linux/tools/perf/util/bpf_lock_contention.c

196 lines
4.2 KiB
C
Raw Permalink Normal View History

// SPDX-License-Identifier: GPL-2.0
#include "util/debug.h"
#include "util/evlist.h"
#include "util/machine.h"
#include "util/map.h"
#include "util/symbol.h"
#include "util/target.h"
#include "util/thread_map.h"
#include "util/lock-contention.h"
#include <linux/zalloc.h>
perf lock contention: Show full callstack with -v option Currently it shows a caller function for each entry, but users need to see the full call stacks sometimes. Use -v/--verbose option to do that. # perf lock con -a -b -v sleep 3 Looking at the vmlinux_path (8 entries long) symsrc__init: cannot get elf header. Using /proc/kcore for kernel data Using /proc/kallsyms for symbols contended total wait max wait avg wait type caller 1 10.74 us 10.74 us 10.74 us spinlock __bpf_trace_contention_begin+0xb 0xffffffffc03b5c47 bpf_prog_bf07ae9e2cbd02c5_contention_begin+0x117 0xffffffffc03b5c47 bpf_prog_bf07ae9e2cbd02c5_contention_begin+0x117 0xffffffffbb8b8e75 bpf_trace_run2+0x35 0xffffffffbb7eab9b __bpf_trace_contention_begin+0xb 0xffffffffbb7ebe75 queued_spin_lock_slowpath+0x1f5 0xffffffffbc1c26ff _raw_spin_lock+0x1f 0xffffffffbb841015 tick_do_update_jiffies64+0x25 0xffffffffbb8409ee tick_irq_enter+0x9e 1 7.70 us 7.70 us 7.70 us spinlock __bpf_trace_contention_begin+0xb 0xffffffffc03b5c47 bpf_prog_bf07ae9e2cbd02c5_contention_begin+0x117 0xffffffffc03b5c47 bpf_prog_bf07ae9e2cbd02c5_contention_begin+0x117 0xffffffffbb8b8e75 bpf_trace_run2+0x35 0xffffffffbb7eab9b __bpf_trace_contention_begin+0xb 0xffffffffbb7ebe75 queued_spin_lock_slowpath+0x1f5 0xffffffffbc1c26ff _raw_spin_lock+0x1f 0xffffffffbb7bc27e raw_spin_rq_lock_nested+0xe 0xffffffffbb7cef9c load_balance+0x66c Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Song Liu <songliubraving@fb.com> Cc: bpf@vger.kernel.org Link: https://lore.kernel.org/r/20220912055314.744552-3-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2022-09-12 05:53:12 +00:00
#include <linux/string.h>
#include <bpf/bpf.h>
#include "bpf_skel/lock_contention.skel.h"
static struct lock_contention_bpf *skel;
struct lock_contention_data {
u64 total_time;
u64 min_time;
u64 max_time;
u32 count;
u32 flags;
};
int lock_contention_prepare(struct lock_contention *con)
{
int i, fd;
int ncpus = 1, ntasks = 1;
struct evlist *evlist = con->evlist;
struct target *target = con->target;
skel = lock_contention_bpf__open();
if (!skel) {
pr_err("Failed to open lock-contention BPF skeleton\n");
return -1;
}
perf lock contention: Allow to change stack depth and skip It needs stack traces to find callers of locks. To minimize the performance overhead it only collects up to 8 entries for each stack trace. And it skips first 3 entries as they came from BPF, tracepoint and lock functions which are not interested for most users. But it turned out that those numbers are different in some configuration. Using fixed number can result in non meaningful caller names. Let's make them adjustable with --stack-depth and --skip-stack options. On my setup, the default output is like below: # /perf lock con -ab -F contended,wait_total sleep 3 contended total wait type caller 28 4.55 ms rwlock:W __bpf_trace_contention_begin+0xb 33 1.67 ms rwlock:W __bpf_trace_contention_begin+0xb 12 580.28 us spinlock __bpf_trace_contention_begin+0xb 60 240.54 us rwsem:R __bpf_trace_contention_begin+0xb 27 64.45 us spinlock __bpf_trace_contention_begin+0xb If I change the stack skip to 5, the result will be like: # perf lock con -ab -F contended,wait_total --stack-skip 5 sleep 3 contended total wait type caller 32 715.45 us spinlock folio_lruvec_lock_irqsave+0x61 26 550.22 us spinlock folio_lruvec_lock_irqsave+0x61 15 486.93 us rwsem:R mmap_read_lock+0x13 12 139.66 us rwsem:W vm_mmap_pgoff+0x93 1 7.04 us spinlock tick_do_update_jiffies64+0x25 Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Song Liu <songliubraving@fb.com> Cc: bpf@vger.kernel.org Link: https://lore.kernel.org/r/20220912055314.744552-4-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2022-09-12 05:53:13 +00:00
