[PATCH 01/14] dma-buf: add dma_resv_for_each_fence_unlocked

[Christian König]

Am 14.09.21 um 19:04 schrieb Daniel Vetter:
> On Fri, Sep 10, 2021 at 10:26:42AM +0200, Christian König wrote:
>> Abstract the complexity of iterating over all the fences
>> in a dma_resv object.
>>
>> The new loop handles the whole RCU and retry dance and
>> returns only fences where we can be sure we grabbed the
>> right one.
>>
>> Signed-off-by: Christian König <christian.koenig at amd.com>
>> ---
>>   drivers/dma-buf/dma-resv.c | 63 ++++++++++++++++++++++++++++++++++++++
>>   include/linux/dma-resv.h   | 36 ++++++++++++++++++++++
>>   2 files changed, 99 insertions(+)
>>
>> diff --git a/drivers/dma-buf/dma-resv.c b/drivers/dma-buf/dma-resv.c
>> index 84fbe60629e3..213a9b7251ca 100644
>> --- a/drivers/dma-buf/dma-resv.c
>> +++ b/drivers/dma-buf/dma-resv.c
>> @@ -323,6 +323,69 @@ void dma_resv_add_excl_fence(struct dma_resv *obj, struct dma_fence *fence)
>>   }
>>   EXPORT_SYMBOL(dma_resv_add_excl_fence);
>>   
>> +/**
>> + * dma_resv_walk_unlocked - walk over fences in a dma_resv obj
>> + * @obj: the dma_resv object
>> + * @cursor: cursor to record the current position
>> + * @all_fences: true returns also the shared fences
>> + * @first: if we should start over
>> + *
>> + * Return all the fences in the dma_resv object which are not yet signaled.
>> + * The returned fence has an extra local reference so will stay alive.
>> + * If a concurrent modify is detected the whole iterator is started over again.
>> + */
>> +struct dma_fence *dma_resv_walk_unlocked(struct dma_resv *obj,
>> +					 struct dma_resv_cursor *cursor,
>> +					 bool all_fences, bool first)
>> +{
>> +	struct dma_fence *fence = NULL;
>> +
>> +	do {
>> +		/* Drop the reference from the previous round */
>> +		dma_fence_put(fence);
>> +
>> +		cursor->is_first = first;
>> +		if (first) {
>> +			cursor->seq = read_seqcount_begin(&obj->seq);
>> +			cursor->index = -1;
>> +			cursor->fences = dma_resv_shared_list(obj);
>> +			cursor->is_exclusive = true;
>> +
>> +			fence = dma_resv_excl_fence(obj);
>> +			if (fence && test_bit(DMA_FENCE_FLAG_SIGNALED_BIT,
>> +					      &fence->flags))
>> +				fence = NULL;
>> +		} else {
>> +			fence = NULL;
>> +		}
>> +
>> +		if (fence) {
>> +			fence = dma_fence_get_rcu(fence);
>> +		} else if (all_fences && cursor->fences) {
>> +			struct dma_resv_list *fences = cursor->fences;
>> +
>> +			cursor->is_exclusive = false;
>> +			while (++cursor->index < fences->shared_count) {
>> +				fence = rcu_dereference(fences->shared[
>> +							cursor->index]);
>> +				if (!test_bit(DMA_FENCE_FLAG_SIGNALED_BIT,
>> +					      &fence->flags))
>> +					break;
>> +			}
>> +			if (cursor->index < fences->shared_count)
>> +				fence = dma_fence_get_rcu(fence);
>> +			else
>> +				fence = NULL;
>> +		}
>> +
>> +		/* For the eventually next round */
>> +		first = true;
>> +	} while (read_seqcount_retry(&obj->seq, cursor->seq));
>> +
>> +	return fence;
>> +}
>> +EXPORT_SYMBOL_GPL(dma_resv_walk_unlocked);
>> +
>>   /**
>>    * dma_resv_copy_fences - Copy all fences from src to dst.
>>    * @dst: the destination reservation object
>> diff --git a/include/linux/dma-resv.h b/include/linux/dma-resv.h
>> index 9100dd3dc21f..f5b91c292ee0 100644
>> --- a/include/linux/dma-resv.h
>> +++ b/include/linux/dma-resv.h
>> @@ -149,6 +149,39 @@ struct dma_resv {
>>   	struct dma_resv_list __rcu *fence;
>>   };
>>   
>> +/**
>> + * struct dma_resv_cursor - current position into the dma_resv fences
>> + * @seq: sequence number to check
>> + * @index: index into the shared fences
>> + * @shared: the shared fences
>> + * @is_first: true if this is the first returned fence
>> + * @is_exclusive: if the current fence is the exclusive one
>> + */
>> +struct dma_resv_cursor {
>> +	unsigned int seq;
>> +	unsigned int index;
>> +	struct dma_resv_list *fences;
>> +	bool is_first;
>> +	bool is_exclusive;
>> +};
> A bit a bikeshed, but I think I'd be nice to align this with the other
> iterators we have, e.g. for the drm_connector list.
>
> So struct dma_resv_fence_iter, dma_resv_fence_iter_begin/next/end().

