtc, 0x15); /* Vertical blank start */ 144 out_idx(0xe7, crtc, 0x15); /* Vertical blank start */
149 out_idx(0x04, crtc, 0x16); /* Vertical blank end */ 145 out_idx(0x04, crtc, 0x16); /* Vertical blank end */
150 csel = inb(0x3cc); 146 csel = inb(0x3cc);
@@ -153,21 +149,38 @@ static void vga_set_480_scanlines(int lines)
153 outb(csel, 0x3c2); 149 outb(csel, 0x3c2);
154} 150}
155 151
152static void vga_set_vertical_end(int lines)
153{
154 u16 crtc; /* CRTC base address */
155 u8 ovfw; /* CRTC overflow register */
156 int end = lines-1;
157
158 crtc = vga_crtc();
159
160 ovfw = 0x3c | ((end >> (8-1)) & 0x02) | ((end >> (9-6)) & 0x40);
161
162 out_idx(ovfw, crtc, 0x07); /* Vertical overflow */
163 out_idx(end, crtc, 0x12); /* Vertical display end */
164}
165
156static void vga_set_80x30(void) 166static void vga_set_80x30(void)
157{ 167{
158 vga_set_480_scanlines(30*16); 168 vga_set_480_scanlines();
169 vga_set_vertical_end(30*16);
159} 170}
160 171
161static void vga_set_80x34(void) 172static void vga_set_80x34(void)
162{ 173{
174 vga_set_480_scanlines();
163 vga_set_14font(); 175 vga_set_14font();
164 vga_set_480_scanlines(34*14); 176 vga_set_vertical_end(34*14);
165} 177}
166 178
167static void vga_set_80x60(void) 179static void vga_set_80x60(void)
168{ 180{
181 vga_set_480_scanlines();
169 vga_set_8font(); 182 vga_set_8font();
170 vga_set_480_scanlines(60*8); 183 vga_set_vertical_end(60*8);
171} 184}
172 185
173static int vga_set_mode(struct mode_info *mode) 186static int vga_set_mode(struct mode_info *mode)
diff --git a/arch/x86/include/asm/apic.h b/arch/x86/include/asm/apic.h
index df8a300dfe6c..42f2f8377422 100644
--- a/arch/x86/include/asm/apic.h
+++ b/arch/x86/include/asm/apic.h
@@ -107,6 +107,9 @@ extern u32 native_safe_apic_wait_icr_idle(void);
107extern void native_apic_icr_write(u32 low, u32 id); 107extern void native_apic_icr_write(u32 low, u32 id);
108extern u64 native_apic_icr_read(void); 108extern u64 native_apic_icr_read(void);
109 109
110#define EIM_8BIT_APIC_ID 0
111#define EIM_32BIT_APIC_ID 1
112
110#ifdef CONFIG_X86_X2APIC 113#ifdef CONFIG_X86_X2APIC
111/* 114/*
112 * Make previous memory operations globally visible before 115 * Make previous memory operations globally visible before
diff --git a/arch/x86/include/asm/cacheflush.h b/arch/x86/include/asm/cacheflush.h
index b3894bf52fcd..e55dfc1ad453 100644
--- a/arch/x86/include/asm/cacheflush.h
+++ b/arch/x86/include/asm/cacheflush.h
@@ -126,6 +126,11 @@ void clflush_cache_range(void *addr, unsigned int size);
126#ifdef CONFIG_DEBUG_RODATA 126#ifdef CONFIG_DEBUG_RODATA
127void mark_rodata_ro(void); 127void mark_rodata_ro(void);
128extern const int rodata_test_data; 128extern const int rodata_test_data;
129void set_kernel_text_rw(void);
130void set_kernel_text_ro(void);
131#else
132static inline void set_kernel_text_rw(void) { }
133static inline void set_kernel_text_ro(void) { }
129#endif 134#endif
130 135
131#ifdef CONFIG_DEBUG_RODATA_TEST 136#ifdef CONFIG_DEBUG_RODATA_TEST
diff --git a/arch/x86/include/asm/cpu_debug.h b/arch/x86/include/asm/cpu_debug.h
index 222802029fa6..222802029fa6 100755..100644
--- a/arch/x86/include/asm/cpu_debug.h
+++ b/arch/x86/include/asm/cpu_debug.h
diff --git a/arch/x86/include/asm/cpufeature.h b/arch/x86/include/asm/cpufeature.h
index 0beba0d1468d..bb83b1c397aa 100644
--- a/arch/x86/include/asm/cpufeature.h
+++ b/arch/x86/include/asm/cpufeature.h
@@ -154,6 +154,7 @@
154 * CPUID levels like 0x6, 0xA etc 154 * CPUID levels like 0x6, 0xA etc
155 */ 155 */
156#define X86_FEATURE_IDA (7*32+ 0) /* Intel Dynamic Acceleration */ 156#define X86_FEATURE_IDA (7*32+ 0) /* Intel Dynamic Acceleration */
157#define X86_FEATURE_ARAT (7*32+ 1) /* Always Running APIC Timer */
157 158
158/* Virtualization flags: Linux defined */ 159/* Virtualization flags: Linux defined */
159#define X86_FEATURE_TPR_SHADOW (8*32+ 0) /* Intel TPR Shadow */ 160#define X86_FEATURE_TPR_SHADOW (8*32+ 0) /* Intel TPR Shadow */
diff --git a/arch/x86/include/asm/cpumask.h b/arch/x86/include/asm/cpumask.h
index a7f3c75f8ad7..61c852fa346b 100644
--- a/arch/x86/include/asm/cpumask.h
+++ b/arch/x86/include/asm/cpumask.h
@@ -3,8 +3,6 @@
3#ifndef __ASSEMBLY__ 3#ifndef __ASSEMBLY__
4#include <linux/cpumask.h> 4#include <linux/cpumask.h>
5 5
6#ifdef CONFIG_X86_64
7
8extern cpumask_var_t cpu_callin_mask; 6extern cpumask_var_t cpu_callin_mask;
9extern cpumask_var_t cpu_callout_mask; 7extern cpumask_var_t cpu_callout_mask;
10extern cpumask_var_t cpu_initialized_mask; 8extern cpumask_var_t cpu_initialized_mask;
@@ -12,21 +10,5 @@ extern cpumask_var_t cpu_sibling_setup_mask;
12 10
13extern void setup_cpu_local_masks(void); 11extern void setup_cpu_local_masks(void);
14 12
15#else /* CONFIG_X86_32 */
16
17extern cpumask_t cpu_callin_map;
18extern cpumask_t cpu_callout_map;
19extern cpumask_t cpu_initialized;
20extern cpumask_t cpu_sibling_setup_map;
21
22#define cpu_callin_mask ((struct cpumask *)&cpu_callin_map)
23#define cpu_callout_mask ((struct cpumask *)&cpu_callout_map)
24#define cpu_initialized_mask ((struct cpumask *)&cpu_initialized)
25#define cpu_sibling_setup_mask ((struct cpumask *)&cpu_sibling_setup_map)
26
27static inline void setup_cpu_local_masks(void) { }
28
29#endif /* CONFIG_X86_32 */
30
31#endif /* __ASSEMBLY__ */ 13#endif /* __ASSEMBLY__ */
32#endif /* _ASM_X86_CPUMASK_H */ 14#endif /* _ASM_X86_CPUMASK_H */
diff --git a/arch/x86/include/asm/dma-mapping.h b/arch/x86/include/asm/dma-mapping.h
index cea7b74963e9..f82fdc412c64 100644
--- a/arch/x86/include/asm/dma-mapping.h
+++ b/arch/x86/include/asm/dma-mapping.h
@@ -238,7 +238,7 @@ static inline unsigned long dma_alloc_coherent_mask(struct device *dev,
238 238
239 dma_mask = dev->coherent_dma_mask; 239 dma_mask = dev->coherent_dma_mask;
240 if (!dma_mask) 240 if (!dma_mask)
241 dma_mask = (gfp & GFP_DMA) ? DMA_24BIT_MASK : DMA_32BIT_MASK; 241 dma_mask = (gfp & GFP_DMA) ? DMA_BIT_MASK(24) : DMA_BIT_MASK(32);
242 242
243 return dma_mask; 243 return dma_mask;
244} 244}
@@ -247,10 +247,10 @@ static inline gfp_t dma_alloc_coherent_gfp_flags(struct device *dev, gfp_t gfp)
247{ 247{
248 unsigned long dma_mask = dma_alloc_coherent_mask(dev, gfp); 248 unsigned long dma_mask = dma_alloc_coherent_mask(dev, gfp);
249 249
250 if (dma_mask <= DMA_24BIT_MASK) 250 if (dma_mask <= DMA_BIT_MASK(24))
251 gfp |= GFP_DMA; 251 gfp |= GFP_DMA;
252#ifdef CONFIG_X86_64 252#ifdef CONFIG_X86_64
253 if (dma_mask <= DMA_32BIT_MASK && !(gfp & GFP_DMA)) 253 if (dma_mask <= DMA_BIT_MASK(32) && !(gfp & GFP_DMA))
254 gfp |= GFP_DMA32; 254 gfp |= GFP_DMA32;
255#endif 255#endif
256 return gfp; 256 return gfp;
diff --git a/arch/x86/include/asm/fixmap.h b/arch/x86/include/asm/fixmap.h
index 63a79c77d220..2d81af3974a0 100644
--- a/arch/x86/include/asm/fixmap.h
+++ b/arch/x86/include/asm/fixmap.h
@@ -111,6 +111,8 @@ enum fixed_addresses {
111#ifdef CONFIG_PARAVIRT 111#ifdef CONFIG_PARAVIRT
112 FIX_PARAVIRT_BOOTMAP, 112 FIX_PARAVIRT_BOOTMAP,
113#endif 113#endif
114 FIX_TEXT_POKE0, /* reserve 2 pages for text_poke() */
115 FIX_TEXT_POKE1,
114 __end_of_permanent_fixed_addresses, 116 __end_of_permanent_fixed_addresses,
115#ifdef CONFIG_PROVIDE_OHCI1394_DMA_INIT 117#ifdef CONFIG_PROVIDE_OHCI1394_DMA_INIT
116 FIX_OHCI1394_BASE, 118 FIX_OHCI1394_BASE,
@@ -149,11 +151,11 @@ extern pte_t *pkmap_page_table;
149 151
150void __native_set_fixmap(enum fixed_addresses idx, pte_t pte); 152void __native_set_fixmap(enum fixed_addresses idx, pte_t pte);
151void native_set_fixmap(enum fixed_addresses idx, 153void native_set_fixmap(enum fixed_addresses idx,
152 unsigned long phys, pgprot_t flags); 154 phys_addr_t phys, pgprot_t flags);
153 155
154#ifndef CONFIG_PARAVIRT 156#ifndef CONFIG_PARAVIRT
155static inline void __set_fixmap(enum fixed_addresses idx, 157static inline void __set_fixmap(enum fixed_addresses idx,
156 unsigned long phys, pgprot_t flags) 158 phys_addr_t phys, pgprot_t flags)
157{ 159{
158 native_set_fixmap(idx, phys, flags); 160 native_set_fixmap(idx, phys, flags);
159} 161}
diff --git a/arch/x86/include/asm/ftrace.h b/arch/x86/include/asm/ftrace.h
index db24c2278be0..bd2c6511c887 100644
--- a/arch/x86/include/asm/ftrace.h
+++ b/arch/x86/include/asm/ftrace.h
@@ -28,6 +28,13 @@
28 28
29#endif 29#endif
30 30
31/* FIXME: I don't want to stay hardcoded */
32#ifdef CONFIG_X86_64
33# define FTRACE_SYSCALL_MAX 296
34#else
35# define FTRACE_SYSCALL_MAX 333
36#endif
37
31#ifdef CONFIG_FUNCTION_TRACER 38#ifdef CONFIG_FUNCTION_TRACER
32#define MCOUNT_ADDR ((long)(mcount)) 39#define MCOUNT_ADDR ((long)(mcount))
33#define MCOUNT_INSN_SIZE 5 /* sizeof mcount call */ 40#define MCOUNT_INSN_SIZE 5 /* sizeof mcount call */
diff --git a/arch/x86/include/asm/io.h b/arch/x86/include/asm/io.h
index e5383e3d2f8c..73739322b6d0 100644
--- a/arch/x86/include/asm/io.h
+++ b/arch/x86/include/asm/io.h
@@ -193,8 +193,10 @@ extern void __iomem *ioremap_wc(resource_size_t offset, unsigned long size);
193 */ 193 */
194extern void early_ioremap_init(void); 194extern void early_ioremap_init(void);
195extern void early_ioremap_reset(void); 195extern void early_ioremap_reset(void);
196extern void __iomem *early_ioremap(unsigned long offset, unsigned long size); 196extern void __iomem *early_ioremap(resource_size_t phys_addr,
197extern void __iomem *early_memremap(unsigned long offset, unsigned long size); 197 unsigned long size);
198extern void __iomem *early_memremap(resource_size_t phys_addr,
199 unsigned long size);
198extern void early_iounmap(void __iomem *addr, unsigned long size); 200extern void early_iounmap(void __iomem *addr, unsigned long size);
199 201
200#define IO_SPACE_LIMIT 0xffff 202#define IO_SPACE_LIMIT 0xffff
diff --git a/arch/x86/include/asm/io_apic.h b/arch/x86/include/asm/io_apic.h
index 373cc2bbcad2..9d826e436010 100644
--- a/arch/x86/include/asm/io_apic.h
+++ b/arch/x86/include/asm/io_apic.h
@@ -162,10 +162,13 @@ extern int (*ioapic_renumber_irq)(int ioapic, int irq);
162extern void ioapic_init_mappings(void); 162extern void ioapic_init_mappings(void);
163 163
164#ifdef CONFIG_X86_64 164#ifdef CONFIG_X86_64
165extern int save_IO_APIC_setup(void); 165extern struct IO_APIC_route_entry **alloc_ioapic_entries(void);
166extern void mask_IO_APIC_setup(void); 166extern void free_ioapic_entries(struct IO_APIC_route_entry **ioapic_entries);
167extern void restore_IO_APIC_setup(void); 167extern int save_IO_APIC_setup(struct IO_APIC_route_entry **ioapic_entries);
168extern void reinit_intr_remapped_IO_APIC(int); 168extern void mask_IO_APIC_setup(struct IO_APIC_route_entry **ioapic_entries);
169extern int restore_IO_APIC_setup(struct IO_APIC_route_entry **ioapic_entries);
170extern void reinit_intr_remapped_IO_APIC(int intr_remapping,
171 struct IO_APIC_route_entry **ioapic_entries);
169#endif 172#endif
170 173
171extern void probe_nr_irqs_gsi(void); 174extern void probe_nr_irqs_gsi(void);
diff --git a/arch/x86/include/asm/paravirt.h b/arch/x86/include/asm/paravirt.h
index 7727aa8b7dda..378e3691c08c 100644
--- a/arch/x86/include/asm/paravirt.h
+++ b/arch/x86/include/asm/paravirt.h
@@ -347,7 +347,7 @@ struct pv_mmu_ops {
347 /* Sometimes the physical address is a pfn, and sometimes its 347 /* Sometimes the physical address is a pfn, and sometimes its
348 an mfn. We can tell which is which from the index. */ 348 an mfn. We can tell which is which from the index. */
349 void (*set_fixmap)(unsigned /* enum fixed_addresses */ idx, 349 void (*set_fixmap)(unsigned /* enum fixed_addresses */ idx,
350 unsigned long phys, pgprot_t flags); 350 phys_addr_t phys, pgprot_t flags);
351}; 351};
352 352
353struct raw_spinlock; 353struct raw_spinlock;
@@ -1432,7 +1432,7 @@ static inline void arch_leave_lazy_mmu_mode(void)
1432void arch_flush_lazy_mmu_mode(void); 1432void arch_flush_lazy_mmu_mode(void);
1433 1433
1434static inline void __set_fixmap(unsigned /* enum fixed_addresses */ idx, 1434static inline void __set_fixmap(unsigned /* enum fixed_addresses */ idx,
1435 unsigned long phys, pgprot_t flags) 1435 phys_addr_t phys, pgprot_t flags)
1436{ 1436{
1437 pv_mmu_ops.set_fixmap(idx, phys, flags); 1437 pv_mmu_ops.set_fixmap(idx, phys, flags);
1438} 1438}
diff --git a/arch/x86/include/asm/pci.h b/arch/x86/include/asm/pci.h
index e545ea01abcf..b51a1e8b0baf 100644
--- a/arch/x86/include/asm/pci.h
+++ b/arch/x86/include/asm/pci.h
@@ -140,11 +140,6 @@ static inline int __pcibus_to_node(const struct pci_bus *bus)
140 return sd->node; 140 return sd->node;
141} 141}
142 142
143static inline cpumask_t __pcibus_to_cpumask(struct pci_bus *bus)
144{
145 return node_to_cpumask(__pcibus_to_node(bus));
146}
147
148static inline const struct cpumask * 143static inline const struct cpumask *
149cpumask_of_pcibus(const struct pci_bus *bus) 144cpumask_of_pcibus(const struct pci_bus *bus)
150{ 145{
diff --git a/arch/x86/include/asm/processor.h b/arch/x86/include/asm/processor.h
index ae85a8d66a30..fcf4d92e7e04 100644
--- a/arch/x86/include/asm/processor.h
+++ b/arch/x86/include/asm/processor.h
@@ -94,7 +94,7 @@ struct cpuinfo_x86 {
94 unsigned long loops_per_jiffy; 94 unsigned long loops_per_jiffy;
95#ifdef CONFIG_SMP 95#ifdef CONFIG_SMP
96 /* cpus sharing the last level cache: */ 96 /* cpus sharing the last level cache: */
97 cpumask_t llc_shared_map; 97 cpumask_var_t llc_shared_map;
98#endif 98#endif
99 /* cpuid returned max cores value: */ 99 /* cpuid returned max cores value: */
100 u16 x86_max_cores; 100 u16 x86_max_cores;
@@ -352,6 +352,11 @@ struct i387_soft_struct {
352 u32 entry_eip; 352 u32 entry_eip;
353}; 353};
354 354
355struct ymmh_struct {
356 /* 16 * 16 bytes for each YMMH-reg = 256 bytes */
357 u32 ymmh_space[64];
358};
359
355struct xsave_hdr_struct { 360struct xsave_hdr_struct {
356 u64 xstate_bv; 361 u64 xstate_bv;
357 u64 reserved1[2]; 362 u64 reserved1[2];
@@ -361,6 +366,7 @@ struct xsave_hdr_struct {
361struct xsave_struct { 366struct xsave_struct {
362 struct i387_fxsave_struct i387; 367 struct i387_fxsave_struct i387;
363 struct xsave_hdr_struct xsave_hdr; 368 struct xsave_hdr_struct xsave_hdr;
369 struct ymmh_struct ymmh;
364 /* new processor state extensions will go here */ 370 /* new processor state extensions will go here */
365} __attribute__ ((packed, aligned (64))); 371} __attribute__ ((packed, aligned (64)));
366 372
@@ -736,6 +742,7 @@ static inline void __sti_mwait(unsigned long eax, unsigned long ecx)
736extern void mwait_idle_with_hints(unsigned long eax, unsigned long ecx); 742extern void mwait_idle_with_hints(unsigned long eax, unsigned long ecx);
737 743
738extern void select_idle_routine(const struct cpuinfo_x86 *c); 744extern void select_idle_routine(const struct cpuinfo_x86 *c);
745extern void init_c1e_mask(void);
739 746
740extern unsigned long boot_option_idle_override; 747extern unsigned long boot_option_idle_override;
741extern unsigned long idle_halt; 748extern unsigned long idle_halt;
diff --git a/arch/x86/include/asm/ptrace-abi.h b/arch/x86/include/asm/ptrace-abi.h
index 8e0f8d199e05..86723035a515 100644
--- a/arch/x86/include/asm/ptrace-abi.h
+++ b/arch/x86/include/asm/ptrace-abi.h
@@ -80,8 +80,6 @@
80 80
81#define PTRACE_SINGLEBLOCK 33 /* resume execution until next branch */ 81#define PTRACE_SINGLEBLOCK 33 /* resume execution until next branch */
82 82
83#ifdef CONFIG_X86_PTRACE_BTS
84
85#ifndef __ASSEMBLY__ 83#ifndef __ASSEMBLY__
86#include <linux/types.h> 84#include <linux/types.h>
87 85
@@ -140,6 +138,5 @@ struct ptrace_bts_config {
140 BTS records are read from oldest to newest. 138 BTS records are read from oldest to newest.
141 Returns number of BTS records drained. 139 Returns number of BTS records drained.
142*/ 140*/
143#endif /* CONFIG_X86_PTRACE_BTS */
144 141
145#endif /* _ASM_X86_PTRACE_ABI_H */ 142#endif /* _ASM_X86_PTRACE_ABI_H */
diff --git a/arch/x86/include/asm/required-features.h b/arch/x86/include/asm/required-features.h
index d5cd6c586881..a4737dddfd58 100644
--- a/arch/x86/include/asm/required-features.h
+++ b/arch/x86/include/asm/required-features.h
@@ -50,7 +50,7 @@
50#ifdef CONFIG_X86_64 50#ifdef CONFIG_X86_64
51#define NEED_PSE 0 51#define NEED_PSE 0
52#define NEED_MSR (1<<(X86_FEATURE_MSR & 31)) 52#define NEED_MSR (1<<(X86_FEATURE_MSR & 31))
53#define NEED_PGE (1<<(X86_FEATURE_PGE & 31)) 53#define NEED_PGE 0
54#define NEED_FXSR (1<<(X86_FEATURE_FXSR & 31)) 54#define NEED_FXSR (1<<(X86_FEATURE_FXSR & 31))
55#define NEED_XMM (1<<(X86_FEATURE_XMM & 31)) 55#define NEED_XMM (1<<(X86_FEATURE_XMM & 31))
56#define NEED_XMM2 (1<<(X86_FEATURE_XMM2 & 31)) 56#define NEED_XMM2 (1<<(X86_FEATURE_XMM2 & 31))
diff --git a/arch/x86/include/asm/sigcontext.h b/arch/x86/include/asm/sigcontext.h
index ec666491aaa4..72e5a4491661 100644
--- a/arch/x86/include/asm/sigcontext.h
+++ b/arch/x86/include/asm/sigcontext.h
@@ -269,6 +269,11 @@ struct _xsave_hdr {
269 __u64 reserved2[5]; 269 __u64 reserved2[5];
270}; 270};
271 271
272struct _ymmh_state {
273 /* 16 * 16 bytes for each YMMH-reg */
274 __u32 ymmh_space[64];
275};
276
272/* 277/*
273 * Extended state pointed by the fpstate pointer in the sigcontext. 278 * Extended state pointed by the fpstate pointer in the sigcontext.
274 * In addition to the fpstate, information encoded in the xstate_hdr 279 * In addition to the fpstate, information encoded in the xstate_hdr
@@ -278,6 +283,7 @@ struct _xsave_hdr {
278struct _xstate { 283struct _xstate {
279 struct _fpstate fpstate; 284 struct _fpstate fpstate;
280 struct _xsave_hdr xstate_hdr; 285 struct _xsave_hdr xstate_hdr;
286 struct _ymmh_state ymmh;
281 /* new processor state extensions go here */ 287 /* new processor state extensions go here */
282}; 288};
283 289
diff --git a/arch/x86/include/asm/smp.h b/arch/x86/include/asm/smp.h
index 47d0e21f2b9e..19e0d88b966d 100644
--- a/arch/x86/include/asm/smp.h
+++ b/arch/x86/include/asm/smp.h
@@ -21,19 +21,19 @@
21extern int smp_num_siblings; 21extern int smp_num_siblings;
22extern unsigned int num_processors; 22extern unsigned int num_processors;
23 23
24DECLARE_PER_CPU(cpumask_t, cpu_sibling_map); 24DECLARE_PER_CPU(cpumask_var_t, cpu_sibling_map);
25DECLARE_PER_CPU(cpumask_t, cpu_core_map); 25DECLARE_PER_CPU(cpumask_var_t, cpu_core_map);
26DECLARE_PER_CPU(u16, cpu_llc_id); 26DECLARE_PER_CPU(u16, cpu_llc_id);
27DECLARE_PER_CPU(int, cpu_number); 27DECLARE_PER_CPU(int, cpu_number);
28 28
29static inline struct cpumask *cpu_sibling_mask(int cpu) 29static inline struct cpumask *cpu_sibling_mask(int cpu)
30{ 30{
31 return &per_cpu(cpu_sibling_map, cpu); 31 return per_cpu(cpu_sibling_map, cpu);
32} 32}
33 33
34static inline struct cpumask *cpu_core_mask(int cpu) 34static inline struct cpumask *cpu_core_mask(int cpu)
35{ 35{
36 return &per_cpu(cpu_core_map, cpu); 36 return per_cpu(cpu_core_map, cpu);
37} 37}
38 38
39DECLARE_EARLY_PER_CPU(u16, x86_cpu_to_apicid); 39DECLARE_EARLY_PER_CPU(u16, x86_cpu_to_apicid);
@@ -121,9 +121,10 @@ static inline void arch_send_call_function_single_ipi(int cpu)
121 smp_ops.send_call_func_single_ipi(cpu); 121 smp_ops.send_call_func_single_ipi(cpu);
122} 122}
123 123
124static inline void arch_send_call_function_ipi(cpumask_t mask) 124#define arch_send_call_function_ipi_mask arch_send_call_function_ipi_mask
125static inline void arch_send_call_function_ipi_mask(const struct cpumask *mask)
125{ 126{
126 smp_ops.send_call_func_ipi(&mask); 127 smp_ops.send_call_func_ipi(mask);
127} 128}
128 129
129void cpu_disable_common(void); 130void cpu_disable_common(void);
diff --git a/arch/x86/include/asm/thread_info.h b/arch/x86/include/asm/thread_info.h
index df9d5f78385e..8820a73ae090 100644
--- a/arch/x86/include/asm/thread_info.h
+++ b/arch/x86/include/asm/thread_info.h
@@ -94,6 +94,7 @@ struct thread_info {
94#define TIF_FORCED_TF 24 /* true if TF in eflags artificially */ 94#define TIF_FORCED_TF 24 /* true if TF in eflags artificially */
95#define TIF_DEBUGCTLMSR 25 /* uses thread_struct.debugctlmsr */ 95#define TIF_DEBUGCTLMSR 25 /* uses thread_struct.debugctlmsr */
96#define TIF_DS_AREA_MSR 26 /* uses thread_struct.ds_area_msr */ 96#define TIF_DS_AREA_MSR 26 /* uses thread_struct.ds_area_msr */
97#define TIF_SYSCALL_FTRACE 27 /* for ftrace syscall instrumentation */
97 98
98#define _TIF_SYSCALL_TRACE (1 << TIF_SYSCALL_TRACE) 99#define _TIF_SYSCALL_TRACE (1 << TIF_SYSCALL_TRACE)
99#define _TIF_NOTIFY_RESUME (1 << TIF_NOTIFY_RESUME) 100#define _TIF_NOTIFY_RESUME (1 << TIF_NOTIFY_RESUME)
@@ -115,15 +116,17 @@ struct thread_info {
115#define _TIF_FORCED_TF (1 << TIF_FORCED_TF) 116#define _TIF_FORCED_TF (1 << TIF_FORCED_TF)
116#define _TIF_DEBUGCTLMSR (1 << TIF_DEBUGCTLMSR) 117#define _TIF_DEBUGCTLMSR (1 << TIF_DEBUGCTLMSR)
117#define _TIF_DS_AREA_MSR (1 << TIF_DS_AREA_MSR) 118#define _TIF_DS_AREA_MSR (1 << TIF_DS_AREA_MSR)
119#define _TIF_SYSCALL_FTRACE (1 << TIF_SYSCALL_FTRACE)
118 120
119/* work to do in syscall_trace_enter() */ 121/* work to do in syscall_trace_enter() */
120#define _TIF_WORK_SYSCALL_ENTRY \ 122#define _TIF_WORK_SYSCALL_ENTRY \
121 (_TIF_SYSCALL_TRACE | _TIF_SYSCALL_EMU | \ 123 (_TIF_SYSCALL_TRACE | _TIF_SYSCALL_EMU | _TIF_SYSCALL_FTRACE | \
122 _TIF_SYSCALL_AUDIT | _TIF_SECCOMP | _TIF_SINGLESTEP) 124 _TIF_SYSCALL_AUDIT | _TIF_SECCOMP | _TIF_SINGLESTEP)
123 125
124/* work to do in syscall_trace_leave() */ 126/* work to do in syscall_trace_leave() */
125#define _TIF_WORK_SYSCALL_EXIT \ 127#define _TIF_WORK_SYSCALL_EXIT \
126 (_TIF_SYSCALL_TRACE | _TIF_SYSCALL_AUDIT | _TIF_SINGLESTEP) 128 (_TIF_SYSCALL_TRACE | _TIF_SYSCALL_AUDIT | _TIF_SINGLESTEP | \
129 _TIF_SYSCALL_FTRACE)
127 130
128/* work to do on interrupt/exception return */ 131/* work to do on interrupt/exception return */
129#define _TIF_WORK_MASK \ 132#define _TIF_WORK_MASK \
@@ -132,7 +135,7 @@ struct thread_info {
132 _TIF_SINGLESTEP|_TIF_SECCOMP|_TIF_SYSCALL_EMU)) 135 _TIF_SINGLESTEP|_TIF_SECCOMP|_TIF_SYSCALL_EMU))
133 136
134/* work to do on any return to user space */ 137/* work to do on any return to user space */
135#define _TIF_ALLWORK_MASK (0x0000FFFF & ~_TIF_SECCOMP) 138#define _TIF_ALLWORK_MASK ((0x0000FFFF & ~_TIF_SECCOMP) | _TIF_SYSCALL_FTRACE)
136 139
137/* Only used for 64 bit */ 140/* Only used for 64 bit */
138#define _TIF_DO_NOTIFY_MASK \ 141#define _TIF_DO_NOTIFY_MASK \
diff --git a/arch/x86/include/asm/topology.h b/arch/x86/include/asm/topology.h
index 744299c0b774..892b119dba6f 100644
--- a/arch/x86/include/asm/topology.h
+++ b/arch/x86/include/asm/topology.h
@@ -44,9 +44,6 @@
44 44
45#ifdef CONFIG_X86_32 45#ifdef CONFIG_X86_32
46 46
47/* Mappings between node number and cpus on that node. */
48extern cpumask_t node_to_cpumask_map[];
49
50/* Mappings between logical cpu number and node number */ 47/* Mappings between logical cpu number and node number */
51extern int cpu_to_node_map[]; 48extern int cpu_to_node_map[];
52 49
@@ -57,30 +54,8 @@ static inline int cpu_to_node(int cpu)
57} 54}
58#define early_cpu_to_node(cpu) cpu_to_node(cpu) 55#define early_cpu_to_node(cpu) cpu_to_node(cpu)
59 56
60/* Returns a bitmask of CPUs on Node 'node'.
61 *
62 * Side note: this function creates the returned cpumask on the stack
63 * so with a high NR_CPUS count, excessive stack space is used. The
64 * cpumask_of_node function should be used whenever possible.
65 */
66static inline cpumask_t node_to_cpumask(int node)
67{
68 return node_to_cpumask_map[node];
69}
70
71/* Returns a bitmask of CPUs on Node 'node'. */
72static inline const struct cpumask *cpumask_of_node(int node)
73{
74 return &node_to_cpumask_map[node];
75}
76
77static inline void setup_node_to_cpumask_map(void) { }
78
79#else /* CONFIG_X86_64 */ 57#else /* CONFIG_X86_64 */
80 58
81/* Mappings between node number and cpus on that node. */
82extern cpumask_t *node_to_cpumask_map;
83
84/* Mappings between logical cpu number and node number */ 59/* Mappings between logical cpu number and node number */
85DECLARE_EARLY_PER_CPU(int, x86_cpu_to_node_map); 60DECLARE_EARLY_PER_CPU(int, x86_cpu_to_node_map);
86 61
@@ -91,8 +66,6 @@ DECLARE_PER_CPU(int, node_number);
91#ifdef CONFIG_DEBUG_PER_CPU_MAPS 66#ifdef CONFIG_DEBUG_PER_CPU_MAPS
92extern int cpu_to_node(int cpu); 67extern int cpu_to_node(int cpu);
93extern int early_cpu_to_node(int cpu); 68extern int early_cpu_to_node(int cpu);
94extern const cpumask_t *cpumask_of_node(int node);
95extern cpumask_t node_to_cpumask(int node);
96 69
97#else /* !CONFIG_DEBUG_PER_CPU_MAPS */ 70#else /* !CONFIG_DEBUG_PER_CPU_MAPS */
98 71
@@ -108,42 +81,32 @@ static inline int early_cpu_to_node(int cpu)
108 return early_per_cpu(x86_cpu_to_node_map, cpu); 81 return early_per_cpu(x86_cpu_to_node_map, cpu);
109} 82}
110 83
111/* Returns a pointer to the cpumask of CPUs on Node 'node'. */ 84#endif /* !CONFIG_DEBUG_PER_CPU_MAPS */
112static inline const cpumask_t *cpumask_of_node(int node) 85
113{ 86#endif /* CONFIG_X86_64 */
114 return &node_to_cpumask_map[node];
115}
116 87
117/* Returns a bitmask of CPUs on Node 'node'. */ 88/* Mappings between node number and cpus on that node. */
118static inline cpumask_t node_to_cpumask(int node) 89extern cpumask_var_t node_to_cpumask_map[MAX_NUMNODES];
90
91#ifdef CONFIG_DEBUG_PER_CPU_MAPS
92extern const struct cpumask *cpumask_of_node(int node);
93#else
94/* Returns a pointer to the cpumask of CPUs on Node 'node'. */
95static inline const struct cpumask *cpumask_of_node(int node)
119{ 96{
120 return node_to_cpumask_map[node]; 97 return node_to_cpumask_map[node];
121} 98}
122 99#endif
123#endif /* !CONFIG_DEBUG_PER_CPU_MAPS */
124 100
125extern void setup_node_to_cpumask_map(void); 101extern void setup_node_to_cpumask_map(void);
126 102
127/* 103/*
128 * Replace default node_to_cpumask_ptr with optimized version
129 * Deprecated: use "const struct cpumask *mask = cpumask_of_node(node)"
130 */
131#define node_to_cpumask_ptr(v, node) \
132 const cpumask_t *v = cpumask_of_node(node)
133
134#define node_to_cpumask_ptr_next(v, node) \
135 v = cpumask_of_node(node)
136
137#endif /* CONFIG_X86_64 */
138
139/*
140 * Returns the number of the node containing Node 'node'. This 104 * Returns the number of the node containing Node 'node'. This
141 * architecture is flat, so it is a pretty simple function! 105 * architecture is flat, so it is a pretty simple function!
