drm/amd/display: Set default brightness according to ACPI

Currently, amdgpu will always set up the brightness at 100% when it
loads.  However this is jarring when the BIOS has it previously
programmed to a much lower value.

The ACPI ATIF method includes two members for "ac_level" and "dc_level".
These represent the default values that should be used if the system is
brought up in AC and DC respectively.

Use these values to set up the default brightness when the backlight
device is registered.

v2: squash in ACPI fix

Reviewed-by: Leo Li <sunpeng.li@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: drop some kernel messages in VCN code

Similar to commit 813e7d4cd05e where some kernel log
messages are dropped. With this commit, more log
messages in older version of VCN/JPEG code are dropped.

Acked-by: Leo Liu <leo.liu@amd.com>
Signed-off-by: David (Ming Qiang) Wu <David.Wu3@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Revert "drm/amd/display: Add NULL check for 'afb' before dereferencing in amdgpu_dm_plane_handle_cursor_update"

[WHY]
This patch is a dupplicate implementation of 14bcf29b, which we
are reverting due to a regression with kms_plane_cursor IGT tests.

This reverts commit 38e6f715b02b572f74677eb2f29d3b4bc6f1ddff.

Reviewed-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Tested-by: George Zhang <George.zhang@amd.com>
Signed-off-by: Ivan Lipski <ivlipski@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Revert "drm/amd/display: Add null check for 'afb' in amdgpu_dm_update_cursor"

[WHY]
This commit introduced a regression with kms_plane_cursor IGT
tests, causing AMD Asics to fail these tests.

This reverts commit ce66ffd9813c8bd73e21b96d6476e5fb32b727c7.

Reviewed-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Tested-by: George Zhang <George.zhang@amd.com>
Signed-off-by: Ivan Lipski <ivlipski@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: refine gpu_info firmware loading

refine gpu_info firmware loading

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

drm/amd/display: prevent register access while in IPS

We can't read/write to DCN registers while in IPS. Since, that can cause
the system to hang. So, before proceeding with the access in that
scenario, force the system out of IPS.

Cc: stable@vger.kernel.org # 6.6+
Reviewed-by: Roman Li <roman.li@amd.com>
Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: enhance amdgpu_ucode_request() function flexibility

v1:
Adding formatting string feature to improve function flexibility.

v2:
modify macro name to ADMGPU_UCODE_MAX_NAME.

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

drm/amdgpu: fix the overflowed constant warning for RREG32_SOC15()

To fix potential overflowed constant warning reported by Coverity,
modify the variables to uint32_t.

Signed-off-by: Bob Zhou <bob.zhou@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add lock in amdgpu_gart_invalidate_tlb

We need to take the reset domain lock before flush hdp. We can't put the
lock inside amdgpu_device_flush_hdp itself because it is used during
reset where we already take the write side lock.

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

drm/amdgpu: fix locking scope when flushing tlb

Which method is used to flush tlb does not depend on whether a reset is
in progress or not. We should skip flush altogether if the GPU will get
reset. So put both path under reset_domain read lock.

Signed-off-by: Yunxiang Li <Yunxiang.Li@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
CC: stable@vger.kernel.org

drm/amdgpu: call flush_gpu_tlb directly in gfxhub enable

Here since we are in reset and takes the reset_domain write side lock
already. We can't use the flush tlb helper which tries to take the read
side.

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

drm/amdgpu: use helper in amdgpu_gart_unbind

When amdgpu_gart_invalidate_tlb helper is introduced this part was left
out of the conversion. Avoid the code duplication here.

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

drm/amdgpu: remove tlb flush in amdgpu_gtt_mgr_recover

At this point the gart is not set up, there's no point to invalidate tlb
here and it could even be harmful.

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

drm/amdgpu/kfd: remove is_hws_hang and is_resetting

is_hws_hang and is_resetting serves pretty much the same purpose and
they all duplicates the work of the reset_domain lock, just check that
directly instead. This also eliminate a few bugs listed below and get
rid of dqm->ops.pre_reset.

kfd_hws_hang did not need to avoid scheduling another reset. If the
on-going reset decided to skip GPU reset we have a bad time, otherwise
the extra reset will get cancelled anyway.

remove_queue_mes forgot to check is_resetting flag compared to the
pre-MES path unmap_queue_cpsch, so it did not block hw access during
reset correctly.

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

drm/amdgpu: fix sriov host flr handler

We send back the ready to reset message before we stop anything. This is
wrong. Move it to when we are actually ready for the FLR to happen.

In the current state since we take tens of seconds to stop everything,
it is very likely that host would give up waiting and reset the GPU
before we send ready, so it would be the same as before. But this gets
rid of the hack with reset_domain locking and also let us tell how slow
ready to reset actually is from the host. The ready to reset speed can
be improved later.

Signed-off-by: Yunxiang Li <Yunxiang.Li@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Emily Deng <Emily.Deng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add skip_hw_access checks for sriov

Accessing registers via host is missing the check for skip_hw_access and
the lockdep check that comes with it.

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

drm/amdgpu: add reset source in various cases

To fullfill the reset event description.

Suggested-by: Lijo Lazar <Lijo.Lazar@amd.com>
Signed-off-by: Eric Huang <jinhuieric.huang@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: fix NULL pointer in amdgpu_reset_get_desc

amdgpu_job_ring may return NULL, which causes kernel NULL
pointer error, using another way to print ring name instead
of ring->name.

