drm/amdgpu: move kiq_reg_write_reg_wait() out of amdgpu_virt.c

It's used for more than just SR-IOV now, so move it to
amdgpu_gmc.c and rename it to better match the functionality and
update the comments in the code paths to better document
when each path is used and why.  No functional change.

Reviewed-by: Shaoyun.liu <Shaoyun.liu@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: Shaoyun.Liu@amd.com
Cc: Christian.Koenig@amd.com

drm/amdgpu: add new INFO IOCTL query for input power

Some chips provide both average and input power.  Previously
we just exposed average power, add a new query for input
power.

Example userspace:
https://github.com/Umio-Yasuno/libdrm-amdgpu-sys-rs/tree/input_power

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

drm/amdgpu: Query boot status if boot failed

Check and report firmware boot status if it doesn't
reach steady status.

Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Yang Wang <kevinyang.wang@amd.com>
Reviewed-by: Le Ma <le.ma@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add ACA bank dump debugfs support

add ACA bank dump debugfs support

Signed-off-by: Yang Wang <kevinyang.wang@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add ACA kernel hardware error log support

add ACA kernel hardware error log support.

Signed-off-by: Yang Wang <kevinyang.wang@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Query boot status if discovery failed

Check and report boot status if discovery failed.

Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Yang Wang <kevinyang.wang@amd.com>
Reviewed-by: Le Ma <le.ma@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Add ras helper to query boot errors v2

Add ras helper function to query boot time gpu
errors.
v2: use aqua_vanjaram smn addressing pattern

Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Yang Wang <kevinyang.wang@amd.com>
Reviewed-by: Le Ma <le.ma@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: implement RAS ACA driver framework

v1:
implement new RAS ACA driver code framework.

v2:
- rename aca_bank_set to aca_banks.
- rename aca_source_xxx to aca_handle_xxx.

v3:
Optimize some function implementation details. (from Hawking's suggestion)

Signed-off-by: Yang Wang <kevinyang.wang@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Init pcie_index/data address as fallback (v2)

To allow using this helper for indirect access when
nbio funcs is not available. For instance, in ip
discovery phase.

v2: define macro for pcie_index/data/index_hi fallback.

Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Yang Wang <kevinyang.wang@amd.com>
Reviewed-by: Le Ma <le.ma@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: drop psp v13 query_boot_status implementation

Will replace it with new implementation to cover
boot fails in ip discovery phase.

Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Yang Wang <kevinyang.wang@amd.com>
Reviewed-by: Le Ma <le.ma@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Replace DRM_* with dev_* in amdgpu_psp.c

So kernel message has the device pcie bdf information,
which helps issue debugging especially in multiple GPU
system.

Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Tao Zhou <tao.zhou1@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Bump KFD ioctl version

This is not strictly a change in the IOCTL API. This version bump is meant
to indicate to user mode the presence of a number of changes and fixes
that enable the management of VA mappings in compute VMs using the GEM_VA
ioctl for DMABufs exported from KFD.

Signed-off-by: Felix Kuehling <felix.kuehling@amd.com>
Reviewed-by: Xiaogang Chen<Xiaogang.Chen@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Auto-validate DMABuf imports in compute VMs

DMABuf imports in compute VMs are not wrapped in a kgd_mem object on the
process_info->kfd_bo_list. There is no explicit KFD API call to validate
them or add eviction fences to them.

This patch automatically validates and fences dymanic DMABuf imports when
they are added to a compute VM. Revalidation after evictions is handled
in the VM code.

v2:
* Renamed amdgpu_vm_validate_evicted_bos to amdgpu_vm_validate
* Eliminated evicted_user state, use evicted state for VM BOs and user BOs
* Fixed and simplified amdgpu_vm_fence_imports, depends on reserved BOs
* Moved dma_resv_reserve_fences for amdgpu_vm_fence_imports into
  amdgpu_vm_validate, outside the vm->status_lock
* Added dummy version of amdgpu_amdkfd_bo_validate_and_fence for builds
  without KFD

v4: Eliminate amdgpu_vm_fence_imports. It's not needed because the
reservation with its fences is shared with the export, as long as all
imports are from KFD, with the exports already reserved, validated and
fenced by the KFD restore worker.

v5: Reintroduced separate evicted_user state to simplify the state machine
and CS error handling when amdgpu_vm_validate is called without a ticket.

Signed-off-by: Felix Kuehling <felix.kuehling@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Avoid enum conversion warning

Clang warns (or errors with CONFIG_WERROR=y) when performing arithmetic
with different enumerated types, which is usually a bug:

    drivers/gpu/drm/amd/amdgpu/../display/dc/link/protocols/link_dp_dpia_bw.c:548:24: error: arithmetic between different enumeration types ('const enum dc_link_rate' and 'const enum dc_lane_count') [-Werror,-Wenum-enum-conversion]
      548 |                         link_cap->link_rate * link_cap->lane_count * LINK_RATE_REF_FREQ_IN_KHZ * 8;
          |                         ~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~~~~~~~~~~~~
    1 error generated.

In this case, there is not a problem because the enumerated types are
basically treated as '#define' values. Add an explicit cast to an
integral type to silence the warning.

Closes: https://github.com/ClangBuiltLinux/linux/issues/1976
Fixes: 5f3bce13266e ("drm/amd/display: Request usb4 bw for mst streams")
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: Fix smuv13.0.6 current clock reporting

When current clock is equal to max dpm level clock, the level is not
indicated correctly with *. Fix by comparing current clock against dpm
level value.

Signed-off-by: Lijo Lazar <lijo.lazar@amd.com>
Reviewed-by: Asad Kamal <asad.kamal@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org # 6.7.x

drm/amd/pm: Add error log for smu v13.0.6 reset

For all mode-2 reset fail cases, add error log.

Signed-off-by: Lijo Lazar <lijo.lazar@amd.com>
Reviewed-by: Asad Kamal <asad.kamal@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org # 6.7.x

drm/amdkfd: Fix 'node' NULL check in 'svm_range_get_range_boundaries()'

Range interval [start, last] is ordered by rb_tree, rb_prev, rb_next
return value still needs NULL check, thus modified from "node" to "rb_node".

