aboutsummaryrefslogtreecommitdiffstats
path: root/arch/x86
diff options
context:
space:
mode:
Diffstat (limited to 'arch/x86')
-rw-r--r--arch/x86/Kconfig7
-rw-r--r--arch/x86/boot/video-vga.c33
-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/fixmap.h4
-rw-r--r--arch/x86/include/asm/io.h6
-rw-r--r--arch/x86/include/asm/paravirt.h4
-rw-r--r--arch/x86/include/asm/processor.h6
-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/uv/uv_mmrs.h5
-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/apic/apic.c6
-rw-r--r--arch/x86/kernel/apic/apic_flat_64.c4
-rw-r--r--arch/x86/kernel/apic/io_apic.c5
-rw-r--r--arch/x86/kernel/apic/x2apic_uv_x.c16
-rw-r--r--arch/x86/kernel/bios_uv.c3
-rw-r--r--arch/x86/kernel/cpu/addon_cpuid_features.c1
-rw-r--r--[-rwxr-xr-x]arch/x86/kernel/cpu/cpu_debug.c0
-rw-r--r--arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c81
-rw-r--r--arch/x86/kernel/cpu/cpufreq/longhaul.c1
-rw-r--r--arch/x86/kernel/ftrace.c2
-rw-r--r--arch/x86/kernel/irq.c2
-rw-r--r--arch/x86/kernel/microcode_core.c33
-rw-r--r--arch/x86/kernel/mpparse.c7
-rw-r--r--arch/x86/kernel/ptrace.c3
-rw-r--r--arch/x86/kernel/reboot.c8
-rw-r--r--arch/x86/kernel/tlb_uv.c189
-rw-r--r--arch/x86/kernel/uv_sysfs.c4
-rw-r--r--arch/x86/kernel/xsave.c2
-rw-r--r--arch/x86/mm/gup.c16
-rw-r--r--arch/x86/mm/ioremap.c23
-rw-r--r--arch/x86/mm/pat.c2
-rw-r--r--arch/x86/mm/pgtable.c3
-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.c4
-rw-r--r--arch/x86/xen/xen-ops.h2
40 files changed, 478 insertions, 229 deletions
diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
index 4b3408206091..c9086e6307a5 100644
--- a/arch/x86/Kconfig
+++ b/arch/x86/Kconfig
@@ -252,17 +252,13 @@ config SMP
252 252
253config X86_X2APIC 253config X86_X2APIC
254 bool "Support x2apic" 254 bool "Support x2apic"
255 depends on X86_LOCAL_APIC && X86_64 255 depends on X86_LOCAL_APIC && X86_64 && INTR_REMAP
256 select INTR_REMAP
257 ---help--- 256 ---help---
258 This enables x2apic support on CPUs that have this feature. 257 This enables x2apic support on CPUs that have this feature.
259 258
260 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),
261 and accesses the local apic via MSRs not via mmio. 260 and accesses the local apic via MSRs not via mmio.
262 261
263 ( On certain CPU models you may need to enable INTR_REMAP too,
264 to get functional x2apic mode. )
265
266 If you don't know what to do here, say N. 262 If you don't know what to do here, say N.
267 263
268config SPARSE_IRQ 264config SPARSE_IRQ
@@ -357,6 +353,7 @@ config X86_UV
357 bool "SGI Ultraviolet" 353 bool "SGI Ultraviolet"
358 depends on X86_64 354 depends on X86_64
359 depends on X86_EXTENDED_PLATFORM 355 depends on X86_EXTENDED_PLATFORM
356 depends on NUMA
360 select X86_X2APIC 357 select X86_X2APIC
361 ---help--- 358 ---help---
362 This option is needed in order to support SGI Ultraviolet systems. 359 This option is needed in order to support SGI Ultraviolet systems.
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/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/fixmap.h b/arch/x86/include/asm/fixmap.h
index 81937a5dc77c..2d81af3974a0 100644
--- a/arch/x86/include/asm/fixmap.h
+++ b/arch/x86/include/asm/fixmap.h
@@ -151,11 +151,11 @@ extern pte_t *pkmap_page_table;
151 151
152void __native_set_fixmap(enum fixed_addresses idx, pte_t pte); 152void __native_set_fixmap(enum fixed_addresses idx, pte_t pte);
153void native_set_fixmap(enum fixed_addresses idx, 153void native_set_fixmap(enum fixed_addresses idx,
154 unsigned long phys, pgprot_t flags); 154 phys_addr_t phys, pgprot_t flags);
155 155
156#ifndef CONFIG_PARAVIRT 156#ifndef CONFIG_PARAVIRT
157static inline void __set_fixmap(enum fixed_addresses idx, 157static inline void __set_fixmap(enum fixed_addresses idx,
158 unsigned long phys, pgprot_t flags) 158 phys_addr_t phys, pgprot_t flags)
159{ 159{
160 native_set_fixmap(idx, phys, flags); 160 native_set_fixmap(idx, phys, flags);
161} 161}
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/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/processor.h b/arch/x86/include/asm/processor.h
index 34c52370f2fe..fcf4d92e7e04 100644
--- a/arch/x86/include/asm/processor.h
+++ b/arch/x86/include/asm/processor.h
@@ -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
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/uv/uv_mmrs.h b/arch/x86/include/asm/uv/uv_mmrs.h
index db68ac8a5ac2..2cae46c7c8a2 100644
--- a/arch/x86/include/asm/uv/uv_mmrs.h
+++ b/arch/x86/include/asm/uv/uv_mmrs.h
@@ -17,6 +17,11 @@
17/* ========================================================================= */ 17/* ========================================================================= */
18/* UVH_BAU_DATA_CONFIG */ 18/* UVH_BAU_DATA_CONFIG */
19/* ========================================================================= */ 19/* ========================================================================= */
20#define UVH_LB_BAU_MISC_CONTROL 0x320170UL
