From: Linus Torvalds <torvalds@linux-foundation.org>
To: Andy Lutomirski <luto@amacapital.net>
Cc: Dave Jones <davej@codemonkey.org.uk>, Chris Mason <clm@fb.com>,
Andy Lutomirski <luto@kernel.org>, Jens Axboe <axboe@fb.com>,
Al Viro <viro@zeniv.linux.org.uk>, Josef Bacik <jbacik@fb.com>,
David Sterba <dsterba@suse.com>,
linux-btrfs <linux-btrfs@vger.kernel.org>,
Linux Kernel <linux-kernel@vger.kernel.org>
Subject: Re: bio linked list corruption.
Date: Mon, 24 Oct 2016 14:17:55 -0700 [thread overview]
Message-ID: <CA+55aFyqUvb1BEraMp2LGK7Hm1xwzQzwsWb+KykpZiYSQ5cUEA@mail.gmail.com> (raw)
In-Reply-To: <CA+55aFycKG_7qpSs_pH7ibnOYL9vM85UMGhUFEGC-qpB4qkb5A@mail.gmail.com>
On Mon, Oct 24, 2016 at 1:46 PM, Linus Torvalds
<torvalds@linux-foundation.org> wrote:
>
> So this is all some really subtle code, but I'm not seeing that it
> would be wrong.
Ahh... Except maybe..
The vmalloc/vfree code itself is a bit scary. In particular, we have a
rather insane model of TLB flushing. We leave the virtual area on a
lazy purge-list, and we delay flushing the TLB and actually freeing
the virtual memory for it so that we can batch things up.
But we've free'd the physical pages that are *mapped* by that area
when we do the vfree(). So there can be stale TLB entries that point
to pages that have gotten re-used. They shouldn't matter, because
nothing should be writing to those pages, but it strikes me that this
may also be hurting the DEBUG_PAGEALLOC thing. Maybe we're not getting
the page fautls that we *should* be getting, and there are hidden
reads and writes to those paghes that already got free'd.\
There was some nasty reason why we did that insane thing. I think it
was just that there are a few high-frequency vmalloc/vfree users and
the TLB flushing was killing some performance.
But it does strike me that we are playing very fast and loose with the
TLB on the vmalloc area.
So maybe all the new VMAP code is fine, and it's really vmalloc/vfree
that has been subtly broken but nobody has ever cared before?
Linus
next prev parent reply other threads:[~2016-10-24 21:17 UTC|newest]
Thread overview: 117+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-10-11 14:45 btrfs bio linked list corruption Dave Jones
2016-10-11 15:11 ` Al Viro
2016-10-11 15:19 ` Dave Jones
2016-10-11 15:20 ` Chris Mason
2016-10-11 15:49 ` Dave Jones
2016-10-11 15:54 ` Chris Mason
2016-10-11 16:25 ` Dave Jones
2016-10-12 13:47 ` Dave Jones
2016-10-12 14:40 ` Dave Jones
2016-10-12 14:42 ` Chris Mason
2016-10-13 18:16 ` Dave Jones
2016-10-13 21:18 ` Chris Mason
2016-10-13 21:56 ` Dave Jones
2016-10-16 0:42 ` Dave Jones
2016-10-18 1:07 ` Chris Mason
2016-10-18 22:42 ` Dave Jones
2016-10-18 23:12 ` Jens Axboe
2016-10-18 23:31 ` Chris Mason
2016-10-18 23:36 ` Jens Axboe
2016-10-18 23:39 ` Linus Torvalds
2016-10-18 23:42 ` Chris Mason
2016-10-19 0:10 ` Linus Torvalds
2016-10-19 0:19 ` Chris Mason
2016-10-19 0:28 ` Linus Torvalds
2016-10-20 22:48 ` Dave Jones
2016-10-19 1:05 ` Andy Lutomirski
2016-10-20 22:50 ` Dave Jones
2016-10-20 23:01 ` Andy Lutomirski
2016-10-20 23:03 ` Dave Jones
2016-10-20 23:23 ` Andy Lutomirski
2016-10-21 20:02 ` Dave Jones
2016-10-21 20:17 ` Chris Mason
2016-10-21 20:23 ` Dave Jones
2016-10-21 20:38 ` Chris Mason
2016-10-21 20:41 ` Josef Bacik
