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@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);elsefence = 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);-- 2.25.1