Suggested-by: Lijo Lazar <Lijo.Lazar@amd.com>
Signed-off-by: Eric Huang <jinhuieric.huang@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: dc 3.2.287

This version brings the following changes:
- Add sequential ONO sequencing for DCN35
- Add new GPINT command definitions
- reduce ODM slice count to initial new dc state only when needed
- Enable copying of bounding box data from VBIOS DMUB
- Guard reading 3DLUT registers for dcn32/dcn35

Reviewed-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@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: Fix DSC slice and delay calculations

[why]
There are other factors that determine the number
of DSC slices. The slices should not be determined
in DML but retrieve the value calculated from driver.

[how]
Update the logic to determine DSC slice.
Make DSCDelay per display pipe.

Reviewed-by: Jun Lei <jun.lei@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Sung Joon Kim <sungjoon.kim@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Increase MAX_LINKS by 2

Two additional virtual links are created and thus increasing size
for dc->links by two.

Reviewed-by: Alvin Lee <alvin.lee2@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Alex Hung <alex.hung@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Guard reading 3DLUT registers for dcn32/dcn35

[Why]
3DLUT is not part of the DPP on DCN32/DCN35 ASIC and these registers
now exist in MCM state.

[How]
Add guards when reading DPP state based on whether the register has a
valid offset.

Reviewed-by: Sung joon Kim <sungjoon.kim@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Various DML2 fixes for FAMS2

- Ensure SubVP stream settings match ODM policy
- Fix MALL size calculations when DCC is enabled
- Fail if any stream fails DRR policy check

Reviewed-by: Chaitanya Dhere <chaitanya.dhere@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Dillon Varone <dillon.varone@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Program DIG FE source select for DVI before PHY en

[Description]
In newer DCN's the programming of SYMCLK_FE_SRC_SEL depends on
the value of DIG_FE_SOURCE_SELECT. If DIG_FE_SOURCE_SELECT is not
already programmed at the time of PHY / DIG enable then the FW
sequence will program an incorrect SYMCLK source. Ensure that we
program DIG_FE_SOURCE_SELECT for all DIO scenarios (DVI in this
particular case) before going through the PHY / DIG enable sequence.

Reviewed-by: Wenjing Liu <wenjing.liu@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Alvin Lee <alvin.lee2@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: remove dead code in navi10_emit_clk_levels and navi10_print_clk_levels

Since the range of the varibable i is 0 - 3.
So execution cannot reach this statement: default.

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

drm/amdgpu: remove dead code in atom_get_src_int

Since the range of align is 0~7, the expression is: align = (attr >> 3) & 7.
In the case of ATOM_ARG_IMM, the code cannot reach the default case.
So there is no need for "break".

Signed-off-by: Jesse Zhang <Jesse.Zhang@amd.com>
Suggested-by: Tim Huang <Tim.Huang@amd.com>
Reviewed-by: Tim Huang <Tim.Huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Introduce deferred Replay coasting vtotal update

Add functions to defer updating of coasting vtotal after source refresh rate update.

Reviewed-by: Robin Chen <robin.chen@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: ChunTao Tso <chuntao.tso@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add NULL check for 'afb' before dereferencing in amdgpu_dm_plane_handle_cursor_update

This commit adds a null check for the 'afb' variable in the
amdgpu_dm_plane_handle_cursor_update function. Previously, 'afb' was
assumed to be null, but was used later in the code without a null check.
This could potentially lead to a null pointer dereference.

Fixes the below:
drivers/gpu/drm/amd/amdgpu/../display/amdgpu_dm/amdgpu_dm_plane.c:1298 amdgpu_dm_plane_handle_cursor_update() error: we previously assumed 'afb' could be null (see line 1252)

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

drm/amd/display: Add null check for 'afb' in amdgpu_dm_update_cursor

This commit adds a null check for the 'afb' variable in the
amdgpu_dm_update_cursor function. Previously, 'afb' was assumed to be
null at line 8388, but was used later in the code without a null check.
This could potentially lead to a null pointer dereference.

Fixes the below:
drivers/gpu/drm/amd/amdgpu/../display/amdgpu_dm/amdgpu_dm.c:8433 amdgpu_dm_update_cursor()
error: we previously assumed 'afb' could be null (see line 8388)

drivers/gpu/drm/amd/amdgpu/../display/amdgpu_dm/amdgpu_dm.c
    8379 static void amdgpu_dm_update_cursor(struct drm_plane *plane,
    8380                                     struct drm_plane_state *old_plane_state,
    8381                                     struct dc_stream_update *update)
    8382 {
    8383         struct amdgpu_device *adev = drm_to_adev(plane->dev);
    8384         struct amdgpu_framebuffer *afb = to_amdgpu_framebuffer(plane->state->fb);
    8385         struct drm_crtc *crtc = afb ? plane->state->crtc : old_plane_state->crtc;
                                         ^^^^^

    8386         struct dm_crtc_state *crtc_state = crtc ? to_dm_crtc_state(crtc->state) : NULL;
    8387         struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
    8388         uint64_t address = afb ? afb->address : 0;
                                    ^^^^^ Checks for NULL

