Victor Zhao [Tue, 27 Apr 2021 09:52:56 +0000 (17:52 +0800)]
drm/amdgpu: fix r initial values
Sriov gets suspend of IP block <dce_virtual> failed as return
value was not initialized.
v2: return 0 directly to align original code semantic before this
was broken out into a separate helper function instead of setting
initial values
Signed-off-by: Victor Zhao <Victor.Zhao@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Dennis Li [Tue, 27 Apr 2021 14:21:03 +0000 (22:21 +0800)]
drm/amdgpu: fix no full coverage issue for gprs initialization
The wave's number per simd in aldebaran is changed to 8, so it is
impossible to use old algorithm to initiate all sgprs with one
threadgroup. The new algorithm firstly use three threadgroups to
initiate most sgprs simultaneously and then use another threadgroup with
4 waves to cover other uninitiated sgprs.
v2:
Add more description about the new algorithm to clear sgprs and add some
comment for shader binaries
Signed-off-by: Dennis Li <Dennis.Li@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Philip Yang [Tue, 20 Apr 2021 19:13:59 +0000 (15:13 -0400)]
drm/amdkfd: enable subsequent retry fault
After draining the stale retry fault, or failed to validate the range
to recover, have to remove the fault address from fault filter ring, to
be able to handle subsequent retry interrupt on same address. Otherwise
the retry fault will not be processed to recover until timeout passed.
Signed-off-by: Philip Yang <Philip.Yang@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Philip Yang [Tue, 20 Apr 2021 14:05:44 +0000 (10:05 -0400)]
drm/amdgpu: address remove from fault filter
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: Philip Yang <Philip.Yang@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Philip Yang [Tue, 20 Apr 2021 01:19:57 +0000 (21:19 -0400)]
drm/amdkfd: handle stale retry fault
Retry fault interrupt maybe pending in IH ring after GPU page table
is updated to recover the vm fault, because each page of the range
generate retry fault interrupt. There is race if application unmap
range to remove and free the range first and then retry fault work
restore_pages handle the retry fault interrupt, because range can not be
found, this vm fault can not be recovered and report incorrect GPU vm
fault to application.
Before unmap to remove and free range, drain retry fault interrupt
from IH ring1 to ensure no retry fault comes after the range is removed.
Drain retry fault interrupt skip the range which is on deferred list
to remove, or the range is child range, which is split by unmap, does
not add to svms and have interval notifier.
Signed-off-by: Philip Yang <Philip.Yang@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Philip Yang [Sun, 18 Apr 2021 02:37:21 +0000 (22:37 -0400)]
drm/amdgpu: return IH ring drain finished if ring is empty
Sometimes IH do not setup ring wptr overflow flag after wptr exceed
rptr. As a workaround, if IH rptr equals to wptr, ring is empty,
return true to indicate IH ring checkpoint is processed, IH ring drain
is finished.
Signed-off-by: Philip Yang <Philip.Yang@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Philip Yang [Tue, 20 Apr 2021 01:51:27 +0000 (21:51 -0400)]
drm/amdkfd: retry validation to recover range
GPU vm retry fault recover range need retry validation if
1. range is split in parallel by unmap while recover
2. range migrate to system memory and range is updated in system
memory while recover
Signed-off-by: Philip Yang <Philip.Yang@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Jonathan Kim [Mon, 26 Apr 2021 19:38:18 +0000 (15:38 -0400)]
drm/amdkfd: fix spelling mistake in packet manager
The plural of 'process' should be 'processes'.
Signed-off-by: Jonathan Kim <jonathan.kim@amd.com> Reviewed-by: Felix Kuehling <felix.kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Christian König [Mon, 26 Apr 2021 08:06:13 +0000 (10:06 +0200)]
drm/amdgpu: restructure amdgpu_vram_mgr_new
Merge the two loops, loosen the restriction for big allocations.
This reduces the CPU overhead in the good case, but increases
it a bit under memory pressure.
Signed-off-by: Christian König <christian.koenig@amd.com> Acked-and-Tested-by: Nirmoy Das <nirmoy.das@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Philip Yang [Mon, 26 Apr 2021 18:25:37 +0000 (14:25 -0400)]
drm/amdkfd: fix double free device pgmap resource
Use devm_memunmap_pages instead of memunmap_pages to release pgmap
and remove pgmap from device action, to avoid double free pgmap when
unloading driver module.
Release device memory region if failed to create device memory pages
structure.
Signed-off-by: Philip Yang <Philip.Yang@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
There is a spelling mistake in a pr_debug message. Fix it.
