linux-kernel - [PATCH 0/3] perf-stat: refactor print/formatting into print-ops

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Thu, 26 May 2011 13:50:03 -0600
From:	Jim Cromie <jim.cromie@...il.com>
To:	mingo@...e.hu
Cc:	acme@...stprotocols.net, linux-kernel@...r.kernel.org
Subject: [PATCH 0/3] perf-stat: refactor print/formatting into print-ops

 [PATCH 1/3] perf-stat: refactor print/formatting into print-ops for
 [PATCH 2/3] perf-stat: fix +- nan% in -Aa runs
 [PATCH 3/3] perf-stat: clean up <no count> handling, CPUx prefixing



[PATCH 1/3] perf-stat: refactor print/formatting into print-ops for

produces output like this for csv-output:

[jimc@...po perf]$  sudo ./perf stat -Aa -x'\t' perl -e '$i++ for 1..100000'
CPU0	task-clock-msecs	16.255272
CPU0	context-switches	12
CPU0	CPU-migrations	0
CPU0	page-faults	445
CPU0	cycles	27441481	67.07%
CPU0	instructions	36775396	73.16%
CPU0	branches	8836245	20.72%
CPU0	branch-misses	21646	14.62%
CPU0	cache-references	58221	12.31%
CPU0	cache-misses	6815	12.28%
seconds-time-elapsed	0.016211262

This alters csv-output form by printing event-name in 1st column,
swapping cols 1, 2 from previous table.  This may be problematic for
some existing users, but is more "normal" and therefore better
long-term.  The 3rd - 5th colums match those in pretty

As before, csv-output is triggered by use of -x "sep", where sep can
be any string, not just a char.  This patch adds special case for
"\t", which is converted to a tab.

[PATCH 2/3] perf-stat: fix +- nan% in -Aa runs

without this patch, I get computations like this:

[jimc@...ucho perf]$ sudo ./perf stat -r3 -Aa  perl -e '$i++ for 1..100000'

 Performance counter stats for 'perl -e $i++ for 1..100000' (3 runs):

CPU0             12.391883 task-clock-msecs         #      0.966 CPUs    ( +-    -nan% )
CPU1             12.446571 task-clock-msecs         #      0.970 CPUs    ( +-    -nan% )
CPU2             12.408014 task-clock-msecs         #      0.967 CPUs    ( +-    -nan% )
CPU3             12.422264 task-clock-msecs         #      0.968 CPUs    ( +-    -nan% )

I couldnt see anything wrong in the caller, so fixed it in stddev_stats()

[PATCH 3/3] perf-stat: clean up <no count> handling, CPUx prefixing

push <no-count> printing into print-ops to hide some cruft, and fold
away dynamic formats.  Refactor cpustr string prep into cpustr(int
cpu), use it everywhere, and change fn-sigs accordingly.

Also includes 1-line fixup for overlooked hack in earlier patch:
  -       if (scaled == -1 && !csv_output) {
  +       if (scaled == -1) {

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/