linux-kernel - RE: [PATCH V5 3/3] perf tool: check buildid for symoff

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:	Tue, 25 Nov 2014 17:19:23 +0000
From:	"Liang, Kan" <kan.liang@...el.com>
To:	Jiri Olsa <jolsa@...hat.com>
CC:	"acme@...nel.org" <acme@...nel.org>,
	"namhyung@...nel.org" <namhyung@...nel.org>,
	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
	"ak@...ux.intel.com" <ak@...ux.intel.com>
Subject: RE: [PATCH V5 3/3] perf tool: check buildid for symoff

> > +	data__for_each_file_new(i, d) {
> > +		k_dsos_tmp = &d->session->machines.host.kernel_dsos;
> > +		u_dsos_tmp = &d->session->machines.host.user_dsos;
> > +
> > +		if (!dsos__build_ids_equal(base_k_dsos, k_dsos_tmp))
> > +			pr_warning("The perf.data come from different
> kernel. "
> > +				   "The kernel symbol offset may vary for
> different kernel.\n");
> > +
> > +		if (!dsos__build_ids_equal(base_u_dsos, u_dsos_tmp))
> 
> looks like at this time not all dsos on the list have the buildids read..
> 

Right, I shouldn't assume all dsos have biuldids. I will modify the
dsos__build_ids_equal to check has_build_id.
If neither dsos have build_id, it will print
pr_debug ("Cannot get build_id for all dsos\n") once, and continue to do
next dso compare.

All the buildid compare related warnings will also be printed once.

> I tried to put in here the perf_session__read_build_ids call but it keeps
> showing me warning below:
> 
> > +			pr_warning("The perf.data come from different
> user binary. "
> > +				   "The user space symbol offset may vary
> for different
> > +binaries.\n");
> 
> for following workload:
> 
> [jolsa@...va perf]$ ./perf record ls
> ...
> [ perf record: Woken up 1 times to write data ] [ perf record: Captured and
> wrote 0.012 MB perf.data (~540 samples) ] [jolsa@...va perf]$ ./perf
> record ls ...
> [ perf record: Woken up 1 times to write data ] [ perf record: Captured and
> wrote 0.012 MB perf.data (~540 samples) ] [jolsa@...va perf]$ ./perf diff -
> s symoff The perf.data come from different user binary. The user space
> symbol offset may vary for different binaries.
> # Event 'cycles'
> ...

Another reason for the false warning is that we use long_name to find the dso_b.
However, the long_name is not unique.
The vmlinux dso is added when processing header. Its long name and short name
are same. (E.g. /lib/modules/3.18.0-rc3-01635-g05066a2-dirty/build/vmlinux)
When processing sample, vmlinux is loaded. Two new dsos are added for section
"init.text" and "exit.text".  They also have the long name
/lib/modules/3.18.0-rc3-01635-g05066a2-dirty/build/vmlinux. But their short
name are different, [kernel.vmlinux].init.text and [kernel.vmlinux].exit.text.

So I will change the dsos__build_ids_equal to find the dso by short_name.

Kan
> 
> 
> jirka
--
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/