aboutsummaryrefslogtreecommitdiffstats
path: root/arch/x86/kernel
diff options
context:
space:
mode:
Diffstat (limited to 'arch/x86/kernel')
-rw-r--r--arch/x86/kernel/acpi/boot.c16
-rw-r--r--arch/x86/kernel/apb_timer.c2
-rw-r--r--arch/x86/kernel/apic/apic.c4
-rw-r--r--arch/x86/kernel/cpu/Makefile7
-rw-r--r--arch/x86/kernel/cpu/common.c2
-rw-r--r--arch/x86/kernel/cpu/intel.c5
-rw-r--r--arch/x86/kernel/cpu/microcode/amd_early.c33
-rw-r--r--arch/x86/kernel/cpu/microcode/core.c10
-rw-r--r--arch/x86/kernel/cpu/microcode/core_early.c2
-rw-r--r--arch/x86/kernel/cpu/perf_event.c14
-rw-r--r--arch/x86/kernel/cpu/perf_event.h1
-rw-r--r--arch/x86/kernel/cpu/perf_event_intel.c173
-rw-r--r--arch/x86/kernel/cpu/perf_event_intel_uncore_snbep.c49
-rw-r--r--arch/x86/kernel/dumpstack_64.c1
-rw-r--r--arch/x86/kernel/entry_32.S15
-rw-r--r--arch/x86/kernel/entry_64.S81
-rw-r--r--arch/x86/kernel/i8259.c3
-rw-r--r--arch/x86/kernel/irqinit.c3
-rw-r--r--arch/x86/kernel/ptrace.c2
-rw-r--r--arch/x86/kernel/setup.c2
-rw-r--r--arch/x86/kernel/smpboot.c15
-rw-r--r--arch/x86/kernel/traps.c71
-rw-r--r--arch/x86/kernel/tsc.c5
23 files changed, 206 insertions, 310 deletions
diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c
index b436fc735aa4..a142e77693e1 100644
--- a/arch/x86/kernel/acpi/boot.c
+++ b/arch/x86/kernel/acpi/boot.c
@@ -397,7 +397,7 @@ static int mp_register_gsi(struct device *dev, u32 gsi, int trigger,
397 397
398 /* Don't set up the ACPI SCI because it's already set up */ 398 /* Don't set up the ACPI SCI because it's already set up */
399 if (acpi_gbl_FADT.sci_interrupt == gsi) 399 if (acpi_gbl_FADT.sci_interrupt == gsi)
400 return gsi; 400 return mp_map_gsi_to_irq(gsi, IOAPIC_MAP_ALLOC);
401 401
402 trigger = trigger == ACPI_EDGE_SENSITIVE ? 0 : 1; 402 trigger = trigger == ACPI_EDGE_SENSITIVE ? 0 : 1;
403 polarity = polarity == ACPI_ACTIVE_HIGH ? 0 : 1; 403 polarity = polarity == ACPI_ACTIVE_HIGH ? 0 : 1;
@@ -604,14 +604,18 @@ void __init acpi_pic_sci_set_trigger(unsigned int irq, u16 trigger)
604 604
605int acpi_gsi_to_irq(u32 gsi, unsigned int *irqp) 605int acpi_gsi_to_irq(u32 gsi, unsigned int *irqp)
606{ 606{
607 int irq = mp_map_gsi_to_irq(gsi, IOAPIC_MAP_ALLOC | IOAPIC_MAP_CHECK); 607 int irq;
608 608
609 if (irq >= 0) { 609 if (acpi_irq_model == ACPI_IRQ_MODEL_PIC) {
610 *irqp = gsi;
611 } else {
612 irq = mp_map_gsi_to_irq(gsi,
613 IOAPIC_MAP_ALLOC | IOAPIC_MAP_CHECK);
614 if (irq < 0)
615 return -1;
610 *irqp = irq; 616 *irqp = irq;
611 return 0;
612 } 617 }
613 618 return 0;
614 return -1;
615} 619}
616EXPORT_SYMBOL_GPL(acpi_gsi_to_irq); 620EXPORT_SYMBOL_GPL(acpi_gsi_to_irq);
617 621
diff --git a/arch/x86/kernel/apb_timer.c b/arch/x86/kernel/apb_timer.c
index 5972b108f15a..b708738d016e 100644
--- a/arch/x86/kernel/apb_timer.c
+++ b/arch/x86/kernel/apb_timer.c
@@ -185,8 +185,6 @@ static void apbt_setup_irq(struct apbt_dev *adev)
185 185
186 irq_modify_status(adev->irq, 0, IRQ_MOVE_PCNTXT); 186 irq_modify_status(adev->irq, 0, IRQ_MOVE_PCNTXT);
187 irq_set_affinity(adev->irq, cpumask_of(adev->cpu)); 187 irq_set_affinity(adev->irq, cpumask_of(adev->cpu));
188 /* APB timer irqs are set up as mp_irqs, timer is edge type */
189 __irq_set_handler(adev->irq, handle_edge_irq, 0, "edge");
190} 188}
191 189
192/* Should be called with per cpu */ 190/* Should be called with per cpu */
diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c
index 00853b254ab0..ba6cc041edb1 100644
--- a/arch/x86/kernel/apic/apic.c
+++ b/arch/x86/kernel/apic/apic.c
@@ -1297,7 +1297,7 @@ void setup_local_APIC(void)
1297 unsigned int value, queued; 1297 unsigned int value, queued;
1298 int i, j, acked = 0; 1298 int i, j, acked = 0;
1299 unsigned long long tsc = 0, ntsc; 1299 unsigned long long tsc = 0, ntsc;
1300 long long max_loops = cpu_khz; 1300 long long max_loops = cpu_khz ? cpu_khz : 1000000;
1301 1301
1302 if (cpu_has_tsc) 1302 if (cpu_has_tsc)
1303 rdtscll(tsc); 1303 rdtscll(tsc);
@@ -1383,7 +1383,7 @@ void setup_local_APIC(void)
1383 break; 1383 break;
1384 } 1384 }
1385 if (queued) { 1385 if (queued) {
1386 if (cpu_has_tsc) { 1386 if (cpu_has_tsc && cpu_khz) {
1387 rdtscll(ntsc); 1387 rdtscll(ntsc);
1388 max_loops = (cpu_khz << 10) - (ntsc - tsc); 1388 max_loops = (cpu_khz << 10) - (ntsc - tsc);
1389 } else 1389 } else
diff --git a/arch/x86/kernel/cpu/Makefile b/arch/x86/kernel/cpu/Makefile
index 01d5453b5502..e27b49d7c922 100644
--- a/arch/x86/kernel/cpu/Makefile
+++ b/arch/x86/kernel/cpu/Makefile
@@ -39,9 +39,12 @@ obj-$(CONFIG_CPU_SUP_AMD) += perf_event_amd_iommu.o
39endif 39endif
40obj-$(CONFIG_CPU_SUP_INTEL) += perf_event_p6.o perf_event_knc.o perf_event_p4.o 40obj-$(CONFIG_CPU_SUP_INTEL) += perf_event_p6.o perf_event_knc.o perf_event_p4.o
41obj-$(CONFIG_CPU_SUP_INTEL) += perf_event_intel_lbr.o perf_event_intel_ds.o perf_event_intel.o 41obj-$(CONFIG_CPU_SUP_INTEL) += perf_event_intel_lbr.o perf_event_intel_ds.o perf_event_intel.o
42obj-$(CONFIG_CPU_SUP_INTEL) += perf_event_intel_uncore.o perf_event_intel_uncore_snb.o
43obj-$(CONFIG_CPU_SUP_INTEL) += perf_event_intel_uncore_snbep.o perf_event_intel_uncore_nhmex.o
44obj-$(CONFIG_CPU_SUP_INTEL) += perf_event_intel_rapl.o 42obj-$(CONFIG_CPU_SUP_INTEL) += perf_event_intel_rapl.o
43
44obj-$(CONFIG_PERF_EVENTS_INTEL_UNCORE) += perf_event_intel_uncore.o \
45 perf_event_intel_uncore_snb.o \
46 perf_event_intel_uncore_snbep.o \
47 perf_event_intel_uncore_nhmex.o
45endif 48endif
46 49
47 50
diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c
index 4b4f78c9ba19..cfa9b5b2c27a 100644
--- a/arch/x86/kernel/cpu/common.c
+++ b/arch/x86/kernel/cpu/common.c
@@ -146,6 +146,8 @@ EXPORT_PER_CPU_SYMBOL_GPL(gdt_page);
146 146
147static int __init x86_xsave_setup(char *s) 147static int __init x86_xsave_setup(char *s)
148{ 148{
149 if (strlen(s))
150 return 0;
149 setup_clear_cpu_cap(X86_FEATURE_XSAVE); 151 setup_clear_cpu_cap(X86_FEATURE_XSAVE);
150 setup_clear_cpu_cap(X86_FEATURE_XSAVEOPT); 152 setup_clear_cpu_cap(X86_FEATURE_XSAVEOPT);
151 setup_clear_cpu_cap(X86_FEATURE_XSAVES); 153 setup_clear_cpu_cap(X86_FEATURE_XSAVES);
diff --git a/arch/x86/kernel/cpu/intel.c b/arch/x86/kernel/cpu/intel.c
index 1ef456273172..9cc6b6f25f42 100644
--- a/arch/x86/kernel/cpu/intel.c
+++ b/arch/x86/kernel/cpu/intel.c
@@ -213,12 +213,13 @@ static void intel_workarounds(struct cpuinfo_x86 *c)
213{ 213{
214#ifdef CONFIG_X86_F00F_BUG 214#ifdef CONFIG_X86_F00F_BUG
215 /* 215 /*
216 * All current models of Pentium and Pentium with MMX technology CPUs 216 * All models of Pentium and Pentium with MMX technology CPUs
217 * have the F0 0F bug, which lets nonprivileged users lock up the 217 * have the F0 0F bug, which lets nonprivileged users lock up the
218 * system. Announce that the fault handler will be checking for it. 218 * system. Announce that the fault handler will be checking for it.