Fixes the below:
drivers/gpu/drm/amd/amdgpu/../amdkfd/kfd_svm.c:2691 svm_range_get_range_boundaries() warn: can 'node' even be NULL?

Suggested-by: Philip Yang <Philip.Yang@amd.com>
Cc: Felix Kuehling <Felix.Kuehling@amd.com>
Cc: Christian König <christian.koenig@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Reviewed-by: Felix Kuehling <felix.kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: drop exp hw support check for GC 9.4.3

No longer needed.

Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org # 6.7.x

drm/amdgpu: move debug options init prior to amdgpu device init

To bring debug options into effect in early initialization phase

Signed-off-by: Le Ma <le.ma@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add debug flag to place fw bo on vram for frontdoor loading

Use debug_mask=0x8 param to help isolating data path issues
on new systems in early phase.

v2: rename the flag for explicitness (lijo)

Signed-off-by: Le Ma <le.ma@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Revert "drm/amdgpu: add param to specify fw bo location for front-door loading"

This reverts commit c572abffe9f50c8ba33060865449313b3f588c35.

Will use debug module param instead of independent module param.

Signed-off-by: Le Ma <le.ma@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix variable deferencing before NULL check in edp_setup_replay()

In edp_setup_replay(), 'struct dc *dc' & 'struct dmub_replay *replay'
was dereferenced before the pointer 'link' & 'replay' NULL check.

Fixes the below:
drivers/gpu/drm/amd/amdgpu/../display/dc/link/protocols/link_edp_panel_control.c:947 edp_setup_replay() warn: variable dereferenced before check 'link' (see line 933)

Cc: stable@vger.kernel.org
Cc: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Cc: Harry Wentland <harry.wentland@amd.com>
Cc: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Cc: Aurabindo Pillai <aurabindo.pillai@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Reviewed-by: Aurabindo Pillai <aurabindo.pillai@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: update regGL2C_CTRL4 value in golden setting

This patch to update regGL2C_CTRL4 in golden setting.

Signed-off-by: Yifan Zhang <yifan1.zhang@amd.com>
Reviewed-by: Tim Huang <Tim.Huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org # 6.7.x

drm/amdgpu: Release 'adev->pm.fw' before return in 'amdgpu_device_need_post()'

In function 'amdgpu_device_need_post(struct amdgpu_device *adev)' -
'adev->pm.fw' may not be released before return.

Using the function release_firmware() to release adev->pm.fw.

Thus fixing the below:
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c:1571 amdgpu_device_need_post() warn: 'adev->pm.fw' from request_firmware() not released on lines: 1554.

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

drm/amdgpu: Fix unsigned comparison with less than zero in vpe_u1_8_from_fraction()

The variables 'numerator' and 'denominator', are unsigned 16-bit integer
types, that can never be less than 0.

Thus fixing the below:
drivers/gpu/drm/amd/amdgpu/amdgpu_vpe.c:62 vpe_u1_8_from_fraction() warn: unsigned 'numerator' is never less than zero.
drivers/gpu/drm/amd/amdgpu/amdgpu_vpe.c:63 vpe_u1_8_from_fraction() warn: unsigned 'denominator' is never less than zero.

Cc: Peyton Lee <peytolee@amd.com>
Cc: Lang Yu <lang.yu@amd.com>
Cc: Christian König <christian.koenig@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Reviewed-by: Peyton Lee <peyton.lee@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Fix with right return code '-EIO' in 'amdgpu_gmc_vram_checking()'

The amdgpu_gmc_vram_checking() function in emulation checks whether
all of the memory range of shared system memory could be accessed by
GPU, from this aspect, -EIO is returned for error scenarios.

Fixes the below:
drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c:919 gmc_v6_0_hw_init() warn: missing error code? 'r'
drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c:1103 gmc_v7_0_hw_init() warn: missing error code? 'r'
drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c:1223 gmc_v8_0_hw_init() warn: missing error code? 'r'
drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c:2344 gmc_v9_0_hw_init() warn: missing error code? 'r'

Cc: Xiaojian Du <Xiaojian.Du@amd.com>
Cc: Lijo Lazar <lijo.lazar@amd.com>
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: Christian König <christian.koenig@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Do not program VM_L2_CNTL under SRIOV

VM_L2_CNTL* should not be programmed on driver unload under SRIOV.
These regs are skipped during SRIOV driver init.

Signed-off-by: Victor Lu <victorchengchi.lu@amd.com>
Reviewed-by: Vignesh Chander <Vignesh.Chander@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/powerplay: Fix kzalloc parameter 'ATOM_Tonga_PPM_Table' in 'get_platform_power_management_table()'

In 'struct phm_ppm_table *ptr' allocation using kzalloc, an incorrect
structure type is passed to sizeof() in kzalloc, larger structure types
were used, thus using correct type 'struct phm_ppm_table' fixes the
below:

drivers/gpu/drm/amd/amdgpu/../pm/powerplay/hwmgr/process_pptables_v1_0.c:203 get_platform_power_management_table() warn: struct type mismatch 'phm_ppm_table vs _ATOM_Tonga_PPM_Table'

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

drm/amdgpu: update ATHUB_MISC_CNTL offset for athub v3.3

This patch to update ATHUB_MISC_CNTL offset for athub v3.3

v2: correct a typo (Tim)
v3: correct patch title (Lang)

Signed-off-by: Yifan Zhang <yifan1.zhang@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Tim Huang <Tim.Huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: update headers for nbio v7.11

This patch is to update headers for nbio v7.11.

Signed-off-by: Yifan Zhang <yifan1.zhang@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Tim Huang <Tim.Huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/pm: clarify debugfs pm output

On APUs power is SoC power, not just GPU.
Clarify that for UVD/VCE/VCN the IP is powered down,
not disabled which can confusing and lead to concerns
that the IP is actually not available.

