aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
-rw-r--r--arch/x86/kernel/cpu/perf_event_intel.c66
-rw-r--r--arch/x86/kernel/cpu/perf_event_intel_uncore_snb.c12
-rw-r--r--include/linux/pci_ids.h4
-rw-r--r--tools/lib/api/Makefile2
-rw-r--r--tools/lib/traceevent/event-parse.c2
-rw-r--r--tools/perf/bench/futex-requeue.c15
-rw-r--r--tools/perf/bench/numa.c12
-rw-r--r--tools/perf/builtin-kmem.c58
-rw-r--r--tools/perf/builtin-report.c2
-rw-r--r--tools/perf/builtin-top.c2
-rw-r--r--tools/perf/builtin-trace.c10
-rw-r--r--tools/perf/util/probe-event.c2
-rw-r--r--tools/perf/util/probe-finder.c4
13 files changed, 114 insertions, 77 deletions
diff --git a/arch/x86/kernel/cpu/perf_event_intel.c b/arch/x86/kernel/cpu/perf_event_intel.c
index 219d3fb423a1..960e85de13fb 100644
--- a/arch/x86/kernel/cpu/perf_event_intel.c
+++ b/arch/x86/kernel/cpu/perf_event_intel.c
@@ -2533,34 +2533,6 @@ ssize_t intel_event_sysfs_show(char *page, u64 config)
2533 return x86_event_sysfs_show(page, config, event); 2533 return x86_event_sysfs_show(page, config, event);
2534} 2534}
2535 2535
2536static __initconst const struct x86_pmu core_pmu = {
2537 .name = "core",
2538 .handle_irq = x86_pmu_handle_irq,
2539 .disable_all = x86_pmu_disable_all,
2540 .enable_all = core_pmu_enable_all,
2541 .enable = core_pmu_enable_event,
2542 .disable = x86_pmu_disable_event,
2543 .hw_config = x86_pmu_hw_config,
2544 .schedule_events = x86_schedule_events,
2545 .eventsel = MSR_ARCH_PERFMON_EVENTSEL0,
2546 .perfctr = MSR_ARCH_PERFMON_PERFCTR0,
2547 .event_map = intel_pmu_event_map,
2548 .max_events = ARRAY_SIZE(intel_perfmon_event_map),
2549 .apic = 1,
2550 /*
2551 * Intel PMCs cannot be accessed sanely above 32 bit width,
2552 * so we install an artificial 1<<31 period regardless of
2553 * the generic event period:
2554 */
2555 .max_period = (1ULL << 31) - 1,
2556 .get_event_constraints = intel_get_event_constraints,
2557 .put_event_constraints = intel_put_event_constraints,
2558 .event_constraints = intel_core_event_constraints,
2559 .guest_get_msrs = core_guest_get_msrs,
2560 .format_attrs = intel_arch_formats_attr,
2561 .events_sysfs_show = intel_event_sysfs_show,
2562};
2563
2564struct intel_shared_regs *allocate_shared_regs(int cpu) 2536struct intel_shared_regs *allocate_shared_regs(int cpu)
2565{ 2537{
2566 struct intel_shared_regs *regs; 2538 struct intel_shared_regs *regs;
@@ -2743,6 +2715,44 @@ static struct attribute *intel_arch3_formats_attr[] = {
2743 NULL, 2715 NULL,
2744}; 2716};
2745 2717
2718static __initconst const struct x86_pmu core_pmu = {
2719 .name = "core",
2720 .handle_irq = x86_pmu_handle_irq,
2721 .disable_all = x86_pmu_disable_all,
2722 .enable_all = core_pmu_enable_all,
2723 .enable = core_pmu_enable_event,
2724 .disable = x86_pmu_disable_event,
2725 .hw_config = x86_pmu_hw_config,
2726 .schedule_events = x86_schedule_events,
2727 .eventsel = MSR_ARCH_PERFMON_EVENTSEL0,
2728 .perfctr = MSR_ARCH_PERFMON_PERFCTR0,
2729 .event_map = intel_pmu_event_map,
2730 .max_events = ARRAY_SIZE(intel_perfmon_event_map),
2731 .apic = 1,
2732 /*
2733 * Intel PMCs cannot be accessed sanely above 32-bit width,
2734 * so we install an artificial 1<<31 period regardless of
2735 * the generic event period:
2736 */
2737 .max_period = (1ULL<<31) - 1,
2738 .get_event_constraints = intel_get_event_constraints,
2739 .put_event_constraints = intel_put_event_constraints,
2740 .event_constraints = intel_core_event_constraints,
2741 .guest_get_msrs = core_guest_get_msrs,
2742 .format_attrs = intel_arch_formats_attr,
2743 .events_sysfs_show = intel_event_sysfs_show,
2744
2745 /*
2746 * Virtual (or funny metal) CPU can define x86_pmu.extra_regs
2747 * together with PMU version 1 and thus be using core_pmu with
2748 * shared_regs. We need following callbacks here to allocate
2749 * it properly.
