From: Timothy Pearson <tpearson@raptorengineering.com>
To: Timothy Pearson <tpearson@raptorengineering.com>
Cc: Jens Axboe <axboe@kernel.dk>,
regressions <regressions@lists.linux.dev>,
Pavel Begunkov <asml.silence@gmail.com>
Subject: Re: Regression in io_uring, leading to data corruption
Date: Mon, 13 Nov 2023 23:06:44 -0600 (CST) [thread overview]
Message-ID: <1357741503.47170425.1699938404661.JavaMail.zimbra@raptorengineeringinc.com> (raw)
In-Reply-To: <1270025902.47126081.1699923171139.JavaMail.zimbra@raptorengineeringinc.com>
----- Original Message -----
> From: "Timothy Pearson" <tpearson@raptorengineering.com>
> To: "Jens Axboe" <axboe@kernel.dk>
> Cc: "regressions" <regressions@lists.linux.dev>, "Pavel Begunkov" <asml.silence@gmail.com>
> Sent: Monday, November 13, 2023 6:52:51 PM
> Subject: Re: Regression in io_uring, leading to data corruption
>
> Seemed to do what I wanted, however corruption remains. Was worth a try...
>
> I guess I'll proceed under the assumption that somehow the thread setup is
> stomping on userspace for now.
Finally found it! Patch here:
https://lore.kernel.org/regressions/19221908.47168775.1699937769845.JavaMail.zimbra@raptorengineeringinc.com/T/#u
We were missing the appropriate barrier instruction inside the IPI issued by the task_work_add() call, which was in turn issued during queued io worker creation. The explaination for how userspace was getting corrupted boils down to an inconsistent view of main memory as seen the two cores involved in the worker handoff and new worker creation.
Who doesn't love a one line fix after a week and a half of pulling one's hair out? ;)
next prev parent reply other threads:[~2023-11-14 5:06 UTC|newest]
Thread overview: 95+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-11-07 16:34 Regression in io_uring, leading to data corruption Timothy Pearson
2023-11-07 16:49 ` Jens Axboe
2023-11-07 16:57 ` Timothy Pearson
2023-11-07 17:14 ` Jens Axboe
2023-11-07 21:22 ` Jens Axboe
2023-11-07 21:39 ` Timothy Pearson
2023-11-07 21:46 ` Jens Axboe
2023-11-07 22:07 ` Timothy Pearson
2023-11-07 22:16 ` Jens Axboe
2023-11-07 22:29 ` Timothy Pearson
2023-11-07 22:44 ` Jens Axboe
2023-11-07 23:12 ` Timothy Pearson
2023-11-07 23:16 ` Jens Axboe
2023-11-07 23:34 ` Timothy Pearson
2023-11-07 23:52 ` Jens Axboe
2023-11-08 0:02 ` Timothy Pearson
2023-11-08 0:09 ` Jens Axboe
2023-11-08 3:27 ` Timothy Pearson
2023-11-08 3:30 ` Timothy Pearson
2023-11-08 4:00 ` Timothy Pearson
2023-11-08 15:10 ` Jens Axboe
2023-11-08 15:14 ` Jens Axboe
2023-11-08 17:10 ` Timothy Pearson
2023-11-08 17:26 ` Jens Axboe
2023-11-08 17:40 ` Timothy Pearson
2023-11-08 17:49 ` Jens Axboe
2023-11-08 17:57 ` Jens Axboe
2023-11-08 18:36 ` Timothy Pearson
2023-11-08 18:51 ` Timothy Pearson
2023-11-08 19:08 ` Jens Axboe
2023-11-08 19:06 ` Jens Axboe
2023-11-08 22:05 ` Jens Axboe
2023-11-08 22:15 ` Timothy Pearson
2023-11-08 22:18 ` Jens Axboe
2023-11-08 22:28 ` Timothy Pearson
2023-11-08 23:58 ` Jens Axboe
2023-11-09 15:12 ` Jens Axboe
2023-11-09 17:00 ` Timothy Pearson
2023-11-09 17:17 ` Jens Axboe
2023-11-09 17:24 ` Timothy Pearson
2023-11-09 17:30 ` Jens Axboe
2023-11-09 17:36 ` Timothy Pearson
2023-11-09 17:38 ` Jens Axboe
2023-11-09 17:42 ` Timothy Pearson
2023-11-09 17:45 ` Jens Axboe
2023-11-09 18:20 ` tpearson
2023-11-10 3:51 ` Jens Axboe
2023-11-10 4:35 ` Timothy Pearson
2023-11-10 6:48 ` Timothy Pearson
2023-11-10 14:52 ` Jens Axboe
2023-11-11 18:42 ` Timothy Pearson
2023-11-11 18:58 ` Jens Axboe
2023-11-11 19:04 ` Timothy Pearson
2023-11-11 19:11 ` Jens Axboe
2023-11-11 19:15 ` Timothy Pearson
2023-11-11 19:23 ` Jens Axboe
2023-11-11 21:57 ` Timothy Pearson
2023-11-13 17:06 ` Timothy Pearson
2023-11-13 17:39 ` Jens Axboe
2023-11-13 19:02 ` Timothy Pearson
2023-11-13 19:29 ` Jens Axboe
2023-11-13 20:58 ` Timothy Pearson
2023-11-13 21:22 ` Timothy Pearson
2023-11-13 22:15 ` Jens Axboe
2023-11-13 23:19 ` Timothy Pearson
2023-11-13 23:48 ` Jens Axboe
2023-11-14 0:04 ` Timothy Pearson
2023-11-14 0:13 ` Jens Axboe
2023-11-14 0:52 ` Timothy Pearson
2023-11-14 5:06 ` Timothy Pearson [this message]
2023-11-14 13:17 ` Jens Axboe
2023-11-14 16:59 ` Timothy Pearson
2023-11-14 17:04 ` Jens Axboe
2023-11-14 17:14 ` Timothy Pearson
2023-11-14 17:17 ` Jens Axboe
2023-11-14 17:21 ` Timothy Pearson
2023-11-14 17:57 ` Timothy Pearson
2023-11-14 18:02 ` Jens Axboe
2023-11-14 18:12 ` Timothy Pearson
2023-11-14 18:26 ` Jens Axboe
2023-11-15 11:03 ` Timothy Pearson
2023-11-15 16:46 ` Jens Axboe
2023-11-15 17:03 ` Timothy Pearson
2023-11-15 18:30 ` Jens Axboe
2023-11-15 18:35 ` Timothy Pearson
2023-11-15 18:37 ` Jens Axboe
2023-11-15 18:40 ` Timothy Pearson
2023-11-15 19:00 ` Jens Axboe
2023-11-16 3:28 ` Timothy Pearson
2023-11-16 3:46 ` Jens Axboe
2023-11-16 3:54 ` Timothy Pearson
2023-11-19 0:16 ` Timothy Pearson
2023-11-13 20:47 ` Jens Axboe
2023-11-13 21:08 ` Timothy Pearson
2023-11-10 14:48 ` Jens Axboe
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=1357741503.47170425.1699938404661.JavaMail.zimbra@raptorengineeringinc.com \
--to=tpearson@raptorengineering.com \
--cc=asml.silence@gmail.com \
--cc=axboe@kernel.dk \
--cc=regressions@lists.linux.dev \
/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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.