Hi, 在 2023/09/26 7:59, Kirill Kirilenko 写道:
On 25.09.2023 05:58 +0300, Yu Kuai wrote:Roman and Kirill, can you test the following patch? Thanks, Kuai diff --git a/drivers/md/raid1.c b/drivers/md/raid1.c index 4b30a1742162..4963f864ef99 100644 --- a/drivers/md/raid1.c +++ b/drivers/md/raid1.c @@ -1345,6 +1345,7 @@ static void raid1_write_request(struct mddev *mddev, struct bio *bio, int first_clone; int max_sectors; bool write_behind = false; + bool is_discard = (bio_op(bio) == REQ_OP_DISCARD); if (mddev_is_clustered(mddev) && md_cluster_ops->area_resyncing(mddev, WRITE, @@ -1405,7 +1406,7 @@ static void raid1_write_request(struct mddev *mddev, struct bio *bio, * write-mostly, which means we could allocate write behind * bio later. */ - if (rdev && test_bit(WriteMostly, &rdev->flags)) + if (!is_discard && rdev && test_bit(WriteMostly, &rdev->flags)) write_behind = true; if (rdev && unlikely(test_bit(Blocked, &rdev->flags))) {Thank you. I can confirm, that your patch eliminates freezes during 'fstrim' execution. Tested on kernel 6.5.0. Still 'fstrim' takes more than 2 minutes, but I believe it's normal to a file system with 1M+ inodes.
Thanks for the test.
Probably I'm wrong here, but to me this doesn't look like a solution, more like a masking the real problem. Even with TRIM operations split in 1MB pieces, I don't expect kernel to freeze.
I still don't quite understand what you mean 'kernel freeze', this patch indeed fix a problem that diskcard bio is treated as normal write bio and it's splitted. Can you explain more by how do you judge 'kernel freeze'? In the meantime dose 'iostat -dmx 1' shows that disk is idle and no dicard io is handled? Thanks, Kuai
.
