drm/amdgpu: Validate TA binary size

Add TA binary size validation to avoid OOB write.

Signed-off-by: Candice Li <candice.li@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Update BadOpcode Interrupt handling with MES

Based on the recommendation of MEC FW, update BadOpcode interrupt
handling by unmapping all queues, removing the queue that got the
interrupt from scheduling and remapping rest of the queues back when
using MES scheduler. This is done to prevent the case where unmapping
of the bad queue can fail thereby causing a GPU reset.

Signed-off-by: Mukul Joshi <mukul.joshi@amd.com>
Acked-by: Harish Kasiviswanathan <Harish.Kasiviswanathan@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Felix Kuehling <felix.kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Update queue unmap after VM fault with MES

MEC FW expects MES to unmap all queues when a VM fault is observed
on a queue and then resumed once the affected process is terminated.
Use the MES Suspend and Resume APIs to achieve this.

Signed-off-by: Mukul Joshi <mukul.joshi@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Harish Kasiviswanathan <Harish.Kasiviswanathan@amd.com>
Reviewed-by: Felix Kuehling <felix.kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Implement MES Suspend and Resume APIs for GFX11

Add implementation for MES Suspend and Resume APIs to unmap/map
all queues for GFX11. Support for GFX12 will be added when the
corresponding firmware support is in place.

Signed-off-by: Mukul Joshi <mukul.joshi@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Harish Kasiviswanathan <Harish.Kasiviswanathan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Enable processes isolation on gfx9

When amdgpu enable enforce_isolation, KFD enables single-process mode in
HWS and sets exec_cleaner_shader bit in MAP_PROCESS.

Signed-off-by: Amber Lin <Amber.Lin@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx_v9_4_3: Apply Isolation Enforcement to GFX & Compute rings

This commit applies isolation enforcement to the GFX and Compute rings
in the gfx_v9_4_3 module.

The commit sets `amdgpu_gfx_enforce_isolation_ring_begin_use` and
`amdgpu_gfx_enforce_isolation_ring_end_use` as the functions to be
called when a ring begins and ends its use, respectively.

`amdgpu_gfx_enforce_isolation_ring_begin_use` is called when a ring
begins its use. This function cancels any scheduled
`enforce_isolation_work` and, if necessary, signals the Kernel Fusion
Driver (KFD) to stop the runqueue.

`amdgpu_gfx_enforce_isolation_ring_end_use` is called when a ring ends
its use. This function schedules `enforce_isolation_work` to be run
after a delay.

These functions are part of the Enforce Isolation Handler, which
enforces shader isolation on AMD GPUs to prevent data leakage between
different processes.

The commit also includes a check for the type of the ring. If the type
of the ring is `AMDGPU_RING_TYPE_COMPUTE`, the `xcp_id` of the
`enforce_isolation` structure in the `gfx` structure of the
`amdgpu_device` is set to the `xcp_id` of the ring. This ensures that
the correct `xcp_id` is used when enforcing isolation on compute rings.
The `xcp_id` is an identifier for an XCP partition, and different rings
can be associated with different XCP partitions.

Cc: Christian König <christian.koenig@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>

drm/amdgpu/gfx9: Apply Isolation Enforcement to GFX & Compute rings

This commit applies isolation enforcement to the GFX and Compute rings
in the gfx_v9_0 module.

The commit sets `amdgpu_gfx_enforce_isolation_ring_begin_use` and
`amdgpu_gfx_enforce_isolation_ring_end_use` as the functions to be
called when a ring begins and ends its use, respectively.

`amdgpu_gfx_enforce_isolation_ring_begin_use` is called when a ring
begins its use. This function cancels any scheduled
`enforce_isolation_work` and, if necessary, signals the Kernel Fusion
Driver (KFD) to stop the runqueue.

`amdgpu_gfx_enforce_isolation_ring_end_use` is called when a ring ends
its use. This function schedules `enforce_isolation_work` to be run
after a delay.

These functions are part of the Enforce Isolation Handler, which
enforces shader isolation on AMD GPUs to prevent data leakage between
different processes.

Cc: Christian König <christian.koenig@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Suggested-by: Christian König <christian.koenig@amd.com>

drm/amdgpu: Implement Enforce Isolation Handler for KGD/KFD serialization

This commit introduces the Enforce Isolation Handler designed to enforce
shader isolation on AMD GPUs, which helps to prevent data leakage
between different processes.

The handler counts the number of emitted fences for each GFX and compute
ring. If there are any fences, it schedules the `enforce_isolation_work`
to be run after a delay of `GFX_SLICE_PERIOD`. If there are no fences,
it signals the Kernel Fusion Driver (KFD) to resume the runqueue.

