public inbox for [email protected]
 help / color / mirror / Atom feed
From: Amir Goldstein <[email protected]>
To: Zorro Lang <[email protected]>
Cc: [email protected], [email protected],
	[email protected], Christian Brauner <[email protected]>,
	Jens Axboe <[email protected]>
Subject: Re: [xfstests generic/617] fsx io_uring dio starts to fail on overlayfs since v6.6-rc1
Date: Sun, 24 Sep 2023 18:45:38 +0300	[thread overview]
Message-ID: <CAOQ4uxgpC-qgE0BvtBYW0JfqyJ-Yf8jj84wLwm+6XzRrjNsXHg@mail.gmail.com> (raw)
In-Reply-To: <20230924142754.ejwsjen5pvyc32l4@dell-per750-06-vm-08.rhts.eng.pek2.redhat.com>

On Sun, Sep 24, 2023 at 6:15 PM Zorro Lang <[email protected]> wrote:
>
> Hi,
>
> The generic/617 of fstests is a test case does IO_URING soak direct-IO
> fsx test, but recently (about from v6.6-rc1 to now) it always fails on
> overlayfs as [1], no matter the underlying fs is ext4 or xfs. But it
> never failed on overlay before, likes [2].
>
> So I thought it might be a regression of overlay or io-uring on current v6.6.
> Please help to review, it's easy to reproduce. My system is Fedora-rawhide/RHEL-9,
> with upstream mainline linux HEAD=dc912ba91b7e2fa74650a0fc22cccf0e0d50f371.
> The generic/617.full output as [3].
>

Hi Zorro,

Thank you for the report.
I am on public holiday, so it may take me a few days to get to look at this.
In the meanwhile could you try to bisect if the regression is introduced
by this vfs PR merge (of my patches):

 *   de16588a7737 - Merge tag 'v6.6-vfs.misc' of
git://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs
|\
| * e6fa4c728fb6 - cachefiles: use kiocb_{start,end}_write() helpers
| * 8f7371268a4b - ovl: use kiocb_{start,end}_write() helpers
| * 8c3cfa80fd1e - aio: use kiocb_{start,end}_write() helpers
| * e484fd73f4bd - io_uring: use kiocb_{start,end}_write() helpers

The io_uring PR merge:

*   c1b7fcf3f6d9 - Merge tag 'for-6.6/io_uring-2023-08-28' of
git://git.kernel.dk/linux

Or the overlayfs PR merge:

*   63580f669d7f - Merge tag 'ovl-update-6.6' of
git://git.kernel.org/pub/scm/linux/kernel/git/overlayfs/vfs

Thanks,
Amir.

