Re: [PATCH] mm: Release a semaphore in 'get_vaddr_frames()'

[Christophe JAILLET <christophe.jaillet@wanadoo.fr>]

Le 10/12/2017 à 10:45, Michal Hocko a écrit :
> On Sat 09-12-17 08:09:41, Christophe JAILLET wrote:
>> A semaphore is acquired before this check, so we must release it before
>> leaving.
>>
>> Fixes: b7f0554a56f2 ("mm: fail get_vaddr_frames() for filesystem-dax mappings")
>> Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
>> ---
>> -- Untested --
>>
>> The wording of the commit entry and log description could be improved
>> but I didn't find something better.
> The changelog is ok imo.
>
>> ---
>>   mm/frame_vector.c | 4 +++-
>>   1 file changed, 3 insertions(+), 1 deletion(-)
>>
>> diff --git a/mm/frame_vector.c b/mm/frame_vector.c
>> index 297c7238f7d4..e0c5e659fa82 100644
>> --- a/mm/frame_vector.c
>> +++ b/mm/frame_vector.c
>> @@ -62,8 +62,10 @@ int get_vaddr_frames(unsigned long start, unsigned int nr_frames,
>>   	 * get_user_pages_longterm() and disallow it for filesystem-dax
>>   	 * mappings.
>>   	 */
>> -	if (vma_is_fsdax(vma))
>> +	if (vma_is_fsdax(vma)) {
>> +		up_read(&mm->mmap_sem);
>>   		return -EOPNOTSUPP;
>> +	}
> Is there any reason to do a different error handling than other error
> paths? Namely not going without goto out?

You are right, I misread the code after out:.
I thought it would override the returned value, but I was wrong.
'goto out' is definitively better, IMHO. I'll propose a v2.

CJ