linux-toradex.git/drivers/dma-buf/sync_file.c, branch v4.11

dma-fence: Wrap querying the fence->status

2017-01-09T14:36:58+00:00

The fence->status is an optional field that is only valid once the fence
has been signaled. (Driver may fill the fence->status with an error code
prior to calling dma_fence_signal().) Given the restriction upon its
validity, wrap querying of the fence->status into a helper
dma_fence_get_status().

Signed-off-by: Chris Wilson 
Reviewed-by: Daniel Vetter 
Reviewed-by: Sumit Semwal 
Signed-off-by: Sumit Semwal 
Link: http://patchwork.freedesktop.org/patch/msgid/20170104141222.6992-2-chris@chris-wilson.co.uk

dma-buf: Reorganize device dma access docs

2016-12-13T11:39:51+00:00

- Put the initial overview for dma-buf into dma-buf.rst.
- Put all the comments about detailed semantics into the right
  kernel-doc comment for functions or ops structure member.
- To allow that detail, switch the reworked kerneldoc to inline style
  for dma_buf_ops.
- Tie everything together into a much more streamlined overview
  comment, relying on the hyperlinks for all the details.
- Also sprinkle some links into the kerneldoc for dma_buf and
  dma_buf_attachment to tie it all together.

Cc: linux-doc@vger.kernel.org
Cc: Jonathan Corbet 
Cc: Sumit Semwal 
Signed-off-by: Daniel Vetter 
Signed-off-by: Sumit Semwal 
Link: http://patchwork.freedesktop.org/patch/msgid/20161209185309.1682-4-daniel.vetter@ffwll.ch

dma-buf: Update kerneldoc for sync_file_create

2016-12-13T11:39:39+00:00

This was missed when adding a dma_fence_get call. While at it also
remove the kerneldoc for the static inline helper - no point
documenting internals down to every detail.

Fixes: 30cd85dd6edc ("dma-buf/sync_file: hold reference to fence when creating sync_file")
Cc: Gustavo Padovan 
Cc: Sean Paul 
Cc: linux-doc@vger.kernel.org
Cc: Jonathan Corbet 
Cc: Sumit Semwal 
Signed-off-by: Daniel Vetter 
Reviewed-by: Gustavo Padovan 
Signed-off-by: Sumit Semwal 
Link: http://patchwork.freedesktop.org/patch/msgid/20161209185309.1682-3-daniel.vetter@ffwll.ch

Revert "dma-buf/sync-file: Avoid enable fence signaling if poll(.timeout=0)"

2016-11-18T09:35:58+00:00

This reverts commit ecebca79f6976ddaddfd054d699272515869ea28.

Do not enable fence callback on poll() when using fence_array causes the
fence_array to not signal.

For now we will revert the change and enable signaling everytime time
poll is called with timeout=0 as well.

Cc: Chris Wilson 
Signed-off-by: Gustavo Padovan 
Acked-by: Chris Wilson 
Signed-off-by: Daniel Vetter 
Link: http://patchwork.freedesktop.org/patch/msgid/1479457603-30758-1-git-send-email-gustavo@padovan.org

dma-buf: Rename struct fence to dma_fence

2016-10-25T12:40:39+00:00

I plan to usurp the short name of struct fence for a core kernel struct,
and so I need to rename the specialised fence/timeline for DMA
operations to make room.

A consensus was reached in
https://lists.freedesktop.org/archives/dri-devel/2016-July/113083.html
that making clear this fence applies to DMA operations was a good thing.
Since then the patch has grown a bit as usage increases, so hopefully it
remains a good thing!

(v2...: rebase, rerun spatch)
v3: Compile on msm, spotted a manual fixup that I broke.
v4: Try again for msm, sorry Daniel

coccinelle script:
@@

@@
- struct fence
+ struct dma_fence
@@

@@
- struct fence_ops
+ struct dma_fence_ops
@@

@@
- struct fence_cb
+ struct dma_fence_cb
@@

@@
- struct fence_array
+ struct dma_fence_array
@@

@@
- enum fence_flag_bits
+ enum dma_fence_flag_bits
@@

@@
(
- fence_init
+ dma_fence_init
|
- fence_release
+ dma_fence_release
|
- fence_free
+ dma_fence_free
|
- fence_get
+ dma_fence_get
|
- fence_get_rcu
+ dma_fence_get_rcu
|
- fence_put
+ dma_fence_put
|
- fence_signal
+ dma_fence_signal
|
- fence_signal_locked
+ dma_fence_signal_locked
|
- fence_default_wait
+ dma_fence_default_wait
|
- fence_add_callback
+ dma_fence_add_callback
|
- fence_remove_callback
+ dma_fence_remove_callback
|
- fence_enable_sw_signaling
+ dma_fence_enable_sw_signaling
|
- fence_is_signaled_locked
+ dma_fence_is_signaled_locked
|
- fence_is_signaled
+ dma_fence_is_signaled
|
- fence_is_later
+ dma_fence_is_later
|
- fence_later
+ dma_fence_later
|
- fence_wait_timeout
+ dma_fence_wait_timeout
|
- fence_wait_any_timeout
+ dma_fence_wait_any_timeout
|
- fence_wait
+ dma_fence_wait
|
- fence_context_alloc
+ dma_fence_context_alloc
|
- fence_array_create
+ dma_fence_array_create
|
- to_fence_array
+ to_dma_fence_array
|
- fence_is_array
+ dma_fence_is_array
|
- trace_fence_emit
+ trace_dma_fence_emit
|
- FENCE_TRACE
+ DMA_FENCE_TRACE
|
- FENCE_WARN
+ DMA_FENCE_WARN
|
- FENCE_ERR
+ DMA_FENCE_ERR
)
 (
 ...
 )

