linux-kernel - Re: [PATCH/RFC 3/9] perf symbols: Discard symbols in kallsyms for loaded modules

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:   Sun, 25 Jun 2017 22:47:13 +0900
From:   Namhyung Kim <namhyung@...nel.org>
To:     Arnaldo Carvalho de Melo <acme@...nel.org>
Cc:     Ingo Molnar <mingo@...nel.org>,
        Peter Zijlstra <a.p.zijlstra@...llo.nl>,
        Jiri Olsa <jolsa@...nel.org>,
        LKML <linux-kernel@...r.kernel.org>, kernel-team@....com,
        Masami Hiramatsu <mhiramat@...nel.org>,
        Andi Kleen <andi@...stfloor.org>,
        Adrian Hunter <adrian.hunter@...el.com>,
        Wang Nan <wangnan0@...wei.com>
Subject: Re: [PATCH/RFC 3/9] perf symbols: Discard symbols in kallsyms for
 loaded modules

On Fri, Jun 23, 2017 at 10:51:05AM -0300, Arnaldo Carvalho de Melo wrote:
> Em Fri, Jun 23, 2017 at 02:48:21PM +0900, Namhyung Kim escreveu:
> > If a module is already loaded, it should have symbols and no need to
> > load new symbols from kallsyms.  Actually kallsyms can have different
> > addresses if the module was reloaded.
> 
> Well, if it is loaded, then it should match what is in kallsyms, no?

No, the perf.data was recorded with a module, and then the module was
modified and reloaded at the same address.

Maybe the word "loaded" can be confusing - if perf.data file has
symbols in the module v1 but kallsyms has symbols in the module v2.
So it needs to read symbols from the module v1 (in the build-id cache).

>  
> > Current code just discards the first symbols only, but it should do the
> > same for all symbols in the module.  Note that the kernel doesn't set
> > the dso->loaded bit so simply checking it would do the job IMHO.
> 
> The kernel sets dso->loaded? Can you rephrase this part? And if this
> procedure doesn't set dso->loaded how can simply checking it do the job?
> 
> Can you clarify the above sentence? Please elaborate the scenario more
> fully.

Oh, I made it confusing, sorry..  I mean that it's in the middle of
the dso__load() so the (kernel) dso->loaded bit was not set yet.

Thanks,
Namhyung


> 
> - Arnaldo
>  
> > Cc: Adrian Hunter <adrian.hunter@...el.com>
> > Cc: Wang Nan <wangnan0@...wei.com>
> > Signed-off-by: Namhyung Kim <namhyung@...nel.org>
> > ---
> >  tools/perf/util/symbol.c | 9 +++++----
> >  1 file changed, 5 insertions(+), 4 deletions(-)
> > 
> > diff --git a/tools/perf/util/symbol.c b/tools/perf/util/symbol.c
> > index e7a98dbd2aed..74078ba595b3 100644
> > --- a/tools/perf/util/symbol.c
> > +++ b/tools/perf/util/symbol.c
> > @@ -787,11 +787,12 @@ static int dso__split_kallsyms(struct dso *dso, struct map *map, u64 delta)
> >  					curr_map = map;
> >  					goto discard_symbol;
> >  				}
> > -
> > -				if (curr_map->dso->loaded &&
> > -				    !machine__is_default_guest(machine))
> > -					goto discard_symbol;
> >  			}
> > +
> > +			if (curr_map->dso->loaded &&
> > +			    !machine__is_default_guest(machine))
> > +				goto discard_symbol;
> > +
> >  			/*
> >  			 * So that we look just like we get from .ko files,
> >  			 * i.e. not prelinked, relative to map->start.
> > -- 
> > 2.13.1