Skip to contentSkip to conversation

perf: Multiple task contexts

tip:perf/core tip-bot for Peter Zijlstra 1 participant 1 message Sep 9, 2010

Subsystems (2)
lore
Share & tools
lore
#0 lore raw
Commit-ID:  8dc85d547285668e509f86c177bcd4ea055bcaaf
Gitweb:     http://git.kernel.org/tip/8dc85d547285668e509f86c177bcd4ea055bcaaf
Author:     Peter Zijlstra <[email protected]>
AuthorDate: Thu, 2 Sep 2010 16:50:03 +0200
Committer:  Ingo Molnar <[email protected]>
CommitDate: Thu, 9 Sep 2010 20:46:33 +0200

perf: Multiple task contexts

Provide the infrastructure for multiple task contexts.

A more flexible approach would have resulted in more pointer chases in the scheduling hot-paths. This approach has the limitation of a static number of task contexts.

Since I expect most external PMUs to be system wide, or at least node wide (as per the intel uncore unit) they won't actually need a task context.

Cc paulus <[email protected]>
Cc stephane eranian <[email protected]>
Cc Robert Richter <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Lin Ming <[email protected]>
Cc Yanmin <[email protected]>

LKML-Reference: <new-submission>

---
 include/linux/perf_event.h |    1 +
 include/linux/sched.h      |    8 +-
 kernel/perf_event.c        |  336 ++++++++++++++++++++++++++++++--------------
 3 files changed, 239 insertions(+), 106 deletions(-)
include/linux/perf_event.h +1 -0
index 22155ef..9ecfd85 100644
@@ -572,6 +572,7 @@ struct pmu {
 
 	int * __percpu			pmu_disable_count;
 	struct perf_cpu_context * __percpu pmu_cpu_context;
+	int				task_ctx_nr;
 
 	/*
 	 * Fully disable/enable this PMU, can be used to protect from the PMI
include/linux/sched.h +7 -1
index 1e2a6db..89d6023 100644
@@ -1160,6 +1160,12 @@ struct sched_rt_entity {
 
 struct rcu_node;
 
+enum perf_event_task_context {
+	perf_invalid_context = -1,
+	perf_hw_context = 0,
+	perf_nr_task_contexts,
+};
+
 struct task_struct {
 	volatile long state;	/* -1 unrunnable, 0 runnable, >0 stopped */
 	void *stack;
@@ -1431,7 +1437,7 @@ struct task_struct {
 	struct futex_pi_state *pi_state_cache;
 #endif
 #ifdef CONFIG_PERF_EVENTS
-	struct perf_event_context *perf_event_ctxp;
+	struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
 	struct mutex perf_event_mutex;
 	struct list_head perf_event_list;
 #endif
kernel/perf_event.c +231 -105

658 lines not on this page · raw message