>
> [1]
> FSTYP         -- overlay
> PLATFORM      -- Linux/x86_64 dell-xxxx-xxx 6.6.0-rc2+ #1 SMP PREEMPT_DYNAMIC Fri Sep 22 15:41:10 EDT 2023
> MKFS_OPTIONS  -- -m crc=1,finobt=1,rmapbt=0,reflink=1,inobtcount=1,bigtime=1 /mnt/xfstests/scratch
> MOUNT_OPTIONS -- -o context=system_u:object_r:root_t:s0 /mnt/xfstests/scratch /mnt/xfstests/scratch/ovl-mnt
>
> generic/617       - output mismatch (see /var/lib/xfstests/results//generic/617.out.bad)
>     --- tests/generic/617.out   2023-09-22 16:08:35.444572181 -0400
>     +++ /var/lib/xfstests/results//generic/617.out.bad  2023-09-22 19:33:29.240901008 -0400
>     @@ -1,2 +1,54 @@
>      QA output created by 617
>     +uring write bad io length: 0 instead of 8192
>     +short write: 0x0 bytes instead of 0x2000
>     +LOG DUMP (46 total operations):
>     +1(  1 mod 256): FALLOC   0x28c0d thru 0x2953a      (0x92d bytes) EXTENDING
>     +2(  2 mod 256): TRUNCATE UP        from 0x2953a to 0x81000
>     +3(  3 mod 256): ZERO     0x6f07 thru 0x14790       (0xd88a bytes)
>     ...
>     (Run 'diff -u /var/lib/xfstests/tests/generic/617.out /var/lib/xfstests/results//generic/617.out.bad'  to see the entire diff)
> Ran: generic/617
> Failures: generic/617
> Failed 1 of 1 tests
>
> [2]
> FSTYP         -- overlay
> PLATFORM      -- Linux/x86_64 hp-xxxxxx-xx 6.5.0-rc6+ #1 SMP PREEMPT_DYNAMIC Mon Aug 14 12:45:06 UTC 2023
> MKFS_OPTIONS  -- /mnt/scratch
> MOUNT_OPTIONS -- -o context=system_u:object_r:root_t:s0 /mnt/scratch /mnt/scratch/ovl-mnt
>
> generic/617        5s
> Ran: generic/617
> Passed all 1 tests
>
> [3]
> /var/lib/xfstests/ltp/fsx -S 0 -U -q -N 20000 -p 200 -o 128000 -l 600000 -r 4096 -t 4096 -w 4096 -Z /mnt/xfstests/test/ovl-mnt/junk
> uring write bad io length: 0 instead of 8192
> short write: 0x0 bytes instead of 0x2000
> LOG DUMP (46 total operations):
> 1(  1 mod 256): FALLOC   0x28c0d thru 0x2953a   (0x92d bytes) EXTENDING
> 2(  2 mod 256): TRUNCATE UP     from 0x2953a to 0x81000
> 3(  3 mod 256): ZERO     0x6f07 thru 0x14790    (0xd88a bytes)
> 4(  4 mod 256): ZERO     0x4a097 thru 0x5dfc8   (0x13f32 bytes)
> 5(  5 mod 256): TRUNCATE DOWN   from 0x81000 to 0x10000
> 6(  6 mod 256): MAPWRITE 0x74000 thru 0x78de7   (0x4de8 bytes)
> 7(  7 mod 256): SKIPPED (no operation)
> 8(  8 mod 256): MAPWRITE 0x52000 thru 0x57bf5   (0x5bf6 bytes)
> 9(  9 mod 256): DEDUPE 0x19000 thru 0x20fff     (0x8000 bytes) to 0x57000 thru 0x5efff
> 10( 10 mod 256): COPY 0x64000 thru 0x6bfff      (0x8000 bytes) to 0x1c000 thru 0x23fff
> 11( 11 mod 256): INSERT 0x38000 thru 0x46fff    (0xf000 bytes)
> 12( 12 mod 256): WRITE    0x16000 thru 0x1cfff  (0x7000 bytes)
> 13( 13 mod 256): MAPWRITE 0x3f000 thru 0x444d5  (0x54d6 bytes)
> 14( 14 mod 256): INSERT 0x83000 thru 0x8cfff    (0xa000 bytes)
> 15( 15 mod 256): COPY 0x54000 thru 0x5dfff      (0xa000 bytes) to 0x6d000 thru 0x76fff
> 16( 16 mod 256): PUNCH    0x34f3d thru 0x48c45  (0x13d09 bytes)