The function is synchronized using the `enforce_isolation_mutex`.

This commit also introduces a reference count mechanism
(kfd_sch_req_count) to keep track of the number of requests to enable
the KFD scheduler. When a request to enable the KFD scheduler is made,
the reference count is decremented. When the reference count reaches
zero, a delayed work is scheduled to enforce isolation after a delay of
GFX_SLICE_PERIOD.

When a request to disable the KFD scheduler is made, the function first
checks if the reference count is zero. If it is, it cancels the delayed
work for enforcing isolation and checks if the KFD scheduler is active.
If the KFD scheduler is active, it sends a request to stop the KFD
scheduler and sets the KFD scheduler state to inactive. Then, it
increments the reference count.

The function is synchronized using the kfd_sch_mutex to ensure that the
KFD scheduler state and reference count are updated atomically.

Cc: Christian König <christian.koenig@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Suggested-by: Christian König <christian.koenig@amd.com>
Suggested-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: APIs to stop/start KFD scheduling

Provide amdgpu_amdkfd_stop_sched() for amdgpu to stop KFD scheduling
compute work on HIQ. amdgpu_amdkfd_start_sched() resumes the scheduling.
When amdgpu_amdkfd_stop_sched is called, KFD will unmap queues from
runlist. If users send ioctls to KFD to create queues, they'll be added
but those queues won't be mapped to runlist (so not scheduled) until
amdgpu_amdkfd_start_sched is called.

v2: fix build (Alex)

Signed-off-by: Amber Lin <Amber.Lin@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx9: Add cleaner shader support for GFX9.4.4 hardware

This commit extends the cleaner shader feature to support GFX9.4.4
hardware.

The cleaner shader feature is used to clear or initialize certain GPU
resources, such as Local Data Share (LDS), Vector General Purpose
Registers (VGPRs), and Scalar General Purpose Registers (SGPRs). This
operation needs to be performed in isolation, while no other tasks
should be running on the GPU at the same time.

Previously, the cleaner shader feature was implemented for GFX9.4.3
hardware. This commit adds support for GFX9.4.4 hardware by allowing the
cleaner shader to be used with this hardware version.

Cc: Christian König <christian.koenig@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx9: Add cleaner shader for GFX9.4.3

This commit adds the cleaner shader microcode for GFX9.4.3 GPUs. The
cleaner shader is a piece of GPU code that is used to clear or
initialize certain GPU resources, such as Local Data Share (LDS), Vector
General Purpose Registers (VGPRs), and Scalar General Purpose Registers
(SGPRs).

Clearing these resources is important for ensuring data isolation
between different workloads running on the GPU. Without the cleaner
shader, residual data from a previous workload could potentially be
accessed by a subsequent workload, leading to data leaks and incorrect
computation results.

The cleaner shader microcode is represented as an array of 32-bit words
(`gfx_9_4_3_cleaner_shader_hex`). This array is the binary
representation of the cleaner shader code, which is written in a
low-level GPU instruction set.

When the cleaner shader feature is enabled, the AMDGPU driver loads this
array into a specific location in the GPU memory. The GPU then reads
this memory location to fetch and execute the cleaner shader
instructions.

The cleaner shader is executed automatically by the GPU at the end of
each workload, before the next workload starts. This ensures that all
GPU resources are in a clean state before the start of each workload.

This addition is part of the cleaner shader feature implementation. The
cleaner shader feature helps improve GPU performance and resource
utilization by cleaning up GPU resources after they are used. It also
enhances security and reliability by preventing data leaks between
workloads.

v2: fix copyright date (Alex)

Cc: Christian König <christian.koenig@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx9: Implement cleaner shader support for GFX9.4.3 hardware

The patch modifies the gfx_v9_4_3_kiq_set_resources function to write
the cleaner shader's memory controller address to the ring buffer. It
also adds a new function, gfx_v9_4_3_ring_emit_cleaner_shader, which
emits the PACKET3_RUN_CLEANER_SHADER packet to the ring buffer.

This patch adds support for the PACKET3_RUN_CLEANER_SHADER packet in the
gfx_v9_4_3 module. This packet is used to emit the cleaner shader, which
is used to clear GPU memory before it's reused, helping to prevent data
leakage between different processes.

Finally, the patch updates the ring function structures to include the
new gfx_v9_4_3_ring_emit_cleaner_shader function. This allows the
cleaner shader to be emitted as part of the ring's operations.

