Re: [PATCH] block: update locking context for blk_free_devt

[Jens Axboe <axboe@fb.com>]

On 2014-10-13 14:26, Dmitry Monakhov wrote:
> Jens Axboe <axboe@fb.com> writes:
>
>> On 2014-10-13 12:35, Dmitry Monakhov wrote:
>>> After 2da78092 this function will not longer sleepy
>>>
>>> Signed-off-by: Dmitry Monakhov <dmonakhov@openvz.org>
>>> ---
>>>    block/genhd.c |    4 +---
>>>    1 files changed, 1 insertions(+), 3 deletions(-)
>>>
>>> diff --git a/block/genhd.c b/block/genhd.c
>>> index 09da5e4..ae9aaa5 100644
>>> --- a/block/genhd.c
>>> +++ b/block/genhd.c
>>> @@ -441,12 +441,10 @@ int blk_alloc_devt(struct hd_struct *part, dev_t *devt)
>>>     * Free @devt which was allocated using blk_alloc_devt().
>>>     *
>>>     * CONTEXT:
>>> - * Might sleep.
>>> + * Don't care.
>>>     */
>>>    void blk_free_devt(dev_t devt)
>>>    {
>>> -	might_sleep();
>>> -
>>>    	if (devt == MKDEV(0, 0))
>>>    		return;
>>
>> What's this against? The might_sleep() was killed about a month ago by
>> commit 46f341ffcfb5. The comment wasn't updated though, it should be.
> Against 764f612c6c3c23  linux-block/for-next.
> I've found this during development other blockdev related feature.
> can you please point what is your development branch.

Ah, I see. The devel branches are always for-3.x/core and 
for-3.x/drivers, so right now the current ones are for-3.18/core and so 
forth. That fix went into master, and I generally don't pull into the 
devel branches unless I really have to. This particular one should 
probably have gone in, since it spews a warning. Generally I expect 
people running them to use for-next, which really should be merged with 
master. I will just pull it in. for-next can and will be rebased 
sometimes though, only the real devel branches are more or less set in 
stone when it comes to history. So for development purposes, those 
should be used and not for-next.

-- 
Jens Axboe