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

drm/amdgpu: fall back to INPUT power for AVG power via INFO IOCTL

For backwards compatibility with userspace.

Fixes: 47f1724db4fe ("drm/amd: Introduce `AMDGPU_PP_SENSOR_GPU_INPUT_POWER`")
Link: https://gitlab.freedesktop.org/drm/amd/-/issues/2897
Reviewed-by: Yang Wang <kevinyang.wang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: fix avg vs input power reporting on smu7

Hawaii, Bonaire, Fiji, and Tonga support average power, the others
support current power.

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

drm/amdkfd: fixes for HMM mem allocation

Fix err return value and reset pgmap->type after checking it.

Fixes: c83dee9b6394 ("drm/amdkfd: add SPM support for SVM")
Reviewed-by: Felix Kuehling <felix.kuehling@amd.com>
Signed-off-by: Dafna Hirschfeld <dhirschfeld@habana.ai>
Signed-off-by: Felix Kuehling <felix.kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: cleanup inconsistent indenting in amdgpu_dm_color

smatch warnings:
amdgpu_dm_update_plane_color_mgmt() warn: inconsistent indenting

Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202401051643.PPdbmG1U-lkp@intel.com/
Signed-off-by: Melissa Wen <mwen@igalia.com>
Reviewed-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: make a correction on comment

Use a generic comment for AMDGPU_VM_RESERVED_VRAM size.

Signed-off-by: James Zhu <James.Zhu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Revert "drm/amd/display: fix bandwidth validation failure on DCN 2.1"

This commit causes dmesg-warn on several IGT tests on DCN 3.1.6: *ERROR*
link_enc_cfg_validate: Invalid link encoder assignments - 0x1c

Affected IGT tests include:
- amdgpu/[amd_assr|amd_plane|amd_hotplug]
- kms_atomic
- kms_color
- kms_flip
- kms_properties
- kms_universal_plane

and some other tests

This reverts commit 3a0fa3bc245ef92838a8296e0055569b8dff94c4.

Cc: Melissa Wen <mwen@igalia.com>
Cc: Hamza Mahfooz <hamza.mahfooz@amd.com>
Reviewed-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Ivan Lipski <ivlipski@amd.com>
Signed-off-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Fix sparse __rcu annotation warnings

Properly mark kfd_process->ef as __rcu and consistently use the right
accessor functions.

Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202312052245.yFpBSgNH-lkp@intel.com/
Signed-off-by: Felix Kuehling <felix.kuehling@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Fix lock dependency warning with srcu

======================================================
WARNING: possible circular locking dependency detected
6.5.0-kfd-yangp #2289 Not tainted
------------------------------------------------------
kworker/0:2/996 is trying to acquire lock:
        (srcu){.+.+}-{0:0}, at: __synchronize_srcu+0x5/0x1a0

but task is already holding lock:
        ((work_completion)(&svms->deferred_list_work)){+.+.}-{0:0}, at:
process_one_work+0x211/0x560

which lock already depends on the new lock.

the existing dependency chain (in reverse order) is:

-> #3 ((work_completion)(&svms->deferred_list_work)){+.+.}-{0:0}:
        __flush_work+0x88/0x4f0
        svm_range_list_lock_and_flush_work+0x3d/0x110 [amdgpu]
        svm_range_set_attr+0xd6/0x14c0 [amdgpu]
        kfd_ioctl+0x1d1/0x630 [amdgpu]
        __x64_sys_ioctl+0x88/0xc0

-> #2 (&info->lock#2){+.+.}-{3:3}:
        __mutex_lock+0x99/0xc70
        amdgpu_amdkfd_gpuvm_restore_process_bos+0x54/0x740 [amdgpu]
        restore_process_helper+0x22/0x80 [amdgpu]
        restore_process_worker+0x2d/0xa0 [amdgpu]
        process_one_work+0x29b/0x560
        worker_thread+0x3d/0x3d0

-> #1 ((work_completion)(&(&process->restore_work)->work)){+.+.}-{0:0}:
        __flush_work+0x88/0x4f0
        __cancel_work_timer+0x12c/0x1c0
        kfd_process_notifier_release_internal+0x37/0x1f0 [amdgpu]
        __mmu_notifier_release+0xad/0x240
        exit_mmap+0x6a/0x3a0
        mmput+0x6a/0x120
        do_exit+0x322/0xb90
        do_group_exit+0x37/0xa0
        __x64_sys_exit_group+0x18/0x20
        do_syscall_64+0x38/0x80

-> #0 (srcu){.+.+}-{0:0}:
        __lock_acquire+0x1521/0x2510
        lock_sync+0x5f/0x90
        __synchronize_srcu+0x4f/0x1a0
        __mmu_notifier_release+0x128/0x240
        exit_mmap+0x6a/0x3a0
        mmput+0x6a/0x120
        svm_range_deferred_list_work+0x19f/0x350 [amdgpu]
        process_one_work+0x29b/0x560
        worker_thread+0x3d/0x3d0

other info that might help us debug this:
Chain exists of:
  srcu --> &info->lock#2 --> (work_completion)(&svms->deferred_list_work)

Possible unsafe locking scenario:

        CPU0                    CPU1
        ----                    ----
        lock((work_completion)(&svms->deferred_list_work));
                        lock(&info->lock#2);
lock((work_completion)(&svms->deferred_list_work));
        sync(srcu);

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>

drm/amdgpu: Packed socket_id to ras feature mask

Initialize RAS feature mask bit[31:29] with socket_id.

Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Tao Zhou <tao.zhou1@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Support poison error injection via ras_ctrl debugfs

Support poison error injection.

Signed-off-by: Candice Li <candice.li@amd.com>
Reviewed-by: Tao Zhou <tao.zhou1@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: correct the cu count for gfx v11

Correct the algorithm of active CU to skip disabled
sa for gfx v11.

Signed-off-by: Likun Gao <Likun.Gao@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org

drm/amdgpu: Drop unnecessary sentences about CE and deferred error.