2750 */
2751 .cpu_prepare = intel_pmu_cpu_prepare,
2752 .cpu_starting = intel_pmu_cpu_starting,
2753 .cpu_dying = intel_pmu_cpu_dying,
2754};
2755
2746static __initconst const struct x86_pmu intel_pmu = { 2756static __initconst const struct x86_pmu intel_pmu = {
2747 .name = "Intel", 2757 .name = "Intel",
2748 .handle_irq = intel_pmu_handle_irq, 2758 .handle_irq = intel_pmu_handle_irq,
diff --git a/arch/x86/kernel/cpu/perf_event_intel_uncore_snb.c b/arch/x86/kernel/cpu/perf_event_intel_uncore_snb.c
index 3001015b755c..4562e9e22c60 100644
--- a/arch/x86/kernel/cpu/perf_event_intel_uncore_snb.c
+++ b/arch/x86/kernel/cpu/perf_event_intel_uncore_snb.c
@@ -1,6 +1,13 @@
1/* Nehalem/SandBridge/Haswell uncore support */ 1/* Nehalem/SandBridge/Haswell uncore support */
2#include "perf_event_intel_uncore.h" 2#include "perf_event_intel_uncore.h"
3 3
4/* Uncore IMC PCI IDs */
5#define PCI_DEVICE_ID_INTEL_SNB_IMC 0x0100
6#define PCI_DEVICE_ID_INTEL_IVB_IMC 0x0154
7#define PCI_DEVICE_ID_INTEL_IVB_E3_IMC 0x0150
8#define PCI_DEVICE_ID_INTEL_HSW_IMC 0x0c00
9#define PCI_DEVICE_ID_INTEL_HSW_U_IMC 0x0a04
10
4/* SNB event control */ 11/* SNB event control */
5#define SNB_UNC_CTL_EV_SEL_MASK 0x000000ff 12#define SNB_UNC_CTL_EV_SEL_MASK 0x000000ff
6#define SNB_UNC_CTL_UMASK_MASK 0x0000ff00 13#define SNB_UNC_CTL_UMASK_MASK 0x0000ff00
@@ -472,6 +479,10 @@ static const struct pci_device_id hsw_uncore_pci_ids[] = {
472 PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_HSW_IMC), 479 PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_HSW_IMC),
473 .driver_data = UNCORE_PCI_DEV_DATA(SNB_PCI_UNCORE_IMC, 0), 480 .driver_data = UNCORE_PCI_DEV_DATA(SNB_PCI_UNCORE_IMC, 0),
474 }, 481 },
482 { /* IMC */
483 PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_HSW_U_IMC),
484 .driver_data = UNCORE_PCI_DEV_DATA(SNB_PCI_UNCORE_IMC, 0),
485 },
475 { /* end: all zeroes */ }, 486 { /* end: all zeroes */ },
476}; 487};
477 488
@@ -502,6 +513,7 @@ static const struct imc_uncore_pci_dev desktop_imc_pci_ids[] = {
502 IMC_DEV(IVB_IMC, &ivb_uncore_pci_driver), /* 3rd Gen Core processor */ 513 IMC_DEV(IVB_IMC, &ivb_uncore_pci_driver), /* 3rd Gen Core processor */
503 IMC_DEV(IVB_E3_IMC, &ivb_uncore_pci_driver), /* Xeon E3-1200 v2/3rd Gen Core processor */ 514 IMC_DEV(IVB_E3_IMC, &ivb_uncore_pci_driver), /* Xeon E3-1200 v2/3rd Gen Core processor */
504 IMC_DEV(HSW_IMC, &hsw_uncore_pci_driver), /* 4th Gen Core Processor */ 515 IMC_DEV(HSW_IMC, &hsw_uncore_pci_driver), /* 4th Gen Core Processor */
516 IMC_DEV(HSW_U_IMC, &hsw_uncore_pci_driver), /* 4th Gen Core ULT Mobile Processor */
505 { /* end marker */ } 517 { /* end marker */ }
506}; 518};
507 519
diff --git a/include/linux/pci_ids.h b/include/linux/pci_ids.h
index 38cff8f6716d..2f7b9a40f627 100644
--- a/include/linux/pci_ids.h
+++ b/include/linux/pci_ids.h
@@ -2541,10 +2541,6 @@
2541 2541
2542#define PCI_VENDOR_ID_INTEL 0x8086 2542#define PCI_VENDOR_ID_INTEL 0x8086