142 */ 106 */
143#define parent_node(node) (node) 107#define parent_node(node) (node)
144 108
145#define pcibus_to_node(bus) __pcibus_to_node(bus) 109#define pcibus_to_node(bus) __pcibus_to_node(bus)
146#define pcibus_to_cpumask(bus) __pcibus_to_cpumask(bus)
147 110
148#ifdef CONFIG_X86_32 111#ifdef CONFIG_X86_32
149extern unsigned long node_start_pfn[]; 112extern unsigned long node_start_pfn[];
@@ -209,40 +172,24 @@ static inline int early_cpu_to_node(int cpu)
209 return 0; 172 return 0;
210} 173}
211 174
212static inline const cpumask_t *cpumask_of_node(int node) 175static inline const struct cpumask *cpumask_of_node(int node)
213{
214 return &cpu_online_map;
215}
216static inline cpumask_t node_to_cpumask(int node)
217{ 176{
218 return cpu_online_map; 177 return cpu_online_mask;
219} 178}
220 179
221static inline void setup_node_to_cpumask_map(void) { } 180static inline void setup_node_to_cpumask_map(void) { }
222 181
223/*
224 * Replace default node_to_cpumask_ptr with optimized version
225 * Deprecated: use "const struct cpumask *mask = cpumask_of_node(node)"
226 */
227#define node_to_cpumask_ptr(v, node) \
228 const cpumask_t *v = cpumask_of_node(node)
229
230#define node_to_cpumask_ptr_next(v, node) \
231 v = cpumask_of_node(node)
232#endif 182#endif
233 183
234#include <asm-generic/topology.h> 184#include <asm-generic/topology.h>
235 185
236extern cpumask_t cpu_coregroup_map(int cpu);
237extern const struct cpumask *cpu_coregroup_mask(int cpu); 186extern const struct cpumask *cpu_coregroup_mask(int cpu);
238 187
239#ifdef ENABLE_TOPO_DEFINES 188#ifdef ENABLE_TOPO_DEFINES
240#define topology_physical_package_id(cpu) (cpu_data(cpu).phys_proc_id) 189#define topology_physical_package_id(cpu) (cpu_data(cpu).phys_proc_id)
241#define topology_core_id(cpu) (cpu_data(cpu).cpu_core_id) 190#define topology_core_id(cpu) (cpu_data(cpu).cpu_core_id)
242#define topology_core_siblings(cpu) (per_cpu(cpu_core_map, cpu)) 191#define topology_core_cpumask(cpu) (per_cpu(cpu_core_map, cpu))
243#define topology_thread_siblings(cpu) (per_cpu(cpu_sibling_map, cpu)) 192#define topology_thread_cpumask(cpu) (per_cpu(cpu_sibling_map, cpu))
244#define topology_core_cpumask(cpu) (&per_cpu(cpu_core_map, cpu))
245#define topology_thread_cpumask(cpu) (&per_cpu(cpu_sibling_map, cpu))
246 193
247/* indicates that pointers to the topology cpumask_t maps are valid */ 194/* indicates that pointers to the topology cpumask_t maps are valid */
248#define arch_provides_topology_pointers yes 195#define arch_provides_topology_pointers yes
@@ -256,7 +203,7 @@ struct pci_bus;
256void set_pci_bus_resources_arch_default(struct pci_bus *b); 203void set_pci_bus_resources_arch_default(struct pci_bus *b);
257 204
258#ifdef CONFIG_SMP 205#ifdef CONFIG_SMP
259#define mc_capable() (cpus_weight(per_cpu(cpu_core_map, 0)) != nr_cpu_ids) 206#define mc_capable() (cpumask_weight(cpu_core_mask(0)) != nr_cpu_ids)
260#define smt_capable() (smp_num_siblings > 1) 207#define smt_capable() (smp_num_siblings > 1)
261#endif 208#endif
262 209
diff --git a/arch/x86/include/asm/xen/page.h b/arch/x86/include/asm/xen/page.h
index 1a918dde46b5..018a0a400799 100644
--- a/arch/x86/include/asm/xen/page.h
+++ b/arch/x86/include/asm/xen/page.h
@@ -124,7 +124,8 @@ static inline unsigned long mfn_to_local_pfn(unsigned long mfn)
124 124
125/* VIRT <-> MACHINE conversion */ 125/* VIRT <-> MACHINE conversion */
126#define virt_to_machine(v) (phys_to_machine(XPADDR(__pa(v)))) 126#define virt_to_machine(v) (phys_to_machine(XPADDR(__pa(v))))
127#define virt_to_mfn(v) (pfn_to_mfn(PFN_DOWN(__pa(v)))) 127#define virt_to_pfn(v) (PFN_DOWN(__pa(v)))
128#define virt_to_mfn(v) (pfn_to_mfn(virt_to_pfn(v)))
128#define mfn_to_virt(m) (__va(mfn_to_pfn(m) << PAGE_SHIFT)) 129#define mfn_to_virt(m) (__va(mfn_to_pfn(m) << PAGE_SHIFT))
129 130
130static inline unsigned long pte_mfn(pte_t pte) 131static inline unsigned long pte_mfn(pte_t pte)
diff --git a/arch/x86/include/asm/xsave.h b/arch/x86/include/asm/xsave.h
index 08e9a1ac07a9..727acc152344 100644
--- a/arch/x86/include/asm/xsave.h
+++ b/arch/x86/include/asm/xsave.h
@@ -7,6 +7,7 @@
7 7
8#define XSTATE_FP 0x1 8#define XSTATE_FP 0x1
9#define XSTATE_SSE 0x2 9#define XSTATE_SSE 0x2
10#define XSTATE_YMM 0x4
10 11
11#define XSTATE_FPSSE (XSTATE_FP | XSTATE_SSE) 12#define XSTATE_FPSSE (XSTATE_FP | XSTATE_SSE)
12 13
@@ -15,7 +16,7 @@
15/* 16/*
16 * These are the features that the OS can handle currently. 17 * These are the features that the OS can handle currently.
17 */ 18 */
18#define XCNTXT_MASK (XSTATE_FP | XSTATE_SSE) 19#define XCNTXT_MASK (XSTATE_FP | XSTATE_SSE | XSTATE_YMM)
19 20
20#ifdef CONFIG_X86_64 21#ifdef CONFIG_X86_64
21#define REX_PREFIX "0x48, " 22#define REX_PREFIX "0x48, "
diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile
index c611ad64137f..145cce75cda7 100644
--- a/arch/x86/kernel/Makefile
+++ b/arch/x86/kernel/Makefile
@@ -66,7 +66,8 @@ obj-$(CONFIG_X86_MPPARSE) += mpparse.o
66obj-y += apic/ 66obj-y += apic/
67obj-$(CONFIG_X86_REBOOTFIXUPS) += reboot_fixups_32.o 67obj-$(CONFIG_X86_REBOOTFIXUPS) += reboot_fixups_32.o
68obj-$(CONFIG_DYNAMIC_FTRACE) += ftrace.o 68obj-$(CONFIG_DYNAMIC_FTRACE) += ftrace.o
69obj-$(CONFIG_FUNCTION_GRAPH_TRACER) += ftrace.o 69obj-$(CONFIG_FUNCTION_GRAPH_TRACER) += ftrace.o
70obj-$(CONFIG_FTRACE_SYSCALLS) += ftrace.o
70obj-$(CONFIG_KEXEC) += machine_kexec_$(BITS).o 71obj-$(CONFIG_KEXEC) += machine_kexec_$(BITS).o
71obj-$(CONFIG_KEXEC) += relocate_kernel_$(BITS).o crash.o 72obj-$(CONFIG_KEXEC) += relocate_kernel_$(BITS).o crash.o
72obj-$(CONFIG_CRASH_DUMP) += crash_dump_$(BITS).o 73obj-$(CONFIG_CRASH_DUMP) += crash_dump_$(BITS).o
diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c
index a18eb7ce2236..723989d7f802 100644
--- a/arch/x86/kernel/acpi/boot.c
+++ b/arch/x86/kernel/acpi/boot.c
@@ -230,6 +230,35 @@ static void __cpuinit acpi_register_lapic(int id, u8 enabled)
230} 230}
231 231
232static int __init 232static int __init
233acpi_parse_x2apic(struct acpi_subtable_header *header, const unsigned long end)
234{
235 struct acpi_madt_local_x2apic *processor = NULL;
236
237 processor = (struct acpi_madt_local_x2apic *)header;
238
239 if (BAD_MADT_ENTRY(processor, end))
240 return -EINVAL;
241
242 acpi_table_print_madt_entry(header);
243
244#ifdef CONFIG_X86_X2APIC
245 /*
246 * We need to register disabled CPU as well to permit
247 * counting disabled CPUs. This allows us to size
248 * cpus_possible_map more accurately, to permit
249 * to not preallocating memory for all NR_CPUS
250 * when we use CPU hotplug.
251 */
252 acpi_register_lapic(processor->local_apic_id, /* APIC ID */
253 processor->lapic_flags & ACPI_MADT_ENABLED);
254#else
255 printk(KERN_WARNING PREFIX "x2apic entry ignored\n");
256#endif
257
258 return 0;
259}
260
261static int __init
233acpi_parse_lapic(struct acpi_subtable_header * header, const unsigned long end) 262acpi_parse_lapic(struct acpi_subtable_header * header, const unsigned long end)
234{ 263{
235 struct acpi_madt_local_apic *processor = NULL; 264 struct acpi_madt_local_apic *processor = NULL;
@@ -289,6 +318,25 @@ acpi_parse_lapic_addr_ovr(struct acpi_subtable_header * header,
289} 318}
290 319
291static int __init 320static int __init
321acpi_parse_x2apic_nmi(struct acpi_subtable_header *header,
322 const unsigned long end)
323{
324 struct acpi_madt_local_x2apic_nmi *x2apic_nmi = NULL;
325
326 x2apic_nmi = (struct acpi_madt_local_x2apic_nmi *)header;
327
328 if (BAD_MADT_ENTRY(x2apic_nmi, end))
329 return -EINVAL;
330
331 acpi_table_print_madt_entry(header);
332
333 if (x2apic_nmi->lint != 1)
334 printk(KERN_WARNING PREFIX "NMI not connected to LINT 1!\n");
335
336 return 0;
337}
338
339static int __init
292acpi_parse_lapic_nmi(struct acpi_subtable_header * header, const unsigned long end) 340acpi_parse_lapic_nmi(struct acpi_subtable_header * header, const unsigned long end)
293{ 341{
294 struct acpi_madt_local_apic_nmi *lapic_nmi = NULL; 342 struct acpi_madt_local_apic_nmi *lapic_nmi = NULL;
@@ -793,6 +841,7 @@ static int __init early_acpi_parse_madt_lapic_addr_ovr(void)
793static int __init acpi_parse_madt_lapic_entries(void) 841static int __init acpi_parse_madt_lapic_entries(void)
794{ 842{
795 int count; 843 int count;
844 int x2count = 0;
796 845
797 if (!cpu_has_apic) 846 if (!cpu_has_apic)
798 return -ENODEV; 847 return -ENODEV;
@@ -816,22 +865,28 @@ static int __init acpi_parse_madt_lapic_entries(void)
816 count = acpi_table_parse_madt(ACPI_MADT_TYPE_LOCAL_SAPIC, 865 count = acpi_table_parse_madt(ACPI_MADT_TYPE_LOCAL_SAPIC,
817 acpi_parse_sapic, MAX_APICS); 866 acpi_parse_sapic, MAX_APICS);
818 867
819 if (!count) 868 if (!count) {
869 x2count = acpi_table_parse_madt(ACPI_MADT_TYPE_LOCAL_X2APIC,
870 acpi_parse_x2apic, MAX_APICS);
820 count = acpi_table_parse_madt(ACPI_MADT_TYPE_LOCAL_APIC, 871 count = acpi_table_parse_madt(ACPI_MADT_TYPE_LOCAL_APIC,
821 acpi_parse_lapic, MAX_APICS); 872 acpi_parse_lapic, MAX_APICS);
822 if (!count) { 873 }
874 if (!count && !x2count) {
823 printk(KERN_ERR PREFIX "No LAPIC entries present\n"); 875 printk(KERN_ERR PREFIX "No LAPIC entries present\n");
824 /* TBD: Cleanup to allow fallback to MPS */ 876 /* TBD: Cleanup to allow fallback to MPS */
825 return -ENODEV; 877 return -ENODEV;
826 } else if (count < 0) { 878 } else if (count < 0 || x2count < 0) {
827 printk(KERN_ERR PREFIX "Error parsing LAPIC entry\n"); 879 printk(KERN_ERR PREFIX "Error parsing LAPIC entry\n");
828 /* TBD: Cleanup to allow fallback to MPS */ 880 /* TBD: Cleanup to allow fallback to MPS */
829 return count; 881 return count;
830 } 882 }
831 883
884 x2count =
885 acpi_table_parse_madt(ACPI_MADT_TYPE_LOCAL_X2APIC_NMI,
886 acpi_parse_x2apic_nmi, 0);
832 count = 887 count =
833 acpi_table_parse_madt(ACPI_MADT_TYPE_LOCAL_APIC_NMI, acpi_parse_lapic_nmi, 0); 888 acpi_table_parse_madt(ACPI_MADT_TYPE_LOCAL_APIC_NMI, acpi_parse_lapic_nmi, 0);
834 if (count < 0) { 889 if (count < 0 || x2count < 0) {
835 printk(KERN_ERR PREFIX "Error parsing LAPIC NMI entry\n"); 890 printk(KERN_ERR PREFIX "Error parsing LAPIC NMI entry\n");
836 /* TBD: Cleanup to allow fallback to MPS */ 891 /* TBD: Cleanup to allow fallback to MPS */
837 return count; 892 return count;
@@ -1470,7 +1525,7 @@ static int __init dmi_ignore_irq0_timer_override(const struct dmi_system_id *d)
1470 1525
1471/* 1526/*
1472 * If your system is blacklisted here, but you find that acpi=force 1527 * If your system is blacklisted here, but you find that acpi=force
1473 * works for you, please contact acpi-devel@sourceforge.net 1528 * works for you, please contact linux-acpi@vger.kernel.org
1474 */ 1529 */
1475static struct dmi_system_id __initdata acpi_dmi_table[] = { 1530static struct dmi_system_id __initdata acpi_dmi_table[] = {
1476 /* 1531 /*
diff --git a/arch/x86/kernel/alternative.c b/arch/x86/kernel/alternative.c
index 4c80f1557433..f57658702571 100644
--- a/arch/x86/kernel/alternative.c
+++ b/arch/x86/kernel/alternative.c
@@ -5,6 +5,7 @@
5#include <linux/kprobes.h> 5#include <linux/kprobes.h>
6#include <linux/mm.h> 6#include <linux/mm.h>
7#include <linux/vmalloc.h> 7#include <linux/vmalloc.h>
8#include <linux/memory.h>
8#include <asm/alternative.h> 9#include <asm/alternative.h>
9#include <asm/sections.h> 10#include <asm/sections.h>
10#include <asm/pgtable.h> 11#include <asm/pgtable.h>
@@ -12,7 +13,9 @@
12#include <asm/nmi.h> 13#include <asm/nmi.h>
13#include <asm/vsyscall.h> 14#include <asm/vsyscall.h>
14#include <asm/cacheflush.h> 15#include <asm/cacheflush.h>
16#include <asm/tlbflush.h>
15#include <asm/io.h> 17#include <asm/io.h>
18#include <asm/fixmap.h>
16 19
17#define MAX_PATCH_LEN (255-1) 20#define MAX_PATCH_LEN (255-1)
18 21
@@ -226,6 +229,7 @@ static void alternatives_smp_lock(u8 **start, u8 **end, u8 *text, u8 *text_end)
226{ 229{
227 u8 **ptr; 230 u8 **ptr;
228 231
232 mutex_lock(&text_mutex);
229 for (ptr = start; ptr < end; ptr++) { 233 for (ptr = start; ptr < end; ptr++) {
230 if (*ptr < text) 234 if (*ptr < text)
231 continue; 235 continue;
@@ -234,6 +238,7 @@ static void alternatives_smp_lock(u8 **start, u8 **end, u8 *text, u8 *text_end)
234 /* turn DS segment override prefix into lock prefix */ 238 /* turn DS segment override prefix into lock prefix */
235 text_poke(*ptr, ((unsigned char []){0xf0}), 1); 239 text_poke(*ptr, ((unsigned char []){0xf0}), 1);
236 }; 240 };
241 mutex_unlock(&text_mutex);
237} 242}
238 243
239static void alternatives_smp_unlock(u8 **start, u8 **end, u8 *text, u8 *text_end) 244static void alternatives_smp_unlock(u8 **start, u8 **end, u8 *text, u8 *text_end)
@@ -243,6 +248,7 @@ static void alternatives_smp_unlock(u8 **start, u8 **end, u8 *text, u8 *text_end
243 if (noreplace_smp) 248 if (noreplace_smp)
244 return; 249 return;
245 250
251 mutex_lock(&text_mutex);
246 for (ptr = start; ptr < end; ptr++) { 252 for (ptr = start; ptr < end; ptr++) {
247 if (*ptr < text) 253 if (*ptr < text)
248 continue; 254 continue;
@@ -251,6 +257,7 @@ static void alternatives_smp_unlock(u8 **start, u8 **end, u8 *text, u8 *text_end
251 /* turn lock prefix into DS segment override prefix */ 257 /* turn lock prefix into DS segment override prefix */
252 text_poke(*ptr, ((unsigned char []){0x3E}), 1); 258 text_poke(*ptr, ((unsigned char []){0x3E}), 1);
253 }; 259 };
260 mutex_unlock(&text_mutex);
254} 261}
255 262
256struct smp_alt_module { 263struct smp_alt_module {
@@ -500,15 +507,16 @@ void *text_poke_early(void *addr, const void *opcode, size_t len)
500 * It means the size must be writable atomically and the address must be aligned 507 * It means the size must be writable atomically and the address must be aligned
501 * in a way that permits an atomic write. It also makes sure we fit on a single 508 * in a way that permits an atomic write. It also makes sure we fit on a single
502 * page. 509 * page.
510 *
511 * Note: Must be called under text_mutex.
503 */ 512 */
504void *__kprobes text_poke(void *addr, const void *opcode, size_t len) 513void *__kprobes text_poke(void *addr, const void *opcode, size_t len)
505{ 514{
515 unsigned long flags;
506 char *vaddr; 516 char *vaddr;
507 int nr_pages = 2;
508 struct page *pages[2]; 517 struct page *pages[2];
509 int i; 518 int i;
510 519
511 might_sleep();
512 if (!core_kernel_text((unsigned long)addr)) { 520 if (!core_kernel_text((unsigned long)addr)) {
513 pages[0] = vmalloc_to_page(addr); 521 pages[0] = vmalloc_to_page(addr);
514 pages[1] = vmalloc_to_page(addr + PAGE_SIZE); 522 pages[1] = vmalloc_to_page(addr + PAGE_SIZE);
@@ -518,18 +526,21 @@ void *__kprobes text_poke(void *addr, const void *opcode, size_t len)
518 pages[1] = virt_to_page(addr + PAGE_SIZE); 526 pages[1] = virt_to_page(addr + PAGE_SIZE);
519 } 527 }
520 BUG_ON(!pages[0]); 528 BUG_ON(!pages[0]);
521 if (!pages[1]) 529 local_irq_save(flags);
522 nr_pages = 1; 530 set_fixmap(FIX_TEXT_POKE0, page_to_phys(pages[0]));
523 vaddr = vmap(pages, nr_pages, VM_MAP, PAGE_KERNEL); 531 if (pages[1])
524 BUG_ON(!vaddr); 532 set_fixmap(FIX_TEXT_POKE1, page_to_phys(pages[1]));
525 local_irq_disable(); 533 vaddr = (char *)fix_to_virt(FIX_TEXT_POKE0);
526 memcpy(&vaddr[(unsigned long)addr & ~PAGE_MASK], opcode, len); 534 memcpy(&vaddr[(unsigned long)addr & ~PAGE_MASK], opcode, len);
527 local_irq_enable(); 535 clear_fixmap(FIX_TEXT_POKE0);
528 vunmap(vaddr); 536 if (pages[1])
537 clear_fixmap(FIX_TEXT_POKE1);
538 local_flush_tlb();
529 sync_core(); 539 sync_core();
530 /* Could also do a CLFLUSH here to speed up CPU recovery; but 540 /* Could also do a CLFLUSH here to speed up CPU recovery; but
531 that causes hangs on some VIA CPUs. */ 541 that causes hangs on some VIA CPUs. */
532 for (i = 0; i < len; i++) 542 for (i = 0; i < len; i++)
533 BUG_ON(((char *)addr)[i] != ((char *)opcode)[i]); 543 BUG_ON(((char *)addr)[i] != ((char *)opcode)[i]);
544 local_irq_restore(flags);
534 return addr; 545 return addr;
535} 546}
diff --git a/arch/x86/kernel/amd_iommu.c b/arch/x86/kernel/amd_iommu.c
index c5962fe3796f..a97db99dad52 100644
--- a/arch/x86/kernel/amd_iommu.c
+++ b/arch/x86/kernel/amd_iommu.c
@@ -1928,6 +1928,12 @@ static phys_addr_t amd_iommu_iova_to_phys(struct iommu_domain *dom,
1928 return paddr; 1928 return paddr;
1929} 1929}
1930 1930
1931static int amd_iommu_domain_has_cap(struct iommu_domain *domain,
1932 unsigned long cap)
1933{
1934 return 0;
1935}
1936
1931static struct iommu_ops amd_iommu_ops = { 1937static struct iommu_ops amd_iommu_ops = {
1932 .domain_init = amd_iommu_domain_init, 1938 .domain_init = amd_iommu_domain_init,
1933 .domain_destroy = amd_iommu_domain_destroy, 1939 .domain_destroy = amd_iommu_domain_destroy,
@@ -1936,5 +1942,6 @@ static struct iommu_ops amd_iommu_ops = {
1936 .map = amd_iommu_map_range, 1942 .map = amd_iommu_map_range,
1937 .unmap = amd_iommu_unmap_range, 1943 .unmap = amd_iommu_unmap_range,
1938 .iova_to_phys = amd_iommu_iova_to_phys, 1944 .iova_to_phys = amd_iommu_iova_to_phys,
1945 .domain_has_cap = amd_iommu_domain_has_cap,
1939}; 1946};
1940 1947
diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c
index 85eb8e100818..f2870920f246 100644
--- a/arch/x86/kernel/apic/apic.c
+++ b/arch/x86/kernel/apic/apic.c
@@ -431,6 +431,12 @@ static void __cpuinit setup_APIC_timer(void)
431{ 431{
432 struct clock_event_device *levt = &__get_cpu_var(lapic_events); 432 struct clock_event_device *levt = &__get_cpu_var(lapic_events);
433 433
434 if (cpu_has(&current_cpu_data, X86_FEATURE_ARAT)) {
435 lapic_clockevent.features &= ~CLOCK_EVT_FEAT_C3STOP;
436 /* Make LAPIC timer preferrable over percpu HPET */
437 lapic_clockevent.rating = 150;
438 }
439
434 memcpy(levt, &lapic_clockevent, sizeof(*levt)); 440 memcpy(levt, &lapic_clockevent, sizeof(*levt));
435 levt->cpumask = cpumask_of(smp_processor_id()); 441 levt->cpumask = cpumask_of(smp_processor_id());
436 442
@@ -1304,6 +1310,7 @@ void __init enable_IR_x2apic(void)
1304#ifdef CONFIG_INTR_REMAP 1310#ifdef CONFIG_INTR_REMAP
1305 int ret; 1311 int ret;
1306 unsigned long flags; 1312 unsigned long flags;
1313 struct IO_APIC_route_entry **ioapic_entries = NULL;
1307 1314
1308 if (!cpu_has_x2apic) 1315 if (!cpu_has_x2apic)
1309 return; 1316 return;
@@ -1334,17 +1341,23 @@ void __init enable_IR_x2apic(void)
1334 return; 1341 return;
1335 } 1342 }
1336 1343
1337 ret = save_IO_APIC_setup(); 1344 ioapic_entries = alloc_ioapic_entries();
1345 if (!ioapic_entries) {
1346 pr_info("Allocate ioapic_entries failed: %d\n", ret);
1347 goto end;
1348 }
1349
1350 ret = save_IO_APIC_setup(ioapic_entries);
1338 if (ret) { 1351 if (ret) {
1339 pr_info("Saving IO-APIC state failed: %d\n", ret); 1352 pr_info("Saving IO-APIC state failed: %d\n", ret);
1340 goto end; 1353 goto end;
1341 } 1354 }
1342 1355
1343 local_irq_save(flags); 1356 local_irq_save(flags);
1344 mask_IO_APIC_setup(); 1357 mask_IO_APIC_setup(ioapic_entries);
1345 mask_8259A(); 1358 mask_8259A();
1346 1359
1347 ret = enable_intr_remapping(1); 1360 ret = enable_intr_remapping(EIM_32BIT_APIC_ID);
1348 1361
1349 if (ret && x2apic_preenabled) { 1362 if (ret && x2apic_preenabled) {
1350 local_irq_restore(flags); 1363 local_irq_restore(flags);
@@ -1364,9 +1377,9 @@ end_restore:
1364 /* 1377 /*
1365 * IR enabling failed 1378 * IR enabling failed
1366 */ 1379 */
1367 restore_IO_APIC_setup(); 1380 restore_IO_APIC_setup(ioapic_entries);
1368 else 1381 else
1369 reinit_intr_remapped_IO_APIC(x2apic_preenabled); 1382 reinit_intr_remapped_IO_APIC(x2apic_preenabled, ioapic_entries);
1370 1383
1371 unmask_8259A(); 1384 unmask_8259A();
1372 local_irq_restore(flags); 1385 local_irq_restore(flags);
@@ -1379,6 +1392,8 @@ end:
1379 pr_info("Enabled Interrupt-remapping\n"); 1392 pr_info("Enabled Interrupt-remapping\n");
1380 } else 1393 } else
1381 pr_err("Failed to enable Interrupt-remapping and x2apic\n"); 1394 pr_err("Failed to enable Interrupt-remapping and x2apic\n");
1395 if (ioapic_entries)
1396 free_ioapic_entries(ioapic_entries);
1382#else 1397#else
1383 if (!cpu_has_x2apic) 1398 if (!cpu_has_x2apic)
1384 return; 1399 return;
@@ -1954,6 +1969,10 @@ static int lapic_suspend(struct sys_device *dev, pm_message_t state)
1954 1969
1955 local_irq_save(flags); 1970 local_irq_save(flags);
1956 disable_local_APIC(); 1971 disable_local_APIC();
1972#ifdef CONFIG_INTR_REMAP
1973 if (intr_remapping_enabled)
1974 disable_intr_remapping();
1975#endif
1957 local_irq_restore(flags); 1976 local_irq_restore(flags);
1958 return 0; 1977 return 0;
1959} 1978}
@@ -1964,15 +1983,41 @@ static int lapic_resume(struct sys_device *dev)
1964 unsigned long flags; 1983 unsigned long flags;
1965 int maxlvt; 1984 int maxlvt;
1966 1985
1986#ifdef CONFIG_INTR_REMAP
1987 int ret;
1988 struct IO_APIC_route_entry **ioapic_entries = NULL;
1989
1967 if (!apic_pm_state.active) 1990 if (!apic_pm_state.active)
1968 return 0; 1991 return 0;
1969 1992
1970 maxlvt = lapic_get_maxlvt();
1971
1972 local_irq_save(flags); 1993 local_irq_save(flags);
1994 if (x2apic) {
1995 ioapic_entries = alloc_ioapic_entries();
1996 if (!ioapic_entries) {
1997 WARN(1, "Alloc ioapic_entries in lapic resume failed.");
1998 return -ENOMEM;
1999 }
2000
2001 ret = save_IO_APIC_setup(ioapic_entries);
2002 if (ret) {
2003 WARN(1, "Saving IO-APIC state failed: %d\n", ret);
2004 free_ioapic_entries(ioapic_entries);
2005 return ret;
2006 }
2007
2008 mask_IO_APIC_setup(ioapic_entries);
2009 mask_8259A();
2010 enable_x2apic();
2011 }
2012#else
2013 if (!apic_pm_state.active)
2014 return 0;
1973 2015
2016 local_irq_save(flags);
1974 if (x2apic) 2017 if (x2apic)
1975 enable_x2apic(); 2018 enable_x2apic();
2019#endif
2020
1976 else { 2021 else {
1977 /* 2022 /*
1978 * Make sure the APICBASE points to the right address 2023 * Make sure the APICBASE points to the right address
@@ -1986,6 +2031,7 @@ static int lapic_resume(struct sys_device *dev)
1986 wrmsr(MSR_IA32_APICBASE, l, h); 2031 wrmsr(MSR_IA32_APICBASE, l, h);
1987 } 2032 }
1988 2033
2034 maxlvt = lapic_get_maxlvt();
1989 apic_write(APIC_LVTERR, ERROR_APIC_VECTOR | APIC_LVT_MASKED); 2035 apic_write(APIC_LVTERR, ERROR_APIC_VECTOR | APIC_LVT_MASKED);
1990 apic_write(APIC_ID, apic_pm_state.apic_id); 2036 apic_write(APIC_ID, apic_pm_state.apic_id);
1991 apic_write(APIC_DFR, apic_pm_state.apic_dfr); 2037 apic_write(APIC_DFR, apic_pm_state.apic_dfr);
@@ -2009,8 +2055,20 @@ static int lapic_resume(struct sys_device *dev)
2009 apic_write(APIC_ESR, 0); 2055 apic_write(APIC_ESR, 0);
2010 apic_read(APIC_ESR); 2056 apic_read(APIC_ESR);
2011 2057
2058#ifdef CONFIG_INTR_REMAP
2059 if (intr_remapping_enabled)
2060 reenable_intr_remapping(EIM_32BIT_APIC_ID);
2061
2062 if (x2apic) {
2063 unmask_8259A();
2064 restore_IO_APIC_setup(ioapic_entries);
2065 free_ioapic_entries(ioapic_entries);
2066 }
2067#endif
2068
2012 local_irq_restore(flags); 2069 local_irq_restore(flags);
2013 2070
2071
2014 return 0; 2072 return 0;
2015} 2073}
2016 2074
@@ -2048,7 +2106,9 @@ static int __init init_lapic_sysfs(void)
2048 error = sysdev_register(&device_lapic); 2106 error = sysdev_register(&device_lapic);
2049 return error; 2107 return error;
2050} 2108}
2051device_initcall(init_lapic_sysfs); 2109
2110/* local apic needs to resume before other devices access its registers. */
2111core_initcall(init_lapic_sysfs);
2052 2112
2053#else /* CONFIG_PM */ 2113#else /* CONFIG_PM */
2054 2114
diff --git a/arch/x86/kernel/apic/apic_flat_64.c b/arch/x86/kernel/apic/apic_flat_64.c
index 0014714ea97b..306e5e88fb6f 100644
--- a/arch/x86/kernel/apic/apic_flat_64.c
+++ b/arch/x86/kernel/apic/apic_flat_64.c
@@ -212,7 +212,7 @@ struct apic apic_flat = {
212 .trampoline_phys_high = DEFAULT_TRAMPOLINE_PHYS_HIGH, 212 .trampoline_phys_high = DEFAULT_TRAMPOLINE_PHYS_HIGH,
213 .wait_for_init_deassert = NULL, 213 .wait_for_init_deassert = NULL,
214 .smp_callin_clear_local_apic = NULL, 214 .smp_callin_clear_local_apic = NULL,
215 .inquire_remote_apic = NULL, 215 .inquire_remote_apic = default_inquire_remote_apic,
216 216
217 .read = native_apic_mem_read, 217 .read = native_apic_mem_read,
218 .write = native_apic_mem_write, 218 .write = native_apic_mem_write,
@@ -362,7 +362,7 @@ struct apic apic_physflat = {
362 .trampoline_phys_high = DEFAULT_TRAMPOLINE_PHYS_HIGH, 362 .trampoline_phys_high = DEFAULT_TRAMPOLINE_PHYS_HIGH,
363 .wait_for_init_deassert = NULL, 363 .wait_for_init_deassert = NULL,
364 .smp_callin_clear_local_apic = NULL, 364 .smp_callin_clear_local_apic = NULL,
365 .inquire_remote_apic = NULL, 365 .inquire_remote_apic = default_inquire_remote_apic,
366 366
367 .read = native_apic_mem_read, 367 .read = native_apic_mem_read,
368 .write = native_apic_mem_write, 368 .write = native_apic_mem_write,
diff --git a/arch/x86/kernel/apic/bigsmp_32.c b/arch/x86/kernel/apic/bigsmp_32.c
index d806ecaa948f..676cdac385c0 100644
--- a/arch/x86/kernel/apic/bigsmp_32.c
+++ b/arch/x86/kernel/apic/bigsmp_32.c
@@ -26,12 +26,12 @@ static int bigsmp_apic_id_registered(void)
26 return 1; 26 return 1;
27} 27}
28 28
29static const cpumask_t *bigsmp_target_cpus(void) 29static const struct cpumask *bigsmp_target_cpus(void)
30{ 30{
31#ifdef CONFIG_SMP 31#ifdef CONFIG_SMP
32 return &cpu_online_map; 32 return cpu_online_mask;
33#else 33#else
34 return &cpumask_of_cpu(0); 34 return cpumask_of(0);
35#endif 35#endif
36} 36}
37 37
@@ -118,9 +118,9 @@ static int bigsmp_check_phys_apicid_present(int boot_cpu_physical_apicid)
118} 118}
119 119
120/* As we are using single CPU as destination, pick only one CPU here */ 120/* As we are using single CPU as destination, pick only one CPU here */
121static unsigned int bigsmp_cpu_mask_to_apicid(const cpumask_t *cpumask) 121static unsigned int bigsmp_cpu_mask_to_apicid(const struct cpumask *cpumask)
122{ 122{
123 return bigsmp_cpu_to_logical_apicid(first_cpu(*cpumask)); 123 return bigsmp_cpu_to_logical_apicid(cpumask_first(cpumask));
124} 124}
125 125
126static unsigned int bigsmp_cpu_mask_to_apicid_and(const struct cpumask *cpumask, 126static unsigned int bigsmp_cpu_mask_to_apicid_and(const struct cpumask *cpumask,
@@ -188,10 +188,10 @@ static const struct dmi_system_id bigsmp_dmi_table[] = {
188 { } /* NULL entry stops DMI scanning */ 188 { } /* NULL entry stops DMI scanning */
189}; 189};
190 190
191static void bigsmp_vector_allocation_domain(int cpu, cpumask_t *retmask) 191static void bigsmp_vector_allocation_domain(int cpu, struct cpumask *retmask)
192{ 192{
193 cpus_clear(*retmask); 193 cpumask_clear(retmask);
194 cpu_set(cpu, *retmask); 194 cpumask_set_cpu(cpu, retmask);
195} 195}
196 196
197static int probe_bigsmp(void) 197static int probe_bigsmp(void)
diff --git a/arch/x86/kernel/apic/es7000_32.c b/arch/x86/kernel/apic/es7000_32.c
index 19588f2770ee..1c11b819f245 100644
--- a/arch/x86/kernel/apic/es7000_32.c
+++ b/arch/x86/kernel/apic/es7000_32.c
@@ -410,7 +410,7 @@ static void es7000_enable_apic_mode(void)
410 WARN(1, "Command failed, status = %x\n", mip_status); 410 WARN(1, "Command failed, status = %x\n", mip_status);
411} 411}
412 412
413static void es7000_vector_allocation_domain(int cpu, cpumask_t *retmask) 413static void es7000_vector_allocation_domain(int cpu, struct cpumask *retmask)
414{ 414{
415 /* Careful. Some cpus do not strictly honor the set of cpus 415 /* Careful. Some cpus do not strictly honor the set of cpus
416 * specified in the interrupt destination when using lowest 416 * specified in the interrupt destination when using lowest
@@ -420,7 +420,8 @@ static void es7000_vector_allocation_domain(int cpu, cpumask_t *retmask)
420 * deliver interrupts to the wrong hyperthread when only one 420 * deliver interrupts to the wrong hyperthread when only one
421 * hyperthread was specified in the interrupt desitination. 421 * hyperthread was specified in the interrupt desitination.
422 */ 422 */
423 *retmask = (cpumask_t){ { [0] = APIC_ALL_CPUS, } }; 423 cpumask_clear(retmask);
424 cpumask_bits(retmask)[0] = APIC_ALL_CPUS;
424} 425}
425 426
426 427
@@ -455,14 +456,14 @@ static int es7000_apic_id_registered(void)
455 return 1; 456 return 1;
456} 457}
457 458
458static const cpumask_t *target_cpus_cluster(void) 459static const struct cpumask *target_cpus_cluster(void)
459{ 460{
460 return &CPU_MASK_ALL; 461 return cpu_all_mask;
461} 462}
462 463
463static const cpumask_t *es7000_target_cpus(void) 464static const struct cpumask *es7000_target_cpus(void)
464{ 465{
465 return &cpumask_of_cpu(smp_processor_id()); 466 return cpumask_of(smp_processor_id());
466} 467}
467 468
468static unsigned long 469static unsigned long
@@ -517,7 +518,7 @@ static void es7000_setup_apic_routing(void)
517 "Enabling APIC mode: %s. Using %d I/O APICs, target cpus %lx\n", 518 "Enabling APIC mode: %s. Using %d I/O APICs, target cpus %lx\n",
518 (apic_version[apic] == 0x14) ? 519 (apic_version[apic] == 0x14) ?