Remove "no user action is needed" for correctable and deferred error
to avoid confusion.

Signed-off-by: Candice Li <candice.li@amd.com>
Reviewed-by: Tao Zhou <tao.zhou1@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: 3.2.266

This version brings along following fixes:

- Improve z8/z10 support.
- Revert some of the VRR optimization.
- Improve usb4 when using MST.

Acked-by: Rodrigo Siqueira <rodrigo.siqueira@amd.com>
Signed-off-by: Aric Cyr <aric.cyr@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Dpia hpd status not in sync after S4

[Why]
Dpia hpd status not in sync causing driver not enabling BW Alloc after
S4.

[How]
Update hpd_status of the link when querying hpd state from dmub in
dpia_query_hpd_status().

Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Acked-by: Rodrigo Siqueira <rodrigo.siqueira@amd.com>
Signed-off-by: Meenakshikumar Somasundaram <meenakshikumar.somasundaram@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Update z8 latency

Adjust z8 latency for performance.

Reviewed-by: Muhammad Ahmed <ahmed.ahmed@amd.com>
Acked-by: Rodrigo Siqueira <rodrigo.siqueira@amd.com>
Signed-off-by: Charlene Liu <charlene.liu@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Revert "drm/amd/display: Fix conversions between bytes and KB"

This reverts commit d0f639c5869399bf6dde4d694d5f8c0ab8c0ec46.

The previous commit causes failure to light up for 1080p
eDP + 8k HDMI panel combo.

Reviewed-by: Charlene Liu <charlene.liu@amd.com>
Acked-by: Rodrigo Siqueira <rodrigo.siqueira@amd.com>
Signed-off-by: Daniel Miess <daniel.miess@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Request usb4 bw for mst streams

[WHY]
When usb4 bandwidth allocation mode is enabled, driver need to request
bandwidth from connection manager. For mst link,  the requested
bandwidth should be big enough for all remote streams.

[HOW]
- If mst link, the requested bandwidth should be the sum of all mst
  streams bandwidth added with dp MTPH overhead.
- Allocate/deallcate usb4 bandwidth when setting dpms on/off.
- When doing display mode validation, driver also need to consider total
  bandwidth of all mst streams for mst link.

Reviewed-by: Cruise Hung <cruise.hung@amd.com>
Acked-by: Rodrigo Siqueira <rodrigo.siqueira@amd.com>
Signed-off-by: Peichen Huang <peichen.huang@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: revert "Optimize VRR updates to only necessary ones"

This reverts commit 6e4337f695c25162f0296934152506ad596fcebf.

The original commit causes regression in corner case with HDMI at
specific timings. Reverting from staging to get the full suite to
retest.

Reviewed-by: Rodrigo Siqueira <rodrigo.siqueira@amd.com>
Signed-off-by: Martin Leung <martin.leung@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: revert "for FPO & SubVP/DRR config program vmin/max"

This reverts commit 6b2b782ad6a25734ae847d1659bea3f613dbb563.

The original commit causes issues with certain features when DRR is
disabled, need to revisit this change later after resolving issues with
new DRR policy.

Reviewed-by: Rodrigo Siqueira <rodrigo.siqueira@amd.com>
Signed-off-by: Martin Leung <martin.leung@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Disconnect phantom pipe OPP from OPTC being disabled

[Why]
If an OPP is used for a different OPTC without first being disconnected
from the previous OPTC, unexpected behaviour can occur. This also
applies to phantom pipes, which is what the current logic missed.

[How]
Disconnect OPPs from OPTC for phantom pipes before disabling OTG master.

Also move the disconnection to before the OTG master disable, since the
register is double buffered.

Reviewed-by: Dillon Varone <dillon.varone@amd.com>
Acked-by: Rodrigo Siqueira <rodrigo.siqueira@amd.com>
Signed-off-by: George Shen <george.shen@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: To adjust dprefclk by down spread percentage

[Why]
Panels show corruption with high refresh rate timings when ssc is
enabled.

[How]
Read down-spread percentage from lut to adjust dprefclk. Issues come
from S0i3 with this commit has been fixed by SMU.

Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Acked-by: Rodrigo Siqueira <rodrigo.siqueira@amd.com>
Signed-off-by: Martin Tsai <martin.tsai@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Fix lock dependency warning

======================================================
WARNING: possible circular locking dependency detected
6.5.0-kfd-fkuehlin #276 Not tainted
------------------------------------------------------
kworker/8:2/2676 is trying to acquire lock:
ffff9435aae95c88 ((work_completion)(&svm_bo->eviction_work)){+.+.}-{0:0}, at: __flush_work+0x52/0x550

but task is already holding lock:
ffff9435cd8e1720 (&svms->lock){+.+.}-{3:3}, at: svm_range_deferred_list_work+0xe8/0x340 [amdgpu]

which lock already depends on the new lock.

the existing dependency chain (in reverse order) is:

-> #2 (&svms->lock){+.+.}-{3:3}:
       __mutex_lock+0x97/0xd30
       kfd_ioctl_alloc_memory_of_gpu+0x6d/0x3c0 [amdgpu]
       kfd_ioctl+0x1b2/0x5d0 [amdgpu]
       __x64_sys_ioctl+0x86/0xc0
       do_syscall_64+0x39/0x80
       entry_SYSCALL_64_after_hwframe+0x63/0xcd

-> #1 (&mm->mmap_lock){++++}-{3:3}:
       down_read+0x42/0x160
       svm_range_evict_svm_bo_worker+0x8b/0x340 [amdgpu]
       process_one_work+0x27a/0x540
       worker_thread+0x53/0x3e0
       kthread+0xeb/0x120
       ret_from_fork+0x31/0x50
       ret_from_fork_asm+0x11/0x20

