Re: [PATCH v3 1/2] gnttab: Introduce rwlock to protect updates to grant table state

["Egger, Christoph" <chegger@amazon.de>]

On 2015/01/09 15:10, Jan Beulich wrote:
>>>> On 09.01.15 at 14:05, <chegger@amazon.de> wrote:
>> On 2014/12/18 12:37, Jan Beulich wrote:
>>>>>> On 03.12.14 at 15:29, <chegger@amazon.de> wrote:
>>>> --- a/xen/common/grant_table.c
>>>> +++ b/xen/common/grant_table.c
>> [...]
>>>> @@ -944,6 +944,7 @@ __gnttab_unmap_common(
>>>>      }
>>>>  
>>>>      op->rd = rd;
>>>> +    read_lock(&rgt->lock);
>>>>      act = &active_entry(rgt, op->map->ref);
>>>>  
>>>>      if ( op->frame == 0 )
>>>
>>> The nesting of the two locks should be mentioned in the doc change.
>>
>> Where do you see the nesting coming from?
> 
> Reproducing the subsequent hunk I had commented on
> 
>> @@ -1004,6 +1005,7 @@ __gnttab_unmap_common(
>>  
>>   unmap_out:
>>      double_gt_unlock(lgt, rgt);
>> +    read_unlock(&rgt->lock);
> 
> there obviously are two (or really three) locks in simultaneous use
> here.

Ah, I see now. My mistake is I was looking for multiple subsequent
read_lock(&rgt->lock); calls and found nothing.

Christoph