Cc: Christian König <christian.koenig@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Suggested-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx9: Implement cleaner shader support for GFX9 hardware

The patch modifies the gfx_v9_0_kiq_set_resources function to write
the cleaner shader's memory controller address to the ring buffer. It
also adds a new function, gfx_v9_0_ring_emit_cleaner_shader, which
emits the PACKET3_RUN_CLEANER_SHADER packet to the ring buffer.

This patch adds support for the PACKET3_RUN_CLEANER_SHADER packet in the
gfx_v9_0 module. This packet is used to emit the cleaner shader, which
is used to clear GPU memory before it's reused, helping to prevent data
leakage between different processes.

Finally, the patch updates the ring function structures to include the
new gfx_v9_0_ring_emit_cleaner_shader function. This allows the
cleaner shader to be emitted as part of the ring's operations.

Cc: Christian König <christian.koenig@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Suggested-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Add PACKET3_RUN_CLEANER_SHADER for cleaner shader execution

This commit adds the PACKET3_RUN_CLEANER_SHADER definition. This packet
is a command packet used to instruct the GPU to execute the cleaner
shader.

The cleaner shader is a piece of GPU code that is used to clear or
initialize certain GPU resources, such as Local Data Share (LDS), Vector
General Purpose Registers (VGPRs), and Scalar General Purpose Registers
(SGPRs). Clearing these resources is important for ensuring data
isolation between different workloads running on the GPU.

The PACKET3_RUN_CLEANER_SHADER packet is used to trigger the execution
of the cleaner shader on the GPU. The packet consists of a header
followed by a RESERVED field, which is programmed to zero. When the GPU
receives this packet, it fetches and executes the cleaner shader
instructions from the location specified in the packet.

The cleaner shader feature helps to enhances security and reliability by
preventing data leaks between workloads.

Cc: Christian König <christian.koenig@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Add sysfs interface for running cleaner shader

This patch adds a new sysfs interface for running the cleaner shader on
AMD GPUs. The cleaner shader is used to clear GPU memory before it's
reused, which can help prevent data leakage between different processes.

The new sysfs file is write-only and is named `run_cleaner_shader`.
Write the number of the partition to this file to trigger the cleaner shader
on that partition. There is only one partition on GPUs which do not
support partitioning.

Changes made in this patch:

- Added `amdgpu_set_run_cleaner_shader` function to handle writes to the
  `run_cleaner_shader` sysfs file.
- Added `run_cleaner_shader` to the list of device attributes in
  `amdgpu_device_attrs`.
- Updated `default_attr_update` to handle `run_cleaner_shader`.
- Added `AMDGPU_DEVICE_ATTR_WO` macro to create write-only device
  attributes.

v2: fix error handling (Alex)

Cc: Christian König <christian.koenig@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>

drm/amdgpu: Add enforce_isolation sysfs attribute

This commit adds a new sysfs attribute 'enforce_isolation' to control
the 'enforce_isolation' setting per GPU. The attribute can be read and
written, and accepts values 0 (disabled) and 1 (enabled).

When 'enforce_isolation' is enabled, reserved VMIDs are allocated for
each ring. When it's disabled, the reserved VMIDs are freed.

The set function locks a mutex before changing the 'enforce_isolation'
flag and the VMIDs, and unlocks it afterwards. This ensures that these
operations are atomic and prevents race conditions and other concurrency
issues.

Cc: Christian König <christian.koenig@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Suggested-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Enforce isolation as part of the job

This patch adds a new parameter 'enforce_isolation' to the amdgpu_job
structure. This parameter is used to determine whether shader isolation
should be enforced for a job. The enforce_isolation parameter is then
stored in the amdgpu_job structure and used when flushing the VM.

The enforce_isolation field of the amdgpu_job structure is set directly
after the job is allocated

This change allows more fine-grained control over shader isolation,
making it possible to enforce isolation on a per-job basis rather than
globally. This can be useful in scenarios where only certain jobs
require isolation.

Cc: Christian König <christian.koenig@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Suggested-by: Christian König <christian.koenig@amd.com>

drm/amdgpu: abort KIQ waits when there is a pending reset

Stop waiting for the KIQ to return back when there is a reset pending.
It's quite likely that the KIQ will never response.

Signed-off-by: Koenig Christian <Christian.Koenig@amd.com>
Suggested-by: Lazar Lijo <Lijo.Lazar@amd.com>
Tested-by: Victor Skvortsov <victor.skvortsov@amd.com>
Signed-off-by: Victor Skvortsov <victor.skvortsov@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Make enforce_isolation setting per GPU

This commit makes enforce_isolation setting to be per GPU and per
partition by adding the enforce_isolation array to the adev structure.
The adev variable is set based on the global enforce_isolation module
parameter during device initialization.