219 * The Quark is also family 5, but does not have the same bug.
219 */ 220 */
220 clear_cpu_bug(c, X86_BUG_F00F); 221 clear_cpu_bug(c, X86_BUG_F00F);
221 if (!paravirt_enabled() && c->x86 == 5) { 222 if (!paravirt_enabled() && c->x86 == 5 && c->x86_model < 9) {
222 static int f00f_workaround_enabled; 223 static int f00f_workaround_enabled;
223 224
224 set_cpu_bug(c, X86_BUG_F00F); 225 set_cpu_bug(c, X86_BUG_F00F);
diff --git a/arch/x86/kernel/cpu/microcode/amd_early.c b/arch/x86/kernel/cpu/microcode/amd_early.c
index 7aa1acc79789..06674473b0e6 100644
--- a/arch/x86/kernel/cpu/microcode/amd_early.c
+++ b/arch/x86/kernel/cpu/microcode/amd_early.c
@@ -108,12 +108,13 @@ static size_t compute_container_size(u8 *data, u32 total_size)
108 * load_microcode_amd() to save equivalent cpu table and microcode patches in 108 * load_microcode_amd() to save equivalent cpu table and microcode patches in
109 * kernel heap memory. 109 * kernel heap memory.
110 */ 110 */
111static void apply_ucode_in_initrd(void *ucode, size_t size) 111static void apply_ucode_in_initrd(void *ucode, size_t size, bool save_patch)
112{ 112{
113 struct equiv_cpu_entry *eq; 113 struct equiv_cpu_entry *eq;
114 size_t *cont_sz; 114 size_t *cont_sz;
115 u32 *header; 115 u32 *header;
116 u8 *data, **cont; 116 u8 *data, **cont;
117 u8 (*patch)[PATCH_MAX_SIZE];
117 u16 eq_id = 0; 118 u16 eq_id = 0;
118 int offset, left; 119 int offset, left;
119 u32 rev, eax, ebx, ecx, edx; 120 u32 rev, eax, ebx, ecx, edx;
@@ -123,10 +124,12 @@ static void apply_ucode_in_initrd(void *ucode, size_t size)
123 new_rev = (u32 *)__pa_nodebug(&ucode_new_rev); 124 new_rev = (u32 *)__pa_nodebug(&ucode_new_rev);
124 cont_sz = (size_t *)__pa_nodebug(&container_size); 125 cont_sz = (size_t *)__pa_nodebug(&container_size);
125 cont = (u8 **)__pa_nodebug(&container); 126 cont = (u8 **)__pa_nodebug(&container);
127 patch = (u8 (*)[PATCH_MAX_SIZE])__pa_nodebug(&amd_ucode_patch);
126#else 128#else
127 new_rev = &ucode_new_rev; 129 new_rev = &ucode_new_rev;
128 cont_sz = &container_size; 130 cont_sz = &container_size;
129 cont = &container; 131 cont = &container;
132 patch = &amd_ucode_patch;
130#endif 133#endif
131 134
132 data = ucode; 135 data = ucode;
@@ -213,9 +216,9 @@ static void apply_ucode_in_initrd(void *ucode, size_t size)
213 rev = mc->hdr.patch_id; 216 rev = mc->hdr.patch_id;
214 *new_rev = rev; 217 *new_rev = rev;
215 218
216 /* save ucode patch */ 219 if (save_patch)
217 memcpy(amd_ucode_patch, mc, 220 memcpy(patch, mc,
218 min_t(u32, header[1], PATCH_MAX_SIZE)); 221 min_t(u32, header[1], PATCH_MAX_SIZE));
219 } 222 }
220 } 223 }
221 224
@@ -246,7 +249,7 @@ void __init load_ucode_amd_bsp(void)
246 *data = cp.data; 249 *data = cp.data;
247 *size = cp.size; 250 *size = cp.size;
248 251
249 apply_ucode_in_initrd(cp.data, cp.size); 252 apply_ucode_in_initrd(cp.data, cp.size, true);
250} 253}
251 254
252#ifdef CONFIG_X86_32 255#ifdef CONFIG_X86_32
@@ -263,7 +266,7 @@ void load_ucode_amd_ap(void)
263 size_t *usize; 266 size_t *usize;
264 void **ucode; 267 void **ucode;
265 268
266 mc = (struct microcode_amd *)__pa(amd_ucode_patch); 269 mc = (struct microcode_amd *)__pa_nodebug(amd_ucode_patch);
267 if (mc->hdr.patch_id && mc->hdr.processor_rev_id) { 270 if (mc->hdr.patch_id && mc->hdr.processor_rev_id) {
268 __apply_microcode_amd(mc); 271 __apply_microcode_amd(mc);
269 return; 272 return;
@@ -275,7 +278,7 @@ void load_ucode_amd_ap(void)
275 if (!*ucode || !*usize) 278 if (!*ucode || !*usize)
276 return; 279 return;
277 280
278 apply_ucode_in_initrd(*ucode, *usize); 281 apply_ucode_in_initrd(*ucode, *usize, false);
279} 282}
280 283
281static void __init collect_cpu_sig_on_bsp(void *arg) 284static void __init collect_cpu_sig_on_bsp(void *arg)
@@ -339,7 +342,7 @@ void load_ucode_amd_ap(void)
339 * AP has a different equivalence ID than BSP, looks like 342 * AP has a different equivalence ID than BSP, looks like
340 * mixed-steppings silicon so go through the ucode blob anew. 343 * mixed-steppings silicon so go through the ucode blob anew.
341 */ 344 */
342 apply_ucode_in_initrd(ucode_cpio.data, ucode_cpio.size); 345 apply_ucode_in_initrd(ucode_cpio.data, ucode_cpio.size, false);
343 } 346 }
344} 347}
345#endif 348#endif
@@ -347,7 +350,9 @@ void load_ucode_amd_ap(void)
347int __init save_microcode_in_initrd_amd(void) 350int __init save_microcode_in_initrd_amd(void)
348{ 351{
349 unsigned long cont; 352 unsigned long cont;
353 int retval = 0;
350 enum ucode_state ret; 354 enum ucode_state ret;
355 u8 *cont_va;
351 u32 eax; 356 u32 eax;
352 357
353 if (!container) 358 if (!container)
@@ -355,13 +360,15 @@ int __init save_microcode_in_initrd_amd(void)
355 360
356#ifdef CONFIG_X86_32 361#ifdef CONFIG_X86_32
357 get_bsp_sig(); 362 get_bsp_sig();
358 cont = (unsigned long)container; 363 cont = (unsigned long)container;
364 cont_va = __va(container);
359#else 365#else
360 /* 366 /*
361 * We need the physical address of the container for both bitness since 367 * We need the physical address of the container for both bitness since
362 * boot_params.hdr.ramdisk_image is a physical address. 368 * boot_params.hdr.ramdisk_image is a physical address.