Signed-off-by: Colin Ian King <colin.king@canonical.com> Reviewed-by: Nirmoy Das <nirmoy.das@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Aric Cyr <aric.cyr@amd.com> Acked-by: Wayne Lin <waynelin@amd.com> Tested-by: Daniel Wheeler <daniel.wheeler@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Anthony Koo [Sat, 17 Apr 2021 19:13:15 +0000 (15:13 -0400)]
drm/amd/display: [FW Promotion] Release 0.0.63
Signed-off-by: Anthony Koo <Anthony.Koo@amd.com> Acked-by: Wayne Lin <waynelin@amd.com> Tested-by: Daniel Wheeler <daniel.wheeler@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
drm/amd/display: Add SE_DCN3_REG_LIST for control SDP num
[Why] New platform. Need to add corresponding register control
Signed-off-by: Max.Tseng <Max.Tseng@amd.com> Reviewed-by: Anthony Koo <Anthony.Koo@amd.com> Acked-by: Wayne Lin <waynelin@amd.com> Tested-by: Daniel Wheeler <daniel.wheeler@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
drm/amd/display: avoid to authentication when DEVICE_COUNT=0
[why]
we don't support authentication with DEVICE_COUNT=0
[how]
check value DEVICE_COUNT before doing authentication
Signed-off-by: Yu-ting Shen <Yu-ting.Shen@amd.com> Reviewed-by: Wenjing Liu <Wenjing.Liu@amd.com> Acked-by: Wayne Lin <waynelin@amd.com> Tested-by: Daniel Wheeler <daniel.wheeler@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Wenjing Liu <wenjing.liu@amd.com> Reviewed-by: Eric Bernstein <Eric.Bernstein@amd.com> Acked-by: Wayne Lin <waynelin@amd.com> Tested-by: Daniel Wheeler <daniel.wheeler@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
drm/amd/display: fix wrong statement in mst hpd debugfs
[why]
Previous statement would always evaluate to true
making it meaningless
[how]
Just check if a connector is MST by checking if its port exists.
Reported-by: kernel test robot <lkp@intel.com> Signed-off-by: Mikita Lipski <mikita.lipski@amd.com> Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com> Acked-by: Wayne Lin <waynelin@amd.com> Tested-by: Daniel Wheeler <daniel.wheeler@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Robin Chen [Thu, 15 Apr 2021 02:08:10 +0000 (10:08 +0800)]
drm/amd/display: Clear MASTER_UPDATE_LOCK_DB_EN when disable doublebuffer lock
[Why]
It seems there's a typo to set MASTER_UPDATE_LOCK_DB_EN when disable
doublebuffer lock.
[How]
Clear MASTER_UPDATE_LOCK_DB_EN when disable doublebuffer lock
Signed-off-by: Robin Chen <robin.chen@amd.com> Reviewed-by: Joshua Aberback <Joshua.Aberback@amd.com> Acked-by: Wayne Lin <waynelin@amd.com> Tested-by: Daniel Wheeler <daniel.wheeler@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
drm/amd/display: Add new DP_SEC registers for programming SDP Line number
[Why]
Add register for programming in new platform
Signed-off-by: Max.Tseng <Max.Tseng@amd.com> Reviewed-by: Anthony Koo <Anthony.Koo@amd.com> Acked-by: Wayne Lin <waynelin@amd.com> Tested-by: Daniel Wheeler <daniel.wheeler@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Chris Park [Wed, 14 Apr 2021 02:25:23 +0000 (22:25 -0400)]
drm/amd/display: Fix BSOD with NULL check
[Why]
CLK mgr is null for server settings.
[How]
Guard the function with NULL check.
Signed-off-by: Chris Park <Chris.Park@amd.com> Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com> Acked-by: Wayne Lin <waynelin@amd.com> Tested-by: Daniel Wheeler <daniel.wheeler@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Jake Wang [Mon, 12 Apr 2021 21:46:49 +0000 (17:46 -0400)]
drm/amd/display: Added multi instance support for ABM
[WHY & HOW]
ABM assumes only 1 eDP is connected. Refactored existing
ABM interface to support multiple instances.
Signed-off-by: Jake Wang <haonan.wang2@amd.com> Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com> Acked-by: Wayne Lin <waynelin@amd.com> Tested-by: Daniel Wheeler <daniel.wheeler@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
drm/amd/display: ddc resource data need to be initialized
[WHY]
to initial ddc structure before we use them to avoid error checking.
[HOW]
add 0 as structure initialization value in DCN resource construct
Signed-off-by: Yu-ting Shen <Yu-ting.Shen@amd.com> Reviewed-by: Meenakshikumar Somasundaram <Meenakshikumar.Somasundaram@amd.com> Acked-by: Wayne Lin <waynelin@amd.com> Tested-by: Daniel Wheeler <daniel.wheeler@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Stylon Wang [Wed, 7 Apr 2021 10:14:55 +0000 (18:14 +0800)]
drm/amd/display: Expose internal display flag via debugfs
[Why & How]
Add a per-connector debugfs entry to expose internal display flag,
which is indication that the display is "internally connected"
and not hotpluggable.
Signed-off-by: Stylon Wang <stylon.wang@amd.com> Reviewed-by: Wayne Lin <Wayne.Lin@amd.com> Tested-by: Daniel Wheeler <daniel.wheeler@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Lewis Huang [Fri, 9 Apr 2021 10:39:43 +0000 (18:39 +0800)]
drm/amd/display: skip program clock when allow seamless boot
[Why]
Driver program dpp clock calculate by pipe split config but hw config is single pipe.
[How]
Skip programming clock when allow seamless boot.
After porgramming pipe config, seamless boot flag will be clear.
