Bug: amdgpu_job_free_cb was accessing s_job->s_entity when the allocated
amdgpu_ctx (and the entity inside it) were already deallocated from
amdgpu_cs_parser_fini.

Fix: Save job's priority on it's creation instead of accessing it from
s_entity later on.
Signed-off-by: NAndrey Grodzovsky <Andrey.Grodzovsky@amd.com>
Reviewed-by: NAndres Rodriguez <andresx7@gmail.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Use _INVALID to identify bad parameters and _UNSET to represent the
lack of interest in a specific value.
Signed-off-by: NAndres Rodriguez <andresx7@gmail.com>
Acked-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

This is useful for changing an entity's priority at runtime.

v2: don't modify the order of amd_sched_entity members
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAndres Rodriguez <andresx7@gmail.com>
Acked-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Returning invalid priorities as _NORMAL is a backwards compatibility
quirk of amdgpu_ctx_ioctl(). Move this detail one layer up where it
belongs.
Signed-off-by: NAndres Rodriguez <andresx7@gmail.com>
Acked-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Add an initial framework for changing the HW priorities of rings. The
framework allows requesting priority changes for the lifetime of an
amdgpu_job. After the job completes the priority will decay to the next
lowest priority for which a request is still valid.

A new ring function set_priority() can now be populated to take care of
the HW specific programming sequence for priority changes.

v2: set priority before emitting IB, and take a ref on amdgpu_job
v3: use AMD_SCHED_PRIORITY_* instead of AMDGPU_CTX_PRIORITY_*
v4: plug amdgpu_ring_restore_priority_cb into amdgpu_job_free_cb
v5: use atomic for tracking job priorities instead of last_job
v6: rename amdgpu_ring_priority_[get/put]() and align parameters
v7: replace spinlocks with mutexes for KIQ compatibility
v8: raise ring priority during cs_ioctl, instead of job_run
v9: priority_get() before push_job()
Reviewed-by: NChristian König <christian.koenig@amd.com>
Acked-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAndres Rodriguez <andresx7@gmail.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Add a new context creation parameter to express a global context priority.

The priority ranking in descending order is as follows:
 * AMDGPU_CTX_PRIORITY_HIGH_HW
 * AMDGPU_CTX_PRIORITY_HIGH_SW
 * AMDGPU_CTX_PRIORITY_NORMAL
 * AMDGPU_CTX_PRIORITY_LOW_SW
 * AMDGPU_CTX_PRIORITY_LOW_HW

The driver will attempt to schedule work to the hardware according to
the priorities. No latency or throughput guarantees are provided by
this patch.

This interface intends to service the EGL_IMG_context_priority
extension, and vulkan equivalents.

Setting a priority above NORMAL requires CAP_SYS_NICE or DRM_MASTER.

v2: Instead of using flags, repurpose __pad
v3: Swap enum values of _NORMAL _HIGH for backwards compatibility
v4: Validate usermode priority and store it
v5: Move priority validation into amdgpu_ctx_ioctl(), headline reword
v6: add UAPI note regarding priorities requiring CAP_SYS_ADMIN
v7: remove ctx->priority
v8: added AMDGPU_CTX_PRIORITY_LOW, s/CAP_SYS_ADMIN/CAP_SYS_NICE
v9: change the priority parameter to __s32
v10: split priorities into _SW and _HW
v11: Allow DRM_MASTER without CAP_SYS_NICE
Reviewed-by: NEmil Velikov <emil.l.velikov@gmail.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAndres Rodriguez <andresx7@gmail.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

1,TDR will kickout guilty job if it hang exceed the threshold
of the given one from kernel paramter "job_hang_limit", that
way a bad command stream will not infinitly cause GPU hang.

by default this threshold is 1 so a job will be kicked out
after it hang.

2,if a job timeout TDR routine will not reset all sched/ring,
instead if will only reset on the givn one which is indicated
by @job of amdgpu_sriov_gpu_reset, that way we don't need to
reset and recover each sched/ring if we already know which job
cause GPU hang.

3,unblock sriov_gpu_reset for AI family.

V2:
1:put kickout guilty job after sched parked.
2:since parking scheduler prior to kickout already occupies a
while, we can do last check on the in question job before
doing hw_reset.

TODO:
1:when a job is considered as guilty, we should mark some flag
in its fence status flag, and let UMD side aware that this
fence signaling is not due to job complete but job hang.

2:if gpu reset cause all video memory lost, we need introduce
a new policy to implement TDR, like drop all jobs not yet
signaled, and all IOCTL on this device will return ERROR
DEVICE_LOST.
this will be implemented later.
Signed-off-by: NMonk Liu <Monk.Liu@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

The problem is that executing the jobs in the right order doesn't give you the right result
because consecutive jobs executed on the same engine are pipelined.
In other words job B does it buffer read before job A has written it's result.
Signed-off-by: NChunming Zhou <David1.Zhou@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

big number is to high priority.
Signed-off-by: NChunming Zhou <David1.Zhou@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

A unique id is useful for debugging and tracing. Intended to replace
pointers in ftrace output.
Reviewed-by: NChunming Zhou <david1.zhou@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAndres Rodriguez <andresx7@gmail.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Some fences might be alive even after we have stopped the scheduler leading
to warnings about leaked objects from the SLUB allocator.