In amdgpu_ids.c, the adev->enforce_isolation value for the current GPU
is used to determine whether to enforce isolation between graphics and
compute processes on that GPU.

In amdgpu_ids.c, the adev->enforce_isolation value for the current GPU
and partition is used to determine whether to enforce isolation between
graphics and compute processes on that GPU and partition.

This allows the enforce_isolation setting to be controlled individually
for each GPU and each partition, which is useful in a system with
multiple GPUs and partitions where different isolation settings might be
desired for different GPUs and partitions.

v2: fix loop in amdgpu_vmid_mgr_init() (Alex)

Cc: Christian König <christian.koenig@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Suggested-by: Christian König <christian.koenig@amd.com>

drm/amdgpu: Emit cleaner shader at end of IB submission

This commit introduces the emission of a cleaner shader at the end of
the IB submission process. This is achieved by adding a new function
pointer, `emit_cleaner_shader`, to the `amdgpu_ring_funcs` structure. If
the `emit_cleaner_shader` function is set in the ring functions, it is
called during the VM flush process.

The cleaner shader is only emitted if the `enable_cleaner_shader` flag
is set in the `amdgpu_device` structure. This allows the cleaner shader
emission to be controlled on a per-device basis.

By emitting a cleaner shader at the end of the IB submission, we can
ensure that the VM state is properly cleaned up after each submission.

Cc: Christian König <christian.koenig@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Suggested-by: Christian König <christian.koenig@amd.com>

drm/amdgpu: Add infrastructure for Cleaner Shader feature

The cleaner shader is used by the CP firmware to clean LDS and GPRs
between processes on the CUs.

This adds an internal API for GFX IP code to allocate and initialize the
cleaner shader.

Cc: Christian König <christian.koenig@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Suggested-by: Christian König <christian.koenig@amd.com>

drm/amdgpu: handle enforce isolation on non-0 gfxhub

Some chips have more than one gfxhub so check if we
are a gfxhub rather than just gfxhub 0.

Acked-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/sdma5.2: limit wptr workaround to sdma 5.2.1

The workaround seems to cause stability issues on other
SDMA 5.2.x IPs.

Fixes: a03ebf116303 ("drm/amdgpu/sdma5.2: Update wptr registers as well as doorbell")
Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/3556
Acked-by: Ruijing Dong <ruijing.dong@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add vcn ip dump support for vcn_v2_6

Add support for logging the registers in devcoredump
buffer for vcn_v2_6.

Signed-off-by: Sunil Khatri <sunil.khatri@amd.com>
Acked-by: Leo Liu <leo.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add print support for vcn_v2_5 ip dump

Add support for logging the registers in devcoredump
buffer for vcn_v2_5.

Signed-off-by: Sunil Khatri <sunil.khatri@amd.com>
Acked-by: Leo Liu <leo.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add vcn_v2_5 ip dump support

Add support of vcn ip dump in the devcoredump
for vcn_v2_5.

Signed-off-by: Sunil Khatri <sunil.khatri@amd.com>
Acked-by: Leo Liu <leo.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add print support for vcn_v2_0 ip dump

Add support for logging the registers in devcoredump
buffer for vcn_v2_0.

Signed-off-by: Sunil Khatri <sunil.khatri@amd.com>
Acked-by: Leo Liu <leo.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add vcn_v2_0 ip dump support

Add support of vcn ip dump in the devcoredump
for vcn_v2_0.

Signed-off-by: Sunil Khatri <sunil.khatri@amd.com>
Acked-by: Leo Liu <leo.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add print support for vcn_v1_0 ip dump

Add support for logging the registers in devcoredump
buffer for vcn_v1_0.

Signed-off-by: Sunil Khatri <sunil.khatri@amd.com>
Acked-by: Leo Liu <leo.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add vcn_v1_0 ip dump support

Add support of vcn ip dump in the devcoredump
for vcn_v1_0.

Signed-off-by: Sunil Khatri <sunil.khatri@amd.com>
Acked-by: Leo Liu <leo.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add print support for vcn_v4_0_5 ip dump

Add support for logging the registers in devcoredump
buffer for vcn_v4_0_5.

Signed-off-by: Sunil Khatri <sunil.khatri@amd.com>
Acked-by: Leo Liu <leo.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add print support for vcn_v4_0 ip dump

Add support for logging the registers in devcoredump
buffer for vcn_v4_0.

