lkml.org 
[lkml]   [2017]   [Jan]   [10]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
From
Subject[RFC 4/6] perf/core: avoid rb-tree traversal when no inactive events
Date
Avoid unnecessary rb-tree queries.

Signed-off-by: David Carrillo-Cisneros <davidcc@google.com>
---
include/linux/perf_event.h | 2 ++
kernel/events/core.c | 17 +++++++++++++++--
2 files changed, 17 insertions(+), 2 deletions(-)

diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h
index fd32ecc37d33..2e0c94a770e2 100644
--- a/include/linux/perf_event.h
+++ b/include/linux/perf_event.h
@@ -748,6 +748,8 @@ struct perf_event_context {
struct list_head event_list;
int nr_events;
int nr_active;
+ int nr_inactive_pinned;
+ int nr_inactive_flexible;
int is_active;
int nr_stat;
int nr_freq;
diff --git a/kernel/events/core.c b/kernel/events/core.c
index 302f13ca75dc..c7715b2627a9 100644
--- a/kernel/events/core.c
+++ b/kernel/events/core.c
@@ -1531,6 +1531,10 @@ rbtree_add_inactive(struct perf_event *event,
/* Add new node and rebalance tree. */
rb_link_node(&event->rbtree_node, parent, pos);
rb_insert_color(&event->rbtree_node, &ctx->rbtree_root);
+ if (event->attr.pinned)
+ ctx->nr_inactive_pinned++;
+ else
+ ctx->nr_inactive_flexible++;
}

static void
@@ -1754,6 +1758,11 @@ static void ctx_sched_groups_del(struct perf_event *group,
WARN_ON(group->state != PERF_EVENT_STATE_INACTIVE);
rb_erase(&group->rbtree_node, &ctx->rbtree_root);
list_del_init(&group->ctx_active_entry);
+ if (group->attr.pinned)
+ ctx->nr_inactive_pinned--;
+ else
+ ctx->nr_inactive_flexible--;
+
}

/*
@@ -2142,6 +2151,10 @@ ctx_sched_groups_to_active(struct perf_event *event, struct perf_event_context *
WARN_ON(event->state != PERF_EVENT_STATE_ACTIVE);
rb_erase(&event->rbtree_node, &ctx->rbtree_root);
list_move_tail(&event->ctx_active_entry, h);
+ if (event->attr.pinned)
+ ctx->nr_inactive_pinned--;
+ else
+ ctx->nr_inactive_flexible--;
}

static int
@@ -3371,11 +3384,11 @@ ctx_sched_in(struct perf_event_context *ctx,
* First go through the list and put on any pinned groups
* in order to give them the best chance of going on.
*/
- if (is_active & EVENT_PINNED)
+ if (is_active & EVENT_PINNED && ctx->nr_inactive_pinned)
ctx_pinned_sched_in(ctx, cpuctx);

/* Then walk through the lower prio flexible groups */
- if (is_active & EVENT_FLEXIBLE)
+ if (is_active & EVENT_FLEXIBLE && ctx->nr_inactive_flexible)
ctx_flexible_sched_in(ctx, cpuctx);
}

--
2.11.0.390.gc69c2f50cf-goog
\
 
 \ /
  Last update: 2017-01-10 11:27    [W:0.194 / U:0.304 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site