363 */ 369 */
364 cont = __pa(container); 370 cont = __pa(container);
371 cont_va = container;
365#endif 372#endif
366 373
367 /* 374 /*
@@ -372,6 +379,8 @@ int __init save_microcode_in_initrd_amd(void)
372 if (relocated_ramdisk) 379 if (relocated_ramdisk)
373 container = (u8 *)(__va(relocated_ramdisk) + 380 container = (u8 *)(__va(relocated_ramdisk) +
374 (cont - boot_params.hdr.ramdisk_image)); 381 (cont - boot_params.hdr.ramdisk_image));
382 else
383 container = cont_va;
375 384
376 if (ucode_new_rev) 385 if (ucode_new_rev)
377 pr_info("microcode: updated early to new patch_level=0x%08x\n", 386 pr_info("microcode: updated early to new patch_level=0x%08x\n",
@@ -382,7 +391,7 @@ int __init save_microcode_in_initrd_amd(void)
382 391
383 ret = load_microcode_amd(eax, container, container_size); 392 ret = load_microcode_amd(eax, container, container_size);
384 if (ret != UCODE_OK) 393 if (ret != UCODE_OK)
385 return -EINVAL; 394 retval = -EINVAL;
386 395
387 /* 396 /*
388 * This will be freed any msec now, stash patches for the current 397 * This will be freed any msec now, stash patches for the current
@@ -391,5 +400,5 @@ int __init save_microcode_in_initrd_amd(void)
391 container = NULL; 400 container = NULL;
392 container_size = 0; 401 container_size = 0;
393 402
394 return 0; 403 return retval;
395} 404}
diff --git a/arch/x86/kernel/cpu/microcode/core.c b/arch/x86/kernel/cpu/microcode/core.c
index dd9d6190b08d..08fe6e8a726e 100644
--- a/arch/x86/kernel/cpu/microcode/core.c
+++ b/arch/x86/kernel/cpu/microcode/core.c
@@ -465,6 +465,16 @@ static void mc_bp_resume(void)
465 465
466 if (uci->valid && uci->mc) 466 if (uci->valid && uci->mc)
467 microcode_ops->apply_microcode(cpu); 467 microcode_ops->apply_microcode(cpu);
468#ifdef CONFIG_X86_64
469 else if (!uci->mc)
470 /*
471 * We might resume and not have applied late microcode but still
472 * have a newer patch stashed from the early loader. We don't
473 * have it in uci->mc so we have to load it the same way we're
474 * applying patches early on the APs.
475 */
476 load_ucode_ap();
477#endif
468} 478}
469 479
470static struct syscore_ops mc_syscore_ops = { 480static struct syscore_ops mc_syscore_ops = {
diff --git a/arch/x86/kernel/cpu/microcode/core_early.c b/arch/x86/kernel/cpu/microcode/core_early.c
index 5f28a64e71ea..2c017f242a78 100644
--- a/arch/x86/kernel/cpu/microcode/core_early.c
+++ b/arch/x86/kernel/cpu/microcode/core_early.c
@@ -124,7 +124,7 @@ void __init load_ucode_bsp(void)
124static bool check_loader_disabled_ap(void) 124static bool check_loader_disabled_ap(void)
125{ 125{
126#ifdef CONFIG_X86_32 126#ifdef CONFIG_X86_32
127 return __pa_nodebug(dis_ucode_ldr); 127 return *((bool *)__pa_nodebug(&dis_ucode_ldr));
128#else 128#else
129 return dis_ucode_ldr; 129 return dis_ucode_ldr;
130#endif 130#endif
diff --git a/arch/x86/kernel/cpu/perf_event.c b/arch/x86/kernel/cpu/perf_event.c
index 1b8299dd3d91..143e5f5dc855 100644
--- a/arch/x86/kernel/cpu/perf_event.c
+++ b/arch/x86/kernel/cpu/perf_event.c
@@ -243,8 +243,9 @@ static bool check_hw_exists(void)
243 243
244msr_fail: 244msr_fail:
245 printk(KERN_CONT "Broken PMU hardware detected, using software events only.\n"); 245 printk(KERN_CONT "Broken PMU hardware detected, using software events only.\n");
246 printk(boot_cpu_has(X86_FEATURE_HYPERVISOR) ? KERN_INFO : KERN_ERR 246 printk("%sFailed to access perfctr msr (MSR %x is %Lx)\n",
247 "Failed to access perfctr msr (MSR %x is %Lx)\n", reg, val_new); 247 boot_cpu_has(X86_FEATURE_HYPERVISOR) ? KERN_INFO : KERN_ERR,
248 reg, val_new);
248 249
249 return false; 250 return false;
250} 251}
@@ -444,12 +445,6 @@ int x86_pmu_hw_config(struct perf_event *event)
444 if (event->attr.type == PERF_TYPE_RAW) 445 if (event->attr.type == PERF_TYPE_RAW)
445 event->hw.config |= event->attr.config & X86_RAW_EVENT_MASK; 446 event->hw.config |= event->attr.config & X86_RAW_EVENT_MASK;
446 447
447 if (event->attr.sample_period && x86_pmu.limit_period) {
448 if (x86_pmu.limit_period(event, event->attr.sample_period) >
449 event->attr.sample_period)
450 return -EINVAL;
451 }
452
453 return x86_setup_perfctr(event); 448 return x86_setup_perfctr(event);
454} 449}
455 450
@@ -987,9 +982,6 @@ int x86_perf_event_set_period(struct perf_event *event)
987 if (left > x86_pmu.max_period) 982 if (left > x86_pmu.max_period)
988 left = x86_pmu.max_period; 983 left = x86_pmu.max_period;
989 984
990 if (x86_pmu.limit_period)
991 left = x86_pmu.limit_period(event, left);
992
993 per_cpu(pmc_prev_left[idx], smp_processor_id()) = left; 985 per_cpu(pmc_prev_left[idx], smp_processor_id()) = left;
994 986
995 /* 987 /*
diff --git a/arch/x86/kernel/cpu/perf_event.h b/arch/x86/kernel/cpu/perf_event.h
index d98a34d435d7..fc5eb390b368 100644
--- a/arch/x86/kernel/cpu/perf_event.h
+++ b/arch/x86/kernel/cpu/perf_event.h
@@ -445,7 +445,6 @@ struct x86_pmu {
445 struct x86_pmu_quirk *quirks; 445 struct x86_pmu_quirk *quirks;
446 int perfctr_second_write; 446 int perfctr_second_write;
447 bool late_ack; 447 bool late_ack;
448 unsigned (*limit_period)(struct perf_event *event, unsigned l);
449 448
450 /* 449 /*
451 * sysfs attrs 450 * sysfs attrs