2543#define PCI_DEVICE_ID_INTEL_EESSC 0x0008 2543#define PCI_DEVICE_ID_INTEL_EESSC 0x0008
2544#define PCI_DEVICE_ID_INTEL_SNB_IMC 0x0100
2545#define PCI_DEVICE_ID_INTEL_IVB_IMC 0x0154
2546#define PCI_DEVICE_ID_INTEL_IVB_E3_IMC 0x0150
2547#define PCI_DEVICE_ID_INTEL_HSW_IMC 0x0c00
2548#define PCI_DEVICE_ID_INTEL_PXHD_0 0x0320 2544#define PCI_DEVICE_ID_INTEL_PXHD_0 0x0320
2549#define PCI_DEVICE_ID_INTEL_PXHD_1 0x0321 2545#define PCI_DEVICE_ID_INTEL_PXHD_1 0x0321
2550#define PCI_DEVICE_ID_INTEL_PXH_0 0x0329 2546#define PCI_DEVICE_ID_INTEL_PXH_0 0x0329
diff --git a/tools/lib/api/Makefile b/tools/lib/api/Makefile
index d8fe29fc19a4..8bd960658463 100644
--- a/tools/lib/api/Makefile
+++ b/tools/lib/api/Makefile
@@ -16,7 +16,7 @@ MAKEFLAGS += --no-print-directory
16LIBFILE = $(OUTPUT)libapi.a 16LIBFILE = $(OUTPUT)libapi.a
17 17
18CFLAGS := $(EXTRA_WARNINGS) $(EXTRA_CFLAGS) 18CFLAGS := $(EXTRA_WARNINGS) $(EXTRA_CFLAGS)
19CFLAGS += -ggdb3 -Wall -Wextra -std=gnu99 -Werror -O6 -D_FORTIFY_SOURCE=2 -fPIC 19CFLAGS += -ggdb3 -Wall -Wextra -std=gnu99 -Werror -O6 -U_FORTIFY_SOURCE -D_FORTIFY_SOURCE=2 -fPIC
20CFLAGS += -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64 20CFLAGS += -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64
21 21
22RM = rm -f 22RM = rm -f
diff --git a/tools/lib/traceevent/event-parse.c b/tools/lib/traceevent/event-parse.c
index e0917c0f5d9f..29f94f6f0d9e 100644
--- a/tools/lib/traceevent/event-parse.c
+++ b/tools/lib/traceevent/event-parse.c
@@ -3865,7 +3865,7 @@ static void print_str_arg(struct trace_seq *s, void *data, int size,
3865 } else if (el_size == 4) { 3865 } else if (el_size == 4) {
3866 trace_seq_printf(s, "%u", *(uint32_t *)num); 3866 trace_seq_printf(s, "%u", *(uint32_t *)num);
3867 } else if (el_size == 8) { 3867 } else if (el_size == 8) {
3868 trace_seq_printf(s, "%lu", *(uint64_t *)num); 3868 trace_seq_printf(s, "%"PRIu64, *(uint64_t *)num);
3869 } else { 3869 } else {
3870 trace_seq_printf(s, "BAD SIZE:%d 0x%x", 3870 trace_seq_printf(s, "BAD SIZE:%d 0x%x",
3871 el_size, *(uint8_t *)num); 3871 el_size, *(uint8_t *)num);
diff --git a/tools/perf/bench/futex-requeue.c b/tools/perf/bench/futex-requeue.c
index bedff6b5b3cf..ad0d9b5342fb 100644
--- a/tools/perf/bench/futex-requeue.c
+++ b/tools/perf/bench/futex-requeue.c
@@ -132,6 +132,9 @@ int bench_futex_requeue(int argc, const char **argv,
132 if (!fshared) 132 if (!fshared)
133 futex_flag = FUTEX_PRIVATE_FLAG; 133 futex_flag = FUTEX_PRIVATE_FLAG;
134 134
135 if (nrequeue > nthreads)
136 nrequeue = nthreads;
137
135 printf("Run summary [PID %d]: Requeuing %d threads (from [%s] %p to %p), " 138 printf("Run summary [PID %d]: Requeuing %d threads (from [%s] %p to %p), "
136 "%d at a time.\n\n", getpid(), nthreads, 139 "%d at a time.\n\n", getpid(), nthreads,
137 fshared ? "shared":"private", &futex1, &futex2, nrequeue); 140 fshared ? "shared":"private", &futex1, &futex2, nrequeue);
@@ -161,20 +164,18 @@ int bench_futex_requeue(int argc, const char **argv,
161 164