Signed-off-by: Sunil Khatri <sunil.khatri@amd.com>
Acked-by: Leo Liu <leo.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add print support for vcn_v4_0_3 ip dump

Add support for logging the registers in devcoredump
buffer for vcn_v4_0_3.

Signed-off-by: Sunil Khatri <sunil.khatri@amd.com>
Acked-by: Leo Liu <leo.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add vcn_v4_0_5 ip dump support

Add support of vcn ip dump in the devcoredump
for vcn_v4_0_5.

Signed-off-by: Sunil Khatri <sunil.khatri@amd.com>
Acked-by: Leo Liu <leo.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add vcn_v4_0 ip dump support

Add support of vcn ip dump in the devcoredump
for vcn_v4_0.

Signed-off-by: Sunil Khatri <sunil.khatri@amd.com>
Acked-by: Leo Liu <leo.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add vcn_v4_0_3 ip dump support

Add support of vcn ip dump in the devcoredump
for vcn_v4_0_3.

Signed-off-by: Sunil Khatri <sunil.khatri@amd.com>
Acked-by: Leo Liu <leo.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add print support for vcn_v5_0 ip dump

Add support for logging the registers in devcoredump
buffer for vcn_v5_0.

Signed-off-by: Sunil Khatri <sunil.khatri@amd.com>
Acked-by: Leo Liu <leo.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/mes12: add API for user queue reset

Add API for resetting user queues.

Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/mes11: add API for user queue reset

Add API for resetting user queues.

Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/mes: add API for user queue reset

Add API for resetting user queues.

Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx11: export gfx_v11_0_request_gfx_index_mutex()

It will be used by the queue reset code.

Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx11: add a mutex for the gfx semaphore

This will be used in more places in the future so
add a mutex.

Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx11: enter safe mode before touching CP_INT_CNTL

Need to enter safe mode before touching GC MMIO.

Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx7: add ring reset callback for gfx

Add ring reset callback for gfx.

v2: fix operator precedence (kernel test robot)

Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx8: add ring reset callback for gfx

Add ring reset callback for gfx.

v2: fix operator precedence (kernel test robot)

Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add vcn_v5_0 ip dump support

Add support of vcn ip dump in the devcoredump
for vcn_v5_0.

Signed-off-by: Sunil Khatri <sunil.khatri@amd.com>
Acked-by: Leo Liu <leo.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add print support for vcn_v3_0 ip dump

Add support for logging the registers in devcoredump
buffer for vcn_v3_0.

Signed-off-by: Sunil Khatri <sunil.khatri@amd.com>
Acked-by: Leo Liu <leo.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add vcn_v3_0 ip dump support

Add support of vcn ip dump in the devcoredump
for vcn_v3_0.

Signed-off-by: Sunil Khatri <sunil.khatri@amd.com>
Acked-by: Leo Liu <leo.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add vcn ip dump ptr in vcn global struct

Add pointer to the vcn ip dump in the vcn global structure
to be accessible for all vcn version via global adev.

Signed-off-by: Sunil Khatri <sunil.khatri@amd.com>
Acked-by: Leo Liu <leo.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd: Remove unused declarations

amdgpu_gart_table_vram_pin() and amdgpu_gart_table_vram_unpin() has
been removed since commit 575e55ee4fbc ("drm/amdgpu: recover gart table
at resume") remain the declarations untouched in the header files.

Besides, amdgpu_dm_display_resume() has also beed removed since
commit a80aa93de1a0 ("drm/amd/display: Unify dm resume sequence into a
single call"). So, let's remove this unused declarations.

Signed-off-by: Zhang Zekun <zhangzekun11@huawei.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/radeon/evergreen_cs: fix int overflow errors in cs track offsets

Several cs track offsets (such as 'track->db_s_read_offset')
either are initialized with or plainly take big enough values that,
once shifted 8 bits left, may be hit with integer overflow if the
resulting values end up going over u32 limit.

Same goes for a few instances of 'surf.layer_size * mslice'
multiplications that are added to 'offset' variable - they may
potentially overflow as well and need to be validated properly.

While some debug prints in this code section take possible overflow
issues into account, simply casting to (unsigned long) may be
erroneous in its own way, as depending on CPU architecture one is
liable to get different results.

Fix said problems by:
 - casting 'offset' to fixed u64 data type instead of
 ambiguous unsigned long.
 - casting one of the operands in vulnerable to integer
 overflow cases to u64.
 - adjust format specifiers in debug prints to properly
 represent 'offset' values.

Found by Linux Verification Center (linuxtesting.org) with static
analysis tool SVACE.