diff --git a/arch/x86/kernel/cpu/perf_event_intel.c b/arch/x86/kernel/cpu/perf_event_intel.c
index a73947c53b65..944bf019b74f 100644
--- a/arch/x86/kernel/cpu/perf_event_intel.c
+++ b/arch/x86/kernel/cpu/perf_event_intel.c
@@ -220,15 +220,6 @@ static struct event_constraint intel_hsw_event_constraints[] = {
220 EVENT_CONSTRAINT_END 220 EVENT_CONSTRAINT_END
221}; 221};
222 222
223static struct event_constraint intel_bdw_event_constraints[] = {
224 FIXED_EVENT_CONSTRAINT(0x00c0, 0), /* INST_RETIRED.ANY */
225 FIXED_EVENT_CONSTRAINT(0x003c, 1), /* CPU_CLK_UNHALTED.CORE */
226 FIXED_EVENT_CONSTRAINT(0x0300, 2), /* CPU_CLK_UNHALTED.REF */
227 INTEL_UEVENT_CONSTRAINT(0x148, 0x4), /* L1D_PEND_MISS.PENDING */
228 INTEL_EVENT_CONSTRAINT(0xa3, 0x4), /* CYCLE_ACTIVITY.* */
229 EVENT_CONSTRAINT_END
230};
231
232static u64 intel_pmu_event_map(int hw_event) 223static u64 intel_pmu_event_map(int hw_event)
233{ 224{
234 return intel_perfmon_event_map[hw_event]; 225 return intel_perfmon_event_map[hw_event];
@@ -424,126 +415,6 @@ static __initconst const u64 snb_hw_cache_event_ids
424 415
425}; 416};
426 417
427static __initconst const u64 hsw_hw_cache_event_ids
428 [PERF_COUNT_HW_CACHE_MAX]
429 [PERF_COUNT_HW_CACHE_OP_MAX]
430 [PERF_COUNT_HW_CACHE_RESULT_MAX] =
431{
432 [ C(L1D ) ] = {
433 [ C(OP_READ) ] = {
434 [ C(RESULT_ACCESS) ] = 0x81d0, /* MEM_UOPS_RETIRED.ALL_LOADS */
435 [ C(RESULT_MISS) ] = 0x151, /* L1D.REPLACEMENT */
436 },
437 [ C(OP_WRITE) ] = {
438 [ C(RESULT_ACCESS) ] = 0x82d0, /* MEM_UOPS_RETIRED.ALL_STORES */
439 [ C(RESULT_MISS) ] = 0x0,
440 },
441 [ C(OP_PREFETCH) ] = {
442 [ C(RESULT_ACCESS) ] = 0x0,
443 [ C(RESULT_MISS) ] = 0x0,
444 },
445 },
446 [ C(L1I ) ] = {
447 [ C(OP_READ) ] = {
448 [ C(RESULT_ACCESS) ] = 0x0,
449 [ C(RESULT_MISS) ] = 0x280, /* ICACHE.MISSES */
450 },
451 [ C(OP_WRITE) ] = {
452 [ C(RESULT_ACCESS) ] = -1,
453 [ C(RESULT_MISS) ] = -1,
454 },
455 [ C(OP_PREFETCH) ] = {
456 [ C(RESULT_ACCESS) ] = 0x0,
457 [ C(RESULT_MISS) ] = 0x0,
458 },
459 },
460 [ C(LL ) ] = {
461 [ C(OP_READ) ] = {
462 /* OFFCORE_RESPONSE:ALL_DATA_RD|ALL_CODE_RD */
463 [ C(RESULT_ACCESS) ] = 0x1b7,
464 /* OFFCORE_RESPONSE:ALL_DATA_RD|ALL_CODE_RD|SUPPLIER_NONE|
465 L3_MISS|ANY_SNOOP */
466 [ C(RESULT_MISS) ] = 0x1b7,
467 },
468 [ C(OP_WRITE) ] = {
469 [ C(RESULT_ACCESS) ] = 0x1b7, /* OFFCORE_RESPONSE:ALL_RFO */
470 /* OFFCORE_RESPONSE:ALL_RFO|SUPPLIER_NONE|L3_MISS|ANY_SNOOP */
471 [ C(RESULT_MISS) ] = 0x1b7,
472 },
473 [ C(OP_PREFETCH) ] = {
474 [ C(RESULT_ACCESS) ] = 0x0,
475 [ C(RESULT_MISS) ] = 0x0,
476 },
477 },
478 [ C(DTLB) ] = {
479 [ C(OP_READ) ] = {
480 [ C(RESULT_ACCESS) ] = 0x81d0, /* MEM_UOPS_RETIRED.ALL_LOADS */
481 [ C(RESULT_MISS) ] = 0x108, /* DTLB_LOAD_MISSES.MISS_CAUSES_A_WALK */
482 },
483 [ C(OP_WRITE) ] = {
484 [ C(RESULT_ACCESS) ] = 0x82d0, /* MEM_UOPS_RETIRED.ALL_STORES */
485 [ C(RESULT_MISS) ] = 0x149, /* DTLB_STORE_MISSES.MISS_CAUSES_A_WALK */
486 },
487 [ C(OP_PREFETCH) ] = {
488 [ C(RESULT_ACCESS) ] = 0x0,
489 [ C(RESULT_MISS) ] = 0x0,
490 },
491 },
492 [ C(ITLB) ] = {
493 [ C(OP_READ) ] = {
494 [ C(RESULT_ACCESS) ] = 0x6085, /* ITLB_MISSES.STLB_HIT */
495 [ C(RESULT_MISS) ] = 0x185, /* ITLB_MISSES.MISS_CAUSES_A_WALK */
496 },
497 [ C(OP_WRITE) ] = {
498 [ C(RESULT_ACCESS) ] = -1,
499 [ C(RESULT_MISS) ] = -1,
500 },
501 [ C(OP_PREFETCH) ] = {
502 [ C(RESULT_ACCESS) ] = -1,
503 [ C(RESULT_MISS) ] = -1,
504 },
505 },
506 [ C(BPU ) ] = {
507 [ C(OP_READ) ] = {
508 [ C(RESULT_ACCESS) ] = 0xc4, /* BR_INST_RETIRED.ALL_BRANCHES */
509 [ C(RESULT_MISS) ] = 0xc5, /* BR_MISP_RETIRED.ALL_BRANCHES */
510 },
511 [ C(OP_WRITE) ] = {
512 [ C(RESULT_ACCESS) ] = -1,
513 [ C(RESULT_MISS) ] = -1,
514 },
515 [ C(OP_PREFETCH) ] = {
516 [ C(RESULT_ACCESS) ] = -1,
517 [ C(RESULT_MISS) ] = -1,
518 },
519 },
520};
521
522static __initconst const u64 hsw_hw_cache_extra_regs
523 [PERF_COUNT_HW_CACHE_MAX]
524 [PERF_COUNT_HW_CACHE_OP_MAX]
525 [PERF_COUNT_HW_CACHE_RESULT_MAX] =
526{
527 [ C(LL ) ] = {
528 [ C(OP_READ) ] = {
529 /* OFFCORE_RESPONSE:ALL_DATA_RD|ALL_CODE_RD */
530 [ C(RESULT_ACCESS) ] = 0x2d5,
531 /* OFFCORE_RESPONSE:ALL_DATA_RD|ALL_CODE_RD|SUPPLIER_NONE|
532 L3_MISS|ANY_SNOOP */
533 [ C(RESULT_MISS) ] = 0x3fbc0202d5ull,
534 },
535 [ C(OP_WRITE) ] = {
536 [ C(RESULT_ACCESS) ] = 0x122, /* OFFCORE_RESPONSE:ALL_RFO */
537 /* OFFCORE_RESPONSE:ALL_RFO|SUPPLIER_NONE|L3_MISS|ANY_SNOOP */
538 [ C(RESULT_MISS) ] = 0x3fbc020122ull,
539 },
540 [ C(OP_PREFETCH) ] = {
541 [ C(RESULT_ACCESS) ] = 0x0,
542 [ C(RESULT_MISS) ] = 0x0,
543 },
544 },
545};
546
547static __initconst const u64 westmere_hw_cache_event_ids 418static __initconst const u64 westmere_hw_cache_event_ids
548 [PERF_COUNT_HW_CACHE_MAX] 419 [PERF_COUNT_HW_CACHE_MAX]
549 [PERF_COUNT_HW_CACHE_OP_MAX] 420 [PERF_COUNT_HW_CACHE_OP_MAX]
@@ -2034,24 +1905,6 @@ hsw_get_event_constraints(struct cpu_hw_events *cpuc, struct perf_event *event)
2034 return c; 1905 return c;
2035} 1906}
2036 1907
2037/*
2038 * Broadwell:
2039 * The INST_RETIRED.ALL period always needs to have lowest
2040 * 6bits cleared (BDM57). It shall not use a period smaller
2041 * than 100 (BDM11). We combine the two to enforce
2042 * a min-period of 128.
