This version brings along following fixes:
- correct unknown plane state patch
- Revert "Refactor DC update checks"
- Revert "Add 3DLUT DMA broadcast support"
- Remove invalid DPSTREAMCLK mask usage
- enable eDP DSC seamless boot support
- Revert "Rework HDMI link training and YCbCr422 with DSC policy"
- Disable PSR & Replay CRTC disable by default
- Fix Silence Compiler Warnings
- Add link output control for DPIA
- eliminate clock manager code duplication
- Don't set 4to1MPC config dynamically
- Merge pipes for validate
- Fix bounds checking in dml2_0 clock table array
- Avoid turning off the PHY when OTG is running for DVI
- Should support p-state under dcn21
- Enable Replay support for dcn42
- Remove check for DC_DMCUB_ENABLE on DCN42
Acked-by: Wayne Lin <Wayne.Lin@amd.com>
Signed-off-by: Taimur Hassan <Syed.Hassan@amd.com>
Signed-off-by: Chuanyu Tseng <chuanyu.tseng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
Compiler build generates type conversion warnings throughout dc/dml2_0
where values are implicitly narrowed (e.g. int/uint32_t/uint64_t assigned
to uint8_t, unsigned char, char, bool, or dml_bool_t), cluttering build
output and masking genuine issues.
[How]
Add explicit casts at each narrowing assignment with ASSERT guards
to catch out-of-range values in debug builds:
- uint8_t: otg_inst, num_planes, pipe_idx, vblank_index fields
- unsigned char: pipe_dlg_param.otg_inst from tg->inst
- char: mcache num_pipes from num_dpps_required
- bool/dml_bool_t: INTERLACE bitfield and fams2 enable flag use != 0
- uint64_t: widen min_hardware_refresh_in_uhz to hold div64_u64 result,
then cast to unsigned long for min_refresh_uhz with ASSERT
Reviewed-by: Austin Zheng <austin.zheng@amd.com>
Signed-off-by: Gaghik Khachatrian <gaghik.khachatrian@amd.com>
Signed-off-by: Chuanyu Tseng <chuanyu.tseng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
Resolve compiler warnings by marking unused parameters explicitly.
[How]
In .c and .h files, keep parameter names in signatures and add a
line with`(void)param;` inside the function body
Preserved function signatures and avoids breaking code paths that
may reference the parameter under conditional compilation.
Reviewed-by: Aric Cyr <aric.cyr@amd.com>
Signed-off-by: Gaghik Khachatrian <gaghik.khachatrian@amd.com>
Signed-off-by: Chuanyu Tseng <chuanyu.tseng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
Resolve compiler warnings by marking unused parameters explicitly.
[How]
In .c and .h function definitions, keep parameter names
in signatures and add a line with `(void)param;` in function body
Preserved function signatures and avoids breaking code paths that
may reference the parameter under conditional compilation.
Reviewed-by: Dillon Varone <dillon.varone@amd.com>
Reviewed-by: Austin Zheng <austin.zheng@amd.com>
Signed-off-by: Gaghik Khachatrian <gaghik.khachatrian@amd.com>
Signed-off-by: Chuanyu Tseng <chuanyu.tseng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
FPU guards (DC_FP_START/DC_FP_END) are required to wrap around code that
can manipulates floats. To do this properly, the FPU guards must be used
in a file that is not compiled as a FPU unit. If the guards are used in
a file that is a FPU unit, other sections in the file that aren't guarded
may be end up being compiled to use FPU operations.
[How]
Added DC_FP_START and DC_FP_END to DC functions that call DML functions
using FPU.
Reviewed-by: Dillon Varone <dillon.varone@amd.com>
Signed-off-by: Rafal Ostrowski <rafal.ostrowski@amd.com>
Signed-off-by: Alex Hung <alex.hung@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
FPU guards (DC_FP_START/DC_FP_END) are required to wrap around code that
can manipulates floats. To do this properly, the FPU guards must be used
in a file that is not compiled as a FPU unit. If the guards are used in
a file that is a FPU unit, other sections in the file that aren't guarded
may be end up being compiled to use FPU operations.
[How]
Removed DC_FP_START and DC_FP_END.