162 /* Ok, all threads are patiently blocked, start requeueing */ 165 /* Ok, all threads are patiently blocked, start requeueing */
163 gettimeofday(&start, NULL); 166 gettimeofday(&start, NULL);
164 for (nrequeued = 0; nrequeued < nthreads; nrequeued += nrequeue) { 167 while (nrequeued < nthreads) {
165 /* 168 /*
166 * Do not wakeup any tasks blocked on futex1, allowing 169 * Do not wakeup any tasks blocked on futex1, allowing
167 * us to really measure futex_wait functionality. 170 * us to really measure futex_wait functionality.
168 */ 171 */
169 futex_cmp_requeue(&futex1, 0, &futex2, 0, 172 nrequeued += futex_cmp_requeue(&futex1, 0, &futex2, 0,
170 nrequeue, futex_flag); 173 nrequeue, futex_flag);
171 } 174 }
175
172 gettimeofday(&end, NULL); 176 gettimeofday(&end, NULL);
173 timersub(&end, &start, &runtime); 177 timersub(&end, &start, &runtime);
174 178
175 if (nrequeued > nthreads)
176 nrequeued = nthreads;
177
178 update_stats(&requeued_stats, nrequeued); 179 update_stats(&requeued_stats, nrequeued);
179 update_stats(&requeuetime_stats, runtime.tv_usec); 180 update_stats(&requeuetime_stats, runtime.tv_usec);
180 181
@@ -184,7 +185,7 @@ int bench_futex_requeue(int argc, const char **argv,
184 } 185 }
185 186
186 /* everybody should be blocked on futex2, wake'em up */ 187 /* everybody should be blocked on futex2, wake'em up */
187 nrequeued = futex_wake(&futex2, nthreads, futex_flag); 188 nrequeued = futex_wake(&futex2, nrequeued, futex_flag);
188 if (nthreads != nrequeued) 189 if (nthreads != nrequeued)
189 warnx("couldn't wakeup all tasks (%d/%d)", nrequeued, nthreads); 190 warnx("couldn't wakeup all tasks (%d/%d)", nrequeued, nthreads);
190 191
diff --git a/tools/perf/bench/numa.c b/tools/perf/bench/numa.c
index ebfa163b80b5..ba5efa4710b5 100644
--- a/tools/perf/bench/numa.c
+++ b/tools/perf/bench/numa.c
@@ -180,7 +180,7 @@ static const struct option options[] = {
180 OPT_INTEGER('H', "thp" , &p0.thp, "MADV_NOHUGEPAGE < 0 < MADV_HUGEPAGE"), 180 OPT_INTEGER('H', "thp" , &p0.thp, "MADV_NOHUGEPAGE < 0 < MADV_HUGEPAGE"),
181 OPT_BOOLEAN('c', "show_convergence", &p0.show_convergence, "show convergence details"), 181 OPT_BOOLEAN('c', "show_convergence", &p0.show_convergence, "show convergence details"),
182 OPT_BOOLEAN('m', "measure_convergence", &p0.measure_convergence, "measure convergence latency"), 182 OPT_BOOLEAN('m', "measure_convergence", &p0.measure_convergence, "measure convergence latency"),
183 OPT_BOOLEAN('q', "quiet" , &p0.show_quiet, "bzero the initial allocations"), 183 OPT_BOOLEAN('q', "quiet" , &p0.show_quiet, "quiet mode"),
184 OPT_BOOLEAN('S', "serialize-startup", &p0.serialize_startup,"serialize thread startup"), 184 OPT_BOOLEAN('S', "serialize-startup", &p0.serialize_startup,"serialize thread startup"),
185 185
186 /* Special option string parsing callbacks: */ 186 /* Special option string parsing callbacks: */
@@ -828,6 +828,9 @@ static int count_process_nodes(int process_nr)