2043 */
2044static unsigned bdw_limit_period(struct perf_event *event, unsigned left)
2045{
2046 if ((event->hw.config & INTEL_ARCH_EVENT_MASK) ==
2047 X86_CONFIG(.event=0xc0, .umask=0x01)) {
2048 if (left < 128)
2049 left = 128;
2050 left &= ~0x3fu;
2051 }
2052 return left;
2053}
2054
2055PMU_FORMAT_ATTR(event, "config:0-7" ); 1908PMU_FORMAT_ATTR(event, "config:0-7" );
2056PMU_FORMAT_ATTR(umask, "config:8-15" ); 1909PMU_FORMAT_ATTR(umask, "config:8-15" );
2057PMU_FORMAT_ATTR(edge, "config:18" ); 1910PMU_FORMAT_ATTR(edge, "config:18" );
@@ -2692,8 +2545,8 @@ __init int intel_pmu_init(void)
2692 case 69: /* 22nm Haswell ULT */ 2545 case 69: /* 22nm Haswell ULT */
2693 case 70: /* 22nm Haswell + GT3e (Intel Iris Pro graphics) */ 2546 case 70: /* 22nm Haswell + GT3e (Intel Iris Pro graphics) */
2694 x86_pmu.late_ack = true; 2547 x86_pmu.late_ack = true;
2695 memcpy(hw_cache_event_ids, hsw_hw_cache_event_ids, sizeof(hw_cache_event_ids)); 2548 memcpy(hw_cache_event_ids, snb_hw_cache_event_ids, sizeof(hw_cache_event_ids));
2696 memcpy(hw_cache_extra_regs, hsw_hw_cache_extra_regs, sizeof(hw_cache_extra_regs)); 2549 memcpy(hw_cache_extra_regs, snb_hw_cache_extra_regs, sizeof(hw_cache_extra_regs));
2697 2550
2698 intel_pmu_lbr_init_snb(); 2551 intel_pmu_lbr_init_snb();
2699 2552
@@ -2712,28 +2565,6 @@ __init int intel_pmu_init(void)
2712 pr_cont("Haswell events, "); 2565 pr_cont("Haswell events, ");
2713 break; 2566 break;
2714 2567
2715 case 61: /* 14nm Broadwell Core-M */
2716 x86_pmu.late_ack = true;
2717 memcpy(hw_cache_event_ids, hsw_hw_cache_event_ids, sizeof(hw_cache_event_ids));
2718 memcpy(hw_cache_extra_regs, hsw_hw_cache_extra_regs, sizeof(hw_cache_extra_regs));
2719
2720 intel_pmu_lbr_init_snb();
2721
2722 x86_pmu.event_constraints = intel_bdw_event_constraints;
2723 x86_pmu.pebs_constraints = intel_hsw_pebs_event_constraints;
2724 x86_pmu.extra_regs = intel_snbep_extra_regs;
2725 x86_pmu.pebs_aliases = intel_pebs_aliases_snb;
2726 /* all extra regs are per-cpu when HT is on */
2727 x86_pmu.er_flags |= ERF_HAS_RSP_1;
2728 x86_pmu.er_flags |= ERF_NO_HT_SHARING;
2729
2730 x86_pmu.hw_config = hsw_hw_config;
2731 x86_pmu.get_event_constraints = hsw_get_event_constraints;
2732 x86_pmu.cpu_events = hsw_events_attrs;
2733 x86_pmu.limit_period = bdw_limit_period;
2734 pr_cont("Broadwell events, ");
2735 break;
2736
2737 default: 2568 default:
2738 switch (x86_pmu.version) { 2569 switch (x86_pmu.version) {
2739 case 1: 2570 case 1:
diff --git a/arch/x86/kernel/cpu/perf_event_intel_uncore_snbep.c b/arch/x86/kernel/cpu/perf_event_intel_uncore_snbep.c
index adf138eac85c..f9ed429d6e4f 100644
--- a/arch/x86/kernel/cpu/perf_event_intel_uncore_snbep.c
+++ b/arch/x86/kernel/cpu/perf_event_intel_uncore_snbep.c
@@ -486,14 +486,17 @@ static struct attribute_group snbep_uncore_qpi_format_group = {
486 .attrs = snbep_uncore_qpi_formats_attr, 486 .attrs = snbep_uncore_qpi_formats_attr,
487}; 487};
488 488
489#define SNBEP_UNCORE_MSR_OPS_COMMON_INIT() \ 489#define __SNBEP_UNCORE_MSR_OPS_COMMON_INIT() \
490 .init_box = snbep_uncore_msr_init_box, \
491 .disable_box = snbep_uncore_msr_disable_box, \ 490 .disable_box = snbep_uncore_msr_disable_box, \
492 .enable_box = snbep_uncore_msr_enable_box, \ 491 .enable_box = snbep_uncore_msr_enable_box, \
493 .disable_event = snbep_uncore_msr_disable_event, \ 492 .disable_event = snbep_uncore_msr_disable_event, \
494 .enable_event = snbep_uncore_msr_enable_event, \ 493 .enable_event = snbep_uncore_msr_enable_event, \
495 .read_counter = uncore_msr_read_counter 494 .read_counter = uncore_msr_read_counter
496 495
496#define SNBEP_UNCORE_MSR_OPS_COMMON_INIT() \
497 __SNBEP_UNCORE_MSR_OPS_COMMON_INIT(), \
498 .init_box = snbep_uncore_msr_init_box \
499
497static struct intel_uncore_ops snbep_uncore_msr_ops = { 500static struct intel_uncore_ops snbep_uncore_msr_ops = {
498 SNBEP_UNCORE_MSR_OPS_COMMON_INIT(), 501 SNBEP_UNCORE_MSR_OPS_COMMON_INIT(),
499}; 502};
@@ -1919,6 +1922,30 @@ static struct intel_uncore_type hswep_uncore_cbox = {
1919 .format_group = &hswep_uncore_cbox_format_group, 1922 .format_group = &hswep_uncore_cbox_format_group,
1920}; 1923};
1921 1924
1925/*
1926 * Write SBOX Initialization register bit by bit to avoid spurious #GPs
1927 */
1928static void hswep_uncore_sbox_msr_init_box(struct intel_uncore_box *box)
1929{
1930 unsigned msr = uncore_msr_box_ctl(box);
1931
1932 if (msr) {
1933 u64 init = SNBEP_PMON_BOX_CTL_INT;
1934 u64 flags = 0;
1935 int i;
1936
1937 for_each_set_bit(i, (unsigned long *)&init, 64) {
1938 flags |= (1ULL << i);
1939 wrmsrl(msr, flags);
1940 }
1941 }
1942}
1943
1944static struct intel_uncore_ops hswep_uncore_sbox_msr_ops = {
1945 __SNBEP_UNCORE_MSR_OPS_COMMON_INIT(),
1946 .init_box = hswep_uncore_sbox_msr_init_box
1947};
1948
1922static struct attribute *hswep_uncore_sbox_formats_attr[] = { 1949static struct attribute *hswep_uncore_sbox_formats_attr[] = {
1923 &format_attr_event.attr, 1950 &format_attr_event.attr,
1924 &format_attr_umask.attr, 1951 &format_attr_umask.attr,
@@ -1944,7 +1971,7 @@ static struct intel_uncore_type hswep_uncore_sbox = {
1944 .event_mask = HSWEP_S_MSR_PMON_RAW_EVENT_MASK, 1971 .event_mask = HSWEP_S_MSR_PMON_RAW_EVENT_MASK,
1945 .box_ctl = HSWEP_S0_MSR_PMON_BOX_CTL, 1972 .box_ctl = HSWEP_S0_MSR_PMON_BOX_CTL,
1946 .msr_offset = HSWEP_SBOX_MSR_OFFSET, 1973 .msr_offset = HSWEP_SBOX_MSR_OFFSET,
1947 .ops = &snbep_uncore_msr_ops, 1974 .ops = &hswep_uncore_sbox_msr_ops,
1948 .format_group = &hswep_uncore_sbox_format_group, 1975 .format_group = &hswep_uncore_sbox_format_group,
1949}; 1976};
1950 1977
@@ -2025,13 +2052,27 @@ static struct intel_uncore_type hswep_uncore_imc = {
2025 SNBEP_UNCORE_PCI_COMMON_INIT(), 2052 SNBEP_UNCORE_PCI_COMMON_INIT(),
2026}; 2053};
2027 2054
2055static unsigned hswep_uncore_irp_ctrs[] = {0xa0, 0xa8, 0xb0, 0xb8};
2056
2057static u64 hswep_uncore_irp_read_counter(struct intel_uncore_box *box, struct perf_event *event)
2058{
2059 struct pci_dev *pdev = box->pci_dev;
2060 struct hw_perf_event *hwc = &event->hw;
2061 u64 count = 0;
2062
2063 pci_read_config_dword(pdev, hswep_uncore_irp_ctrs[hwc->idx], (u32 *)&count);
2064 pci_read_config_dword(pdev, hswep_uncore_irp_ctrs[hwc->idx] + 4, (u32 *)&count + 1);
2065
2066 return count;
2067}
2068
2028static struct intel_uncore_ops hswep_uncore_irp_ops = { 2069static struct intel_uncore_ops hswep_uncore_irp_ops = {
2029 .init_box = snbep_uncore_pci_init_box, 2070 .init_box = snbep_uncore_pci_init_box,
2030 .disable_box = snbep_uncore_pci_disable_box, 2071 .disable_box = snbep_uncore_pci_disable_box,
2031 .enable_box = snbep_uncore_pci_enable_box, 2072 .enable_box = snbep_uncore_pci_enable_box,
2032 .disable_event = ivbep_uncore_irp_disable_event, 2073 .disable_event = ivbep_uncore_irp_disable_event,
2033 .enable_event = ivbep_uncore_irp_enable_event, 2074 .enable_event = ivbep_uncore_irp_enable_event,
2034 .read_counter = ivbep_uncore_irp_read_counter, 2075 .read_counter = hswep_uncore_irp_read_counter,
2035}; 2076};
2036 2077
2037static struct intel_uncore_type hswep_uncore_irp = { 2078static struct intel_uncore_type hswep_uncore_irp = {
diff --git a/arch/x86/kernel/dumpstack_64.c b/arch/x86/kernel/dumpstack_64.c
index 1abcb50b48ae..ff86f19b5758 100644
--- a/arch/x86/kernel/dumpstack_64.c
+++ b/arch/x86/kernel/dumpstack_64.c
@@ -24,7 +24,6 @@ static char x86_stack_ids[][8] = {
24 [ DEBUG_STACK-1 ] = "#DB", 24 [ DEBUG_STACK-1 ] = "#DB",
25 [ NMI_STACK-1 ] = "NMI", 25 [ NMI_STACK-1 ] = "NMI",
26 [ DOUBLEFAULT_STACK-1 ] = "#DF", 26 [ DOUBLEFAULT_STACK-1 ] = "#DF",
27 [ STACKFAULT_STACK-1 ] = "#SS",
28 [ MCE_STACK-1 ] = "#MC", 27 [ MCE_STACK-1 ] = "#MC",
29#if DEBUG_STKSZ > EXCEPTION_STKSZ 28#if DEBUG_STKSZ > EXCEPTION_STKSZ
30 [ N_EXCEPTION_STACKS ... 29 [ N_EXCEPTION_STACKS ...