519 "Physical Cluster" : "Logical Cluster", 520 "Physical Cluster" : "Logical Cluster",
520 nr_ioapics, cpus_addr(*es7000_target_cpus())[0]); 521 nr_ioapics, cpumask_bits(es7000_target_cpus())[0]);
521} 522}
522 523
523static int es7000_apicid_to_node(int logical_apicid) 524static int es7000_apicid_to_node(int logical_apicid)
@@ -572,7 +573,7 @@ static int es7000_check_phys_apicid_present(int cpu_physical_apicid)
572 return 1; 573 return 1;
573} 574}
574 575
575static unsigned int es7000_cpu_mask_to_apicid(const cpumask_t *cpumask) 576static unsigned int es7000_cpu_mask_to_apicid(const struct cpumask *cpumask)
576{ 577{
577 unsigned int round = 0; 578 unsigned int round = 0;
578 int cpu, uninitialized_var(apicid); 579 int cpu, uninitialized_var(apicid);
diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c
index 1bb5c6cee3eb..a2789e42e162 100644
--- a/arch/x86/kernel/apic/io_apic.c
+++ b/arch/x86/kernel/apic/io_apic.c
@@ -851,63 +851,74 @@ __setup("pirq=", ioapic_pirq_setup);
851#endif /* CONFIG_X86_32 */ 851#endif /* CONFIG_X86_32 */
852 852
853#ifdef CONFIG_INTR_REMAP 853#ifdef CONFIG_INTR_REMAP
854/* I/O APIC RTE contents at the OS boot up */ 854struct IO_APIC_route_entry **alloc_ioapic_entries(void)
855static struct IO_APIC_route_entry *early_ioapic_entries[MAX_IO_APICS]; 855{
856 int apic;
857 struct IO_APIC_route_entry **ioapic_entries;
858
859 ioapic_entries = kzalloc(sizeof(*ioapic_entries) * nr_ioapics,
860 GFP_ATOMIC);
861 if (!ioapic_entries)
862 return 0;
863
864 for (apic = 0; apic < nr_ioapics; apic++) {
865 ioapic_entries[apic] =
866 kzalloc(sizeof(struct IO_APIC_route_entry) *
867 nr_ioapic_registers[apic], GFP_ATOMIC);
868 if (!ioapic_entries[apic])
869 goto nomem;
870 }
871
872 return ioapic_entries;
873
874nomem:
875 while (--apic >= 0)
876 kfree(ioapic_entries[apic]);
877 kfree(ioapic_entries);
878
879 return 0;
880}
856 881
857/* 882/*
858 * Saves all the IO-APIC RTE's 883 * Saves all the IO-APIC RTE's
859 */ 884 */
860int save_IO_APIC_setup(void) 885int save_IO_APIC_setup(struct IO_APIC_route_entry **ioapic_entries)
861{ 886{
862 union IO_APIC_reg_01 reg_01;
863 unsigned long flags;
864 int apic, pin; 887 int apic, pin;
865 888
866 /* 889 if (!ioapic_entries)
867 * The number of IO-APIC IRQ registers (== #pins): 890 return -ENOMEM;
868 */
869 for (apic = 0; apic < nr_ioapics; apic++) {
870 spin_lock_irqsave(&ioapic_lock, flags);
871 reg_01.raw = io_apic_read(apic, 1);
872 spin_unlock_irqrestore(&ioapic_lock, flags);
873 nr_ioapic_registers[apic] = reg_01.bits.entries+1;
874 }
875 891
876 for (apic = 0; apic < nr_ioapics; apic++) { 892 for (apic = 0; apic < nr_ioapics; apic++) {
877 early_ioapic_entries[apic] = 893 if (!ioapic_entries[apic])
878 kzalloc(sizeof(struct IO_APIC_route_entry) * 894 return -ENOMEM;
879 nr_ioapic_registers[apic], GFP_KERNEL);
880 if (!early_ioapic_entries[apic])
881 goto nomem;
882 }
883 895
884 for (apic = 0; apic < nr_ioapics; apic++)
885 for (pin = 0; pin < nr_ioapic_registers[apic]; pin++) 896 for (pin = 0; pin < nr_ioapic_registers[apic]; pin++)
886 early_ioapic_entries[apic][pin] = 897 ioapic_entries[apic][pin] =
887 ioapic_read_entry(apic, pin); 898 ioapic_read_entry(apic, pin);
899 }
888 900
889 return 0; 901 return 0;
890
891nomem:
892 while (apic >= 0)
893 kfree(early_ioapic_entries[apic--]);
894 memset(early_ioapic_entries, 0,
895 ARRAY_SIZE(early_ioapic_entries));
896
897 return -ENOMEM;
898} 902}
899 903
900void mask_IO_APIC_setup(void) 904/*
905 * Mask all IO APIC entries.
906 */
907void mask_IO_APIC_setup(struct IO_APIC_route_entry **ioapic_entries)
901{ 908{
902 int apic, pin; 909 int apic, pin;
903 910
911 if (!ioapic_entries)
912 return;
913
904 for (apic = 0; apic < nr_ioapics; apic++) { 914 for (apic = 0; apic < nr_ioapics; apic++) {
905 if (!early_ioapic_entries[apic]) 915 if (!ioapic_entries[apic])
906 break; 916 break;
917
907 for (pin = 0; pin < nr_ioapic_registers[apic]; pin++) { 918 for (pin = 0; pin < nr_ioapic_registers[apic]; pin++) {
908 struct IO_APIC_route_entry entry; 919 struct IO_APIC_route_entry entry;
909 920
910 entry = early_ioapic_entries[apic][pin]; 921 entry = ioapic_entries[apic][pin];
911 if (!entry.mask) { 922 if (!entry.mask) {
912 entry.mask = 1; 923 entry.mask = 1;
913 ioapic_write_entry(apic, pin, entry); 924 ioapic_write_entry(apic, pin, entry);
@@ -916,22 +927,30 @@ void mask_IO_APIC_setup(void)
916 } 927 }
917} 928}
918 929
919void restore_IO_APIC_setup(void) 930/*
931 * Restore IO APIC entries which was saved in ioapic_entries.
932 */
933int restore_IO_APIC_setup(struct IO_APIC_route_entry **ioapic_entries)
920{ 934{
921 int apic, pin; 935 int apic, pin;
922 936
937 if (!ioapic_entries)
938 return -ENOMEM;
939
923 for (apic = 0; apic < nr_ioapics; apic++) { 940 for (apic = 0; apic < nr_ioapics; apic++) {
924 if (!early_ioapic_entries[apic]) 941 if (!ioapic_entries[apic])
925 break; 942 return -ENOMEM;
943
926 for (pin = 0; pin < nr_ioapic_registers[apic]; pin++) 944 for (pin = 0; pin < nr_ioapic_registers[apic]; pin++)
927 ioapic_write_entry(apic, pin, 945 ioapic_write_entry(apic, pin,
928 early_ioapic_entries[apic][pin]); 946 ioapic_entries[apic][pin]);
929 kfree(early_ioapic_entries[apic]);
930 early_ioapic_entries[apic] = NULL;
931 } 947 }
948 return 0;
932} 949}
933 950
934void reinit_intr_remapped_IO_APIC(int intr_remapping) 951void reinit_intr_remapped_IO_APIC(int intr_remapping,
952 struct IO_APIC_route_entry **ioapic_entries)
953
935{ 954{
936 /* 955 /*
937 * for now plain restore of previous settings. 956 * for now plain restore of previous settings.
@@ -940,7 +959,17 @@ void reinit_intr_remapped_IO_APIC(int intr_remapping)
940 * table entries. for now, do a plain restore, and wait for 959 * table entries. for now, do a plain restore, and wait for
941 * the setup_IO_APIC_irqs() to do proper initialization. 960 * the setup_IO_APIC_irqs() to do proper initialization.
942 */ 961 */
943 restore_IO_APIC_setup(); 962 restore_IO_APIC_setup(ioapic_entries);
963}
964
965void free_ioapic_entries(struct IO_APIC_route_entry **ioapic_entries)
966{
967 int apic;
968
969 for (apic = 0; apic < nr_ioapics; apic++)
970 kfree(ioapic_entries[apic]);
971
972 kfree(ioapic_entries);
944} 973}
945#endif 974#endif
946 975
@@ -2495,7 +2524,6 @@ static void irq_complete_move(struct irq_desc **descp)
2495static inline void irq_complete_move(struct irq_desc **descp) {} 2524static inline void irq_complete_move(struct irq_desc **descp) {}
2496#endif 2525#endif
2497 2526
2498#ifdef CONFIG_INTR_REMAP
2499static void __eoi_ioapic_irq(unsigned int irq, struct irq_cfg *cfg) 2527static void __eoi_ioapic_irq(unsigned int irq, struct irq_cfg *cfg)
2500{ 2528{
2501 int apic, pin; 2529 int apic, pin;
@@ -2529,6 +2557,7 @@ eoi_ioapic_irq(struct irq_desc *desc)
2529 spin_unlock_irqrestore(&ioapic_lock, flags); 2557 spin_unlock_irqrestore(&ioapic_lock, flags);
2530} 2558}
2531 2559
2560#ifdef CONFIG_X86_X2APIC
2532static void ack_x2apic_level(unsigned int irq) 2561static void ack_x2apic_level(unsigned int irq)
2533{ 2562{
2534 struct irq_desc *desc = irq_to_desc(irq); 2563 struct irq_desc *desc = irq_to_desc(irq);
@@ -2540,7 +2569,6 @@ static void ack_x2apic_edge(unsigned int irq)
2540{ 2569{
2541 ack_x2APIC_irq(); 2570 ack_x2APIC_irq();
2542} 2571}
2543
2544#endif 2572#endif
2545 2573
2546static void ack_apic_edge(unsigned int irq) 2574static void ack_apic_edge(unsigned int irq)
@@ -2606,6 +2634,9 @@ static void ack_apic_level(unsigned int irq)
2606 */ 2634 */
2607 ack_APIC_irq(); 2635 ack_APIC_irq();
2608 2636
2637 if (irq_remapped(irq))
2638 eoi_ioapic_irq(desc);
2639
2609 /* Now we can move and renable the irq */ 2640 /* Now we can move and renable the irq */
2610 if (unlikely(do_unmask_irq)) { 2641 if (unlikely(do_unmask_irq)) {
2611 /* Only migrate the irq if the ack has been received. 2642 /* Only migrate the irq if the ack has been received.
@@ -2651,6 +2682,26 @@ static void ack_apic_level(unsigned int irq)
2651#endif 2682#endif
2652} 2683}
2653 2684
2685#ifdef CONFIG_INTR_REMAP
2686static void ir_ack_apic_edge(unsigned int irq)
2687{
2688#ifdef CONFIG_X86_X2APIC
2689 if (x2apic_enabled())
2690 return ack_x2apic_edge(irq);
2691#endif
2692 return ack_apic_edge(irq);
2693}
2694
2695static void ir_ack_apic_level(unsigned int irq)
2696{
2697#ifdef CONFIG_X86_X2APIC
2698 if (x2apic_enabled())
2699 return ack_x2apic_level(irq);
2700#endif
2701 return ack_apic_level(irq);
2702}
2703#endif /* CONFIG_INTR_REMAP */
2704
2654static struct irq_chip ioapic_chip __read_mostly = { 2705static struct irq_chip ioapic_chip __read_mostly = {
2655 .name = "IO-APIC", 2706 .name = "IO-APIC",
2656 .startup = startup_ioapic_irq, 2707 .startup = startup_ioapic_irq,
@@ -2670,8 +2721,8 @@ static struct irq_chip ir_ioapic_chip __read_mostly = {
2670 .mask = mask_IO_APIC_irq, 2721 .mask = mask_IO_APIC_irq,
2671 .unmask = unmask_IO_APIC_irq, 2722 .unmask = unmask_IO_APIC_irq,
2672#ifdef CONFIG_INTR_REMAP 2723#ifdef CONFIG_INTR_REMAP
2673 .ack = ack_x2apic_edge, 2724 .ack = ir_ack_apic_edge,
2674 .eoi = ack_x2apic_level, 2725 .eoi = ir_ack_apic_level,
2675#ifdef CONFIG_SMP 2726#ifdef CONFIG_SMP
2676 .set_affinity = set_ir_ioapic_affinity_irq, 2727 .set_affinity = set_ir_ioapic_affinity_irq,
2677#endif 2728#endif
@@ -3397,7 +3448,7 @@ static struct irq_chip msi_ir_chip = {
3397 .unmask = unmask_msi_irq, 3448 .unmask = unmask_msi_irq,
3398 .mask = mask_msi_irq, 3449 .mask = mask_msi_irq,
3399#ifdef CONFIG_INTR_REMAP 3450#ifdef CONFIG_INTR_REMAP
3400 .ack = ack_x2apic_edge, 3451 .ack = ir_ack_apic_edge,
3401#ifdef CONFIG_SMP 3452#ifdef CONFIG_SMP
3402 .set_affinity = ir_set_msi_irq_affinity, 3453 .set_affinity = ir_set_msi_irq_affinity,
3403#endif 3454#endif
diff --git a/arch/x86/kernel/apic/nmi.c b/arch/x86/kernel/apic/nmi.c
index bdfad80c3cf1..d6bd62407152 100644
--- a/arch/x86/kernel/apic/nmi.c
+++ b/arch/x86/kernel/apic/nmi.c
@@ -39,7 +39,7 @@
39int unknown_nmi_panic; 39int unknown_nmi_panic;
40int nmi_watchdog_enabled; 40int nmi_watchdog_enabled;
41 41
42static cpumask_t backtrace_mask = CPU_MASK_NONE; 42static cpumask_var_t backtrace_mask;
43 43
44/* nmi_active: 44/* nmi_active:
45 * >0: the lapic NMI watchdog is active, but can be disabled 45 * >0: the lapic NMI watchdog is active, but can be disabled
@@ -138,6 +138,7 @@ int __init check_nmi_watchdog(void)
138 if (!prev_nmi_count) 138 if (!prev_nmi_count)
139 goto error; 139 goto error;
140 140
141 alloc_cpumask_var(&backtrace_mask, GFP_KERNEL);
141 printk(KERN_INFO "Testing NMI watchdog ... "); 142 printk(KERN_INFO "Testing NMI watchdog ... ");
142 143
143#ifdef CONFIG_SMP 144#ifdef CONFIG_SMP
@@ -413,14 +414,14 @@ nmi_watchdog_tick(struct pt_regs *regs, unsigned reason)
413 touched = 1; 414 touched = 1;
414 } 415 }
415 416
416 if (cpu_isset(cpu, backtrace_mask)) { 417 if (cpumask_test_cpu(cpu, backtrace_mask)) {
417 static DEFINE_SPINLOCK(lock); /* Serialise the printks */ 418 static DEFINE_SPINLOCK(lock); /* Serialise the printks */
418 419
419 spin_lock(&lock); 420 spin_lock(&lock);
420 printk(KERN_WARNING "NMI backtrace for cpu %d\n", cpu); 421 printk(KERN_WARNING "NMI backtrace for cpu %d\n", cpu);
421 dump_stack(); 422 dump_stack();
422 spin_unlock(&lock); 423 spin_unlock(&lock);
423 cpu_clear(cpu, backtrace_mask); 424 cpumask_clear_cpu(cpu, backtrace_mask);
424 } 425 }
425 426
426 /* Could check oops_in_progress here too, but it's safer not to */ 427 /* Could check oops_in_progress here too, but it's safer not to */
@@ -554,10 +555,10 @@ void __trigger_all_cpu_backtrace(void)
554{ 555{
555 int i; 556 int i;
556 557
557 backtrace_mask = cpu_online_map; 558 cpumask_copy(backtrace_mask, cpu_online_mask);
558 /* Wait for up to 10 seconds for all CPUs to do the backtrace */ 559 /* Wait for up to 10 seconds for all CPUs to do the backtrace */
559 for (i = 0; i < 10 * 1000; i++) { 560 for (i = 0; i < 10 * 1000; i++) {
560 if (cpus_empty(backtrace_mask)) 561 if (cpumask_empty(backtrace_mask))
561 break; 562 break;
562 mdelay(1); 563 mdelay(1);
563 } 564 }
diff --git a/arch/x86/kernel/apic/numaq_32.c b/arch/x86/kernel/apic/numaq_32.c
index ba2fc6465534..533e59c6fc82 100644
--- a/arch/x86/kernel/apic/numaq_32.c
+++ b/arch/x86/kernel/apic/numaq_32.c
@@ -334,9 +334,9 @@ static inline void numaq_smp_callin_clear_local_apic(void)
334 clear_local_APIC(); 334 clear_local_APIC();
335} 335}
336 336
337static inline const cpumask_t *numaq_target_cpus(void) 337static inline const struct cpumask *numaq_target_cpus(void)
338{ 338{
339 return &CPU_MASK_ALL; 339 return cpu_all_mask;
340} 340}
341 341
342static inline unsigned long 342static inline unsigned long
@@ -427,7 +427,7 @@ static inline int numaq_check_phys_apicid_present(int boot_cpu_physical_apicid)
427 * We use physical apicids here, not logical, so just return the default 427 * We use physical apicids here, not logical, so just return the default
428 * physical broadcast to stop people from breaking us 428 * physical broadcast to stop people from breaking us
429 */ 429 */
430static inline unsigned int numaq_cpu_mask_to_apicid(const cpumask_t *cpumask) 430static unsigned int numaq_cpu_mask_to_apicid(const struct cpumask *cpumask)
431{ 431{
432 return 0x0F; 432 return 0x0F;
433} 433}
@@ -462,7 +462,7 @@ static int probe_numaq(void)
462 return found_numaq; 462 return found_numaq;
463} 463}
464 464
465static void numaq_vector_allocation_domain(int cpu, cpumask_t *retmask) 465static void numaq_vector_allocation_domain(int cpu, struct cpumask *retmask)
466{ 466{
467 /* Careful. Some cpus do not strictly honor the set of cpus 467 /* Careful. Some cpus do not strictly honor the set of cpus
468 * specified in the interrupt destination when using lowest 468 * specified in the interrupt destination when using lowest
@@ -472,7 +472,8 @@ static void numaq_vector_allocation_domain(int cpu, cpumask_t *retmask)
472 * deliver interrupts to the wrong hyperthread when only one 472 * deliver interrupts to the wrong hyperthread when only one
473 * hyperthread was specified in the interrupt desitination. 473 * hyperthread was specified in the interrupt desitination.
474 */ 474 */
475 *retmask = (cpumask_t){ { [0] = APIC_ALL_CPUS, } }; 475 cpumask_clear(retmask);
476 cpumask_bits(retmask)[0] = APIC_ALL_CPUS;
476} 477}
477 478
478static void numaq_setup_portio_remap(void) 479static void numaq_setup_portio_remap(void)
diff --git a/arch/x86/kernel/apic/probe_32.c b/arch/x86/kernel/apic/probe_32.c
index 141c99a1c264..01eda2ac65e4 100644
--- a/arch/x86/kernel/apic/probe_32.c
+++ b/arch/x86/kernel/apic/probe_32.c
@@ -83,7 +83,8 @@ static void default_vector_allocation_domain(int cpu, struct cpumask *retmask)
83 * deliver interrupts to the wrong hyperthread when only one 83 * deliver interrupts to the wrong hyperthread when only one
84 * hyperthread was specified in the interrupt desitination. 84 * hyperthread was specified in the interrupt desitination.
85 */ 85 */
86 *retmask = (cpumask_t) { { [0] = APIC_ALL_CPUS } }; 86 cpumask_clear(retmask);
87 cpumask_bits(retmask)[0] = APIC_ALL_CPUS;
87} 88}
88 89
89/* should be called last. */ 90/* should be called last. */
diff --git a/arch/x86/kernel/apic/summit_32.c b/arch/x86/kernel/apic/summit_32.c
index aac52fa873ff..9cfe1f415d81 100644
--- a/arch/x86/kernel/apic/summit_32.c
+++ b/arch/x86/kernel/apic/summit_32.c
@@ -53,23 +53,19 @@ static unsigned summit_get_apic_id(unsigned long x)
53 return (x >> 24) & 0xFF; 53 return (x >> 24) & 0xFF;
54} 54}
55 55
56static inline void summit_send_IPI_mask(const cpumask_t *mask, int vector) 56static inline void summit_send_IPI_mask(const struct cpumask *mask, int vector)
57{ 57{
58 default_send_IPI_mask_sequence_logical(mask, vector); 58 default_send_IPI_mask_sequence_logical(mask, vector);
59} 59}
60 60
61static void summit_send_IPI_allbutself(int vector) 61static void summit_send_IPI_allbutself(int vector)
62{ 62{
63 cpumask_t mask = cpu_online_map; 63 default_send_IPI_mask_allbutself_logical(cpu_online_mask, vector);
64 cpu_clear(smp_processor_id(), mask);
65
66 if (!cpus_empty(mask))
67 summit_send_IPI_mask(&mask, vector);
68} 64}
69 65
70static void summit_send_IPI_all(int vector) 66static void summit_send_IPI_all(int vector)
71{ 67{
72 summit_send_IPI_mask(&cpu_online_map, vector); 68 summit_send_IPI_mask(cpu_online_mask, vector);
73} 69}
74 70
75#include <asm/tsc.h> 71#include <asm/tsc.h>
@@ -186,13 +182,13 @@ static inline int is_WPEG(struct rio_detail *rio){
186 182
187#define SUMMIT_APIC_DFR_VALUE (APIC_DFR_CLUSTER) 183#define SUMMIT_APIC_DFR_VALUE (APIC_DFR_CLUSTER)
188 184
189static const cpumask_t *summit_target_cpus(void) 185static const struct cpumask *summit_target_cpus(void)
190{ 186{
191 /* CPU_MASK_ALL (0xff) has undefined behaviour with 187 /* CPU_MASK_ALL (0xff) has undefined behaviour with
192 * dest_LowestPrio mode logical clustered apic interrupt routing 188 * dest_LowestPrio mode logical clustered apic interrupt routing
193 * Just start on cpu 0. IRQ balancing will spread load 189 * Just start on cpu 0. IRQ balancing will spread load
194 */ 190 */
195 return &cpumask_of_cpu(0); 191 return cpumask_of(0);
196} 192}
197 193
198static unsigned long summit_check_apicid_used(physid_mask_t bitmap, int apicid) 194static unsigned long summit_check_apicid_used(physid_mask_t bitmap, int apicid)
@@ -289,7 +285,7 @@ static int summit_check_phys_apicid_present(int boot_cpu_physical_apicid)
289 return 1; 285 return 1;
290} 286}
291 287
292static unsigned int summit_cpu_mask_to_apicid(const cpumask_t *cpumask) 288static unsigned int summit_cpu_mask_to_apicid(const struct cpumask *cpumask)
293{ 289{
294 unsigned int round = 0; 290 unsigned int round = 0;
295 int cpu, apicid = 0; 291 int cpu, apicid = 0;
@@ -346,7 +342,7 @@ static int probe_summit(void)
346 return 0; 342 return 0;
347} 343}
348 344
349static void summit_vector_allocation_domain(int cpu, cpumask_t *retmask) 345static void summit_vector_allocation_domain(int cpu, struct cpumask *retmask)
350{ 346{
351 /* Careful. Some cpus do not strictly honor the set of cpus 347 /* Careful. Some cpus do not strictly honor the set of cpus
352 * specified in the interrupt destination when using lowest 348 * specified in the interrupt destination when using lowest
@@ -356,7 +352,8 @@ static void summit_vector_allocation_domain(int cpu, cpumask_t *retmask)
356 * deliver interrupts to the wrong hyperthread when only one 352 * deliver interrupts to the wrong hyperthread when only one
357 * hyperthread was specified in the interrupt desitination. 353 * hyperthread was specified in the interrupt desitination.
358 */ 354 */
359 *retmask = (cpumask_t){ { [0] = APIC_ALL_CPUS, } }; 355 cpumask_clear(retmask);
356 cpumask_bits(retmask)[0] = APIC_ALL_CPUS;
360} 357}
361 358
362#ifdef CONFIG_X86_SUMMIT_NUMA 359#ifdef CONFIG_X86_SUMMIT_NUMA
diff --git a/arch/x86/kernel/apm_32.c b/arch/x86/kernel/apm_32.c
index ac7783a67432..49e0939bac42 100644
--- a/arch/x86/kernel/apm_32.c
+++ b/arch/x86/kernel/apm_32.c
@@ -466,7 +466,7 @@ static const lookup_t error_table[] = {
466 * @err: APM BIOS return code 466 * @err: APM BIOS return code
467 * 467 *
468 * Write a meaningful log entry to the kernel log in the event of 468 * Write a meaningful log entry to the kernel log in the event of
469 * an APM error. 469 * an APM error. Note that this also handles (negative) kernel errors.
470 */ 470 */
471 471
472static void apm_error(char *str, int err) 472static void apm_error(char *str, int err)
@@ -478,43 +478,14 @@ static void apm_error(char *str, int err)
478 break; 478 break;
479 if (i < ERROR_COUNT) 479 if (i < ERROR_COUNT)
480 printk(KERN_NOTICE "apm: %s: %s\n", str, error_table[i].msg); 480 printk(KERN_NOTICE "apm: %s: %s\n", str, error_table[i].msg);
481 else if (err < 0)
482 printk(KERN_NOTICE "apm: %s: linux error code %i\n", str, err);
481 else 483 else
482 printk(KERN_NOTICE "apm: %s: unknown error code %#2.2x\n", 484 printk(KERN_NOTICE "apm: %s: unknown error code %#2.2x\n",
483 str, err); 485 str, err);
484} 486}
485 487
486/* 488/*
487 * Lock APM functionality to physical CPU 0
488 */
489
490#ifdef CONFIG_SMP
491
492static cpumask_t apm_save_cpus(void)
493{
494 cpumask_t x = current->cpus_allowed;
495 /* Some bioses don't like being called from CPU != 0 */
496 set_cpus_allowed(current, cpumask_of_cpu(0));
497 BUG_ON(smp_processor_id() != 0);
498 return x;
499}
500
501static inline void apm_restore_cpus(cpumask_t mask)
502{
503 set_cpus_allowed(current, mask);
504}
505
506#else
507
508/*
509 * No CPU lockdown needed on a uniprocessor
510 */
511
512#define apm_save_cpus() (current->cpus_allowed)
513#define apm_restore_cpus(x) (void)(x)
514
515#endif
516
517/*
518 * These are the actual BIOS calls. Depending on APM_ZERO_SEGS and 489 * These are the actual BIOS calls. Depending on APM_ZERO_SEGS and
519 * apm_info.allow_ints, we are being really paranoid here! Not only 490 * apm_info.allow_ints, we are being really paranoid here! Not only
520 * are interrupts disabled, but all the segment registers (except SS) 491 * are interrupts disabled, but all the segment registers (except SS)
@@ -568,16 +539,23 @@ static inline void apm_irq_restore(unsigned long flags)
568# define APM_DO_RESTORE_SEGS 539# define APM_DO_RESTORE_SEGS
569#endif 540#endif
570 541
542struct apm_bios_call {
543 u32 func;
544 /* In and out */
545 u32 ebx;
546 u32 ecx;
547 /* Out only */
548 u32 eax;
549 u32 edx;
550 u32 esi;
551
552 /* Error: -ENOMEM, or bits 8-15 of eax */
553 int err;
554};
555
571/** 556/**
572 * apm_bios_call - Make an APM BIOS 32bit call 557 * __apm_bios_call - Make an APM BIOS 32bit call
573 * @func: APM function to execute 558 * @_call: pointer to struct apm_bios_call.
574 * @ebx_in: EBX register for call entry
575 * @ecx_in: ECX register for call entry
576 * @eax: EAX register return
577 * @ebx: EBX register return
578 * @ecx: ECX register return
579 * @edx: EDX register return
580 * @esi: ESI register return
581 * 559 *
582 * Make an APM call using the 32bit protected mode interface. The 560 * Make an APM call using the 32bit protected mode interface. The
583 * caller is responsible for knowing if APM BIOS is configured and 561 * caller is responsible for knowing if APM BIOS is configured and
@@ -586,80 +564,142 @@ static inline void apm_irq_restore(unsigned long flags)
586 * flag is loaded into AL. If there is an error, then the error 564 * flag is loaded into AL. If there is an error, then the error
587 * code is returned in AH (bits 8-15 of eax) and this function 565 * code is returned in AH (bits 8-15 of eax) and this function
588 * returns non-zero. 566 * returns non-zero.
567 *
568 * Note: this makes the call on the current CPU.
589 */ 569 */
590 570static long __apm_bios_call(void *_call)
591static u8 apm_bios_call(u32 func, u32 ebx_in, u32 ecx_in,
592 u32 *eax, u32 *ebx, u32 *ecx, u32 *edx, u32 *esi)
593{ 571{
594 APM_DECL_SEGS 572 APM_DECL_SEGS
595 unsigned long flags; 573 unsigned long flags;
596 cpumask_t cpus;
597 int cpu; 574 int cpu;
598 struct desc_struct save_desc_40; 575 struct desc_struct save_desc_40;
599 struct desc_struct *gdt; 576 struct desc_struct *gdt;
600 577 struct apm_bios_call *call = _call;
601 cpus = apm_save_cpus();
602 578
603 cpu = get_cpu(); 579 cpu = get_cpu();
580 BUG_ON(cpu != 0);
604 gdt = get_cpu_gdt_table(cpu); 581 gdt = get_cpu_gdt_table(cpu);
605 save_desc_40 = gdt[0x40 / 8]; 582 save_desc_40 = gdt[0x40 / 8];
606 gdt[0x40 / 8] = bad_bios_desc; 583 gdt[0x40 / 8] = bad_bios_desc;
607 584
608 apm_irq_save(flags); 585 apm_irq_save(flags);
609 APM_DO_SAVE_SEGS; 586 APM_DO_SAVE_SEGS;
610 apm_bios_call_asm(func, ebx_in, ecx_in, eax, ebx, ecx, edx, esi); 587 apm_bios_call_asm(call->func, call->ebx, call->ecx,
588 &call->eax, &call->ebx, &call->ecx, &call->edx,
589 &call->esi);
611 APM_DO_RESTORE_SEGS; 590 APM_DO_RESTORE_SEGS;
612 apm_irq_restore(flags); 591 apm_irq_restore(flags);
613 gdt[0x40 / 8] = save_desc_40; 592 gdt[0x40 / 8] = save_desc_40;
614 put_cpu(); 593 put_cpu();
615 apm_restore_cpus(cpus);
616 594
617 return *eax & 0xff; 595 return call->eax & 0xff;
596}
597
598/* Run __apm_bios_call or __apm_bios_call_simple on CPU 0 */
599static int on_cpu0(long (*fn)(void *), struct apm_bios_call *call)
600{
601 int ret;
602
603 /* Don't bother with work_on_cpu in the common case, so we don't
604 * have to worry about OOM or overhead. */
605 if (get_cpu() == 0) {
606 ret = fn(call);
607 put_cpu();
608 } else {
609 put_cpu();
610 ret = work_on_cpu(0, fn, call);
611 }
612
613 /* work_on_cpu can fail with -ENOMEM */
614 if (ret < 0)
615 call->err = ret;
616 else
617 call->err = (call->eax >> 8) & 0xff;
618
619 return ret;
618} 620}
619 621
620/** 622/**
621 * apm_bios_call_simple - make a simple APM BIOS 32bit call 623 * apm_bios_call - Make an APM BIOS 32bit call (on CPU 0)
622 * @func: APM function to invoke 624 * @call: the apm_bios_call registers.
623 * @ebx_in: EBX register value for BIOS call 625 *
624 * @ecx_in: ECX register value for BIOS call 626 * If there is an error, it is returned in @call.err.
625 * @eax: EAX register on return from the BIOS call 627 */
628static int apm_bios_call(struct apm_bios_call *call)
629{
630 return on_cpu0(__apm_bios_call, call);
631}
632
633/**
634 * __apm_bios_call_simple - Make an APM BIOS 32bit call (on CPU 0)
635 * @_call: pointer to struct apm_bios_call.
626 * 636 *
627 * Make a BIOS call that returns one value only, or just status. 637 * Make a BIOS call that returns one value only, or just status.
628 * If there is an error, then the error code is returned in AH 638 * If there is an error, then the error code is returned in AH
629 * (bits 8-15 of eax) and this function returns non-zero. This is 639 * (bits 8-15 of eax) and this function returns non-zero (it can
630 * used for simpler BIOS operations. This call may hold interrupts 640 * also return -ENOMEM). This is used for simpler BIOS operations.
631 * off for a long time on some laptops. 641 * This call may hold interrupts off for a long time on some laptops.
642 *
643 * Note: this makes the call on the current CPU.
632 */ 644 */
633 645static long __apm_bios_call_simple(void *_call)
634static u8 apm_bios_call_simple(u32 func, u32 ebx_in, u32 ecx_in, u32 *eax)
635{ 646{
636 u8 error; 647 u8 error;
637 APM_DECL_SEGS 648 APM_DECL_SEGS
638 unsigned long flags; 649 unsigned long flags;
639 cpumask_t cpus;
640 int cpu; 650 int cpu;
641 struct desc_struct save_desc_40; 651 struct desc_struct save_desc_40;
642 struct desc_struct *gdt; 652 struct desc_struct *gdt;
643 653 struct apm_bios_call *call = _call;
644 cpus = apm_save_cpus();
645 654
646 cpu = get_cpu(); 655 cpu = get_cpu();
656 BUG_ON(cpu != 0);
647 gdt = get_cpu_gdt_table(cpu); 657 gdt = get_cpu_gdt_table(cpu);
648 save_desc_40 = gdt[0x40 / 8]; 658 save_desc_40 = gdt[0x40 / 8];
649 gdt[0x40 / 8] = bad_bios_desc; 659 gdt[0x40 / 8] = bad_bios_desc;
650 660
651 apm_irq_save(flags); 661 apm_irq_save(flags);
652 APM_DO_SAVE_SEGS; 662 APM_DO_SAVE_SEGS;
653 error = apm_bios_call_simple_asm(func, ebx_in, ecx_in, eax); 663 error = apm_bios_call_simple_asm(call->func, call->ebx, call->ecx,
664 &call->eax);
654 APM_DO_RESTORE_SEGS; 665 APM_DO_RESTORE_SEGS;
655 apm_irq_restore(flags); 666 apm_irq_restore(flags);
656 gdt[0x40 / 8] = save_desc_40; 667 gdt[0x40 / 8] = save_desc_40;
657 put_cpu(); 668 put_cpu();
658 apm_restore_cpus(cpus);
659 return error; 669 return error;
660} 670}
661 671
662/** 672/**
673 * apm_bios_call_simple - make a simple APM BIOS 32bit call
674 * @func: APM function to invoke
675 * @ebx_in: EBX register value for BIOS call
676 * @ecx_in: ECX register value for BIOS call
677 * @eax: EAX register on return from the BIOS call
678 * @err: bits
679 *
680 * Make a BIOS call that returns one value only, or just status.
681 * If there is an error, then the error code is returned in @err
682 * and this function returns non-zero. This is used for simpler
683 * BIOS operations. This call may hold interrupts off for a long
684 * time on some laptops.
685 */
686static int apm_bios_call_simple(u32 func, u32 ebx_in, u32 ecx_in, u32 *eax,
687 int *err)
688{
689 struct apm_bios_call call;
690 int ret;
691
692 call.func = func;
693 call.ebx = ebx_in;
694 call.ecx = ecx_in;
695
696 ret = on_cpu0(__apm_bios_call_simple, &call);
697 *eax = call.eax;
698 *err = call.err;
699 return ret;
700}
701
702/**
663 * apm_driver_version - APM driver version 703 * apm_driver_version - APM driver version
664 * @val: loaded with the APM version on return 704 * @val: loaded with the APM version on return
665 * 705 *
@@ -678,9 +718,10 @@ static u8 apm_bios_call_simple(u32 func, u32 ebx_in, u32 ecx_in, u32 *eax)
678static int apm_driver_version(u_short *val) 718static int apm_driver_version(u_short *val)
679{ 719{
680 u32 eax; 720 u32 eax;
721 int err;
681 722
682 if (apm_bios_call_simple(APM_FUNC_VERSION, 0, *val, &eax)) 723 if (apm_bios_call_simple(APM_FUNC_VERSION, 0, *val, &eax, &err))
683 return (eax >> 8) & 0xff; 724 return err;
684 *val = eax; 725 *val = eax;
685 return APM_SUCCESS; 726 return APM_SUCCESS;
686} 727}
@@ -701,22 +742,21 @@ static int apm_driver_version(u_short *val)
701 * that APM 1.2 is in use. If no messges are pending the value 0x80 742 * that APM 1.2 is in use. If no messges are pending the value 0x80
702 * is returned (No power management events pending). 743 * is returned (No power management events pending).