    8389         struct dc_cursor_position position = {0};
    8390         struct dc_cursor_attributes attributes;
    8391         int ret;
    8392
    8393         if (!plane->state->fb && !old_plane_state->fb)
    8394                 return;
    8395
    8396         drm_dbg_atomic(plane->dev, "crtc_id=%d with size %d to %d\n",
    8397                        amdgpu_crtc->crtc_id, plane->state->crtc_w,
    8398                        plane->state->crtc_h);
    8399
    8400         ret = amdgpu_dm_plane_get_cursor_position(plane, crtc, &position);
    8401         if (ret)
    8402                 return;
    8403
    8404         if (!position.enable) {
    8405                 /* turn off cursor */
    8406                 if (crtc_state && crtc_state->stream) {
    8407                         dc_stream_set_cursor_position(crtc_state->stream,
    8408                                                       &position);
    8409                         update->cursor_position = &crtc_state->stream->cursor_position;
    8410                 }
    8411                 return;
    8412         }
    8413
    8414         amdgpu_crtc->cursor_width = plane->state->crtc_w;
    8415         amdgpu_crtc->cursor_height = plane->state->crtc_h;
    8416
    8417         memset(&attributes, 0, sizeof(attributes));
    8418         attributes.address.high_part = upper_32_bits(address);
    8419         attributes.address.low_part  = lower_32_bits(address);
    8420         attributes.width             = plane->state->crtc_w;
    8421         attributes.height            = plane->state->crtc_h;
    8422         attributes.color_format      = CURSOR_MODE_COLOR_PRE_MULTIPLIED_ALPHA;
    8423         attributes.rotation_angle    = 0;
    8424         attributes.attribute_flags.value = 0;
    8425
    8426         /* Enable cursor degamma ROM on DCN3+ for implicit sRGB degamma in DRM
    8427          * legacy gamma setup.
    8428          */
    8429         if (crtc_state->cm_is_degamma_srgb &&
    8430             adev->dm.dc->caps.color.dpp.gamma_corr)
    8431                 attributes.attribute_flags.bits.ENABLE_CURSOR_DEGAMMA = 1;
    8432
--> 8433         attributes.pitch = afb->base.pitches[0] / afb->base.format->cpp[0];
                                    ^^^^^                  ^^^^^
Unchecked dereferences

    8434
    8435         if (crtc_state->stream) {
    8436                 if (!dc_stream_set_cursor_attributes(crtc_state->stream,
    8437                                                      &attributes))
    8438                         DRM_ERROR("DC failed to set cursor attributes\n");
    8439
    8440                 update->cursor_attributes = &crtc_state->stream->cursor_attributes;
    8441
    8442                 if (!dc_stream_set_cursor_position(crtc_state->stream,
    8443                                                    &position))
    8444                         DRM_ERROR("DC failed to set cursor position\n");
    8445
    8446                 update->cursor_position = &crtc_state->stream->cursor_position;
    8447         }
    8448 }

Fixes: 66eba12a5482 ("drm/amd/display: Do cursor programming with rest of pipe")
Reported-by: Dan Carpenter <dan.carpenter@linaro.org>
Cc: Tom Chung <chiahsuan.chung@amd.com>
Cc: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Cc: Roman Li <roman.li@amd.com>
Cc: Hersen Wu <hersenxs.wu@amd.com>
Cc: Alex Hung <alex.hung@amd.com>
Cc: Aurabindo Pillai <aurabindo.pillai@amd.com>
Cc: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add monitor patch skip disable crtc during psr and ips1

[Why]
For some panel, it cannot handle pseudo vblank set by otg resync
when leave psr

[How]
The monitor patch will keep otg_on during enter IPS1.
And then we don't need to do otg resync when wake up.

Reviewed-by: Duncan Ma <duncan.ma@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Lewis Huang <lewis.huang@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: add set ips disable

[How&Why]
Once IPS active, all the DCN resources are
not be allowed to access.
It needs to a function for 3rd party to
on/off IPS.

Reviewed-by: Duncan Ma <duncan.ma@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Chiawen Huang <chiawen.huang@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add recovery timeout to FAMS2

[WHY&HOW]
Add 5ms timeout to trigger recovery and force allow P-State in DMUB.

Reviewed-by: Alvin Lee <alvin.lee2@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Dillon Varone <dillon.varone@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Force max clocks unconditionally when p-state is unsupported

[WHY&HOW]
UCLK and FCLK are updated together, so an FCLK update can also cause UCLK update
to SMU.  When this happens, the UCLK provided should be max if switching is
unsupported.

Reviewed-by: Alvin Lee <alvin.lee2@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Dillon Varone <dillon.varone@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Change the order of setting DP_IS_USB_C flag

[Why]
enc10->base.features.flags.bits.DP_IS_USB_C will be overwritten if we set it
before initializing enc10->base.features

[How]
Determine DP_IS_USB_C after enc10->base.features is initialized. Besides,
bp_cap_info.DP_IS_USB_C will never be set in get_connector_speed_cap_info().
Remove the redudant code.

Reviewed-by: Roman Li <roman.li@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Wayne Lin <wayne.lin@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: bypass ODM before CRTC off

[WHY]
OPPs couldn't disconnect from the ODM that cause the double buffer pending not being latched due to missing VUPDATE.

[HOW]
Moving memory blanking before OTG turn off to make sure double buffer latched correctly.

Reviewed-by: Dmytro Laktyushkin <dmytro.laktyushkin@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Yihan Zhu <yihan.zhu@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display/dcn401: use pre-allocated temp structure for bounding box

This mirrors what the driver does for older DCN generations.

