Age | Commit message (Collapse) | Author | Files | Lines |
|
When pinning a buffer, we should check to see if there are any
additional restrictions imposed by bo->preferred_domains. This will
prevent the BO from being moved to an invalid domain when pinning.
For example, this can happen if the user requests to create a BO in GTT
domain for display scanout. amdgpu_dm will allow pinning to either VRAM
or GTT domains, since DCN can scanout from either or. However, in
amdgpu_bo_pin_restricted(), pinning to VRAM is preferred if there is
adequate carveout. This can lead to pinning to VRAM despite the user
requesting GTT placement for the BO.
v2: Allow the kernel to override the domain, which can happen when
exporting a BO to a V4L camera (for example).
Signed-off-by: Leo Li <[email protected]>
Reviewed-by: Alex Deucher <[email protected]>
Reviewed-by: Christian König <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
Cc: [email protected]
|
|
Port aggregated doorbell support to gfx11.
Signed-off-by: Jack Xiao <[email protected]>
Reviewed-by: Hawking Zhang <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Port aggregated doorbell support to sdma6.
Signed-off-by: Jack Xiao <[email protected]>
Reviewed-by: Hawking Zhang <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Ring aggregated doorbel to make unmapped queue scheduled in mes firmware.
Signed-off-by: Le Ma <[email protected]>
Reviewed-by: Hawking Zhang <[email protected]>
Reviewed-by: Jack Xiao <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Allocate and enable aggregated doorbell.
Signed-off-by: Jack Xiao <[email protected]>
Reviewed-by: Hawking Zhang <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Allocate and enable aggregated doorbell.
Signed-off-by: Le Ma <[email protected]>
Reviewed-by: Hawking Zhang <[email protected]>
Reviewed-by: Jack Xiao <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Move reset_context out of gpu recover function to make it configurable
for different reset purpose.
For the reset way of call gpu_recovery sysfs, force to use full reset
method. Otherwise, try soft reset by default if the related ASIC
supportted, if soft reset failed, will use full reset.
Signed-off-by: Likun Gao <[email protected]>
Reviewed-by: Hawking Zhang <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Support SDMA soft reset for SDMA v6.
V3: use ib test to check soft reset.
V4: squash in unused variable fix (Alex)
Signed-off-by: Likun Gao <[email protected]>
Reviewed-by: Hawking Zhang <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Enable soft reset for gfx 11.
V2: enable both gfx v11.0.0 and gfx v11.0.2.
Signed-off-by: Likun Gao <[email protected]>
Reviewed-by: Hawking Zhang <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Support GFX soft reset for gfx v11.
V3: use ib test check soft reset.
V4: squash in unused variable fix (Alex)
Signed-off-by: Likun Gao <[email protected]>
Reviewed-by: Hawking Zhang <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Set corresponding ready flag for mes ring when enable or disable
mes ring.
Signed-off-by: Jack Xiao <[email protected]>
Reviewed-by: Hawking Zhang <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
That has been done in BO release notify.
Signed-off-by: xinhui pan <[email protected]>
Acked-by: Christian König <[email protected]>
Reviewed-by: Felix Kuehling <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
For GMC 10 parts which support scatter/gather display (display
from system memory), we should allocate a larger gart size
to better handler larger displays. This mirrors what we already
do for GMC 9 parts.
v2: fix typo (Alex)
Acked-by: Christian König <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Need reserve buffers before unmap mes ctx bo va.
v2: fix removal of dma_resv_excl_fence() (Alex)
v3: fix dma_resv_usage (Alex)
Signed-off-by: Jack Xiao <[email protected]>
Reviewed-by: Hawking Zhang <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
For some cases (accessing registers, unmap legacy queue), it needs
access mes in atomic context. Use spinlock to protect agaist mes
ring buffer race condition.
Signed-off-by: Jack Xiao <[email protected]>
Reviewed-by: Hawking Zhang <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Seems to break hibernation. Disable for now until we can root
cause it.
Fixes: 087451f372bf ("drm/amdgpu: use generic fb helpers instead of setting up AMD own's.")
Bug: https://bugzilla.kernel.org/show_bug.cgi?id=216119
Acked-by: Evan Quan <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Was dropped when we converted to the generic helpers.
Fixes: 087451f372bf ("drm/amdgpu: use generic fb helpers instead of setting up AMD own's.")
Acked-by: Evan Quan <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
To avoid code repetition, unify the function exit path when possible. No
functional changes.
Acked-by: Christian König <[email protected]>
Signed-off-by: André Almeida <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
if amdgpu_mes_ctx_alloc_meta_data() fails, we should call amdgpu_vm_fini()
to handle amdgpu_vm_init().
Add a new lable before amdgpu_vm_init() and goto this lable when
amdgpu_mes_ctx_alloc_meta_data() fails.
Signed-off-by: Jianglei Nie <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
MES fw updated to support unmapping legacy gfx/compute queue.
Signed-off-by: Jack Xiao <[email protected]>
Reviewed-by: Hawking Zhang <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
It should not init whole ras bad page framework on sriov guest side
due to it is handled on host side.
