diff options
| author | Sebastian Andrzej Siewior <bigeasy@linutronix.de> | 2024-07-04 19:03:39 +0200 |
|---|---|---|
| committer | Peter Zijlstra <peterz@infradead.org> | 2024-07-09 13:26:36 +0200 |
| commit | 0d40a6d83e3e6751f1107ba33587262d937c969f (patch) | |
| tree | 691c1d8f6993be280af09b4da362eb714ce34e6e /include/linux | |
| parent | 5af42f928f3ac555c228740fb4a92d05b19fdd49 (diff) | |
| download | linux-next-0d40a6d83e3e6751f1107ba33587262d937c969f.tar.gz linux-next-0d40a6d83e3e6751f1107ba33587262d937c969f.zip | |
perf: Move swevent_htable::recursion into task_struct.
The swevent_htable::recursion counter is used to avoid creating an
swevent while an event is processed to avoid recursion. The counter is
per-CPU and preemption must be disabled to have a stable counter.
perf_pending_task() disables preemption to access the counter and then
signal. This is problematic on PREEMPT_RT because sending a signal uses
a spinlock_t which must not be acquired in atomic on PREEMPT_RT because
it becomes a sleeping lock.
The atomic context can be avoided by moving the counter into the
task_struct. There is a 4 byte hole between futex_state (usually always
on) and the following perf pointer (perf_event_ctxp). After the
recursion lost some weight it fits perfectly.
Move swevent_htable::recursion into task_struct.
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Marco Elver <elver@google.com>
Link: https://lore.kernel.org/r/20240704170424.1466941-6-bigeasy@linutronix.de
Diffstat (limited to 'include/linux')
| -rw-r--r-- | include/linux/perf_event.h | 6 | ||||
| -rw-r--r-- | include/linux/sched.h | 7 |
2 files changed, 7 insertions, 6 deletions
diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h index ea0d82418d85..99a7ea1d29ed 100644 --- a/include/linux/perf_event.h +++ b/include/linux/perf_event.h @@ -970,12 +970,6 @@ struct perf_event_context { local_t nr_pending; }; -/* - * Number of contexts where an event can trigger: - * task, softirq, hardirq, nmi. - */ -#define PERF_NR_CONTEXTS 4 - struct perf_cpu_pmu_context { struct perf_event_pmu_context epc; struct perf_event_pmu_context *task_epc; diff --git a/include/linux/sched.h b/include/linux/sched.h index 61591ac6eab6..afb1087f5831 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -734,6 +734,12 @@ enum perf_event_task_context { perf_nr_task_contexts, }; +/* + * Number of contexts where an event can trigger: + * task, softirq, hardirq, nmi. + */ +#define PERF_NR_CONTEXTS 4 + struct wake_q_node { struct wake_q_node *next; }; @@ -1256,6 +1262,7 @@ struct task_struct { unsigned int futex_state; #endif #ifdef CONFIG_PERF_EVENTS + u8 perf_recursion[PERF_NR_CONTEXTS]; struct perf_event_context *perf_event_ctxp; struct mutex perf_event_mutex; struct list_head perf_event_list; |