Signed-off-by: Chris Wilson 
Reviewed-by: Gustavo Padovan 
Acked-by: Sumit Semwal 
Acked-by: Christian König 
Signed-off-by: Daniel Vetter 
Link: http://patchwork.freedesktop.org/patch/msgid/20161025120045.28839-1-chris@chris-wilson.co.uk

dma-buf/sync_file: hold reference to fence when creating sync_file

2016-10-21T14:49:06+00:00

fence referencing was out of balance. It was not taking any ref to the
fence at creating time, but it was putting a reference when freeing the
sync file.

This patch fixes the balancing issue by getting a reference for the fence
when creating the sync_file.

Signed-off-by: Gustavo Padovan 
Signed-off-by: Sean Paul 
Link: http://patchwork.freedesktop.org/patch/msgid/1476899313-22241-1-git-send-email-gustavo@padovan.org

dma-buf/sync_file: free fences array in num_fences is 1

2016-09-22T11:06:16+00:00

When merging sync_files there is a case when we can end up with only one
fence in the merged sync_file: when all fences belong to the same
timeline.

So for this case a fence_array is not created instead we just assigned the
fence to sync_file->fence. Then we do not use the fences array anymore nor
does free it.

This patch frees the array.

Reported-by:  Chris Wilson 
Reviewed-by: Chris Wilson 
Signed-off-by: Gustavo Padovan 
Signed-off-by: Sean Paul 
Link: http://patchwork.freedesktop.org/patch/msgid/1474442419-6040-1-git-send-email-gustavo@padovan.org

dma-buf/sync_file: Increment refcount of fence when all are signaled.

2016-09-20T12:23:39+00:00

When we merge several fences, if all of them are signaled already, we
still keep one of them. So instead of using add_fence(), which will not
increase the refcount of signaled fences, we should explicitly call
fence_get() for the fence we are keeping.

This patch fixes a kernel panic that can be triggered by creating a fence
that is expired (or increasing the timeline until it expires), then
creating a merged fence out of it, and deleting the merged fence. This
will make the original expired fence's refcount go to zero.

Testcase: igt/sw_sync/sync_expired_merge
Signed-off-by: Rafael Antognolli 
Reviewed-by: Chris Wilson 
Reviewed-by: Gustavo Padovan 
Signed-off-by: Sumit Semwal 
Link: http://patchwork.freedesktop.org/patch/msgid/1473966865-4508-1-git-send-email-rafael.antognolli@intel.com

dma-buf/sync-file: Avoid enable fence signaling if poll(.timeout=0)

2016-09-13T14:39:49+00:00

If we being polled with a timeout of zero, a nonblocking busy query,
we don't need to install any fence callbacks as we will not be waiting.
As we only install the callback once, the overhead comes from the atomic
bit test that also causes serialisation between threads.

Signed-off-by: Chris Wilson 
Cc: Sumit Semwal 
Cc: Gustavo Padovan 
Cc: linux-media@vger.kernel.org
Cc: dri-devel@lists.freedesktop.org
Cc: linaro-mm-sig@lists.linaro.org
Reviewed-by: Gustavo Padovan 
Signed-off-by: Sumit Semwal 
Link: http://patchwork.freedesktop.org/patch/msgid/20160829181613.30722-1-chris@chris-wilson.co.uk

dma-buf/sync_file: only enable fence signalling on poll()

2016-08-11T10:09:38+00:00

Signalling doesn't need to be enabled at sync_file creation, it is only
required if userspace waiting the fence to signal through poll().

Thus we delay fence_add_callback() until poll is called. It only adds the
callback the first time poll() is called. This avoid re-adding the same
callback multiple times.

v2: rebase and update to work with new fence support for sync_file

v3: use atomic operation to set enabled and protect fence_add_callback()

v4: use user bit from fence flags (comment from Chris Wilson)

v5: use ternary if on poll return (comment from Chris Wilson)

Signed-off-by: Gustavo Padovan 
Reviewed-by: Chris Wilson 
Signed-off-by: Sumit Semwal 
  [sumits: remove unused var status]
Link: http://patchwork.freedesktop.org/patch/msgid/1470404378-27961-1-git-send-email-gustavo@padovan.org