Signed-off-by: Stanley.Yang <[email protected]>
Reviewed-by: Hawking Zhang <[email protected]>
Reviewed-by: Tao Zhou <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
GFX is the only IP block that RAS TA needs to program
the hardware when receiving enable_feature command.
Changed from V1:
remove amdgpu_ras_need_send_ras_feature inline function,
use GFX RAS block check directly.
Signed-off-by: Stanley.Yang <[email protected]>
Reviewed-by: Hawking Zhang <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
We don't need to validate and map root PD specially here,
it would be validated and mapped by amdgpu_vm_validate_pt_bos
if it is evicted.
The special case is when turning a GFX VM to a compute VM,
if vm_update_mode changed, we should make sure root PD gets
mapped. So just map root PD after updating vm->update_funcs
in amdgpu_vm_make_compute whether the vm_update_mode changed
or not.
v3:
- Add some comments suggested by Christian.
v2:
- Don't rename vm_validate_pt_pd_bos and make it public.
Signed-off-by: Lang Yu <[email protected]>
Acked-by: Christian König <[email protected]>
Reviewed-by: Felix Kuehling <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Output user queue eviction and restore event. User queue eviction may be
triggered by svm or userptr MMU notifier, TTM eviction, device suspend
and CRIU checkpoint and restore.
User queue restore may be rescheduled if eviction happens again while
restore.
Signed-off-by: Philip Yang <[email protected]>
Reviewed-by: Felix Kuehling <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
This reverts commit 8748de873fedf4d55bdd99bbb738ee7ddf329792
since drv enabled mes to access registers.
Signed-off-by: Jack Xiao <[email protected]>
Reviewed-by: Hawking Zhang <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Enable mes to access registers.
v2: squash mes sched ring enablement flag
Signed-off-by: Jack Xiao <[email protected]>
Reviewed-by: Hawking Zhang <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Add mes register access routines:
1. read register
2. write register
3. wait register
4. write and wait register
Signed-off-by: Jack Xiao <[email protected]>
Reviewed-by: Hawking Zhang <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Add misc op commands in mes11.
Signed-off-by: Jack Xiao <[email protected]>
Reviewed-by: Hawking Zhang <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Add common interface for mes misc op, including accessing register
interface.
Signed-off-by: Jack Xiao <[email protected]>
Reviewed-by: Hawking Zhang <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Fixes this issue:
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c:5094: warning: expecting prototype for amdgpu_device_gpu_recover_imp(). Prototype was for amdgpu_device_gpu_recover() instead
Fixes: cf727044144d ("drm/amdgpu: Rename amdgpu_device_gpu_recover_imp back to amdgpu_device_gpu_recover")
Reviewed-by: Kent Russell <[email protected]>
Reported-by: Stephen Rothwell <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
[Why&How]
Some userspace expect a backwards compatible modifier on DCN32/321. For
hardware with num_pipes more than 16, we expose the most efficient
modifier first. As a fall back method, we need to expose slightly inefficient
modifier AMD_FMT_MOD_TILE_GFX9_64K_R_X after the best option.
Also set the number of packers to fixed value as required per hardware
documentation. This value is cached during hardware initialization and
can be read through the base driver.
Signed-off-by: Aurabindo Pillai <[email protected]>
Reviewed-by: Bas Nieuwenhuizen <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
[Why&How]
TA firmware is needed to enable HDCP.
Changes in v2:
Load separate firmware for PSP 13.0.0
Signed-off-by: Aurabindo Pillai <[email protected]>
Acked-by: Alex Deucher <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Change amdggpu to amdgpu and pedning to pending
Signed-off-by: Kent Russell <[email protected]>
Reviewed-by: Alex Deucher <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Align refcount behaviour for amdgpu_job embedded HW fence with
classic pointer style HW fences by increasing refcount each
time emit is called so amdgpu code doesn't need to make workarounds
using amdgpu_job.job_run_counter to keep the HW fence refcount balanced.
Also since in the previous patch we resumed setting s_fence->parent to NULL
in drm_sched_stop switch to directly checking if job->hw_fence is
signaled to short circuit reset if already signed.
Signed-off-by: Andrey Grodzovsky <[email protected]>
Tested-by: Yiqing Yao <[email protected]>
Acked-by: Christian König <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Problem:
After we start handling timed out jobs we assume there fences won't be
signaled but we cannot be sure and sometimes they fire late. We need
to prevent concurrent accesses to fence array from
amdgpu_fence_driver_clear_job_fences during GPU reset and amdgpu_fence_process
from a late EOP interrupt.
Fix:
Before accessing fence array in GPU disable EOP interrupt and flush
all pending interrupt handlers for amdgpu device's interrupt line.
v2: Switch from irq_get/put to full enable/disable_irq for amdgpu
Signed-off-by: Andrey Grodzovsky <[email protected]>
Acked-by: Christian König <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
This function should drop the fence refcount when it extracts the
fence from the fence array, just as it's done in amdgpu_fence_process.
