[Intel-gfx] [PATCH 7/8] dma-buf: add reservation_object_fences helper

Christian König ckoenig.leichtzumerken at gmail.com
Wed Aug 7 09:06:57 UTC 2019


Am 06.08.19 um 21:24 schrieb Chris Wilson:
> Quoting Christian König (2019-08-06 16:01:33)
>> Add a new helper to get a consistent set of pointers from the reservation
>> object. While at it group all access helpers together in the header file.
> Ah, needs to be earlier :)

Ah, crap. That got incorrectly reordered while moving the fixes to the 
beginning of the set.

>   
>> +/**
>> + * reservation_object_fences - read consistent fence pointers
>> + * @obj: reservation object where we get the fences from
>> + * @excl: pointer for the exclusive fence
>> + * @list: pointer for the shared fence list
>> + *
>> + * Make sure we have a consisten exclusive fence and shared fence list.
>> + * Must be called with rcu read side lock held.
>> + */
>> +static inline void
>> +reservation_object_fences(struct reservation_object *obj,
>> +                         struct dma_fence **excl,
>> +                         struct reservation_object_list **list)
>> +{
>> +       unsigned int seq;
>> +
>> +       do {
>> +               seq = read_seqcount_begin(&obj->seq);
>> +               *excl = rcu_dereference(obj->fence_excl);
>> +               *list = rcu_dereference(obj->fence);
>> +       } while (read_seqcount_retry(&obj->seq, seq));
>> +}
> I would personally prefer return excl rather than have it as a second
> outparam, but I'd leave that to gcc to decide.
>
> Having stared at this, I agree this does the right thing. The important
> point from all callers' perspective is that the combination of pointers
> is consistent for this rcu_read_lock. And rcu_dereference enforces the
> callers do hold rcu_read_lock.
>
> I didn't check all the conversions, just stared at the heart of the
> problem.
>
> Reviewed-by: Chris Wilson <chris at chris-wilson.co.uk>

Thanks.

Going to fix that up,
Christian.

> -Chris



More information about the Intel-gfx mailing list