[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Message-Id: <20161025235200.28781-1-alexis.berlemont@gmail.com>
Date: Wed, 26 Oct 2016 01:51:58 +0200
From: Alexis Berlemont <alexis.berlemont@...il.com>
To: linux-kernel@...r.kernel.org
Cc: Alexis Berlemont <alexis.berlemont@...il.com>,
peterz@...radead.org, mingo@...hat.com, acme@...nel.org,
alexander.shishkin@...ux.intel.com
Subject: [PATCH 0/2] perf: measure page fault duration in perf trace
Hi,
Here are 2 small patches which try to fulfill a point in the perf todo
list:
* Forward port the page fault tracepoints and use it in 'trace'.
http://git.kernel.org/?p=linux/kernel/git/acme/linux.git;a=commitdiff;h=d53b11976093b6d8afeb8181db53aaffc754920d;hp=32ba4abf60ae1b710d22a75725491815de649bc5
There are some questionable points:
* With luck I think I found the patch related with the todo item (the
link in the todo wiki page is broken); I hope I am not wrong...
* In the patch mentioned above, I found only changes related with
tracepoints creations and calls; the tracepoints were declared
generic (in include/trace/events/kmem.h) but were only called in x86
(arch/x86/mm/fault.c); as in x86, the tracepoint
"mm_pagefault_start" looks fairly like "page_fault_user" and
"page_fault_kernel", I decided to only create one x86-specific
tracepoint: "page_fault_exit"; maybe, you would prefer declaring
generic tracepoints;
* No option has been added for activating page-fault duration
calculation: if the needed tracepoints are available, durations will
be printed; maybe, that was not what you were looking for.
The patches were generated against tip/perf/core.
Alexis.
Alexis Berlemont (2):
perf, x86-mm: Add exit-fault tracing
perf: add page fault duration measures in perf trace
arch/x86/include/asm/trace/exceptions.h | 21 +++
arch/x86/mm/fault.c | 1 +
tools/perf/builtin-trace.c | 225 ++++++++++++++++++++++++++++----
3 files changed, 221 insertions(+), 26 deletions(-)
--
2.10.1
Powered by blists - more mailing lists