21#define UV_ENABLE_INTD_SOFT_ACK_MODE_SHIFT 15
22#define UV_INTD_SOFT_ACK_TIMEOUT_PERIOD_SHIFT 16
23#define UV_INTD_SOFT_ACK_TIMEOUT_PERIOD 0x000000000bUL
24/* 1011 timebase 7 (168millisec) * 3 ticks -> 500ms */
20#define UVH_BAU_DATA_CONFIG 0x61680UL 25#define UVH_BAU_DATA_CONFIG 0x61680UL
21#define UVH_BAU_DATA_CONFIG_32 0x0438 26#define UVH_BAU_DATA_CONFIG_32 0x0438
22 27
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/apic/apic.c b/arch/x86/kernel/apic/apic.c
index 4b48ff9163ca..83e47febcc89 100644
--- a/arch/x86/kernel/apic/apic.c
+++ b/arch/x86/kernel/apic/apic.c
@@ -472,6 +472,12 @@ static void __cpuinit setup_APIC_timer(void)
472{ 472{
473 struct clock_event_device *levt = &__get_cpu_var(lapic_events); 473 struct clock_event_device *levt = &__get_cpu_var(lapic_events);
474 474
475 if (cpu_has(&current_cpu_data, X86_FEATURE_ARAT)) {
476 lapic_clockevent.features &= ~CLOCK_EVT_FEAT_C3STOP;
477 /* Make LAPIC timer preferrable over percpu HPET */
478 lapic_clockevent.rating = 150;
479 }
480
475 memcpy(levt, &lapic_clockevent, sizeof(*levt)); 481 memcpy(levt, &lapic_clockevent, sizeof(*levt));
476 levt->cpumask = cpumask_of(smp_processor_id()); 482 levt->cpumask = cpumask_of(smp_processor_id());
477 483
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/io_apic.c b/arch/x86/kernel/apic/io_apic.c
index 7c9d045ac834..849900022407 100644
--- a/arch/x86/kernel/apic/io_apic.c
+++ b/arch/x86/kernel/apic/io_apic.c
@@ -2519,7 +2519,6 @@ static void irq_complete_move(struct irq_desc **descp)
2519static inline void irq_complete_move(struct irq_desc **descp) {} 2519static inline void irq_complete_move(struct irq_desc **descp) {}
2520#endif 2520#endif
2521 2521
2522#ifdef CONFIG_X86_X2APIC
2523static void __eoi_ioapic_irq(unsigned int irq, struct irq_cfg *cfg) 2522static void __eoi_ioapic_irq(unsigned int irq, struct irq_cfg *cfg)
2524{ 2523{
2525 int apic, pin; 2524 int apic, pin;
@@ -2553,6 +2552,7 @@ eoi_ioapic_irq(struct irq_desc *desc)
2553 spin_unlock_irqrestore(&ioapic_lock, flags); 2552 spin_unlock_irqrestore(&ioapic_lock, flags);
2554} 2553}
2555 2554
2555#ifdef CONFIG_X86_X2APIC
2556static void ack_x2apic_level(unsigned int irq) 2556static void ack_x2apic_level(unsigned int irq)
2557{ 2557{
2558 struct irq_desc *desc = irq_to_desc(irq); 2558 struct irq_desc *desc = irq_to_desc(irq);
@@ -2629,6 +2629,9 @@ static void ack_apic_level(unsigned int irq)
2629 */ 2629 */
2630 ack_APIC_irq(); 2630 ack_APIC_irq();
2631 2631
2632 if (irq_remapped(irq))
2633 eoi_ioapic_irq(desc);
2634
2632 /* Now we can move and renable the irq */ 2635 /* Now we can move and renable the irq */
2633 if (unlikely(do_unmask_irq)) { 2636 if (unlikely(do_unmask_irq)) {
2634 /* Only migrate the irq if the ack has been received. 2637 /* Only migrate the irq if the ack has been received.
diff --git a/arch/x86/kernel/apic/x2apic_uv_x.c b/arch/x86/kernel/apic/x2apic_uv_x.c
index 1248318436e8..de1a50af807b 100644
--- a/arch/x86/kernel/apic/x2apic_uv_x.c
+++ b/arch/x86/kernel/apic/x2apic_uv_x.c
@@ -549,7 +549,8 @@ void __init uv_system_init(void)
549 unsigned long gnode_upper, lowmem_redir_base, lowmem_redir_size; 549 unsigned long gnode_upper, lowmem_redir_base, lowmem_redir_size;
550 int bytes, nid, cpu, lcpu, pnode, blade, i, j, m_val, n_val; 550 int bytes, nid, cpu, lcpu, pnode, blade, i, j, m_val, n_val;
551 int max_pnode = 0; 551 int max_pnode = 0;
552 unsigned long mmr_base, present; 552 unsigned long mmr_base, present, paddr;
553 unsigned short pnode_mask;
553 554
554 map_low_mmrs(); 555 map_low_mmrs();
555 556
@@ -592,6 +593,7 @@ void __init uv_system_init(void)
592 } 593 }
593 } 594 }
594 595
596 pnode_mask = (1 << n_val) - 1;
595 node_id.v = uv_read_local_mmr(UVH_NODE_ID); 597 node_id.v = uv_read_local_mmr(UVH_NODE_ID);
596 gnode_upper = (((unsigned long)node_id.s.node_id) & 598 gnode_upper = (((unsigned long)node_id.s.node_id) &
597 ~((1 << n_val) - 1)) << m_val; 599 ~((1 << n_val) - 1)) << m_val;
@@ -615,7 +617,7 @@ void __init uv_system_init(void)
615 uv_cpu_hub_info(cpu)->numa_blade_id = blade; 617 uv_cpu_hub_info(cpu)->numa_blade_id = blade;
616 uv_cpu_hub_info(cpu)->blade_processor_id = lcpu; 618 uv_cpu_hub_info(cpu)->blade_processor_id = lcpu;
617 uv_cpu_hub_info(cpu)->pnode = pnode; 619 uv_cpu_hub_info(cpu)->pnode = pnode;
618 uv_cpu_hub_info(cpu)->pnode_mask = (1 << n_val) - 1; 620 uv_cpu_hub_info(cpu)->pnode_mask = pnode_mask;
619 uv_cpu_hub_info(cpu)->gpa_mask = (1 << (m_val + n_val)) - 1; 621 uv_cpu_hub_info(cpu)->gpa_mask = (1 << (m_val + n_val)) - 1;
620 uv_cpu_hub_info(cpu)->gnode_upper = gnode_upper; 622 uv_cpu_hub_info(cpu)->gnode_upper = gnode_upper;
621 uv_cpu_hub_info(cpu)->global_mmr_base = mmr_base; 623 uv_cpu_hub_info(cpu)->global_mmr_base = mmr_base;
@@ -631,6 +633,16 @@ void __init uv_system_init(void)
631 lcpu, blade); 633 lcpu, blade);
632 } 634 }
633 635
636 /* Add blade/pnode info for nodes without cpus */
637 for_each_online_node(nid) {
638 if (uv_node_to_blade[nid] >= 0)
