From: Andrew Burgess <aburgess@redhat.com>
To: Bernd Edlinger <bernd.edlinger@hotmail.de>,
"gdb-patches@sourceware.org" <gdb-patches@sourceware.org>,
Thiago Jung Bauermann <thiago.bauermann@linaro.org>
Subject: Re: [PATCH v2] Fix sporadic XFAILs in gdb.threads/attach-many-short-lived-threads.exp
Date: Sat, 06 Apr 2024 18:40:39 +0100 [thread overview]
Message-ID: <87cyr22xuw.fsf@redhat.com> (raw)
In-Reply-To: <AS8P193MB1285AC3E42F9767F851AE50BE4022@AS8P193MB1285.EURP193.PROD.OUTLOOK.COM>
Bernd Edlinger <bernd.edlinger@hotmail.de> writes:
Thanks for looking into these failures.
> This is about random test failures like those:
>
> XFAIL: gdb.threads/attach-many-short-lived-threads.exp: iter 6: attach (EPERM)
> XFAIL: gdb.threads/attach-many-short-lived-threads.exp: iter 7: attach (EPERM)
> XFAIL: gdb.threads/attach-many-short-lived-threads.exp: iter 8: attach (EPERM)
> XFAIL: gdb.threads/attach-many-short-lived-threads.exp: iter 9: attach (EPERM)
> XFAIL: gdb.threads/attach-many-short-lived-threads.exp: iter 10: attach (EPERM)
>
> The reason for this effect is apparently as follows:
>
> There is a race condition when gdb tries to attach a thread but the
> thread exits at the same time. Normally when that happens the return
> code of ptrace(PTRACE_ATTACH, x) is EPERM, which could also have other
> reasons. To detect the true reason, we try to open /proc/<pid>/status
> which normally fails in that situation, but it may happen that the
> fopen succeeds, and the thread disappears while reading the content,
> then the read has the errno=ESRCH, use that as an indication that the
> thread has exited between fopen and reading of the status file.
I assume this would be the commit message for this change? This text
seems to be out of date given the changes in V2 and probably needs
updating.
Thanks,
Andrew
> ---
> gdb/nat/linux-procfs.c | 11 +++--------
> 1 file changed, 3 insertions(+), 8 deletions(-)
>
> v2: from kernel code review, it seems the missing "State:"
> can only happen if the thread disappeared, so no need to
> look at errno at all here.
>
> diff --git a/gdb/nat/linux-procfs.c b/gdb/nat/linux-procfs.c
> index e2086952ce6..8d46d5bf289 100644
> --- a/gdb/nat/linux-procfs.c
> +++ b/gdb/nat/linux-procfs.c
> @@ -157,17 +157,12 @@ linux_proc_pid_is_gone (pid_t pid)
> enum proc_state state;
>
> have_state = linux_proc_pid_get_state (pid, 0, &state);
> - if (have_state < 0)
> + if (have_state <= 0)
> {
> - /* If we can't open the status file, assume the thread has
> - disappeared. */
> + /* If we can't open the status file or there is no "State:" line,
> + assume the thread has disappeared. */
> return 1;
> }
> - else if (have_state == 0)
> - {
> - /* No "State:" line, assume thread is alive. */
> - return 0;
> - }
> else
> return (state == PROC_STATE_ZOMBIE || state == PROC_STATE_DEAD);
> }
> --
> 2.39.2
prev parent reply other threads:[~2024-04-06 17:40 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-04-06 4:46 Bernd Edlinger
2024-04-06 17:40 ` Andrew Burgess [this message]
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=87cyr22xuw.fsf@redhat.com \
--to=aburgess@redhat.com \
--cc=bernd.edlinger@hotmail.de \
--cc=gdb-patches@sourceware.org \
--cc=thiago.bauermann@linaro.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;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).