Fixes: 285484e2d55e ("drm/radeon: add support for evergreen/ni tiling informations v11")
Signed-off-by: Nikita Zhandarovich <n.zhandarovich@fintech.ru>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: fixing rlc firmware loading failure issue

Skip rlc firmware validation to ignore firmware header size mismatch issues.
This restores the workaround added in
commit 849e133c973c ("drm/amdgpu: Fix the null pointer when load rlc firmware")

Fixes: 3af2c80ae2f5 ("drm/amdgpu: refine gfx10 firmware loading")
Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/3551
Signed-off-by: Yang Wang <kevinyang.wang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: remove ME0 registers from  mi300 dump

Remove ME0 registers from MI300 gfx_9_4_3 ipdump
MI300 does not have  gfx ME and hence those register
are just empty one and could be dropped.

Signed-off-by: Sunil Khatri <sunil.khatri@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx9: use rlc safe mode for soft recovery

Protect the MMIO access with safe mode.

Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx9.4.3: use rlc safe mode for soft recovery

Protect the MMIO access with safe mode.

Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx9.4.3: use proper rlc safe mode helpers

Rather than open coding it for the queue reset.

Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx9: use proper rlc safe mode helpers

Rather than open coding it for the queue reset.

Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx9: add ring reset callback for gfx

Add ring reset callback for gfx.

Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx9: per queue reset only on bare metal

It's not supported under SR-IOV at the moment.

Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx9.4.3: implement reset_hw_queue for gfx9.4.3

Using mmio to do queue reset. Enter safe mode
before writing mmio registers.

v2: set register instance offset according to xcc id.

Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Jiadong Zhu <Jiadong.Zhu@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx9: implement reset_hw_queue for gfx9

Using mmio to do queue reset. Enter safe mode
when writing registers.

Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Jiadong Zhu <Jiadong.Zhu@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx: add a new kiq_pm4_funcs callback for reset_hw_queue

Add reset_hw_queue in kiq_pm4_funcs callbacks.

Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Jiadong Zhu <Jiadong.Zhu@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx_9.4.3: wait for reset done before remap

There is a racing condition that cp firmware modifies
MQD in reset sequence after driver updates it for
remapping. We have to wait till CP_HQD_ACTIVE becoming
false then remap the queue.

v2: fix KIQ locking (Alex)
v3: fix KIQ locking harder

Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Jiadong Zhu <Jiadong.Zhu@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx9.4.3: remap queue after reset successfully

Kiq command unmap_queues only does the dequeueing action.
We have to map the queue back with clean mqd.

Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Jiadong Zhu <Jiadong.Zhu@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx9.4.3: add ring reset callback

Add ring reset callback for compute.

Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx9: wait for reset done before remap

There is a racing condition that cp firmware modifies
MQD in reset sequence after driver updates it for
remapping. We have to wait till CP_HQD_ACTIVE becoming
false then remap the queue.

v2: fix KIQ locking (Alex)
v3: fix KIQ locking harder

Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Jiadong Zhu <Jiadong.Zhu@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx9: remap queue after reset successfully

Kiq command unmap_queues only does the dequeueing action.
We have to map the queue back with clean mqd.

Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Jiadong Zhu <Jiadong.Zhu@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx9: add ring reset callback

Add ring reset callback for compute.

Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: increase the reset counter for the queue reset

Update the reset counter for the amdgpu_cs_query_reset_state()

Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Prike Liang <Prike.Liang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add per ring reset support (v5)

If a specific job is hung, try and reset just the
ring associated with the job.

v2: move to amdgpu_job.c
v3: fix drm_sched_stop() handling when ring reset fails
v4: drop unnecessary amdgpu_fence_driver_clear_job_fences() and
    drm_sched_increase_karma()
v5: rework sched_stop handling

Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add new ring reset callback

Use this to reset just a single ring.

Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Return earlier in amdgpu_sw_ring_ib_end if mcbp is off

As we don't trigger preemption is sw ring muxer when mcbp is
disabled,so return earlier in amdgpu_sw_ring_ib_end function
if mcbp is disabled ,not required to call amdgpu_ring_mux_end_ib

Signed-off-by: Soham Dandapat <sdandapa@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add cp queue registers print for gfx9_4_3

Add gfx9_4_3 print support of CP queue registers
for all queues to be used by devcoredump.

Signed-off-by: Sunil Khatri <sunil.khatri@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add cp queue registers for gfx9_4_3 ipdump

Add gfx9 support of CP queue registers for all queues
to be used by devcoredump.

