From 0db61388b389f43c1ba2f1cee3613feb4fd12150 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Tue, 22 Apr 2025 15:33:18 -0700 Subject: perf/core: Change to POLLERR for pinned events with error Commit: f4b07fd62d4d11d5 ("perf/core: Use POLLHUP for pinned events in error") started to emit POLLHUP for pinned events in an error state. But the POLLHUP is also used to signal events that the attached task is terminated. To distinguish pinned per-task events in the error state it would need to check if the task is live. Change it to POLLERR to make it clear. Suggested-by: Gabriel Marin Signed-off-by: Namhyung Kim Signed-off-by: Ingo Molnar Cc: Alexander Shishkin Cc: Arnaldo Carvalho de Melo Cc: Kan Liang Cc: Mark Rutland Cc: Peter Zijlstra Link: https://lore.kernel.org/r/20250422223318.180343-1-namhyung@kernel.org --- kernel/events/core.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/kernel/events/core.c b/kernel/events/core.c index e93c19565914..95e703891b24 100644 --- a/kernel/events/core.c +++ b/kernel/events/core.c @@ -3943,7 +3943,7 @@ static int merge_sched_in(struct perf_event *event, void *data) perf_event_set_state(event, PERF_EVENT_STATE_ERROR); if (*perf_event_fasync(event)) - event->pending_kill = POLL_HUP; + event->pending_kill = POLL_ERR; perf_event_wakeup(event); } else { @@ -6075,7 +6075,7 @@ static __poll_t perf_poll(struct file *file, poll_table *wait) if (unlikely(READ_ONCE(event->state) == PERF_EVENT_STATE_ERROR && event->attr.pinned)) - return events; + return EPOLLERR; /* * Pin the event->rb by taking event->mmap_mutex; otherwise -- cgit v1.2.3 From 1a97fea9db9e9b9c4839d4232dde9f505ff5b4cc Mon Sep 17 00:00:00 2001 From: Luo Gengkun Date: Wed, 23 Apr 2025 06:47:24 +0000 Subject: perf/x86: Fix non-sampling (counting) events on certain x86 platforms Perf doesn't work at perf stat for hardware events on certain x86 platforms: $perf stat -- sleep 1 Performance counter stats for 'sleep 1': 16.44 msec task-clock # 0.016 CPUs utilized 2 context-switches # 121.691 /sec 0 cpu-migrations # 0.000 /sec 54 page-faults # 3.286 K/sec cycles instructions branches branch-misses The reason is that the check in x86_pmu_hw_config() for sampling events is unexpectedly applied to counting events as well. It should only impact x86 platforms with limit_period used for non-PEBS events. For Intel platforms, it should only impact some older platforms, e.g., HSW, BDW and NHM. Fixes: 88ec7eedbbd2 ("perf/x86: Fix low freqency setting issue") Signed-off-by: Luo Gengkun Signed-off-by: Ingo Molnar Reviewed-by: Kan Liang Cc: Alexander Shishkin Cc: Arnaldo Carvalho de Melo Cc: Jiri Olsa Cc: Mark Rutland Cc: Namhyung Kim Cc: Peter Zijlstra Cc: Ravi Bangoria Link: https://lore.kernel.org/r/20250423064724.3716211-1-luogengkun@huaweicloud.com --- arch/x86/events/core.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/arch/x86/events/core.c b/arch/x86/events/core.c index 6866cc5acb0b..3a4f031d2f44 100644 --- a/arch/x86/events/core.c +++ b/arch/x86/events/core.c @@ -629,7 +629,7 @@ int x86_pmu_hw_config(struct perf_event *event) if (event->attr.type == event->pmu->type) event->hw.config |= x86_pmu_get_event_config(event); - if (!event->attr.freq && x86_pmu.limit_period) { + if (is_sampling_event(event) && !event->attr.freq && x86_pmu.limit_period) { s64 left = event->attr.sample_period; x86_pmu.limit_period(event, &left); if (left > event->attr.sample_period) -- cgit v1.2.3