No functional change, but the base for multi level page tables.
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>

Decribes better what this is used for.
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>

We will add the fence to freed buffer objects in a later commit, to ensure
that the underlying memory can only be re-used after all references in
page tables have been cleared.
Signed-off-by: NNicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: NChunming Zhou <david1.zhou@amd.com>
Reviewed-by: NJunwei Zhang <Jerry.Zhang@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NLeo Liu <leo.liu@amd.com>
Reviewed-by: NAlex Deucher <alexander.deucher@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Till GFX8 we can only enable PRT support globally, but with the next hardware
generation we can do this on a per page basis.

Keep the interface consistent by adding PRT mappings and enable
support globally on current hardware when the first mapping is made.

v2: disable PRT support delayed and on all error paths
v3: PRT and other permissions are mutal exclusive,
    PRT mappings don't need a BO.
v4: update PRT mappings durign CS as well, make va_flags 64bit
Signed-off-by: NJunwei Zhang <Jerry.Zhang@amd.com>
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NNicolai Hähnle <nicolai.haehnle@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

If we don't reset the chunk info in the error path, the subsequent
fini path will double free.
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Like ttm_bo_validate(), ttm_bo_init() might need to move BO and
the number of bytes moved by TTM should be reported. This can help
the throttle buffer migration mechanism to make a better decision.

v2: fix computation
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NSamuel Pitoiset <samuel.pitoiset@gmail.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Difference families may have different numbers of rings. Use
the variable rather than a hardcoded number.
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Make sure the CSA is mapped.

v2: agd: rebase.
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>

Return success when the ring is properly initialized, otherwise return
failure.

Tonga SRIOV VF doesn't have UVD and VCE engines, the initialization of
these IPs is bypassed. The system crashes if application submit IB to
their rings which are not ready to use. It could be a common issue if
IP having ring buffer is disabled for some reason on specific ASIC, so
it should check the ring being ready to use.

Bug: amdgpu_test crashes system on Tonga VF.
Signed-off-by: NDing Pixel <Pixel.Ding@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Fixes a rare NULL pointer dereference in amdgpu_ttm_bind.

The issue was found by Nicolai Haehnle.
The patch was tested by Nicolai Haehnle.
Signed-off-by: NAlex Xie <AlexBin.Xie@amd.com>
Reviewed-by: NChunming Zhou <david1.zhou@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org

In fence waiting, it never return -EDEADLK yet, so drop this function
here.
Signed-off-by: NHuang Rui <ray.huang@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Signed-off-by: NHuang Rui <ray.huang@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

v2: agd: rebase and squash in all the previous optimizations and
changes so everything compiles.
v3: squash in Slava's 32bit build fix
v4: rebase on drm-next (fence -> dma_fence),
    squash in Monk's ioctl update patch
Signed-off-by: NJunwei Zhang <Jerry.Zhang@amd.com>
Reviewed-by: NMonk Liu <monk.liu@amd.com>
Reviewed-by: NJammy Zhou <Jammy.Zhou@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>
Signed-off-by: NSumit Semwal <sumit.semwal@linaro.org>
 [sumits: fix checkpatch warnings]
Link: http://patchwork.freedesktop.org/patch/msgid/1478290570-30982-2-git-send-email-alexander.deucher@amd.com

This way we can use parse_cs and still keep VM mode enabled.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-and-Tested by: Leo Liu <leo.liu@amd.com>
Reviewed-by: NAlex Deucher <alexander.deucher@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

It's constant, so it doesn't make to much sense to keep it
with the variable data.

v2: update vce and uvd phys mode ring structures as well
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>

Saves a bunch of CPU cycles when swapping things back in and
allows us to split the VM headers into a separate file.

v2: rename parameters
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>

It's completely pointless to have two pointers to the
device in the same structure.

v2: rename function to amdgpu_ttm_adev, fix typos
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>

Add a flag noting that a BO must be created using linear VRAM
and set this flag on all in kernel users where appropriate.

Hopefully I haven't missed anything.

v2: add it in a few more places, fix CPU mapping.
v3: rename to VRAM_CONTIGUOUS, fix typo in CS code.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NEdward O'Callaghan <funfunctor@folklore1984.net>
Tested-by: NMike Lothian <mike@fireburn.co.uk>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

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

Signed-off-by: NJimQu <Jim.Qu@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Only allocate address space when we really need it.

v2: fix a typo, add correct function description,
    stop leaking the node in the error case.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NAlex Deucher <alexander.deucher@amd.com>
Reviewed-by: NFelix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

We need to validate the offset to make sure that we don't write after the BO.

