We hit soft hang while doing memory pressure test on one numa system.
After a qucik look, this is because kfd invalid/valid userptr memory
frequently with process_info lock hold.
Looks like update page table mapping use too much cpu time.

perf top says below,
75.81%  [kernel]       [k] __srcu_read_unlock
 6.19%  [amdgpu]       [k] amdgpu_gmc_set_pte_pde
 3.56%  [kernel]       [k] __srcu_read_lock
 2.20%  [amdgpu]       [k] amdgpu_vm_cpu_update
 2.20%  [kernel]       [k] __sg_page_iter_dma_next
 2.15%  [drm]          [k] drm_dev_enter
 1.70%  [drm]          [k] drm_prime_sg_to_dma_addr_array
 1.18%  [kernel]       [k] __sg_alloc_table_from_pages
 1.09%  [drm]          [k] drm_dev_exit

So move drm_dev_enter/exit outside gmc code, instead let caller do it.
They are gart_unbind, gart_map, vm_clear_bo, vm_update_pdes and
gmc_init_pdb0. vm_bo_update_mapping already calls it.
Signed-off-by: Nxinhui pan <xinhui.pan@amd.com>
Reviewed-and-tested-by: NAndrey Grodzovsky <andrey.grodzovsky@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

As the user cannot do anything about the unsupported Trusted Memory Zone
(TMZ) feature, do not warn about it, but make it informational, so
demote the log level from warning to info.
Signed-off-by: NPaul Menzel <pmenzel@molgen.mpg.de>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

As the user cannot do anything about the unsupported Trusted Memory Zone
(TMZ) feature, do not warn about it, but make it informational, so
demote the log level from warning to info.
Signed-off-by: NPaul Menzel <pmenzel@molgen.mpg.de>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Add MCA specific IP blocks targetting RAS features
Reviewed-by: NHawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: NJohn Clements <john.clements@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

The tmz functions are verified on yellow carp. So enable it by
default.
Signed-off-by: NAaron Liu <aaron.liu@amd.com>
Reviewed-by: NHuang Rui <ray.huang@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

When IP discovery enabled, the reserved buffer has been alloacted.
Signed-off-by: NAaron Liu <aaron.liu@amd.com>
Reviewed-by: NHuang Rui <ray.huang@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Some ASICs such as Yellow Carp needs to reserve a region of video memory
to avoid access from driver. So this patch is to introduce a stolen
reserved buffer to protect specific buffer region.

v2: free this buffer in amdgpu_ttm_fini.
Signed-off-by: NHuang Rui <ray.huang@amd.com>
Acked-and-Tested-by: NAaron Liu <aaron.liu@amd.com>
Acked-by: NAlex Deucher <alexander.deucher@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

When we want to decouble resource management from buffer management we need to
be able to handle resources separately.

Add a resource pointer and rename bo->mem so that all code needs to
change to access the pointer instead.

No functional change.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NMatthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210430092508.60710-4-christian.koenig@amd.com

Fixes the following W=1 kernel build warning(s):

 drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c:487: warning: expecting prototype for amdgpu_tmz_set(). Prototype was for amdgpu_gmc_tmz_set() instead
 drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c:533: warning: expecting prototype for amdgpu_noretry_set(). Prototype was for amdgpu_gmc_noretry_set() instead

Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: "Christian König" <christian.koenig@amd.com>
Cc: David Airlie <airlied@linux.ie>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: amd-gfx@lists.freedesktop.org
Cc: dri-devel@lists.freedesktop.org
Signed-off-by: NLee Jones <lee.jones@linaro.org>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

In SRIOV environment, KMD should access GC registers
with RLCG if GC indirect access flag enabled.
Signed-off-by: NPeng Ju Zhou <PengJu.Zhou@amd.com>
Reviewed-by: NAlex Deucher <alexander.deucher@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

This should prevent writing to memory or IO ranges possibly
already allocated for other uses after our device is removed.

v5:
Protect more places wher memcopy_to/form_io takes place
Protect IB submissions

v6: Switch to !drm_dev_enter instead of scoping entire code
with brackets.

