On 12/25/24 7:09 PM, Shin'ichiro Kawasaki wrote: > The current null_blk implementation checks if any bad blocks exist in > the target blocks of each IO. If so, the IO fails and data is not > transferred for all of the IO target blocks. However, when real storage > devices have bad blocks, the devices may transfer data partially up to > the first bad blocks. Especially, when the IO is a write operation, such > partial IO leaves partially written data on the device. > > To simulate such partial IO using null_blk, perform the data transfer > from the IO start block to the block just before the first bad block. > Introduce __null_handle_rq() to support partial data transfer. Modify > null_handle_badblocks() to calculate the size of the partial data > transfer and call __null_handle_rq(). We should have an option to control this behavior to be able to mimic actual devices. E.g. SAS devices may do partial data transfers before hitting a bad block, but ATA devices will not (it is always all or nothing with ATA). The current default corresponds to an ATA drive behavior and this change allows emulating a SAS drive behavior. So let's control this with an option. -- Damien Le Moal Western Digital Research
