| Message ID | 20240719042318.265227-1-shinichiro.kawasaki@wdc.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | [blktests,v2] dm/002: avoid device access by udev at dmsetup remove | expand |
On 7/19/24 6:23 AM, Shin'ichiro Kawasaki wrote: > The test case dm/002 rarely fails with the message below: > > dm/002 => nvme0n1 (dm-dust general functionality test) [failed] > runtime 0.204s ... 0.174s > --- tests/dm/002.out 2024-06-14 14:37:40.480794693 +0900 > +++ /home/shin/Blktests/blktests/results/nvme0n1/dm/002.out.bad 2024-06-14 21:38:18.588976499 +0900 > @@ -7,4 +7,6 @@ > countbadblocks: 0 badblock(s) found > countbadblocks: 3 badblock(s) found > countbadblocks: 0 badblock(s) found > +device-mapper: remove ioctl on dust1 failed: Device or resource busy > +Command failed. > Test complete > modprobe: FATAL: Module dm_dust is in use. > > This failure happens when udev opens the dm device at "dmsetup remove" > command. To avoid the failure, call "udevadm settle" before the "dmsetup > remove" command. I think udevadm settle is overkill as it waits for everything, not only that device. Did you consider to use "dmsetup remove --retry <dev>"? This is one liner and you do not need to implement the retry yourself. We have many such situations in cryptsetup tests and --retry was enough to fix it (as the busy comes usually from blkid scan that is fast enough). It will print a few benign messages while retrying, though. (Just a hint, I am not nacking the patch :) Milan > > Signed-off-by: Shin'ichiro Kawasaki <shinichiro.kawasaki@wdc.com> > --- > This patch addresses a failure found during the debug work for another > dm/002 failure [1]. Per discussion on v1 patch [2], do "udevadm settle" > instead of retrying "dmsetup remove". > > [1] https://lore.kernel.org/linux-block/42ecobcsduvlqh77iavjj2p3ewdh7u4opdz4xruauz4u5ddljz@yr7ye4fq72tr/ > [2] https://lore.kernel.org/linux-block/o5wik4yvo2teffpjlwycbaek6znrtde5kml3hkof5r2w5rxttj@bhokt2ksdcbj/ > > tests/dm/002 | 1 + > 1 file changed, 1 insertion(+) > > diff --git a/tests/dm/002 b/tests/dm/002 > index fae3986..afa174a 100755 > --- a/tests/dm/002 > +++ b/tests/dm/002 > @@ -37,6 +37,7 @@ test_device() { > sync > dmsetup message dust1 0 countbadblocks > sync > + udevadm settle > dmsetup remove dust1 > > echo "Test complete"
On Jul 19, 2024 / 09:14, Milan Broz wrote: > On 7/19/24 6:23 AM, Shin'ichiro Kawasaki wrote: > > The test case dm/002 rarely fails with the message below: > > > > dm/002 => nvme0n1 (dm-dust general functionality test) [failed] > > runtime 0.204s ... 0.174s > > --- tests/dm/002.out 2024-06-14 14:37:40.480794693 +0900 > > +++ /home/shin/Blktests/blktests/results/nvme0n1/dm/002.out.bad 2024-06-14 21:38:18.588976499 +0900 > > @@ -7,4 +7,6 @@ > > countbadblocks: 0 badblock(s) found > > countbadblocks: 3 badblock(s) found > > countbadblocks: 0 badblock(s) found > > +device-mapper: remove ioctl on dust1 failed: Device or resource busy > > +Command failed. > > Test complete > > modprobe: FATAL: Module dm_dust is in use. > > > > This failure happens when udev opens the dm device at "dmsetup remove" > > command. To avoid the failure, call "udevadm settle" before the "dmsetup > > remove" command. > > I think udevadm settle is overkill as it waits for everything, not only that device. > > Did you consider to use "dmsetup remove --retry <dev>"? This is one > liner and you do not need to implement the retry yourself. > > We have many such situations in cryptsetup tests and --retry was enough > to fix it (as the busy comes usually from blkid scan that is fast enough). > It will print a few benign messages while retrying, though. > > (Just a hint, I am not nacking the patch :) Thanks Milan, I was not aware of the --retry option. I confirmed it avoids the failure. I think it's the better than "udevadm settle". Will post v3.
diff --git a/tests/dm/002 b/tests/dm/002 index fae3986..afa174a 100755 --- a/tests/dm/002 +++ b/tests/dm/002 @@ -37,6 +37,7 @@ test_device() { sync dmsetup message dust1 0 countbadblocks sync + udevadm settle dmsetup remove dust1 echo "Test complete"
The test case dm/002 rarely fails with the message below: dm/002 => nvme0n1 (dm-dust general functionality test) [failed] runtime 0.204s ... 0.174s --- tests/dm/002.out 2024-06-14 14:37:40.480794693 +0900 +++ /home/shin/Blktests/blktests/results/nvme0n1/dm/002.out.bad 2024-06-14 21:38:18.588976499 +0900 @@ -7,4 +7,6 @@ countbadblocks: 0 badblock(s) found countbadblocks: 3 badblock(s) found countbadblocks: 0 badblock(s) found +device-mapper: remove ioctl on dust1 failed: Device or resource busy +Command failed. Test complete modprobe: FATAL: Module dm_dust is in use. This failure happens when udev opens the dm device at "dmsetup remove" command. To avoid the failure, call "udevadm settle" before the "dmsetup remove" command. Signed-off-by: Shin'ichiro Kawasaki <shinichiro.kawasaki@wdc.com> --- This patch addresses a failure found during the debug work for another dm/002 failure [1]. Per discussion on v1 patch [2], do "udevadm settle" instead of retrying "dmsetup remove". [1] https://lore.kernel.org/linux-block/42ecobcsduvlqh77iavjj2p3ewdh7u4opdz4xruauz4u5ddljz@yr7ye4fq72tr/ [2] https://lore.kernel.org/linux-block/o5wik4yvo2teffpjlwycbaek6znrtde5kml3hkof5r2w5rxttj@bhokt2ksdcbj/ tests/dm/002 | 1 + 1 file changed, 1 insertion(+)