639 continue;
640 paddr = node_start_pfn(nid) << PAGE_SHIFT;
641 pnode = (paddr >> m_val) & pnode_mask;
642 blade = boot_pnode_to_blade(pnode);
643 uv_node_to_blade[nid] = blade;
644 }
645
634 map_gru_high(max_pnode); 646 map_gru_high(max_pnode);
635 map_mmr_high(max_pnode); 647 map_mmr_high(max_pnode);
636 map_config_high(max_pnode); 648 map_config_high(max_pnode);
diff --git a/arch/x86/kernel/bios_uv.c b/arch/x86/kernel/bios_uv.c
index f63882728d91..63a88e1f987d 100644
--- a/arch/x86/kernel/bios_uv.c
+++ b/arch/x86/kernel/bios_uv.c
@@ -182,7 +182,8 @@ void uv_bios_init(void)
182 memcpy(&uv_systab, tab, sizeof(struct uv_systab)); 182 memcpy(&uv_systab, tab, sizeof(struct uv_systab));
183 iounmap(tab); 183 iounmap(tab);
184 184
185 printk(KERN_INFO "EFI UV System Table Revision %d\n", tab->revision); 185 printk(KERN_INFO "EFI UV System Table Revision %d\n",
186 uv_systab.revision);
186} 187}
187#else /* !CONFIG_EFI */ 188#else /* !CONFIG_EFI */
188 189
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/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 19f6b9d27e83..ecdb682ab516 100644
--- a/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c
+++ b/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c
@@ -68,6 +68,7 @@ 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);
@@ -152,7 +153,8 @@ struct drv_cmd {
152 u32 val; 153 u32 val;
153}; 154};
154 155
155static long do_drv_read(void *_cmd) 156/* Called via smp_call_function_single(), on the target CPU */
157static void do_drv_read(void *_cmd)
156{ 158{
157 struct drv_cmd *cmd = _cmd; 159 struct drv_cmd *cmd = _cmd;
158 u32 h; 160 u32 h;
@@ -169,10 +171,10 @@ static long do_drv_read(void *_cmd)
169 default: 171 default:
170 break; 172 break;
171 } 173 }
172 return 0;
173} 174}
174 175
175static long do_drv_write(void *_cmd) 176/* Called via smp_call_function_many(), on the target CPUs */
177static void do_drv_write(void *_cmd)
176{ 178{
177 struct drv_cmd *cmd = _cmd; 179 struct drv_cmd *cmd = _cmd;
178 u32 lo, hi; 180 u32 lo, hi;
@@ -191,23 +193,24 @@ static long do_drv_write(void *_cmd)
191 default: 193 default:
192 break; 194 break;
193 } 195 }
194 return 0;
195} 196}
196 197
197static void drv_read(struct drv_cmd *cmd) 198static void drv_read(struct drv_cmd *cmd)
198{ 199{
199 cmd->val = 0; 200 cmd->val = 0;
200 201
201 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);
202} 203}
203 204
204static void drv_write(struct drv_cmd *cmd) 205static void drv_write(struct drv_cmd *cmd)
205{ 206{
206 unsigned int i; 207 int this_cpu;
207 208
208 for_each_cpu(i, cmd->mask) { 209 this_cpu = get_cpu();
209 work_on_cpu(i, do_drv_write, cmd); 210 if (cpumask_test_cpu(this_cpu, cmd->mask))
210 } 211 do_drv_write(cmd);
212 smp_call_function_many(cmd->mask, do_drv_write, cmd, 1);
213 put_cpu();
211} 214}
212 215
213static u32 get_cur_val(const struct cpumask *mask) 216static u32 get_cur_val(const struct cpumask *mask)
@@ -241,28 +244,23 @@ static u32 get_cur_val(const struct cpumask *mask)
241 return cmd.val; 244 return cmd.val;
242} 245}
243 246
244struct perf_cur { 247struct perf_pair {
245 union { 248 union {
246 struct { 249 struct {
247 u32 lo; 250 u32 lo;
248 u32 hi; 251 u32 hi;
249 } split; 252 } split;
250 u64 whole; 253 u64 whole;
251 } aperf_cur, mperf_cur; 254 } aperf, mperf;
252}; 255};
253 256
254 257/* Called via smp_call_function_single(), on the target CPU */
255static long read_measured_perf_ctrs(void *_cur) 258static void read_measured_perf_ctrs(void *_cur)
256{ 259{
257 struct perf_cur *cur = _cur; 260 struct perf_pair *cur = _cur;
258 261
259 rdmsr(MSR_IA32_APERF, cur->aperf_cur.split.lo, cur->aperf_cur.split.hi); 262 rdmsr(MSR_IA32_APERF, cur->aperf.split.lo, cur->aperf.split.hi);
260 rdmsr(MSR_IA32_MPERF, cur->mperf_cur.split.lo, cur->mperf_cur.split.hi); 263 rdmsr(MSR_IA32_MPERF, cur->mperf.split.lo, cur->mperf.split.hi);
261
262 wrmsr(MSR_IA32_APERF, 0, 0);
263 wrmsr(MSR_IA32_MPERF, 0, 0);
264
265 return 0;
266} 264}
267 265
268/* 266/*
@@ -281,52 +279,57 @@ static long read_measured_perf_ctrs(void *_cur)
281static unsigned int get_measured_perf(struct cpufreq_policy *policy, 279static unsigned int get_measured_perf(struct cpufreq_policy *policy,
282 unsigned int cpu) 280 unsigned int cpu)
283{ 281{
284 struct perf_cur cur; 282 struct perf_pair readin, cur;
285 unsigned int perf_percent; 283 unsigned int perf_percent;
286 unsigned int retval; 284 unsigned int retval;
287 285
288 if (!work_on_cpu(cpu, read_measured_perf_ctrs, &cur)) 286 if (smp_call_function_single(cpu, read_measured_perf_ctrs, &readin, 1))
289 return 0; 287 return 0;
290 288
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
291#ifdef __i386__ 296#ifdef __i386__
292 /* 297 /*
293 * 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
294 * Get an approximate value. Return failure in case we cannot get 299 * Get an approximate value. Return failure in case we cannot get
295 * an approximate value. 300 * an approximate value.