828 td = g->threads + task_nr; 828 td = g->threads + task_nr;
829 829
830 node = numa_node_of_cpu(td->curr_cpu); 830 node = numa_node_of_cpu(td->curr_cpu);
831 if (node < 0) /* curr_cpu was likely still -1 */
832 return 0;
833
831 node_present[node] = 1; 834 node_present[node] = 1;
832 } 835 }
833 836
@@ -882,6 +885,11 @@ static void calc_convergence_compression(int *strong)
882 for (p = 0; p < g->p.nr_proc; p++) { 885 for (p = 0; p < g->p.nr_proc; p++) {
883 unsigned int nodes = count_process_nodes(p); 886 unsigned int nodes = count_process_nodes(p);
884 887
888 if (!nodes) {
889 *strong = 0;
890 return;
891 }
892
885 nodes_min = min(nodes, nodes_min); 893 nodes_min = min(nodes, nodes_min);
886 nodes_max = max(nodes, nodes_max); 894 nodes_max = max(nodes, nodes_max);
887 } 895 }
@@ -1395,7 +1403,7 @@ static void print_res(const char *name, double val,
1395 if (!name) 1403 if (!name)
1396 name = "main,"; 1404 name = "main,";
1397 1405
1398 if (g->p.show_quiet) 1406 if (!g->p.show_quiet)
1399 printf(" %-30s %15.3f, %-15s %s\n", name, val, txt_unit, txt_short); 1407 printf(" %-30s %15.3f, %-15s %s\n", name, val, txt_unit, txt_short);
1400 else 1408 else
1401 printf(" %14.3f %s\n", val, txt_long); 1409 printf(" %14.3f %s\n", val, txt_long);
diff --git a/tools/perf/builtin-kmem.c b/tools/perf/builtin-kmem.c
index 63ea01349b6e..1634186d537c 100644
--- a/tools/perf/builtin-kmem.c
+++ b/tools/perf/builtin-kmem.c
@@ -319,7 +319,7 @@ static int page_stat_cmp(struct page_stat *a, struct page_stat *b)
319 return 0; 319 return 0;
320} 320}
321 321
322static struct page_stat *search_page_alloc_stat(struct page_stat *stat, bool create) 322static struct page_stat *search_page_alloc_stat(struct page_stat *pstat, bool create)
323{ 323{
324 struct rb_node **node = &page_alloc_tree.rb_node; 324 struct rb_node **node = &page_alloc_tree.rb_node;
325 struct rb_node *parent = NULL; 325 struct rb_node *parent = NULL;
@@ -331,7 +331,7 @@ static struct page_stat *search_page_alloc_stat(struct page_stat *stat, bool cre
331 parent = *node; 331 parent = *node;
332 data = rb_entry(*node, struct page_stat, node); 332 data = rb_entry(*node, struct page_stat, node);
333 333
334 cmp = page_stat_cmp(data, stat); 334 cmp = page_stat_cmp(data, pstat);
335 if (cmp < 0) 335 if (cmp < 0)
336 node = &parent->rb_left; 336 node = &parent->rb_left;
337 else if (cmp > 0) 337 else if (cmp > 0)
@@ -345,10 +345,10 @@ static struct page_stat *search_page_alloc_stat(struct page_stat *stat, bool cre
345 345
346 data = zalloc(sizeof(*data)); 346 data = zalloc(sizeof(*data));
347 if (data != NULL) { 347 if (data != NULL) {
348 data->page = stat->page; 348 data->page = pstat->page;
349 data->order = stat->order; 349 data->order = pstat->order;
350 data->gfp_flags = stat->gfp_flags; 350 data->gfp_flags = pstat->gfp_flags;
351 data->migrate_type = stat->migrate_type; 351 data->migrate_type = pstat->migrate_type;
352 352
353 rb_link_node(&data->node, parent, node); 353 rb_link_node(&data->node, parent, node);