Fix this by allocating/freeing the SLUB allocator from the module
init/fini functions just like we do it for hw fences.

v2: make variable static, add link to bug

Fixes: https://bugs.freedesktop.org/show_bug.cgi?id=97500Reported-by: NGrazvydas Ignotas <notasas@gmail.com>
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com> (v1)
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org

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: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: NGustavo Padovan <gustavo.padovan@collabora.co.uk>
Acked-by: NSumit Semwal <sumit.semwal@linaro.org>
Acked-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/20161025120045.28839-1-chris@chris-wilson.co.uk

Which is to recover hw jobs when gpu reset.
Signed-off-by: NChunming Zhou <David1.Zhou@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

amd_sched_hw_job_reset will remove callback from hw fence.
Signed-off-by: NChunming Zhou <David1.Zhou@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Parent of sched fence is hw fence which is to signal sched fence.
Signed-off-by: NChunming Zhou <David1.Zhou@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

We return the fence as part of the job structur anyway,
no need to do this twice.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NChunming Zhou <david1.zhou@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Make it two events, one for the job being scheduled and one when it is finished.
Acked-by: NAlex Deucher <alexander.deucher@amd.com>
Signed-off-by: NChristian König <christian.koenig@amd.com>
Acked-by: NChunming Zhou <david1.zhou@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Remove the job reference counting and just properly destroy it from a
work item which blocks on any potential running timeout handler.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NMonk.Liu <monk.liu@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

The driver shouldn't mess with the scheduler internals.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NMonk.Liu <monk.liu@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Remembering the code path in a variable to cleanup
differently is usually not a good idea at all.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NMonk.Liu <monk.liu@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

No need for double housekeeping here.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NMonk.Liu <monk.liu@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Completely pointless and confusing to use a callback
to call into the same code file.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NMonk.Liu <monk.liu@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

v2: fix even more
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NMonk.Liu <monk.liu@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

This is the result of running the following commands:
find drivers/gpu/drm/amd/ -name "*.h" -exec sed -i 's/[ \t]\+$//' {} \;
find drivers/gpu/drm/amd/ -name "*.c" -exec sed -i 's/[ \t]\+$//' {} \;
find drivers/gpu/drm/amd/ -name "*.h" -exec sed -i 's/ \+\t/\t/' {} \;
find drivers/gpu/drm/amd/ -name "*.c" -exec sed -i 's/ \+\t/\t/' {} \;

v2: drop changes to DAL and internal headers
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NAlex Deucher <alexander.deucher@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

This marks the struct amdgpu_sched_ops const and
adjusts amd_sched_init to take a const pointer
for the ops param. The ops member of
struct amd_gpu_scheduler is also changed to const.
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NNils Wallménius <nils.wallmenius@gmail.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

this is to fix fatal page fault error that occured if:
job is signaled/released after its timeout work is already
put to the global queue (in this case the cancel_delayed_work
will return false), which will lead to NX-protection error
page fault during job_timeout_func.
Signed-off-by: NMonk Liu <Monk.Liu@amd.com>
Reviewed-by: NChunming Zhou <david1.zhou@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Add two callbacks to scheduler to maintain jobs, and invoked for
job timeout calculations. Now TDR measures time gap from
job is processed by hw.

v2:
fix typo
Signed-off-by: NMonk Liu <Monk.Liu@amd.com>
Reviewed-by: NChunming Zhou <david1.zhou@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

original time out detect routine is incorrect, cuz it measures
the gap from job scheduled, but we should only measure the
gap from processed by hw.
Signed-off-by: NMonk Liu <Monk.Liu@amd.com>
Reviewed-by: NChunming Zhou <david1.zhou@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

the mirror_list will be used for later time out detect
feature.  This is needed to properly detect a GPU
timeout with the scheduler.
Signed-off-by: NMonk Liu <Monk.Liu@amd.com>
Reviewed-by: NChunming Zhou <david1.zhou@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

for those jobs submitted through scheduler, do not
free it immediately after scheduled, instead free it
in global workqueue by its sched fence signaling
callback function.

v2:
call uf's bo_undef after job_run()
call job's sync free after job_run()
no static inline __amdgpu_job_free() anymore, just use
kfree(job) to replace it.
Signed-off-by: NMonk Liu <Monk.Liu@amd.com>
Reviewed-by: NChunming Zhou <david1.zhou@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Consolidate job initialization in one place rather than
duplicating it in multiple places.
Signed-off-by: NMonk Liu <Monk.Liu@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Allows us to set priorities in the scheduler.
Signed-off-by: NChunming Zhou <David1.Zhou@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NJunwei Zhang <Jerry.Zhang@amd.com>

We only need to wait for jobs to be scheduled when
the dependency is from the same scheduler.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NChunming 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: NChristian König <christian.koenig@amd.com>
Reviewed-by: NAlex Deucher <alexander.deucher@amd.com>

Change-Id: I45bb8ff10ef05dc3b15e31a77fbcf31117705f11
Signed-off-by: NChunming Zhou <David1.Zhou@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>

Change-Id: I67e987db0efdca28faa80b332b75571192130d33
Signed-off-by: NJunwei Zhang <Jerry.Zhang@amd.com>
Reviewed-by: NDavid Zhou <david1.zhou@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>

Embed the scheduler into the ring structure instead of allocating it.
Use the ring name directly instead of the id.

v2: rebased, whitespace cleanup
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NJunwei Zhang <Jerry.Zhang@amd.com>
Reviewed-by: Chunming Zhou<david1.zhou@amd.com>

Just to be consistent with the other members.

v2: rename the ring member as well.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: Junwei Zhang <Jerry.Zhang@amd.com> (v1)
Reviewed-by: Chunming Zhou<david1.zhou@amd.com>

Reorder the fields and properly return the kfifo_alloc error code.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NJunwei Zhang <Jerry.Zhang@amd.com>
Reviewed-by: Chunming Zhou<david1.zhou@amd.com>

Use consistent naming across functions.
Reviewed-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NDavid Zhou <david1.zhou@amd.com>
Signed-off-by: NJunwei Zhang <Jerry.Zhang@amd.com>