Signed-off-by: Lewis Huang <Lewis.Huang@amd.com> Reviewed-by: Eric Yang <eric.yang2@amd.com> Acked-by: Wayne Lin <waynelin@amd.com> Tested-by: Daniel Wheeler <daniel.wheeler@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Lewis Huang [Mon, 12 Apr 2021 23:02:28 +0000 (07:02 +0800)]
drm/amd/display: Revert wait vblank on update dpp clock
[Why]
This change only fix dpp clock switch to lower case.
New solution later can fix both case, which is "dc: skip
program clock when allow seamless boot"
[How]
This reverts commit "dc: wait vblank when stream enabled
and update dpp clock"
Signed-off-by: Lewis Huang <Lewis.Huang@amd.com> Reviewed-by: Eric Yang <eric.yang2@amd.com> Acked-by: Wayne Lin <waynelin@amd.com> Tested-by: Daniel Wheeler <daniel.wheeler@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
drm/amd/display: fix HDCP reset sequence on reinitialize
[why]
When setup is called after hdcp has already setup,
it would cause to disable HDCP flow won’t execute.
[how]
Don't clean up hdcp content to be 0.
Signed-off-by: Brandon Syu <Brandon.Syu@amd.com> Reviewed-by: Wenjing Liu <Wenjing.Liu@amd.com> Acked-by: Wayne Lin <waynelin@amd.com> Tested-by: Daniel Wheeler <daniel.wheeler@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Michael Strauss [Wed, 7 Apr 2021 15:21:44 +0000 (11:21 -0400)]
drm/amd/display: Add new case to get spread spectrum info
[WHY]
New minor revision needs to be handled
[HOW]
Add switch case and assert to catch missing switch cases in the future
Signed-off-by: Michael Strauss <michael.strauss@amd.com> Reviewed-by: Eric Yang <eric.yang2@amd.com> Acked-by: Wayne Lin <waynelin@amd.com> Tested-by: Daniel Wheeler <daniel.wheeler@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Fixed a kernel-doc error in the documentation of a function.
Signed-off-by: Fabio M. De Francesco <fmdefrancesco@gmail.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Darren Powell [Wed, 7 Apr 2021 22:40:34 +0000 (18:40 -0400)]
amdgpu/pm: set pp_dpm_dcefclk to readonly on NAVI10 and newer gpus
v2 : change condition to apply to all chips after NAVI10
Writing to dcefclk causes the gpu to become unresponsive, and requires a reboot.
Patch prevents user from successfully writing to file pp_dpm_dcefclk on parts
NAVI10 and newer, and gives better user feedback that this operation is not allowed.
Signed-off-by: Darren Powell <darren.powell@amd.com> Reviewed-by: Kenneth Feng <kenneth.feng@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Darren Powell [Wed, 7 Apr 2021 04:34:35 +0000 (00:34 -0400)]
amdgpu/pm: Prevent force of DCEFCLK on NAVI10 and SIENNA_CICHLID
Writing to dcefclk causes the gpu to become unresponsive, and requires a reboot.
Patch ignores a .force_clk_levels(SMU_DCEFCLK) call and issues an
info message.
Signed-off-by: Darren Powell <darren.powell@amd.com> Reviewed-by: Kenneth Feng <kenneth.feng@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Yingjie Wang [Fri, 9 Apr 2021 00:57:20 +0000 (17:57 -0700)]
drm/amd/dc: Fix a missing check bug in dm_dp_mst_detect()
In dm_dp_mst_detect(), We should check whether or not @connector
has been unregistered from userspace. If the connector is unregistered,
we should return disconnected status.
Fixes: 4562236b3bc0 ("drm/amd/dc: Add dc display driver (v2)") Signed-off-by: Yingjie Wang <wangyingjie55@126.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Harry Wentland [Thu, 22 Apr 2021 23:10:52 +0000 (19:10 -0400)]
drm/amd/display: Reject non-zero src_y and src_x for video planes
[Why]
This hasn't been well tested and leads to complete system hangs on DCN1
based systems, possibly others.
The system hang can be reproduced by gesturing the video on the YouTube
Android app on ChromeOS into full screen.
[How]
Reject atomic commits with non-zero drm_plane_state.src_x or src_y values.
v2:
- Add code comment describing the reason we're rejecting non-zero
src_x and src_y
- Drop gerrit Change-Id
- Add stable CC
- Based on amd-staging-drm-next
v3: removed trailing whitespace
Signed-off-by: Harry Wentland <harry.wentland@amd.com> Cc: stable@vger.kernel.org Cc: nicholas.kazlauskas@amd.com Cc: amd-gfx@lists.freedesktop.org Cc: alexander.deucher@amd.com Cc: Roman.Li@amd.com Cc: hersenxs.wu@amd.com Cc: danny.wang@amd.com Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com> Acked-by: Christian König <christian.koenig@amd.com> Reviewed-by: Hersen Wu <hersenxs.wu@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Christian König [Thu, 22 Apr 2021 11:11:39 +0000 (13:11 +0200)]
drm/amdgpu: fix concurrent VM flushes on Vega/Navi v2
Starting with Vega the hardware supports concurrent flushes
of VMID which can be used to implement per process VMID
allocation.