Should fix:
[   26.924055] BUG: sleeping function called from invalid context at include/linux/sched/mm.h:306
[   26.924060] in_atomic(): 1, irqs_disabled(): 0, non_block: 0, pid: 1022, name: modprobe
[   26.924063] preempt_count: 2, expected: 0
[   26.924064] RCU nest depth: 0, expected: 0
[   26.924066] Preemption disabled at:
[   26.924067] [<ffffffffc089e5e0>] dc_fpu_begin+0x30/0xd0 [amdgpu]
[   26.924322] CPU: 9 PID: 1022 Comm: modprobe Not tainted 6.8.0+ #20
[   26.924325] Hardware name: System manufacturer System Product Name/CROSSHAIR VI HERO, BIOS 6302 10/23/2018
[   26.924326] Call Trace:
[   26.924327]  <TASK>
[   26.924329]  dump_stack_lvl+0x37/0x50
[   26.924333]  ? dc_fpu_begin+0x30/0xd0 [amdgpu]
[   26.924589]  dump_stack+0x10/0x20
[   26.924592]  __might_resched+0x16a/0x1c0
[   26.924596]  __might_sleep+0x42/0x70
[   26.924598]  __kmalloc_node_track_caller+0x2ad/0x4b0
[   26.924601]  ? dm_helpers_allocate_gpu_mem+0x12/0x20 [amdgpu]
[   26.924855]  ? dcn401_update_bw_bounding_box+0x2a/0xf0 [amdgpu]
[   26.925122]  kmemdup+0x20/0x50
[   26.925124]  ? kernel_fpu_begin_mask+0x6b/0xe0
[   26.925127]  ? kmemdup+0x20/0x50
[   26.925129]  dcn401_update_bw_bounding_box+0x2a/0xf0 [amdgpu]
[   26.925393]  dc_create+0x311/0x670 [amdgpu]
[   26.925649]  amdgpu_dm_init+0x2aa/0x1fa0 [amdgpu]
[   26.925903]  ? irq_work_queue+0x38/0x50
[   26.925907]  ? vprintk_emit+0x1e7/0x270
[   26.925910]  ? dev_printk_emit+0x83/0xb0
[   26.925914]  ? amdgpu_device_rreg+0x17/0x20 [amdgpu]
[   26.926133]  dm_hw_init+0x14/0x30 [amdgpu]

v2: drop extra memcpy

Fixes: 669d6b078ed8 ("drm/amd/display: avoid large on-stack structures")
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Suggested-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: George Zhang <george.zhang@amd.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: harry.wentland@amd.com
Cc: sunpeng.li@amd.com
Cc: Rodrigo.Siqueira@amd.com

drm/amd/display: use pre-allocated temp structure for bounding box

This mirrors what the driver does for older DCN generations.

Should fix:

BUG: sleeping function called from invalid context at include/linux/sched/mm.h:306
in_atomic(): 1, irqs_disabled(): 0, non_block: 0, pid: 449, name: kworker/u64:8
preempt_count: 2, expected: 0
RCU nest depth: 0, expected: 0
Preemption disabled at:
ffffffffc0ce1580>] dc_fpu_begin+0x30/0xd0 [amdgpu]
CPU: 5 PID: 449 Comm: kworker/u64:8 Tainted: G        W          6.8.0+ #35
Hardware name: System manufacturer System Product Name/ROG STRIX X570-E GAMING WIFI II, BIOS 4204 02/24/2022
Workqueue: events_unbound async_run_entry_fn

v2: drop extra memcpy

Fixes: 88c61827cedc ("drm/amd/display: dynamically allocate dml2_configuration_options structures")
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Tested-by: George Zhang <George.zhang@amd.com> (v1)
Suggested-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: George Zhang <george.zhang@amd.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: harry.wentland@amd.com
Cc: sunpeng.li@amd.com
Cc: Rodrigo.Siqueira@amd.com

drm/amdgpu: add sdma 7.0 support for copy dcc buffer

1. Add dcc buffer flag for copy buffer
2. Add sdma 7.0 support copy dcc buffer

Signed-off-by: Likun Gao <Likun.Gao@amd.com>
Signed-off-by: Frank Min <Frank.Min@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: support for DCC feature

Deal with AMDGPU_GEM_CREATE_GFX12_DCC to set DCC bit
when needed.

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>

drm/amdgpu: add additional VM bits

Add additional VM PTE bits.

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

drm/amd/display: Simplify if conditions

The if conditions !A || A && B can be simplified to !A || B.

Fixes the following Coccinelle/coccicheck warnings reported by
excluded_middle.cocci:

WARNING !A || A && B is equivalent to !A || B
WARNING !A || A && B is equivalent to !A || B
WARNING !A || A && B is equivalent to !A || B

Compile-tested only.

Signed-off-by: Thorsten Blum <thorsten.blum@toblux.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Extend PSRSU residency mode

1. To support multiple PSRSU residency measurement mode

Reviewed-by: Wenjing Liu <wenjing.liu@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Jack Chang <jack.chang@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add outbox notification support for HPD redetect

[Why]
HPD sense changes can occur during low power states and need to be
notified from firmware to driver. Upon notification the hotplug
redetection routines should execute.

