mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Masami Hiramatsu <mhiramat@kernel.org>
To: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Arnaldo Carvalho de Melo <acme@kernel.org>,
	linux-kernel@vger.kernel.org, Namhyung Kim <namhyung@kernel.org>,
	Peter Zijlstra <peterz@infradead.org>,
	Ingo Molnar <mingo@redhat.com>,
	Hemant Kumar <hemant@linux.vnet.ibm.com>,
	Ananth N Mavinakayanahalli <ananth@linux.vnet.ibm.com>,
	Brendan Gregg <brendan.d.gregg@gmail.com>
Subject: Re: [PATCH perf/core v7 02/21] perf buildid: Fix to set correct dso name for kallsyms
Date: Thu, 12 May 2016 17:57:40 +0900	[thread overview]
Message-ID: <20160512175740.a7efac4e423d5f78676e7ec7@kernel.org> (raw)
In-Reply-To: <20160512110212.c8bb188b823e1a637e53ae26@kernel.org>

On Thu, 12 May 2016 11:02:12 +0900
Masami Hiramatsu <mhiramat@kernel.org> wrote:

> On Wed, 11 May 2016 12:45:00 -0300
> Arnaldo Carvalho de Melo <acme@kernel.org> wrote:
> 
> > Em Wed, May 11, 2016 at 10:51:37PM +0900, Masami Hiramatsu escreveu:
> > > Fix to set correct dso name ("[kernel.kallsyms]") for
> > > kallsyms, as for vdso does.
> > 
> > Can you rewrite the above comment and also break this down in two
> > patches,
> 
> No, could you explain what parts this consists of?
> I think this patch is done just one thing. Actually I can rewrite
> this as one-line patch like below
> 
>   	if (asprintf(&filename, "%s%s%s", buildid_dir, slash ? "/" : "",
> -		     is_vdso ? DSO__NAME_VDSO : realname) < 0)
> +		     is_vdso ? DSO__NAME_VDSO : (is_kallsyms ? "[kernel.kallsyms]": realname)) < 0)
> 
> But indeed this is not readable so clean it up like below.
> 
> > probably decribing what is the problem that it fixes?
> 
> Ah, indeed. This is actually not a fix for existing bug, instead
> it will prevent buggy behavior. Current function can get any filepath
> with is_vdso = true or is_kallsyms = true, but it seems easily giving
> contradictory combination.
> 
> Hmm, OK, I think I have to solve this issue in another way.

OK, from further investigation, I found this is not a real bug, but
just an unneeded behavior changing. At this moment, I should drop this
from the series. Here is what I understand.

At first, the path of the original binary is used only when storing
build-id caches.
When loading the cache, perf uses build-id symlink to access it.

OK, so this is the callgraph of build_id_cache__dirname_from_path.
'<-' shows the caller, and with what parameters given.
 
build_id_cache__dirname_from_path
	<- build_id_cache__add_s, with (pathname, is_kallsyms, is_vdso)
		<- build_id_cache__add_b, with (sbuild_id, name, is_kallsyms, is_vdso)
			<-dso__cache_build_id
		<- build_id_cache__add_file, with (sbuild_id,filepath,false,false)
		<- build_id_cache__update_file, with (sbuild_id,filepath,false,false)
	<- build_id_cache__list_build_ids, with (pathname, false, false)

Focusing on the is_kallsyms and is_vdso, the callgraph to
build_id_cache__list_build_ids(), build_id_cache__add_file()
and build_id_cache__update_file() are not using those parameters.
These functions just pass the file path, so we can ignore.
(BTW, in this case we can not remove/purge the caches for vdso, kallsyms and
 kcore from buildid directory ...)

The last callgraph is dso__cache_build_id(), and only it uses is_kallsyms and is_vdso.
Those parameters are given as below

  is_kallsyms = dso->kernel && dso->long_name[0] != '/'
  is_vdso = dso__is_vdso(dso) <- check the dso->short_name is DSO__NAME_VDSO*
  pathname = dso->long_name

So, is_vdso == true, build_id_cache__dirname_from_path() stores all
DSO__NAME_VDSO* ("[vdso]", "[vdso32]", and "[vdsox32]") caches under
DSO__NAME_VDSO("[vdso]").