Signed-off-by: Andrey Grodzovsky <[email protected]>
Reviewed-by: Christian König <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Signed-off-by: Leslie Shi <[email protected]>
Reviewed-by: Hawking Zhang <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
MES requires mc wptr address for usermode queues.
Export bo gart address for mc wptr address.
Signed-off-by: Jack Xiao <[email protected]>
Reviewed-by: Hawking Zhang <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Update GFX11 cs related settings.
Signed-off-by: Evan Quan <[email protected]>
Acked-by: Alex Deucher <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
There is a spelling mistake in a pr_debug message. Fix it.
Signed-off-by: Colin Ian King <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
This reverts commit 92020e81ddbeac351ea4a19bcf01743f32b9c800.
This causes stuttering and timeouts with DMCUB for some users
so revert it until we understand why and safely enable it
to save power.
Bug: https://gitlab.freedesktop.org/drm/amd/-/issues/1887
Acked-by: Harry Wentland <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
Cc: Nicholas Kazlauskas <[email protected]>
|
|
there is an unexpected word 'for' in the comments that need to be dropped
file - drivers/gpu/drm/amd/amdgpu/amdgpu_ih.c
line - 245
* position and also advance the position for for Vega10
changed to:
* position and also advance the position for Vega10
Signed-off-by: Jiang Jian <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Update MES API to support oversubscription without aggregated doorbell
for usermode queues.
v2: Change oversubscription_no_aggregated_en to is_kfd_process (align
with MES)
Signed-off-by: Graham Sider <[email protected]>
Acked-by: Felix Kuehling <[email protected]>
Reviewed-by: Jack Xiao <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Starting with GFX11, MES requires wptr BOs to be GTT allocated/mapped to
GART for usermode queues in order to support oversubscription. In the
case that work is submitted to an unmapped queue, MES must have a GART
wptr address to determine whether the queue should be mapped.
This change is accompanied with changes in MES and is applicable for
MES_API_VERSION >= 2.
v3:
- Use amdgpu_vm_bo_lookup_mapping for wptr_bo mapping lookup
- Move wptr_bo refcount increment to amdgpu_amdkfd_map_gtt_bo_to_gart
- Remove list_del_init from amdgpu_amdkfd_map_gtt_bo_to_gart
- Cleanup/fix create_queue wptr_bo error handling
v4:
- Add MES version shift/mask defines to amdgpu_mes.h
- Change version check from MES_VERSION to MES_API_VERSION
- Add check in kfd_ioctl_create_queue before wptr bo pin/GART map to
ensure bo is a single page.
Signed-off-by: Graham Sider <[email protected]>
Acked-by: Alex Deucher <[email protected]>
Acked-by: Christian König <[email protected]>
Reviewed-by: Philip Yang <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Store MES scheduler and MES KIQ version numbers in amdgpu_mes for GFX11.
Signed-off-by: Graham Sider <[email protected]>
Acked-by: Felix Kuehling <[email protected]>
Reviewed-by: Jack Xiao <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
amdgpu: [mmhub0] no-retry page fault (src_id:0 ring:40 vmid:8 pasid:32769, for process test_basic pid 3305 thread test_basic pid 3305)
amdgpu: in page starting at address 0x00007ff990003000 from IH client 0x12 (VMC)
amdgpu: VM_L2_PROTECTION_FAULT_STATUS:0x00840051
amdgpu: Faulty UTCL2 client ID: MP1 (0x0)
amdgpu: MORE_FAULTS: 0x1
amdgpu: WALKER_ERROR: 0x0
amdgpu: PERMISSION_FAULTS: 0x5
amdgpu: MAPPING_ERROR: 0x0
amdgpu: RW: 0x1
When memory is allocated by kfd, no one triggers the tlb flush for MMHUB0.
There is page fault from MMHUB0.
v2:fix indentation
v3:change subject and fix indentation
Signed-off-by: Ruili Ji <[email protected]>
Reviewed-by: Philip Yang <[email protected]>
Reviewed-by: Aaron Liu <[email protected]>
Acked-by: Alex Deucher <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Declare 'static', as the function is not intended to be used
outside of this translation unit.
Fixes: 4ed49c954e35 ("drm/amdgpu/vcn: add unified queue ib test")
Reported-by: kernel test robot <[email protected]>
Reviewed-by: James Zhu <[email protected]>
Signed-off-by: Ruijing Dong <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
there is an unexpected word "the" in the comments that need to be dropped
file: drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
line: 57
* the kernel tells the the ring what VMID to use for that command
changed to
* the kernel tells the ring what VMID to use for that command
Signed-off-by: Jiang Jian <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Use the correct adev variable for the drm_fb_helper in
amdgpu_device_gpu_recover(). Noticed by inspection.
Fixes: 087451f372bf ("drm/amdgpu: use generic fb helpers instead of setting up AMD own's.")
Reviewed-by: Guchun Chen <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
The DRM_AMDGPU Kconfig code contains:
select BACKLIGHT_CLASS_DEVICE
So the condition these ifdefs test for is always true, drop them.
Signed-off-by: Hans de Goede <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|