[How]
Add Support in DMUB srv and DMUB srv stat for receiving these
notifications. DM can hook them up and process the HPD redetection
once received.

Reviewed-by: Duncan Ma <duncan.ma@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add RAS is_rma flag

Set the flag to true if bad page number reaches threshold.

Signed-off-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: Add null checks for 'stream' and 'plane' before dereferencing

This commit adds null checks for the 'stream' and 'plane' variables in
the dcn30_apply_idle_power_optimizations function. These variables were
previously assumed to be null at line 922, but they were used later in
the code without checking if they were null. This could potentially lead
to a null pointer dereference, which would cause a crash.

The null checks ensure that 'stream' and 'plane' are not null before
they are used, preventing potential crashes.

Fixes the below static smatch checker:
drivers/gpu/drm/amd/amdgpu/../display/dc/hwss/dcn30/dcn30_hwseq.c:938 dcn30_apply_idle_power_optimizations() error: we previously assumed 'stream' could be null (see line 922)
drivers/gpu/drm/amd/amdgpu/../display/dc/hwss/dcn30/dcn30_hwseq.c:940 dcn30_apply_idle_power_optimizations() error: we previously assumed 'plane' could be null (see line 922)

Cc: Tom Chung <chiahsuan.chung@amd.com>
Cc: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Cc: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Cc: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Cc: Roman Li <roman.li@amd.com>
Cc: Hersen Wu <hersenxs.wu@amd.com>
Cc: Alex Hung <alex.hung@amd.com>
Cc: Aurabindo Pillai <aurabindo.pillai@amd.com>
Cc: Harry Wentland <harry.wentland@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/amd/pm: remove dead code in si_convert_power_level_to_smc

Since gmc_pg is false, setting mcFlags with SISLANDS_SMC_MC_PG_EN  cannot be reach.

Signed-off-by: Jesse Zhang <Jesse.Zhang@amd.com>
Suggested-by: Tim Huang <Tim.Huang@amd.com>
Reviewed-by: Tim Huang <Tim.Huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Prevent IPX From Link Detect and Set Mode

IPX involvment proven to affect LT, causing link loss. Need to prevent
IPX enabled in LT process in which link detect and set mode are main
procedures that have LT taken place.

Reviewed-by: Roman Li <roman.li@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Fangzhi Zuo <jerry.zuo@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: remove logically dead code

idr_for_each_entry can ensure that mem is not empty during the loop.
So don't need check mem again.

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

drm/amdgpu: fix failure mapping legacy queue when FLR

Flag "mes.ring.shced.ready" will be set as true after mes hw init and set
as false when mes hw fini to avoid duplicate initialization. But hw fini
will not be called when function level reset, which will cause mes hw
init be skipped during FLR, which will leads to mapping legacy queue
fail. Set this flag as false when post reset will fix this issue.

Signed-off-by: Lin.Cao <lincao12@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fetch Mall caps from DC

[Why]
When performing P-State switching with Subvp on 8k (downscaled to 4k).
corruption can be seen on the screen. MALL data was not being fetched
from DC, and the system things there is more MALL space then what is
actually available.

[How]
Read MALL size from dc caps.

Reviewed-by: Dillon Varone <dillon.varone@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Daniel Sa <daniel.sa@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix YUV video color corruption in DCN401

[Why]
Missing check causes sequence error which results in chroma
 filter coefficients not being updated in certain modes
 when we display YUV video in fullscreen.  This results in
 color corruption in video

[How]
Add back chroma_coef_mode check in dscl_set_scl_filter
 so that filter coefficients are calculated and updated when
 we have YUV surface

Reviewed-by: Alvin Lee <alvin.lee2@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Samson Tam <samson.tam@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: add reset cause in gpu pre-reset smi event

reset cause is requested by customer as additional
info for gpu reset smi event.

v2: integerate reset sources suggested by Lijo Lazar

Signed-off-by: Eric Huang <jinhuieric.huang@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Update soc24_enum.h and soc21_enum.h

Update to latest changes.

Signed-off-by: Frank Min <Frank.Min@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Set PTE_IS_PTE bit for gfx12

Set PTE_IS_PTE bit while PRT is enabled on gfx12.

Signed-off-by: Frank Min <Frank.Min@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Updated optc401_set_drr to use dcn401 functions

why:
optc_401_set_drr was using an old optc3 function to update vtotal min and max,
causing crashes when disabling FAMS2

how:
Updated dcn401 to point to opt401 function for vtotal updates. This version of
the function has FAMS2 logic that allows for FAMS2 to be disabled.

Reviewed-by: Dillon Varone <dillon.varone@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Relja Vojvodic <relja.vojvodic@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add reset sources in gpu reset context

reset source or reset cause is very useful info
for reset context, it will be used by events API.

Suggested-by: Lijo Lazar <Lijo.Lazar@amd.com>
Signed-off-by: Eric Huang <jinhuieric.huang@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add UCLK p-state support message to dcn401

[WHY&HOW]
Improves on the SMU interface to explicitly declare P-State support.

Reviewed-by: Alvin Lee <alvin.lee2@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Dillon Varone <dillon.varone@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Use current_state when checking old_pipe subvp type

[Description]
When checking the subvp type of the previous state we must
pass in current_state to the interface instead of context
otherwise we will get the wrong result.