I've renamed the structure to dma_resv_iter.

> Also I think the for_each macro must not include begin/end calls. If we
> include that then it saves 2 lines of code at the cost of a pile of
> awkward bugs because people break; out of the loop or return early  (only
> continue is safe) and we leak a fence. Or worse.
>
> Explicit begin/end is much more robust at a very marginal cost imo.

The key point is that this makes it quite a bunch more complicated to 
implement. See those functions are easiest when you centralize them and 
try to not spread the functionality into begin/end.

The only thing I could see in the end function would be to drop the 
reference for the dma_fence and that is not really something I would 
like to do because we actually need to keep that reference in a bunch of 
cases.

Regards,
Christian.

>
> Otherwise I think this fence iterator is a solid concept that yeah we
> should roll out everywhere.
> -Daniel
>
>> +
>> +/**
>> + * dma_resv_for_each_fence_unlocked - fence iterator
>> + * @obj: a dma_resv object pointer
>> + * @cursor: a struct dma_resv_cursor pointer
>> + * @all_fences: true if all fences should be returned
>> + * @fence: the current fence
>> + *
>> + * Iterate over the fences in a struct dma_resv object without holding the
>> + * dma_resv::lock. The RCU read side lock must be hold when using this, but can
>> + * be dropped and re-taken as necessary inside the loop. @all_fences controls
>> + * if the shared fences are returned as well.
>> + */
>> +#define dma_resv_for_each_fence_unlocked(obj, cursor, all_fences, fence)    \
>> +	for (fence = dma_resv_walk_unlocked(obj, cursor, all_fences, true); \
>> +	     fence; dma_fence_put(fence),				    \
>> +	     fence = dma_resv_walk_unlocked(obj, cursor, all_fences, false))
>> +
>>   #define dma_resv_held(obj) lockdep_is_held(&(obj)->lock.base)
>>   #define dma_resv_assert_held(obj) lockdep_assert_held(&(obj)->lock.base)
>>   
>> @@ -366,6 +399,9 @@ void dma_resv_fini(struct dma_resv *obj);
>>   int dma_resv_reserve_shared(struct dma_resv *obj, unsigned int num_fences);
>>   void dma_resv_add_shared_fence(struct dma_resv *obj, struct dma_fence *fence);
>>   void dma_resv_add_excl_fence(struct dma_resv *obj, struct dma_fence *fence);
>> +struct dma_fence *dma_resv_walk_unlocked(struct dma_resv *obj,
>> +					 struct dma_resv_cursor *cursor,
>> +					 bool first, bool all_fences);
>>   int dma_resv_get_fences(struct dma_resv *obj, struct dma_fence **pfence_excl,
>>   			unsigned *pshared_count, struct dma_fence ***pshared);
>>   int dma_resv_copy_fences(struct dma_resv *dst, struct dma_resv *src);
>> -- 
>> 2.25.1
>>