Re: dm: Fix oops when clone_and_map_rq returns !DM_MAPIO_REMAPPED

[Junichi Nomura <j-nomura@ce.jp.nec.com>]

On 05/27/15 22:50, Mike Snitzer wrote:
>> Hi Junichi,
>>
>> In reviewing this patch I wondered if it better to xplicitly check for a
>> return of DM_MAPIO_REQUEUE in map_request() since that is the only other
>> return that is possible.  I'm still on the fence but your patch is more
>> conservative and at least we won't go on to try to setup_clone, etc if
>> for some reason in the future a new DM_MAPIO_* were invented and
>> returned from clone_and_map_rq().

Either way should work. But I wanted to make it explicit
to call setup_clone() only when DM_MAPIO_REMAPPED is returned.

>> I do intend to revise the header slightly to make explicit references to
>> function names in some places to improve clarity.  I'll have to double
>> check but I _think_ this should cc stable@ too since blk-mq support was
>> added in Linux 4.0 (IIRC).
> 
> FYI, here is the revised header:
> https://git.kernel.org/cgit/linux/kernel/git/device-mapper/linux-dm.git/commit/?h=dm-4.1&id=3a1407559a593d4360af12dd2df5296bf8eb0d28

Thanks for the nice revision.

-- 
Jun'ichi Nomura, NEC Corporation