Reviewed-by: Samson Tam <samson.tam@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Alvin Lee <alvin.lee2@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: add missing error handling in function smu_v13_0_6_allocate_dpm_context

Check return value to avoid null pointer dereference.

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

drm/amd/pm: Fix the null pointer dereference for vega10_hwmgr

Check return value and conduct null pointer handling to avoid null pointer dereference.

Signed-off-by: Bob Zhou <bob.zhou@amd.com>
Reviewed-by: Tim Huang <Tim.Huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Update the impelmentation of AMDGPU_PTE_MTYPE_VG10

This patch changes the implementation of AMDGPU_PTE_MTYPE_VG10,
clear the bits before setting the new one.

Suggested-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: longlyao <Longlong.Yao@amd.com>
Signed-off-by: Shane Xiao <shane.xiao@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: remove dead code in kfd_create_vcrat_image_gpu

kfd_create_vcrat_image_gpu itself checks the avail_size at the start.
So the value of avail_size is at least VCRAT_SIZE_FOR_GPU(16384),
minus struct crat_header(40UL) and struct crat_subtype_compute(40UL) it cannot be less than 0.

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

drm/amdkfd: fix the kdf debugger issue

The expression caps | HSA_CAP_TRAP_DEBUG_PRECISE_MEMORY_OPERATIONS_SUPPORTED
and  caps | HSA_CAP_TRAP_DEBUG_PRECISE_ALU_OPERATIONS_SUPPORTED
are always 1/true regardless of the values of its operand.

Fixes: 9243240bed38 ("drm/amdkfd: enable single alu ops for gfx12")
Signed-off-by: Jesse Zhang <Jesse.Zhang@amd.com>
Suggested-by: Felix Kuehling <felix.kuehling@amd.com>
Reviewed-by: Felix Kuehling <felix.kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Comment out the unused variable use_static in pm_map_queues_v9

To fix the warning about unused value,
remove the use_static and use the parameter is_static directly.

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

drm/amd/display: Move fpo_in_use to stream_status

[Description]
Refactor code and move fpo_in_use into stream_status to avoid
unexpected changes to previous dc_state (i.e., current_state).
Since stream pointers are shared between current and new dc_states,
updating parameters of one stream will update the other as well
which causes unexpected behaviors (i.e., checking that fpo_in_use
isn't set in previous state and set in the new state is invalid).
To avoid incorrect updates to current_state, move the fpo_in_use flag
into dc_stream_status since stream_status is owned by dc and are not
shared between different dc_states.

Reviewed-by: Samson Tam <samson.tam@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Alvin Lee <alvin.lee2@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Only program P-State force if pipe config changed

[Description]
Today for MED update type we do not call update clocks. However, for FPO
the assumption is that update clocks should be called to disable P-State
switch before any HW programming since FPO in FW and driver are not
synchronized. This causes an issue where on a MED update, an FPO P-State
switch could be taking place, then driver forces P-State disallow in the below
code and prevents FPO from completing the sequence. In this case we add a check
to avoid re-programming (and thus re-setting) the P-State force register by
only reprogramming if the pipe was not previously Subvp or FPO. The assumption
is that the P-State force register should be programmed correctly the first
time SubVP / FPO was enabled, so there's no need to update / reset it if the
pipe config has never exited SubVP / FPO.

Reviewed-by: Samson Tam <samson.tam@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Alvin Lee <alvin.lee2@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add retires when read DPCD

[why & how]
Sometimes read DPCD return fail while result not retrieved yet. Add
retries mechanism in Replay handle hpd irq to get real result.

Reviewed-by: Jerry Zuo <jerry.zuo@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Joan Lee <joan.lee@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix DML2 logic to set clk state to min

[Why]

When an eDP with high clock states is going into s0i3, stream_count is
0. This causes DML to not update the clks to the lowest state and
blocking us to enter s0i3 since eDP is out of vmin.

[How]

When stream_count is 0, set all the clocks to the lowest state.

Reviewed-by: Jun Lei <jun.lei@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Nicholas Susanto <nicholas.susanto@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Handle deallocated VPGRs in gfx11+ trap handler

A wavefront may deallocate its VGPRs at the end of a program while
waiting for memory transactions to complete. If it subsequently
receives a context save exception it will be unable to save,
since this requires VGPRs. In this case the trap handler should
terminate the wavefront.

Fixes intermittent VM faults under context switching load.

V2: Use S_ENDPGM instead of S_ENDPGM_SAVED for performance counters

Signed-off-by: Jay Cornwall <jay.cornwall@amd.com>
Reviewed-by: Lancelot Six <lancelot.six@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Support new VA page table block size

[Why]
Page table definition increased up to 2MB.

[How]
Define new use case of page table for VA.

Reviewed-by: Alvin Lee <alvin.lee2@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Chris Park <chris.park@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: workaround for oled eDP not lighting up on DCN401

[Why]
Currently there's an issue on DCN401 that prevents oled eDP panels from
being lit up that is still under investigation. To unblock dev work
while investigating, we can work around the issue by skipping toggling
the enablement of the backlight.

[How]
 - new debug bit that will skip touching backlight enable DPCD for oled

Reviewed-by: Alvin Lee <alvin.lee2@amd.com>
Reviewed-by: Wenjing Liu <wenjing.liu@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Joshua Aberback <joshua.aberback@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add params of set_abm_event for VB Scaling

[Why]
Add parameters for set_abm_event to enable varibright scaling.
VariBright Scaling is a feature to refer to system states like

1. Power mode
2. Battery Life percent
3. FullScreen video
4. Backlight slider

to adjust variBright strength to get low power or user experience.

