linux-perf-users.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: "Alex Xu (Hello71)" <alex_y_xu@yahoo.ca>
To: acme@kernel.org, mingo@redhat.com, peterz@infradead.org,
	Tony Garnock-Jones <tonyg@leastfixedpoint.com>
Cc: linux-perf-users@vger.kernel.org
Subject: Re: [PATCH] tools/perf: Use long-running addr2line per dso
Date: Thu, 09 Sep 2021 12:52:31 -0400	[thread overview]
Message-ID: <1631205749.uhek9ofd69.none@localhost> (raw)
In-Reply-To: <20210909112202.1947499-1-tonyg@leastfixedpoint.com>

Excerpts from Tony Garnock-Jones's message of September 9, 2021 7:22 am:
> Invoking addr2line in a separate subprocess, one for each required
> lookup, takes a terribly long time. This patch introduces a
> long-running addr2line process for each dso, *DRAMATICALLY* speeding
> up runs of perf. What used to take tens of minutes now takes tens of
> seconds.
> 
> Debian bug report about this issue:
> https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=911815
> 
> Signed-off-by: Tony Garnock-Jones <tonyg@leastfixedpoint.com>

This patch seems awfully complicated, especially considering it still 
uses relatively slow stdio instead of direct library calls. Did you look 
into calling elfutils instead of libbfd if the latter is not available? 
It is GPLv2 compatible, so can be linked against perf for distribution. 
Additionally, elfutils is already used by perf for unwinding, but not 
addr2line.

On the patch specifically, why was popen changed to socketpair? It looks 
like it adds significant complexity, and the advanced features of 
socketpair are not being used (SOCK_DGRAM/SOCK_SEQPACKET, SCM_RIGHTS).

Regards,
Alex.

  reply	other threads:[~2021-09-09 16:52 UTC|newest]

Thread overview: 17+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-09-09 11:22 [PATCH] tools/perf: Use long-running addr2line per dso Tony Garnock-Jones
2021-09-09 16:52 ` Alex Xu (Hello71) [this message]
2021-09-09 20:19   ` Arnaldo Carvalho de Melo
2021-09-10  8:58     ` Tony Garnock-Jones
2021-09-10  8:54   ` Tony Garnock-Jones
2021-09-09 19:05 ` Ian Rogers
2021-09-10  8:57   ` Tony Garnock-Jones
2021-09-10 10:23 ` [PATCH v2] " Tony Garnock-Jones
2021-09-10 22:45   ` Ian Rogers
2021-09-10 22:55     ` Tony Garnock-Jones
2021-09-10 22:55   ` [PATCH v3] " Tony Garnock-Jones
2021-09-11  0:25     ` Ian Rogers
2021-09-13 20:05     ` Arnaldo Carvalho de Melo
2021-09-16 12:09       ` [PATCH v4] " Tony Garnock-Jones
2021-10-01  0:29         ` Ian Rogers
2021-10-01  1:20           ` Arnaldo Carvalho de Melo
2021-10-04 12:29         ` Arnaldo Carvalho de Melo

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=1631205749.uhek9ofd69.none@localhost \
    --to=alex_y_xu@yahoo.ca \
    --cc=acme@kernel.org \
    --cc=linux-perf-users@vger.kernel.org \
    --cc=mingo@redhat.com \
    --cc=peterz@infradead.org \
    --cc=tonyg@leastfixedpoint.com \
    /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;
as well as URLs for NNTP newsgroup(s).