Signed-off-by: Sunil Khatri <sunil.khatri@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Align hwss_wait_for_all_blank_complete descriptor with implementation

The descriptor for `hwss_wait_for_all_blank_complete` was previously
misaligned with the actual implementation. This commit refines the
descriptor to reflect the implementation of
`hwss_wait_for_all_blank_complete`

Fixes the below with gcc W=1:
drivers/gpu/drm/amd/amdgpu/../display/dc/core/dc_hw_sequencer.c:991: warning: expecting prototype for hwss_wait_for_blank_complete(). Prototype was for hwss_wait_for_all_blank_complete() instead

Cc: Tom Chung <chiahsuan.chung@amd.com>
Cc: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Cc: Roman Li <roman.li@amd.com>
Cc: Alex Hung <alex.hung@amd.com>
Cc: Aurabindo Pillai <aurabindo.pillai@amd.com>
Cc: Harry Wentland <harry.wentland@amd.com>
Cc: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Reviewed-by: Tom Chung <chiahsuan.chung@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add print support for gfx9_4_3 ipdump

Add support of gfx9_4_3 ipdump print so devcoredump
could trigger it to dump the captured registers
in devcoredump.

Signed-off-by: Sunil Khatri <sunil.khatri@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add gfx9_4_3 register support in ipdump

Add general registers of gfx9_4_3 in ipdump for
devcoredump support.

Signed-off-by: Sunil Khatri <sunil.khatri@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/amdgpu: cleanup parse_cs callbacks

Because gpu_addr is updated in the calling routine
(amdgpu_cs_patch_ibs()),it is removed in the callback.

Use .patch_cs_in_place instead of .parse_cs for
amdgpu_vce_ring_parse_cs_vm() as there is no need for keeping
a temporary IB, therefore ib->sa_bo is NULL and amdgpu_ib_free()
is removed.

Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: David (Ming Qiang) Wu <David.Wu3@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/amdgpu: command submission parser for JPEG

Add JPEG IB command parser to ensure registers
in the command are within the JPEG IP block.

Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: David (Ming Qiang) Wu <David.Wu3@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/mes12: fix suspend issue

Use mes pipe to unmap kcq and kgq.

Signed-off-by: Jack Xiao <Jack.Xiao@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/mes12: sw/hw fini for unified mes

Free memory for two pipes and unmap pipe0 via pipe1.

Signed-off-by: Jack Xiao <Jack.Xiao@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/mes12: configure two pipes hardware resources

Configure two pipes with different hardware resources.

Signed-off-by: Jack Xiao <Jack.Xiao@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/mes12: adjust mes12 sw/hw init for multiple pipes

Adjust mes12 sw/hw initiailization for both pipe0 and pipe1
enablement. The two pipes are almost identical pipe. Pipe0
behaves like schq and pipe1 like kiq, pipe0 was mapped by pipe1.

Signed-off-by: Jack Xiao <Jack.Xiao@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/mes12: add mes pipe switch support

Add mes pipe switch to let caller choose pipe
to submit packet.

Signed-off-by: Jack Xiao <Jack.Xiao@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Block MMR_READ IOCTL in reset

Register access from userspace should be blocked until
reset is complete.

Signed-off-by: Victor Skvortsov <victor.skvortsov@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: fallback to pipe reset on queue reset fail for gfx9

If queue reset fails, tell the CP to reset the pipe.
Since queues multiplex context per pipe and we've issued a device wide
preemption prior to the hang, we can assume the hung pipe only has one
queue to reset on pipe reset.

Signed-off-by: Jonathan Kim <jonathan.kim@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Reorder to read EFI exported ROM first

On EFI BIOSes, PCI ROM may be exported through EFI_PCI_IO_PROTOCOL and
expansion ROM BARs may not be enabled. Choose to read from EFI exported
ROM data before reading PCI Expansion ROM BAR.

Signed-off-by: Lijo Lazar <lijo.lazar@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/mes12: load unified mes fw on pipe0 and pipe1

Enable unified mes firmware to load on pipe0 and pipe1.

Signed-off-by: Jack Xiao <Jack.Xiao@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Disable dpm_enabled flag while VF is in reset

VFs do not perform HW fini/suspend in FLR, so the dpm_enabled
is incorrectly kept enabled. Add interface to disable it in
virt_pre_reset call.

v2: Made implementation generic for all asics
v3: Re-order conditionals so PP_MP1_STATE_FLR is only evaluated on VF

Signed-off-by: Victor Skvortsov <victor.skvortsov@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Revert "drm/amdgpu: Extend KIQ reg polling wait for VF"

KIQ timeouts no longer seen.

This reverts commit 3a19a8af64eaff8a8b230796741a1a8277205344.