[How]
Add parameters of set_abm_event for VB Scaling

Reviewed-by: Jun Lei <jun.lei@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Chun-LiangChang <chuchang@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix swapped dimension calculations

[Why]
The values calculated in optc1_get_otg_active_size are assigned to the
wrong output parameters, vertical blank is being used for horizontal size
and vice versa. This results in DPG test pattern looking wrong during
hardware init, as the DPG dimensions get assigned from this output, and
potentially other issues.

Reviewed-by: Aric Cyr <aric.cyr@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Joshua Aberback <joshua.aberback@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Wait for hardmins to complete on dcn401

[WHY&HOW]
When updating clocks via SMU, DAL needs to wait for requests to be fulfilled
before proceeding.

Reviewed-by: Alvin Lee <alvin.lee2@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Dillon Varone <dillon.varone@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: turn on symclk for dio virtual stream in dpms sequence

[why]
In order to support glitchless display clock ramping for virtual stream,
we must
turn on symclk for stream encoder. The code will power on phy and enable
symclk
for dio encoder during virtual stream dpms sequence.

Reviewed-by: Dillon Varone <dillon.varone@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Wenjing Liu <wenjing.liu@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Keep VBios pixel rate div setting until next mode set

[why]
Vbios & Driver have difference pixel rate div policy.
When enabling fast boot & performing blank & unblank w/o timing setting,
pixel clock & pixel rate dividor are not match.
It would cause too high pixel reate and eDP would be black screen.

[How]
We would keep pixel rate div setting by Vbios until next timing setting.

Reviewed-by: Jun Lei <jun.lei@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: yi-lchen <yi-lchen@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Revert "drm/amdgpu/gfx11: enable gfx pipe1 hardware support"

This reverts commit 6670142d25f3cc3166f2a6c8454acd310bf2776a.

Pierre-Eric reported problems with this on his navi33.  Revert
for now until we understand what is going wrong.

Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: Pierre-eric.Pelloux-prayer@amd.com

drm/amdgpu: update gc_12_0_0 headers

Add some additional registers.

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

drm/amdgpu: define new gfx12 uapi flags

define new gfx12 uapi flags

Signed-off-by: Marek Olšák <marek.olsak@amd.com>
Acked-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: remove dead code in the function svm_range_get_pte_flags

The varible uncached  set false, the condition uncached cannot be true.
So remove the dead code, mapping flags will set the flag AMDGPU_VM_MTYPE_UC in else.

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

drm/amdgpu: Update the impelmentation of AMDGPU_PTE_MTYPE_NV10

This patch changes the implementation of AMDGPU_PTE_MTYPE_NV10,
clear the bits before setting the new one.

Suggested-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: longlyao <Longlong.Yao@amd.com>
Signed-off-by: Shane Xiao <shane.xiao@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: disable lane0 L1TLB and enable lane1 L1TLB

This patch to disable lane0 L1TLB and enable lane1 L1TLB.

Signed-off-by: Yifan Zhang <yifan1.zhang@amd.com>
Reviewed-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: init SAW registers for mmhub v3.3

This patch to configure mmhub3.3 SAW registers

Signed-off-by: Yifan Zhang <yifan1.zhang@amd.com>
Reviewed-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/pptable: Fix UBSAN array-index-out-of-bounds

Flexible arrays used [1] instead of []. Replace the former with the latter
to resolve multiple UBSAN warnings observed on boot with a BONAIRE card.

In addition, use the __counted_by attribute where possible to hint the
length of the arrays to the compiler and any sanitizers.

Signed-off-by: Tasos Sahanidis <tasos@tasossah.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix a handful of spelling mistakes

There are a few spelling mistakes in dml2_printf messages. Fix them.

Reviewed-by: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Colin Ian King <colin.i.king@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: fix comments and error message for ipdump

Fix comments and error messages to rightly represent
the information.

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

drm/amdgpu: rename ip_dump_cp_queues to compute queues

Rename the variable ip_dump_cp_queues to ip_dump_compute_queue
as it represent compute queues.

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

drm/amdgpu: add cp queue registers for gfx9 ipdump

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

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

drm/amdgpu: add print support for gfx9 ipdump

Add support of gfx9 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 register support in ipdump

Add general registers of gfx9 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/amdgpu: Fix type mismatch in amdgpu_gfx_kiq_init_ring

This commit fixes a type mismatch in the amdgpu_gfx_kiq_init_ring
function triggered by the snprintf function expecting unsigned char
arguments due to the '%hhu' format specifier, but receiving int and u32
arguments.

The issue occurred because the arguments xcc_id, ring->me, ring->pipe,
and ring->queue were of type int and u32, not unsigned char. This led to
a type mismatch when these arguments were passed to snprintf.

To resolve this, the snprintf line was modified to cast these arguments
to unsigned char. This ensures that the arguments are of the correct
type for the '%hhu' format specifier and resolves the warning.