-> #0 ((work_completion)(&svm_bo->eviction_work)){+.+.}-{0:0}:
       __lock_acquire+0x1426/0x2200
       lock_acquire+0xc1/0x2b0
       __flush_work+0x80/0x550
       __cancel_work_timer+0x109/0x190
       svm_range_bo_release+0xdc/0x1c0 [amdgpu]
       svm_range_free+0x175/0x180 [amdgpu]
       svm_range_deferred_list_work+0x15d/0x340 [amdgpu]
       process_one_work+0x27a/0x540
       worker_thread+0x53/0x3e0
       kthread+0xeb/0x120
       ret_from_fork+0x31/0x50
       ret_from_fork_asm+0x11/0x20

other info that might help us debug this:

Chain exists of:
  (work_completion)(&svm_bo->eviction_work) --> &mm->mmap_lock --> &svms->lock

 Possible unsafe locking scenario:

       CPU0                    CPU1
       ----                    ----
  lock(&svms->lock);
                               lock(&mm->mmap_lock);
                               lock(&svms->lock);
  lock((work_completion)(&svm_bo->eviction_work));

I believe this cannot really lead to a deadlock in practice, because
svm_range_evict_svm_bo_worker only takes the mmap_read_lock if the BO
refcount is non-0. That means it's impossible that svm_range_bo_release
is running concurrently. However, there is no good way to annotate this.

To avoid the problem, take a BO reference in
svm_range_schedule_evict_svm_bo instead of in the worker. That way it's
impossible for a BO to get freed while eviction work is pending and the
cancel_work_sync call in svm_range_bo_release can be eliminated.

v2: Use svm_bo_ref_unless_zero and explained why that's safe. Also
removed redundant checks that are already done in
amdkfd_fence_enable_signaling.

Signed-off-by: Felix Kuehling <felix.kuehling@amd.com>
Reviewed-by: Philip Yang <philip.yang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Merge tag 'amd-drm-next-6.8-2024-01-05' of https://gitlab.freedesktop.org/agd5f/linux into drm-next

amd-drm-next-6.8-2024-01-05:

amdgpu:
- VRR fixes
- PSR-SU fixes
- SubVP fixes
- DCN 3.5 fixes
- Documentation updates
- DMCUB fixes
- DML2 fixes
- UMC 12.0 updates
- GPUVM fix
- Misc code cleanups and whitespace cleanups
- DP MST fix
- Let KFD sync with GPUVM fences
- GFX11 reset fix
- SMU 13.0.6 fixes
- VSC fix for DP/eDP
- Navi12 display fix
- RN/CZN system aperture fix
- DCN 2.1 bandwidth validation fix
- DCN INIT cleanup

amdkfd:
- SVM fixes
- Revert TBA/TMA location change

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240105220522.4976-1-alexander.deucher@amd.com

drm/amd/display: Allow z8/z10 from driver

Copy StutterPeriod from DML2 into DML1 StutterPeriod parameter.

Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Reviewed-by: Muhammad Ahmed <ahmed.ahmed@amd.com>
Reviewed-by: Aric Cyr <aric.cyr@amd.com>
Acked-by: Rodrigo Siqueira <rodrigo.siqueira@amd.com>
Signed-off-by: Charlene Liu <charlene.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix bandwidth validation failure on DCN 2.1

IGT `amdgpu/amd_color/crtc-lut-accuracy` fails right at the beginning of
the test execution, during atomic check, because DC rejects the
bandwidth state for a fb sizing 64x64. The test was previously working
with the deprecated dc_commit_state(). Now using
dc_validate_with_context() approach, the atomic check needs to perform a
full state validation. Therefore, set fast_validation to false in the
dc_validate_global_state call for atomic check.

Cc: stable@vger.kernel.org
Fixes: b8272241ff9d ("drm/amd/display: Drop dc_commit_state in favor of dc_commit_streams")
Signed-off-by: Melissa Wen <mwen@igalia.com>
Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: apply the RV2 system aperture fix to RN/CZN as well

These chips needs the same fix.  This was previously not seen
on then since the AGP aperture expanded the system aperture,
but this showed up again when AGP was disabled.

Reviewed-and-tested-by: Jiadong Zhu <Jiadong.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Move fixpt_from_s3132 to amdgpu_dm

Other environments don't like the unary minus operator on
an unsigned value.

Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix recent checkpatch errors in amdgpu_dm

 - Use tabs, not spaces.
 - Brace and parentheses placement

Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Revert "drm/amdkfd: Relocate TBA/TMA to opposite side of VM hole"

That commit causes NULL pointer dereferences in dmesgs when
running applications using ROCm, including clinfo, blender,
and PyTorch, since v6.6.1. Revert it to fix blender again.

This reverts commit 96c211f1f9ef82183493f4ceed4e347b52849149.

Closes: https://github.com/ROCm/ROCm/issues/2596
Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/2991
Reviewed-by: Jay Cornwall <jay.cornwall@amd.com>
Signed-off-by: Kaibo Ma <ent3rm4n@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: avoid stringop-overflow warnings for dp_decide_lane_settings()

gcc prints a warning about a possible array overflow for a couple of
callers of dp_decide_lane_settings() after commit 1b56c90018f0 ("Makefile:
Enable -Wstringop-overflow globally"):

drivers/gpu/drm/amd/amdgpu/../display/dc/link/protocols/link_dp_training_fixed_vs_pe_retimer.c: In function 'dp_perform_fixed_vs_pe_training_sequence_legacy':
drivers/gpu/drm/amd/amdgpu/../display/dc/link/protocols/link_dp_training_fixed_vs_pe_retimer.c:426:25: error: 'dp_decide_lane_settings' accessing 4 bytes in a region of size 1 [-Werror=stringop-overflow=]
  426 |                         dp_decide_lane_settings(lt_settings, dpcd_lane_adjust,
      |                         ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  427 |                                         lt_settings->hw_lane_settings, lt_settings->dpcd_lane_settings);
      |                                         ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
drivers/gpu/drm/amd/amdgpu/../display/dc/link/protocols/link_dp_training_fixed_vs_pe_retimer.c:426:25: note: referencing argument 4 of type 'union dpcd_training_lane[4]'

I'm not entirely sure what caused this, but changing the prototype to expect
a pointer instead of an array avoids the warnings.

