On 8/17/20 2:52 AM, Ming Lei wrote: > Hello, > > The 1st patch checks max discard segment limit in request merge code, > and one discard request failure is fixed for virtio_blk. > > The 2nd patch fixes handling of single max discard segment for virtio_blk, > and potential memory corruption is fixed. > > The 3rd patch renames the confusing blk_discard_mergable(). > > The biggest problem is that virtio_blk handles discard request in > multi-range style, however it(at least qemu) may claim that multi range > discard isn't supported by returning 1 for max discard segment. I have applied this, but dropped patch #3. We can do the rename for 5.10 if need be. -- Jens Axboe