v7:
Drop guard of HW ring commands emission protection since they
are in GART and not in MMIO.
Signed-off-by: NAndrey Grodzovsky <andrey.grodzovsky@amd.com>
Reviewed-by: NAlex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210512142648.666476-10-andrey.grodzovsky@amd.com

invoke callback function for mmhub ras fini
Signed-off-by: NHawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: NJohn Clements <John.Clements@amd.com>
Reviewed-by: NDennis Li <Dennis.Li@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

invoke hdp v4_0 ras init in gmc late_init phase
while ras fini in gmc sw_fini phase
Signed-off-by: NHawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: NJohn Clements <John.Clements@amd.com>
Reviewed-by: NDennis Li <Dennis.Li@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Add interface to remove address from fault filter ring by resetting
fault ring entry key, then future vm fault on the address will be
processed to recover.

Define fault key as atomic64_t type to use atomic read/set/cmpxchg key
to protect fault ring access by interrupt handler and interrupt deferred
work for vg20. Change fault->timestamp to 48-bit to share same uint64_t
with 8-bit fault->next, it is enough for 48bit IH timestamp.
Signed-off-by: NPhilip Yang <Philip.Yang@amd.com>
Reviewed-by: NFelix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

The tmz functions are verified on renoir chips as well. So enable it by
default.
Signed-off-by: NHuang Rui <ray.huang@amd.com>
Tested-by: NLang Yu <Lang.Yu@amd.com>
Reviewed-by: NAlex Deucher <alexander.deucher@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Use amdgpu_gmc_vram_pa and amdgpu_gmc_vram_cpu_pa
to simplify codes. No logic change.
Signed-off-by: NOak Zeng <Oak.Zeng@amd.com>
Signed-off-by: NHarish Kasiviswanathan <Harish.Kasiviswanathan@amd.com>
Acked-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Add one function to calculate BO's GPU physical address.
And another function to calculate BO's CPU physical address.

v2: Use functions vs macros (Christian)
    Use more proper function names (Christian)
Signed-off-by: NOak Zeng <Oak.Zeng@amd.com>
Suggested-by: NLijo Lazar <lijo.lazar@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

mmhub ras is always owned by gpu driver. ras_funcs
initialization shall be done at ip level, instead of
putting it in common gmc interface file
Signed-off-by: NHawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: NLijo Lazar <lijo.lazar@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

mmhub ras is only avaiable in cerntain mmhub ip
generation.
Signed-off-by: NHawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: NDennis Li <Dennis.Li@amd.com>
Reviewed-by: NJohn Clements <John.Clements@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

umc ras is not managed by gpu driver when gpu is
connected to cpu through xgmi. split umc callbacks
into ras and non-ras ones so gpu driver only
initializes umc ras callbacks when it manages
umc ras.
Signed-off-by: NHawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: NDennis Li <Dennis.Li@amd.com>
Reviewed-by: NJohn Clements <John.Clements@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

xgmi ras is not managed by gpu driver when gpu is
connected to cpu through xgmi. move all xgmi ras
functions to xgmi_ras_funcs so gpu driver only
initializes xgmi ras functions when it manages
xgmi ras.
Signed-off-by: NHawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: NDennis Li <Dennis.Li@amd.com>
Reviewed-by: NJohn Clements <John.Clements@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Allow allocating BO structures with different structure size
than struct amdgpu_bo.

v2: Check bo_ptr_size in all amdgpu_bo_create() caller.
Signed-off-by: NNirmoy Das <nirmoy.das@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Setting amdgpu.noretry=1 as default for Arcturus.
Signed-off-by: NPhilip Cox <Philip.Cox@amd.com>
Reviewed-by: NKent Russell <kent.russell@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

This is needed for best machine learning performance. XNACK can still
be enabled per-process if needed.
Signed-off-by: NFelix Kuehling <Felix.Kuehling@amd.com>
Reviewed-by: NAlex Deucher <alexander.deucher@amd.com>
Reviewed-by: NPhilip Yang <Philip.Yang@amd.com>
Tested-by: NAlex Sierra <alex.sierra@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

For the new 2-level GART table, the last PDE0 points
to PTB. Since PTB is in vram and right now we are
runing under s=0 mode (vram is treated as FB carveout),
so the s bit of this PDE0 should be set to 0.
Signed-off-by: NOak Zeng <Oak.Zeng@amd.com>
Reviewed-by: NFelix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Add functions to allocate PDB0, map it for CPU access,
and fill it.