354 rb_insert_color(&data->node, &page_alloc_tree); 354 rb_insert_color(&data->node, &page_alloc_tree);
@@ -375,7 +375,7 @@ static int perf_evsel__process_page_alloc_event(struct perf_evsel *evsel,
375 unsigned int migrate_type = perf_evsel__intval(evsel, sample, 375 unsigned int migrate_type = perf_evsel__intval(evsel, sample,
376 "migratetype"); 376 "migratetype");
377 u64 bytes = kmem_page_size << order; 377 u64 bytes = kmem_page_size << order;
378 struct page_stat *stat; 378 struct page_stat *pstat;
379 struct page_stat this = { 379 struct page_stat this = {
380 .order = order, 380 .order = order,
381 .gfp_flags = gfp_flags, 381 .gfp_flags = gfp_flags,
@@ -401,21 +401,21 @@ static int perf_evsel__process_page_alloc_event(struct perf_evsel *evsel,
401 * This is to find the current page (with correct gfp flags and 401 * This is to find the current page (with correct gfp flags and
402 * migrate type) at free event. 402 * migrate type) at free event.
403 */ 403 */
404 stat = search_page(page, true); 404 pstat = search_page(page, true);
405 if (stat == NULL) 405 if (pstat == NULL)
406 return -ENOMEM; 406 return -ENOMEM;
407 407
408 stat->order = order; 408 pstat->order = order;
409 stat->gfp_flags = gfp_flags; 409 pstat->gfp_flags = gfp_flags;
410 stat->migrate_type = migrate_type; 410 pstat->migrate_type = migrate_type;
411 411
412 this.page = page; 412 this.page = page;
413 stat = search_page_alloc_stat(&this, true); 413 pstat = search_page_alloc_stat(&this, true);
414 if (stat == NULL) 414 if (pstat == NULL)
415 return -ENOMEM; 415 return -ENOMEM;
416 416
417 stat->nr_alloc++; 417 pstat->nr_alloc++;
418 stat->alloc_bytes += bytes; 418 pstat->alloc_bytes += bytes;
419 419
420 order_stats[order][migrate_type]++; 420 order_stats[order][migrate_type]++;
421 421
@@ -428,7 +428,7 @@ static int perf_evsel__process_page_free_event(struct perf_evsel *evsel,
428 u64 page; 428 u64 page;
429 unsigned int order = perf_evsel__intval(evsel, sample, "order"); 429 unsigned int order = perf_evsel__intval(evsel, sample, "order");
430 u64 bytes = kmem_page_size << order; 430 u64 bytes = kmem_page_size << order;
431 struct page_stat *stat; 431 struct page_stat *pstat;
432 struct page_stat this = { 432 struct page_stat this = {
433 .order = order, 433 .order = order,
434 }; 434 };
@@ -441,8 +441,8 @@ static int perf_evsel__process_page_free_event(struct perf_evsel *evsel,
441 nr_page_frees++; 441 nr_page_frees++;
442 total_page_free_bytes += bytes; 442 total_page_free_bytes += bytes;
443 443
444 stat = search_page(page, false); 444 pstat = search_page(page, false);
445 if (stat == NULL) { 445 if (pstat == NULL) {
446 pr_debug2("missing free at page %"PRIx64" (order: %d)\n", 446 pr_debug2("missing free at page %"PRIx64" (order: %d)\n",
447 page, order); 447 page, order);
448 448
@@ -453,18 +453,18 @@ static int perf_evsel__process_page_free_event(struct perf_evsel *evsel,
453 } 453 }
454 454