bpf_map__set_value_size(skel->maps.stacks, con->max_stack * sizeof(u64));
bpf_map__set_max_entries(skel->maps.stacks, con->map_nr_entries);
bpf_map__set_max_entries(skel->maps.lock_stat, con->map_nr_entries);
if (target__has_cpu(target))
ncpus = perf_cpu_map__nr(evlist->core.user_requested_cpus);
if (target__has_task(target))
ntasks = perf_thread_map__nr(evlist->core.threads);
bpf_map__set_max_entries(skel->maps.cpu_filter, ncpus);
bpf_map__set_max_entries(skel->maps.task_filter, ntasks);
if (lock_contention_bpf__load(skel) < 0) {
pr_err("Failed to load lock-contention BPF skeleton\n");
return -1;
}
if (target__has_cpu(target)) {
u32 cpu;
u8 val = 1;
skel->bss->has_cpu = 1;
fd = bpf_map__fd(skel->maps.cpu_filter);
for (i = 0; i < ncpus; i++) {
cpu = perf_cpu_map__cpu(evlist->core.user_requested_cpus, i).cpu;
bpf_map_update_elem(fd, &cpu, &val, BPF_ANY);
}
}
if (target__has_task(target)) {
u32 pid;
u8 val = 1;
skel->bss->has_task = 1;
fd = bpf_map__fd(skel->maps.task_filter);
for (i = 0; i < ntasks; i++) {
pid = perf_thread_map__pid(evlist->core.threads, i);
bpf_map_update_elem(fd, &pid, &val, BPF_ANY);
}
}
if (target__none(target) && evlist->workload.pid > 0) {
u32 pid = evlist->workload.pid;
u8 val = 1;
skel->bss->has_task = 1;
fd = bpf_map__fd(skel->maps.task_filter);
bpf_map_update_elem(fd, &pid, &val, BPF_ANY);
}
skel->bss->stack_skip = con->stack_skip;
lock_contention_bpf__attach(skel);
return 0;
}
int lock_contention_start(void)
{
skel->bss->enabled = 1;
return 0;
}
int lock_contention_stop(void)
{
skel->bss->enabled = 0;
return 0;
}
int lock_contention_read(struct lock_contention *con)
{
int fd, stack;
s32 prev_key, key;
struct lock_contention_data data;
struct lock_stat *st;
struct machine *machine = con->machine;
perf lock contention: Allow to change stack depth and skip It needs stack traces to find callers of locks. To minimize the performance overhead it only collects up to 8 entries for each stack trace. And it skips first 3 entries as they came from BPF, tracepoint and lock functions which are not interested for most users. But it turned out that those numbers are different in some configuration. Using fixed number can result in non meaningful caller names. Let's make them adjustable with --stack-depth and --skip-stack options. On my setup, the default output is like below: # /perf lock con -ab -F contended,wait_total sleep 3 contended total wait type caller 28 4.55 ms rwlock:W __bpf_trace_contention_begin+0xb 33 1.67 ms rwlock:W __bpf_trace_contention_begin+0xb 12 580.28 us spinlock __bpf_trace_contention_begin+0xb 60 240.54 us rwsem:R __bpf_trace_contention_begin+0xb 27 64.45 us spinlock __bpf_trace_contention_begin+0xb If I change the stack skip to 5, the result will be like: # perf lock con -ab -F contended,wait_total --stack-skip 5 sleep 3 contended total wait type caller 32 715.45 us spinlock folio_lruvec_lock_irqsave+0x61 26 550.22 us spinlock folio_lruvec_lock_irqsave+0x61 15 486.93 us rwsem:R mmap_read_lock+0x13 12 139.66 us rwsem:W vm_mmap_pgoff+0x93 1 7.04 us spinlock tick_do_update_jiffies64+0x25 Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Song Liu <songliubraving@fb.com> Cc: bpf@vger.kernel.org Link: https://lore.kernel.org/r/20220912055314.744552-4-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2022-09-12 05:53:13 +00:00
u64 stack_trace[con->max_stack];
fd = bpf_map__fd(skel->maps.lock_stat);
stack = bpf_map__fd(skel->maps.stacks);
con->lost = skel->bss->lost;
prev_key = 0;
while (!bpf_map_get_next_key(fd, &prev_key, &key)) {
struct map *kmap;
struct symbol *sym;
int idx = 0;
bpf_map_lookup_elem(fd, &key, &data);
st = zalloc(sizeof(*st));
if (st == NULL)
return -1;
st->nr_contended = data.count;
st->wait_time_total = data.total_time;
st->wait_time_max = data.max_time;
st->wait_time_min = data.min_time;
if (data.count)
st->avg_wait_time = data.total_time / data.count;
st->flags = data.flags;
bpf_map_lookup_elem(stack, &key, stack_trace);
/* skip lock internal functions */