Those functions are only used for 2-level vmid0 page
table construction
Signed-off-by: NOak Zeng <Oak.Zeng@amd.com>
Reviewed-by: NFelix Kuehling <felix.kuehling@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

If use GART for FB translation, place both vram and gart to sysvm
aperture. AGP aperture is not set up in this case because it
is not used
Signed-off-by: NOak Zeng <Oak.Zeng@amd.com>
Reviewed-by: NChristian Konig <christian.koenig@amd.com>
Reviewed-by: NFelix Kuehling <felix.kuehling@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

In amdgpu_gmc_gart_location function, gart_size is adjusted
by a smu_prv_buffer_size. This logic shouldn't belong to
this function. Move the logic to the mc_init functions
Signed-off-by: NOak Zeng <Oak.Zeng@amd.com>
Reviewed-by: NChristian Konig <christian.koenig@amd.com>
Reviewed-by: NFelix Kuehling <felix.kuehling@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

This has been stable for a while.
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

There is no reserveration so set the size to 0.  Fixes
a regression on SR-IOV.
Reviewed-by: NGuchun Chen <guchun.chen@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

We need to move the check under the non-headless case, otherwise
we always reserve the VGA save size.

Fixes: 157fe68d ("drm/amdgpu: fix size calculation with stolen vga memory")
Reviewed-by: NGuchun Chen <guchun.chen@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Based on an idea from Dave, but cleaned up a bit.

We had multiple fields for essentially the same thing.

Now bo->base.size is the original size of the BO in
arbitrary units, usually bytes.

bo->mem.num_pages is the size in number of pages in the
resource domain of bo->mem.mem_type.

v2: use the GEM object size instead of the BO size
v3: fix printks in some places
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: Huang Rui <ray.huang@amd.com> (v1)
Acked-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/406831/

If we need to keep the stolen vga memory, make sure it is
at least as big as the legacy vga size.
Acked-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org

If we need to keep the stolen vga memory, make sure it is
at least as big as the legacy vga size.
Acked-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

There are no performance advantages to setting it to 1 and
it causes stability issues in some cases.

v2: simplify the code

Bug: https://gitlab.freedesktop.org/drm/amd/-/issues/1374Reviewed-by: NFelix Kuehling <Felix.Kuehling@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

noretry = 0 casue KFDGraphicsInterop test failed
on SRIOV platform for vega10, so set noretry to 1
for vega10.
Signed-off-by: NStanley.Yang <Stanley.Yang@amd.com>
Reviewed-by: NHawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Fixes the following W=1 kernel build warning(s):

 drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c:69: warning: Function parameter or member 'bo' not described in 'amdgpu_gmc_pd_addr'
 drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c:121: warning: Function parameter or member 'bo' not described in 'amdgpu_gmc_agp_addr'
 drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c:121: warning: Excess function parameter 'tbo' description in 'amdgpu_gmc_agp_addr'

Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: "Christian König" <christian.koenig@amd.com>
Cc: David Airlie <airlied@linux.ie>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: amd-gfx@lists.freedesktop.org
Cc: dri-devel@lists.freedesktop.org
Signed-off-by: NLee Jones <lee.jones@linaro.org>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Uses the same pathes as navi.
Reviewed-by: NAaron Liu <aaron.liu@amd.com>
Reviewed-by: NLuben Tuikov <luben.tuikov@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

It makes no difference to kmalloc if the structure
is 48 or 64 bytes in size.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NDave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/396950/

set noretry default value to 1 for
sienna_cichlid/navy_founder/dimgrey_cavefish.
Signed-off-by: NChengming Gui <Jack.Gui@amd.com>
Reviewed-by: NGuchun Chen <guchun.chen@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>