703 */ 744 */
704
705static int apm_get_event(apm_event_t *event, apm_eventinfo_t *info) 745static int apm_get_event(apm_event_t *event, apm_eventinfo_t *info)
706{ 746{
707 u32 eax; 747 struct apm_bios_call call;
708 u32 ebx;
709 u32 ecx;
710 u32 dummy;
711 748
712 if (apm_bios_call(APM_FUNC_GET_EVENT, 0, 0, &eax, &ebx, &ecx, 749 call.func = APM_FUNC_GET_EVENT;
713 &dummy, &dummy)) 750 call.ebx = call.ecx = 0;
714 return (eax >> 8) & 0xff; 751
715 *event = ebx; 752 if (apm_bios_call(&call))
753 return call.err;
754
755 *event = call.ebx;
716 if (apm_info.connection_version < 0x0102) 756 if (apm_info.connection_version < 0x0102)
717 *info = ~0; /* indicate info not valid */ 757 *info = ~0; /* indicate info not valid */
718 else 758 else
719 *info = ecx; 759 *info = call.ecx;
720 return APM_SUCCESS; 760 return APM_SUCCESS;
721} 761}
722 762
@@ -737,9 +777,10 @@ static int apm_get_event(apm_event_t *event, apm_eventinfo_t *info)
737static int set_power_state(u_short what, u_short state) 777static int set_power_state(u_short what, u_short state)
738{ 778{
739 u32 eax; 779 u32 eax;
780 int err;
740 781
741 if (apm_bios_call_simple(APM_FUNC_SET_STATE, what, state, &eax)) 782 if (apm_bios_call_simple(APM_FUNC_SET_STATE, what, state, &eax, &err))
742 return (eax >> 8) & 0xff; 783 return err;
743 return APM_SUCCESS; 784 return APM_SUCCESS;
744} 785}
745 786
@@ -770,6 +811,7 @@ static int apm_do_idle(void)
770 u8 ret = 0; 811 u8 ret = 0;
771 int idled = 0; 812 int idled = 0;
772 int polling; 813 int polling;
814 int err;
773 815
774 polling = !!(current_thread_info()->status & TS_POLLING); 816 polling = !!(current_thread_info()->status & TS_POLLING);
775 if (polling) { 817 if (polling) {
@@ -782,7 +824,7 @@ static int apm_do_idle(void)
782 } 824 }
783 if (!need_resched()) { 825 if (!need_resched()) {
784 idled = 1; 826 idled = 1;
785 ret = apm_bios_call_simple(APM_FUNC_IDLE, 0, 0, &eax); 827 ret = apm_bios_call_simple(APM_FUNC_IDLE, 0, 0, &eax, &err);
786 } 828 }
787 if (polling) 829 if (polling)
788 current_thread_info()->status |= TS_POLLING; 830 current_thread_info()->status |= TS_POLLING;
@@ -797,8 +839,7 @@ static int apm_do_idle(void)
797 * Only report the failure the first 5 times. 839 * Only report the failure the first 5 times.
798 */ 840 */
799 if (++t < 5) { 841 if (++t < 5) {
800 printk(KERN_DEBUG "apm_do_idle failed (%d)\n", 842 printk(KERN_DEBUG "apm_do_idle failed (%d)\n", err);
801 (eax >> 8) & 0xff);
802 t = jiffies; 843 t = jiffies;
803 } 844 }
804 return -1; 845 return -1;
@@ -816,9 +857,10 @@ static int apm_do_idle(void)
816static void apm_do_busy(void) 857static void apm_do_busy(void)
817{ 858{
818 u32 dummy; 859 u32 dummy;
860 int err;
819 861
820 if (clock_slowed || ALWAYS_CALL_BUSY) { 862 if (clock_slowed || ALWAYS_CALL_BUSY) {
821 (void)apm_bios_call_simple(APM_FUNC_BUSY, 0, 0, &dummy); 863 (void)apm_bios_call_simple(APM_FUNC_BUSY, 0, 0, &dummy, &err);
822 clock_slowed = 0; 864 clock_slowed = 0;
823 } 865 }
824} 866}
@@ -937,7 +979,7 @@ static void apm_power_off(void)
937 979
938 /* Some bioses don't like being called from CPU != 0 */ 980 /* Some bioses don't like being called from CPU != 0 */
939 if (apm_info.realmode_power_off) { 981 if (apm_info.realmode_power_off) {
940 (void)apm_save_cpus(); 982 set_cpus_allowed_ptr(current, cpumask_of(0));
941 machine_real_restart(po_bios_call, sizeof(po_bios_call)); 983 machine_real_restart(po_bios_call, sizeof(po_bios_call));
942 } else { 984 } else {
943 (void)set_system_power_state(APM_STATE_OFF); 985 (void)set_system_power_state(APM_STATE_OFF);
@@ -956,12 +998,13 @@ static void apm_power_off(void)
956static int apm_enable_power_management(int enable) 998static int apm_enable_power_management(int enable)
957{ 999{
958 u32 eax; 1000 u32 eax;
1001 int err;
959 1002
960 if ((enable == 0) && (apm_info.bios.flags & APM_BIOS_DISENGAGED)) 1003 if ((enable == 0) && (apm_info.bios.flags & APM_BIOS_DISENGAGED))
961 return APM_NOT_ENGAGED; 1004 return APM_NOT_ENGAGED;
962 if (apm_bios_call_simple(APM_FUNC_ENABLE_PM, APM_DEVICE_BALL, 1005 if (apm_bios_call_simple(APM_FUNC_ENABLE_PM, APM_DEVICE_BALL,
963 enable, &eax)) 1006 enable, &eax, &err))
964 return (eax >> 8) & 0xff; 1007 return err;
965 if (enable) 1008 if (enable)
966 apm_info.bios.flags &= ~APM_BIOS_DISABLED; 1009 apm_info.bios.flags &= ~APM_BIOS_DISABLED;
967 else 1010 else
@@ -986,24 +1029,23 @@ static int apm_enable_power_management(int enable)
986 1029
987static int apm_get_power_status(u_short *status, u_short *bat, u_short *life) 1030static int apm_get_power_status(u_short *status, u_short *bat, u_short *life)
988{ 1031{
989 u32 eax; 1032 struct apm_bios_call call;
990 u32 ebx; 1033
991 u32 ecx; 1034 call.func = APM_FUNC_GET_STATUS;
992 u32 edx; 1035 call.ebx = APM_DEVICE_ALL;
993 u32 dummy; 1036 call.ecx = 0;
994 1037
995 if (apm_info.get_power_status_broken) 1038 if (apm_info.get_power_status_broken)
996 return APM_32_UNSUPPORTED; 1039 return APM_32_UNSUPPORTED;
997 if (apm_bios_call(APM_FUNC_GET_STATUS, APM_DEVICE_ALL, 0, 1040 if (apm_bios_call(&call))
998 &eax, &ebx, &ecx, &edx, &dummy)) 1041 return call.err;
999 return (eax >> 8) & 0xff; 1042 *status = call.ebx;
1000 *status = ebx; 1043 *bat = call.ecx;
1001 *bat = ecx;
1002 if (apm_info.get_power_status_swabinminutes) { 1044 if (apm_info.get_power_status_swabinminutes) {
1003 *life = swab16((u16)edx); 1045 *life = swab16((u16)call.edx);
1004 *life |= 0x8000; 1046 *life |= 0x8000;
1005 } else 1047 } else
1006 *life = edx; 1048 *life = call.edx;
1007 return APM_SUCCESS; 1049 return APM_SUCCESS;
1008} 1050}
1009 1051
@@ -1048,12 +1090,14 @@ static int apm_get_battery_status(u_short which, u_short *status,
1048static int apm_engage_power_management(u_short device, int enable) 1090static int apm_engage_power_management(u_short device, int enable)
1049{ 1091{
1050 u32 eax; 1092 u32 eax;
1093 int err;
1051 1094
1052 if ((enable == 0) && (device == APM_DEVICE_ALL) 1095 if ((enable == 0) && (device == APM_DEVICE_ALL)
1053 && (apm_info.bios.flags & APM_BIOS_DISABLED)) 1096 && (apm_info.bios.flags & APM_BIOS_DISABLED))
1054 return APM_DISABLED; 1097 return APM_DISABLED;
1055 if (apm_bios_call_simple(APM_FUNC_ENGAGE_PM, device, enable, &eax)) 1098 if (apm_bios_call_simple(APM_FUNC_ENGAGE_PM, device, enable,
1056 return (eax >> 8) & 0xff; 1099 &eax, &err))
1100 return err;
1057 if (device == APM_DEVICE_ALL) { 1101 if (device == APM_DEVICE_ALL) {
1058 if (enable) 1102 if (enable)
1059 apm_info.bios.flags &= ~APM_BIOS_DISENGAGED; 1103 apm_info.bios.flags &= ~APM_BIOS_DISENGAGED;
@@ -1689,16 +1733,14 @@ static int apm(void *unused)
1689 char *power_stat; 1733 char *power_stat;
1690 char *bat_stat; 1734 char *bat_stat;
1691 1735
1692#ifdef CONFIG_SMP
1693 /* 2002/08/01 - WT 1736 /* 2002/08/01 - WT
1694 * This is to avoid random crashes at boot time during initialization 1737 * This is to avoid random crashes at boot time during initialization
1695 * on SMP systems in case of "apm=power-off" mode. Seen on ASUS A7M266D. 1738 * on SMP systems in case of "apm=power-off" mode. Seen on ASUS A7M266D.
1696 * Some bioses don't like being called from CPU != 0. 1739 * Some bioses don't like being called from CPU != 0.
1697 * Method suggested by Ingo Molnar. 1740 * Method suggested by Ingo Molnar.
1698 */ 1741 */
1699 set_cpus_allowed(current, cpumask_of_cpu(0)); 1742 set_cpus_allowed_ptr(current, cpumask_of(0));
1700 BUG_ON(smp_processor_id() != 0); 1743 BUG_ON(smp_processor_id() != 0);
1701#endif
1702 1744
1703 if (apm_info.connection_version == 0) { 1745 if (apm_info.connection_version == 0) {
1704 apm_info.connection_version = apm_info.bios.version; 1746 apm_info.connection_version = apm_info.bios.version;
diff --git a/arch/x86/kernel/cpu/addon_cpuid_features.c b/arch/x86/kernel/cpu/addon_cpuid_features.c
index 8220ae69849d..c965e5212714 100644
--- a/arch/x86/kernel/cpu/addon_cpuid_features.c
+++ b/arch/x86/kernel/cpu/addon_cpuid_features.c
@@ -31,6 +31,7 @@ void __cpuinit init_scattered_cpuid_features(struct cpuinfo_x86 *c)
31 31
32 static const struct cpuid_bit __cpuinitconst cpuid_bits[] = { 32 static const struct cpuid_bit __cpuinitconst cpuid_bits[] = {
33 { X86_FEATURE_IDA, CR_EAX, 1, 0x00000006 }, 33 { X86_FEATURE_IDA, CR_EAX, 1, 0x00000006 },
34 { X86_FEATURE_ARAT, CR_EAX, 2, 0x00000006 },
34 { 0, 0, 0, 0 } 35 { 0, 0, 0, 0 }
35 }; 36 };
36 37
diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c
index e2962cc1e27b..c4f667896c28 100644
--- a/arch/x86/kernel/cpu/common.c
+++ b/arch/x86/kernel/cpu/common.c
@@ -41,8 +41,6 @@
41 41
42#include "cpu.h" 42#include "cpu.h"
43 43
44#ifdef CONFIG_X86_64
45
46/* all of these masks are initialized in setup_cpu_local_masks() */ 44/* all of these masks are initialized in setup_cpu_local_masks() */
47cpumask_var_t cpu_initialized_mask; 45cpumask_var_t cpu_initialized_mask;
48cpumask_var_t cpu_callout_mask; 46cpumask_var_t cpu_callout_mask;
@@ -60,16 +58,6 @@ void __init setup_cpu_local_masks(void)
60 alloc_bootmem_cpumask_var(&cpu_sibling_setup_mask); 58 alloc_bootmem_cpumask_var(&cpu_sibling_setup_mask);
61} 59}
62 60
63#else /* CONFIG_X86_32 */
64
65cpumask_t cpu_sibling_setup_map;
66cpumask_t cpu_callout_map;
67cpumask_t cpu_initialized;
68cpumask_t cpu_callin_map;
69
70#endif /* CONFIG_X86_32 */
71
72
73static const struct cpu_dev *this_cpu __cpuinitdata; 61static const struct cpu_dev *this_cpu __cpuinitdata;
74 62
75DEFINE_PER_CPU_PAGE_ALIGNED(struct gdt_page, gdt_page) = { .gdt = { 63DEFINE_PER_CPU_PAGE_ALIGNED(struct gdt_page, gdt_page) = { .gdt = {
@@ -859,6 +847,7 @@ static void vgetcpu_set_mode(void)
859void __init identify_boot_cpu(void) 847void __init identify_boot_cpu(void)
860{ 848{
861 identify_cpu(&boot_cpu_data); 849 identify_cpu(&boot_cpu_data);
850 init_c1e_mask();
862#ifdef CONFIG_X86_32 851#ifdef CONFIG_X86_32
863 sysenter_setup(); 852 sysenter_setup();
864 enable_sep_cpu(); 853 enable_sep_cpu();
diff --git a/arch/x86/kernel/cpu/cpu_debug.c b/arch/x86/kernel/cpu/cpu_debug.c
index 46e29ab96c6a..46e29ab96c6a 100755..100644
--- a/arch/x86/kernel/cpu/cpu_debug.c
+++ b/arch/x86/kernel/cpu/cpu_debug.c
diff --git a/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c b/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c
index 23da96e57b17..ecdb682ab516 100644
--- a/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c
+++ b/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c
@@ -33,7 +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/ftrace.h> 36#include <trace/power.h>
37 37
38#include <linux/acpi.h> 38#include <linux/acpi.h>
39#include <linux/io.h> 39#include <linux/io.h>
@@ -68,10 +68,13 @@ struct acpi_cpufreq_data {
68 unsigned int max_freq; 68 unsigned int max_freq;
69 unsigned int resume; 69 unsigned int resume;
70 unsigned int cpu_feature; 70 unsigned int cpu_feature;
71 u64 saved_aperf, saved_mperf;
71}; 72};
72 73
73static DEFINE_PER_CPU(struct acpi_cpufreq_data *, drv_data); 74static DEFINE_PER_CPU(struct acpi_cpufreq_data *, drv_data);
74 75
76DEFINE_TRACE(power_mark);
77
75/* acpi_perf_data is a pointer to percpu data. */ 78/* acpi_perf_data is a pointer to percpu data. */
76static struct acpi_processor_performance *acpi_perf_data; 79static struct acpi_processor_performance *acpi_perf_data;
77 80
@@ -150,7 +153,8 @@ struct drv_cmd {
150 u32 val; 153 u32 val;
151}; 154};
152 155
153static long do_drv_read(void *_cmd) 156/* Called via smp_call_function_single(), on the target CPU */
157static void do_drv_read(void *_cmd)
154{ 158{
155 struct drv_cmd *cmd = _cmd; 159 struct drv_cmd *cmd = _cmd;
156 u32 h; 160 u32 h;
@@ -167,10 +171,10 @@ static long do_drv_read(void *_cmd)
167 default: 171 default:
168 break; 172 break;
169 } 173 }
170 return 0;
171} 174}
172 175
173static long do_drv_write(void *_cmd) 176/* Called via smp_call_function_many(), on the target CPUs */
177static void do_drv_write(void *_cmd)
174{ 178{
175 struct drv_cmd *cmd = _cmd; 179 struct drv_cmd *cmd = _cmd;
176 u32 lo, hi; 180 u32 lo, hi;
@@ -189,23 +193,24 @@ static long do_drv_write(void *_cmd)
189 default: 193 default:
190 break; 194 break;
191 } 195 }
192 return 0;
193} 196}
194 197
195static void drv_read(struct drv_cmd *cmd) 198static void drv_read(struct drv_cmd *cmd)
196{ 199{
197 cmd->val = 0; 200 cmd->val = 0;
198 201
199 work_on_cpu(cpumask_any(cmd->mask), do_drv_read, cmd); 202 smp_call_function_single(cpumask_any(cmd->mask), do_drv_read, cmd, 1);
200} 203}
201 204
202static void drv_write(struct drv_cmd *cmd) 205static void drv_write(struct drv_cmd *cmd)
203{ 206{
204 unsigned int i; 207 int this_cpu;
205 208
206 for_each_cpu(i, cmd->mask) { 209 this_cpu = get_cpu();
207 work_on_cpu(i, do_drv_write, cmd); 210 if (cpumask_test_cpu(this_cpu, cmd->mask))
208 } 211 do_drv_write(cmd);
212 smp_call_function_many(cmd->mask, do_drv_write, cmd, 1);
213 put_cpu();
209} 214}
210 215
211static u32 get_cur_val(const struct cpumask *mask) 216static u32 get_cur_val(const struct cpumask *mask)
@@ -239,28 +244,23 @@ static u32 get_cur_val(const struct cpumask *mask)
239 return cmd.val; 244 return cmd.val;
240} 245}
241 246
242struct perf_cur { 247struct perf_pair {
243 union { 248 union {
244 struct { 249 struct {
245 u32 lo; 250 u32 lo;
246 u32 hi; 251 u32 hi;
247 } split; 252 } split;
248 u64 whole; 253 u64 whole;
249 } aperf_cur, mperf_cur; 254 } aperf, mperf;
250}; 255};
251 256
252 257/* Called via smp_call_function_single(), on the target CPU */
253static long read_measured_perf_ctrs(void *_cur) 258static void read_measured_perf_ctrs(void *_cur)
254{ 259{
255 struct perf_cur *cur = _cur; 260 struct perf_pair *cur = _cur;
256
257 rdmsr(MSR_IA32_APERF, cur->aperf_cur.split.lo, cur->aperf_cur.split.hi);
258 rdmsr(MSR_IA32_MPERF, cur->mperf_cur.split.lo, cur->mperf_cur.split.hi);
259 261
260 wrmsr(MSR_IA32_APERF, 0, 0); 262 rdmsr(MSR_IA32_APERF, cur->aperf.split.lo, cur->aperf.split.hi);
261 wrmsr(MSR_IA32_MPERF, 0, 0); 263 rdmsr(MSR_IA32_MPERF, cur->mperf.split.lo, cur->mperf.split.hi);
262
263 return 0;
264} 264}
265 265
266/* 266/*
@@ -279,52 +279,57 @@ static long read_measured_perf_ctrs(void *_cur)
279static unsigned int get_measured_perf(struct cpufreq_policy *policy, 279static unsigned int get_measured_perf(struct cpufreq_policy *policy,
280 unsigned int cpu) 280 unsigned int cpu)
281{ 281{
282 struct perf_cur cur; 282 struct perf_pair readin, cur;
283 unsigned int perf_percent; 283 unsigned int perf_percent;
284 unsigned int retval; 284 unsigned int retval;
285 285
286 if (!work_on_cpu(cpu, read_measured_perf_ctrs, &cur)) 286 if (smp_call_function_single(cpu, read_measured_perf_ctrs, &readin, 1))
287 return 0; 287 return 0;
288 288
289 cur.aperf.whole = readin.aperf.whole -
290 per_cpu(drv_data, cpu)->saved_aperf;
291 cur.mperf.whole = readin.mperf.whole -
292 per_cpu(drv_data, cpu)->saved_mperf;
293 per_cpu(drv_data, cpu)->saved_aperf = readin.aperf.whole;
294 per_cpu(drv_data, cpu)->saved_mperf = readin.mperf.whole;
295
289#ifdef __i386__ 296#ifdef __i386__
290 /* 297 /*
291 * We dont want to do 64 bit divide with 32 bit kernel 298 * We dont want to do 64 bit divide with 32 bit kernel
292 * Get an approximate value. Return failure in case we cannot get 299 * Get an approximate value. Return failure in case we cannot get
293 * an approximate value. 300 * an approximate value.
294 */ 301 */
295 if (unlikely(cur.aperf_cur.split.hi || cur.mperf_cur.split.hi)) { 302 if (unlikely(cur.aperf.split.hi || cur.mperf.split.hi)) {
296 int shift_count; 303 int shift_count;
297 u32 h; 304 u32 h;
298 305
299 h = max_t(u32, cur.aperf_cur.split.hi, cur.mperf_cur.split.hi); 306 h = max_t(u32, cur.aperf.split.hi, cur.mperf.split.hi);
300 shift_count = fls(h); 307 shift_count = fls(h);
301 308
302 cur.aperf_cur.whole >>= shift_count; 309 cur.aperf.whole >>= shift_count;
303 cur.mperf_cur.whole >>= shift_count; 310 cur.mperf.whole >>= shift_count;
304 } 311 }
305 312
306 if (((unsigned long)(-1) / 100) < cur.aperf_cur.split.lo) { 313 if (((unsigned long)(-1) / 100) < cur.aperf.split.lo) {
307 int shift_count = 7; 314 int shift_count = 7;
308 cur.aperf_cur.split.lo >>= shift_count; 315 cur.aperf.split.lo >>= shift_count;
309 cur.mperf_cur.split.lo >>= shift_count; 316 cur.mperf.split.lo >>= shift_count;
310 } 317 }
311 318
312 if (cur.aperf_cur.split.lo && cur.mperf_cur.split.lo) 319 if (cur.aperf.split.lo && cur.mperf.split.lo)
313 perf_percent = (cur.aperf_cur.split.lo * 100) / 320 perf_percent = (cur.aperf.split.lo * 100) / cur.mperf.split.lo;
314 cur.mperf_cur.split.lo;
315 else 321 else
316 perf_percent = 0; 322 perf_percent = 0;
317 323
318#else 324#else
319 if (unlikely(((unsigned long)(-1) / 100) < cur.aperf_cur.whole)) { 325 if (unlikely(((unsigned long)(-1) / 100) < cur.aperf.whole)) {
320 int shift_count = 7; 326 int shift_count = 7;
321 cur.aperf_cur.whole >>= shift_count; 327 cur.aperf.whole >>= shift_count;
322 cur.mperf_cur.whole >>= shift_count; 328 cur.mperf.whole >>= shift_count;
323 } 329 }
324 330
325 if (cur.aperf_cur.whole && cur.mperf_cur.whole) 331 if (cur.aperf.whole && cur.mperf.whole)
326 perf_percent = (cur.aperf_cur.whole * 100) / 332 perf_percent = (cur.aperf.whole * 100) / cur.mperf.whole;
327 cur.mperf_cur.whole;
328 else 333 else
329 perf_percent = 0; 334 perf_percent = 0;
330 335
@@ -680,6 +685,18 @@ static int acpi_cpufreq_cpu_init(struct cpufreq_policy *policy)
680 perf->states[i].transition_latency * 1000; 685 perf->states[i].transition_latency * 1000;
681 } 686 }
682 687
688 /* Check for high latency (>20uS) from buggy BIOSes, like on T42 */
689 if (perf->control_register.space_id == ACPI_ADR_SPACE_FIXED_HARDWARE &&
690 policy->cpuinfo.transition_latency > 20 * 1000) {
691 static int print_once;
692 policy->cpuinfo.transition_latency = 20 * 1000;
693 if (!print_once) {
694 print_once = 1;
695 printk(KERN_INFO "Capping off P-state tranision latency"
696 " at 20 uS\n");
697 }
698 }
699
683 data->max_freq = perf->states[0].core_frequency * 1000; 700 data->max_freq = perf->states[0].core_frequency * 1000;
684 /* table init */ 701 /* table init */
685 for (i = 0; i < perf->state_count; i++) { 702 for (i = 0; i < perf->state_count; i++) {
diff --git a/arch/x86/kernel/cpu/cpufreq/longhaul.c b/arch/x86/kernel/cpu/cpufreq/longhaul.c
index f1c51aea064d..ce2ed3e4aad9 100644
--- a/arch/x86/kernel/cpu/cpufreq/longhaul.c
+++ b/arch/x86/kernel/cpu/cpufreq/longhaul.c
@@ -33,7 +33,6 @@
33#include <linux/timex.h> 33#include <linux/timex.h>
34#include <linux/io.h> 34#include <linux/io.h>
35#include <linux/acpi.h> 35#include <linux/acpi.h>
36#include <linux/kernel.h>
37 36
38#include <asm/msr.h> 37#include <asm/msr.h>
39#include <acpi/processor.h> 38#include <acpi/processor.h>
@@ -305,7 +304,7 @@ retry_loop:
305 outb(3, 0x22); 304 outb(3, 0x22);
306 } else if ((pr != NULL) && pr->flags.bm_control) { 305 } else if ((pr != NULL) && pr->flags.bm_control) {
307 /* Disable bus master arbitration */ 306 /* Disable bus master arbitration */
308 acpi_set_register(ACPI_BITREG_ARB_DISABLE, 1); 307 acpi_write_bit_register(ACPI_BITREG_ARB_DISABLE, 1);
309 } 308 }
310 switch (longhaul_version) { 309 switch (longhaul_version) {
311 310
@@ -328,7 +327,7 @@ retry_loop:
328 case TYPE_POWERSAVER: 327 case TYPE_POWERSAVER:
329 if (longhaul_flags & USE_ACPI_C3) { 328 if (longhaul_flags & USE_ACPI_C3) {
330 /* Don't allow wakeup */ 329 /* Don't allow wakeup */
331 acpi_set_register(ACPI_BITREG_BUS_MASTER_RLD, 0); 330 acpi_write_bit_register(ACPI_BITREG_BUS_MASTER_RLD, 0);
332 do_powersaver(cx->address, mults_index, dir); 331 do_powersaver(cx->address, mults_index, dir);
333 } else { 332 } else {
334 do_powersaver(0, mults_index, dir); 333 do_powersaver(0, mults_index, dir);
@@ -341,7 +340,7 @@ retry_loop:
341 outb(0, 0x22); 340 outb(0, 0x22);
342 } else if ((pr != NULL) && pr->flags.bm_control) { 341 } else if ((pr != NULL) && pr->flags.bm_control) {
343 /* Enable bus master arbitration */ 342 /* Enable bus master arbitration */
344 acpi_set_register(ACPI_BITREG_ARB_DISABLE, 0); 343 acpi_write_bit_register(ACPI_BITREG_ARB_DISABLE, 0);
345 } 344 }
346 outb(pic2_mask, 0xA1); /* restore mask */ 345 outb(pic2_mask, 0xA1); /* restore mask */
347 outb(pic1_mask, 0x21); 346 outb(pic1_mask, 0x21);
diff --git a/arch/x86/kernel/cpu/cpufreq/p4-clockmod.c b/arch/x86/kernel/cpu/cpufreq/p4-clockmod.c
index 41ed94915f97..6ac55bd341ae 100644
--- a/arch/x86/kernel/cpu/cpufreq/p4-clockmod.c
+++ b/arch/x86/kernel/cpu/cpufreq/p4-clockmod.c
@@ -211,7 +211,7 @@ static int cpufreq_p4_cpu_init(struct cpufreq_policy *policy)
211 unsigned int i; 211 unsigned int i;
212 212
213#ifdef CONFIG_SMP 213#ifdef CONFIG_SMP
214 cpumask_copy(policy->cpus, &per_cpu(cpu_sibling_map, policy->cpu)); 214 cpumask_copy(policy->cpus, cpu_sibling_mask(policy->cpu));
215#endif 215#endif
216 216
217 /* Errata workaround */ 217 /* Errata workaround */
diff --git a/arch/x86/kernel/cpu/cpufreq/powernow-k8.c b/arch/x86/kernel/cpu/cpufreq/powernow-k8.c
index a15ac94e0b9b..4709ead2db52 100644
--- a/arch/x86/kernel/cpu/cpufreq/powernow-k8.c
+++ b/arch/x86/kernel/cpu/cpufreq/powernow-k8.c
@@ -54,7 +54,10 @@ static DEFINE_PER_CPU(struct powernow_k8_data *, powernow_data);
54static int cpu_family = CPU_OPTERON; 54static int cpu_family = CPU_OPTERON;
55 55
56#ifndef CONFIG_SMP 56#ifndef CONFIG_SMP
57DEFINE_PER_CPU(cpumask_t, cpu_core_map); 57static inline const struct cpumask *cpu_core_mask(int cpu)
58{
59 return cpumask_of(0);
60}
58#endif 61#endif
59 62
60/* Return a frequency in MHz, given an input fid */ 63/* Return a frequency in MHz, given an input fid */
@@ -699,7 +702,7 @@ static int fill_powernow_table(struct powernow_k8_data *data,
699 702
700 dprintk("cfid 0x%x, cvid 0x%x\n", data->currfid, data->currvid); 703 dprintk("cfid 0x%x, cvid 0x%x\n", data->currfid, data->currvid);
701 data->powernow_table = powernow_table; 704 data->powernow_table = powernow_table;
702 if (first_cpu(per_cpu(cpu_core_map, data->cpu)) == data->cpu) 705 if (cpumask_first(cpu_core_mask(data->cpu)) == data->cpu)
703 print_basics(data); 706 print_basics(data);
704 707
705 for (j = 0; j < data->numps; j++) 708 for (j = 0; j < data->numps; j++)
@@ -862,7 +865,7 @@ static int powernow_k8_cpu_init_acpi(struct powernow_k8_data *data)
862 865
863 /* fill in data */ 866 /* fill in data */
864 data->numps = data->acpi_data.state_count; 867 data->numps = data->acpi_data.state_count;
865 if (first_cpu(per_cpu(cpu_core_map, data->cpu)) == data->cpu) 868 if (cpumask_first(cpu_core_mask(data->cpu)) == data->cpu)
866 print_basics(data); 869 print_basics(data);
867 powernow_k8_acpi_pst_values(data, 0); 870 powernow_k8_acpi_pst_values(data, 0);
868 871
@@ -1300,7 +1303,7 @@ static int __cpuinit powernowk8_cpu_init(struct cpufreq_policy *pol)
1300 if (cpu_family == CPU_HW_PSTATE) 1303 if (cpu_family == CPU_HW_PSTATE)
1301 cpumask_copy(pol->cpus, cpumask_of(pol->cpu)); 1304 cpumask_copy(pol->cpus, cpumask_of(pol->cpu));
1302 else 1305 else
1303 cpumask_copy(pol->cpus, &per_cpu(cpu_core_map, pol->cpu)); 1306 cpumask_copy(pol->cpus, cpu_core_mask(pol->cpu));
1304 data->available_cores = pol->cpus; 1307 data->available_cores = pol->cpus;
1305 1308
1306 if (cpu_family == CPU_HW_PSTATE) 1309 if (cpu_family == CPU_HW_PSTATE)
@@ -1365,7 +1368,7 @@ static unsigned int powernowk8_get(unsigned int cpu)
1365 unsigned int khz = 0; 1368 unsigned int khz = 0;
1366 unsigned int first; 1369 unsigned int first;
1367 1370
1368 first = first_cpu(per_cpu(cpu_core_map, cpu)); 1371 first = cpumask_first(cpu_core_mask(cpu));
1369 data = per_cpu(powernow_data, first); 1372 data = per_cpu(powernow_data, first);
1370 1373
1371 if (!data) 1374 if (!data)
diff --git a/arch/x86/kernel/cpu/cpufreq/speedstep-ich.c b/arch/x86/kernel/cpu/cpufreq/speedstep-ich.c
index 8bbb11adb315..016c1a4fa3fc 100644
--- a/arch/x86/kernel/cpu/cpufreq/speedstep-ich.c
+++ b/arch/x86/kernel/cpu/cpufreq/speedstep-ich.c
@@ -321,7 +321,7 @@ static int speedstep_cpu_init(struct cpufreq_policy *policy)
321 321
322 /* only run on CPU to be set, or on its sibling */ 322 /* only run on CPU to be set, or on its sibling */
323#ifdef CONFIG_SMP 323#ifdef CONFIG_SMP
324 cpumask_copy(policy->cpus, &per_cpu(cpu_sibling_map, policy->cpu)); 324 cpumask_copy(policy->cpus, cpu_sibling_mask(policy->cpu));
325#endif 325#endif
326 326
327 cpus_allowed = current->cpus_allowed; 327 cpus_allowed = current->cpus_allowed;
diff --git a/arch/x86/kernel/cpu/intel_cacheinfo.c b/arch/x86/kernel/cpu/intel_cacheinfo.c
index c471eb1a389c..483eda96e102 100644
--- a/arch/x86/kernel/cpu/intel_cacheinfo.c
+++ b/arch/x86/kernel/cpu/intel_cacheinfo.c
@@ -159,7 +159,7 @@ struct _cpuid4_info_regs {
159 unsigned long can_disable; 159 unsigned long can_disable;
160}; 160};
161 161
162#ifdef CONFIG_PCI 162#if defined(CONFIG_PCI) && defined(CONFIG_SYSFS)
163static struct pci_device_id k8_nb_id[] = { 163static struct pci_device_id k8_nb_id[] = {
164 { PCI_DEVICE(PCI_VENDOR_ID_AMD, 0x1103) }, 164 { PCI_DEVICE(PCI_VENDOR_ID_AMD, 0x1103) },
165 { PCI_DEVICE(PCI_VENDOR_ID_AMD, 0x1203) }, 165 { PCI_DEVICE(PCI_VENDOR_ID_AMD, 0x1203) },
@@ -324,15 +324,6 @@ __cpuinit cpuid4_cache_lookup_regs(int index,
324 return 0; 324 return 0;
325} 325}
326 326
327static int
328__cpuinit cpuid4_cache_lookup(int index, struct _cpuid4_info *this_leaf)
329{
330 struct _cpuid4_info_regs *leaf_regs =
331 (struct _cpuid4_info_regs *)this_leaf;
332
333 return cpuid4_cache_lookup_regs(index, leaf_regs);
334}
335
336static int __cpuinit find_num_cache_leaves(void) 327static int __cpuinit find_num_cache_leaves(void)
337{ 328{
338 unsigned int eax, ebx, ecx, edx; 329 unsigned int eax, ebx, ecx, edx;
@@ -508,6 +499,8 @@ unsigned int __cpuinit init_intel_cacheinfo(struct cpuinfo_x86 *c)
508 return l2; 499 return l2;
509} 500}
510 501
502#ifdef CONFIG_SYSFS
503
511/* pointer to _cpuid4_info array (for each cache leaf) */ 504/* pointer to _cpuid4_info array (for each cache leaf) */
512static DEFINE_PER_CPU(struct _cpuid4_info *, cpuid4_info); 505static DEFINE_PER_CPU(struct _cpuid4_info *, cpuid4_info);
513#define CPUID4_INFO_IDX(x, y) (&((per_cpu(cpuid4_info, x))[y])) 506#define CPUID4_INFO_IDX(x, y) (&((per_cpu(cpuid4_info, x))[y]))
@@ -571,6 +564,15 @@ static void __cpuinit free_cache_attributes(unsigned int cpu)
571 per_cpu(cpuid4_info, cpu) = NULL; 564 per_cpu(cpuid4_info, cpu) = NULL;
572} 565}
573 566
567static int
568__cpuinit cpuid4_cache_lookup(int index, struct _cpuid4_info *this_leaf)
569{
570 struct _cpuid4_info_regs *leaf_regs =
571 (struct _cpuid4_info_regs *)this_leaf;
572
573 return cpuid4_cache_lookup_regs(index, leaf_regs);
574}
575
574static void __cpuinit get_cpu_leaves(void *_retval) 576static void __cpuinit get_cpu_leaves(void *_retval)
575{ 577{
576 int j, *retval = _retval, cpu = smp_processor_id(); 578 int j, *retval = _retval, cpu = smp_processor_id();
@@ -612,8 +614,6 @@ static int __cpuinit detect_cache_attributes(unsigned int cpu)
612 return retval; 614 return retval;
613} 615}
614 616
615#ifdef CONFIG_SYSFS
616
617#include <linux/kobject.h> 617#include <linux/kobject.h>
618#include <linux/sysfs.h> 618#include <linux/sysfs.h>
619 619
diff --git a/arch/x86/kernel/cpu/mcheck/mce_64.c b/arch/x86/kernel/cpu/mcheck/mce_64.c
index ca14604611ec..863f89568b1a 100644
--- a/arch/x86/kernel/cpu/mcheck/mce_64.c
+++ b/arch/x86/kernel/cpu/mcheck/mce_64.c
@@ -990,7 +990,7 @@ static struct sysdev_attribute *mce_attributes[] = {
990 NULL 990 NULL
991}; 991};
992 992
993static cpumask_t mce_device_initialized = CPU_MASK_NONE; 993static cpumask_var_t mce_device_initialized;
994 994
995/* Per cpu sysdev init. All of the cpus still share the same ctl bank */ 995/* Per cpu sysdev init. All of the cpus still share the same ctl bank */
996static __cpuinit int mce_create_device(unsigned int cpu) 996static __cpuinit int mce_create_device(unsigned int cpu)
@@ -1021,7 +1021,7 @@ static __cpuinit int mce_create_device(unsigned int cpu)
1021 if (err) 1021 if (err)
1022 goto error2; 1022 goto error2;
1023 } 1023 }
1024 cpu_set(cpu, mce_device_initialized); 1024 cpumask_set_cpu(cpu, mce_device_initialized);
1025 1025
1026 return 0; 1026 return 0;
1027error2: 1027error2:
@@ -1043,7 +1043,7 @@ static __cpuinit void mce_remove_device(unsigned int cpu)
1043{ 1043{
1044 int i; 1044 int i;
1045 1045
1046 if (!cpu_isset(cpu, mce_device_initialized)) 1046 if (!cpumask_test_cpu(cpu, mce_device_initialized))
1047 return; 1047 return;
1048 1048
1049 for (i = 0; mce_attributes[i]; i++) 1049 for (i = 0; mce_attributes[i]; i++)
@@ -1053,7 +1053,7 @@ static __cpuinit void mce_remove_device(unsigned int cpu)
1053 sysdev_remove_file(&per_cpu(device_mce, cpu), 1053 sysdev_remove_file(&per_cpu(device_mce, cpu),
1054 &bank_attrs[i]); 1054 &bank_attrs[i]);
1055 sysdev_unregister(&per_cpu(device_mce,cpu)); 1055 sysdev_unregister(&per_cpu(device_mce,cpu));
1056 cpu_clear(cpu, mce_device_initialized); 1056 cpumask_clear_cpu(cpu, mce_device_initialized);
1057} 1057}
1058 1058
1059/* Make sure there are no machine checks on offlined CPUs. */ 1059/* Make sure there are no machine checks on offlined CPUs. */
@@ -1162,6 +1162,8 @@ static __init int mce_init_device(void)
1162 if (!mce_available(&boot_cpu_data)) 1162 if (!mce_available(&boot_cpu_data))
1163 return -EIO; 1163 return -EIO;
1164 1164
1165 alloc_cpumask_var(&mce_device_initialized, GFP_KERNEL);
1166
1165 err = mce_init_banks(); 1167 err = mce_init_banks();
1166 if (err) 1168 if (err)
1167 return err; 1169 return err;
diff --git a/arch/x86/kernel/cpu/mcheck/mce_amd_64.c b/arch/x86/kernel/cpu/mcheck/mce_amd_64.c
index 7d01be868870..56dde9c4bc96 100644
--- a/arch/x86/kernel/cpu/mcheck/mce_amd_64.c
+++ b/arch/x86/kernel/cpu/mcheck/mce_amd_64.c
@@ -485,7 +485,7 @@ static __cpuinit int threshold_create_bank(unsigned int cpu, unsigned int bank)
485 485
486#ifdef CONFIG_SMP 486#ifdef CONFIG_SMP
487 if (cpu_data(cpu).cpu_core_id && shared_bank[bank]) { /* symlink */ 487 if (cpu_data(cpu).cpu_core_id && shared_bank[bank]) { /* symlink */
488 i = cpumask_first(&per_cpu(cpu_core_map, cpu)); 488 i = cpumask_first(cpu_core_mask(cpu));
489 489
490 /* first core not up yet */ 490 /* first core not up yet */
491 if (cpu_data(i).cpu_core_id) 491 if (cpu_data(i).cpu_core_id)
@@ -505,7 +505,7 @@ static __cpuinit int threshold_create_bank(unsigned int cpu, unsigned int bank)
505 if (err) 505 if (err)
506 goto out; 506 goto out;
507 507
508 cpumask_copy(b->cpus, &per_cpu(cpu_core_map, cpu)); 508 cpumask_copy(b->cpus, cpu_core_mask(cpu));
509 per_cpu(threshold_banks, cpu)[bank] = b; 509 per_cpu(threshold_banks, cpu)[bank] = b;
510 goto out; 510 goto out;
511 } 511 }