But concurrent flushes are mutual exclusive with back to
back VMID allocations, fix this to avoid a VMID used in
two ways at the same time.
v2: don't set ring to NULL
Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: James Zhu <James.Zhu@amd.com> Tested-by: James Zhu <James.Zhu@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Nirmoy Das [Thu, 22 Apr 2021 15:20:20 +0000 (17:20 +0200)]
drm/amdgpu: remove AMDGPU_GEM_CREATE_SHADOW flag
Remove unused AMDGPU_GEM_CREATE_SHADOW flag.
Userspace is never allowed to use this.
Signed-off-by: Nirmoy Das <nirmoy.das@amd.com> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Nirmoy Das [Wed, 21 Apr 2021 16:09:46 +0000 (18:09 +0200)]
drm/amdgpu: cleanup amdgpu_bo_create()
Remove shadow bo related code as vm code is creating shadow bo
using proper API. Without shadow bo code, amdgpu_bo_create() is basically
a wrapper around amdgpu_bo_do_create(). So rename amdgpu_bo_do_create()
to amdgpu_bo_create().
Signed-off-by: Nirmoy Das <nirmoy.das@amd.com> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Nirmoy Das [Thu, 22 Apr 2021 14:54:01 +0000 (16:54 +0200)]
drm/amdgpu: create shadow bo using amdgpu_bo_create_shadow()
Shadow BOs are only needed for vm code so call amdgpu_bo_create_shadow()
directly instead of depending on amdgpu_bo_create().
Signed-off-by: Nirmoy Das <nirmoy.das@amd.com> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Nirmoy Das [Thu, 22 Apr 2021 14:38:58 +0000 (16:38 +0200)]
drm/amdgpu: remove unused vm context flags
Remove unused AMDGPU_VM_CONTEXT_GFX and AMDGPU_VM_CONTEXT_COMPUTE
flags.
Signed-off-by: Nirmoy Das <nirmoy.das@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Nirmoy Das [Thu, 22 Apr 2021 10:53:16 +0000 (12:53 +0200)]
drm/amdgpu: cleanup amdgpu_vm_init()
Currently only way to create compute vm is through
amdgpu_vm_make_compute(). So vm_context isn't required
anymore for amdgpu_vm_init().
Signed-off-by: Nirmoy Das <nirmoy.das@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Nirmoy Das [Wed, 21 Apr 2021 15:27:46 +0000 (17:27 +0200)]
drm/amdgpu: expose amdgpu_bo_create_shadow()
Exposed amdgpu_bo_create_shadow() will be needed
for amdgpu_vm handling.
Signed-off-by: Nirmoy Das <nirmoy.das@amd.com> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Christian König [Fri, 12 Feb 2021 09:46:30 +0000 (10:46 +0100)]
drm/amdgpu: fix coding style and documentation in amdgpu_vram_mgr.c
No functional changes, just cleaning up some leftovers and improve
documentation.
Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Nirmoy Das <nirmoy.aiemd@gmail.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Christian König [Thu, 11 Feb 2021 15:32:04 +0000 (16:32 +0100)]
drm/amdgpu: fix coding style and documentation in amdgpu_gtt_mgr.c
Avoid the forward define, fix coding style, add documentation.
No functional change.
Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Nirmoy Das <nirmoy.aiemd@gmail.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Colin Ian King [Thu, 22 Apr 2021 12:44:52 +0000 (13:44 +0100)]
drm/amdkfd: remove redundant initialization to variable r
The variable r is being initialized with a value that is never read
and it is being updated later with a new value. The initialization is
redundant and can be removed.
Addresses-Coverity: ("Unused value") Signed-off-by: Colin Ian King <colin.king@canonical.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Colin Ian King [Thu, 22 Apr 2021 12:31:58 +0000 (13:31 +0100)]
drm/amdkfd: fix uint32 variable compared to less than zero
Currently the call to kfd_process_gpuidx_from_gpuid is returning an
int value and this is being assigned to a uint32_t variable gpuidx
and this is being checked for a negative error return which is always
going to be false. Fix this by making gpuidx an int32_t. This makes
gpuidx also type consistent with the use of gpuidx from the callers.
Addresses-Coverity: ("Unsigned compared against 0") Fixes: cda0f85bfa5e ("drm/amdkfd: refine migration policy with xnack on") Signed-off-by: Colin Ian King <colin.king@canonical.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Alex Sierra [Wed, 21 Apr 2021 18:54:09 +0000 (13:54 -0500)]
drm/amdkfd: set attribute access for default ranges
Attribute access value for default ranges is set, based on
process xnack on/off.
XNACK ON has GPU access attribute for unregistered ranges through page
fault. While XNACK OFF has no access attribute for unregistered ranges.
Signed-off-by: Alex Sierra <alex.sierra@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Alex Sierra [Mon, 12 Apr 2021 18:35:18 +0000 (13:35 -0500)]
drm/amdkfd: svm ranges creation for unregistered memory
SVM ranges are created for unregistered memory, triggered
by page faults. These ranges are migrated/mapped to
GPU VRAM memory.
Signed-off-by: Alex Sierra <alex.sierra@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Alex Sierra [Mon, 12 Apr 2021 18:34:57 +0000 (13:34 -0500)]
drm/amdgpu: extend xnack limit page fault timeout
Extending this timeout will prevent IH from storm interrupts coming
from SDMA while a page fault is active. Currently, on Aldebaran,
handling that many interrupts can take a lot of CPU time
(up to 4 seconds).
