[dm-devel] [PATCH 3/7] dm-crypt: avoid deadlock in mempools

[Ondrej Kozina]

On 02/16/2015 02:58 PM, Mike Snitzer wrote:
> On Mon, Feb 16 2015 at  4:31am -0500,
> Milan Broz <mbroz at redhat.com> wrote:
>
>> On 02/14/2015 02:14 AM, Mike Snitzer wrote:
>>> On Fri, Feb 13 2015 at  5:09P -0500,
>>> Mikulas Patocka <mpatocka at redhat.com> wrote:
>>>
>>>>
>>>>
>>>> On Fri, 13 Feb 2015, Mike Snitzer wrote:
>>>>
>>>>>    * In order to not degrade performance with excessive locking, we try
>>>>> - * non-blocking allocations without a mutex first and if it fails, we fallback
>>>>> + * non-blocking allocation without a mutex first and if it fails, we fallback
>>>>>    * to a blocking allocation with a mutex.
>>>>>    */
>>>>>   static struct bio *crypt_alloc_buffer(struct dm_crypt_io *io, unsigned size)
>>>>
>>>> There are multiple allocations, so I would leave plural there.
>>>
>>> Fixed, and tweaked the headers (already did that last time around so
>>> nothing new, you just didn't pick up my headers for your v2).  I pushed
>>> your patchset to linux-next (for 3.21), see:
>>> https://git.kernel.org/cgit/linux/kernel/git/device-mapper/linux-dm.git/log/?h=for-next
>>>
>>> Ondrej and Milan, please let us know if you hit any problems with this
>>> patchset and/or branch.
>>
>> Will try to test it soon on some strange configurations :)
>
> OK, see if you and Ondrej can re-test with urgency over the next few
> days.
>

I'll post summary of results as a reply to the top post, including short 
commentary on why we decided to implement the switch for dmtable.

For anyone interested I'll also give links to full test results.

AFAIK the only thing that should be tested again at the moment is the 
switch itself.

O.