Additional to that a page should be enough and can make address space
handling much easier.
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>

We get a few warnings when building kernel with W=1:
drivers/gpu/drm/amd/amdgpu/cz_smc.c:51:5: warning: no previous prototype for 'cz_send_msg_to_smc_async' [-Wmissing-prototypes]
drivers/gpu/drm/amd/amdgpu/cz_smc.c:143:5: warning: no previous prototype for 'cz_write_smc_sram_dword' [-Wmissing-prototypes]
drivers/gpu/drm/amd/amdgpu/iceland_smc.c:124:6: warning: no previous prototype for 'iceland_start_smc' [-Wmissing-prototypes]
drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c:3926:6: warning: no previous prototype for 'gfx_v8_0_rlc_stop' [-Wmissing-prototypes]
drivers/gpu/drm/amd/amdgpu/amdgpu_job.c:94:6: warning: no previous prototype for 'amdgpu_job_free_cb' [-Wmissing-prototypes]
....

In fact, these functions are only used in the file in which they are
declared and don't need a declaration, but can be made static.
So this patch marks these functions with 'static'.
Reviewed-by: NChristian König <christian.koenig@amd.com>
Acked-by: NHuang Rui <ray.huang@amd.com>
Reviewed-by: NEdward O'Callaghan <funfunctor@folklore1984.net>
Signed-off-by: NBaoyou Xie <baoyou.xie@linaro.org>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

We don't really need the GTT table any more most of the time. So bind it
only on demand.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NAlex Deucher <alexander.deucher@amd.com>
Acked-by: NFelix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

v1:
for gfx8, use CONTEXT_CONTROL package to dynamically
skip preamble CEIB and other load_xxx command in sequence.

v2:
support GFX7 as well.
remove cntxcntl in compute ring funcs because CPC doesn't
support this packet.

v3: fix reduntant judgement in cntxcntl.
v4: some cleanups, don't change cs_submit()
v5: keep old MESA supported & bump up KMS version.
Signed-off-by: NMonk Liu <Monk.Liu@amd.com>
Ack-by: NChunming Zhou <David1.Zhou@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

job->ctx actually is a fence_context of the entity
it belongs to, naming it as ctx is too vague, and
we'll need add amdgpu_ctx into the job structure
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>

As last resort try to evict BOs from the current working set into other
memory domains. This effectively prevents command submission failures when
VM page tables have been swapped out.

v2: fix typos
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>

All other errors can't be fixed by using a different memory domain.
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>

The old mechanism used a per-submission limit that didn't take previous
submissions within the same time frame into account. It also filled VRAM
slowly when VRAM usage dropped due to a big eviction or buffer deallocation.

This new method establishes a configurable MBps limit that is obeyed when
VRAM usage is very high. When VRAM usage is not very high, it gives
the driver the freedom to fill it quickly. The result is more consistent
performance.

It can't keep the BO move rate low if lots of evictions are happening due
to VRAM fragmentation, or if a big buffer is being migrated.

The amdgpu.moverate parameter can be used to set a non-default limit.
Measurements can be done to find out which amdgpu.moverate setting gives
the best results.

Mainly APUs and cards with small VRAM will benefit from this. For F1 2015,
anything with 2 GB VRAM or less will benefit.

Some benchmark results - F1 2015 (Tonga 2GB):

Limit      MinFPS AvgFPS
Old code:  14     32.6
128 MB/s:  28     41
64 MB/s:   15.5   43
32 MB/s:   28.7   43.4
8 MB/s:    27.8   44.4
8 MB/s:    21.9   42.8 (different run)

Random drops in Min FPS can still occur (due to fragmented VRAM?), but
the average FPS is much better. 8 MB/s is probably a good limit for this
game & the current VRAM management. The random FPS drops are still to be
tackled.

v2: use a spinlock
Signed-off-by: NMarek Olšák <marek.olsak@amd.com>
Acked-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Make it more obvious what we are doing here.
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>

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>

It's useful for debugging.
Signed-off-by: NMarek Olšák <marek.olsak@amd.com>
Reviewed-by: NAlex Deucher <alexander.deucher@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>

Keep the time we don't have a fence associated with the resource smaller.
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>

Same problem as with the VM page tables. The user fence address must be
determined before the job is scheduled, not when the IB is executed.

This fixes a security problem where user fences could be used to overwrite
any part of VRAM.
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>

It's just overhead to do so and allocating a VMID
when we don't need one is actually a bit dangerous.
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>

We don't need to validate them again if the eviction counter didn't changed.
Reviewed-by: NAlex Deucher <alexander.deucher@amd.com>
Signed-off-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

When we pipeline evictions the page directory could already be
moving somewhere else when grab_id is called.
Reviewed-by: NAlex Deucher <alexander.deucher@amd.com>
Signed-off-by: NChristian König <christian.koenig@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>