Re: [PATCH v2] md/raid1,raid10: don't handle IO error for REQ_RAHEAD and REQ_NOWAIT

[Zdenek Kabelac <zdenek.kabelac@gmail.com>]

Dne 28. 05. 25 v 3:28 Xiao Ni napsal(a):
> CC Heinz who is a dm-raid expert. But he is on holiday this week.
> 
> On Tue, May 27, 2025 at 4:24 PM Yu Kuai <yukuai1@huaweicloud.com> wrote:
>>
>> Hi, Zdenek Kabelac
>>
>> 在 2025/05/27 16:14, Yu Kuai 写道:
>>> From: Yu Kuai <yukuai3@huawei.com>
>>>
>>> IO with REQ_RAHEAD or REQ_NOWAIT can fail early, even if the storage medium
>>> is fine, hence record badblocks or remove the disk from array does not
>>> make sense.
>>>
>>> This problem if found by lvm2 test lvcreate-large-raid, where dm-zero
>>> will fail read ahead IO directly.
>>>
>>> Reported-and-tested-by: Mikulas Patocka <mpatocka@redhat.com>
>>> Closes: https://lore.kernel.org/all/34fa755d-62c8-4588-8ee1-33cb1249bdf2@redhat.com/
>>> Signed-off-by: Yu Kuai <yukuai3@huawei.com>
>>> ---
>>> Changes in v2:
>>>    - handle REQ_NOWAIT as well.
>>>
>>>    drivers/md/raid1-10.c | 10 ++++++++++
>>>    drivers/md/raid1.c    | 19 ++++++++++---------
>>>    drivers/md/raid10.c   | 11 ++++++-----
>>>    3 files changed, 26 insertions(+), 14 deletions(-)
>>>
>>
>> Just to let you know that while testing lvcreate-large-raid, the test
>> can fail sometime:
>>
>> [ 0:12.021] ## DEBUG0: 08:11:43.596775 lvcreate[8576]
>> device_mapper/ioctl/libdm-iface.c:2118  device-mapper: create ioctl on
>> LVMTEST8371vg1-LV1_rmeta_0
>> LVM-iqJjW9HItbME2d4DC2S7D58zd2omecjf0yQN83foinyxHaPoZqGEnX4rRUN7i0kH
>> failed: Device or resource busy

Hi

This issue is caused by occasional race with udev - which holds open our 
internal device open while we are deactivating raid LV.

We are currently trying to resolve this issue.

Regards

Zdenek