Fixes: 7727e7b60f82 ("drm/amd/display: Improve robustness of FIXED_VS link training at DP1 rates")
Acked-by: Randy Dunlap <rdunlap@infradead.org>
Tested-by: Randy Dunlap <rdunlap@infradead.org> # build-tested
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix power_helpers.c codestyle

Place define macro expression inside () in power_helpers.c file

Signed-off-by: Marcelo Mendes Spessoto Junior <marcelomspessoto@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix hdcp_log.h codestyle

Place HDCP_EVENT_TRACE(hdcp, event) macro content inside do while loop
to avoid if-else issues in hdcp_log.h file

v2: fix up build (Alex)

Signed-off-by: Marcelo Mendes Spessoto Junior <marcelomspessoto@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix hdcp2_execution.c codestyle

Remove braces for single statement if expressions and change comparison
order for hdcp2_execution.c file

Signed-off-by: Marcelo Mendes Spessoto Junior <marcelomspessoto@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix hdcp_psp.h codestyle

Fix identation inside enum and place expressions in define macros inside
() for hdcp_psp.h file

Signed-off-by: Marcelo Mendes Spessoto Junior <marcelomspessoto@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix freesync.c codestyle

Remove braces for single statement if expression for freesync.c file

Signed-off-by: Marcelo Mendes Spessoto Junior <marcelomspessoto@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix hdcp_psp.c codestyle

Fix identation for hdcp_psp.c file

Signed-off-by: Marcelo Mendes Spessoto Junior <marcelomspessoto@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix hdcp1_execution.c codestyle

Remove braces from single statement if expression in hdcp1_execution.c
file

Signed-off-by: Marcelo Mendes Spessoto Junior <marcelomspessoto@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm/smu7: fix a memleak in smu7_hwmgr_backend_init

The hwmgr->backend, (i.e. data) allocated by kzalloc is not freed in
the error-handling paths of smu7_get_evv_voltages and
smu7_update_edc_leakage_table. However, it did be freed in the
error-handling of phm_initializa_dynamic_state_adjustment_rule_settings,
by smu7_hwmgr_backend_fini. So the lack of free in smu7_get_evv_voltages
and smu7_update_edc_leakage_table is considered a memleak in this patch.

Fixes: 599a7e9fe1b6 ("drm/amd/powerplay: implement smu7 hwmgr to manager asics with smu ip version 7.")
Fixes: 8f0804c6b7d0 ("drm/amd/pm: add edc leakage controller setting")
Signed-off-by: Zhipeng Lu <alexious@zju.edu.cn>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Fix iterator used outside loop in 'kfd_add_peer_prop()'

Fix the following about iterator use:
drivers/gpu/drm/amd/amdgpu/../amdkfd/kfd_topology.c:1456 kfd_add_peer_prop() warn: iterator used outside loop: 'iolink3'

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

drm/amdgpu: Drop 'fence' check in 'to_amdgpu_amdkfd_fence()'

Return value of container_of(...) can't be null, so null check is not
required for 'fence'. Hence drop its NULL check.

Fixes the below:
drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_fence.c:93 to_amdgpu_amdkfd_fence() warn: can 'fence' even be NULL?

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

drm/amdkfd: Confirm list is non-empty before utilizing list_first_entry in kfd_topology.c

Before using list_first_entry, make sure to check that list is not
empty, if list is empty return -ENODATA.

Fixes the below:
drivers/gpu/drm/amd/amdgpu/../amdkfd/kfd_topology.c:1347 kfd_create_indirect_link_prop() warn: can 'gpu_link' even be NULL?
drivers/gpu/drm/amd/amdgpu/../amdkfd/kfd_topology.c:1428 kfd_add_peer_prop() warn: can 'iolink1' even be NULL?
drivers/gpu/drm/amd/amdgpu/../amdkfd/kfd_topology.c:1433 kfd_add_peer_prop() warn: can 'iolink2' even be NULL?

Fixes: 0f28cca87e9a ("drm/amdkfd: Extend KFD device topology to surface peer-to-peer links")
Cc: Felix Kuehling <Felix.Kuehling@amd.com>
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: Felix Kuehling <Felix.Kuehling@amd.com>
Suggested-by: Lijo Lazar <lijo.lazar@amd.com>
Reviewed-by: Felix Kuehling <felix.kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Fix '*fw' from request_firmware() not released in 'amdgpu_ucode_request()'

Fixes the below:
drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c:1404 amdgpu_ucode_request() warn: '*fw' from request_firmware() not released on lines: 1404.

Cc: Mario Limonciello <mario.limonciello@amd.com>
Cc: Lijo Lazar <lijo.lazar@amd.com>
Cc: Christian König <christian.koenig@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Fix variable 'mca_funcs' dereferenced before NULL check in 'amdgpu_mca_smu_get_mca_entry()'

Fixes the below:

drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c:377 amdgpu_mca_smu_get_mca_entry() warn: variable dereferenced before check 'mca_funcs' (see line 368)

357 int amdgpu_mca_smu_get_mca_entry(struct amdgpu_device *adev,
     enum amdgpu_mca_error_type type,
358                                  int idx, struct mca_bank_entry *entry)
359 {
360         const struct amdgpu_mca_smu_funcs *mca_funcs =
adev->mca.mca_funcs;
361         int count;
362
363         switch (type) {
364         case AMDGPU_MCA_ERROR_TYPE_UE:
365                 count = mca_funcs->max_ue_count;

mca_funcs is dereferenced here.

366                 break;
367         case AMDGPU_MCA_ERROR_TYPE_CE:
368                 count = mca_funcs->max_ce_count;

mca_funcs is dereferenced here.

369                 break;
370         default:
371                 return -EINVAL;
372         }
373
374         if (idx >= count)
375                 return -EINVAL;
376
377         if (mca_funcs && mca_funcs->mca_get_mca_entry)
        ^^^^^^^^^

Checked too late!