455 this.page = page; 455 this.page = page;
456 this.gfp_flags = stat->gfp_flags; 456 this.gfp_flags = pstat->gfp_flags;
457 this.migrate_type = stat->migrate_type; 457 this.migrate_type = pstat->migrate_type;
458 458
459 rb_erase(&stat->node, &page_tree); 459 rb_erase(&pstat->node, &page_tree);
460 free(stat); 460 free(pstat);
461 461
462 stat = search_page_alloc_stat(&this, false); 462 pstat = search_page_alloc_stat(&this, false);
463 if (stat == NULL) 463 if (pstat == NULL)
464 return -ENOENT; 464 return -ENOENT;
465 465
466 stat->nr_free++; 466 pstat->nr_free++;
467 stat->free_bytes += bytes; 467 pstat->free_bytes += bytes;
468 468
469 return 0; 469 return 0;
470} 470}
@@ -640,9 +640,9 @@ static void print_page_summary(void)
640 nr_page_frees, total_page_free_bytes / 1024); 640 nr_page_frees, total_page_free_bytes / 1024);
641 printf("\n"); 641 printf("\n");
642 642
643 printf("%-30s: %'16lu [ %'16"PRIu64" KB ]\n", "Total alloc+freed requests", 643 printf("%-30s: %'16"PRIu64" [ %'16"PRIu64" KB ]\n", "Total alloc+freed requests",
644 nr_alloc_freed, (total_alloc_freed_bytes) / 1024); 644 nr_alloc_freed, (total_alloc_freed_bytes) / 1024);
645 printf("%-30s: %'16lu [ %'16"PRIu64" KB ]\n", "Total alloc-only requests", 645 printf("%-30s: %'16"PRIu64" [ %'16"PRIu64" KB ]\n", "Total alloc-only requests",
646 nr_page_allocs - nr_alloc_freed, 646 nr_page_allocs - nr_alloc_freed,
647 (total_page_alloc_bytes - total_alloc_freed_bytes) / 1024); 647 (total_page_alloc_bytes - total_alloc_freed_bytes) / 1024);
648 printf("%-30s: %'16lu [ %'16"PRIu64" KB ]\n", "Total free-only requests", 648 printf("%-30s: %'16lu [ %'16"PRIu64" KB ]\n", "Total free-only requests",
diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c
index 476cdf7afcca..b63aeda719be 100644
--- a/tools/perf/builtin-report.c
+++ b/tools/perf/builtin-report.c
@@ -329,7 +329,7 @@ static int perf_evlist__tty_browse_hists(struct perf_evlist *evlist,
329 fprintf(stdout, "\n\n"); 329 fprintf(stdout, "\n\n");
330 } 330 }
331 331
332 if (sort_order == default_sort_order && 332 if (sort_order == NULL &&
333 parent_pattern == default_parent_pattern) { 333 parent_pattern == default_parent_pattern) {
334 fprintf(stdout, "#\n# (%s)\n#\n", help); 334 fprintf(stdout, "#\n# (%s)\n#\n", help);
335 335
diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c
index 1cb3436276d1..6a4d5d41c671 100644
--- a/tools/perf/builtin-top.c
+++ b/tools/perf/builtin-top.c
@@ -733,7 +733,7 @@ static void perf_event__process_sample(struct perf_tool *tool,
733"Kernel address maps (/proc/{kallsyms,modules}) are restricted.\n\n" 733"Kernel address maps (/proc/{kallsyms,modules}) are restricted.\n\n"
734"Check /proc/sys/kernel/kptr_restrict.\n\n" 734"Check /proc/sys/kernel/kptr_restrict.\n\n"
735"Kernel%s samples will not be resolved.\n", 735"Kernel%s samples will not be resolved.\n",
736 !RB_EMPTY_ROOT(&al.map->dso->symbols[MAP__FUNCTION]) ? 736 al.map && !RB_EMPTY_ROOT(&al.map->dso->symbols[MAP__FUNCTION]) ?