This eventually causes timeouts in other tasks.
Signed-off-by: Alex Sierra <alex.sierra@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
drm/amdgpu: disable gfx ras by default in aldebaran
aldebaran gfx ras is still under development
Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com> Reviewed-by: Dennis Li <Dennis.Li@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
In order to support multi-process debugging, HWS PM4 packet
MAP_PROCESS requires an extension of 5 DWORDS to support targeting of
per-vmid SPI debug control registers as well as watch points per process.
Signed-off-by: Jonathan Kim <jonathan.kim@amd.com> Reviewed-by: Felix Kuehling <felix.kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Dennis Li [Wed, 21 Apr 2021 04:50:22 +0000 (12:50 +0800)]
drm/amdgpu: refine gprs init shaders to check coverage
Add codes to check whether all SIMDs are covered, make sure that all
GPRs are initialized.
Signed-off-by: Dennis Li <Dennis.Li@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Lee Jones [Fri, 16 Apr 2021 14:37:20 +0000 (15:37 +0100)]
drm/amd/amdgpu/amdgpu_cs: Repair some function naming disparity
Fixes the following W=1 kernel build warning(s):
drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c:685: warning: expecting prototype for cs_parser_fini(). Prototype was for amdgpu_cs_parser_fini() instead
drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c:1502: warning: expecting prototype for amdgpu_cs_wait_all_fence(). Prototype was for amdgpu_cs_wait_all_fences() instead
drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c:1656: warning: expecting prototype for amdgpu_cs_find_bo_va(). Prototype was for amdgpu_cs_find_mapping() 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: Sumit Semwal <sumit.semwal@linaro.org> Cc: Jerome Glisse <glisse@freedesktop.org> Cc: amd-gfx@lists.freedesktop.org Cc: dri-devel@lists.freedesktop.org Cc: linux-media@vger.kernel.org Cc: linaro-mm-sig@lists.linaro.org Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Lee Jones <lee.jones@linaro.org> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Lee Jones [Fri, 16 Apr 2021 14:37:18 +0000 (15:37 +0100)]
drm/amd/amdgpu/amdgpu_ring: Provide description for 'sched_score'
Fixes the following W=1 kernel build warning(s):
drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c:169: warning: Function parameter or member 'sched_score' not described in 'amdgpu_ring_init'
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: Sumit Semwal <sumit.semwal@linaro.org> Cc: amd-gfx@lists.freedesktop.org Cc: dri-devel@lists.freedesktop.org Cc: linux-media@vger.kernel.org Cc: linaro-mm-sig@lists.linaro.org Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Lee Jones <lee.jones@linaro.org> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Lee Jones [Fri, 16 Apr 2021 14:37:17 +0000 (15:37 +0100)]
drm/amd/amdgpu/amdgpu_ttm: Fix incorrectly documented function 'amdgpu_ttm_copy_mem_to_mem()'
Fixes the following W=1 kernel build warning(s):
drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c:311: warning: expecting prototype for amdgpu_copy_ttm_mem_to_mem(). Prototype was for amdgpu_ttm_copy_mem_to_mem() 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: Sumit Semwal <sumit.semwal@linaro.org> Cc: Jerome Glisse <glisse@freedesktop.org> Cc: amd-gfx@lists.freedesktop.org Cc: dri-devel@lists.freedesktop.org Cc: linux-media@vger.kernel.org Cc: linaro-mm-sig@lists.linaro.org Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Lee Jones <lee.jones@linaro.org> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Lee Jones [Fri, 16 Apr 2021 14:37:16 +0000 (15:37 +0100)]
drm/amd/amdgpu/amdgpu_gart: Correct a couple of function names in the docs
Fixes the following W=1 kernel build warning(s):
drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c:73: warning: expecting prototype for amdgpu_dummy_page_init(). Prototype was for amdgpu_gart_dummy_page_init() instead
drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c:96: warning: expecting prototype for amdgpu_dummy_page_fini(). Prototype was for amdgpu_gart_dummy_page_fini() 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: Nirmoy Das <nirmoy.das@amd.com> Cc: amd-gfx@lists.freedesktop.org Cc: dri-devel@lists.freedesktop.org Reviewed-by: Nirmoy Das <nirmoy.das@amd.com> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Lee Jones <lee.jones@linaro.org> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Lee Jones [Fri, 16 Apr 2021 14:37:10 +0000 (15:37 +0100)]
drm/radeon/radeon_device: Provide function name in kernel-doc header
Fixes the following W=1 kernel build warning(s):
drivers/gpu/drm/radeon/radeon_device.c:1101: warning: This comment starts with '/**', but isn't a kernel-doc comment. Refer Documentation/doc-guide/kernel-doc.rst
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 Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Lee Jones <lee.jones@linaro.org> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c: In function ‘amdgpu_device_suspend’:
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c:3733:6: warning: variable ‘r’ set but not used [-Wunused-but-set-variable]
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: Sumit Semwal <sumit.semwal@linaro.org> Cc: amd-gfx@lists.freedesktop.org Cc: dri-devel@lists.freedesktop.org Cc: linux-media@vger.kernel.org Cc: linaro-mm-sig@lists.linaro.org Signed-off-by: Lee Jones <lee.jones@linaro.org> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Felix Kuehling [Mon, 29 Mar 2021 22:49:12 +0000 (18:49 -0400)]
drm/amdkfd: Add CONFIG_HSA_AMD_SVM
Control whether to build SVM support into amdgpu with a Kconfig option.