Reviewed-by: Dillon Varone <dillon.varone@amd.com>
Signed-off-by: Rafal Ostrowski <rafal.ostrowski@amd.com>
Signed-off-by: Alex Hung <alex.hung@amd.com>
Signed-off-by: Chuanyu Tseng <Chuanyu.Tseng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
FPU guards (DC_FP_START/DC_FP_END) are required to wrap around code that
can manipulates floats. To do this properly, the FPU guards must be used
in a file that is not compiled as a FPU unit. If the guards are used in
a file that is a FPU unit, other sections in the file that aren't guarded
may be end up being compiled to use FPU operations.
[How]
Added DC_FP_START and DC_FP_END to DC functions that call DML functions
using FPU.
Reviewed-by: Dillon Varone <dillon.varone@amd.com>
Signed-off-by: Rafal Ostrowski <rafal.ostrowski@amd.com>
Signed-off-by: Alex Hung <alex.hung@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
add support to query vram info from firmware
v2: change APU vram type, add multi-aid check
v3: seperate vram info query function into 3 parts and
call them in a helper func when requirements
are met.
v4: calculate vram_width for v9.x
Signed-off-by: Gangliang Xie <ganglxie@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Revert commit 7d59465de3 ("drm/amd/display: Add 3DLUT DMA broadcast support")
[WHY&HOW]
Dependencies of this change are still causing issues, so reverting until
those can be fixed.
Reviewed-by: Martin Leung <Martin.Leung@amd.com>
Signed-off-by: Dillon Varone <Dillon.Varone@amd.com>
Signed-off-by: Chuanyu Tseng <chuanyu.tseng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
The invalid register field access causes ASSERT(mask != 0) to fire
in set_reg_field_values() during display enable.
WARNING: at drivers/gpu/drm/amd/amdgpu/../display/dc/dc_helper.c:100
set_reg_field_values.isra.0+0xcf/0xf0 [amdgpu]
Call Trace:
<TASK>
generic_reg_update_ex+0x66/0x1d0 [amdgpu]
dccg401_set_dpstreamclk+0xed/0x350 [amdgpu]
dcn401_enable_stream+0x165/0x370 [amdgpu]
link_set_dpms_on+0x6e9/0xe90 [amdgpu]
dce110_apply_single_controller_ctx_to_hw+0x343/0x530 [amdgpu]
dce110_apply_ctx_to_hw+0x1f6/0x2d0 [amdgpu]
dc_commit_state_no_check+0x49a/0xe20 [amdgpu]
dc_commit_streams+0x354/0x570 [amdgpu]
amdgpu_dm_atomic_commit_tail+0x6f8/0x3fc0 [amdgpu]
DCN4.x hardware does not have DPSTREAMCLK_GATE_DISABLE and
DPSTREAMCLK_ROOT_GATE_DISABLE fields in DCCG_GATE_DISABLE_CNTL3.
These global fields only exist in DCN3.1.x hardware.
[How]
Remove the call that tries to update non-existent fields in CNTL3.
DCN4.x uses per-instance fields in CNTL5 instead,
which are already correctly programmed in the switch cases above.
Reviewed-by: Dillon Varone <dillon.varone@amd.com>
Signed-off-by: Roman Li <Roman.Li@amd.com>
Signed-off-by: Chuanyu Tseng <chuanyu.tseng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
To support specific sequencing requirements for DPIA link output
[How]
Implement the dpia_link_hwss structure and define the necessary
control function pointers. The initialization order is
aligned with the core link_hwss definition to ensure consistency
Reviewed-by: Wenjing Liu <wenjing.liu@amd.com>
Signed-off-by: Lincheng Ku <LinCheng.Ku@amd.com>
Signed-off-by: Chuanyu Tseng <chuanyu.tseng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
Implicit signed-to-unsigned conversions caused compiler
warnings in DC paths.
[How]
Added explicit (unsigned int)/(uint32_t) casts for sentinel -1
assignments and IRQ ~MASK initializers, with small cast alignment
in logging/DPCD code.
Functionality and behavior is unchanged; only type intent is explicit.
Reviewed-by: Dillon Varone <dillon.varone@amd.com>
Signed-off-by: Gaghik Khachatrian <gaghik.khachatrian@amd.com>
Signed-off-by: Chuanyu Tseng <chuanyu.tseng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
Clock manager contained significant duplicate code between
variants with identical logic for functions using only SMU
calls or shared registers. This increases maintenance overhead
and potential for bugs.