@@ -529,7 +529,7 @@ static __cpuinit int threshold_create_bank(unsigned int cpu, unsigned int bank)
529#ifndef CONFIG_SMP 529#ifndef CONFIG_SMP
530 cpumask_setall(b->cpus); 530 cpumask_setall(b->cpus);
531#else 531#else
532 cpumask_copy(b->cpus, &per_cpu(cpu_core_map, cpu)); 532 cpumask_copy(b->cpus, cpu_core_mask(cpu));
533#endif 533#endif
534 534
535 per_cpu(threshold_banks, cpu)[bank] = b; 535 per_cpu(threshold_banks, cpu)[bank] = b;
diff --git a/arch/x86/kernel/cpu/mcheck/mce_intel_64.c b/arch/x86/kernel/cpu/mcheck/mce_intel_64.c
index 57df3d383470..d6b72df89d69 100644
--- a/arch/x86/kernel/cpu/mcheck/mce_intel_64.c
+++ b/arch/x86/kernel/cpu/mcheck/mce_intel_64.c
@@ -249,7 +249,7 @@ void cmci_rediscover(int dying)
249 for_each_online_cpu (cpu) { 249 for_each_online_cpu (cpu) {
250 if (cpu == dying) 250 if (cpu == dying)
251 continue; 251 continue;
252 if (set_cpus_allowed_ptr(current, &cpumask_of_cpu(cpu))) 252 if (set_cpus_allowed_ptr(current, cpumask_of(cpu)))
253 continue; 253 continue;
254 /* Recheck banks in case CPUs don't all have the same */ 254 /* Recheck banks in case CPUs don't all have the same */
255 if (cmci_supported(&banks)) 255 if (cmci_supported(&banks))
diff --git a/arch/x86/kernel/cpu/mtrr/generic.c b/arch/x86/kernel/cpu/mtrr/generic.c
index 37f28fc7cf95..0b776c09aff3 100644
--- a/arch/x86/kernel/cpu/mtrr/generic.c
+++ b/arch/x86/kernel/cpu/mtrr/generic.c
@@ -462,9 +462,6 @@ static void generic_get_mtrr(unsigned int reg, unsigned long *base,
462 *base = base_hi << (32 - PAGE_SHIFT) | base_lo >> PAGE_SHIFT; 462 *base = base_hi << (32 - PAGE_SHIFT) | base_lo >> PAGE_SHIFT;
463 *type = base_lo & 0xff; 463 *type = base_lo & 0xff;
464 464
465 printk(KERN_DEBUG " get_mtrr: cpu%d reg%02d base=%010lx size=%010lx %s\n",
466 cpu, reg, *base, *size,
467 mtrr_attrib_to_str(*type & 0xff));
468out_put_cpu: 465out_put_cpu:
469 put_cpu(); 466 put_cpu();
470} 467}
diff --git a/arch/x86/kernel/cpu/proc.c b/arch/x86/kernel/cpu/proc.c
index d67e0e48bc2d..f93047fed791 100644
--- a/arch/x86/kernel/cpu/proc.c
+++ b/arch/x86/kernel/cpu/proc.c
@@ -14,7 +14,7 @@ static void show_cpuinfo_core(struct seq_file *m, struct cpuinfo_x86 *c,
14 if (c->x86_max_cores * smp_num_siblings > 1) { 14 if (c->x86_max_cores * smp_num_siblings > 1) {
15 seq_printf(m, "physical id\t: %d\n", c->phys_proc_id); 15 seq_printf(m, "physical id\t: %d\n", c->phys_proc_id);
16 seq_printf(m, "siblings\t: %d\n", 16 seq_printf(m, "siblings\t: %d\n",
17 cpus_weight(per_cpu(cpu_core_map, cpu))); 17 cpumask_weight(cpu_sibling_mask(cpu)));
18 seq_printf(m, "core id\t\t: %d\n", c->cpu_core_id); 18 seq_printf(m, "core id\t\t: %d\n", c->cpu_core_id);
19 seq_printf(m, "cpu cores\t: %d\n", c->booted_cores); 19 seq_printf(m, "cpu cores\t: %d\n", c->booted_cores);
20 seq_printf(m, "apicid\t\t: %d\n", c->apicid); 20 seq_printf(m, "apicid\t\t: %d\n", c->apicid);
@@ -143,9 +143,9 @@ static int show_cpuinfo(struct seq_file *m, void *v)
143static void *c_start(struct seq_file *m, loff_t *pos) 143static void *c_start(struct seq_file *m, loff_t *pos)
144{ 144{
145 if (*pos == 0) /* just in case, cpu 0 is not the first */ 145 if (*pos == 0) /* just in case, cpu 0 is not the first */
146 *pos = first_cpu(cpu_online_map); 146 *pos = cpumask_first(cpu_online_mask);
147 else 147 else
148 *pos = next_cpu_nr(*pos - 1, cpu_online_map); 148 *pos = cpumask_next(*pos - 1, cpu_online_mask);
149 if ((*pos) < nr_cpu_ids) 149 if ((*pos) < nr_cpu_ids)
150 return &cpu_data(*pos); 150 return &cpu_data(*pos);
151 return NULL; 151 return NULL;
diff --git a/arch/x86/kernel/dumpstack.c b/arch/x86/kernel/dumpstack.c
index dd2130b0fb3e..95ea5fa7d444 100644
--- a/arch/x86/kernel/dumpstack.c
+++ b/arch/x86/kernel/dumpstack.c
@@ -15,6 +15,7 @@
15#include <linux/bug.h> 15#include <linux/bug.h>
16#include <linux/nmi.h> 16#include <linux/nmi.h>
17#include <linux/sysfs.h> 17#include <linux/sysfs.h>
18#include <linux/ftrace.h>
18 19
19#include <asm/stacktrace.h> 20#include <asm/stacktrace.h>
20 21
@@ -196,6 +197,11 @@ unsigned __kprobes long oops_begin(void)
196 int cpu; 197 int cpu;
197 unsigned long flags; 198 unsigned long flags;
198 199
200 /* notify the hw-branch tracer so it may disable tracing and
201 add the last trace to the trace buffer -
202 the earlier this happens, the more useful the trace. */
203 trace_hw_branch_oops();
204
199 oops_enter(); 205 oops_enter();
200 206
201 /* racy, but better than risking deadlock. */ 207 /* racy, but better than risking deadlock. */
diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c
index 76f7141e0f91..18dfa30795c9 100644
--- a/arch/x86/kernel/ftrace.c
+++ b/arch/x86/kernel/ftrace.c
@@ -18,14 +18,28 @@
18#include <linux/init.h> 18#include <linux/init.h>
19#include <linux/list.h> 19#include <linux/list.h>
20 20
21#include <trace/syscall.h>
22
23#include <asm/cacheflush.h>
21#include <asm/ftrace.h> 24#include <asm/ftrace.h>
22#include <linux/ftrace.h>
23#include <asm/nops.h> 25#include <asm/nops.h>
24#include <asm/nmi.h> 26#include <asm/nmi.h>
25 27
26 28
27#ifdef CONFIG_DYNAMIC_FTRACE 29#ifdef CONFIG_DYNAMIC_FTRACE
28 30
31int ftrace_arch_code_modify_prepare(void)
32{
33 set_kernel_text_rw();
34 return 0;
35}
36
37int ftrace_arch_code_modify_post_process(void)
38{
39 set_kernel_text_ro();
40 return 0;
41}
42
29union ftrace_code_union { 43union ftrace_code_union {
30 char code[MCOUNT_INSN_SIZE]; 44 char code[MCOUNT_INSN_SIZE];
31 struct { 45 struct {
@@ -66,11 +80,11 @@ static unsigned char *ftrace_call_replace(unsigned long ip, unsigned long addr)
66 * 80 *
67 * 1) Put the instruction pointer into the IP buffer 81 * 1) Put the instruction pointer into the IP buffer
68 * and the new code into the "code" buffer. 82 * and the new code into the "code" buffer.
69 * 2) Set a flag that says we are modifying code 83 * 2) Wait for any running NMIs to finish and set a flag that says
70 * 3) Wait for any running NMIs to finish. 84 * we are modifying code, it is done in an atomic operation.
71 * 4) Write the code 85 * 3) Write the code
72 * 5) clear the flag. 86 * 4) clear the flag.
73 * 6) Wait for any running NMIs to finish. 87 * 5) Wait for any running NMIs to finish.
74 * 88 *
75 * If an NMI is executed, the first thing it does is to call 89 * If an NMI is executed, the first thing it does is to call
76 * "ftrace_nmi_enter". This will check if the flag is set to write 90 * "ftrace_nmi_enter". This will check if the flag is set to write
@@ -82,9 +96,9 @@ static unsigned char *ftrace_call_replace(unsigned long ip, unsigned long addr)
82 * are the same as what exists. 96 * are the same as what exists.
83 */ 97 */
84 98
85static atomic_t in_nmi = ATOMIC_INIT(0); 99#define MOD_CODE_WRITE_FLAG (1 << 31) /* set when NMI should do the write */
100static atomic_t nmi_running = ATOMIC_INIT(0);
86static int mod_code_status; /* holds return value of text write */ 101static int mod_code_status; /* holds return value of text write */
87static int mod_code_write; /* set when NMI should do the write */
88static void *mod_code_ip; /* holds the IP to write to */ 102static void *mod_code_ip; /* holds the IP to write to */
89static void *mod_code_newcode; /* holds the text to write to the IP */ 103static void *mod_code_newcode; /* holds the text to write to the IP */
90 104
@@ -101,6 +115,20 @@ int ftrace_arch_read_dyn_info(char *buf, int size)
101 return r; 115 return r;
102} 116}
103 117
118static void clear_mod_flag(void)
119{
120 int old = atomic_read(&nmi_running);
121
122 for (;;) {
123 int new = old & ~MOD_CODE_WRITE_FLAG;
124
125 if (old == new)
126 break;
127
128 old = atomic_cmpxchg(&nmi_running, old, new);
129 }
130}
131
104static void ftrace_mod_code(void) 132static void ftrace_mod_code(void)
105{ 133{
106 /* 134 /*
@@ -111,37 +139,52 @@ static void ftrace_mod_code(void)
111 */ 139 */
112 mod_code_status = probe_kernel_write(mod_code_ip, mod_code_newcode, 140 mod_code_status = probe_kernel_write(mod_code_ip, mod_code_newcode,
113 MCOUNT_INSN_SIZE); 141 MCOUNT_INSN_SIZE);
142
143 /* if we fail, then kill any new writers */
144 if (mod_code_status)
145 clear_mod_flag();
114} 146}
115 147
116void ftrace_nmi_enter(void) 148void ftrace_nmi_enter(void)
117{ 149{
118 atomic_inc(&in_nmi); 150 if (atomic_inc_return(&nmi_running) & MOD_CODE_WRITE_FLAG) {
119 /* Must have in_nmi seen before reading write flag */ 151 smp_rmb();
120 smp_mb();
121 if (mod_code_write) {
122 ftrace_mod_code(); 152 ftrace_mod_code();
123 atomic_inc(&nmi_update_count); 153 atomic_inc(&nmi_update_count);
124 } 154 }
155 /* Must have previous changes seen before executions */
156 smp_mb();
125} 157}
126 158
127void ftrace_nmi_exit(void) 159void ftrace_nmi_exit(void)
128{ 160{
129 /* Finish all executions before clearing in_nmi */ 161 /* Finish all executions before clearing nmi_running */
130 smp_wmb(); 162 smp_mb();
131 atomic_dec(&in_nmi); 163 atomic_dec(&nmi_running);
164}
165
166static void wait_for_nmi_and_set_mod_flag(void)
167{
168 if (!atomic_cmpxchg(&nmi_running, 0, MOD_CODE_WRITE_FLAG))
169 return;
170
171 do {
172 cpu_relax();
173 } while (atomic_cmpxchg(&nmi_running, 0, MOD_CODE_WRITE_FLAG));
174
175 nmi_wait_count++;
132} 176}
133 177
134static void wait_for_nmi(void) 178static void wait_for_nmi(void)
135{ 179{
136 int waited = 0; 180 if (!atomic_read(&nmi_running))
181 return;
137 182
138 while (atomic_read(&in_nmi)) { 183 do {
139 waited = 1;
140 cpu_relax(); 184 cpu_relax();
141 } 185 } while (atomic_read(&nmi_running));
142 186
143 if (waited) 187 nmi_wait_count++;
144 nmi_wait_count++;
145} 188}
146 189
147static int 190static int
@@ -151,14 +194,9 @@ do_ftrace_mod_code(unsigned long ip, void *new_code)
151 mod_code_newcode = new_code; 194 mod_code_newcode = new_code;
152 195
153 /* The buffers need to be visible before we let NMIs write them */ 196 /* The buffers need to be visible before we let NMIs write them */
154 smp_wmb();
155
156 mod_code_write = 1;
157
158 /* Make sure write bit is visible before we wait on NMIs */
159 smp_mb(); 197 smp_mb();
160 198
161 wait_for_nmi(); 199 wait_for_nmi_and_set_mod_flag();
162 200
163 /* Make sure all running NMIs have finished before we write the code */ 201 /* Make sure all running NMIs have finished before we write the code */
164 smp_mb(); 202 smp_mb();
@@ -166,13 +204,9 @@ do_ftrace_mod_code(unsigned long ip, void *new_code)
166 ftrace_mod_code(); 204 ftrace_mod_code();
167 205
168 /* Make sure the write happens before clearing the bit */ 206 /* Make sure the write happens before clearing the bit */
169 smp_wmb();
170
171 mod_code_write = 0;
172
173 /* make sure NMIs see the cleared bit */
174 smp_mb(); 207 smp_mb();
175 208
209 clear_mod_flag();
176 wait_for_nmi(); 210 wait_for_nmi();
177 211
178 return mod_code_status; 212 return mod_code_status;
@@ -368,25 +402,6 @@ int ftrace_disable_ftrace_graph_caller(void)
368 return ftrace_mod_jmp(ip, old_offset, new_offset); 402 return ftrace_mod_jmp(ip, old_offset, new_offset);
369} 403}
370 404
371#else /* CONFIG_DYNAMIC_FTRACE */
372
373/*
374 * These functions are picked from those used on
375 * this page for dynamic ftrace. They have been
376 * simplified to ignore all traces in NMI context.
377 */
378static atomic_t in_nmi;
379
380void ftrace_nmi_enter(void)
381{
382 atomic_inc(&in_nmi);
383}
384
385void ftrace_nmi_exit(void)
386{
387 atomic_dec(&in_nmi);
388}
389
390#endif /* !CONFIG_DYNAMIC_FTRACE */ 405#endif /* !CONFIG_DYNAMIC_FTRACE */
391 406
392/* 407/*
@@ -396,14 +411,13 @@ void ftrace_nmi_exit(void)
396void prepare_ftrace_return(unsigned long *parent, unsigned long self_addr) 411void prepare_ftrace_return(unsigned long *parent, unsigned long self_addr)
397{ 412{
398 unsigned long old; 413 unsigned long old;
399 unsigned long long calltime;
400 int faulted; 414 int faulted;
401 struct ftrace_graph_ent trace; 415 struct ftrace_graph_ent trace;
402 unsigned long return_hooker = (unsigned long) 416 unsigned long return_hooker = (unsigned long)
403 &return_to_handler; 417 &return_to_handler;
404 418
405 /* Nmi's are currently unsupported */ 419 /* Nmi's are currently unsupported */
406 if (unlikely(atomic_read(&in_nmi))) 420 if (unlikely(in_nmi()))
407 return; 421 return;
408 422
409 if (unlikely(atomic_read(&current->tracing_graph_pause))) 423 if (unlikely(atomic_read(&current->tracing_graph_pause)))
@@ -439,17 +453,7 @@ void prepare_ftrace_return(unsigned long *parent, unsigned long self_addr)
439 return; 453 return;
440 } 454 }
441 455
442 if (unlikely(!__kernel_text_address(old))) { 456 if (ftrace_push_return_trace(old, self_addr, &trace.depth) == -EBUSY) {
443 ftrace_graph_stop();
444 *parent = old;
445 WARN_ON(1);
446 return;
447 }
448
449 calltime = cpu_clock(raw_smp_processor_id());
450
451 if (ftrace_push_return_trace(old, calltime,
452 self_addr, &trace.depth) == -EBUSY) {
453 *parent = old; 457 *parent = old;
454 return; 458 return;
455 } 459 }
@@ -463,3 +467,66 @@ void prepare_ftrace_return(unsigned long *parent, unsigned long self_addr)
463 } 467 }
464} 468}
465#endif /* CONFIG_FUNCTION_GRAPH_TRACER */ 469#endif /* CONFIG_FUNCTION_GRAPH_TRACER */
470
471#ifdef CONFIG_FTRACE_SYSCALLS
472
473extern unsigned long __start_syscalls_metadata[];
474extern unsigned long __stop_syscalls_metadata[];
475extern unsigned long *sys_call_table;
476
477static struct syscall_metadata **syscalls_metadata;
478
479static struct syscall_metadata *find_syscall_meta(unsigned long *syscall)
480{
481 struct syscall_metadata *start;
482 struct syscall_metadata *stop;
483 char str[KSYM_SYMBOL_LEN];
484
485
486 start = (struct syscall_metadata *)__start_syscalls_metadata;
487 stop = (struct syscall_metadata *)__stop_syscalls_metadata;
488 kallsyms_lookup((unsigned long) syscall, NULL, NULL, NULL, str);
489
490 for ( ; start < stop; start++) {
491 if (start->name && !strcmp(start->name, str))
492 return start;
493 }
494 return NULL;
495}
496
497struct syscall_metadata *syscall_nr_to_meta(int nr)
498{
499 if (!syscalls_metadata || nr >= FTRACE_SYSCALL_MAX || nr < 0)
500 return NULL;
501
502 return syscalls_metadata[nr];
503}
504
505void arch_init_ftrace_syscalls(void)
506{
507 int i;
508 struct syscall_metadata *meta;
509 unsigned long **psys_syscall_table = &sys_call_table;
510 static atomic_t refs;
511
512 if (atomic_inc_return(&refs) != 1)
513 goto end;
514
515 syscalls_metadata = kzalloc(sizeof(*syscalls_metadata) *
516 FTRACE_SYSCALL_MAX, GFP_KERNEL);
517 if (!syscalls_metadata) {
518 WARN_ON(1);
519 return;
520 }
521
522 for (i = 0; i < FTRACE_SYSCALL_MAX; i++) {
523 meta = find_syscall_meta(psys_syscall_table[i]);
524 syscalls_metadata[i] = meta;
525 }
526 return;
527
528 /* Paranoid: avoid overflow */
529end:
530 atomic_dec(&refs);
531}
532#endif
diff --git a/arch/x86/kernel/irq.c b/arch/x86/kernel/irq.c
index 3aaf7b9e3a8b..c3fe010d74c8 100644
--- a/arch/x86/kernel/irq.c
+++ b/arch/x86/kernel/irq.c
@@ -65,7 +65,7 @@ static int show_other_interrupts(struct seq_file *p, int prec)
65 seq_printf(p, " Spurious interrupts\n"); 65 seq_printf(p, " Spurious interrupts\n");
66#endif 66#endif
67 if (generic_interrupt_extension) { 67 if (generic_interrupt_extension) {
68 seq_printf(p, "PLT: "); 68 seq_printf(p, "%*s: ", prec, "PLT");
69 for_each_online_cpu(j) 69 for_each_online_cpu(j)
70 seq_printf(p, "%10u ", irq_stats(j)->generic_irqs); 70 seq_printf(p, "%10u ", irq_stats(j)->generic_irqs);
71 seq_printf(p, " Platform interrupts\n"); 71 seq_printf(p, " Platform interrupts\n");
diff --git a/arch/x86/kernel/kprobes.c b/arch/x86/kernel/kprobes.c
index 55b94614e348..7b5169d2b000 100644
--- a/arch/x86/kernel/kprobes.c
+++ b/arch/x86/kernel/kprobes.c
@@ -638,13 +638,13 @@ static void __used __kprobes kretprobe_trampoline_holder(void)
638#else 638#else
639 " pushf\n" 639 " pushf\n"
640 /* 640 /*
641 * Skip cs, ip, orig_ax. 641 * Skip cs, ip, orig_ax and gs.
642 * trampoline_handler() will plug in these values 642 * trampoline_handler() will plug in these values
643 */ 643 */
644 " subl $12, %esp\n" 644 " subl $16, %esp\n"
645 " pushl %fs\n" 645 " pushl %fs\n"
646 " pushl %ds\n"
647 " pushl %es\n" 646 " pushl %es\n"
647 " pushl %ds\n"
648 " pushl %eax\n" 648 " pushl %eax\n"
649 " pushl %ebp\n" 649 " pushl %ebp\n"
650 " pushl %edi\n" 650 " pushl %edi\n"
@@ -655,10 +655,10 @@ static void __used __kprobes kretprobe_trampoline_holder(void)
655 " movl %esp, %eax\n" 655 " movl %esp, %eax\n"
656 " call trampoline_handler\n" 656 " call trampoline_handler\n"
657 /* Move flags to cs */ 657 /* Move flags to cs */
658 " movl 52(%esp), %edx\n" 658 " movl 56(%esp), %edx\n"
659 " movl %edx, 48(%esp)\n" 659 " movl %edx, 52(%esp)\n"
660 /* Replace saved flags with true return address. */ 660 /* Replace saved flags with true return address. */
661 " movl %eax, 52(%esp)\n" 661 " movl %eax, 56(%esp)\n"
662 " popl %ebx\n" 662 " popl %ebx\n"
663 " popl %ecx\n" 663 " popl %ecx\n"
664 " popl %edx\n" 664 " popl %edx\n"
@@ -666,8 +666,8 @@ static void __used __kprobes kretprobe_trampoline_holder(void)
666 " popl %edi\n" 666 " popl %edi\n"
667 " popl %ebp\n" 667 " popl %ebp\n"
668 " popl %eax\n" 668 " popl %eax\n"
669 /* Skip ip, orig_ax, es, ds, fs */ 669 /* Skip ds, es, fs, gs, orig_ax and ip */
670 " addl $20, %esp\n" 670 " addl $24, %esp\n"
671 " popf\n" 671 " popf\n"
672#endif 672#endif
673 " ret\n"); 673 " ret\n");
@@ -691,6 +691,7 @@ static __used __kprobes void *trampoline_handler(struct pt_regs *regs)
691 regs->cs = __KERNEL_CS; 691 regs->cs = __KERNEL_CS;
692#else 692#else
693 regs->cs = __KERNEL_CS | get_kernel_rpl(); 693 regs->cs = __KERNEL_CS | get_kernel_rpl();
694 regs->gs = 0;
694#endif 695#endif
695 regs->ip = trampoline_address; 696 regs->ip = trampoline_address;
696 regs->orig_ax = ~0UL; 697 regs->orig_ax = ~0UL;
diff --git a/arch/x86/kernel/microcode_amd.c b/arch/x86/kernel/microcode_amd.c
index c25fdb382292..453b5795a5c6 100644
--- a/arch/x86/kernel/microcode_amd.c
+++ b/arch/x86/kernel/microcode_amd.c
@@ -12,31 +12,30 @@
12 * 12 *
13 * Licensed under the terms of the GNU General Public 13 * Licensed under the terms of the GNU General Public
14 * License version 2. See file COPYING for details. 14 * License version 2. See file COPYING for details.
15*/ 15 */
16 16#include <linux/platform_device.h>
17#include <linux/capability.h> 17#include <linux/capability.h>
18#include <linux/kernel.h>
19#include <linux/init.h>
20#include <linux/sched.h>
21#include <linux/cpumask.h>
22#include <linux/module.h>
23#include <linux/slab.h>
24#include <linux/vmalloc.h>
25#include <linux/miscdevice.h> 18#include <linux/miscdevice.h>
19#include <linux/firmware.h>
26#include <linux/spinlock.h> 20#include <linux/spinlock.h>
27#include <linux/mm.h> 21#include <linux/cpumask.h>
28#include <linux/fs.h> 22#include <linux/pci_ids.h>
23#include <linux/uaccess.h>
24#include <linux/vmalloc.h>
25#include <linux/kernel.h>
26#include <linux/module.h>
29#include <linux/mutex.h> 27#include <linux/mutex.h>
28#include <linux/sched.h>
29#include <linux/init.h>
30#include <linux/slab.h>
30#include <linux/cpu.h> 31#include <linux/cpu.h>
31#include <linux/firmware.h>
32#include <linux/platform_device.h>
33#include <linux/pci.h> 32#include <linux/pci.h>
34#include <linux/pci_ids.h> 33#include <linux/fs.h>
35#include <linux/uaccess.h> 34#include <linux/mm.h>
36 35
37#include <asm/msr.h>
38#include <asm/processor.h>
39#include <asm/microcode.h> 36#include <asm/microcode.h>
37#include <asm/processor.h>
38#include <asm/msr.h>
40 39
41MODULE_DESCRIPTION("AMD Microcode Update Driver"); 40MODULE_DESCRIPTION("AMD Microcode Update Driver");
42MODULE_AUTHOR("Peter Oruba"); 41MODULE_AUTHOR("Peter Oruba");
@@ -72,8 +71,8 @@ struct microcode_header_amd {
72} __attribute__((packed)); 71} __attribute__((packed));
73 72
74struct microcode_amd { 73struct microcode_amd {
75 struct microcode_header_amd hdr; 74 struct microcode_header_amd hdr;
76 unsigned int mpb[0]; 75 unsigned int mpb[0];
77}; 76};
78 77
79#define UCODE_MAX_SIZE 2048 78#define UCODE_MAX_SIZE 2048
@@ -184,8 +183,8 @@ static int get_ucode_data(void *to, const u8 *from, size_t n)
184 return 0; 183 return 0;
185} 184}
186 185
187static void *get_next_ucode(const u8 *buf, unsigned int size, 186static void *
188 unsigned int *mc_size) 187get_next_ucode(const u8 *buf, unsigned int size, unsigned int *mc_size)
189{ 188{
190 unsigned int total_size; 189 unsigned int total_size;
191 u8 section_hdr[UCODE_CONTAINER_SECTION_HDR]; 190 u8 section_hdr[UCODE_CONTAINER_SECTION_HDR];
@@ -223,7 +222,6 @@ static void *get_next_ucode(const u8 *buf, unsigned int size,
223 return mc; 222 return mc;
224} 223}
225 224
226
227static int install_equiv_cpu_table(const u8 *buf) 225static int install_equiv_cpu_table(const u8 *buf)
228{ 226{
229 u8 *container_hdr[UCODE_CONTAINER_HEADER_SIZE]; 227 u8 *container_hdr[UCODE_CONTAINER_HEADER_SIZE];
@@ -372,4 +370,3 @@ struct microcode_ops * __init init_amd_microcode(void)
372{ 370{
373 return &microcode_amd_ops; 371 return &microcode_amd_ops;
374} 372}
375
diff --git a/arch/x86/kernel/microcode_core.c b/arch/x86/kernel/microcode_core.c
index c9b721ba968c..2e0eb4140951 100644
--- a/arch/x86/kernel/microcode_core.c
+++ b/arch/x86/kernel/microcode_core.c
@@ -70,41 +70,41 @@
70 * Fix sigmatch() macro to handle old CPUs with pf == 0. 70 * Fix sigmatch() macro to handle old CPUs with pf == 0.
71 * Thanks to Stuart Swales for pointing out this bug. 71 * Thanks to Stuart Swales for pointing out this bug.
72 */ 72 */
73#include <linux/platform_device.h>
73#include <linux/capability.h> 74#include <linux/capability.h>
74#include <linux/kernel.h> 75#include <linux/miscdevice.h>
75#include <linux/init.h> 76#include <linux/firmware.h>
76#include <linux/sched.h>
77#include <linux/smp_lock.h> 77#include <linux/smp_lock.h>
78#include <linux/spinlock.h>
78#include <linux/cpumask.h> 79#include <linux/cpumask.h>
79#include <linux/module.h> 80#include <linux/uaccess.h>
80#include <linux/slab.h>
81#include <linux/vmalloc.h> 81#include <linux/vmalloc.h>
82#include <linux/miscdevice.h> 82#include <linux/kernel.h>
83#include <linux/spinlock.h> 83#include <linux/module.h>
84#include <linux/mm.h>
85#include <linux/fs.h>
86#include <linux/mutex.h> 84#include <linux/mutex.h>
85#include <linux/sched.h>
86#include <linux/init.h>
87#include <linux/slab.h>
87#include <linux/cpu.h> 88#include <linux/cpu.h>
88#include <linux/firmware.h> 89#include <linux/fs.h>
89#include <linux/platform_device.h> 90#include <linux/mm.h>
90 91
91#include <asm/msr.h>
92#include <asm/uaccess.h>
93#include <asm/processor.h>
94#include <asm/microcode.h> 92#include <asm/microcode.h>
93#include <asm/processor.h>
94#include <asm/msr.h>
95 95
96MODULE_DESCRIPTION("Microcode Update Driver"); 96MODULE_DESCRIPTION("Microcode Update Driver");
97MODULE_AUTHOR("Tigran Aivazian <tigran@aivazian.fsnet.co.uk>"); 97MODULE_AUTHOR("Tigran Aivazian <tigran@aivazian.fsnet.co.uk>");
98MODULE_LICENSE("GPL"); 98MODULE_LICENSE("GPL");
99 99
100#define MICROCODE_VERSION "2.00" 100#define MICROCODE_VERSION "2.00"
101 101
102static struct microcode_ops *microcode_ops; 102static struct microcode_ops *microcode_ops;
103 103
104/* no concurrent ->write()s are allowed on /dev/cpu/microcode */ 104/* no concurrent ->write()s are allowed on /dev/cpu/microcode */
105static DEFINE_MUTEX(microcode_mutex); 105static DEFINE_MUTEX(microcode_mutex);
106 106
107struct ucode_cpu_info ucode_cpu_info[NR_CPUS]; 107struct ucode_cpu_info ucode_cpu_info[NR_CPUS];
108EXPORT_SYMBOL_GPL(ucode_cpu_info); 108EXPORT_SYMBOL_GPL(ucode_cpu_info);
109 109
110#ifdef CONFIG_MICROCODE_OLD_INTERFACE 110#ifdef CONFIG_MICROCODE_OLD_INTERFACE
@@ -198,18 +198,33 @@ static void microcode_dev_exit(void)
198 198
199MODULE_ALIAS_MISCDEV(MICROCODE_MINOR); 199MODULE_ALIAS_MISCDEV(MICROCODE_MINOR);
200#else 200#else
201#define microcode_dev_init() 0 201#define microcode_dev_init() 0
202#define microcode_dev_exit() do { } while (0) 202#define microcode_dev_exit() do { } while (0)
203#endif 203#endif
204 204
205/* fake device for request_firmware */ 205/* fake device for request_firmware */
206static struct platform_device *microcode_pdev; 206static struct platform_device *microcode_pdev;
207
208static long reload_for_cpu(void *unused)
209{
210 struct ucode_cpu_info *uci = ucode_cpu_info + smp_processor_id();
211 int err = 0;
212
213 mutex_lock(&microcode_mutex);
214 if (uci->valid) {
215 err = microcode_ops->request_microcode_fw(smp_processor_id(),
216 &microcode_pdev->dev);
217 if (!err)
218 microcode_ops->apply_microcode(smp_processor_id());
219 }
220 mutex_unlock(&microcode_mutex);
221 return err;
222}
207 223
208static ssize_t reload_store(struct sys_device *dev, 224static ssize_t reload_store(struct sys_device *dev,
209 struct sysdev_attribute *attr, 225 struct sysdev_attribute *attr,
210 const char *buf, size_t sz) 226 const char *buf, size_t sz)
211{ 227{
212 struct ucode_cpu_info *uci = ucode_cpu_info + dev->id;
213 char *end; 228 char *end;
214 unsigned long val = simple_strtoul(buf, &end, 0); 229 unsigned long val = simple_strtoul(buf, &end, 0);
215 int err = 0; 230 int err = 0;
@@ -218,21 +233,9 @@ static ssize_t reload_store(struct sys_device *dev,
218 if (end == buf) 233 if (end == buf)
219 return -EINVAL; 234 return -EINVAL;
220 if (val == 1) { 235 if (val == 1) {
221 cpumask_t old = current->cpus_allowed;
222
223 get_online_cpus(); 236 get_online_cpus();
224 if (cpu_online(cpu)) { 237 if (cpu_online(cpu))
225 set_cpus_allowed_ptr(current, &cpumask_of_cpu(cpu)); 238 err = work_on_cpu(cpu, reload_for_cpu, NULL);
226 mutex_lock(&microcode_mutex);
227 if (uci->valid) {
228 err = microcode_ops->request_microcode_fw(cpu,
229 &microcode_pdev->dev);
230 if (!err)
231 microcode_ops->apply_microcode(cpu);
232 }
233 mutex_unlock(&microcode_mutex);
234 set_cpus_allowed_ptr(current, &old);
235 }
236 put_online_cpus(); 239 put_online_cpus();
237 } 240 }
238 if (err) 241 if (err)
@@ -268,8 +271,8 @@ static struct attribute *mc_default_attrs[] = {
268}; 271};
269 272
270static struct attribute_group mc_attr_group = { 273static struct attribute_group mc_attr_group = {
271 .attrs = mc_default_attrs, 274 .attrs = mc_default_attrs,
272 .name = "microcode", 275 .name = "microcode",
273}; 276};
274 277
275static void __microcode_fini_cpu(int cpu) 278static void __microcode_fini_cpu(int cpu)
@@ -328,9 +331,9 @@ static int microcode_resume_cpu(int cpu)
328 return 0; 331 return 0;
329} 332}
330 333
331static void microcode_update_cpu(int cpu) 334static long microcode_update_cpu(void *unused)
332{ 335{
333 struct ucode_cpu_info *uci = ucode_cpu_info + cpu; 336 struct ucode_cpu_info *uci = ucode_cpu_info + smp_processor_id();
334 int err = 0; 337 int err = 0;
335 338
336 /* 339 /*
@@ -338,30 +341,27 @@ static void microcode_update_cpu(int cpu)
338 * otherwise just request a firmware: 341 * otherwise just request a firmware:
339 */ 342 */
340 if (uci->valid) { 343 if (uci->valid) {
341 err = microcode_resume_cpu(cpu); 344 err = microcode_resume_cpu(smp_processor_id());
342 } else { 345 } else {
343 collect_cpu_info(cpu); 346 collect_cpu_info(smp_processor_id());
344 if (uci->valid && system_state == SYSTEM_RUNNING) 347 if (uci->valid && system_state == SYSTEM_RUNNING)
345 err = microcode_ops->request_microcode_fw(cpu, 348 err = microcode_ops->request_microcode_fw(
349 smp_processor_id(),
346 &microcode_pdev->dev); 350 &microcode_pdev->dev);
347 } 351 }
348 if (!err) 352 if (!err)
349 microcode_ops->apply_microcode(cpu); 353 microcode_ops->apply_microcode(smp_processor_id());
354 return err;
350} 355}
351 356
352static void microcode_init_cpu(int cpu) 357static int microcode_init_cpu(int cpu)
353{ 358{
354 cpumask_t old = current->cpus_allowed; 359 int err;
355
356 set_cpus_allowed_ptr(current, &cpumask_of_cpu(cpu));
357 /* We should bind the task to the CPU */
358 BUG_ON(raw_smp_processor_id() != cpu);
359
360 mutex_lock(&microcode_mutex); 360 mutex_lock(&microcode_mutex);
361 microcode_update_cpu(cpu); 361 err = work_on_cpu(cpu, microcode_update_cpu, NULL);
362 mutex_unlock(&microcode_mutex); 362 mutex_unlock(&microcode_mutex);
363 363
364 set_cpus_allowed_ptr(current, &old); 364 return err;
365} 365}
366 366
367static int mc_sysdev_add(struct sys_device *sys_dev) 367static int mc_sysdev_add(struct sys_device *sys_dev)
@@ -379,8 +379,11 @@ static int mc_sysdev_add(struct sys_device *sys_dev)
379 if (err) 379 if (err)
380 return err; 380 return err;
381 381
382 microcode_init_cpu(cpu); 382 err = microcode_init_cpu(cpu);
383 return 0; 383 if (err)
384 sysfs_remove_group(&sys_dev->kobj, &mc_attr_group);
385
386 return err;
384} 387}
385 388
386static int mc_sysdev_remove(struct sys_device *sys_dev) 389static int mc_sysdev_remove(struct sys_device *sys_dev)
@@ -404,14 +407,14 @@ static int mc_sysdev_resume(struct sys_device *dev)
404 return 0; 407 return 0;
405 408
406 /* only CPU 0 will apply ucode here */ 409 /* only CPU 0 will apply ucode here */
407 microcode_update_cpu(0); 410 microcode_update_cpu(NULL);
408 return 0; 411 return 0;
409} 412}
410 413
411static struct sysdev_driver mc_sysdev_driver = { 414static struct sysdev_driver mc_sysdev_driver = {
412 .add = mc_sysdev_add, 415 .add = mc_sysdev_add,
413 .remove = mc_sysdev_remove, 416 .remove = mc_sysdev_remove,
414 .resume = mc_sysdev_resume, 417 .resume = mc_sysdev_resume,
415}; 418};
416 419
417static __cpuinit int 420static __cpuinit int
@@ -424,7 +427,9 @@ mc_cpu_callback(struct notifier_block *nb, unsigned long action, void *hcpu)
424 switch (action) { 427 switch (action) {
425 case CPU_ONLINE: 428 case CPU_ONLINE:
426 case CPU_ONLINE_FROZEN: 429 case CPU_ONLINE_FROZEN:
427 microcode_init_cpu(cpu); 430 if (microcode_init_cpu(cpu))
431 printk(KERN_ERR "microcode: failed to init CPU%d\n",
432 cpu);
428 case CPU_DOWN_FAILED: 433 case CPU_DOWN_FAILED:
429 case CPU_DOWN_FAILED_FROZEN: 434 case CPU_DOWN_FAILED_FROZEN:
430 pr_debug("microcode: CPU%d added\n", cpu); 435 pr_debug("microcode: CPU%d added\n", cpu);
@@ -448,7 +453,7 @@ mc_cpu_callback(struct notifier_block *nb, unsigned long action, void *hcpu)
448} 453}
449 454
450static struct notifier_block __refdata mc_cpu_notifier = { 455static struct notifier_block __refdata mc_cpu_notifier = {
451 .notifier_call = mc_cpu_callback, 456 .notifier_call = mc_cpu_callback,
452}; 457};
453 458
454static int __init microcode_init(void) 459static int __init microcode_init(void)
diff --git a/arch/x86/kernel/microcode_intel.c b/arch/x86/kernel/microcode_intel.c
index 5e9f4fc51385..149b9ec7c1ab 100644
--- a/arch/x86/kernel/microcode_intel.c
+++ b/arch/x86/kernel/microcode_intel.c
@@ -70,28 +70,28 @@
70 * Fix sigmatch() macro to handle old CPUs with pf == 0. 70 * Fix sigmatch() macro to handle old CPUs with pf == 0.