diff --git a/arch/x86/kernel/entry_32.S b/arch/x86/kernel/entry_32.S
index b553ed89e5f5..344b63f18d14 100644
--- a/arch/x86/kernel/entry_32.S
+++ b/arch/x86/kernel/entry_32.S
@@ -447,15 +447,14 @@ sysenter_exit:
447sysenter_audit: 447sysenter_audit:
448 testl $(_TIF_WORK_SYSCALL_ENTRY & ~_TIF_SYSCALL_AUDIT),TI_flags(%ebp) 448 testl $(_TIF_WORK_SYSCALL_ENTRY & ~_TIF_SYSCALL_AUDIT),TI_flags(%ebp)
449 jnz syscall_trace_entry 449 jnz syscall_trace_entry
450 addl $4,%esp 450 /* movl PT_EAX(%esp), %eax already set, syscall number: 1st arg to audit */
451 CFI_ADJUST_CFA_OFFSET -4 451 movl PT_EBX(%esp), %edx /* ebx/a0: 2nd arg to audit */
452 movl %esi,4(%esp) /* 5th arg: 4th syscall arg */ 452 /* movl PT_ECX(%esp), %ecx already set, a1: 3nd arg to audit */
453 movl %edx,(%esp) /* 4th arg: 3rd syscall arg */ 453 pushl_cfi PT_ESI(%esp) /* a3: 5th arg */
454 /* %ecx already in %ecx 3rd arg: 2nd syscall arg */ 454 pushl_cfi PT_EDX+4(%esp) /* a2: 4th arg */
455 movl %ebx,%edx /* 2nd arg: 1st syscall arg */
456 /* %eax already in %eax 1st arg: syscall number */
457 call __audit_syscall_entry 455 call __audit_syscall_entry
458 pushl_cfi %ebx 456 popl_cfi %ecx /* get that remapped edx off the stack */
457 popl_cfi %ecx /* get that remapped esi off the stack */
459 movl PT_EAX(%esp),%eax /* reload syscall number */ 458 movl PT_EAX(%esp),%eax /* reload syscall number */
460 jmp sysenter_do_call 459 jmp sysenter_do_call
461 460
diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S
index df088bb03fb3..c0226ab54106 100644
--- a/arch/x86/kernel/entry_64.S
+++ b/arch/x86/kernel/entry_64.S
@@ -828,9 +828,15 @@ ENTRY(native_iret)
828 jnz native_irq_return_ldt 828 jnz native_irq_return_ldt
829#endif 829#endif
830 830
831.global native_irq_return_iret
831native_irq_return_iret: 832native_irq_return_iret:
833 /*
834 * This may fault. Non-paranoid faults on return to userspace are
835 * handled by fixup_bad_iret. These include #SS, #GP, and #NP.
836 * Double-faults due to espfix64 are handled in do_double_fault.
837 * Other faults here are fatal.
838 */
832 iretq 839 iretq
833 _ASM_EXTABLE(native_irq_return_iret, bad_iret)
834 840
835#ifdef CONFIG_X86_ESPFIX64 841#ifdef CONFIG_X86_ESPFIX64
836native_irq_return_ldt: 842native_irq_return_ldt:
@@ -858,25 +864,6 @@ native_irq_return_ldt:
858 jmp native_irq_return_iret 864 jmp native_irq_return_iret
859#endif 865#endif
860 866
861 .section .fixup,"ax"
862bad_iret:
863 /*
864 * The iret traps when the %cs or %ss being restored is bogus.
865 * We've lost the original trap vector and error code.
866 * #GPF is the most likely one to get for an invalid selector.
867 * So pretend we completed the iret and took the #GPF in user mode.
868 *
869 * We are now running with the kernel GS after exception recovery.
870 * But error_entry expects us to have user GS to match the user %cs,
871 * so swap back.
872 */
873 pushq $0
874
875 SWAPGS
876 jmp general_protection
877
878 .previous
879
880 /* edi: workmask, edx: work */ 867 /* edi: workmask, edx: work */
881retint_careful: 868retint_careful:
882 CFI_RESTORE_STATE 869 CFI_RESTORE_STATE
@@ -922,37 +909,6 @@ ENTRY(retint_kernel)
922 CFI_ENDPROC 909 CFI_ENDPROC
923END(common_interrupt) 910END(common_interrupt)
924 911
925 /*
926 * If IRET takes a fault on the espfix stack, then we
927 * end up promoting it to a doublefault. In that case,
928 * modify the stack to make it look like we just entered
929 * the #GP handler from user space, similar to bad_iret.
930 */
931#ifdef CONFIG_X86_ESPFIX64
932 ALIGN
933__do_double_fault:
934 XCPT_FRAME 1 RDI+8
935 movq RSP(%rdi),%rax /* Trap on the espfix stack? */
936 sarq $PGDIR_SHIFT,%rax
937 cmpl $ESPFIX_PGD_ENTRY,%eax
938 jne do_double_fault /* No, just deliver the fault */
939 cmpl $__KERNEL_CS,CS(%rdi)
940 jne do_double_fault
941 movq RIP(%rdi),%rax
942 cmpq $native_irq_return_iret,%rax
943 jne do_double_fault /* This shouldn't happen... */
944 movq PER_CPU_VAR(kernel_stack),%rax
945 subq $(6*8-KERNEL_STACK_OFFSET),%rax /* Reset to original stack */
946 movq %rax,RSP(%rdi)
947 movq $0,(%rax) /* Missing (lost) #GP error code */
948 movq $general_protection,RIP(%rdi)
949 retq
950 CFI_ENDPROC
951END(__do_double_fault)
952#else
953# define __do_double_fault do_double_fault
954#endif
955
956/* 912/*
957 * APIC interrupts. 913 * APIC interrupts.
958 */ 914 */
@@ -1124,7 +1080,7 @@ idtentry overflow do_overflow has_error_code=0
1124idtentry bounds do_bounds has_error_code=0 1080idtentry bounds do_bounds has_error_code=0
1125idtentry invalid_op do_invalid_op has_error_code=0 1081idtentry invalid_op do_invalid_op has_error_code=0
1126idtentry device_not_available do_device_not_available has_error_code=0 1082idtentry device_not_available do_device_not_available has_error_code=0
1127idtentry double_fault __do_double_fault has_error_code=1 paranoid=1 1083idtentry double_fault do_double_fault has_error_code=1 paranoid=1
1128idtentry coprocessor_segment_overrun do_coprocessor_segment_overrun has_error_code=0 1084idtentry coprocessor_segment_overrun do_coprocessor_segment_overrun has_error_code=0
1129idtentry invalid_TSS do_invalid_TSS has_error_code=1 1085idtentry invalid_TSS do_invalid_TSS has_error_code=1
1130idtentry segment_not_present do_segment_not_present has_error_code=1 1086idtentry segment_not_present do_segment_not_present has_error_code=1
@@ -1289,7 +1245,7 @@ apicinterrupt3 HYPERVISOR_CALLBACK_VECTOR \
1289 1245
1290idtentry debug do_debug has_error_code=0 paranoid=1 shift_ist=DEBUG_STACK 1246idtentry debug do_debug has_error_code=0 paranoid=1 shift_ist=DEBUG_STACK
1291idtentry int3 do_int3 has_error_code=0 paranoid=1 shift_ist=DEBUG_STACK 1247idtentry int3 do_int3 has_error_code=0 paranoid=1 shift_ist=DEBUG_STACK
1292idtentry stack_segment do_stack_segment has_error_code=1 paranoid=1 1248idtentry stack_segment do_stack_segment has_error_code=1
1293#ifdef CONFIG_XEN 1249#ifdef CONFIG_XEN
1294idtentry xen_debug do_debug has_error_code=0 1250idtentry xen_debug do_debug has_error_code=0
1295idtentry xen_int3 do_int3 has_error_code=0 1251idtentry xen_int3 do_int3 has_error_code=0
@@ -1399,17 +1355,16 @@ error_sti:
1399 1355
1400/* 1356/*
1401 * There are two places in the kernel that can potentially fault with 1357 * There are two places in the kernel that can potentially fault with
1402 * usergs. Handle them here. The exception handlers after iret run with 1358 * usergs. Handle them here. B stepping K8s sometimes report a
1403 * kernel gs again, so don't set the user space flag. B stepping K8s 1359 * truncated RIP for IRET exceptions returning to compat mode. Check
1404 * sometimes report an truncated RIP for IRET exceptions returning to 1360 * for these here too.