296 */ 301 */
297 if (unlikely(cur.aperf_cur.split.hi || cur.mperf_cur.split.hi)) { 302 if (unlikely(cur.aperf.split.hi || cur.mperf.split.hi)) {
298 int shift_count; 303 int shift_count;
299 u32 h; 304 u32 h;
300 305
301 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);
302 shift_count = fls(h); 307 shift_count = fls(h);
303 308
304 cur.aperf_cur.whole >>= shift_count; 309 cur.aperf.whole >>= shift_count;
305 cur.mperf_cur.whole >>= shift_count; 310 cur.mperf.whole >>= shift_count;
306 } 311 }
307 312
308 if (((unsigned long)(-1) / 100) < cur.aperf_cur.split.lo) { 313 if (((unsigned long)(-1) / 100) < cur.aperf.split.lo) {
309 int shift_count = 7; 314 int shift_count = 7;
310 cur.aperf_cur.split.lo >>= shift_count; 315 cur.aperf.split.lo >>= shift_count;
311 cur.mperf_cur.split.lo >>= shift_count; 316 cur.mperf.split.lo >>= shift_count;
312 } 317 }
313 318
314 if (cur.aperf_cur.split.lo && cur.mperf_cur.split.lo) 319 if (cur.aperf.split.lo && cur.mperf.split.lo)
315 perf_percent = (cur.aperf_cur.split.lo * 100) / 320 perf_percent = (cur.aperf.split.lo * 100) / cur.mperf.split.lo;
316 cur.mperf_cur.split.lo;
317 else 321 else
318 perf_percent = 0; 322 perf_percent = 0;
319 323
320#else 324#else
321 if (unlikely(((unsigned long)(-1) / 100) < cur.aperf_cur.whole)) { 325 if (unlikely(((unsigned long)(-1) / 100) < cur.aperf.whole)) {
322 int shift_count = 7; 326 int shift_count = 7;
323 cur.aperf_cur.whole >>= shift_count; 327 cur.aperf.whole >>= shift_count;
324 cur.mperf_cur.whole >>= shift_count; 328 cur.mperf.whole >>= shift_count;
325 } 329 }
326 330
327 if (cur.aperf_cur.whole && cur.mperf_cur.whole) 331 if (cur.aperf.whole && cur.mperf.whole)
328 perf_percent = (cur.aperf_cur.whole * 100) / 332 perf_percent = (cur.aperf.whole * 100) / cur.mperf.whole;
329 cur.mperf_cur.whole;
330 else 333 else
331 perf_percent = 0; 334 perf_percent = 0;
332 335
diff --git a/arch/x86/kernel/cpu/cpufreq/longhaul.c b/arch/x86/kernel/cpu/cpufreq/longhaul.c
index 0bd48e65a0ca..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>
diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c
index 70a10ca100f6..18dfa30795c9 100644
--- a/arch/x86/kernel/ftrace.c
+++ b/arch/x86/kernel/ftrace.c
@@ -18,6 +18,8 @@
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
21#include <asm/cacheflush.h> 23#include <asm/cacheflush.h>
22#include <asm/ftrace.h> 24#include <asm/ftrace.h>
23#include <asm/nops.h> 25#include <asm/nops.h>
diff --git a/arch/x86/kernel/irq.c b/arch/x86/kernel/irq.c
index fd57bf35d0fc..c1739ac29708 100644
--- a/arch/x86/kernel/irq.c
+++ b/arch/x86/kernel/irq.c
@@ -63,7 +63,7 @@ static int show_other_interrupts(struct seq_file *p, int prec)
63 seq_printf(p, " Spurious interrupts\n"); 63 seq_printf(p, " Spurious interrupts\n");
64#endif 64#endif
65 if (generic_interrupt_extension) { 65 if (generic_interrupt_extension) {
66 seq_printf(p, "PLT: "); 66 seq_printf(p, "%*s: ", prec, "PLT");
67 for_each_online_cpu(j) 67 for_each_online_cpu(j)
68 seq_printf(p, "%10u ", irq_stats(j)->generic_irqs); 68 seq_printf(p, "%10u ", irq_stats(j)->generic_irqs);
69 seq_printf(p, " Platform interrupts\n"); 69 seq_printf(p, " Platform interrupts\n");
diff --git a/arch/x86/kernel/microcode_core.c b/arch/x86/kernel/microcode_core.c
index a0f3851ef310..2e0eb4140951 100644
--- a/arch/x86/kernel/microcode_core.c
+++ b/arch/x86/kernel/microcode_core.c
@@ -108,40 +108,29 @@ struct 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
111struct update_for_cpu {
112 const void __user *buf;
113 size_t size;
114};
115
116static long update_for_cpu(void *_ufc)
117{
118 struct update_for_cpu *ufc = _ufc;
119 int error;
120
121 error = microcode_ops->request_microcode_user(smp_processor_id(),
122 ufc->buf, ufc->size);
123 if (error < 0)
124 return error;
125 if (!error)
126 microcode_ops->apply_microcode(smp_processor_id());
127 return error;
128}
129
130static int do_microcode_update(const void __user *buf, size_t size) 111static int do_microcode_update(const void __user *buf, size_t size)
131{ 112{
113 cpumask_t old;
132 int error = 0; 114 int error = 0;
133 int cpu; 115 int cpu;
134 struct update_for_cpu ufc = { .buf = buf, .size = size }; 116
117 old = current->cpus_allowed;
135 118
136 for_each_online_cpu(cpu) { 119 for_each_online_cpu(cpu) {
137 struct ucode_cpu_info *uci = ucode_cpu_info + cpu; 120 struct ucode_cpu_info *uci = ucode_cpu_info + cpu;
138 121
139 if (!uci->valid) 122 if (!uci->valid)
140 continue; 123 continue;
141 error = work_on_cpu(cpu, update_for_cpu, &ufc); 124
125 set_cpus_allowed_ptr(current, &cpumask_of_cpu(cpu));
