[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20210723125634.213430293@goodmis.org>
Date: Fri, 23 Jul 2021 08:54:59 -0400
From: Steven Rostedt <rostedt@...dmis.org>
To: linux-kernel@...r.kernel.org
Cc: Ingo Molnar <mingo@...nel.org>,
Andrew Morton <akpm@...ux-foundation.org>,
Nicolas Saenz Julienne <nsaenzju@...hat.com>
Subject: [for-linus][PATCH 5/7] ftrace: Avoid synchronize_rcu_tasks_rude() call when not necessary
From: Nicolas Saenz Julienne <nsaenzju@...hat.com>
synchronize_rcu_tasks_rude() triggers IPIs and forces rescheduling on
all CPUs. It is a costly operation and, when targeting nohz_full CPUs,
very disrupting (hence the name). So avoid calling it when 'old_hash'
doesn't need to be freed.
Link: https://lkml.kernel.org/r/20210721114726.1545103-1-nsaenzju@redhat.com
Signed-off-by: Nicolas Saenz Julienne <nsaenzju@...hat.com>
Signed-off-by: Steven Rostedt (VMware) <rostedt@...dmis.org>
---
kernel/trace/ftrace.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c
index e6fb3e6e1ffc..4fbcf560dd03 100644
--- a/kernel/trace/ftrace.c
+++ b/kernel/trace/ftrace.c
@@ -5985,7 +5985,8 @@ ftrace_graph_release(struct inode *inode, struct file *file)
* infrastructure to do the synchronization, thus we must do it
* ourselves.
*/
- synchronize_rcu_tasks_rude();
+ if (old_hash != EMPTY_HASH)
+ synchronize_rcu_tasks_rude();
free_ftrace_hash(old_hash);
}
--
2.30.2
Powered by blists - more mailing lists