Fixes the below:
>> drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c:333:4: warning: format
>> specifies type 'unsigned char' but the argument has type 'int'
>> [-Wformat]
                    xcc_id, ring->me, ring->pipe, ring->queue);
                    ^~~~~~
>> drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c:333:12: warning: format
>> specifies type 'unsigned char' but the argument has type 'u32' (aka
>> 'unsigned int') [-Wformat]
                    xcc_id, ring->me, ring->pipe, ring->queue);
                            ^~~~~~~~
   drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c:333:22: warning: format specifies type 'unsigned char' but the argument has type 'u32' (aka 'unsigned int') [-Wformat]
                    xcc_id, ring->me, ring->pipe, ring->queue);
                                      ^~~~~~~~~~
   drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c:333:34: warning: format specifies type 'unsigned char' but the argument has type 'u32' (aka 'unsigned int') [-Wformat]
                    xcc_id, ring->me, ring->pipe, ring->queue);
                                                  ^~~~~~~~~~~
   4 warnings generated.

Fixes: 0ea554455542 ("drm/amdgpu: Fix snprintf usage in amdgpu_gfx_kiq_init_ring")
Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202405250446.XeaWe66u-lkp@intel.com/
Cc: Lijo Lazar <lijo.lazar@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Christian König <christian.koenig@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/amdgu: fix Unintentional integer overflow for mall size

Potentially overflowing expression mall_size_per_umc * adev->gmc.num_umc with type unsigned int (32 bits, unsigned)
is evaluated using 32-bit arithmetic,and then used in a context that expects an expression of type u64 (64 bits, unsigned).

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

drm/amdgpu: Update programming for boot error reporting

AMDGPU_RAS_GPU_ERR_BOOT_STATUS field is no longer valid.
The polling sequence is also simplifed according to
the latest firmware change.

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/radeon: Remove __counted_by from StateArray.states[]

Work for __counted_by on generic pointers in structures (not just
flexible array members) has started landing in Clang 19 (current tip of
tree). During the development of this feature, a restriction was added
to __counted_by to prevent the flexible array member's element type from
including a flexible array member itself such as:

  struct foo {
    int count;
    char buf[];
  };

  struct bar {
    int count;
    struct foo data[] __counted_by(count);
  };

because the size of data cannot be calculated with the standard array
size formula:

  sizeof(struct foo) * count

This restriction was downgraded to a warning but due to CONFIG_WERROR,
it can still break the build. The application of __counted_by on the
states member of 'struct _StateArray' triggers this restriction,
resulting in:

  drivers/gpu/drm/radeon/pptable.h:442:5: error: 'counted_by' should not be applied to an array with element of unknown size because 'ATOM_PPLIB_STATE_V2' (aka 'struct _ATOM_PPLIB_STATE_V2') is a struct type with a flexible array member. This will be an error in a future compiler version [-Werror,-Wbounds-safety-counted-by-elt-type-unknown-size]
    442 |     ATOM_PPLIB_STATE_V2 states[] __counted_by(ucNumEntries);
        |     ^~~~~~~~~~~~~~~~~~~~~~~~~~~~
  1 error generated.

Remove this use of __counted_by to fix the warning/error. However,
rather than remove it altogether, leave it commented, as it may be
possible to support this in future compiler releases.

Cc: stable@vger.kernel.org
Closes: https://github.com/ClangBuiltLinux/linux/issues/2028
Fixes: efade6fe50e7 ("drm/radeon: silence UBSAN warning (v3)")
Signed-off-by: Bill Wendling <morbo@google.com>
Co-developed-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/soc24: use common nbio callback to set remap offset

This fixes HDP flushes on systems with non-4K pages.

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

drm/amd/display: Convert some legacy DRM debug macros into appropriate categories

Currently when one enables driver debugging dmesg gets spammed, at I
suspect vblank rate, with messages like:

 [drm:amdgpu_dm_atomic_check [amdgpu]] MPO enablement requested on crtc:[00000000f073c3bb]

Fix if by converting some logging from deprecated and incorrect
DRM_DEBUG_DRIVER to drm_dbg_atomic. Plus some localized drive-by changes
to drm_dbg_kms.

By no means an exhaustive conversion but at least it allows turning on
driver debug selectively.

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@igalia.com>
Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Estimate RAS reservation when report capacity v2

Add estimate of how much vram we need to reserve for RAS
when caculating the total available vram.

v2: apply the change to MP0 v13_0_2 and v13_0_14

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: use u32 for buf size in __amdgpu_eeprom_xfer

And also make sure the value of msg[1].len should be in the range of u16.

Signed-off-by: Tao Zhou <tao.zhou1@amd.com>
Reviewed-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/amdkfd: gfx12 context save/restore trap handler fixes

Fix LDS size interpretation: 512 bytes (>= gfx12) vs 256 (< gfx12).

Ensure STATE_PRIV.BARRIER_COMPLETE cannot change after reading or
before writing. Other waves in the threadgroup may cause this field
to assert if they complete the barrier.

Do not overwrite EXCP_FLAG_PRIV.{SAVE_CONTEXT,HOST_TRAP} when
restoring this register. Both of these fields can assert while the
wavefront is running the trap handler.

Signed-off-by: Jay Cornwall <jay.cornwall@amd.com>
Reviewed-by: Lancelot Six <lancelot.six@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: drop some kernel messages in VCN code

We have messages when the VCN fails to initialize and
there is no need to report on success.
Also PSP loading FWs is the default for production.

Acked-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Sonny Jiang <sonjiang@amd.com>
Signed-off-by: David (Ming Qiang) Wu <David.Wu3@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>