126 error = microcode_ops->request_microcode_user(cpu, buf, size);
142 if (error < 0) 127 if (error < 0)
143 break; 128 goto out;
129 if (!error)
130 microcode_ops->apply_microcode(cpu);
144 } 131 }
132out:
133 set_cpus_allowed_ptr(current, &old);
145 return error; 134 return error;
146} 135}
147 136
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/ptrace.c b/arch/x86/kernel/ptrace.c
index fe9345c967de..23b7c8f017e2 100644
--- a/arch/x86/kernel/ptrace.c
+++ b/arch/x86/kernel/ptrace.c
@@ -21,7 +21,6 @@
21#include <linux/audit.h> 21#include <linux/audit.h>
22#include <linux/seccomp.h> 22#include <linux/seccomp.h>
23#include <linux/signal.h> 23#include <linux/signal.h>
24#include <linux/ftrace.h>
25 24
26#include <asm/uaccess.h> 25#include <asm/uaccess.h>
27#include <asm/pgtable.h> 26#include <asm/pgtable.h>
@@ -35,6 +34,8 @@
35#include <asm/proto.h> 34#include <asm/proto.h>
36#include <asm/ds.h> 35#include <asm/ds.h>
37 36
37#include <trace/syscall.h>
38
38#include "tls.h" 39#include "tls.h"
39 40
40enum x86_regset { 41enum x86_regset {
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/tlb_uv.c b/arch/x86/kernel/tlb_uv.c
index deb5ebb32c3b..ed0c33761e6d 100644
--- a/arch/x86/kernel/tlb_uv.c
+++ b/arch/x86/kernel/tlb_uv.c
@@ -25,6 +25,8 @@ static int uv_bau_retry_limit __read_mostly;
25 25
26/* position of pnode (which is nasid>>1): */ 26/* position of pnode (which is nasid>>1): */
27static int uv_nshift __read_mostly; 27static int uv_nshift __read_mostly;
28/* base pnode in this partition */
29static int uv_partition_base_pnode __read_mostly;
28 30
29static unsigned long uv_mmask __read_mostly; 31static unsigned long uv_mmask __read_mostly;
30 32
@@ -32,6 +34,34 @@ static DEFINE_PER_CPU(struct ptc_stats, ptcstats);
32static DEFINE_PER_CPU(struct bau_control, bau_control); 34static DEFINE_PER_CPU(struct bau_control, bau_control);
33 35
34/* 36/*
37 * Determine the first node on a blade.
38 */
39static int __init blade_to_first_node(int blade)
40{
41 int node, b;
42
43 for_each_online_node(node) {
44 b = uv_node_to_blade_id(node);
45 if (blade == b)
46 return node;
47 }
48 return -1; /* shouldn't happen */
49}
50
51/*
52 * Determine the apicid of the first cpu on a blade.
53 */
54static int __init blade_to_first_apicid(int blade)
55{
56 int cpu;
57
58 for_each_present_cpu(cpu)
59 if (blade == uv_cpu_to_blade_id(cpu))
60 return per_cpu(x86_cpu_to_apicid, cpu);
61 return -1;
62}
63
64/*
35 * Free a software acknowledge hardware resource by clearing its Pending 65 * Free a software acknowledge hardware resource by clearing its Pending
36 * bit. This will return a reply to the sender. 66 * bit. This will return a reply to the sender.
37 * If the message has timed out, a reply has already been sent by the 67 * If the message has timed out, a reply has already been sent by the
@@ -67,7 +97,7 @@ static void uv_bau_process_message(struct bau_payload_queue_entry *msg,
67 msp = __get_cpu_var(bau_control).msg_statuses + msg_slot; 97 msp = __get_cpu_var(bau_control).msg_statuses + msg_slot;
68 cpu = uv_blade_processor_id(); 98 cpu = uv_blade_processor_id();
69 msg->number_of_cpus = 99 msg->number_of_cpus =
70 uv_blade_nr_online_cpus(uv_node_to_blade_id(numa_node_id())); 100 uv_blade_nr_online_cpus(uv_node_to_blade_id(numa_node_id()));
71 this_cpu_mask = 1UL << cpu; 101 this_cpu_mask = 1UL << cpu;
72 if (msp->seen_by.bits & this_cpu_mask) 102 if (msp->seen_by.bits & this_cpu_mask)
73 return; 103 return;
@@ -215,14 +245,14 @@ static int uv_wait_completion(struct bau_desc *bau_desc,
215 * Returns @flush_mask if some remote flushing remains to be done. The 245 * Returns @flush_mask if some remote flushing remains to be done. The
216 * mask will have some bits still set. 246 * mask will have some bits still set.
217 */ 247 */
218const struct cpumask *uv_flush_send_and_wait(int cpu, int this_blade, 248const struct cpumask *uv_flush_send_and_wait(int cpu, int this_pnode,
219 struct bau_desc *bau_desc, 249 struct bau_desc *bau_desc,
220 struct cpumask *flush_mask) 250 struct cpumask *flush_mask)
221{ 251{
222 int completion_status = 0; 252 int completion_status = 0;
223 int right_shift; 253 int right_shift;
224 int tries = 0; 254 int tries = 0;
225 int blade; 255 int pnode;
226 int bit; 256 int bit;
227 unsigned long mmr_offset; 257 unsigned long mmr_offset;
228 unsigned long index; 258 unsigned long index;
@@ -265,8 +295,8 @@ const struct cpumask *uv_flush_send_and_wait(int cpu, int this_blade,
265 * use the IPI method of shootdown on them. 295 * use the IPI method of shootdown on them.