Signed-off-by: Victor Skvortsov <victor.skvortsov@amd.com>
Reviewed-by: Zhigang Luo <zhigang.luo@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Update kmd_fw_shared for VCN5

kmd_fw_shared changed in VCN5

Signed-off-by: Yinjie Yao <yinjie.yao@amd.com>
Reviewed-by: Ruijing Dong <ruijing.dong@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Add node_id to location_id generically

If there are multiple nodes per kfd device, add nodeid to location_id to
differentiate.

Signed-off-by: Lijo Lazar <lijo.lazar@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/amdgpu: add HDP_SD support on gc 12.0.0/1

add HDP_SD support on gc 12.0.0/1

Signed-off-by: Kenneth Feng <kenneth.feng@amd.com>
Reviewed-by: Yang Wang <kevinyang.wang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/sriov: extend NV_MAILBOX_POLL_MSG_TIMEDOUT

on MI300/MI308 UBB products, when doing mode1 reset, since 1 gpu need to
wait all 8 gpus finish mode1 reset and then do re-init. As observed,
sometimes the gpu which triggered the reset need to wait 15s for all
gpus to finish.

If poll msg timeout, guest driver will send the reset message again, and
may mess up the following reinit sequence on other gpus.

So extend the time to cover the maximum time needed to recover.

Signed-off-by: Victor Zhao <Victor.Zhao@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Promote DAL to 3.2.296

This version brings along following fixes:
- Fix some cursor issue
- Fix print format specifiers in DC_LOG_IPS
- Fix minor coding errors in dml21 phase 5
- Fix MST BW calculation Regression
- Improve FAM control for DCN401
- Add null pointer checks for some code
- Refactor 3DLUT for non-DMA
- Optimize vstartup position for AS-SDP
- Update to using new dccg callbacks
- Enable otg synchronization logic for DCN321
- Disable DCN401 UCLK P-State support on full updates

Acked-by: Wayne Lin <wayne.lin@amd.com>
Signed-off-by: Martin Leung <Martin.Leung@amd.com>
Signed-off-by: Tom Chung <chiahsuan.chung@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Remove unnecessary call to REG_SEQ_SUBMIT|WAIT_DONE

[why & how]
Remove unnecessary call to REG_SEQ_SUBMIT and REG_SEQ_WAIT_DONE, since
those macros are not necessary anymore at the dpp1 set degamma. Those
are part of an old implementation.

Acked-by: Wayne Lin <wayne.lin@amd.com>
Signed-off-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Tom Chung <chiahsuan.chung@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Adjust cursor position

[why & how]
When the commit 9d84c7ef8a87 ("drm/amd/display: Correct cursor position
on horizontal mirror") was introduced, it used the wrong calculation for
the position copy for X. This commit uses the correct calculation for that
based on the original patch.

Fixes: 9d84c7ef8a87 ("drm/amd/display: Correct cursor position on horizontal mirror")
Cc: Mario Limonciello <mario.limonciello@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Acked-by: Wayne Lin <wayne.lin@amd.com>
Signed-off-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Tom Chung <chiahsuan.chung@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix cursor offset on rotation 180

[why & how]
Cursor gets clipped off in the middle of the screen with hw
rotation 180. Fix a miscalculation of cursor offset when it's
placed near the edges in the pipe split case.

Cursor bugs with hw rotation were reported on AMD issue
tracker:
https://gitlab.freedesktop.org/drm/amd/-/issues/2247

The issues on rotation 270 was fixed by:
https://lore.kernel.org/amd-gfx/20221118125935.4013669-22-Brian.Chang@amd.com/
that partially addressed the rotation 180 too. So, this patch is the
final bits for rotation 180.

Reported-by: Xaver Hugl <xaver.hugl@gmail.com>
Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/2247
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Fixes: 9d84c7ef8a87 ("drm/amd/display: Correct cursor position on horizontal mirror")
Signed-off-by: Melissa Wen <mwen@igalia.com>
Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Tom Chung <chiahsuan.chung@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Improve FAM control for DCN401

[why & how]
When the commit 5324e2b205a2 ("drm/amd/display: Add driver support for
future FAMS versions") was introduced, it missed some of the FAM2 code.
This commit introduces the code that control the FAM enable and disable.

Fixes: 5324e2b205a2 ("drm/amd/display: Add driver support for future FAMS versions")
Acked-by: Wayne Lin <wayne.lin@amd.com>
Signed-off-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Tom Chung <chiahsuan.chung@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Remove unused field

[why & how]
Remove force_backlight_start_level since it is never used.

Acked-by: Wayne Lin <wayne.lin@amd.com>
Signed-off-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Tom Chung <chiahsuan.chung@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>