2016-10-21 21:11 ` Dave Jones
2016-10-22 15:20 ` Dave Jones
2016-10-23 21:32 ` Chris Mason
2016-10-24 4:40 ` Dave Jones
2016-10-24 20:06 ` Andy Lutomirski
2016-10-24 20:46 ` Linus Torvalds
2016-10-24 21:17 ` Linus Torvalds [this message]
2016-10-24 21:50 ` Linus Torvalds
2016-10-24 22:02 ` Chris Mason
2016-10-24 22:42 ` Andy Lutomirski
2016-10-25 0:00 ` Linus Torvalds
2016-10-25 1:09 ` Andy Lutomirski
[not found] ` <77d9983d-a00a-1dc1-a9a1-631de1d0c146@fb.com>
2016-10-26 0:27 ` Dave Jones
2016-10-26 1:33 ` Linus Torvalds
2016-10-26 1:39 ` Linus Torvalds
2016-10-26 16:30 ` Dave Jones
2016-10-26 16:48 ` Linus Torvalds
2016-10-26 18:18 ` Dave Jones
2016-10-26 18:42 ` Dave Jones
2016-10-26 19:06 ` Linus Torvalds
2016-10-26 20:00 ` Chris Mason
2016-10-26 21:52 ` Chris Mason
2016-10-26 22:21 ` Linus Torvalds
2016-10-26 22:40 ` Dave Jones
2016-10-26 22:51 ` Linus Torvalds
2016-10-26 22:55 ` Jens Axboe
2016-10-26 22:58 ` Linus Torvalds
2016-10-26 23:03 ` Jens Axboe
2016-10-26 23:07 ` Dave Jones
2016-10-26 23:08 ` Linus Torvalds
2016-10-26 23:20 ` Jens Axboe
2016-10-26 23:38 ` Chris Mason
2016-10-26 23:47 ` Dave Jones
2016-10-27 0:00 ` Jens Axboe
2016-10-27 13:33 ` Chris Mason
2016-10-31 18:55 ` Dave Jones
2016-10-31 19:35 ` Linus Torvalds
2016-10-31 19:44 ` Chris Mason
2016-11-06 16:55 ` btrfs btree_ctree_super fault Dave Jones
2016-11-08 14:59 ` Dave Jones
2016-11-08 15:08 ` Chris Mason
2016-11-10 14:35 ` Dave Jones
2016-11-10 15:27 ` Chris Mason
2016-11-23 19:34 ` bio linked list corruption Dave Jones
2016-11-23 19:58 ` Dave Jones
2016-12-01 15:32 ` btrfs_destroy_inode warn (outstanding extents) Dave Jones
2016-12-03 16:48 ` Dave Jones
2016-12-07 16:15 ` Dave Jones
2016-12-09 21:12 ` Steven Rostedt
2016-12-04 23:04 ` bio linked list corruption Vegard Nossum
2016-12-05 11:10 ` Vegard Nossum
2016-12-05 17:09 ` Vegard Nossum
2016-12-05 17:21 ` Dave Jones
2016-12-05 17:55 ` Linus Torvalds
2016-12-05 19:11 ` Vegard Nossum
2016-12-05 20:10 ` Linus Torvalds
2016-12-05 20:35 ` Linus Torvalds
2016-12-05 21:33 ` Vegard Nossum
2016-12-06 8:42 ` Vegard Nossum
2016-12-06 8:16 ` Peter Zijlstra
2016-12-06 8:36 ` Ingo Molnar
2016-12-06 16:33 ` Linus Torvalds
2016-12-05 20:10 ` Vegard Nossum
2016-12-05 18:11 ` Andy Lutomirski
2016-12-05 18:25 ` Linus Torvalds
2016-12-05 18:26 ` Vegard Nossum
2016-10-26 23:19 ` Chris Mason
2016-10-26 23:21 ` Jens Axboe
2016-10-27 6:33 ` Christoph Hellwig
2016-10-27 16:34 ` Linus Torvalds
2016-10-27 16:36 ` Jens Axboe
2016-10-26 23:01 ` Dave Jones
2016-10-26 23:05 ` Jens Axboe
2016-10-26 22:52 ` Jens Axboe
2016-10-26 22:07 ` Linus Torvalds
2016-10-26 22:54 ` Chris Mason
2016-10-27 5:41 ` Dave Chinner
2016-10-27 17:23 ` Dave Jones
2016-10-19 17:09 ` Philipp Hahn
2016-10-19 17:43 ` Linus Torvalds
2016-10-20 6:52 ` Ingo Molnar
2016-10-20 7:17 ` Thomas Gleixner
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=CA+55aFyqUvb1BEraMp2LGK7Hm1xwzQzwsWb+KykpZiYSQ5cUEA@mail.gmail.com \
--to=torvalds@linux-foundation.org \
--cc=axboe@fb.com \
--cc=clm@fb.com \
--cc=davej@codemonkey.org.uk \
--cc=dsterba@suse.com \
--cc=jbacik@fb.com \
--cc=linux-btrfs@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=luto@amacapital.net \
--cc=luto@kernel.org \
--cc=viro@zeniv.linux.org.uk \
/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).