This makes it easier to disable it in production kernels if this new
feature causes problems in production environments.
Use "depends on" instead of "select" for DEVICE_PRIVATE, as is
recommended for visible options.
Reviewed-by: Philip Yang <Philip.Yang@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Philip Yang [Thu, 10 Dec 2020 02:51:50 +0000 (21:51 -0500)]
drm/amdkfd: Add SVM API support capability bits
SVMAPISupported property added to HSA_CAPABILITY, the value match
HSA_CAPABILITY defined in Thunk spec:
SVMAPISupported: it will not be supported on older kernels that don't
have HMM or on systems with GFXv8 or older GPUs without support for
48-bit virtual addresses.
CoherentHostAccess property added to HSA_MEMORYPROPERTY, the value match
HSA_MEMORYPROPERTY defined in Thunk spec:
CoherentHostAccess: whether or not device memory can be coherently
accessed by the host CPU.
Signed-off-by: Philip Yang <Philip.Yang@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Felix Kuehling [Thu, 25 Feb 2021 04:57:33 +0000 (23:57 -0500)]
drm/amdkfd: multiple gpu migrate vram to vram
If prefetch range to gpu with acutal location is another gpu, or GPU
retry fault restore pages to migrate the range with acutal location is
gpu, then migrate from one gpu to another gpu.
Use system memory as bridge because sdma engine may not able to access
another gpu vram, use sdma of source gpu to migrate to system memory,
then use sdma of destination gpu to migrate from system memory to gpu.
Print out gpuid or gpuidx in debug messages.
Signed-off-by: Philip Yang <Philip.Yang@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Felix Kuehling [Thu, 25 Feb 2021 04:55:27 +0000 (23:55 -0500)]
drm/amdkfd: add svm range validate timestamp
With xnack on, add validate timestamp in order to handle GPU vm fault
from multiple GPUs.
If GPU retry fault need migrate the range to the best restore location,
use range validate timestamp to record system timestamp after range is
restored to update GPU page table.
Because multiple pages of same range have multiple retry fault, define
AMDGPU_SVM_RANGE_RETRY_FAULT_PENDING to the long time period that
pending retry fault may still comes after page table update, to skip
duplicate retry fault of same range.
If difference between system timestamp and range last validate timestamp
is bigger than AMDGPU_SVM_RANGE_RETRY_FAULT_PENDING, that means the
retry fault is from another GPU, then continue to handle retry fault
recover.
Signed-off-by: Philip Yang <Philip.Yang@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Felix Kuehling [Thu, 25 Feb 2021 04:46:28 +0000 (23:46 -0500)]
drm/amdkfd: refine migration policy with xnack on
With xnack on, GPU vm fault handler decide the best restore location,
then migrate range to the best restore location and update GPU mapping
to recover the GPU vm fault.
Signed-off-by: Philip Yang <Philip.Yang@amd.com> Signed-off-by: Alex Sierra <alex.sierra@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Alex Sierra [Tue, 28 Jul 2020 21:03:05 +0000 (16:03 -0500)]
drm/amdgpu: add svm_bo eviction to enable_signal cb
Add to amdgpu_amdkfd_fence.enable_signal callback, support
for svm_bo fence eviction.
Signed-off-by: Alex Sierra <alex.sierra@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Alex Sierra [Tue, 28 Jul 2020 21:04:41 +0000 (16:04 -0500)]
drm/amdgpu: svm bo enable_signal call condition
[why]
To support svm bo eviction mechanism.
[how]
If the BO crated has AMDGPU_AMDKFD_CREATE_SVM_BO flag set,
enable_signal callback will be called inside amdgpu_evict_flags.
This also causes gutting of the BO by removing all placements,
so that TTM won't actually do an eviction. Instead it will discard
the memory held by the BO. This is needed for HMM migration to user
mode system memory pages.
Signed-off-by: Alex Sierra <alex.sierra@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Felix Kuehling [Thu, 25 Feb 2021 04:24:53 +0000 (23:24 -0500)]
drm/amdkfd: add svm_bo eviction mechanism support
svm_bo eviction mechanism is different from regular BOs.
Every SVM_BO created contains one eviction fence and one
worker item for eviction process.
SVM_BOs can be attached to one or more pranges.
For SVM_BO eviction mechanism, TTM will start to call
enable_signal callback for every SVM_BO until VRAM space
is available.
Here, all the ttm_evict calls are synchronous, this guarantees
that each eviction has completed and the fence has signaled before
it returns.
Signed-off-by: Alex Sierra <alex.sierra@amd.com> Signed-off-by: Philip Yang <Philip.Yang@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Alex Sierra [Tue, 28 Jul 2020 18:38:29 +0000 (13:38 -0500)]
drm/amdgpu: add param bit flag to create SVM BOs
Add CREATE_SVM_BO define bit for SVM BOs.
Another define flag was moved to concentrate these
KFD type flags in one include file.