266 */ 296 */
267 for_each_cpu(bit, flush_mask) { 297 for_each_cpu(bit, flush_mask) {
268 blade = uv_cpu_to_blade_id(bit); 298 pnode = uv_cpu_to_pnode(bit);
269 if (blade == this_blade) 299 if (pnode == this_pnode)
270 continue; 300 continue;
271 cpumask_clear_cpu(bit, flush_mask); 301 cpumask_clear_cpu(bit, flush_mask);
272 } 302 }
@@ -309,16 +339,16 @@ const struct cpumask *uv_flush_tlb_others(const struct cpumask *cpumask,
309 struct cpumask *flush_mask = __get_cpu_var(uv_flush_tlb_mask); 339 struct cpumask *flush_mask = __get_cpu_var(uv_flush_tlb_mask);
310 int i; 340 int i;
311 int bit; 341 int bit;
312 int blade; 342 int pnode;
313 int uv_cpu; 343 int uv_cpu;
314 int this_blade; 344 int this_pnode;
315 int locals = 0; 345 int locals = 0;
316 struct bau_desc *bau_desc; 346 struct bau_desc *bau_desc;
317 347
318 cpumask_andnot(flush_mask, cpumask, cpumask_of(cpu)); 348 cpumask_andnot(flush_mask, cpumask, cpumask_of(cpu));
319 349
320 uv_cpu = uv_blade_processor_id(); 350 uv_cpu = uv_blade_processor_id();
321 this_blade = uv_numa_blade_id(); 351 this_pnode = uv_hub_info->pnode;
322 bau_desc = __get_cpu_var(bau_control).descriptor_base; 352 bau_desc = __get_cpu_var(bau_control).descriptor_base;
323 bau_desc += UV_ITEMS_PER_DESCRIPTOR * uv_cpu; 353 bau_desc += UV_ITEMS_PER_DESCRIPTOR * uv_cpu;
324 354
@@ -326,13 +356,14 @@ const struct cpumask *uv_flush_tlb_others(const struct cpumask *cpumask,
326 356
327 i = 0; 357 i = 0;
328 for_each_cpu(bit, flush_mask) { 358 for_each_cpu(bit, flush_mask) {
329 blade = uv_cpu_to_blade_id(bit); 359 pnode = uv_cpu_to_pnode(bit);
330 BUG_ON(blade > (UV_DISTRIBUTION_SIZE - 1)); 360 BUG_ON(pnode > (UV_DISTRIBUTION_SIZE - 1));
331 if (blade == this_blade) { 361 if (pnode == this_pnode) {
332 locals++; 362 locals++;
333 continue; 363 continue;
334 } 364 }
335 bau_node_set(blade, &bau_desc->distribution); 365 bau_node_set(pnode - uv_partition_base_pnode,
366 &bau_desc->distribution);
336 i++; 367 i++;
337 } 368 }
338 if (i == 0) { 369 if (i == 0) {
@@ -350,7 +381,7 @@ const struct cpumask *uv_flush_tlb_others(const struct cpumask *cpumask,
350 bau_desc->payload.address = va; 381 bau_desc->payload.address = va;
351 bau_desc->payload.sending_cpu = cpu; 382 bau_desc->payload.sending_cpu = cpu;
352 383
353 return uv_flush_send_and_wait(uv_cpu, this_blade, bau_desc, flush_mask); 384 return uv_flush_send_and_wait(uv_cpu, this_pnode, bau_desc, flush_mask);
354} 385}
355 386
356/* 387/*
@@ -418,24 +449,58 @@ void uv_bau_message_interrupt(struct pt_regs *regs)
418 set_irq_regs(old_regs); 449 set_irq_regs(old_regs);
419} 450}
420 451
452/*
453 * uv_enable_timeouts
454 *
455 * Each target blade (i.e. blades that have cpu's) needs to have
456 * shootdown message timeouts enabled. The timeout does not cause
457 * an interrupt, but causes an error message to be returned to
458 * the sender.
459 */
421static void uv_enable_timeouts(void) 460static void uv_enable_timeouts(void)
422{ 461{
423 int i;
424 int blade; 462 int blade;
425 int last_blade; 463 int nblades;
426 int pnode; 464 int pnode;
427 int cur_cpu = 0; 465 unsigned long mmr_image;
428 unsigned long apicid;
429 466
430 last_blade = -1; 467 nblades = uv_num_possible_blades();
431 for_each_online_node(i) { 468
432 blade = uv_node_to_blade_id(i); 469 for (blade = 0; blade < nblades; blade++) {
433 if (blade == last_blade) 470 if (!uv_blade_nr_possible_cpus(blade))
434 continue; 471 continue;
435 last_blade = blade; 472
436 apicid = per_cpu(x86_cpu_to_apicid, cur_cpu);
437 pnode = uv_blade_to_pnode(blade); 473 pnode = uv_blade_to_pnode(blade);
438 cur_cpu += uv_blade_nr_possible_cpus(i); 474 mmr_image =
475 uv_read_global_mmr64(pnode, UVH_LB_BAU_MISC_CONTROL);
476 /*
477 * Set the timeout period and then lock it in, in three
478 * steps; captures and locks in the period.
479 *
480 * To program the period, the SOFT_ACK_MODE must be off.
481 */
482 mmr_image &= ~((unsigned long)1 <<
483 UV_ENABLE_INTD_SOFT_ACK_MODE_SHIFT);
484 uv_write_global_mmr64
485 (pnode, UVH_LB_BAU_MISC_CONTROL, mmr_image);
486 /*
487 * Set the 4-bit period.
488 */
489 mmr_image &= ~((unsigned long)0xf <<
490 UV_INTD_SOFT_ACK_TIMEOUT_PERIOD_SHIFT);
491 mmr_image |= (UV_INTD_SOFT_ACK_TIMEOUT_PERIOD <<
492 UV_INTD_SOFT_ACK_TIMEOUT_PERIOD_SHIFT);
493 uv_write_global_mmr64
494 (pnode, UVH_LB_BAU_MISC_CONTROL, mmr_image);
495 /*
496 * Subsequent reversals of the timebase bit (3) cause an
497 * immediate timeout of one or all INTD resources as
498 * indicated in bits 2:0 (7 causes all of them to timeout).
