From: Arnaldo Carvalho de Melo <acme@kernel.org>
To: Noel Grandin <noelgrandin@gmail.com>
Cc: "linux-perf-use." <linux-perf-users@vger.kernel.org>
Subject: Re: why is 'perf record --call-graph' using addr2line?
Date: Mon, 9 Feb 2026 16:39:11 -0300 [thread overview]
Message-ID: <aYo334TnekI8UgpY@x1> (raw)
In-Reply-To: <CAFYHVnXddP=V-q=0Z3Na1FOBaQRT3RA55ruDY=O0pNpbaSc4jg@mail.gmail.com>
On Mon, Feb 09, 2026 at 09:24:47PM +0200, Noel Grandin wrote:
> Hi
>
> I am running
>
> $ perf record --call-graph=dwarf,32000 my_program.bin
>
> quite a lot.
>
> Until a couple of days ago, I would do this, hit ctrl-C, it would
> finish writing stuff to disk in a few seconds, and then I would
> proceed with using KDAB hotspot to analyse the data.
>
> Then I did a routine 'apt update' of my ubuntu box, and now....
> Now, it takes 30+ minutes before perf exits, and it is spending all of
> that time running addr2line.
Are you sure? addr2line can be used when using '-s srcline' or '-s
srcfile', but then in 'perf report', not on record.
Doing some systemwide recording here I don't see such calls to
addr2line, be it the function inside perf or the addr2line utility:
root@number:/# perf probe -x ~/bin/perf addr2line
Added new event:
probe_perf:addr2line (on addr2line in /home/acme/bin/perf)
You can now use it in all perf tools, such as:
perf record -e probe_perf:addr2line -aR sleep 1
root@number:/# perf trace -e execve,probe_perf:addr2line perf record --call-graph=dwarf,32000
? ( ): perf/1708890 ... [continued]: execve()) = 0
^C[ perf record: Woken up 302 times to write data ]
root@number:/# ls -[ perf record: Captured and wrote 89.499 MB perf.data (5420 samples) ]
la perf.data
-rw-------. 1 root root 93924024 Feb 9 16:37 perf.data
root@number:/# ls -lah perf.data
-rw-------. 1 root root 90M Feb 9 16:37 perf.data
root@number:/# perf evlist -v
cpu/cycles/P: type: 0 (PERF_TYPE_HARDWARE), size: 144, config: 0 (PERF_COUNT_HW_CPU_CYCLES), { sample_period, sample_freq }: 4000, sample_type: IP|TID|TIME|CALLCHAIN|CPU|PERIOD|REGS_USER|STACK_USER|IDENTIFIER, read_format: ID|LOST, disabled: 1, freq: 1, precise_ip: 2, sample_id_all: 1, exclude_callchain_user: 1, sample_regs_user: 0xff0fff, sample_stack_user: 32000
dummy:u: type: 1 (PERF_TYPE_SOFTWARE), size: 144, config: 0x9 (PERF_COUNT_SW_DUMMY), { sample_period, sample_freq }: 1, sample_type: IP|TID|TIME|CPU|IDENTIFIER, read_format: ID|LOST, exclude_kernel: 1, exclude_hv: 1, mmap: 1, comm: 1, task: 1, mmap_data: 1, sample_id_all: 1, exclude_guest: 1, mmap2: 1, comm_exec: 1, ksymbol: 1, bpf_event: 1, build_id: 1
root@number:/#
> Any ideas?
Can you give more details? Do you have anything in ~/.perfconfig, etc?
> Thanks, Noel Grandin
next prev parent reply other threads:[~2026-02-09 19:39 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-02-09 19:24 why is 'perf record --call-graph' using addr2line? Noel Grandin
2026-02-09 19:39 ` Arnaldo Carvalho de Melo [this message]
2026-02-17 11:47 ` Noel Grandin
2026-02-18 2:05 ` Namhyung Kim
2026-02-18 6:47 ` Noel Grandin
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=aYo334TnekI8UgpY@x1 \
--to=acme@kernel.org \
--cc=linux-perf-users@vger.kernel.org \
--cc=noelgrandin@gmail.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