On Wed, 2016-07-27 at 20:44 +0200, Michal Hocko wrote:

On Wed 27-07-16 14:16:22, Rik van Riel wrote:

quoted

On Wed, 2016-07-27 at 18:33 +0200, Michal Hocko wrote:

quoted

On Wed 27-07-16 10:47:59, Janani Ravichandran wrote:

quoted

Add tracepoints to the slowpath code to gather some
information.
The tracepoints can also be used to find out how much time was
spent in
the slowpath.

I do not think this is a right thing to measure.
__alloc_pages_slowpath
is more a code organization thing. The fast path might perform an
expensive operations like zone reclaim (if node_reclaim_mode > 0)
so
these trace point would miss it.

It doesn't look like it does. The fast path either
returns an allocated page to the caller, or calls
into the slow path.

I must be missing something here but what prevents
__alloc_pages_nodemask->get_page_from_freelist from doing
zone_reclaim?

You are right!

Guess the script may need to collect all the
tracing output from __alloc_pages_nodemask on
up, and then filter the output so only the
interesting (read: long duration) traces get
dumped out to a file.

-- 
All rights reversed