499 */
500 mmr_image |= ((unsigned long)1 <<
501 UV_ENABLE_INTD_SOFT_ACK_MODE_SHIFT);
502 uv_write_global_mmr64
503 (pnode, UVH_LB_BAU_MISC_CONTROL, mmr_image);
439 } 504 }
440} 505}
441 506
@@ -482,8 +547,7 @@ static int uv_ptc_seq_show(struct seq_file *file, void *data)
482 stat->requestee, stat->onetlb, stat->alltlb, 547 stat->requestee, stat->onetlb, stat->alltlb,
483 stat->s_retry, stat->d_retry, stat->ptc_i); 548 stat->s_retry, stat->d_retry, stat->ptc_i);
484 seq_printf(file, "%lx %ld %ld %ld %ld %ld %ld\n", 549 seq_printf(file, "%lx %ld %ld %ld %ld %ld %ld\n",
485 uv_read_global_mmr64(uv_blade_to_pnode 550 uv_read_global_mmr64(uv_cpu_to_pnode(cpu),
486 (uv_cpu_to_blade_id(cpu)),
487 UVH_LB_BAU_INTD_SOFTWARE_ACKNOWLEDGE), 551 UVH_LB_BAU_INTD_SOFTWARE_ACKNOWLEDGE),
488 stat->sflush, stat->dflush, 552 stat->sflush, stat->dflush,
489 stat->retriesok, stat->nomsg, 553 stat->retriesok, stat->nomsg,
@@ -617,16 +681,18 @@ static struct bau_control * __init uv_table_bases_init(int blade, int node)
617 * finish the initialization of the per-blade control structures 681 * finish the initialization of the per-blade control structures
618 */ 682 */
619static void __init 683static void __init
620uv_table_bases_finish(int blade, int node, int cur_cpu, 684uv_table_bases_finish(int blade,
621 struct bau_control *bau_tablesp, 685 struct bau_control *bau_tablesp,
622 struct bau_desc *adp) 686 struct bau_desc *adp)
623{ 687{
624 struct bau_control *bcp; 688 struct bau_control *bcp;
625 int i; 689 int cpu;
626 690
627 for (i = cur_cpu; i < cur_cpu + uv_blade_nr_possible_cpus(blade); i++) { 691 for_each_present_cpu(cpu) {
628 bcp = (struct bau_control *)&per_cpu(bau_control, i); 692 if (blade != uv_cpu_to_blade_id(cpu))
693 continue;
629 694
695 bcp = (struct bau_control *)&per_cpu(bau_control, cpu);
630 bcp->bau_msg_head = bau_tablesp->va_queue_first; 696 bcp->bau_msg_head = bau_tablesp->va_queue_first;
631 bcp->va_queue_first = bau_tablesp->va_queue_first; 697 bcp->va_queue_first = bau_tablesp->va_queue_first;
632 bcp->va_queue_last = bau_tablesp->va_queue_last; 698 bcp->va_queue_last = bau_tablesp->va_queue_last;
@@ -649,11 +715,10 @@ uv_activation_descriptor_init(int node, int pnode)
649 struct bau_desc *adp; 715 struct bau_desc *adp;
650 struct bau_desc *ad2; 716 struct bau_desc *ad2;
651 717
652 adp = (struct bau_desc *) 718 adp = (struct bau_desc *)kmalloc_node(16384, GFP_KERNEL, node);
653 kmalloc_node(16384, GFP_KERNEL, node);
654 BUG_ON(!adp); 719 BUG_ON(!adp);
655 720
656 pa = __pa((unsigned long)adp); 721 pa = uv_gpa(adp); /* need the real nasid*/
657 n = pa >> uv_nshift; 722 n = pa >> uv_nshift;
658 m = pa & uv_mmask; 723 m = pa & uv_mmask;
659 724
@@ -667,8 +732,12 @@ uv_activation_descriptor_init(int node, int pnode)
667 for (i = 0, ad2 = adp; i < UV_ACTIVATION_DESCRIPTOR_SIZE; i++, ad2++) { 732 for (i = 0, ad2 = adp; i < UV_ACTIVATION_DESCRIPTOR_SIZE; i++, ad2++) {
668 memset(ad2, 0, sizeof(struct bau_desc)); 733 memset(ad2, 0, sizeof(struct bau_desc));
669 ad2->header.sw_ack_flag = 1; 734 ad2->header.sw_ack_flag = 1;
670 ad2->header.base_dest_nodeid = 735 /*
671 uv_blade_to_pnode(uv_cpu_to_blade_id(0)); 736 * base_dest_nodeid is the first node in the partition, so
737 * the bit map will indicate partition-relative node numbers.
738 * note that base_dest_nodeid is actually a nasid.
739 */
740 ad2->header.base_dest_nodeid = uv_partition_base_pnode << 1;
672 ad2->header.command = UV_NET_ENDPOINT_INTD; 741 ad2->header.command = UV_NET_ENDPOINT_INTD;
673 ad2->header.int_both = 1; 742 ad2->header.int_both = 1;
674 /* 743 /*
@@ -686,6 +755,8 @@ static struct bau_payload_queue_entry * __init
686uv_payload_queue_init(int node, int pnode, struct bau_control *bau_tablesp) 755uv_payload_queue_init(int node, int pnode, struct bau_control *bau_tablesp)
687{ 756{
688 struct bau_payload_queue_entry *pqp; 757 struct bau_payload_queue_entry *pqp;
758 unsigned long pa;
759 int pn;
689 char *cp; 760 char *cp;
690 761
691 pqp = (struct bau_payload_queue_entry *) kmalloc_node( 762 pqp = (struct bau_payload_queue_entry *) kmalloc_node(
@@ -696,10 +767,14 @@ uv_payload_queue_init(int node, int pnode, struct bau_control *bau_tablesp)
696 cp = (char *)pqp + 31; 767 cp = (char *)pqp + 31;
697 pqp = (struct bau_payload_queue_entry *)(((unsigned long)cp >> 5) << 5); 768 pqp = (struct bau_payload_queue_entry *)(((unsigned long)cp >> 5) << 5);
698 bau_tablesp->va_queue_first = pqp; 769 bau_tablesp->va_queue_first = pqp;
770 /*
771 * need the pnode of where the memory was really allocated
772 */
773 pa = uv_gpa(pqp);
774 pn = pa >> uv_nshift;
699 uv_write_global_mmr64(pnode, 775 uv_write_global_mmr64(pnode,
700 UVH_LB_BAU_INTD_PAYLOAD_QUEUE_FIRST, 776 UVH_LB_BAU_INTD_PAYLOAD_QUEUE_FIRST,
701 ((unsigned long)pnode << 777 ((unsigned long)pn << UV_PAYLOADQ_PNODE_SHIFT) |
702 UV_PAYLOADQ_PNODE_SHIFT) |
703 uv_physnodeaddr(pqp)); 778 uv_physnodeaddr(pqp));
704 uv_write_global_mmr64(pnode, UVH_LB_BAU_INTD_PAYLOAD_QUEUE_TAIL, 779 uv_write_global_mmr64(pnode, UVH_LB_BAU_INTD_PAYLOAD_QUEUE_TAIL,
705 uv_physnodeaddr(pqp)); 780 uv_physnodeaddr(pqp));
