linux-kernel - Re: [PATCH 0/2] perf evsel: add support for inlined function in callchains

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] [day] [month] [year] [list]

Date:   Wed, 20 Feb 2019 15:34:54 +0100
From:   Jiri Olsa <jolsa@...hat.com>
To:     Jonas Rabenstein <jonas.rabenstein@...dium.uni-erlangen.de>
Cc:     linux-perf-users@...r.kernel.org,
        Peter Zijlstra <peterz@...radead.org>,
        Ingo Molnar <mingo@...hat.com>,
        Arnaldo Carvalho de Melo <acme@...nel.org>,
        Alexander Shishkin <alexander.shishkin@...ux.intel.com>,
        Namhyung Kim <namhyung@...nel.org>,
        Andi Kleen <ak@...ux.intel.com>,
        Thomas Richter <tmricht@...ux.ibm.com>,
        Stephane Eranian <eranian@...gle.com>,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH 0/2] perf evsel: add support for inlined function in
 callchains

On Tue, Feb 19, 2019 at 07:38:08PM +0100, Jonas Rabenstein wrote:
> Hi,
> sample__fprintf_callchain currently did not use the already available
> code to get the symbols of an inlined function if such information is
> available in a dso. This patchset adds the required logic to add
> appropriate lines.
> 
> As I am quite new to the code base of perf I am not sure how to test
> that changeset in a correct way. At least the codes builds and the tools
> that make use of sample__fprintf_callchain (perf-script, perf-trace and
> perf-sched as far as I can see) did not fail to run - also I did not get
> into the details of perf-trace and perf-sched as I have never used them
> before.

you could provide some examples that shows what u changed

I can see your changes change the perf script callchains,
displaying less or different callchains:

	--- old 2019-02-20 15:29:34.872312007 +0100
	+++ new 2019-02-20 15:30:47.784725456 +0100
	@@ -23,10 +23,9 @@ yes 11807 203482.590491:     246238 cycl
			      40 [unknown] ([unknown])

	 yes 11807 203482.590654:     534592 cycles:uppp:
	-           7f83e3c7fd0c _dl_addr+0x11c (/usr/lib64/libc-2.27.so)
	+           7f83e3c7fd0c __GI__dl_addr+0x11c (inlined)

	 yes 11807 203482.592371:     455976 cycles:uppp:
	-           55ddec8a3f6e [unknown] (/usr/bin/yes)
		 a790a790a790a79 [unknown] ([unknown])

	 yes 11807 203482.594176:     326514 cycles:uppp:
	@@ -38,11 +37,9 @@ yes 11807 203482.595464:     262867 cycl
		 a790a790a790a79 [unknown] ([unknown])

	 yes 11807 203482.596502:     214762 cycles:uppp:
	-           55ddec8a3f6e [unknown] (/usr/bin/yes)
		 a790a790a790a79 [unknown] ([unknown])


> 
> Another thing I am not sure how to deal with are some warnings of
> checkpatch.pl due to the 80 character line limit. Due to the long
> function names in use the current implementation already exceeded that
> limit in the same spots by even more characters as I have taken the
> inner loop and put it into a separate function.

don't worry much about the long lines warnings

> 
> I hope to expand my work to perf-report but thought it might be usefull
> to get already early feedback on those patches.

yep, good idea

thanks,
jirka