71 * Thanks to Stuart Swales for pointing out this bug. 71 * Thanks to Stuart Swales for pointing out this bug.
72 */ 72 */
73#include <linux/platform_device.h>
73#include <linux/capability.h> 74#include <linux/capability.h>
74#include <linux/kernel.h> 75#include <linux/miscdevice.h>
75#include <linux/init.h> 76#include <linux/firmware.h>
76#include <linux/sched.h>
77#include <linux/smp_lock.h> 77#include <linux/smp_lock.h>
78#include <linux/spinlock.h>
78#include <linux/cpumask.h> 79#include <linux/cpumask.h>
79#include <linux/module.h> 80#include <linux/uaccess.h>
80#include <linux/slab.h>
81#include <linux/vmalloc.h> 81#include <linux/vmalloc.h>
82#include <linux/miscdevice.h> 82#include <linux/kernel.h>
83#include <linux/spinlock.h> 83#include <linux/module.h>
84#include <linux/mm.h>
85#include <linux/fs.h>
86#include <linux/mutex.h> 84#include <linux/mutex.h>
85#include <linux/sched.h>
86#include <linux/init.h>
87#include <linux/slab.h>
87#include <linux/cpu.h> 88#include <linux/cpu.h>
88#include <linux/firmware.h> 89#include <linux/fs.h>
89#include <linux/platform_device.h> 90#include <linux/mm.h>
90#include <linux/uaccess.h>
91 91
92#include <asm/msr.h>
93#include <asm/processor.h>
94#include <asm/microcode.h> 92#include <asm/microcode.h>
93#include <asm/processor.h>
94#include <asm/msr.h>
95 95
96MODULE_DESCRIPTION("Microcode Update Driver"); 96MODULE_DESCRIPTION("Microcode Update Driver");
97MODULE_AUTHOR("Tigran Aivazian <tigran@aivazian.fsnet.co.uk>"); 97MODULE_AUTHOR("Tigran Aivazian <tigran@aivazian.fsnet.co.uk>");
@@ -129,12 +129,13 @@ struct extended_sigtable {
129 struct extended_signature sigs[0]; 129 struct extended_signature sigs[0];
130}; 130};
131 131
132#define DEFAULT_UCODE_DATASIZE (2000) 132#define DEFAULT_UCODE_DATASIZE (2000)
133#define MC_HEADER_SIZE (sizeof(struct microcode_header_intel)) 133#define MC_HEADER_SIZE (sizeof(struct microcode_header_intel))
134#define DEFAULT_UCODE_TOTALSIZE (DEFAULT_UCODE_DATASIZE + MC_HEADER_SIZE) 134#define DEFAULT_UCODE_TOTALSIZE (DEFAULT_UCODE_DATASIZE + MC_HEADER_SIZE)
135#define EXT_HEADER_SIZE (sizeof(struct extended_sigtable)) 135#define EXT_HEADER_SIZE (sizeof(struct extended_sigtable))
136#define EXT_SIGNATURE_SIZE (sizeof(struct extended_signature)) 136#define EXT_SIGNATURE_SIZE (sizeof(struct extended_signature))
137#define DWSIZE (sizeof(u32)) 137#define DWSIZE (sizeof(u32))
138
138#define get_totalsize(mc) \ 139#define get_totalsize(mc) \
139 (((struct microcode_intel *)mc)->hdr.totalsize ? \ 140 (((struct microcode_intel *)mc)->hdr.totalsize ? \
140 ((struct microcode_intel *)mc)->hdr.totalsize : \ 141 ((struct microcode_intel *)mc)->hdr.totalsize : \
@@ -197,30 +198,31 @@ static inline int update_match_cpu(struct cpu_signature *csig, int sig, int pf)
197} 198}
198 199
199static inline int 200static inline int
200update_match_revision(struct microcode_header_intel *mc_header, int rev) 201update_match_revision(struct microcode_header_intel *mc_header, int rev)
201{ 202{
202 return (mc_header->rev <= rev) ? 0 : 1; 203 return (mc_header->rev <= rev) ? 0 : 1;
203} 204}
204 205
205static int microcode_sanity_check(void *mc) 206static int microcode_sanity_check(void *mc)
206{ 207{
208 unsigned long total_size, data_size, ext_table_size;
207 struct microcode_header_intel *mc_header = mc; 209 struct microcode_header_intel *mc_header = mc;
208 struct extended_sigtable *ext_header = NULL; 210 struct extended_sigtable *ext_header = NULL;
209 struct extended_signature *ext_sig;
210 unsigned long total_size, data_size, ext_table_size;
211 int sum, orig_sum, ext_sigcount = 0, i; 211 int sum, orig_sum, ext_sigcount = 0, i;
212 struct extended_signature *ext_sig;
212 213
213 total_size = get_totalsize(mc_header); 214 total_size = get_totalsize(mc_header);
214 data_size = get_datasize(mc_header); 215 data_size = get_datasize(mc_header);
216
215 if (data_size + MC_HEADER_SIZE > total_size) { 217 if (data_size + MC_HEADER_SIZE > total_size) {
216 printk(KERN_ERR "microcode: error! " 218 printk(KERN_ERR "microcode: error! "
217 "Bad data size in microcode data file\n"); 219 "Bad data size in microcode data file\n");
218 return -EINVAL; 220 return -EINVAL;
219 } 221 }
220 222
221 if (mc_header->ldrver != 1 || mc_header->hdrver != 1) { 223 if (mc_header->ldrver != 1 || mc_header->hdrver != 1) {
222 printk(KERN_ERR "microcode: error! " 224 printk(KERN_ERR "microcode: error! "
223 "Unknown microcode update format\n"); 225 "Unknown microcode update format\n");
224 return -EINVAL; 226 return -EINVAL;
225 } 227 }
226 ext_table_size = total_size - (MC_HEADER_SIZE + data_size); 228 ext_table_size = total_size - (MC_HEADER_SIZE + data_size);
@@ -318,11 +320,15 @@ get_matching_microcode(struct cpu_signature *cpu_sig, void *mc, int rev)
318 320
319static void apply_microcode(int cpu) 321static void apply_microcode(int cpu)
320{ 322{
323 struct microcode_intel *mc_intel;
324 struct ucode_cpu_info *uci;
321 unsigned long flags; 325 unsigned long flags;
322 unsigned int val[2]; 326 unsigned int val[2];
323 int cpu_num = raw_smp_processor_id(); 327 int cpu_num;
324 struct ucode_cpu_info *uci = ucode_cpu_info + cpu; 328
325 struct microcode_intel *mc_intel = uci->mc; 329 cpu_num = raw_smp_processor_id();
330 uci = ucode_cpu_info + cpu;
331 mc_intel = uci->mc;
326 332
327 /* We should bind the task to the CPU */ 333 /* We should bind the task to the CPU */
328 BUG_ON(cpu_num != cpu); 334 BUG_ON(cpu_num != cpu);
@@ -348,15 +354,17 @@ static void apply_microcode(int cpu)
348 spin_unlock_irqrestore(&microcode_update_lock, flags); 354 spin_unlock_irqrestore(&microcode_update_lock, flags);
349 if (val[1] != mc_intel->hdr.rev) { 355 if (val[1] != mc_intel->hdr.rev) {
350 printk(KERN_ERR "microcode: CPU%d update from revision " 356 printk(KERN_ERR "microcode: CPU%d update from revision "
351 "0x%x to 0x%x failed\n", cpu_num, uci->cpu_sig.rev, val[1]); 357 "0x%x to 0x%x failed\n",
358 cpu_num, uci->cpu_sig.rev, val[1]);
352 return; 359 return;
353 } 360 }
354 printk(KERN_INFO "microcode: CPU%d updated from revision " 361 printk(KERN_INFO "microcode: CPU%d updated from revision "
355 "0x%x to 0x%x, date = %04x-%02x-%02x \n", 362 "0x%x to 0x%x, date = %04x-%02x-%02x \n",
356 cpu_num, uci->cpu_sig.rev, val[1], 363 cpu_num, uci->cpu_sig.rev, val[1],
357 mc_intel->hdr.date & 0xffff, 364 mc_intel->hdr.date & 0xffff,
358 mc_intel->hdr.date >> 24, 365 mc_intel->hdr.date >> 24,
359 (mc_intel->hdr.date >> 16) & 0xff); 366 (mc_intel->hdr.date >> 16) & 0xff);
367
360 uci->cpu_sig.rev = val[1]; 368 uci->cpu_sig.rev = val[1];
361} 369}
362 370
@@ -404,18 +412,23 @@ static int generic_load_microcode(int cpu, void *data, size_t size,
404 leftover -= mc_size; 412 leftover -= mc_size;
405 } 413 }
406 414
407 if (new_mc) { 415 if (!new_mc)
408 if (!leftover) { 416 goto out;
409 if (uci->mc) 417
410 vfree(uci->mc); 418 if (leftover) {
411 uci->mc = (struct microcode_intel *)new_mc; 419 vfree(new_mc);
412 pr_debug("microcode: CPU%d found a matching microcode update with" 420 goto out;
413 " version 0x%x (current=0x%x)\n",
414 cpu, new_rev, uci->cpu_sig.rev);
415 } else
416 vfree(new_mc);
417 } 421 }
418 422
423 if (uci->mc)
424 vfree(uci->mc);
425 uci->mc = (struct microcode_intel *)new_mc;
426
427 pr_debug("microcode: CPU%d found a matching microcode update with"
428 " version 0x%x (current=0x%x)\n",
429 cpu, new_rev, uci->cpu_sig.rev);
430
431 out:
419 return (int)leftover; 432 return (int)leftover;
420} 433}
421 434
diff --git a/arch/x86/kernel/mpparse.c b/arch/x86/kernel/mpparse.c
index dce99dca6cf8..70fd7e414c15 100644
--- a/arch/x86/kernel/mpparse.c
+++ b/arch/x86/kernel/mpparse.c
@@ -679,7 +679,7 @@ void __init get_smp_config(void)
679 __get_smp_config(0); 679 __get_smp_config(0);
680} 680}
681 681
682static void smp_reserve_bootmem(struct mpf_intel *mpf) 682static void __init smp_reserve_bootmem(struct mpf_intel *mpf)
683{ 683{
684 unsigned long size = get_mpc_size(mpf->physptr); 684 unsigned long size = get_mpc_size(mpf->physptr);
685#ifdef CONFIG_X86_32 685#ifdef CONFIG_X86_32
@@ -838,7 +838,7 @@ static int __init get_MP_intsrc_index(struct mpc_intsrc *m)
838 838
839static struct mpc_intsrc __initdata *m_spare[SPARE_SLOT_NUM]; 839static struct mpc_intsrc __initdata *m_spare[SPARE_SLOT_NUM];
840 840
841static void check_irq_src(struct mpc_intsrc *m, int *nr_m_spare) 841static void __init check_irq_src(struct mpc_intsrc *m, int *nr_m_spare)
842{ 842{
843 int i; 843 int i;
844 844
@@ -866,7 +866,8 @@ static void check_irq_src(struct mpc_intsrc *m, int *nr_m_spare)
866 } 866 }
867} 867}
868#else /* CONFIG_X86_IO_APIC */ 868#else /* CONFIG_X86_IO_APIC */
869static inline void check_irq_src(struct mpc_intsrc *m, int *nr_m_spare) {} 869static
870inline void __init check_irq_src(struct mpc_intsrc *m, int *nr_m_spare) {}
870#endif /* CONFIG_X86_IO_APIC */ 871#endif /* CONFIG_X86_IO_APIC */
871 872
872static int check_slot(unsigned long mpc_new_phys, unsigned long mpc_new_length, 873static int check_slot(unsigned long mpc_new_phys, unsigned long mpc_new_length,
diff --git a/arch/x86/kernel/pci-dma.c b/arch/x86/kernel/pci-dma.c
index 90f5b9ef5def..745579bc8256 100644
--- a/arch/x86/kernel/pci-dma.c
+++ b/arch/x86/kernel/pci-dma.c
@@ -40,7 +40,7 @@ EXPORT_SYMBOL(bad_dma_address);
40 to older i386. */ 40 to older i386. */
41struct device x86_dma_fallback_dev = { 41struct device x86_dma_fallback_dev = {
42 .init_name = "fallback device", 42 .init_name = "fallback device",
43 .coherent_dma_mask = DMA_32BIT_MASK, 43 .coherent_dma_mask = DMA_BIT_MASK(32),
44 .dma_mask = &x86_dma_fallback_dev.coherent_dma_mask, 44 .dma_mask = &x86_dma_fallback_dev.coherent_dma_mask,
45}; 45};
46EXPORT_SYMBOL(x86_dma_fallback_dev); 46EXPORT_SYMBOL(x86_dma_fallback_dev);
@@ -148,7 +148,7 @@ again:
148 if (!is_buffer_dma_capable(dma_mask, addr, size)) { 148 if (!is_buffer_dma_capable(dma_mask, addr, size)) {
149 __free_pages(page, get_order(size)); 149 __free_pages(page, get_order(size));
150 150
151 if (dma_mask < DMA_32BIT_MASK && !(flag & GFP_DMA)) { 151 if (dma_mask < DMA_BIT_MASK(32) && !(flag & GFP_DMA)) {
152 flag = (flag & ~GFP_DMA32) | GFP_DMA; 152 flag = (flag & ~GFP_DMA32) | GFP_DMA;
153 goto again; 153 goto again;
154 } 154 }
@@ -243,7 +243,7 @@ int dma_supported(struct device *dev, u64 mask)
243 /* Copied from i386. Doesn't make much sense, because it will 243 /* Copied from i386. Doesn't make much sense, because it will
244 only work for pci_alloc_coherent. 244 only work for pci_alloc_coherent.
245 The caller just has to use GFP_DMA in this case. */ 245 The caller just has to use GFP_DMA in this case. */
246 if (mask < DMA_24BIT_MASK) 246 if (mask < DMA_BIT_MASK(24))
247 return 0; 247 return 0;
248 248
249 /* Tell the device to use SAC when IOMMU force is on. This 249 /* Tell the device to use SAC when IOMMU force is on. This
@@ -258,7 +258,7 @@ int dma_supported(struct device *dev, u64 mask)
258 SAC for these. Assume all masks <= 40 bits are of this 258 SAC for these. Assume all masks <= 40 bits are of this
259 type. Normally this doesn't make any difference, but gives 259 type. Normally this doesn't make any difference, but gives
260 more gentle handling of IOMMU overflow. */ 260 more gentle handling of IOMMU overflow. */
261 if (iommu_sac_force && (mask >= DMA_40BIT_MASK)) { 261 if (iommu_sac_force && (mask >= DMA_BIT_MASK(40))) {
262 dev_info(dev, "Force SAC with mask %Lx\n", mask); 262 dev_info(dev, "Force SAC with mask %Lx\n", mask);
263 return 0; 263 return 0;
264 } 264 }
diff --git a/arch/x86/kernel/pci-nommu.c b/arch/x86/kernel/pci-nommu.c
index c6d703b39326..71d412a09f30 100644
--- a/arch/x86/kernel/pci-nommu.c
+++ b/arch/x86/kernel/pci-nommu.c
@@ -15,7 +15,7 @@ static int
15check_addr(char *name, struct device *hwdev, dma_addr_t bus, size_t size) 15check_addr(char *name, struct device *hwdev, dma_addr_t bus, size_t size)
16{ 16{
17 if (hwdev && !is_buffer_dma_capable(*hwdev->dma_mask, bus, size)) { 17 if (hwdev && !is_buffer_dma_capable(*hwdev->dma_mask, bus, size)) {
18 if (*hwdev->dma_mask >= DMA_32BIT_MASK) 18 if (*hwdev->dma_mask >= DMA_BIT_MASK(32))
19 printk(KERN_ERR 19 printk(KERN_ERR
20 "nommu_%s: overflow %Lx+%zu of device mask %Lx\n", 20 "nommu_%s: overflow %Lx+%zu of device mask %Lx\n",
21 name, (long long)bus, size, 21 name, (long long)bus, size,
diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c
index 156f87582c6c..ca989158e847 100644
--- a/arch/x86/kernel/process.c
+++ b/arch/x86/kernel/process.c
@@ -8,7 +8,7 @@
8#include <linux/module.h> 8#include <linux/module.h>
9#include <linux/pm.h> 9#include <linux/pm.h>
10#include <linux/clockchips.h> 10#include <linux/clockchips.h>
11#include <linux/ftrace.h> 11#include <trace/power.h>
12#include <asm/system.h> 12#include <asm/system.h>
13#include <asm/apic.h> 13#include <asm/apic.h>
14#include <asm/idle.h> 14#include <asm/idle.h>
@@ -22,6 +22,9 @@ EXPORT_SYMBOL(idle_nomwait);
22 22
23struct kmem_cache *task_xstate_cachep; 23struct kmem_cache *task_xstate_cachep;
24 24
25DEFINE_TRACE(power_start);
26DEFINE_TRACE(power_end);
27
25int arch_dup_task_struct(struct task_struct *dst, struct task_struct *src) 28int arch_dup_task_struct(struct task_struct *dst, struct task_struct *src)
26{ 29{
27 *dst = *src; 30 *dst = *src;
@@ -325,7 +328,7 @@ void stop_this_cpu(void *dummy)
325 /* 328 /*
326 * Remove this CPU: 329 * Remove this CPU:
327 */ 330 */
328 cpu_clear(smp_processor_id(), cpu_online_map); 331 set_cpu_online(smp_processor_id(), false);
329 disable_local_APIC(); 332 disable_local_APIC();
330 333
331 for (;;) { 334 for (;;) {
@@ -475,12 +478,13 @@ static int __cpuinit check_c1e_idle(const struct cpuinfo_x86 *c)
475 return 1; 478 return 1;
476} 479}
477 480
478static cpumask_t c1e_mask = CPU_MASK_NONE; 481static cpumask_var_t c1e_mask;
479static int c1e_detected; 482static int c1e_detected;
480 483
481void c1e_remove_cpu(int cpu) 484void c1e_remove_cpu(int cpu)
482{ 485{
483 cpu_clear(cpu, c1e_mask); 486 if (c1e_mask != NULL)
487 cpumask_clear_cpu(cpu, c1e_mask);
484} 488}
485 489
486/* 490/*
@@ -509,8 +513,8 @@ static void c1e_idle(void)
509 if (c1e_detected) { 513 if (c1e_detected) {
510 int cpu = smp_processor_id(); 514 int cpu = smp_processor_id();
511 515
512 if (!cpu_isset(cpu, c1e_mask)) { 516 if (!cpumask_test_cpu(cpu, c1e_mask)) {
513 cpu_set(cpu, c1e_mask); 517 cpumask_set_cpu(cpu, c1e_mask);
514 /* 518 /*
515 * Force broadcast so ACPI can not interfere. Needs 519 * Force broadcast so ACPI can not interfere. Needs
516 * to run with interrupts enabled as it uses 520 * to run with interrupts enabled as it uses
@@ -562,6 +566,15 @@ void __cpuinit select_idle_routine(const struct cpuinfo_x86 *c)
562 pm_idle = default_idle; 566 pm_idle = default_idle;
563} 567}
564 568
569void __init init_c1e_mask(void)
570{
571 /* If we're using c1e_idle, we need to allocate c1e_mask. */
572 if (pm_idle == c1e_idle) {
573 alloc_cpumask_var(&c1e_mask, GFP_KERNEL);
574 cpumask_clear(c1e_mask);
575 }
576}
577
565static int __init idle_setup(char *str) 578static int __init idle_setup(char *str)
566{ 579{
567 if (!str) 580 if (!str)
diff --git a/arch/x86/kernel/ptrace.c b/arch/x86/kernel/ptrace.c
index b7cc21bc6ae0..23b7c8f017e2 100644
--- a/arch/x86/kernel/ptrace.c
+++ b/arch/x86/kernel/ptrace.c
@@ -34,6 +34,8 @@
34#include <asm/proto.h> 34#include <asm/proto.h>
35#include <asm/ds.h> 35#include <asm/ds.h>
36 36
37#include <trace/syscall.h>
38
37#include "tls.h" 39#include "tls.h"
38 40
39enum x86_regset { 41enum x86_regset {
@@ -1415,6 +1417,9 @@ asmregparm long syscall_trace_enter(struct pt_regs *regs)
1415 tracehook_report_syscall_entry(regs)) 1417 tracehook_report_syscall_entry(regs))
1416 ret = -1L; 1418 ret = -1L;
1417 1419
1420 if (unlikely(test_thread_flag(TIF_SYSCALL_FTRACE)))
1421 ftrace_syscall_enter(regs);
1422
1418 if (unlikely(current->audit_context)) { 1423 if (unlikely(current->audit_context)) {
1419 if (IS_IA32) 1424 if (IS_IA32)
1420 audit_syscall_entry(AUDIT_ARCH_I386, 1425 audit_syscall_entry(AUDIT_ARCH_I386,
@@ -1438,6 +1443,9 @@ asmregparm void syscall_trace_leave(struct pt_regs *regs)
1438 if (unlikely(current->audit_context)) 1443 if (unlikely(current->audit_context))
1439 audit_syscall_exit(AUDITSC_RESULT(regs->ax), regs->ax); 1444 audit_syscall_exit(AUDITSC_RESULT(regs->ax), regs->ax);
1440 1445
1446 if (unlikely(test_thread_flag(TIF_SYSCALL_FTRACE)))
1447 ftrace_syscall_exit(regs);
1448
1441 if (test_thread_flag(TIF_SYSCALL_TRACE)) 1449 if (test_thread_flag(TIF_SYSCALL_TRACE))
1442 tracehook_report_syscall_exit(regs, 0); 1450 tracehook_report_syscall_exit(regs, 0);
1443 1451
diff --git a/arch/x86/kernel/reboot.c b/arch/x86/kernel/reboot.c
index 2aef36d8aca2..1340dad417f4 100644
--- a/arch/x86/kernel/reboot.c
+++ b/arch/x86/kernel/reboot.c
@@ -224,6 +224,14 @@ static struct dmi_system_id __initdata reboot_dmi_table[] = {
224 DMI_MATCH(DMI_PRODUCT_NAME, "Dell XPS710"), 224 DMI_MATCH(DMI_PRODUCT_NAME, "Dell XPS710"),
225 }, 225 },
226 }, 226 },
227 { /* Handle problems with rebooting on Dell DXP061 */
228 .callback = set_bios_reboot,
229 .ident = "Dell DXP061",
230 .matches = {
231 DMI_MATCH(DMI_SYS_VENDOR, "Dell Inc."),
232 DMI_MATCH(DMI_PRODUCT_NAME, "Dell DXP061"),
233 },
234 },
227 { } 235 { }
228}; 236};
229 237
diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c
index ef7d10170c30..58d24ef917d8 100644
--- a/arch/x86/kernel/smpboot.c
+++ b/arch/x86/kernel/smpboot.c
@@ -101,11 +101,11 @@ EXPORT_SYMBOL(smp_num_siblings);
101DEFINE_PER_CPU(u16, cpu_llc_id) = BAD_APICID; 101DEFINE_PER_CPU(u16, cpu_llc_id) = BAD_APICID;
102 102
103/* representing HT siblings of each logical CPU */ 103/* representing HT siblings of each logical CPU */
104DEFINE_PER_CPU(cpumask_t, cpu_sibling_map); 104DEFINE_PER_CPU(cpumask_var_t, cpu_sibling_map);
105EXPORT_PER_CPU_SYMBOL(cpu_sibling_map); 105EXPORT_PER_CPU_SYMBOL(cpu_sibling_map);
106 106
107/* representing HT and core siblings of each logical CPU */ 107/* representing HT and core siblings of each logical CPU */
108DEFINE_PER_CPU(cpumask_t, cpu_core_map); 108DEFINE_PER_CPU(cpumask_var_t, cpu_core_map);
109EXPORT_PER_CPU_SYMBOL(cpu_core_map); 109EXPORT_PER_CPU_SYMBOL(cpu_core_map);
110 110
111/* Per CPU bogomips and other parameters */ 111/* Per CPU bogomips and other parameters */
@@ -115,11 +115,6 @@ EXPORT_PER_CPU_SYMBOL(cpu_info);
115atomic_t init_deasserted; 115atomic_t init_deasserted;
116 116
117#if defined(CONFIG_NUMA) && defined(CONFIG_X86_32) 117#if defined(CONFIG_NUMA) && defined(CONFIG_X86_32)
118
119/* which logical CPUs are on which nodes */
120cpumask_t node_to_cpumask_map[MAX_NUMNODES] __read_mostly =
121 { [0 ... MAX_NUMNODES-1] = CPU_MASK_NONE };
122EXPORT_SYMBOL(node_to_cpumask_map);
123/* which node each logical CPU is on */ 118/* which node each logical CPU is on */
124int cpu_to_node_map[NR_CPUS] __read_mostly = { [0 ... NR_CPUS-1] = 0 }; 119int cpu_to_node_map[NR_CPUS] __read_mostly = { [0 ... NR_CPUS-1] = 0 };
125EXPORT_SYMBOL(cpu_to_node_map); 120EXPORT_SYMBOL(cpu_to_node_map);
@@ -128,7 +123,7 @@ EXPORT_SYMBOL(cpu_to_node_map);
128static void map_cpu_to_node(int cpu, int node) 123static void map_cpu_to_node(int cpu, int node)
129{ 124{
130 printk(KERN_INFO "Mapping cpu %d to node %d\n", cpu, node); 125 printk(KERN_INFO "Mapping cpu %d to node %d\n", cpu, node);
131 cpumask_set_cpu(cpu, &node_to_cpumask_map[node]); 126 cpumask_set_cpu(cpu, node_to_cpumask_map[node]);
132 cpu_to_node_map[cpu] = node; 127 cpu_to_node_map[cpu] = node;
133} 128}
134 129
@@ -139,7 +134,7 @@ static void unmap_cpu_to_node(int cpu)
139 134
140 printk(KERN_INFO "Unmapping cpu %d from all nodes\n", cpu); 135 printk(KERN_INFO "Unmapping cpu %d from all nodes\n", cpu);
141 for (node = 0; node < MAX_NUMNODES; node++) 136 for (node = 0; node < MAX_NUMNODES; node++)
142 cpumask_clear_cpu(cpu, &node_to_cpumask_map[node]); 137 cpumask_clear_cpu(cpu, node_to_cpumask_map[node]);
143 cpu_to_node_map[cpu] = 0; 138 cpu_to_node_map[cpu] = 0;
144} 139}
145#else /* !(CONFIG_NUMA && CONFIG_X86_32) */ 140#else /* !(CONFIG_NUMA && CONFIG_X86_32) */
@@ -301,7 +296,7 @@ notrace static void __cpuinit start_secondary(void *unused)
301 __flush_tlb_all(); 296 __flush_tlb_all();
302#endif 297#endif
303 298
304 /* This must be done before setting cpu_online_map */ 299 /* This must be done before setting cpu_online_mask */
305 set_cpu_sibling_map(raw_smp_processor_id()); 300 set_cpu_sibling_map(raw_smp_processor_id());
306 wmb(); 301 wmb();
307 302
@@ -334,6 +329,23 @@ notrace static void __cpuinit start_secondary(void *unused)
334 cpu_idle(); 329 cpu_idle();
335} 330}
336 331
332#ifdef CONFIG_CPUMASK_OFFSTACK
333/* In this case, llc_shared_map is a pointer to a cpumask. */
334static inline void copy_cpuinfo_x86(struct cpuinfo_x86 *dst,
335 const struct cpuinfo_x86 *src)
336{
337 struct cpumask *llc = dst->llc_shared_map;
338 *dst = *src;
339 dst->llc_shared_map = llc;
340}
341#else
342static inline void copy_cpuinfo_x86(struct cpuinfo_x86 *dst,
343 const struct cpuinfo_x86 *src)
344{
345 *dst = *src;
346}
347#endif /* CONFIG_CPUMASK_OFFSTACK */
348
337/* 349/*
338 * The bootstrap kernel entry code has set these up. Save them for 350 * The bootstrap kernel entry code has set these up. Save them for
339 * a given CPU 351 * a given CPU
@@ -343,7 +355,7 @@ void __cpuinit smp_store_cpu_info(int id)
343{ 355{
344 struct cpuinfo_x86 *c = &cpu_data(id); 356 struct cpuinfo_x86 *c = &cpu_data(id);
345 357
346 *c = boot_cpu_data; 358 copy_cpuinfo_x86(c, &boot_cpu_data);
347 c->cpu_index = id; 359 c->cpu_index = id;
348 if (id != 0) 360 if (id != 0)
349 identify_secondary_cpu(c); 361 identify_secondary_cpu(c);
@@ -367,15 +379,15 @@ void __cpuinit set_cpu_sibling_map(int cpu)
367 cpumask_set_cpu(cpu, cpu_sibling_mask(i)); 379 cpumask_set_cpu(cpu, cpu_sibling_mask(i));
368 cpumask_set_cpu(i, cpu_core_mask(cpu)); 380 cpumask_set_cpu(i, cpu_core_mask(cpu));
369 cpumask_set_cpu(cpu, cpu_core_mask(i)); 381 cpumask_set_cpu(cpu, cpu_core_mask(i));
370 cpumask_set_cpu(i, &c->llc_shared_map); 382 cpumask_set_cpu(i, c->llc_shared_map);
371 cpumask_set_cpu(cpu, &o->llc_shared_map); 383 cpumask_set_cpu(cpu, o->llc_shared_map);
372 } 384 }
373 } 385 }
374 } else { 386 } else {
375 cpumask_set_cpu(cpu, cpu_sibling_mask(cpu)); 387 cpumask_set_cpu(cpu, cpu_sibling_mask(cpu));
376 } 388 }
377 389
378 cpumask_set_cpu(cpu, &c->llc_shared_map); 390 cpumask_set_cpu(cpu, c->llc_shared_map);
379 391
380 if (current_cpu_data.x86_max_cores == 1) { 392 if (current_cpu_data.x86_max_cores == 1) {
381 cpumask_copy(cpu_core_mask(cpu), cpu_sibling_mask(cpu)); 393 cpumask_copy(cpu_core_mask(cpu), cpu_sibling_mask(cpu));
@@ -386,8 +398,8 @@ void __cpuinit set_cpu_sibling_map(int cpu)
386 for_each_cpu(i, cpu_sibling_setup_mask) { 398 for_each_cpu(i, cpu_sibling_setup_mask) {
387 if (per_cpu(cpu_llc_id, cpu) != BAD_APICID && 399 if (per_cpu(cpu_llc_id, cpu) != BAD_APICID &&
388 per_cpu(cpu_llc_id, cpu) == per_cpu(cpu_llc_id, i)) { 400 per_cpu(cpu_llc_id, cpu) == per_cpu(cpu_llc_id, i)) {
389 cpumask_set_cpu(i, &c->llc_shared_map); 401 cpumask_set_cpu(i, c->llc_shared_map);
390 cpumask_set_cpu(cpu, &cpu_data(i).llc_shared_map); 402 cpumask_set_cpu(cpu, cpu_data(i).llc_shared_map);
391 } 403 }
392 if (c->phys_proc_id == cpu_data(i).phys_proc_id) { 404 if (c->phys_proc_id == cpu_data(i).phys_proc_id) {
393 cpumask_set_cpu(i, cpu_core_mask(cpu)); 405 cpumask_set_cpu(i, cpu_core_mask(cpu));
@@ -425,12 +437,7 @@ const struct cpumask *cpu_coregroup_mask(int cpu)
425 if (sched_mc_power_savings || sched_smt_power_savings) 437 if (sched_mc_power_savings || sched_smt_power_savings)
426 return cpu_core_mask(cpu); 438 return cpu_core_mask(cpu);
427 else 439 else
428 return &c->llc_shared_map; 440 return c->llc_shared_map;
429}
430
431cpumask_t cpu_coregroup_map(int cpu)
432{
433 return *cpu_coregroup_mask(cpu);
434} 441}
435 442
436static void impress_friends(void) 443static void impress_friends(void)
@@ -897,9 +904,8 @@ int __cpuinit native_cpu_up(unsigned int cpu)
897 */ 904 */
898static __init void disable_smp(void) 905static __init void disable_smp(void)
899{ 906{
900 /* use the read/write pointers to the present and possible maps */ 907 init_cpu_present(cpumask_of(0));
901 cpumask_copy(&cpu_present_map, cpumask_of(0)); 908 init_cpu_possible(cpumask_of(0));
902 cpumask_copy(&cpu_possible_map, cpumask_of(0));
903 smpboot_clear_io_apic_irqs(); 909 smpboot_clear_io_apic_irqs();
904 910
905 if (smp_found_config) 911 if (smp_found_config)
@@ -1031,6 +1037,8 @@ static void __init smp_cpu_index_default(void)
1031 */ 1037 */
1032void __init native_smp_prepare_cpus(unsigned int max_cpus) 1038void __init native_smp_prepare_cpus(unsigned int max_cpus)
1033{ 1039{
1040 unsigned int i;
1041
1034 preempt_disable(); 1042 preempt_disable();
1035 smp_cpu_index_default(); 1043 smp_cpu_index_default();
1036 current_cpu_data = boot_cpu_data; 1044 current_cpu_data = boot_cpu_data;
@@ -1044,6 +1052,14 @@ void __init native_smp_prepare_cpus(unsigned int max_cpus)
1044 boot_cpu_logical_apicid = logical_smp_processor_id(); 1052 boot_cpu_logical_apicid = logical_smp_processor_id();
1045#endif 1053#endif
1046 current_thread_info()->cpu = 0; /* needed? */ 1054 current_thread_info()->cpu = 0; /* needed? */
1055 for_each_possible_cpu(i) {
1056 alloc_cpumask_var(&per_cpu(cpu_sibling_map, i), GFP_KERNEL);
1057 alloc_cpumask_var(&per_cpu(cpu_core_map, i), GFP_KERNEL);
1058 alloc_cpumask_var(&cpu_data(i).llc_shared_map, GFP_KERNEL);
1059 cpumask_clear(per_cpu(cpu_core_map, i));
1060 cpumask_clear(per_cpu(cpu_sibling_map, i));
1061 cpumask_clear(cpu_data(i).llc_shared_map);
1062 }
1047 set_cpu_sibling_map(0); 1063 set_cpu_sibling_map(0);
1048 1064
1049 enable_IR_x2apic(); 1065 enable_IR_x2apic();
@@ -1132,11 +1148,11 @@ early_param("possible_cpus", _setup_possible_cpus);
1132 1148
1133 1149
1134/* 1150/*
1135 * cpu_possible_map should be static, it cannot change as cpu's 1151 * cpu_possible_mask should be static, it cannot change as cpu's
1136 * are onlined, or offlined. The reason is per-cpu data-structures 1152 * are onlined, or offlined. The reason is per-cpu data-structures
1137 * are allocated by some modules at init time, and dont expect to 1153 * are allocated by some modules at init time, and dont expect to
1138 * do this dynamically on cpu arrival/departure. 1154 * do this dynamically on cpu arrival/departure.