This seems is_kallsyms == true, the pathname has no real filename.

And the kernel dso will be made by machine__get_kernel(), which sets the
kernel dso's long name is whether the user given (guest)vmlinux name or
the name given by machine__mmap_name(). The machine__mmap_name() returns
[kernel.kallsyms], [guest.kernel.kallsyms] or [guest.kernel.kallsyms.<PID>].

OK, so the question is, should we do same thing on kallsyms as vdso?
This patch actually tried it, all the "[*kernel.kallsyms*]" dso caches
stored under "[kernel.kallsyms]".

Thank you,


> >  
> > > Signed-off-by: Masami Hiramatsu <mhiramat@kernel.org>
> > > ---
> > >  tools/perf/util/build-id.c |   12 ++++++++----
> > >  1 file changed, 8 insertions(+), 4 deletions(-)
> > > 
> > > diff --git a/tools/perf/util/build-id.c b/tools/perf/util/build-id.c
> > > index b6ecf87..234d8a1 100644
> > > --- a/tools/perf/util/build-id.c
> > > +++ b/tools/perf/util/build-id.c
> > > @@ -343,21 +343,25 @@ void disable_buildid_cache(void)
> > >  static char *build_id_cache__dirname_from_path(const char *name,
> > >  					       bool is_kallsyms, bool is_vdso)
> > >  {
> > > -	char *realname = (char *)name, *filename;
> > > +	const char *realname = name;
> > > +	char *filename;
> > >  	bool slash = is_kallsyms || is_vdso;
> > >  
> > >  	if (!slash) {
> > >  		realname = realpath(name, NULL);
> > >  		if (!realname)
> > >  			return NULL;
> > > -	}
> > > +	} else if (is_vdso)
> > > +		realname = DSO__NAME_VDSO;
> > > +	else
> > > +		realname = "[kernel.kallsyms]";
> > >  
> > >  	if (asprintf(&filename, "%s%s%s", buildid_dir, slash ? "/" : "",
> > > -		     is_vdso ? DSO__NAME_VDSO : realname) < 0)
> > > +		     realname) < 0)
> > >  		filename = NULL;
> > >  
> > >  	if (!slash)
> > > -		free(realname);
> > > +		free((char *)realname);
> > >  
> > >  	return filename;
> > >  }
> 
> 
> -- 
> Masami Hiramatsu <mhiramat@kernel.org>


-- 
Masami Hiramatsu <mhiramat@kernel.org>

  reply	other threads:[~2016-05-12  8:57 UTC|newest]