737 " modules" : ""); 737 " modules" : "");
738 if (use_browser <= 0) 738 if (use_browser <= 0)
739 sleep(5); 739 sleep(5);
diff --git a/tools/perf/builtin-trace.c b/tools/perf/builtin-trace.c
index e124741be187..e122970361f2 100644
--- a/tools/perf/builtin-trace.c
+++ b/tools/perf/builtin-trace.c
@@ -2241,10 +2241,11 @@ static int trace__run(struct trace *trace, int argc, const char **argv)
2241 if (err < 0) 2241 if (err < 0)
2242 goto out_error_mmap; 2242 goto out_error_mmap;
2243 2243
2244 if (!target__none(&trace->opts.target))
2245 perf_evlist__enable(evlist);
2246
2244 if (forks) 2247 if (forks)
2245 perf_evlist__start_workload(evlist); 2248 perf_evlist__start_workload(evlist);
2246 else
2247 perf_evlist__enable(evlist);
2248 2249
2249 trace->multiple_threads = evlist->threads->map[0] == -1 || 2250 trace->multiple_threads = evlist->threads->map[0] == -1 ||
2250 evlist->threads->nr > 1 || 2251 evlist->threads->nr > 1 ||
@@ -2272,6 +2273,11 @@ next_event:
2272 2273
2273 if (interrupted) 2274 if (interrupted)
2274 goto out_disable; 2275 goto out_disable;
2276
2277 if (done && !draining) {
2278 perf_evlist__disable(evlist);
2279 draining = true;
2280 }
2275 } 2281 }
2276 } 2282 }
2277 2283
diff --git a/tools/perf/util/probe-event.c b/tools/perf/util/probe-event.c
index d8bb616ff57c..d05b77cf35f7 100644
--- a/tools/perf/util/probe-event.c
+++ b/tools/perf/util/probe-event.c
@@ -1084,6 +1084,8 @@ static int parse_perf_probe_point(char *arg, struct perf_probe_event *pev)
1084 * 1084 *
1085 * TODO:Group name support 1085 * TODO:Group name support
1086 */ 1086 */
1087 if (!arg)
1088 return -EINVAL;
1087 1089
1088 ptr = strpbrk(arg, ";=@+%"); 1090 ptr = strpbrk(arg, ";=@+%");
1089 if (ptr && *ptr == '=') { /* Event name */ 1091 if (ptr && *ptr == '=') { /* Event name */
diff --git a/tools/perf/util/probe-finder.c b/tools/perf/util/probe-finder.c
index b5bf9d5efeaf..2a76e14db732 100644
--- a/tools/perf/util/probe-finder.c
+++ b/tools/perf/util/probe-finder.c
@@ -578,10 +578,12 @@ static int find_variable(Dwarf_Die *sc_die, struct probe_finder *pf)
578 /* Search child die for local variables and parameters. */ 578 /* Search child die for local variables and parameters. */
579 if (!die_find_variable_at(sc_die, pf->pvar->var, pf->addr, &vr_die)) { 579 if (!die_find_variable_at(sc_die, pf->pvar->var, pf->addr, &vr_die)) {
580 /* Search again in global variables */ 580 /* Search again in global variables */
581 if (!die_find_variable_at(&pf->cu_die, pf->pvar->var, 0, &vr_die)) 581 if (!die_find_variable_at(&pf->cu_die, pf->pvar->var,
582 0, &vr_die)) {
582 pr_warning("Failed to find '%s' in this function.\n", 583 pr_warning("Failed to find '%s' in this function.\n",
583 pf->pvar->var); 584 pf->pvar->var);
584 ret = -ENOENT; 585 ret = -ENOENT;
586 }
585 } 587 }
586 if (ret >= 0) 588 if (ret >= 0)
587 ret = convert_variable(&vr_die, pf); 589 ret = convert_variable(&vr_die, pf);