Signed-off-by: Alex Sierra <alex.sierra@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Alex Sierra [Tue, 28 Jul 2020 18:35:32 +0000 (13:35 -0500)]
drm/amdkfd: add svm_bo reference for eviction fence
[why]
As part of the SVM functionality, the eviction mechanism used for
SVM_BOs is different. This mechanism uses one eviction fence per prange,
instead of one fence per kfd_process.
[how]
A svm_bo reference to amdgpu_amdkfd_fence to allow differentiate between
SVM_BO or regular BO evictions. This also include modifications to set the
reference at the fence creation call.
Signed-off-by: Alex Sierra <alex.sierra@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Alex Sierra [Tue, 12 May 2020 03:01:24 +0000 (22:01 -0500)]
drm/amdkfd: SVM API call to restore page tables
Use SVM API to restore page tables when retry fault and
compute context are enabled.
Signed-off-by: Alex Sierra <alex.sierra@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Felix Kuehling [Thu, 25 Feb 2021 04:02:21 +0000 (23:02 -0500)]
drm/amdkfd: page table restore through svm API
Page table restore implementation in SVM API. This is called from
the fault handler at amdgpu_vm. To update page tables through
the page fault retry IH.
Signed-off-by: Alex Sierra <alex.sierra@amd.com> Signed-off-by: Philip Yang <Philip.Yang@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Alex Sierra [Fri, 15 Jan 2021 23:03:18 +0000 (17:03 -0600)]
drm/amdgpu: enable 48-bit IH timestamp counter
By default this timestamp is 32 bit counter. It gets
overflowed in around 10 minutes.
Signed-off-by: Alex Sierra <alex.sierra@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Philip Yang <Philip.Yang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Felix Kuehling [Thu, 25 Feb 2021 03:34:11 +0000 (22:34 -0500)]
drm/amdkfd: invalidate tables on page retry fault
GPU page tables are invalidated by unmapping prange directly at
the mmu notifier, when page fault retry is enabled through
amdgpu_noretry global parameter. The restore page table is
performed at the page fault handler.
If xnack is on, we update GPU mappings after migration to avoid
unnecessary GPUVM faults.
Signed-off-by: Alex Sierra <alex.sierra@amd.com> Signed-off-by: Philip Yang <Philip.Yang@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Felix Kuehling [Wed, 17 Mar 2021 04:24:12 +0000 (00:24 -0400)]
drm/amdkfd: HMM migrate vram to ram
If CPU page fault happens, HMM pgmap_ops callback migrate_to_ram start
migrate memory from vram to ram in steps:
1. migrate_vma_pages get vram pages, and notify HMM to invalidate the
pages, HMM interval notifier callback evict process queues
2. Allocate system memory pages
3. Use svm copy memory to migrate data from vram to ram
4. migrate_vma_pages copy pages structure from vram pages to ram pages
5. Return VM_FAULT_SIGBUS if migration failed, to notify application
6. migrate_vma_finalize put vram pages, page_free callback free vram
pages and vram nodes
7. Restore work wait for migration is finished, then update GPU page
table mapping to system memory, and resume process queues
Signed-off-by: Philip Yang <Philip.Yang@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Felix Kuehling [Thu, 25 Feb 2021 01:40:20 +0000 (20:40 -0500)]
drm/amdkfd: HMM migrate ram to vram
Register svm range with same address and size but perferred_location
is changed from CPU to GPU or from GPU to CPU, trigger migration the svm
range from ram to vram or from vram to ram.
If svm range prefetch location is GPU with flags
KFD_IOCTL_SVM_FLAG_HOST_ACCESS, validate the svm range on ram first,
then migrate it from ram to vram.
After migrating to vram is done, CPU access will have cpu page fault,
page fault handler migrate it back to ram and resume cpu access.
Migration steps:
1. migrate_vma_pages get svm range ram pages, notify the
interval is invalidated and unmap from CPU page table, HMM interval
notifier callback evict process queues
2. Allocate new pages in vram using TTM
3. Use svm copy memory to sdma copy data from ram to vram
4. migrate_vma_pages copy ram pages structure to vram pages structure
5. migrate_vma_finalize put ram pages to free ram pages and memory
6. Restore work wait for migration is finished, then update GPUs page
table mapping to new vram pages, resume process queues
If migrate_vma_setup failed to collect all ram pages of range, retry 3
times until success to start migration.
Signed-off-by: Philip Yang <Philip.Yang@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Philip Yang [Fri, 7 Feb 2020 22:08:04 +0000 (17:08 -0500)]
drm/amdkfd: copy memory through gart table
Use sdma linear copy to migrate data between ram and vram. The sdma
linear copy command uses kernel buffer function queue to access system
memory through gart table.
Use reserved gart table window 0 to map system page address, and vram
page address is direct mapping. Use the same kernel buffer function to
fill in gart table mapping, so this is serialized with memory copy by
sdma job submit. We only need wait for the last memory copy sdma fence
for larger buffer migration.
Signed-off-by: Philip Yang <Philip.Yang@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Philip Yang [Tue, 23 Jun 2020 21:06:53 +0000 (17:06 -0400)]
drm/amdkfd: support xgmi same hive mapping
amdgpu_gmc_get_vm_pte use bo_va->is_xgmi same hive information to set
pte flags to update GPU mapping. Add local structure variable bo_va, and
update bo_va.is_xgmi, pass it to mapping->bo_va while mapping to GPU.