1405 * compat mode. Check for these here too.
1406 */ 1361 */
1407error_kernelspace: 1362error_kernelspace:
1408 CFI_REL_OFFSET rcx, RCX+8 1363 CFI_REL_OFFSET rcx, RCX+8
1409 incl %ebx 1364 incl %ebx
1410 leaq native_irq_return_iret(%rip),%rcx 1365 leaq native_irq_return_iret(%rip),%rcx
1411 cmpq %rcx,RIP+8(%rsp) 1366 cmpq %rcx,RIP+8(%rsp)
1412 je error_swapgs 1367 je error_bad_iret
1413 movl %ecx,%eax /* zero extend */ 1368 movl %ecx,%eax /* zero extend */
1414 cmpq %rax,RIP+8(%rsp) 1369 cmpq %rax,RIP+8(%rsp)
1415 je bstep_iret 1370 je bstep_iret
@@ -1420,7 +1375,15 @@ error_kernelspace:
1420bstep_iret: 1375bstep_iret:
1421 /* Fix truncated RIP */ 1376 /* Fix truncated RIP */
1422 movq %rcx,RIP+8(%rsp) 1377 movq %rcx,RIP+8(%rsp)
1423 jmp error_swapgs 1378 /* fall through */
1379
1380error_bad_iret:
1381 SWAPGS
1382 mov %rsp,%rdi
1383 call fixup_bad_iret
1384 mov %rax,%rsp
1385 decl %ebx /* Return to usergs */
1386 jmp error_sti
1424 CFI_ENDPROC 1387 CFI_ENDPROC
1425END(error_entry) 1388END(error_entry)
1426 1389
diff --git a/arch/x86/kernel/i8259.c b/arch/x86/kernel/i8259.c
index 8af817105e29..e7cc5370cd2f 100644
--- a/arch/x86/kernel/i8259.c
+++ b/arch/x86/kernel/i8259.c
@@ -111,8 +111,7 @@ static void make_8259A_irq(unsigned int irq)
111{ 111{
112 disable_irq_nosync(irq); 112 disable_irq_nosync(irq);
113 io_apic_irqs &= ~(1<<irq); 113 io_apic_irqs &= ~(1<<irq);
114 irq_set_chip_and_handler_name(irq, &i8259A_chip, handle_level_irq, 114 irq_set_chip_and_handler(irq, &i8259A_chip, handle_level_irq);
115 i8259A_chip.name);
116 enable_irq(irq); 115 enable_irq(irq);
117} 116}
118 117
diff --git a/arch/x86/kernel/irqinit.c b/arch/x86/kernel/irqinit.c
index 44f1ed42fdf2..4de73ee78361 100644
--- a/arch/x86/kernel/irqinit.c
+++ b/arch/x86/kernel/irqinit.c
@@ -70,7 +70,6 @@ int vector_used_by_percpu_irq(unsigned int vector)
70void __init init_ISA_irqs(void) 70void __init init_ISA_irqs(void)
71{ 71{
72 struct irq_chip *chip = legacy_pic->chip; 72 struct irq_chip *chip = legacy_pic->chip;
73 const char *name = chip->name;
74 int i; 73 int i;
75 74
76#if defined(CONFIG_X86_64) || defined(CONFIG_X86_LOCAL_APIC) 75#if defined(CONFIG_X86_64) || defined(CONFIG_X86_LOCAL_APIC)
@@ -79,7 +78,7 @@ void __init init_ISA_irqs(void)
79 legacy_pic->init(0); 78 legacy_pic->init(0);
80 79
81 for (i = 0; i < nr_legacy_irqs(); i++) 80 for (i = 0; i < nr_legacy_irqs(); i++)
82 irq_set_chip_and_handler_name(i, chip, handle_level_irq, name); 81 irq_set_chip_and_handler(i, chip, handle_level_irq);
83} 82}
84 83
85void __init init_IRQ(void) 84void __init init_IRQ(void)
diff --git a/arch/x86/kernel/ptrace.c b/arch/x86/kernel/ptrace.c
index 749b0e423419..e510618b2e91 100644
--- a/arch/x86/kernel/ptrace.c
+++ b/arch/x86/kernel/ptrace.c
@@ -1484,7 +1484,7 @@ unsigned long syscall_trace_enter_phase1(struct pt_regs *regs, u32 arch)
1484 */ 1484 */
1485 if (work & _TIF_NOHZ) { 1485 if (work & _TIF_NOHZ) {
1486 user_exit(); 1486 user_exit();
1487 work &= ~TIF_NOHZ; 1487 work &= ~_TIF_NOHZ;
1488 } 1488 }
1489 1489
1490#ifdef CONFIG_SECCOMP 1490#ifdef CONFIG_SECCOMP
diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c
index 235cfd39e0d7..ab08aa2276fb 100644
--- a/arch/x86/kernel/setup.c
+++ b/arch/x86/kernel/setup.c
@@ -1128,7 +1128,6 @@ void __init setup_arch(char **cmdline_p)
1128 setup_real_mode(); 1128 setup_real_mode();
1129 1129
1130 memblock_set_current_limit(get_max_mapped()); 1130 memblock_set_current_limit(get_max_mapped());
1131 dma_contiguous_reserve(max_pfn_mapped << PAGE_SHIFT);
1132 1131
1133 /* 1132 /*
1134 * NOTE: On x86-32, only from this point on, fixmaps are ready for use. 1133 * NOTE: On x86-32, only from this point on, fixmaps are ready for use.
@@ -1159,6 +1158,7 @@ void __init setup_arch(char **cmdline_p)
1159 early_acpi_boot_init(); 1158 early_acpi_boot_init();
1160 1159
1161 initmem_init(); 1160 initmem_init();
1161 dma_contiguous_reserve(max_pfn_mapped << PAGE_SHIFT);
1162 1162
1163 /* 1163 /*
1164 * Reserve memory for crash kernel after SRAT is parsed so that it 1164 * Reserve memory for crash kernel after SRAT is parsed so that it
diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c
index 2d5200e56357..668d8f2a8781 100644
--- a/arch/x86/kernel/smpboot.c
+++ b/arch/x86/kernel/smpboot.c
@@ -102,8 +102,6 @@ DEFINE_PER_CPU_READ_MOSTLY(cpumask_var_t, cpu_llc_shared_map);
102DEFINE_PER_CPU_SHARED_ALIGNED(struct cpuinfo_x86, cpu_info); 102DEFINE_PER_CPU_SHARED_ALIGNED(struct cpuinfo_x86, cpu_info);
103EXPORT_PER_CPU_SYMBOL(cpu_info); 103EXPORT_PER_CPU_SYMBOL(cpu_info);
104 104
105static DEFINE_PER_CPU(struct completion, die_complete);
106
107atomic_t init_deasserted; 105atomic_t init_deasserted;
108 106
109/* 107/*
@@ -1305,10 +1303,14 @@ static void __ref remove_cpu_from_maps(int cpu)
1305 numa_remove_cpu(cpu); 1303 numa_remove_cpu(cpu);
1306} 1304}
1307 1305
1306static DEFINE_PER_CPU(struct completion, die_complete);
1307
1308void cpu_disable_common(void) 1308void cpu_disable_common(void)
1309{ 1309{
1310 int cpu = smp_processor_id(); 1310 int cpu = smp_processor_id();
1311 1311
1312 init_completion(&per_cpu(die_complete, smp_processor_id()));
1313
1312 remove_siblinginfo(cpu); 1314 remove_siblinginfo(cpu);
1313 1315
1314 /* It's now safe to remove this processor from the online map */ 1316 /* It's now safe to remove this processor from the online map */
@@ -1327,16 +1329,21 @@ int native_cpu_disable(void)
1327 return ret; 1329 return ret;
1328 1330
1329 clear_local_APIC(); 1331 clear_local_APIC();
1330 init_completion(&per_cpu(die_complete, smp_processor_id()));
1331 cpu_disable_common(); 1332 cpu_disable_common();
1332 1333
1333 return 0; 1334 return 0;
1334} 1335}
1335 1336
1337void cpu_die_common(unsigned int cpu)
1338{
1339 wait_for_completion_timeout(&per_cpu(die_complete, cpu), HZ);
1340}
1341
1336void native_cpu_die(unsigned int cpu) 1342void native_cpu_die(unsigned int cpu)
1337{ 1343{
1338 /* We don't do anything here: idle task is faking death itself. */ 1344 /* We don't do anything here: idle task is faking death itself. */
1339 wait_for_completion_timeout(&per_cpu(die_complete, cpu), HZ); 1345
1346 cpu_die_common(cpu);
1340 1347