while (is_lock_function(machine, stack_trace[idx]) &&
perf lock contention: Allow to change stack depth and skip It needs stack traces to find callers of locks. To minimize the performance overhead it only collects up to 8 entries for each stack trace. And it skips first 3 entries as they came from BPF, tracepoint and lock functions which are not interested for most users. But it turned out that those numbers are different in some configuration. Using fixed number can result in non meaningful caller names. Let's make them adjustable with --stack-depth and --skip-stack options. On my setup, the default output is like below: # /perf lock con -ab -F contended,wait_total sleep 3 contended total wait type caller 28 4.55 ms rwlock:W __bpf_trace_contention_begin+0xb 33 1.67 ms rwlock:W __bpf_trace_contention_begin+0xb 12 580.28 us spinlock __bpf_trace_contention_begin+0xb 60 240.54 us rwsem:R __bpf_trace_contention_begin+0xb 27 64.45 us spinlock __bpf_trace_contention_begin+0xb If I change the stack skip to 5, the result will be like: # perf lock con -ab -F contended,wait_total --stack-skip 5 sleep 3 contended total wait type caller 32 715.45 us spinlock folio_lruvec_lock_irqsave+0x61 26 550.22 us spinlock folio_lruvec_lock_irqsave+0x61 15 486.93 us rwsem:R mmap_read_lock+0x13 12 139.66 us rwsem:W vm_mmap_pgoff+0x93 1 7.04 us spinlock tick_do_update_jiffies64+0x25 Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Song Liu <songliubraving@fb.com> Cc: bpf@vger.kernel.org Link: https://lore.kernel.org/r/20220912055314.744552-4-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2022-09-12 05:53:13 +00:00
idx < con->max_stack - 1)
idx++;
st->addr = stack_trace[idx];
sym = machine__find_kernel_symbol(machine, st->addr, &kmap);
if (sym) {
unsigned long offset;
int ret = 0;
offset = kmap->map_ip(kmap, st->addr) - sym->start;
if (offset)
ret = asprintf(&st->name, "%s+%#lx", sym->name, offset);
else
st->name = strdup(sym->name);
if (ret < 0 || st->name == NULL)
return -1;
} else if (asprintf(&st->name, "%#lx", (unsigned long)st->addr) < 0) {
free(st);
return -1;
}
perf lock contention: Show full callstack with -v option Currently it shows a caller function for each entry, but users need to see the full call stacks sometimes. Use -v/--verbose option to do that. # perf lock con -a -b -v sleep 3 Looking at the vmlinux_path (8 entries long) symsrc__init: cannot get elf header. Using /proc/kcore for kernel data Using /proc/kallsyms for symbols contended total wait max wait avg wait type caller 1 10.74 us 10.74 us 10.74 us spinlock __bpf_trace_contention_begin+0xb 0xffffffffc03b5c47 bpf_prog_bf07ae9e2cbd02c5_contention_begin+0x117 0xffffffffc03b5c47 bpf_prog_bf07ae9e2cbd02c5_contention_begin+0x117 0xffffffffbb8b8e75 bpf_trace_run2+0x35 0xffffffffbb7eab9b __bpf_trace_contention_begin+0xb 0xffffffffbb7ebe75 queued_spin_lock_slowpath+0x1f5 0xffffffffbc1c26ff _raw_spin_lock+0x1f 0xffffffffbb841015 tick_do_update_jiffies64+0x25 0xffffffffbb8409ee tick_irq_enter+0x9e 1 7.70 us 7.70 us 7.70 us spinlock __bpf_trace_contention_begin+0xb 0xffffffffc03b5c47 bpf_prog_bf07ae9e2cbd02c5_contention_begin+0x117 0xffffffffc03b5c47 bpf_prog_bf07ae9e2cbd02c5_contention_begin+0x117 0xffffffffbb8b8e75 bpf_trace_run2+0x35 0xffffffffbb7eab9b __bpf_trace_contention_begin+0xb 0xffffffffbb7ebe75 queued_spin_lock_slowpath+0x1f5 0xffffffffbc1c26ff _raw_spin_lock+0x1f 0xffffffffbb7bc27e raw_spin_rq_lock_nested+0xe 0xffffffffbb7cef9c load_balance+0x66c Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Song Liu <songliubraving@fb.com> Cc: bpf@vger.kernel.org Link: https://lore.kernel.org/r/20220912055314.744552-3-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2022-09-12 05:53:12 +00:00
if (verbose) {
st->callstack = memdup(stack_trace, sizeof(stack_trace));
if (st->callstack == NULL) {
free(st);
return -1;
}
}
hlist_add_head(&st->hash_entry, con->result);
prev_key = key;
}
return 0;
}
int lock_contention_finish(void)
{
if (skel) {
skel->bss->enabled = 0;
lock_contention_bpf__destroy(skel);
}
return 0;
}