On Wed, Nov 23, Olaf Hering wrote:

> > > +        if (!blk_split_discard(ioreq, req->sector_number, req->nr_sectors)) {
> > > +            goto err;
> > How is error handling supposed to work here?

In the guest the cmd is stuck, instead of getting an IO error:

[   91.966404] mkfs.ext4       D 0000000000000000     0  2878   2831 0x00000000
[   91.966406]  ffff88002204bc48 ffff880030530480 ffff88002fae5800 ffff88002204c000
[   91.966407]  0000000000000000 7fffffffffffffff 0000000000008000 00000000024000c0
[   91.966409]  ffff88002204bc60 ffffffff815dd985 ffff880038815c00 ffff88002204bd08
[   91.966409] Call Trace:
[   91.966413]  [<ffffffff815dd985>] schedule+0x35/0x80
[   91.966416]  [<ffffffff815e02c7>] schedule_timeout+0x237/0x2d0
[   91.966419]  [<ffffffff815dcf46>] io_schedule_timeout+0xa6/0x110
[   91.966421]  [<ffffffff815de2f3>] wait_for_completion_io+0xa3/0x110
[   91.966425]  [<ffffffff812d7b00>] submit_bio_wait+0x50/0x60
[   91.966430]  [<ffffffff812e9168>] blkdev_issue_discard+0x78/0xb0
[   91.966433]  [<ffffffff812eee2b>] blk_ioctl_discard+0x7b/0xa0
[   91.966436]  [<ffffffff812efa20>] blkdev_ioctl+0x730/0x920
[   91.966440]  [<ffffffff812318fd>] block_ioctl+0x3d/0x40
[   91.966444]  [<ffffffff8120cd6d>] do_vfs_ioctl+0x2cd/0x4a0
[   91.966453]  [<ffffffff8120cfb4>] SyS_ioctl+0x74/0x80
[   91.966456]  [<ffffffff815e142e>] entry_SYSCALL_64_fastpath+0x12/0x6d

Olaf