Thread overview: 37+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-05-11 13:51 [PATCH perf/core v7 00/21] perf-probe --cache and SDT support Masami Hiramatsu
2016-05-11 13:51 ` [PATCH perf/core v7 01/21] tools/perf: Fix lsdir to set errno correctly Masami Hiramatsu
2016-05-11 13:59   ` Arnaldo Carvalho de Melo
2016-05-12  1:46     ` Masami Hiramatsu
2016-05-12 10:25   ` [tip:perf/core] perf tools: " tip-bot for Masami Hiramatsu
2016-05-11 13:51 ` [PATCH perf/core v7 02/21] perf buildid: Fix to set correct dso name for kallsyms Masami Hiramatsu
2016-05-11 15:45   ` Arnaldo Carvalho de Melo
2016-05-12  2:02     ` Masami Hiramatsu
2016-05-12  8:57       ` Masami Hiramatsu [this message]
2016-05-11 13:51 ` [PATCH perf/core v7 03/21] perf buildid: Introduce DSO__NAME_KALLSYMS and DSO__NAME_KCORE Masami Hiramatsu
2016-05-11 15:47   ` Arnaldo Carvalho de Melo
2016-05-11 13:51 ` [PATCH perf/core v7 04/21] perf: Use SBUILD_ID_SIZE macro instead of BUILD_ID_SIZE macro Masami Hiramatsu
2016-05-11 15:47   ` Arnaldo Carvalho de Melo
2016-05-12 10:25   ` [tip:perf/core] perf tools: Use SBUILD_ID_SIZE where applicable tip-bot for Masami Hiramatsu
2016-05-11 13:52 ` [PATCH perf/core v7 05/21] perf symbol: Use lsdir for search in kcore cache directory Masami Hiramatsu
2016-05-11 15:49   ` Arnaldo Carvalho de Melo
2016-05-12 10:26   ` [tip:perf/core] perf symbols: Use lsdir() for the " tip-bot for Masami Hiramatsu
2016-05-11 13:52 ` [PATCH perf/core v7 06/21] perf-buildid-cache: Use lsdir for looking up buildid caches Masami Hiramatsu
2016-05-11 15:50   ` Arnaldo Carvalho de Melo
2016-05-12 10:26   ` [tip:perf/core] perf buildid-cache: Use lsdir() " tip-bot for Masami Hiramatsu
2016-05-11 13:52 ` [PATCH perf/core v7 07/21] perf symbol: Cleanup the code flow of dso__find_kallsyms Masami Hiramatsu
2016-05-11 15:55   ` Arnaldo Carvalho de Melo
2016-05-12  1:46     ` Masami Hiramatsu
2016-05-11 13:52 ` [PATCH perf/core v7 08/21] perf-buildid-cache: Use path/to/bin/buildid/elf instead of path/to/bin/buildid Masami Hiramatsu
2016-05-11 13:52 ` [PATCH perf/core v7 09/21] perf probe: Add --cache option to cache the probe definitions Masami Hiramatsu
2016-05-11 13:52 ` [PATCH perf/core v7 10/21] perf probe: Use cache entry if possible Masami Hiramatsu
2016-05-11 13:53 ` [PATCH perf/core v7 11/21] perf probe: Show all cached probes Masami Hiramatsu
2016-05-11 13:53 ` [PATCH perf/core v7 12/21] perf probe: Remove caches when --cache is given Masami Hiramatsu
2016-05-11 13:53 ` [PATCH perf/core v7 13/21] perf/sdt: ELF support for SDT Masami Hiramatsu
2016-05-11 13:53 ` [PATCH perf/core v7 14/21] perf probe: Add group name support Masami Hiramatsu
2016-05-11 13:54 ` [PATCH perf/core v7 15/21] perf buildid-cache: Scan and import user SDT events to probe cache Masami Hiramatsu
2016-05-11 13:54 ` [PATCH perf/core v7 16/21] perf probe: Accept %sdt and %cached event name Masami Hiramatsu
2016-05-11 13:54 ` [PATCH perf/core v7 17/21] perf-list: Show SDT and pre-cached events Masami Hiramatsu
2016-05-11 13:54 ` [PATCH perf/core v7 18/21] perf-list: Skip SDTs placed in invalid binaries Masami Hiramatsu
2016-05-11 13:54 ` [PATCH perf/core v7 19/21] perf probe: Allow wildcard for cached events Masami Hiramatsu
2016-05-11 13:54 ` [PATCH perf/core v7 20/21] perf probe: Support @BUILDID or @FILE suffix for SDT events Masami Hiramatsu
2016-05-11 13:55 ` [PATCH perf/core v7 21/21] perf probe: Support a special SDT probe format Masami Hiramatsu

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20160512175740.a7efac4e423d5f78676e7ec7@kernel.org \
    --to=mhiramat@kernel.org \
    --cc=acme@kernel.org \
    --cc=ananth@linux.vnet.ibm.com \
    --cc=brendan.d.gregg@gmail.com \
    --cc=hemant@linux.vnet.ibm.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mingo@redhat.com \
    --cc=namhyung@kernel.org \
    --cc=peterz@infradead.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®