1341 /* They ack this in play_dead() by setting CPU_DEAD */ 1348 /* They ack this in play_dead() by setting CPU_DEAD */
1342 if (per_cpu(cpu_state, cpu) == CPU_DEAD) { 1349 if (per_cpu(cpu_state, cpu) == CPU_DEAD) {
diff --git a/arch/x86/kernel/traps.c b/arch/x86/kernel/traps.c
index 0d0e922fafc1..de801f22128a 100644
--- a/arch/x86/kernel/traps.c
+++ b/arch/x86/kernel/traps.c
@@ -233,32 +233,40 @@ DO_ERROR(X86_TRAP_UD, SIGILL, "invalid opcode", invalid_op)
233DO_ERROR(X86_TRAP_OLD_MF, SIGFPE, "coprocessor segment overrun",coprocessor_segment_overrun) 233DO_ERROR(X86_TRAP_OLD_MF, SIGFPE, "coprocessor segment overrun",coprocessor_segment_overrun)
234DO_ERROR(X86_TRAP_TS, SIGSEGV, "invalid TSS", invalid_TSS) 234DO_ERROR(X86_TRAP_TS, SIGSEGV, "invalid TSS", invalid_TSS)
235DO_ERROR(X86_TRAP_NP, SIGBUS, "segment not present", segment_not_present) 235DO_ERROR(X86_TRAP_NP, SIGBUS, "segment not present", segment_not_present)
236#ifdef CONFIG_X86_32
237DO_ERROR(X86_TRAP_SS, SIGBUS, "stack segment", stack_segment) 236DO_ERROR(X86_TRAP_SS, SIGBUS, "stack segment", stack_segment)
238#endif
239DO_ERROR(X86_TRAP_AC, SIGBUS, "alignment check", alignment_check) 237DO_ERROR(X86_TRAP_AC, SIGBUS, "alignment check", alignment_check)
240 238
241#ifdef CONFIG_X86_64 239#ifdef CONFIG_X86_64
242/* Runs on IST stack */ 240/* Runs on IST stack */
243dotraplinkage void do_stack_segment(struct pt_regs *regs, long error_code)
244{
245 enum ctx_state prev_state;
246
247 prev_state = exception_enter();
248 if (notify_die(DIE_TRAP, "stack segment", regs, error_code,
249 X86_TRAP_SS, SIGBUS) != NOTIFY_STOP) {
250 preempt_conditional_sti(regs);
251 do_trap(X86_TRAP_SS, SIGBUS, "stack segment", regs, error_code, NULL);
252 preempt_conditional_cli(regs);
253 }
254 exception_exit(prev_state);
255}
256
257dotraplinkage void do_double_fault(struct pt_regs *regs, long error_code) 241dotraplinkage void do_double_fault(struct pt_regs *regs, long error_code)
258{ 242{
259 static const char str[] = "double fault"; 243 static const char str[] = "double fault";
260 struct task_struct *tsk = current; 244 struct task_struct *tsk = current;
261 245
246#ifdef CONFIG_X86_ESPFIX64
247 extern unsigned char native_irq_return_iret[];
248
249 /*
250 * If IRET takes a non-IST fault on the espfix64 stack, then we
251 * end up promoting it to a doublefault. In that case, modify
252 * the stack to make it look like we just entered the #GP
253 * handler from user space, similar to bad_iret.
254 */
255 if (((long)regs->sp >> PGDIR_SHIFT) == ESPFIX_PGD_ENTRY &&
256 regs->cs == __KERNEL_CS &&
257 regs->ip == (unsigned long)native_irq_return_iret)
258 {
259 struct pt_regs *normal_regs = task_pt_regs(current);
260
261 /* Fake a #GP(0) from userspace. */
262 memmove(&normal_regs->ip, (void *)regs->sp, 5*8);
263 normal_regs->orig_ax = 0; /* Missing (lost) #GP error code */
264 regs->ip = (unsigned long)general_protection;
265 regs->sp = (unsigned long)&normal_regs->orig_ax;
266 return;
267 }
268#endif
269
262 exception_enter(); 270 exception_enter();
263 /* Return not checked because double check cannot be ignored */ 271 /* Return not checked because double check cannot be ignored */
264 notify_die(DIE_TRAP, str, regs, error_code, X86_TRAP_DF, SIGSEGV); 272 notify_die(DIE_TRAP, str, regs, error_code, X86_TRAP_DF, SIGSEGV);
@@ -399,6 +407,35 @@ asmlinkage __visible struct pt_regs *sync_regs(struct pt_regs *eregs)
399 return regs; 407 return regs;
400} 408}
401NOKPROBE_SYMBOL(sync_regs); 409NOKPROBE_SYMBOL(sync_regs);
410
411struct bad_iret_stack {
412 void *error_entry_ret;
413 struct pt_regs regs;
414};
415
416asmlinkage __visible
417struct bad_iret_stack *fixup_bad_iret(struct bad_iret_stack *s)
418{
419 /*
420 * This is called from entry_64.S early in handling a fault
421 * caused by a bad iret to user mode. To handle the fault
422 * correctly, we want move our stack frame to task_pt_regs
423 * and we want to pretend that the exception came from the
424 * iret target.
425 */
426 struct bad_iret_stack *new_stack =
427 container_of(task_pt_regs(current),
428 struct bad_iret_stack, regs);
429
430 /* Copy the IRET target to the new stack. */
431 memmove(&new_stack->regs.ip, (void *)s->regs.sp, 5*8);
432
433 /* Copy the remainder of the stack from the current stack. */
434 memmove(new_stack, s, offsetof(struct bad_iret_stack, regs.ip));
435
436 BUG_ON(!user_mode_vm(&new_stack->regs));
437 return new_stack;
438}
402#endif 439#endif
403 440
404/* 441/*
@@ -778,7 +815,7 @@ void __init trap_init(void)
778 set_intr_gate(X86_TRAP_OLD_MF, coprocessor_segment_overrun); 815 set_intr_gate(X86_TRAP_OLD_MF, coprocessor_segment_overrun);
779 set_intr_gate(X86_TRAP_TS, invalid_TSS); 816 set_intr_gate(X86_TRAP_TS, invalid_TSS);
780 set_intr_gate(X86_TRAP_NP, segment_not_present); 817 set_intr_gate(X86_TRAP_NP, segment_not_present);
781 set_intr_gate_ist(X86_TRAP_SS, &stack_segment, STACKFAULT_STACK); 818 set_intr_gate(X86_TRAP_SS, stack_segment);
782 set_intr_gate(X86_TRAP_GP, general_protection); 819 set_intr_gate(X86_TRAP_GP, general_protection);
783 set_intr_gate(X86_TRAP_SPURIOUS, spurious_interrupt_bug); 820 set_intr_gate(X86_TRAP_SPURIOUS, spurious_interrupt_bug);
784 set_intr_gate(X86_TRAP_MF, coprocessor_error); 821 set_intr_gate(X86_TRAP_MF, coprocessor_error);
diff --git a/arch/x86/kernel/tsc.c b/arch/x86/kernel/tsc.c
index b6025f9e36c6..b7e50bba3bbb 100644
--- a/arch/x86/kernel/tsc.c
+++ b/arch/x86/kernel/tsc.c
@@ -1166,14 +1166,17 @@ void __init tsc_init(void)
1166 1166
1167 x86_init.timers.tsc_pre_init(); 1167 x86_init.timers.tsc_pre_init();
1168 1168
1169 if (!cpu_has_tsc) 1169 if (!cpu_has_tsc) {
1170 setup_clear_cpu_cap(X86_FEATURE_TSC_DEADLINE_TIMER);
1170 return; 1171 return;
1172 }
1171 1173
1172 tsc_khz = x86_platform.calibrate_tsc(); 1174 tsc_khz = x86_platform.calibrate_tsc();
1173 cpu_khz = tsc_khz; 1175 cpu_khz = tsc_khz;
1174 1176
1175 if (!tsc_khz) { 1177 if (!tsc_khz) {
1176 mark_tsc_unstable("could not calculate TSC khz"); 1178 mark_tsc_unstable("could not calculate TSC khz");
1179 setup_clear_cpu_cap(X86_FEATURE_TSC_DEADLINE_TIMER);
1177 return; 1180 return;
1178 } 1181 }
1179 1182
generated by cgit v1.2.2 (git 2.25.0) at 2025-06-02 16:50:35 -0400