2016-12-24 11:07:48

by Murphy Zhou

[permalink] [raw]
Subject: LTP rwtest01 blocks on DAX mountpoint

Hi lists,

Since around 20161129 tag, LTP rwtest01 on dax mountpoint blocks
on linux-next tree, now on Linus tree.

In "normal", rwtest01 subcase ends in a few minutes, now it keeps
running for hours on dax mountpoint, both ext4 and xfs. Ctrl + c
can interrupt it.

It is always reproducible, blocking following tests.

It does not happen when mounting without dax option.
It does not happen on v4.9.

Bisect point to:

commit 4b4bb46d00b386e1c972890dc5785a7966eaa9c0
Author: Jan Kara <[email protected]>
Date: Wed Dec 14 15:07:53 2016 -0800

dax: clear dirty entry tags on cache flush


Reverting this commit on top of Linus tree "fixes" this issue.

Reproducer:

sh-4.2# cat rwt
rwtest01 export LTPROOT; rwtest -N rwtest01 -c -q -i 60s -f sync 10%25000:$TMPDIR/rw-sync-$$
sh-4.2#
mkfs.xfs /dev/pmem0p1
mount -o dax /dev/pmem0p1 /daxmnt && \
/opt/ltp/runltp -q -d /daxmnt -f rwt -p -b /dev/pmem0p2 -B xfs
umount /daxmnt

Bisect log is attached.

Thanks,
Xiong


Attachments:
(No filename) (964.00 B)
bisect2 (2.75 kB)
Download all attachments

2016-12-30 09:34:01

by Murphy Zhou

[permalink] [raw]
Subject: Re: LTP rwtest01 blocks on DAX mountpoint

On Sat, Dec 24, 2016 at 07:07:14PM +0800, Xiong Zhou wrote:
> Hi lists,
>
> Since around 20161129 tag, LTP rwtest01 on dax mountpoint blocks
> on linux-next tree, now on Linus tree.
>
> In "normal", rwtest01 subcase ends in a few minutes, now it keeps
> running for hours on dax mountpoint, both ext4 and xfs. Ctrl + c
> can interrupt it.

Test programme is waiting for a memcpy call to return.

>From sysrq output, kernel code is not blocking on somewhere,
it just wont return.

>
> It is always reproducible, blocking following tests.
>
> It does not happen when mounting without dax option.
> It does not happen on v4.9.
>
> Bisect point to:
>
> commit 4b4bb46d00b386e1c972890dc5785a7966eaa9c0
> Author: Jan Kara <[email protected]>
> Date: Wed Dec 14 15:07:53 2016 -0800
>
> dax: clear dirty entry tags on cache flush
>
>
> Reverting this commit on top of Linus tree "fixes" this issue.
>
> Reproducer:
>
> sh-4.2# cat rwt
> rwtest01 export LTPROOT; rwtest -N rwtest01 -c -q -i 60s -f sync 10%25000:$TMPDIR/rw-sync-$$
> sh-4.2#
> mkfs.xfs /dev/pmem0p1
> mount -o dax /dev/pmem0p1 /daxmnt && \
> /opt/ltp/runltp -q -d /daxmnt -f rwt -p -b /dev/pmem0p2 -B xfs
> umount /daxmnt
>
> Bisect log is attached.
>
> Thanks,
> Xiong

> git bisect start
> # bad: [50f6584e4c626b8fa39edb66f33fec27bab3996c] Merge tag 'leds_for_4.10_email_update' of git://git.kernel.org/pub/scm/linux/kernel/git/j.anaszewski/linux-leds
> git bisect bad 50f6584e4c626b8fa39edb66f33fec27bab3996c
> # good: [69973b830859bc6529a7a0468ba0d80ee5117826] Linux 4.9
> git bisect good 69973b830859bc6529a7a0468ba0d80ee5117826
> # good: [5266e70335dac35c35b5ca9cea4251c1389d4a68] Merge tag 'tty-4.10-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty
> git bisect good 5266e70335dac35c35b5ca9cea4251c1389d4a68
> # bad: [6df8b74b1720db1133ace0861cb6721bfe57819a] Merge tag 'devicetree-for-4.10' of git://git.kernel.org/pub/scm/linux/kernel/git/robh/linux
> git bisect bad 6df8b74b1720db1133ace0861cb6721bfe57819a
> # good: [f4000cd99750065d5177555c0a805c97174d1b9f] Merge tag 'arm64-upstream' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux
> git bisect good f4000cd99750065d5177555c0a805c97174d1b9f
> # bad: [e1e14ab8411df344a17687821f8f78f0a1e73cbb] radix tree test suite: delete unused rcupdate.c
> git bisect bad e1e14ab8411df344a17687821f8f78f0a1e73cbb
> # good: [f5b893c947151d424a4ab55ea3a8544b81974b31] scsi: qla4xxx: switch to pci_alloc_irq_vectors
> git bisect good f5b893c947151d424a4ab55ea3a8544b81974b31
> # good: [b9f98bd4034a3196ff068eb0fa376c5f41077480] Merge tag 'mmc-v4.10-2' of git://git.kernel.org/pub/scm/linux/kernel/git/ulfh/mmc
> git bisect good b9f98bd4034a3196ff068eb0fa376c5f41077480
> # good: [4d1f0fb096aedea7bb5489af93498a82e467c480] kernel/watchdog: use nmi registers snapshot in hardlockup handler
> git bisect good 4d1f0fb096aedea7bb5489af93498a82e467c480
> # good: [5b56d49fc31dbb0487e14ead790fc81ca9fb2c99] mm: add locked parameter to get_user_pages_remote()
> git bisect good 5b56d49fc31dbb0487e14ead790fc81ca9fb2c99
> # bad: [cfa40bcfd6fed7010b1633bf127ed8571d3b607e] radix tree test suite: benchmark for iterator
> git bisect bad cfa40bcfd6fed7010b1633bf127ed8571d3b607e
> # good: [a41b70d6dfc28b9e1a17c2a9f3181c2b614bfd54] mm: use vmf->page during WP faults
> git bisect good a41b70d6dfc28b9e1a17c2a9f3181c2b614bfd54
> # bad: [4b4bb46d00b386e1c972890dc5785a7966eaa9c0] dax: clear dirty entry tags on cache flush
> git bisect bad 4b4bb46d00b386e1c972890dc5785a7966eaa9c0
> # good: [a19e25536ed3a20845f642ce531e10c27fb2add5] mm: change return values of finish_mkwrite_fault()
> git bisect good a19e25536ed3a20845f642ce531e10c27fb2add5
> # good: [a6abc2c0e77b16480f4d2c1eb7925e5287ae1526] dax: make cache flushing protected by entry lock
> git bisect good a6abc2c0e77b16480f4d2c1eb7925e5287ae1526
> # good: [2f89dc12a25ddf995b9acd7b6543fe892e3473d6] dax: protect PTE modification on WP fault by radix tree entry lock
> git bisect good 2f89dc12a25ddf995b9acd7b6543fe892e3473d6
> # first bad commit: [4b4bb46d00b386e1c972890dc5785a7966eaa9c0] dax: clear dirty entry tags on cache flush

> _______________________________________________
> Linux-nvdimm mailing list
> [email protected]
> https://lists.01.org/mailman/listinfo/linux-nvdimm