1139 * cpu_present_map on the other hand can change dynamically. 1155 * cpu_present_mask on the other hand can change dynamically.
1140 * In case when cpu_hotplug is not compiled, then we resort to current 1156 * In case when cpu_hotplug is not compiled, then we resort to current
1141 * behaviour, which is cpu_possible == cpu_present. 1157 * behaviour, which is cpu_possible == cpu_present.
1142 * - Ashok Raj 1158 * - Ashok Raj
diff --git a/arch/x86/kernel/tlb_uv.c b/arch/x86/kernel/tlb_uv.c
index 78422336ddea..ed0c33761e6d 100644
--- a/arch/x86/kernel/tlb_uv.c
+++ b/arch/x86/kernel/tlb_uv.c
@@ -305,6 +305,8 @@ const struct cpumask *uv_flush_send_and_wait(int cpu, int this_pnode,
305 return NULL; 305 return NULL;
306} 306}
307 307
308static DEFINE_PER_CPU(cpumask_var_t, uv_flush_tlb_mask);
309
308/** 310/**
309 * uv_flush_tlb_others - globally purge translation cache of a virtual 311 * uv_flush_tlb_others - globally purge translation cache of a virtual
310 * address or all TLB's 312 * address or all TLB's
@@ -334,8 +336,7 @@ const struct cpumask *uv_flush_tlb_others(const struct cpumask *cpumask,
334 struct mm_struct *mm, 336 struct mm_struct *mm,
335 unsigned long va, unsigned int cpu) 337 unsigned long va, unsigned int cpu)
336{ 338{
337 static DEFINE_PER_CPU(cpumask_t, flush_tlb_mask); 339 struct cpumask *flush_mask = __get_cpu_var(uv_flush_tlb_mask);
338 struct cpumask *flush_mask = &__get_cpu_var(flush_tlb_mask);
339 int i; 340 int i;
340 int bit; 341 int bit;
341 int pnode; 342 int pnode;
@@ -830,6 +831,10 @@ static int __init uv_bau_init(void)
830 if (!is_uv_system()) 831 if (!is_uv_system())
831 return 0; 832 return 0;
832 833
834 for_each_possible_cpu(cur_cpu)
835 alloc_cpumask_var_node(&per_cpu(uv_flush_tlb_mask, cur_cpu),
836 GFP_KERNEL, cpu_to_node(cur_cpu));
837
833 uv_bau_retry_limit = 1; 838 uv_bau_retry_limit = 1;
834 uv_nshift = uv_hub_info->n_val; 839 uv_nshift = uv_hub_info->n_val;
835 uv_mmask = (1UL << uv_hub_info->n_val) - 1; 840 uv_mmask = (1UL << uv_hub_info->n_val) - 1;
diff --git a/arch/x86/kernel/xsave.c b/arch/x86/kernel/xsave.c
index 2b54fe002e94..0a5b04aa98f1 100644
--- a/arch/x86/kernel/xsave.c
+++ b/arch/x86/kernel/xsave.c
@@ -324,7 +324,7 @@ void __ref xsave_cntxt_init(void)
324 } 324 }
325 325
326 /* 326 /*
327 * for now OS knows only about FP/SSE 327 * Support only the state known to OS.
328 */ 328 */
329 pcntxt_mask = pcntxt_mask & XCNTXT_MASK; 329 pcntxt_mask = pcntxt_mask & XCNTXT_MASK;
330 xsave_init(); 330 xsave_init();
diff --git a/arch/x86/kvm/Kconfig b/arch/x86/kvm/Kconfig
index 0a303c3ed11f..a58504ea78cc 100644
--- a/arch/x86/kvm/Kconfig
+++ b/arch/x86/kvm/Kconfig
@@ -59,7 +59,8 @@ config KVM_AMD
59 59
60config KVM_TRACE 60config KVM_TRACE
61 bool "KVM trace support" 61 bool "KVM trace support"
62 depends on KVM && MARKERS && SYSFS 62 depends on KVM && SYSFS
63 select MARKERS
63 select RELAY 64 select RELAY
64 select DEBUG_FS 65 select DEBUG_FS
65 default n 66 default n
diff --git a/arch/x86/mm/Makefile b/arch/x86/mm/Makefile
index 08537747cb58..fdd30d08ab52 100644
--- a/arch/x86/mm/Makefile
+++ b/arch/x86/mm/Makefile
@@ -14,7 +14,7 @@ obj-$(CONFIG_MMIOTRACE) += mmiotrace.o
14mmiotrace-y := kmmio.o pf_in.o mmio-mod.o 14mmiotrace-y := kmmio.o pf_in.o mmio-mod.o
15obj-$(CONFIG_MMIOTRACE_TEST) += testmmiotrace.o 15obj-$(CONFIG_MMIOTRACE_TEST) += testmmiotrace.o
16 16
17obj-$(CONFIG_NUMA) += numa_$(BITS).o 17obj-$(CONFIG_NUMA) += numa.o numa_$(BITS).o
18obj-$(CONFIG_K8_NUMA) += k8topology_64.o 18obj-$(CONFIG_K8_NUMA) += k8topology_64.o
19obj-$(CONFIG_ACPI_NUMA) += srat_$(BITS).o 19obj-$(CONFIG_ACPI_NUMA) += srat_$(BITS).o
20 20
diff --git a/arch/x86/mm/gup.c b/arch/x86/mm/gup.c
index be54176e9eb2..6340cef6798a 100644
--- a/arch/x86/mm/gup.c
+++ b/arch/x86/mm/gup.c
@@ -219,6 +219,22 @@ static int gup_pud_range(pgd_t pgd, unsigned long addr, unsigned long end,
219 return 1; 219 return 1;
220} 220}
221 221
222/**
223 * get_user_pages_fast() - pin user pages in memory
224 * @start: starting user address
225 * @nr_pages: number of pages from start to pin
226 * @write: whether pages will be written to
227 * @pages: array that receives pointers to the pages pinned.
228 * Should be at least nr_pages long.
229 *
230 * Attempt to pin user pages in memory without taking mm->mmap_sem.
231 * If not successful, it will fall back to taking the lock and
232 * calling get_user_pages().
233 *
234 * Returns number of pages pinned. This may be fewer than the number
235 * requested. If nr_pages is 0 or negative, returns 0. If no pages
236 * were pinned, returns -errno.
237 */
222int get_user_pages_fast(unsigned long start, int nr_pages, int write, 238int get_user_pages_fast(unsigned long start, int nr_pages, int write,
223 struct page **pages) 239 struct page **pages)
224{ 240{
diff --git a/arch/x86/mm/init_32.c b/arch/x86/mm/init_32.c
index db81e9a8556b..749559ed80f5 100644
--- a/arch/x86/mm/init_32.c
+++ b/arch/x86/mm/init_32.c
@@ -1054,17 +1054,47 @@ static noinline int do_test_wp_bit(void)
1054const int rodata_test_data = 0xC3; 1054const int rodata_test_data = 0xC3;
1055EXPORT_SYMBOL_GPL(rodata_test_data); 1055EXPORT_SYMBOL_GPL(rodata_test_data);
1056 1056
1057static int kernel_set_to_readonly;
1058
1059void set_kernel_text_rw(void)
1060{
1061 unsigned long start = PFN_ALIGN(_text);
1062 unsigned long size = PFN_ALIGN(_etext) - start;
1063
1064 if (!kernel_set_to_readonly)
1065 return;
1066
1067 pr_debug("Set kernel text: %lx - %lx for read write\n",
1068 start, start+size);
1069
1070 set_pages_rw(virt_to_page(start), size >> PAGE_SHIFT);
1071}
1072
1073void set_kernel_text_ro(void)
1074{
1075 unsigned long start = PFN_ALIGN(_text);
1076 unsigned long size = PFN_ALIGN(_etext) - start;
1077
1078 if (!kernel_set_to_readonly)
1079 return;
1080
1081 pr_debug("Set kernel text: %lx - %lx for read only\n",
1082 start, start+size);
1083
1084 set_pages_ro(virt_to_page(start), size >> PAGE_SHIFT);
1085}
1086
1057void mark_rodata_ro(void) 1087void mark_rodata_ro(void)
1058{ 1088{
1059 unsigned long start = PFN_ALIGN(_text); 1089 unsigned long start = PFN_ALIGN(_text);
1060 unsigned long size = PFN_ALIGN(_etext) - start; 1090 unsigned long size = PFN_ALIGN(_etext) - start;
1061 1091
1062#ifndef CONFIG_DYNAMIC_FTRACE
1063 /* Dynamic tracing modifies the kernel text section */
1064 set_pages_ro(virt_to_page(start), size >> PAGE_SHIFT); 1092 set_pages_ro(virt_to_page(start), size >> PAGE_SHIFT);
1065 printk(KERN_INFO "Write protecting the kernel text: %luk\n", 1093 printk(KERN_INFO "Write protecting the kernel text: %luk\n",
1066 size >> 10); 1094 size >> 10);
1067 1095
1096 kernel_set_to_readonly = 1;
1097
1068#ifdef CONFIG_CPA_DEBUG 1098#ifdef CONFIG_CPA_DEBUG
1069 printk(KERN_INFO "Testing CPA: Reverting %lx-%lx\n", 1099 printk(KERN_INFO "Testing CPA: Reverting %lx-%lx\n",
1070 start, start+size); 1100 start, start+size);
@@ -1073,7 +1103,6 @@ void mark_rodata_ro(void)
1073 printk(KERN_INFO "Testing CPA: write protecting again\n"); 1103 printk(KERN_INFO "Testing CPA: write protecting again\n");
1074 set_pages_ro(virt_to_page(start), size>>PAGE_SHIFT); 1104 set_pages_ro(virt_to_page(start), size>>PAGE_SHIFT);
1075#endif 1105#endif
1076#endif /* CONFIG_DYNAMIC_FTRACE */
1077 1106
1078 start += size; 1107 start += size;
1079 size = (unsigned long)__end_rodata - start; 1108 size = (unsigned long)__end_rodata - start;
diff --git a/arch/x86/mm/init_64.c b/arch/x86/mm/init_64.c
index 54efa57d1c03..1753e8020df6 100644
--- a/arch/x86/mm/init_64.c
+++ b/arch/x86/mm/init_64.c
@@ -734,21 +734,48 @@ void __init mem_init(void)
734const int rodata_test_data = 0xC3; 734const int rodata_test_data = 0xC3;
735EXPORT_SYMBOL_GPL(rodata_test_data); 735EXPORT_SYMBOL_GPL(rodata_test_data);
736 736
737static int kernel_set_to_readonly;
738
739void set_kernel_text_rw(void)
740{
741 unsigned long start = PFN_ALIGN(_stext);
742 unsigned long end = PFN_ALIGN(__start_rodata);
743
744 if (!kernel_set_to_readonly)
745 return;
746
747 pr_debug("Set kernel text: %lx - %lx for read write\n",
748 start, end);
749
750 set_memory_rw(start, (end - start) >> PAGE_SHIFT);
751}
752
753void set_kernel_text_ro(void)
754{
755 unsigned long start = PFN_ALIGN(_stext);
756 unsigned long end = PFN_ALIGN(__start_rodata);
757
758 if (!kernel_set_to_readonly)
759 return;
760
761 pr_debug("Set kernel text: %lx - %lx for read only\n",
762 start, end);
763
764 set_memory_ro(start, (end - start) >> PAGE_SHIFT);
765}
766
737void mark_rodata_ro(void) 767void mark_rodata_ro(void)
738{ 768{
739 unsigned long start = PFN_ALIGN(_stext), end = PFN_ALIGN(__end_rodata); 769 unsigned long start = PFN_ALIGN(_stext), end = PFN_ALIGN(__end_rodata);
740 unsigned long rodata_start = 770 unsigned long rodata_start =
741 ((unsigned long)__start_rodata + PAGE_SIZE - 1) & PAGE_MASK; 771 ((unsigned long)__start_rodata + PAGE_SIZE - 1) & PAGE_MASK;
742 772
743#ifdef CONFIG_DYNAMIC_FTRACE
744 /* Dynamic tracing modifies the kernel text section */
745 start = rodata_start;
746#endif
747
748 printk(KERN_INFO "Write protecting the kernel read-only data: %luk\n", 773 printk(KERN_INFO "Write protecting the kernel read-only data: %luk\n",
749 (end - start) >> 10); 774 (end - start) >> 10);
750 set_memory_ro(start, (end - start) >> PAGE_SHIFT); 775 set_memory_ro(start, (end - start) >> PAGE_SHIFT);
751 776
777 kernel_set_to_readonly = 1;
778
752 /* 779 /*
753 * The rodata section (but not the kernel text!) should also be 780 * The rodata section (but not the kernel text!) should also be
754 * not-executable. 781 * not-executable.
diff --git a/arch/x86/mm/iomap_32.c b/arch/x86/mm/iomap_32.c
index e331f77348a7..8056545e2d39 100644
--- a/arch/x86/mm/iomap_32.c
+++ b/arch/x86/mm/iomap_32.c
@@ -23,7 +23,7 @@
23 23
24int is_io_mapping_possible(resource_size_t base, unsigned long size) 24int is_io_mapping_possible(resource_size_t base, unsigned long size)
25{ 25{
26#ifndef CONFIG_X86_PAE 26#if !defined(CONFIG_X86_PAE) && defined(CONFIG_PHYS_ADDR_T_64BIT)
27 /* There is no way to map greater than 1 << 32 address without PAE */ 27 /* There is no way to map greater than 1 << 32 address without PAE */
28 if (base + size > 0x100000000ULL) 28 if (base + size > 0x100000000ULL)
29 return 0; 29 return 0;
diff --git a/arch/x86/mm/ioremap.c b/arch/x86/mm/ioremap.c
index 0dfa09d69e80..09daebfdb11c 100644
--- a/arch/x86/mm/ioremap.c
+++ b/arch/x86/mm/ioremap.c
@@ -547,7 +547,7 @@ void __init early_ioremap_reset(void)
547} 547}
548 548
549static void __init __early_set_fixmap(enum fixed_addresses idx, 549static void __init __early_set_fixmap(enum fixed_addresses idx,
550 unsigned long phys, pgprot_t flags) 550 phys_addr_t phys, pgprot_t flags)
551{ 551{
552 unsigned long addr = __fix_to_virt(idx); 552 unsigned long addr = __fix_to_virt(idx);
553 pte_t *pte; 553 pte_t *pte;
@@ -566,7 +566,7 @@ static void __init __early_set_fixmap(enum fixed_addresses idx,
566} 566}
567 567
568static inline void __init early_set_fixmap(enum fixed_addresses idx, 568static inline void __init early_set_fixmap(enum fixed_addresses idx,
569 unsigned long phys, pgprot_t prot) 569 phys_addr_t phys, pgprot_t prot)
570{ 570{
571 if (after_paging_init) 571 if (after_paging_init)
572 __set_fixmap(idx, phys, prot); 572 __set_fixmap(idx, phys, prot);
@@ -607,9 +607,10 @@ static int __init check_early_ioremap_leak(void)
607late_initcall(check_early_ioremap_leak); 607late_initcall(check_early_ioremap_leak);
608 608
609static void __init __iomem * 609static void __init __iomem *
610__early_ioremap(unsigned long phys_addr, unsigned long size, pgprot_t prot) 610__early_ioremap(resource_size_t phys_addr, unsigned long size, pgprot_t prot)
611{ 611{
612 unsigned long offset, last_addr; 612 unsigned long offset;
613 resource_size_t last_addr;
613 unsigned int nrpages; 614 unsigned int nrpages;
614 enum fixed_addresses idx0, idx; 615 enum fixed_addresses idx0, idx;
615 int i, slot; 616 int i, slot;
@@ -625,15 +626,15 @@ __early_ioremap(unsigned long phys_addr, unsigned long size, pgprot_t prot)
625 } 626 }
626 627
627 if (slot < 0) { 628 if (slot < 0) {
628 printk(KERN_INFO "early_iomap(%08lx, %08lx) not found slot\n", 629 printk(KERN_INFO "early_iomap(%08llx, %08lx) not found slot\n",
629 phys_addr, size); 630 (u64)phys_addr, size);
630 WARN_ON(1); 631 WARN_ON(1);
631 return NULL; 632 return NULL;
632 } 633 }
633 634
634 if (early_ioremap_debug) { 635 if (early_ioremap_debug) {
635 printk(KERN_INFO "early_ioremap(%08lx, %08lx) [%d] => ", 636 printk(KERN_INFO "early_ioremap(%08llx, %08lx) [%d] => ",
636 phys_addr, size, slot); 637 (u64)phys_addr, size, slot);
637 dump_stack(); 638 dump_stack();
638 } 639 }
639 640
@@ -680,13 +681,15 @@ __early_ioremap(unsigned long phys_addr, unsigned long size, pgprot_t prot)
680} 681}
681 682
682/* Remap an IO device */ 683/* Remap an IO device */
683void __init __iomem *early_ioremap(unsigned long phys_addr, unsigned long size) 684void __init __iomem *
685early_ioremap(resource_size_t phys_addr, unsigned long size)
684{ 686{
685 return __early_ioremap(phys_addr, size, PAGE_KERNEL_IO); 687 return __early_ioremap(phys_addr, size, PAGE_KERNEL_IO);
686} 688}
687 689
688/* Remap memory */ 690/* Remap memory */
689void __init __iomem *early_memremap(unsigned long phys_addr, unsigned long size) 691void __init __iomem *
692early_memremap(resource_size_t phys_addr, unsigned long size)
690{ 693{
691 return __early_ioremap(phys_addr, size, PAGE_KERNEL); 694 return __early_ioremap(phys_addr, size, PAGE_KERNEL);
692} 695}
diff --git a/arch/x86/mm/mmio-mod.c b/arch/x86/mm/mmio-mod.c
index 2c4baa88f2cb..c9342ed8b402 100644
--- a/arch/x86/mm/mmio-mod.c
+++ b/arch/x86/mm/mmio-mod.c
@@ -378,27 +378,34 @@ static void clear_trace_list(void)
378} 378}
379 379
380#ifdef CONFIG_HOTPLUG_CPU 380#ifdef CONFIG_HOTPLUG_CPU
381static cpumask_t downed_cpus; 381static cpumask_var_t downed_cpus;
382 382
383static void enter_uniprocessor(void) 383static void enter_uniprocessor(void)
384{ 384{
385 int cpu; 385 int cpu;
386 int err; 386 int err;
387 387
388 if (downed_cpus == NULL &&
389 !alloc_cpumask_var(&downed_cpus, GFP_KERNEL)) {
390 pr_notice(NAME "Failed to allocate mask\n");
391 goto out;
392 }
393
388 get_online_cpus(); 394 get_online_cpus();
389 downed_cpus = cpu_online_map; 395 cpumask_copy(downed_cpus, cpu_online_mask);
390 cpu_clear(first_cpu(cpu_online_map), downed_cpus); 396 cpumask_clear_cpu(cpumask_first(cpu_online_mask), downed_cpus);
391 if (num_online_cpus() > 1) 397 if (num_online_cpus() > 1)
392 pr_notice(NAME "Disabling non-boot CPUs...\n"); 398 pr_notice(NAME "Disabling non-boot CPUs...\n");
393 put_online_cpus(); 399 put_online_cpus();
394 400
395 for_each_cpu_mask(cpu, downed_cpus) { 401 for_each_cpu(cpu, downed_cpus) {
396 err = cpu_down(cpu); 402 err = cpu_down(cpu);
397 if (!err) 403 if (!err)
398 pr_info(NAME "CPU%d is down.\n", cpu); 404 pr_info(NAME "CPU%d is down.\n", cpu);
399 else 405 else
400 pr_err(NAME "Error taking CPU%d down: %d\n", cpu, err); 406 pr_err(NAME "Error taking CPU%d down: %d\n", cpu, err);
401 } 407 }
408out:
402 if (num_online_cpus() > 1) 409 if (num_online_cpus() > 1)
403 pr_warning(NAME "multiple CPUs still online, " 410 pr_warning(NAME "multiple CPUs still online, "
404 "may miss events.\n"); 411 "may miss events.\n");
@@ -411,10 +418,10 @@ static void __ref leave_uniprocessor(void)
411 int cpu; 418 int cpu;
412 int err; 419 int err;
413 420
414 if (cpus_weight(downed_cpus) == 0) 421 if (downed_cpus == NULL || cpumask_weight(downed_cpus) == 0)
415 return; 422 return;
416 pr_notice(NAME "Re-enabling CPUs...\n"); 423 pr_notice(NAME "Re-enabling CPUs...\n");
417 for_each_cpu_mask(cpu, downed_cpus) { 424 for_each_cpu(cpu, downed_cpus) {
418 err = cpu_up(cpu); 425 err = cpu_up(cpu);
419 if (!err) 426 if (!err)
420 pr_info(NAME "enabled CPU%d.\n", cpu); 427 pr_info(NAME "enabled CPU%d.\n", cpu);
diff --git a/arch/x86/mm/numa.c b/arch/x86/mm/numa.c
new file mode 100644
index 000000000000..550df481accd
--- /dev/null
+++ b/arch/x86/mm/numa.c
@@ -0,0 +1,67 @@
1/* Common code for 32 and 64-bit NUMA */
2#include <linux/topology.h>
3#include <linux/module.h>
4#include <linux/bootmem.h>
5
6#ifdef CONFIG_DEBUG_PER_CPU_MAPS
7# define DBG(x...) printk(KERN_DEBUG x)
8#else
9# define DBG(x...)
10#endif
11
12/*
13 * Which logical CPUs are on which nodes
14 */
15cpumask_var_t node_to_cpumask_map[MAX_NUMNODES];
16EXPORT_SYMBOL(node_to_cpumask_map);
17
18/*
19 * Allocate node_to_cpumask_map based on number of available nodes
20 * Requires node_possible_map to be valid.
21 *
22 * Note: node_to_cpumask() is not valid until after this is done.
23 * (Use CONFIG_DEBUG_PER_CPU_MAPS to check this.)
24 */
25void __init setup_node_to_cpumask_map(void)
26{
27 unsigned int node, num = 0;
28
29 /* setup nr_node_ids if not done yet */
30 if (nr_node_ids == MAX_NUMNODES) {
31 for_each_node_mask(node, node_possible_map)
32 num = node;
33 nr_node_ids = num + 1;
34 }
35
36 /* allocate the map */
37 for (node = 0; node < nr_node_ids; node++)
38 alloc_bootmem_cpumask_var(&node_to_cpumask_map[node]);
39
40 /* cpumask_of_node() will now work */
41 pr_debug("Node to cpumask map for %d nodes\n", nr_node_ids);
42}
43
44#ifdef CONFIG_DEBUG_PER_CPU_MAPS
45/*
46 * Returns a pointer to the bitmask of CPUs on Node 'node'.
47 */
48const struct cpumask *cpumask_of_node(int node)
49{
50 if (node >= nr_node_ids) {
51 printk(KERN_WARNING
52 "cpumask_of_node(%d): node > nr_node_ids(%d)\n",
53 node, nr_node_ids);
54 dump_stack();
55 return cpu_none_mask;
56 }
57 if (node_to_cpumask_map[node] == NULL) {
58 printk(KERN_WARNING
59 "cpumask_of_node(%d): no node_to_cpumask_map!\n",
60 node);
61 dump_stack();
62 return cpu_online_mask;
63 }
64 return node_to_cpumask_map[node];
65}
66EXPORT_SYMBOL(cpumask_of_node);
67#endif
diff --git a/arch/x86/mm/numa_64.c b/arch/x86/mm/numa_64.c
index 64c9cf043cdd..d73aaa892371 100644
--- a/arch/x86/mm/numa_64.c
+++ b/arch/x86/mm/numa_64.c
@@ -20,12 +20,6 @@
20#include <asm/acpi.h> 20#include <asm/acpi.h>
21#include <asm/k8.h> 21#include <asm/k8.h>
22 22
23#ifdef CONFIG_DEBUG_PER_CPU_MAPS
24# define DBG(x...) printk(KERN_DEBUG x)
25#else
26# define DBG(x...)
27#endif
28
29struct pglist_data *node_data[MAX_NUMNODES] __read_mostly; 23struct pglist_data *node_data[MAX_NUMNODES] __read_mostly;
30EXPORT_SYMBOL(node_data); 24EXPORT_SYMBOL(node_data);
31 25
@@ -49,12 +43,6 @@ DEFINE_EARLY_PER_CPU(int, x86_cpu_to_node_map, NUMA_NO_NODE);
49EXPORT_EARLY_PER_CPU_SYMBOL(x86_cpu_to_node_map); 43EXPORT_EARLY_PER_CPU_SYMBOL(x86_cpu_to_node_map);
50 44
51/* 45/*
52 * Which logical CPUs are on which nodes
53 */
54cpumask_t *node_to_cpumask_map;
55EXPORT_SYMBOL(node_to_cpumask_map);
56
57/*
58 * Given a shift value, try to populate memnodemap[] 46 * Given a shift value, try to populate memnodemap[]
59 * Returns : 47 * Returns :
60 * 1 if OK 48 * 1 if OK
@@ -661,36 +649,6 @@ void __init init_cpu_to_node(void)
661#endif 649#endif
662 650
663 651
664/*
665 * Allocate node_to_cpumask_map based on number of available nodes
666 * Requires node_possible_map to be valid.
667 *
668 * Note: node_to_cpumask() is not valid until after this is done.
669 * (Use CONFIG_DEBUG_PER_CPU_MAPS to check this.)
670 */
671void __init setup_node_to_cpumask_map(void)
672{
673 unsigned int node, num = 0;
674 cpumask_t *map;
675
676 /* setup nr_node_ids if not done yet */
677 if (nr_node_ids == MAX_NUMNODES) {
678 for_each_node_mask(node, node_possible_map)
679 num = node;
680 nr_node_ids = num + 1;
681 }
682
683 /* allocate the map */
684 map = alloc_bootmem_low(nr_node_ids * sizeof(cpumask_t));
685 DBG("node_to_cpumask_map at %p for %d nodes\n", map, nr_node_ids);
686
687 pr_debug("Node to cpumask map at %p for %d nodes\n",
688 map, nr_node_ids);
689
690 /* node_to_cpumask() will now work */
691 node_to_cpumask_map = map;
692}
693
694void __cpuinit numa_set_node(int cpu, int node) 652void __cpuinit numa_set_node(int cpu, int node)
695{ 653{
696 int *cpu_to_node_map = early_per_cpu_ptr(x86_cpu_to_node_map); 654 int *cpu_to_node_map = early_per_cpu_ptr(x86_cpu_to_node_map);
@@ -723,12 +681,12 @@ void __cpuinit numa_clear_node(int cpu)
723 681
724void __cpuinit numa_add_cpu(int cpu) 682void __cpuinit numa_add_cpu(int cpu)
725{ 683{
726 cpu_set(cpu, node_to_cpumask_map[early_cpu_to_node(cpu)]); 684 cpumask_set_cpu(cpu, node_to_cpumask_map[early_cpu_to_node(cpu)]);
727} 685}
728 686
729void __cpuinit numa_remove_cpu(int cpu) 687void __cpuinit numa_remove_cpu(int cpu)
730{ 688{
731 cpu_clear(cpu, node_to_cpumask_map[early_cpu_to_node(cpu)]); 689 cpumask_clear_cpu(cpu, node_to_cpumask_map[early_cpu_to_node(cpu)]);
732} 690}
733 691
734#else /* CONFIG_DEBUG_PER_CPU_MAPS */ 692#else /* CONFIG_DEBUG_PER_CPU_MAPS */
@@ -739,20 +697,20 @@ void __cpuinit numa_remove_cpu(int cpu)
739static void __cpuinit numa_set_cpumask(int cpu, int enable) 697static void __cpuinit numa_set_cpumask(int cpu, int enable)
740{ 698{
741 int node = early_cpu_to_node(cpu); 699 int node = early_cpu_to_node(cpu);
742 cpumask_t *mask; 700 struct cpumask *mask;
743 char buf[64]; 701 char buf[64];
744 702
745 if (node_to_cpumask_map == NULL) { 703 mask = node_to_cpumask_map[node];
746 printk(KERN_ERR "node_to_cpumask_map NULL\n"); 704 if (mask == NULL) {
705 printk(KERN_ERR "node_to_cpumask_map[%i] NULL\n", node);
747 dump_stack(); 706 dump_stack();
748 return; 707 return;
749 } 708 }
750 709
751 mask = &node_to_cpumask_map[node];
752 if (enable) 710 if (enable)
753 cpu_set(cpu, *mask); 711 cpumask_set_cpu(cpu, mask);
754 else 712 else
755 cpu_clear(cpu, *mask); 713 cpumask_clear_cpu(cpu, mask);
756 714
757 cpulist_scnprintf(buf, sizeof(buf), mask); 715 cpulist_scnprintf(buf, sizeof(buf), mask);
758 printk(KERN_DEBUG "%s cpu %d node %d: mask now %s\n", 716 printk(KERN_DEBUG "%s cpu %d node %d: mask now %s\n",
@@ -799,59 +757,6 @@ int early_cpu_to_node(int cpu)
799 return per_cpu(x86_cpu_to_node_map, cpu); 757 return per_cpu(x86_cpu_to_node_map, cpu);
800} 758}
801 759
802
803/* empty cpumask */
804static const cpumask_t cpu_mask_none;
805
806/*
807 * Returns a pointer to the bitmask of CPUs on Node 'node'.
808 */
809const cpumask_t *cpumask_of_node(int node)
810{
811 if (node_to_cpumask_map == NULL) {
812 printk(KERN_WARNING
813 "cpumask_of_node(%d): no node_to_cpumask_map!\n",
814 node);
815 dump_stack();
816 return (const cpumask_t *)&cpu_online_map;
817 }
818 if (node >= nr_node_ids) {
819 printk(KERN_WARNING
820 "cpumask_of_node(%d): node > nr_node_ids(%d)\n",
821 node, nr_node_ids);
822 dump_stack();
823 return &cpu_mask_none;
824 }
825 return &node_to_cpumask_map[node];
826}
827EXPORT_SYMBOL(cpumask_of_node);
828
829/*
830 * Returns a bitmask of CPUs on Node 'node'.
831 *
832 * Side note: this function creates the returned cpumask on the stack
833 * so with a high NR_CPUS count, excessive stack space is used. The
834 * node_to_cpumask_ptr function should be used whenever possible.
