On Wed, Dec 11 2013, Francis Moreau wrote:
> On 12/11/2013 10:52 PM, Greg Thelen wrote:
>> On Wed, Dec 11 2013, Francis Moreau wrote:
>>
>>> On 12/11/2013 06:54 PM, Greg Thelen wrote:
>>> [...]
>>>>
>>>> I don't think that 3.4 suffers the same problem. So I don't see any
>>>> reason to apply "fix shm_file deletion races" to 3.4.x. 3.4 is immune
>>>> because it use exclusive locking (spinlock) for shm-id-to-shm-object
>>>> conversions (shm_lock_check() in shmctl()). Newer kernels use rcu
>>>> protected schemes so more races are possible. Furthermore, I ran some
>>>> quick tests on v3.4.71 and did not find any crashes. So I don't think
>>>> 3.4.x needs this patch.
>>>>
>>>
>>> Ah ok, thank you for your insight.
>>>
>>> Maybe also adding a reference to the commit that introduced the bug
>>> would help but that would add extra work.
>>>
>>> Thanks
>>
>> From patch a399b29dfbaa ("ipc,shm: fix shm_file deletion races") commit
>> log:
>> Fixes: c2c737a0461e ("ipc,shm: shorten critical region for shmat")
>> Fixes: 2caacaa82a51 ("ipc,shm: shorten critical region for shmctl")
>>
>
> One strange thing though: both of those 2 patches have been merged
> during the 3.12 merge window. However the fix has been submitted to 3.10
> and 3.11 stable kernels.
Yes, the 2 patches were merged into 3.12. But they were also applied to
3.10.x and 3.11.x stable kernels. Thus the "ipc,shm: fix shm_file
deletion races" fix is relevant to 3.10.x, 3.11.x, and 3.12.
--
To unsubscribe from this list: send the line "unsubscribe stable" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html