Cc: Yang Wang <kevinyang.wang@amd.com>
Cc: Hawking Zhang <Hawking.Zhang@amd.com>
Cc: Christian König <christian.koenig@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Reviewed-by: Yang Wang <kevinyang.wang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add param to specify fw bo location for front-door loading

This param can help isolating data path issues on new systems in early phase.

Signed-off-by: Le Ma <le.ma@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Remove unreachable code in 'atom_skip_src_int()'

Fixes the below:
drivers/gpu/drm/amd/amdgpu/atom.c:398 atom_skip_src_int() warn: ignoring unreachable code.

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

drm/amd/display: Removing duplicate copyright text

mod_freesync header file has duplicated copyright boilerplate. Drop the
duplicate.

Signed-off-by: Marcelo Mendes Spessoto Junior <marcelomspessoto@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: skip gpu_info fw loading on navi12

It's no longer required.

Link: https://gitlab.freedesktop.org/drm/amd/-/issues/2318
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: add nv12 bounding box

This was included in gpu_info firmware, move it into the
driver for consistency with other nv1x parts.

Link: https://gitlab.freedesktop.org/drm/amd/-/issues/2318
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix sending VSC (+ colorimetry) packets for DP/eDP displays without PSR

The check for sending the vsc infopacket to the display was gated behind
PSR (Panel Self Refresh) being enabled.

The vsc infopacket also contains the colorimetry (specifically the
container color gamut) information for the stream on modern DP.

PSR is typically only supported on mobile phone eDP displays, thus this
was not getting sent for typical desktop monitors or TV screens.

This functionality is needed for proper HDR10 functionality on DP as it
wants BT2020 RGB/YCbCr for the container color space.

Cc: stable@vger.kernel.org
Cc: Harry Wentland <harry.wentland@amd.com>
Cc: Xaver Hugl <xaver.hugl@gmail.com>
Cc: Melissa Wen <mwen@igalia.com>
Fixes: 15f9dfd545a1 ("drm/amd/display: Register Colorspace property for DP and HDMI")
Tested-by: Simon Berz <simon@berz.me>
Tested-by: Xaver Hugl <xaver.hugl@kde.org>
Signed-off-by: Joshua Ashton <joshua@froggi.es>
Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Revert "drm/amdgpu: enable mca debug mode on APU by default"

Not needed any more with firmware fixes

Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Tao Zhou <tao.zhou1@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: Add mem_busy_percent for GCv9.4.3 apu

Expose sysfs entry mem_busy_percent for GC version
9.4.3 APU system

Signed-off-by: Asad Kamal <asad.kamal@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Merge tag 'drm-misc-next-fixes-2024-01-04' of git://anongit.freedesktop.org/drm/drm-misc into drm-next

One fix for drm/plane to avoid a use-after-free and some additional
warnings to prevent more of these occurences, a lock inversion
dependency fix and an indentation fix for drm/rockchip, and some doc
warning fixes for imagination and gpuvm.

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Maxime Ripard <mripard@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/enhl33v2oeihktta2yfyc4exvezdvm3eexcuwxkethc5ommrjo@lkidkv2kwakq

drm/amdkfd: Fix type of 'dbg_flags' in 'struct kfd_process'

dbg_flags looks to be defined with incorrect data type; to process
multiple debug flag options, and hence defined dbg_flags as u32.

Fixes the below:

drivers/gpu/drm/amd/amdgpu/../amdkfd/kfd_packet_manager_v9.c:117 pm_map_process_aldebaran() warn: maybe use && instead of &

Fixes: 0de4ec9a0353 ("drm/amdgpu: prepare map process for multi-process debug devices")
Suggested-by: Lijo Lazar <lijo.lazar@amd.com>
Cc: Felix Kuehling <Felix.Kuehling@amd.com>
Cc: Christian König <christian.koenig@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Fix possible NULL dereference in amdgpu_ras_query_error_status_helper()

Return invalid error code -EINVAL for invalid block id.

Fixes the below:

drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c:1183 amdgpu_ras_query_error_status_helper() error: we previously assumed 'info' could be null (see line 1176)

Suggested-by: Hawking Zhang <Hawking.Zhang@amd.com>
Cc: Tao Zhou <tao.zhou1@amd.com>
Cc: Hawking Zhang <Hawking.Zhang@amd.com>
Cc: Christian König <christian.koenig@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Drop redundant unsigned >=0 comparision 'amdgpu_gfx_rlc_init_microcode()'

unsigned int "version_minor" is always >= 0

Fixes the below:
drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.c:534 amdgpu_gfx_rlc_init_microcode() warn: always true condition '(version_minor >= 0) => (0-u16max >= 0)'

Cc: Christian König <christian.koenig@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Reviewed-by: Tao Zhou <tao.zhou1@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Adjust kdoc for 'dcn35_hw_block_power_down' & 'dcn35_hw_block_power_up'

Fixes the following gcc with W=1:

drivers/gpu/drm/amd/amdgpu/../display/dc/hwss/dcn35/dcn35_hwseq.c:1124: warning: This comment starts with '/**', but isn't a kernel-doc comment. Refer Documentation/doc-guide/kernel-doc.rst

Cc: Charlene Liu <charlene.liu@amd.com>
Cc: Muhammad Ahmed <ahmed.ahmed@amd.com>
Cc: Hamza Mahfooz <hamza.mahfooz@amd.com>
Cc: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Cc: Aurabindo Pillai <aurabindo.pillai@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Srinath Rao <srinath.rao@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Reviewed-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Address function parameter 'context' not described in 'dc_state_rem_all_planes_for_stream' & 'populate_subvp_cmd_drr_info'