835 */
836cpumask_t node_to_cpumask(int node)
837{
838 if (node_to_cpumask_map == NULL) {
839 printk(KERN_WARNING
840 "node_to_cpumask(%d): no node_to_cpumask_map!\n", node);
841 dump_stack();
842 return cpu_online_map;
843 }
844 if (node >= nr_node_ids) {
845 printk(KERN_WARNING
846 "node_to_cpumask(%d): node > nr_node_ids(%d)\n",
847 node, nr_node_ids);
848 dump_stack();
849 return cpu_mask_none;
850 }
851 return node_to_cpumask_map[node];
852}
853EXPORT_SYMBOL(node_to_cpumask);
854
855/* 760/*
856 * --------- end of debug versions of the numa functions --------- 761 * --------- end of debug versions of the numa functions ---------
857 */ 762 */
diff --git a/arch/x86/mm/pat.c b/arch/x86/mm/pat.c
index 640339ee4fb2..c009a241d562 100644
--- a/arch/x86/mm/pat.c
+++ b/arch/x86/mm/pat.c
@@ -31,7 +31,7 @@
31#ifdef CONFIG_X86_PAT 31#ifdef CONFIG_X86_PAT
32int __read_mostly pat_enabled = 1; 32int __read_mostly pat_enabled = 1;
33 33
34void __cpuinit pat_disable(const char *reason) 34static inline void pat_disable(const char *reason)
35{ 35{
36 pat_enabled = 0; 36 pat_enabled = 0;
37 printk(KERN_INFO "%s\n", reason); 37 printk(KERN_INFO "%s\n", reason);
diff --git a/arch/x86/mm/pgtable.c b/arch/x86/mm/pgtable.c
index 5b7c7c8464fe..7aa03a5389f5 100644
--- a/arch/x86/mm/pgtable.c
+++ b/arch/x86/mm/pgtable.c
@@ -345,7 +345,8 @@ void __native_set_fixmap(enum fixed_addresses idx, pte_t pte)
345 fixmaps_set++; 345 fixmaps_set++;
346} 346}
347 347
348void native_set_fixmap(enum fixed_addresses idx, unsigned long phys, pgprot_t flags) 348void native_set_fixmap(enum fixed_addresses idx, phys_addr_t phys,
349 pgprot_t flags)
349{ 350{
350 __native_set_fixmap(idx, pfn_pte(phys >> PAGE_SHIFT, flags)); 351 __native_set_fixmap(idx, pfn_pte(phys >> PAGE_SHIFT, flags));
351} 352}
diff --git a/arch/x86/mm/srat_64.c b/arch/x86/mm/srat_64.c
index 574c8bc95ef0..c7d272b8574c 100644
--- a/arch/x86/mm/srat_64.c
+++ b/arch/x86/mm/srat_64.c
@@ -116,6 +116,36 @@ void __init acpi_numa_slit_init(struct acpi_table_slit *slit)
116 reserve_early(phys, phys + length, "ACPI SLIT"); 116 reserve_early(phys, phys + length, "ACPI SLIT");
117} 117}
118 118
119/* Callback for Proximity Domain -> x2APIC mapping */
120void __init
121acpi_numa_x2apic_affinity_init(struct acpi_srat_x2apic_cpu_affinity *pa)
122{
123 int pxm, node;
124 int apic_id;
125
126 if (srat_disabled())
127 return;
128 if (pa->header.length < sizeof(struct acpi_srat_x2apic_cpu_affinity)) {
129 bad_srat();
130 return;
131 }
132 if ((pa->flags & ACPI_SRAT_CPU_ENABLED) == 0)
133 return;
134 pxm = pa->proximity_domain;
135 node = setup_node(pxm);
136 if (node < 0) {
137 printk(KERN_ERR "SRAT: Too many proximity domains %x\n", pxm);
138 bad_srat();
139 return;
140 }
141
142 apic_id = pa->apic_id;
143 apicid_to_node[apic_id] = node;
144 acpi_numa = 1;
145 printk(KERN_INFO "SRAT: PXM %u -> APIC %u -> Node %u\n",
146 pxm, apic_id, node);
147}
148
119/* Callback for Proximity Domain -> LAPIC mapping */ 149/* Callback for Proximity Domain -> LAPIC mapping */
120void __init 150void __init
121acpi_numa_processor_affinity_init(struct acpi_srat_cpu_affinity *pa) 151acpi_numa_processor_affinity_init(struct acpi_srat_cpu_affinity *pa)
diff --git a/arch/x86/oprofile/op_model_p4.c b/arch/x86/oprofile/op_model_p4.c
index 4c4a51c90bc2..819b131fd752 100644
--- a/arch/x86/oprofile/op_model_p4.c
+++ b/arch/x86/oprofile/op_model_p4.c
@@ -380,7 +380,7 @@ static unsigned int get_stagger(void)
380{ 380{
381#ifdef CONFIG_SMP 381#ifdef CONFIG_SMP
382 int cpu = smp_processor_id(); 382 int cpu = smp_processor_id();
383 return (cpu != first_cpu(per_cpu(cpu_sibling_map, cpu))); 383 return cpu != cpumask_first(__get_cpu_var(cpu_sibling_map));
384#endif 384#endif
385 return 0; 385 return 0;
386} 386}
diff --git a/arch/x86/pci/i386.c b/arch/x86/pci/i386.c
index f234a37bd428..f1817f71e009 100644
--- a/arch/x86/pci/i386.c
+++ b/arch/x86/pci/i386.c
@@ -258,24 +258,7 @@ void pcibios_set_master(struct pci_dev *dev)
258 pci_write_config_byte(dev, PCI_LATENCY_TIMER, lat); 258 pci_write_config_byte(dev, PCI_LATENCY_TIMER, lat);
259} 259}
260 260
261static void pci_unmap_page_range(struct vm_area_struct *vma)
262{
263 u64 addr = (u64)vma->vm_pgoff << PAGE_SHIFT;
264 free_memtype(addr, addr + vma->vm_end - vma->vm_start);
265}
266
267static void pci_track_mmap_page_range(struct vm_area_struct *vma)
268{
269 u64 addr = (u64)vma->vm_pgoff << PAGE_SHIFT;
270 unsigned long flags = pgprot_val(vma->vm_page_prot)
271 & _PAGE_CACHE_MASK;
272
273 reserve_memtype(addr, addr + vma->vm_end - vma->vm_start, flags, NULL);
274}
275
276static struct vm_operations_struct pci_mmap_ops = { 261static struct vm_operations_struct pci_mmap_ops = {
277 .open = pci_track_mmap_page_range,
278 .close = pci_unmap_page_range,
279 .access = generic_access_phys, 262 .access = generic_access_phys,
280}; 263};
281 264
@@ -283,11 +266,6 @@ int pci_mmap_page_range(struct pci_dev *dev, struct vm_area_struct *vma,
283 enum pci_mmap_state mmap_state, int write_combine) 266 enum pci_mmap_state mmap_state, int write_combine)
284{ 267{
285 unsigned long prot; 268 unsigned long prot;
286 u64 addr = vma->vm_pgoff << PAGE_SHIFT;
287 unsigned long len = vma->vm_end - vma->vm_start;
288 unsigned long flags;
289 unsigned long new_flags;
290 int retval;
291 269
292 /* I/O space cannot be accessed via normal processor loads and 270 /* I/O space cannot be accessed via normal processor loads and
293 * stores on this platform. 271 * stores on this platform.
@@ -308,30 +286,6 @@ int pci_mmap_page_range(struct pci_dev *dev, struct vm_area_struct *vma,
308 286
309 vma->vm_page_prot = __pgprot(prot); 287 vma->vm_page_prot = __pgprot(prot);
310 288
311 flags = pgprot_val(vma->vm_page_prot) & _PAGE_CACHE_MASK;
312 retval = reserve_memtype(addr, addr + len, flags, &new_flags);
313 if (retval)
314 return retval;
315
316 if (flags != new_flags) {
317 if (!is_new_memtype_allowed(flags, new_flags)) {
318 free_memtype(addr, addr+len);
319 return -EINVAL;
320 }
321 flags = new_flags;
322 vma->vm_page_prot = __pgprot(
323 (pgprot_val(vma->vm_page_prot) & ~_PAGE_CACHE_MASK) |
324 flags);
325 }
326
327 if (((vma->vm_pgoff < max_low_pfn_mapped) ||
328 (vma->vm_pgoff >= (1UL<<(32 - PAGE_SHIFT)) &&
329 vma->vm_pgoff < max_pfn_mapped)) &&
330 ioremap_change_attr((unsigned long)__va(addr), len, flags)) {
331 free_memtype(addr, addr + len);
332 return -EINVAL;
333 }
334
335 if (io_remap_pfn_range(vma, vma->vm_start, vma->vm_pgoff, 289 if (io_remap_pfn_range(vma, vma->vm_start, vma->vm_pgoff,
336 vma->vm_end - vma->vm_start, 290 vma->vm_end - vma->vm_start,
337 vma->vm_page_prot)) 291 vma->vm_page_prot))
diff --git a/arch/x86/power/Makefile b/arch/x86/power/Makefile
index 9ff4d5b55ad1..58b32db33125 100644
--- a/arch/x86/power/Makefile
+++ b/arch/x86/power/Makefile
@@ -1,2 +1,7 @@
1# __restore_processor_state() restores %gs after S3 resume and so should not
2# itself be stack-protected
3nostackp := $(call cc-option, -fno-stack-protector)
4CFLAGS_cpu_$(BITS).o := $(nostackp)
5
1obj-$(CONFIG_PM_SLEEP) += cpu_$(BITS).o 6obj-$(CONFIG_PM_SLEEP) += cpu_$(BITS).o
2obj-$(CONFIG_HIBERNATION) += hibernate_$(BITS).o hibernate_asm_$(BITS).o 7obj-$(CONFIG_HIBERNATION) += hibernate_$(BITS).o hibernate_asm_$(BITS).o
diff --git a/arch/x86/xen/enlighten.c b/arch/x86/xen/enlighten.c
index 82cd39a6cbd3..f09e8c36ee80 100644
--- a/arch/x86/xen/enlighten.c
+++ b/arch/x86/xen/enlighten.c
@@ -42,6 +42,7 @@
42#include <asm/xen/hypervisor.h> 42#include <asm/xen/hypervisor.h>
43#include <asm/fixmap.h> 43#include <asm/fixmap.h>
44#include <asm/processor.h> 44#include <asm/processor.h>
45#include <asm/proto.h>
45#include <asm/msr-index.h> 46#include <asm/msr-index.h>
46#include <asm/setup.h> 47#include <asm/setup.h>
47#include <asm/desc.h> 48#include <asm/desc.h>
@@ -168,21 +169,23 @@ static void __init xen_banner(void)
168 xen_feature(XENFEAT_mmu_pt_update_preserve_ad) ? " (preserve-AD)" : ""); 169 xen_feature(XENFEAT_mmu_pt_update_preserve_ad) ? " (preserve-AD)" : "");
169} 170}
170 171
172static __read_mostly unsigned int cpuid_leaf1_edx_mask = ~0;
173static __read_mostly unsigned int cpuid_leaf1_ecx_mask = ~0;
174
171static void xen_cpuid(unsigned int *ax, unsigned int *bx, 175static void xen_cpuid(unsigned int *ax, unsigned int *bx,
172 unsigned int *cx, unsigned int *dx) 176 unsigned int *cx, unsigned int *dx)
173{ 177{
178 unsigned maskecx = ~0;
174 unsigned maskedx = ~0; 179 unsigned maskedx = ~0;
175 180
176 /* 181 /*
177 * Mask out inconvenient features, to try and disable as many 182 * Mask out inconvenient features, to try and disable as many
178 * unsupported kernel subsystems as possible. 183 * unsupported kernel subsystems as possible.
179 */ 184 */
180 if (*ax == 1) 185 if (*ax == 1) {
181 maskedx = ~((1 << X86_FEATURE_APIC) | /* disable APIC */ 186 maskecx = cpuid_leaf1_ecx_mask;
182 (1 << X86_FEATURE_ACPI) | /* disable ACPI */ 187 maskedx = cpuid_leaf1_edx_mask;
183 (1 << X86_FEATURE_MCE) | /* disable MCE */ 188 }
184 (1 << X86_FEATURE_MCA) | /* disable MCA */
185 (1 << X86_FEATURE_ACC)); /* thermal monitoring */
186 189
187 asm(XEN_EMULATE_PREFIX "cpuid" 190 asm(XEN_EMULATE_PREFIX "cpuid"
188 : "=a" (*ax), 191 : "=a" (*ax),
@@ -190,9 +193,43 @@ static void xen_cpuid(unsigned int *ax, unsigned int *bx,
190 "=c" (*cx), 193 "=c" (*cx),
191 "=d" (*dx) 194 "=d" (*dx)
192 : "0" (*ax), "2" (*cx)); 195 : "0" (*ax), "2" (*cx));
196
197 *cx &= maskecx;
193 *dx &= maskedx; 198 *dx &= maskedx;
194} 199}
195 200
201static __init void xen_init_cpuid_mask(void)
202{
203 unsigned int ax, bx, cx, dx;
204
205 cpuid_leaf1_edx_mask =
206 ~((1 << X86_FEATURE_MCE) | /* disable MCE */
207 (1 << X86_FEATURE_MCA) | /* disable MCA */
208 (1 << X86_FEATURE_ACC)); /* thermal monitoring */
209
210 if (!xen_initial_domain())
211 cpuid_leaf1_edx_mask &=
212 ~((1 << X86_FEATURE_APIC) | /* disable local APIC */
213 (1 << X86_FEATURE_ACPI)); /* disable ACPI */
214
215 ax = 1;
216 xen_cpuid(&ax, &bx, &cx, &dx);
217
218 /* cpuid claims we support xsave; try enabling it to see what happens */
219 if (cx & (1 << (X86_FEATURE_XSAVE % 32))) {
220 unsigned long cr4;
221
222 set_in_cr4(X86_CR4_OSXSAVE);
223
224 cr4 = read_cr4();
225
226 if ((cr4 & X86_CR4_OSXSAVE) == 0)
227 cpuid_leaf1_ecx_mask &= ~(1 << (X86_FEATURE_XSAVE % 32));
228
229 clear_in_cr4(X86_CR4_OSXSAVE);
230 }
231}
232
196static void xen_set_debugreg(int reg, unsigned long val) 233static void xen_set_debugreg(int reg, unsigned long val)
197{ 234{
198 HYPERVISOR_set_debugreg(reg, val); 235 HYPERVISOR_set_debugreg(reg, val);
@@ -284,12 +321,11 @@ static void xen_set_ldt(const void *addr, unsigned entries)
284 321
285static void xen_load_gdt(const struct desc_ptr *dtr) 322static void xen_load_gdt(const struct desc_ptr *dtr)
286{ 323{
287 unsigned long *frames;
288 unsigned long va = dtr->address; 324 unsigned long va = dtr->address;
289 unsigned int size = dtr->size + 1; 325 unsigned int size = dtr->size + 1;
290 unsigned pages = (size + PAGE_SIZE - 1) / PAGE_SIZE; 326 unsigned pages = (size + PAGE_SIZE - 1) / PAGE_SIZE;
327 unsigned long frames[pages];
291 int f; 328 int f;
292 struct multicall_space mcs;
293 329
294 /* A GDT can be up to 64k in size, which corresponds to 8192 330 /* A GDT can be up to 64k in size, which corresponds to 8192
295 8-byte entries, or 16 4k pages.. */ 331 8-byte entries, or 16 4k pages.. */
@@ -297,19 +333,26 @@ static void xen_load_gdt(const struct desc_ptr *dtr)
297 BUG_ON(size > 65536); 333 BUG_ON(size > 65536);
298 BUG_ON(va & ~PAGE_MASK); 334 BUG_ON(va & ~PAGE_MASK);
299 335
300 mcs = xen_mc_entry(sizeof(*frames) * pages);
301 frames = mcs.args;
302
303 for (f = 0; va < dtr->address + size; va += PAGE_SIZE, f++) { 336 for (f = 0; va < dtr->address + size; va += PAGE_SIZE, f++) {
304 frames[f] = arbitrary_virt_to_mfn((void *)va); 337 int level;
338 pte_t *ptep = lookup_address(va, &level);
339 unsigned long pfn, mfn;
340 void *virt;
341
342 BUG_ON(ptep == NULL);
343
344 pfn = pte_pfn(*ptep);
345 mfn = pfn_to_mfn(pfn);
346 virt = __va(PFN_PHYS(pfn));
347
348 frames[f] = mfn;
305 349
306 make_lowmem_page_readonly((void *)va); 350 make_lowmem_page_readonly((void *)va);
307 make_lowmem_page_readonly(mfn_to_virt(frames[f])); 351 make_lowmem_page_readonly(virt);
308 } 352 }
309 353
310 MULTI_set_gdt(mcs.mc, frames, size / sizeof(struct desc_struct)); 354 if (HYPERVISOR_set_gdt(frames, size / sizeof(struct desc_struct)))
311 355 BUG();
312 xen_mc_issue(PARAVIRT_LAZY_CPU);
313} 356}
314 357
315static void load_TLS_descriptor(struct thread_struct *t, 358static void load_TLS_descriptor(struct thread_struct *t,
@@ -385,7 +428,7 @@ static void xen_write_ldt_entry(struct desc_struct *dt, int entrynum,
385static int cvt_gate_to_trap(int vector, const gate_desc *val, 428static int cvt_gate_to_trap(int vector, const gate_desc *val,
386 struct trap_info *info) 429 struct trap_info *info)
387{ 430{
388 if (val->type != 0xf && val->type != 0xe) 431 if (val->type != GATE_TRAP && val->type != GATE_INTERRUPT)
389 return 0; 432 return 0;
390 433
391 info->vector = vector; 434 info->vector = vector;
@@ -393,8 +436,8 @@ static int cvt_gate_to_trap(int vector, const gate_desc *val,
393 info->cs = gate_segment(*val); 436 info->cs = gate_segment(*val);
394 info->flags = val->dpl; 437 info->flags = val->dpl;
395 /* interrupt gates clear IF */ 438 /* interrupt gates clear IF */
396 if (val->type == 0xe) 439 if (val->type == GATE_INTERRUPT)
397 info->flags |= 4; 440 info->flags |= 1 << 2;
398 441
399 return 1; 442 return 1;
400} 443}
@@ -872,7 +915,6 @@ static const struct machine_ops __initdata xen_machine_ops = {
872 .emergency_restart = xen_emergency_restart, 915 .emergency_restart = xen_emergency_restart,
873}; 916};
874 917
875
876/* First C function to be called on Xen boot */ 918/* First C function to be called on Xen boot */
877asmlinkage void __init xen_start_kernel(void) 919asmlinkage void __init xen_start_kernel(void)
878{ 920{
@@ -897,6 +939,8 @@ asmlinkage void __init xen_start_kernel(void)
897 939
898 xen_init_irq_ops(); 940 xen_init_irq_ops();
899 941
942 xen_init_cpuid_mask();
943
900#ifdef CONFIG_X86_LOCAL_APIC 944#ifdef CONFIG_X86_LOCAL_APIC
901 /* 945 /*
902 * set up the basic apic ops. 946 * set up the basic apic ops.
@@ -938,6 +982,11 @@ asmlinkage void __init xen_start_kernel(void)
938 if (!xen_initial_domain()) 982 if (!xen_initial_domain())
939 __supported_pte_mask &= ~(_PAGE_PWT | _PAGE_PCD); 983 __supported_pte_mask &= ~(_PAGE_PWT | _PAGE_PCD);
940 984
985#ifdef CONFIG_X86_64
986 /* Work out if we support NX */
987 check_efer();
988#endif
989
941 /* Don't do the full vcpu_info placement stuff until we have a 990 /* Don't do the full vcpu_info placement stuff until we have a
942 possible map and a non-dummy shared_info. */ 991 possible map and a non-dummy shared_info. */
943 per_cpu(xen_vcpu, 0) = &HYPERVISOR_shared_info->vcpu_info[0]; 992 per_cpu(xen_vcpu, 0) = &HYPERVISOR_shared_info->vcpu_info[0];
diff --git a/arch/x86/xen/mmu.c b/arch/x86/xen/mmu.c
index db3802fb7b84..9842b1212407 100644
--- a/arch/x86/xen/mmu.c
+++ b/arch/x86/xen/mmu.c
@@ -184,7 +184,7 @@ static inline unsigned p2m_index(unsigned long pfn)
184} 184}
185 185
186/* Build the parallel p2m_top_mfn structures */ 186/* Build the parallel p2m_top_mfn structures */
187void xen_setup_mfn_list_list(void) 187static void __init xen_build_mfn_list_list(void)
188{ 188{
189 unsigned pfn, idx; 189 unsigned pfn, idx;
190 190
@@ -198,7 +198,10 @@ void xen_setup_mfn_list_list(void)
198 unsigned topidx = idx * P2M_ENTRIES_PER_PAGE; 198 unsigned topidx = idx * P2M_ENTRIES_PER_PAGE;
199 p2m_top_mfn_list[idx] = virt_to_mfn(&p2m_top_mfn[topidx]); 199 p2m_top_mfn_list[idx] = virt_to_mfn(&p2m_top_mfn[topidx]);
200 } 200 }
201}
201 202
203void xen_setup_mfn_list_list(void)
204{
202 BUG_ON(HYPERVISOR_shared_info == &xen_dummy_shared_info); 205 BUG_ON(HYPERVISOR_shared_info == &xen_dummy_shared_info);
203 206
204 HYPERVISOR_shared_info->arch.pfn_to_mfn_frame_list_list = 207 HYPERVISOR_shared_info->arch.pfn_to_mfn_frame_list_list =
@@ -218,6 +221,8 @@ void __init xen_build_dynamic_phys_to_machine(void)
218 221
219 p2m_top[topidx] = &mfn_list[pfn]; 222 p2m_top[topidx] = &mfn_list[pfn];
220 } 223 }
224
225 xen_build_mfn_list_list();
221} 226}
222 227
223unsigned long get_phys_to_machine(unsigned long pfn) 228unsigned long get_phys_to_machine(unsigned long pfn)
@@ -233,47 +238,74 @@ unsigned long get_phys_to_machine(unsigned long pfn)
233} 238}
234EXPORT_SYMBOL_GPL(get_phys_to_machine); 239EXPORT_SYMBOL_GPL(get_phys_to_machine);
235 240
236static void alloc_p2m(unsigned long **pp, unsigned long *mfnp) 241/* install a new p2m_top page */
242bool install_p2mtop_page(unsigned long pfn, unsigned long *p)
237{ 243{
238 unsigned long *p; 244 unsigned topidx = p2m_top_index(pfn);
245 unsigned long **pfnp, *mfnp;
239 unsigned i; 246 unsigned i;
240 247
241 p = (void *)__get_free_page(GFP_KERNEL | __GFP_NOFAIL); 248 pfnp = &p2m_top[topidx];
242 BUG_ON(p == NULL); 249 mfnp = &p2m_top_mfn[topidx];
243 250
244 for (i = 0; i < P2M_ENTRIES_PER_PAGE; i++) 251 for (i = 0; i < P2M_ENTRIES_PER_PAGE; i++)
245 p[i] = INVALID_P2M_ENTRY; 252 p[i] = INVALID_P2M_ENTRY;
246 253
247 if (cmpxchg(pp, p2m_missing, p) != p2m_missing) 254 if (cmpxchg(pfnp, p2m_missing, p) == p2m_missing) {
248 free_page((unsigned long)p);
249 else
250 *mfnp = virt_to_mfn(p); 255 *mfnp = virt_to_mfn(p);
256 return true;
257 }
258
259 return false;
251} 260}
252 261
253void set_phys_to_machine(unsigned long pfn, unsigned long mfn) 262static void alloc_p2m(unsigned long pfn)
254{ 263{
255 unsigned topidx, idx; 264 unsigned long *p;
256 265
257 if (unlikely(xen_feature(XENFEAT_auto_translated_physmap))) { 266 p = (void *)__get_free_page(GFP_KERNEL | __GFP_NOFAIL);
258 BUG_ON(pfn != mfn && mfn != INVALID_P2M_ENTRY); 267 BUG_ON(p == NULL);
259 return; 268
260 } 269 if (!install_p2mtop_page(pfn, p))
270 free_page((unsigned long)p);
271}
272
273/* Try to install p2m mapping; fail if intermediate bits missing */
274bool __set_phys_to_machine(unsigned long pfn, unsigned long mfn)
275{
276 unsigned topidx, idx;
261 277
262 if (unlikely(pfn >= MAX_DOMAIN_PAGES)) { 278 if (unlikely(pfn >= MAX_DOMAIN_PAGES)) {
263 BUG_ON(mfn != INVALID_P2M_ENTRY); 279 BUG_ON(mfn != INVALID_P2M_ENTRY);
264 return; 280 return true;
265 } 281 }
266 282
267 topidx = p2m_top_index(pfn); 283 topidx = p2m_top_index(pfn);
268 if (p2m_top[topidx] == p2m_missing) { 284 if (p2m_top[topidx] == p2m_missing) {
269 /* no need to allocate a page to store an invalid entry */
270 if (mfn == INVALID_P2M_ENTRY) 285 if (mfn == INVALID_P2M_ENTRY)
271 return; 286 return true;
272 alloc_p2m(&p2m_top[topidx], &p2m_top_mfn[topidx]); 287 return false;
273 } 288 }
274 289
275 idx = p2m_index(pfn); 290 idx = p2m_index(pfn);
276 p2m_top[topidx][idx] = mfn; 291 p2m_top[topidx][idx] = mfn;
292
293 return true;
294}
295
296void set_phys_to_machine(unsigned long pfn, unsigned long mfn)
297{
298 if (unlikely(xen_feature(XENFEAT_auto_translated_physmap))) {
299 BUG_ON(pfn != mfn && mfn != INVALID_P2M_ENTRY);
300 return;
301 }
302
303 if (unlikely(!__set_phys_to_machine(pfn, mfn))) {
304 alloc_p2m(pfn);
305
306 if (!__set_phys_to_machine(pfn, mfn))
307 BUG();
308 }
277} 309}
278 310
279unsigned long arbitrary_virt_to_mfn(void *vaddr) 311unsigned long arbitrary_virt_to_mfn(void *vaddr)
@@ -987,7 +1019,7 @@ static __init int xen_mark_pinned(struct mm_struct *mm, struct page *page,
987 return 0; 1019 return 0;
988} 1020}
989 1021
990void __init xen_mark_init_mm_pinned(void) 1022static void __init xen_mark_init_mm_pinned(void)
991{ 1023{
992 xen_pgd_walk(&init_mm, xen_mark_pinned, FIXADDR_TOP); 1024 xen_pgd_walk(&init_mm, xen_mark_pinned, FIXADDR_TOP);
993} 1025}
@@ -1270,8 +1302,8 @@ static void xen_flush_tlb_others(const struct cpumask *cpus,
1270 } *args; 1302 } *args;
1271 struct multicall_space mcs; 1303 struct multicall_space mcs;
1272 1304
1273 BUG_ON(cpumask_empty(cpus)); 1305 if (cpumask_empty(cpus))
1274 BUG_ON(!mm); 1306 return; /* nothing to do */
1275 1307
1276 mcs = xen_mc_entry(sizeof(*args)); 1308 mcs = xen_mc_entry(sizeof(*args));
1277 args = mcs.args; 1309 args = mcs.args;
@@ -1438,6 +1470,15 @@ static __init void xen_set_pte_init(pte_t *ptep, pte_t pte)
1438} 1470}
1439#endif 1471#endif
1440 1472
1473static void pin_pagetable_pfn(unsigned cmd, unsigned long pfn)
1474{
1475 struct mmuext_op op;
1476 op.cmd = cmd;
1477 op.arg1.mfn = pfn_to_mfn(pfn);
1478 if (HYPERVISOR_mmuext_op(&op, 1, NULL, DOMID_SELF))
1479 BUG();
1480}
1481
1441/* Early in boot, while setting up the initial pagetable, assume 1482/* Early in boot, while setting up the initial pagetable, assume
1442 everything is pinned. */ 1483 everything is pinned. */
1443static __init void xen_alloc_pte_init(struct mm_struct *mm, unsigned long pfn) 1484static __init void xen_alloc_pte_init(struct mm_struct *mm, unsigned long pfn)
@@ -1446,22 +1487,29 @@ static __init void xen_alloc_pte_init(struct mm_struct *mm, unsigned long pfn)
1446 BUG_ON(mem_map); /* should only be used early */ 1487 BUG_ON(mem_map); /* should only be used early */
1447#endif 1488#endif
1448 make_lowmem_page_readonly(__va(PFN_PHYS(pfn))); 1489 make_lowmem_page_readonly(__va(PFN_PHYS(pfn)));
1490 pin_pagetable_pfn(MMUEXT_PIN_L1_TABLE, pfn);
1491}
1492
1493/* Used for pmd and pud */
1494static __init void xen_alloc_pmd_init(struct mm_struct *mm, unsigned long pfn)
1495{
1496#ifdef CONFIG_FLATMEM
1497 BUG_ON(mem_map); /* should only be used early */
1498#endif
1499 make_lowmem_page_readonly(__va(PFN_PHYS(pfn)));
1449} 1500}
1450 1501
1451/* Early release_pte assumes that all pts are pinned, since there's 1502/* Early release_pte assumes that all pts are pinned, since there's
1452 only init_mm and anything attached to that is pinned. */ 1503 only init_mm and anything attached to that is pinned. */
1453static void xen_release_pte_init(unsigned long pfn) 1504static __init void xen_release_pte_init(unsigned long pfn)
1454{ 1505{
1506 pin_pagetable_pfn(MMUEXT_UNPIN_TABLE, pfn);
1455 make_lowmem_page_readwrite(__va(PFN_PHYS(pfn))); 1507 make_lowmem_page_readwrite(__va(PFN_PHYS(pfn)));
1456} 1508}
1457 1509
1458static void pin_pagetable_pfn(unsigned cmd, unsigned long pfn) 1510static __init void xen_release_pmd_init(unsigned long pfn)
1459{ 1511{
1460 struct mmuext_op op; 1512 make_lowmem_page_readwrite(__va(PFN_PHYS(pfn)));
1461 op.cmd = cmd;
1462 op.arg1.mfn = pfn_to_mfn(pfn);
1463 if (HYPERVISOR_mmuext_op(&op, 1, NULL, DOMID_SELF))
1464 BUG();
1465} 1513}
1466 1514
1467/* This needs to make sure the new pte page is pinned iff its being 1515/* This needs to make sure the new pte page is pinned iff its being
@@ -1750,7 +1798,7 @@ __init pgd_t *xen_setup_kernel_pagetable(pgd_t *pgd,
1750} 1798}
1751#endif /* CONFIG_X86_64 */ 1799#endif /* CONFIG_X86_64 */
1752 1800
1753static void xen_set_fixmap(unsigned idx, unsigned long phys, pgprot_t prot) 1801static void xen_set_fixmap(unsigned idx, phys_addr_t phys, pgprot_t prot)
1754{ 1802{
1755 pte_t pte; 1803 pte_t pte;
1756 1804
@@ -1773,6 +1821,9 @@ static void xen_set_fixmap(unsigned idx, unsigned long phys, pgprot_t prot)
1773#ifdef CONFIG_X86_LOCAL_APIC 1821#ifdef CONFIG_X86_LOCAL_APIC
1774 case FIX_APIC_BASE: /* maps dummy local APIC */ 1822 case FIX_APIC_BASE: /* maps dummy local APIC */
1775#endif 1823#endif
1824 case FIX_TEXT_POKE0:
1825 case FIX_TEXT_POKE1:
1826 /* All local page mappings */
1776 pte = pfn_pte(phys, prot); 1827 pte = pfn_pte(phys, prot);
1777 break; 1828 break;
1778 1829
@@ -1819,7 +1870,6 @@ __init void xen_post_allocator_init(void)
1819 xen_mark_init_mm_pinned(); 1870 xen_mark_init_mm_pinned();
1820} 1871}
1821 1872
1822
1823const struct pv_mmu_ops xen_mmu_ops __initdata = { 1873const struct pv_mmu_ops xen_mmu_ops __initdata = {
1824 .pagetable_setup_start = xen_pagetable_setup_start, 1874 .pagetable_setup_start = xen_pagetable_setup_start,
1825 .pagetable_setup_done = xen_pagetable_setup_done, 1875 .pagetable_setup_done = xen_pagetable_setup_done,
@@ -1843,9 +1893,9 @@ const struct pv_mmu_ops xen_mmu_ops __initdata = {
1843 1893
1844 .alloc_pte = xen_alloc_pte_init, 1894 .alloc_pte = xen_alloc_pte_init,
1845 .release_pte = xen_release_pte_init, 1895 .release_pte = xen_release_pte_init,
1846 .alloc_pmd = xen_alloc_pte_init, 1896 .alloc_pmd = xen_alloc_pmd_init,
1847 .alloc_pmd_clone = paravirt_nop, 1897 .alloc_pmd_clone = paravirt_nop,
1848 .release_pmd = xen_release_pte_init, 1898 .release_pmd = xen_release_pmd_init,
1849 1899
1850#ifdef CONFIG_HIGHPTE 1900#ifdef CONFIG_HIGHPTE
1851 .kmap_atomic_pte = xen_kmap_atomic_pte, 1901 .kmap_atomic_pte = xen_kmap_atomic_pte,
@@ -1883,8 +1933,8 @@ const struct pv_mmu_ops xen_mmu_ops __initdata = {
1883 .make_pud = PV_CALLEE_SAVE(xen_make_pud), 1933 .make_pud = PV_CALLEE_SAVE(xen_make_pud),
1884 .set_pgd = xen_set_pgd_hyper, 1934 .set_pgd = xen_set_pgd_hyper,
1885 1935
1886 .alloc_pud = xen_alloc_pte_init, 1936 .alloc_pud = xen_alloc_pmd_init,
1887 .release_pud = xen_release_pte_init, 1937 .release_pud = xen_release_pmd_init,
1888#endif /* PAGETABLE_LEVELS == 4 */ 1938#endif /* PAGETABLE_LEVELS == 4 */
1889 1939
1890 .activate_mm = xen_activate_mm, 1940 .activate_mm = xen_activate_mm,
diff --git a/arch/x86/xen/mmu.h b/arch/x86/xen/mmu.h
index 24d1b44a337d..da7302624897 100644
--- a/arch/x86/xen/mmu.h
+++ b/arch/x86/xen/mmu.h
@@ -11,6 +11,9 @@ enum pt_level {
11}; 11};
12 12
13 13
14bool __set_phys_to_machine(unsigned long pfn, unsigned long mfn);
15bool install_p2mtop_page(unsigned long pfn, unsigned long *p);
16
14void set_pte_mfn(unsigned long vaddr, unsigned long pfn, pgprot_t flags); 17void set_pte_mfn(unsigned long vaddr, unsigned long pfn, pgprot_t flags);
15 18
16 19
diff --git a/arch/x86/xen/smp.c b/arch/x86/xen/smp.c
index 8d470562ffc9..429834ec1687 100644
--- a/arch/x86/xen/smp.c
+++ b/arch/x86/xen/smp.c
@@ -158,7 +158,7 @@ static void __init xen_fill_possible_map(void)
158 rc = HYPERVISOR_vcpu_op(VCPUOP_is_up, i, NULL); 158 rc = HYPERVISOR_vcpu_op(VCPUOP_is_up, i, NULL);
159 if (rc >= 0) { 159 if (rc >= 0) {
160 num_processors++; 160 num_processors++;
161 cpu_set(i, cpu_possible_map); 161 set_cpu_possible(i, true);
162 } 162 }
163 } 163 }
164} 164}
@@ -197,7 +197,7 @@ static void __init xen_smp_prepare_cpus(unsigned int max_cpus)
197 while ((num_possible_cpus() > 1) && (num_possible_cpus() > max_cpus)) { 197 while ((num_possible_cpus() > 1) && (num_possible_cpus() > max_cpus)) {
198 for (cpu = nr_cpu_ids - 1; !cpu_possible(cpu); cpu--) 198 for (cpu = nr_cpu_ids - 1; !cpu_possible(cpu); cpu--)
199 continue; 199 continue;
200 cpu_clear(cpu, cpu_possible_map); 200 set_cpu_possible(cpu, false);
201 } 201 }
202 202
203 for_each_possible_cpu (cpu) { 203 for_each_possible_cpu (cpu) {
@@ -210,7 +210,7 @@ static void __init xen_smp_prepare_cpus(unsigned int max_cpus)
210 if (IS_ERR(idle)) 210 if (IS_ERR(idle))
211 panic("failed fork for CPU %d", cpu); 211 panic("failed fork for CPU %d", cpu);
212 212
213 cpu_set(cpu, cpu_present_map); 213 set_cpu_present(cpu, true);
214 } 214 }
215} 215}
216 216
@@ -317,7 +317,7 @@ static int __cpuinit xen_cpu_up(unsigned int cpu)
317 BUG_ON(rc); 317 BUG_ON(rc);
318 318
319 while(per_cpu(cpu_state, cpu) != CPU_ONLINE) { 319 while(per_cpu(cpu_state, cpu) != CPU_ONLINE) {
320 HYPERVISOR_sched_op(SCHEDOP_yield, 0); 320 HYPERVISOR_sched_op(SCHEDOP_yield, NULL);
321 barrier(); 321 barrier();
322 } 322 }
323 323
@@ -422,7 +422,7 @@ static void xen_smp_send_call_function_ipi(const struct cpumask *mask)
422 /* Make sure other vcpus get a chance to run if they need to. */ 422 /* Make sure other vcpus get a chance to run if they need to. */
423 for_each_cpu(cpu, mask) { 423 for_each_cpu(cpu, mask) {
424 if (xen_vcpu_stolen(cpu)) { 424 if (xen_vcpu_stolen(cpu)) {
425 HYPERVISOR_sched_op(SCHEDOP_yield, 0); 425 HYPERVISOR_sched_op(SCHEDOP_yield, NULL);
426 break; 426 break;
427 } 427 }
428 } 428 }
diff --git a/arch/x86/xen/xen-ops.h b/arch/x86/xen/xen-ops.h
index 2f5ef2632ea2..20139464943c 100644
--- a/arch/x86/xen/xen-ops.h
+++ b/arch/x86/xen/xen-ops.h
@@ -57,8 +57,6 @@ irqreturn_t xen_debug_interrupt(int irq, void *dev_id);
57 57
58bool xen_vcpu_stolen(int vcpu); 58bool xen_vcpu_stolen(int vcpu);
59 59
60void xen_mark_init_mm_pinned(void);
61
62void xen_setup_vcpu_info_placement(void); 60void xen_setup_vcpu_info_placement(void);
63 61
64#ifdef CONFIG_SMP 62#ifdef CONFIG_SMP
generated by cgit v1.2.2 (git 2.25.0) at 2025-01-20 17:16:55 -0500