We need to clear parser.ibs and num_ibs before amd_sched_fence_create,
otherwise the IB could be freed twice if fence creates fails.
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Chunming Zhou <david1.zhou@amd.com>
Before this patch the scheduler fence was created when we push the job
into the queue, so we could only get the fence after pushing it.
The mutex now was necessary to prevent the thread pushing the jobs to
the hardware from running faster than the thread pushing the jobs into
the queue.
Otherwise the thread pushing jobs into the queue would have accessed
possible freed up memory when it tries to get a reference to the fence.
So what you get in the end is thread A:
mutex_lock(&job->lock);
...
Kick of thread B.
...
mutex_unlock(&job->lock);
And thread B:
mutex_lock(&job->lock);
....
mutex_unlock(&job->lock);
kfree(job);
I'm actually not sure if I'm still up to date on this, but this usage
pattern used to be not allowed with mutexes. See here as well
https://lwn.net/Articles/575460/.
v2: remove unrelated changes, fix missing owner
v3: rebased, add more commit message
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
The code was correct, but getting two references when the ownership
is linearly moved on is a bit awkward and just overhead.
Signed: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
The amdgpu driver has a debugfs interface that shows the amount of
VRAM in use, but the newly added code causes a build error on
all 32-bit architectures:
drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c:1076:17: warning: format '%lu' expects argument of type 'long unsigned int', but argument 4 has type 'long long int' [-Wformat=]
This fixes the format string to use "%llu" for printing 64-bit
numbers, which works everywhere, as long as we also cast to 'u64'.
Unlike atomic64_t, u64 is defined as 'unsigned long long' on
all architectures.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Fixes: a2ef8a9749 ("drm/amdgpu: add vram usage into debugfs")
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The VM default page (used when a VM translation fails) is allocated in
system memory. The VM is misconfigured to interpret the physical address
as referencing a VRAM physical page.
Route default page accesses to system memory.
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Jay Cornwall <jay@jcornwall.me>
Cc: <stable@vger.kernel.org> # v4.2+
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
No need any more to allocate that structure dynamically, just put it on the
stack. This is a start to cleanup some of the scheduler fallouts.
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Junwei Zhang <Jerry.Zhang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Chunming Zhou
Christian König
Junwei Zhang
Flora Cui
Rex Zhu
Arnd Bergmann
Jay Cornwall