Assuming xgmi pstate is hi after boot.
Signed-off-by: Philip Yang <Philip.Yang@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Felix Kuehling [Thu, 25 Feb 2021 00:56:09 +0000 (19:56 -0500)]
drm/amdkfd: validate vram svm range from TTM
If svm range perfetch location is not zero, use TTM to alloc
amdgpu_bo vram nodes to validate svm range, then map vram nodes to GPUs.
Use offset to sub allocate from the same amdgpu_bo to handle overlap
vram range while adding new range or unmapping range.
svm_bo has ref count to trace the shared ranges. If all ranges of shared
amdgpu_bo are migrated to ram, ref count becomes 0, then amdgpu_bo is
released, all ranges svm_bo is set to NULL.
To migrate range from ram back to vram, allocate the same amdgpu_bo
with previous offset if the range has svm_bo.
If prange migrate to VRAM, no CPU mapping exist, then process exit will
not have unmap callback for this prange to free prange and svm bo. Free
outstanding pranges from svms list before process is freed in
svm_range_list_fini.
Signed-off-by: Philip Yang <Philip.Yang@amd.com> Signed-off-by: Alex Sierra <alex.sierra@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Philip Yang [Thu, 15 Apr 2021 21:43:32 +0000 (17:43 -0400)]
drm/amdkfd: set memory limit to avoid OOM with HMM enabled
HMM migration alloc sizeof(struct page) on system memory for each VRAM
page, it is 1GB system memory reserved for 64GB VRAM. To avoid
application OOM, increase system memory used size based on VRAM size of
all GPUs, then application alloc memory will fail if system memory usage
reach the limit.
Signed-off-by: Philip Yang <Philip.Yang@amd.com> Reviewed-by: Oak Zeng <Oak.Zeng@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Philip Yang [Fri, 7 Feb 2020 20:34:33 +0000 (15:34 -0500)]
drm/amdkfd: register HMM device private zone
Register vram memory as MEMORY_DEVICE_PRIVATE type resource, to
allocate vram backing pages for page migration.
Signed-off-by: Philip Yang <Philip.Yang@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Alex Sierra [Thu, 28 May 2020 23:27:05 +0000 (18:27 -0500)]
drm/amdkfd: add ioctl to configure and query xnack retries
Xnack retries are used for page fault recovery. Some AMD chip
families support continuously retry while page table entries are invalid.
The driver must handle the page fault interrupt and fill in a valid entry
for the GPU to continue.
This ioctl allows to enable/disable XNACK retries per KFD process.
Signed-off-by: Alex Sierra <alex.sierra@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Alex Sierra [Thu, 28 May 2020 23:03:15 +0000 (18:03 -0500)]
drm/amdkfd: add xnack enabled flag to kfd_process
XNACK mode controls the SQ RETRY_DISABLE setting that determines,
whether recoverable page faults can be supported on GFXv9 hardware.
Only on Aldebaran we can support different processes running with
different XNACK modes. On older chips all processes must use the same
RETRY_DISABLE setting. However, processes not relying on recoverable
page faults can work with RETRY enabled. This means XNACK off is always
available as a fallback so we can use the same mode on all GPUs in a
process.
Signed-off-by: Alex Sierra <alex.sierra@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Felix Kuehling [Thu, 11 Feb 2021 20:57:20 +0000 (15:57 -0500)]
drm/amdgpu: Enable retry faults unconditionally on Aldebaran
This is needed to allow per-process XNACK mode selection in the SQ when
booting with XNACK off by default.
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Reviewed-by: Philip Yang <Philip.Yang@amd.com> Tested-by: Alex Sierra <alex.sierra@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Felix Kuehling [Wed, 24 Feb 2021 23:47:52 +0000 (18:47 -0500)]
drm/amdkfd: svm range eviction and restore
HMM interval notifier callback notify CPU page table will be updated,
stop process queues if the updated address belongs to svm range
registered in process svms objects tree. Scheduled restore work to
update GPU page table using new pages address in the updated svm range.
The restore worker flushes any deferred work to make sure it restores
an up-to-date svm_range_list.
Signed-off-by: Philip Yang <Philip.Yang@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Felix Kuehling [Wed, 24 Feb 2021 23:29:06 +0000 (18:29 -0500)]
drm/amdkfd: map svm range to GPUs
Use amdgpu_vm_bo_update_mapping to update GPU page table to map or unmap
svm range system memory pages address to GPUs.
Signed-off-by: Philip Yang <Philip.Yang@amd.com> Signed-off-by: Alex Sierra <alex.sierra@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Philip Yang [Mon, 23 Sep 2019 15:53:08 +0000 (11:53 -0400)]
drm/amdgpu: export vm update mapping interface
It will be used by kfd to map svm range to GPU, because svm range does
not have amdgpu_bo and bo_va, cannot use amdgpu_bo_update interface, use
amdgpu vm update interface directly.
Signed-off-by: Philip Yang <Philip.Yang@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
projects / mirror_ubuntu-kernels.git / log