Fixes the following gcc with W=1:

drivers/gpu/drm/amd/amdgpu/../display/dc/core/dc_state.c:524: warning: Function parameter or member 'state' not described in 'dc_state_rem_all_planes_for_stream'
drivers/gpu/drm/amd/amdgpu/../display/dc/core/dc_state.c:524: warning: Excess function parameter 'context' description in 'dc_state_rem_all_planes_for_stream'
drivers/gpu/drm/amd/amdgpu/../display/dc/dc_dmub_srv.c:540: warning: Function parameter or member 'context' not described in 'populate_subvp_cmd_drr_info'

Suggested-by: Aurabindo Pillai <aurabindo.pillai@amd.com>
Cc: Dillon Varone <dillon.varone@amd.com>
Cc: Jun Lei <jun.lei@amd.com>
Cc: Hamza Mahfooz <hamza.mahfooz@amd.com>
Cc: Aurabindo Pillai <aurabindo.pillai@amd.com>
Cc: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Srinath Rao <srinath.rao@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Reviewed-by: Aurabindo Pillai <aurabindo.pillai@amd.com>
Reviewed-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Use kvcalloc instead of kvmalloc_array in amdgpu_cs_parser_bos()

kvmalloc_array + __GFP_ZERO is the same with kvcalloc.

Fixes the below:
drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c:873 amdgpu_cs_parser_bos() warn: Please consider using kvcalloc instead of kvmalloc_array

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

drm/amdgpu: Use kzalloc instead of kmalloc+__GFP_ZERO in amdgpu_ras.c

Fixes the below smatch warnings:

drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c:2543 amdgpu_ras_recovery_init() warn: Please consider using kzalloc instead of kmalloc
drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c:2830 amdgpu_ras_init() warn: Please consider using kzalloc instead of kmalloc

Cc: Christian König <christian.koenig@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Reviewed-by: Tao Zhou <tao.zhou1@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Cleanup indenting in amdgpu_connector_dvi_detect()

drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c:1106 amdgpu_connector_dvi_detect() warn: inconsistent indenting

Fixes: 8a1de314d189 ("drm/amdgpu: Refactor 'amdgpu_connector_dvi_detect' in amdgpu_connectors.c")
Cc: Christian König <christian.koenig@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: "Pan, Xinhui" <Xinhui.Pan@amd.com>
Cc: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Cc: Aurabindo Pillai <aurabindo.pillai@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: Use gpu_metrics_v1_5 for SMUv13.0.6

Use gpu_metrics_v1_5 for SMUv13.0.6 to fill
gpu metric info

Signed-off-by: Asad Kamal <asad.kamal@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Reviewed-by: Le Ma <le.ma@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: Add gpu_metrics_v1_5

Add new gpu_metrics_v1_5 to acquire vcn/jpeg activity
& pcie nak error counters

Signed-off-by: Asad Kamal <asad.kamal@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Reviewed-by: Le Ma <le.ma@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: Update metric table for jpeg/vcn data

Update pmfw metric table to include vcn & jpeg
activity for smu_v_13_0_6

Signed-off-by: Asad Kamal <asad.kamal@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Reviewed-by: Le Ma <le.ma@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: Use separate metric table for APU

Use separate metric table for APU and Non APU
systems for smu_v_13_0_6 to get metric data

Signed-off-by: Asad Kamal <asad.kamal@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Reviewed-by: Le Ma <le.ma@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add some functions for Panel Replay

[WHY]
Prepare for enabling the Panel Replay feature

[HOW]
- Add some Panel Replay setting functions in DC
- Add the Panel Replay resource in dcn35_resource.c
- Add debug masks for Panel Replay

Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Acked-by: Rodrigo Siqueira <rodrigo.siqueira@amd.com>
Reviewed-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Tom Chung <chiahsuan.chung@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Re-revert "drm/amd/display: Enable Replay for static screen use cases"

This reverts commit 44e60b14d5a72f91fd0bdeae8da59ae37a3ca8e5.

Since, it causes a regression in which eDP displays with PSR support,
but no Replay support (Sink support <= 0x03), fail to enable PSR and
consequently all IGT amd_psr tests fail. So, revert this until a more
suitable fix can be found.

This got brought back accidently with the backmerge.

Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Ivan Lipski <ivlipski@amd.com>
Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fixing stream allocation regression

For certain dual display configs that had one display using a 1080p
mode, the DPM level used to drive the configs regressed from DPM 0 to
DPM 3. This was caused by a missing check that should have only limited
the pipe segments on non-phantom pipes. This caused issues with detile
buffer allocation, which dissallow subvp from being used

Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Reviewed-by: Dillon Varone <dillon.varone@amd.com>
Reviewed-by: Martin Leung <martin.leung@amd.com>
Acked-by: Rodrigo Siqueira <rodrigo.siqueira@amd.com>
Signed-off-by: Relja Vojvodic <relja.vojvodic@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Verify disallow bits were cleared for idle

[Why]
A hang was observed where a read-modify-write access occurred due to the
register for idle state being shared between DMCUB and driver.

dmcub read - idle allow / no commit
driver read - idle allow / no commit
driver write - idle disallow / no commit
dmcub write - idle allow / commit

Resulting in DMCUB re-entering IPS after a disable and keeping the allow
high.

[How]
Long term we need to split commit/allow into two registers or use shared
DRAM state, but short term we can reduce the repro rate by ensuring that
the disallow went through by bounding the expected worst case scenario.

Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Reviewed-by: Hansen Dsouza <hansen.dsouza@amd.com>
Reviewed-by: Ovidiu Bunea <ovidiu.bunea@amd.com>
Acked-by: Rodrigo Siqueira <rodrigo.siqueira@amd.com>
Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix Mismatch between pipe and stream

[Why]
Failing mode validation during dc_commit, leading to blackscreen with an
8k DP2 display during mode change.

[What]
Fix mixmatch between pipe and stream, which prevented us from
recognizing the link as DP2.

Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Reviewed-by: Dmytro Laktyushkin <dmytro.laktyushkin@amd.com>
Acked-by: Rodrigo Siqueira <rodrigo.siqueira@amd.com>
Signed-off-by: Gabe Teeger <gabe.teeger@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>