GitLab

Currently there is a possibility of spinlock recursion in
kgsl_timeline_signal and kgsl_ioctl_timeline_destroy. The issue can happen
if two fences are created on a timeline and KGSL_TIMELINE_WAIT_ANY flag is
used. First fence is signalled and once the wait is complete, second fence
is signalled. This results in a spinlock recursion as the release of first
fence is called when the timeline->lock is held. Fix this by making sure
that the lock to protect the list and lock used for dma_fence signalling
are used correctly. Also make fence_lock per timeline to reduce lock
contention among different timelines.

Change-Id: I547d62f29d4407fceca9bd766d8e9f93a88b5c6f
Signed-off-by: Puranam V G Tejaswi <pvgtejas@codeaurora.org>

User supplied count can cause signed integer overflow when passed to
dma_fence_array_create. So make sure that count is less than INT_MAX.

Change-Id: Ie6f57e06a8f23e9fdc10f8a368921b9ad3516eba
Signed-off-by: Puranam V G Tejaswi <pvgtejas@codeaurora.org>

When creating a timeline allocate an ID but don't populate the slot
with the timeline pointer until we are done initializing the timeline.
This avoids a race if somebody tries to free the same identifier before
the create operation is complete. Also, correct kgsl_timeline_by_id
function by doing idr_find and refcount increase under same lock. This
eliminates the chance that a timeline will be destroyed while somebody
is still using it.

Change-Id: Id894baac1bbd6042f668ea4afca39fae3b5a2d86
Signed-off-by: Deepak Kumar <dkumar@codeaurora.org>

Some graphics APIs want the ability to create and use timelines
with 64 bit sequence numbers to synchronize between threads.

Add support for timelines that can be created, signaled and managed from
user space. Timelines can also be signaled from a draw context command
stream via an auxiliary command and a wait can be added as a sync object
to any GPU command.

Change-Id: Ic0dedbad0ffc99cc309951369eaf85b56727802c
Signed-off-by: Jordan Crouse <jcrouse@codeaurora.org>