@@ -715,8 +790,9 @@ uv_payload_queue_init(int node, int pnode, struct bau_control *bau_tablesp)
715/* 790/*
716 * Initialization of each UV blade's structures 791 * Initialization of each UV blade's structures
717 */ 792 */
718static int __init uv_init_blade(int blade, int node, int cur_cpu) 793static int __init uv_init_blade(int blade)
719{ 794{
795 int node;
720 int pnode; 796 int pnode;
721 unsigned long pa; 797 unsigned long pa;
722 unsigned long apicid; 798 unsigned long apicid;
@@ -724,16 +800,17 @@ static int __init uv_init_blade(int blade, int node, int cur_cpu)
724 struct bau_payload_queue_entry *pqp; 800 struct bau_payload_queue_entry *pqp;
725 struct bau_control *bau_tablesp; 801 struct bau_control *bau_tablesp;
726 802
803 node = blade_to_first_node(blade);
727 bau_tablesp = uv_table_bases_init(blade, node); 804 bau_tablesp = uv_table_bases_init(blade, node);
728 pnode = uv_blade_to_pnode(blade); 805 pnode = uv_blade_to_pnode(blade);
729 adp = uv_activation_descriptor_init(node, pnode); 806 adp = uv_activation_descriptor_init(node, pnode);
730 pqp = uv_payload_queue_init(node, pnode, bau_tablesp); 807 pqp = uv_payload_queue_init(node, pnode, bau_tablesp);
731 uv_table_bases_finish(blade, node, cur_cpu, bau_tablesp, adp); 808 uv_table_bases_finish(blade, bau_tablesp, adp);
732 /* 809 /*
733 * the below initialization can't be in firmware because the 810 * the below initialization can't be in firmware because the
734 * messaging IRQ will be determined by the OS 811 * messaging IRQ will be determined by the OS
735 */ 812 */
736 apicid = per_cpu(x86_cpu_to_apicid, cur_cpu); 813 apicid = blade_to_first_apicid(blade);
737 pa = uv_read_global_mmr64(pnode, UVH_BAU_DATA_CONFIG); 814 pa = uv_read_global_mmr64(pnode, UVH_BAU_DATA_CONFIG);
738 if ((pa & 0xff) != UV_BAU_MESSAGE) { 815 if ((pa & 0xff) != UV_BAU_MESSAGE) {
739 uv_write_global_mmr64(pnode, UVH_BAU_DATA_CONFIG, 816 uv_write_global_mmr64(pnode, UVH_BAU_DATA_CONFIG,
@@ -748,9 +825,7 @@ static int __init uv_init_blade(int blade, int node, int cur_cpu)
748static int __init uv_bau_init(void) 825static int __init uv_bau_init(void)
749{ 826{
750 int blade; 827 int blade;
751 int node;
752 int nblades; 828 int nblades;
753 int last_blade;
754 int cur_cpu; 829 int cur_cpu;
755 830
756 if (!is_uv_system()) 831 if (!is_uv_system())
@@ -763,29 +838,21 @@ static int __init uv_bau_init(void)
763 uv_bau_retry_limit = 1; 838 uv_bau_retry_limit = 1;
764 uv_nshift = uv_hub_info->n_val; 839 uv_nshift = uv_hub_info->n_val;
765 uv_mmask = (1UL << uv_hub_info->n_val) - 1; 840 uv_mmask = (1UL << uv_hub_info->n_val) - 1;
766 nblades = 0; 841 nblades = uv_num_possible_blades();
767 last_blade = -1; 842
768 cur_cpu = 0;
769 for_each_online_node(node) {
770 blade = uv_node_to_blade_id(node);
771 if (blade == last_blade)
772 continue;
773 last_blade = blade;
774 nblades++;
775 }
776 uv_bau_table_bases = (struct bau_control **) 843 uv_bau_table_bases = (struct bau_control **)
777 kmalloc(nblades * sizeof(struct bau_control *), GFP_KERNEL); 844 kmalloc(nblades * sizeof(struct bau_control *), GFP_KERNEL);
778 BUG_ON(!uv_bau_table_bases); 845 BUG_ON(!uv_bau_table_bases);
779 846
780 last_blade = -1; 847 uv_partition_base_pnode = 0x7fffffff;
781 for_each_online_node(node) { 848 for (blade = 0; blade < nblades; blade++)
782 blade = uv_node_to_blade_id(node); 849 if (uv_blade_nr_possible_cpus(blade) &&
783 if (blade == last_blade) 850 (uv_blade_to_pnode(blade) < uv_partition_base_pnode))
784 continue; 851 uv_partition_base_pnode = uv_blade_to_pnode(blade);
785 last_blade = blade; 852 for (blade = 0; blade < nblades; blade++)
786 uv_init_blade(blade, node, cur_cpu); 853 if (uv_blade_nr_possible_cpus(blade))
787 cur_cpu += uv_blade_nr_possible_cpus(blade); 854 uv_init_blade(blade);
788 } 855
789 alloc_intr_gate(UV_BAU_MESSAGE, uv_bau_message_intr1); 856 alloc_intr_gate(UV_BAU_MESSAGE, uv_bau_message_intr1);
790 uv_enable_timeouts(); 857 uv_enable_timeouts();
791 858
diff --git a/arch/x86/kernel/uv_sysfs.c b/arch/x86/kernel/uv_sysfs.c
index 67f9b9dbf800..36afb98675a4 100644
--- a/arch/x86/kernel/uv_sysfs.c
+++ b/arch/x86/kernel/uv_sysfs.c
@@ -21,6 +21,7 @@
21 21
22#include <linux/sysdev.h> 22#include <linux/sysdev.h>
23#include <asm/uv/bios.h> 23#include <asm/uv/bios.h>
24#include <asm/uv/uv.h>
24 25
25struct kobject *sgi_uv_kobj; 26struct kobject *sgi_uv_kobj;
26 27
@@ -47,6 +48,9 @@ static int __init sgi_uv_sysfs_init(void)
47{ 48{
48 unsigned long ret; 49 unsigned long ret;
49 50
51 if (!is_uv_system())
52 return -ENODEV;
53
50 if (!sgi_uv_kobj) 54 if (!sgi_uv_kobj)
51 sgi_uv_kobj = kobject_create_and_add("sgi_uv", firmware_kobj); 55 sgi_uv_kobj = kobject_create_and_add("sgi_uv", firmware_kobj);
52 if (!sgi_uv_kobj) { 56 if (!sgi_uv_kobj) {
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/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/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/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/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 585a6e330837..429834ec1687 100644
--- a/arch/x86/xen/smp.c
+++ b/arch/x86/xen/smp.c
@@ -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 2025-01-22 05:20:19 -0500