> 17( 17 mod 256): FALLOC   0x47bd5 thru 0x5a950  (0x12d7b bytes) INTERIOR
> 18( 18 mod 256): READ     0x20000 thru 0x3cfff  (0x1d000 bytes)
> 19( 19 mod 256): READ     0xe000 thru 0x1cfff   (0xf000 bytes)
> 20( 20 mod 256): MAPREAD  0x70000 thru 0x8d512  (0x1d513 bytes)
> 21( 21 mod 256): PUNCH    0x12773 thru 0x1f5b1  (0xce3f bytes)
> 22( 22 mod 256): DEDUPE 0x81000 thru 0x8ffff    (0xf000 bytes) to 0x3b000 thru 0x49fff
> 23( 23 mod 256): CLONE 0x78000 thru 0x7efff     (0x7000 bytes) to 0x1e000 thru 0x24fff
> 24( 24 mod 256): MAPREAD  0x1000 thru 0x1b6c1   (0x1a6c2 bytes)
> 25( 25 mod 256): SKIPPED (no operation)
> 26( 26 mod 256): FALLOC   0x43b4c thru 0x5ecb7  (0x1b16b bytes) INTERIOR
> 27( 27 mod 256): SKIPPED (no operation)
> 28( 28 mod 256): WRITE    0x3000 thru 0x8fff    (0x6000 bytes)
> 29( 29 mod 256): COPY 0x19000 thru 0x20fff      (0x8000 bytes) to 0x79000 thru 0x80fff
> 30( 30 mod 256): WRITE    0x19000 thru 0x28fff  (0x10000 bytes)
> 31( 31 mod 256): DEDUPE 0x3a000 thru 0x3dfff    (0x4000 bytes) to 0x49000 thru 0x4cfff
> 32( 32 mod 256): ZERO     0x8f290 thru 0x927bf  (0x3530 bytes)
> 33( 33 mod 256): PUNCH    0x45aec thru 0x58137  (0x1264c bytes)
> 34( 34 mod 256): SKIPPED (no operation)
> 35( 35 mod 256): FALLOC   0x5b567 thru 0x613ab  (0x5e44 bytes) INTERIOR
> 36( 36 mod 256): ZERO     0x1abc7 thru 0x1bbd6  (0x1010 bytes)
> 37( 37 mod 256): MAPWRITE 0x44000 thru 0x45411  (0x1412 bytes)
> 38( 38 mod 256): FALLOC   0x3b222 thru 0x4de57  (0x12c35 bytes) INTERIOR
> 39( 39 mod 256): COLLAPSE 0x1e000 thru 0x20fff  (0x3000 bytes)
> 40( 40 mod 256): INSERT 0x62000 thru 0x64fff    (0x3000 bytes)
> 41( 41 mod 256): DEDUPE 0xf000 thru 0x16fff     (0x8000 bytes) to 0x78000 thru 0x7ffff
> 42( 42 mod 256): WRITE    0x4000 thru 0x12fff   (0xf000 bytes)
> 43( 43 mod 256): COPY 0x62000 thru 0x79fff      (0x18000 bytes) to 0x2000 thru 0x19fff
> 44( 44 mod 256): SKIPPED (no operation)
> 45( 45 mod 256): MAPWRITE 0x22000 thru 0x297c9  (0x77ca bytes)
> 46( 46 mod 256): WRITE    0x41000 thru 0x42fff  (0x2000 bytes)
> Log of operations saved to "/mnt/xfstests/test/ovl-mnt/junk.fsxops"; replay with --replay-ops
> Correct content saved for comparison
> (maybe hexdump "/mnt/xfstests/test/ovl-mnt/junk" vs "/mnt/xfstests/test/ovl-mnt/junk.fsxgood")
>

  reply	other threads:[~2023-09-24 15:45 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-09-24 14:29 [xfstests generic/617] fsx io_uring dio starts to fail on overlayfs since v6.6-rc1 Zorro Lang
2023-09-24 15:45 ` Amir Goldstein [this message]
2023-09-24 17:52 ` Jens Axboe
2023-09-24 19:16   ` Jens Axboe
2023-09-25  6:35   ` Zorro Lang

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=CAOQ4uxgpC-qgE0BvtBYW0JfqyJ-Yf8jj84wLwm+6XzRrjNsXHg@mail.gmail.com \
    [email protected] \
    [email protected] \
    [email protected] \
    [email protected] \
    [email protected] \
    [email protected] \
    [email protected] \
    /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