[How]
Expose clock constants and internal functions in header for
sharing. Remove duplicate implementations and update function
pointers to use shared functions. Refactor remaining
variant-specific functions to use shared constants and helper
functions. Add compatibility comments for hardware differences.
Reviewed-by: Dmytro Laktyushkin <dmytro.laktyushkin@amd.com>
Signed-off-by: Gabe Teeger <gabe.teeger@amd.com>
Signed-off-by: Chuanyu Tseng <chuanyu.tseng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Fix Conversion that might result in a loss of data warnings in dmub/src/:
- dmub_dcn20/31/32/35/42/60/401.c: Add ASSERT(value <= 0xFF) and
explicit (uint8_t) cast when storing REG_GET results into uint8_t
debug struct fields. Add != 0 for bool assignments from uint32_t
bitfield reads.
- dmub_reg.c: Cast va_arg shift value to uint8_t with ASSERT guard
before passing to set_reg_field_value_masks().
- dmub_srv.c: Widen num_pending to uint64_t to match uint64_t
arithmetic; use != 0 for bool assignments from unsigned expressions.
No functional change intended.
Reviewed-by: Dillon Varone <dillon.varone@amd.com>
Signed-off-by: Gaghik Khachatrian <gaghik.khachatrian@amd.com>
Signed-off-by: Chuanyu Tseng <chuanyu.tseng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
We were previously modifying the global dc->config.enable_4to1MPC
dynamically. These variables are meant as global configs, not to
by dynamically modified. Modifying them dynamically prevents us
from enabling/disabling functionality for debug purposes and can
easily lead to bad things since we're not operating on the current
state but on DC-wide variables.
Instead we should look at the existing split4mpc decision in
dcn20_validate_apply_split_flags and make the decision there,
if the global config.enable_4to1MPC is set to true for the
DCN version we're running.
This fixes corruption that is observed when running a new IGT
kms_colorop test for color-space-conversion that uses a
YUV plane and outputs to a writeback connector.
Co-developed by Claude Sonnet 4.5.
Assisted-by: Claude:claude-sonnet-4.5
Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Chuanyu Tseng <chuanyu.tseng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
dcn401_init_hw() assumes that update_bw_bounding_box() is valid when
entering the update path. However, the existing condition:
((!fams2_enable && update_bw_bounding_box) || freq_changed)
does not guarantee this, as the freq_changed branch can evaluate to true
independently of the callback pointer.
This can result in calling update_bw_bounding_box() when it is NULL.
Fix this by separating the update condition from the pointer checks and
ensuring the callback, dc->clk_mgr, and bw_params are validated before
use.
Fixes the below:
../dc/hwss/dcn401/dcn401_hwseq.c:367 dcn401_init_hw() error: we previously assumed 'dc->res_pool->funcs->update_bw_bounding_box' could be null (see line 362)
Fixes: ca0fb243c3 ("drm/amd/display: Underflow Seen on DCN401 eGPU")
Cc: Daniel Sa <Daniel.Sa@amd.com>
Cc: Alvin Lee <alvin.lee2@amd.com>
Cc: Roman Li <roman.li@amd.com>
Cc: Alex Hung <alex.hung@amd.com>
Cc: Tom Chung <chiahsuan.chung@amd.com>
Cc: Dan Carpenter <dan.carpenter@linaro.org>
Cc: Aurabindo Pillai <aurabindo.pillai@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Reviewed-by: Alex Hung <alex.hung@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Validation expects to operate on non-split pipes. This is
seen in dcn20_fast_validate_bw, which merges pipes for
validation. We weren't doing that in the non-fast path
which lead to validation failures when operating with
4-to-1 MPC and a writeback connector.
Co-developed by Claude Sonnet 4.5
Assisted-by: Claude:claude-sonnet-4.5
Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Chuanyu Tseng <chuanyu.tseng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
clk_mgr_construct() initializes display clock and memory bandwidth
settings during driver bring-up.
As part of this, the driver selects a watermark table based on the
memory type (DDR4, LPDDR4, LPDDR5) from ctx->dc_bios->integrated_info.
The display pipeline continuously reads pixel data from memory,
processes it (such as scaling, color conversion, and blending), and
sends it to the screen. To keep this pipeline running smoothly, the
driver must ensure there is enough memory bandwidth and that clocks are
increased when needed.
Watermark tables define when the GPU should increase clocks to ensure
there is enough bandwidth to feed pixel data without underflow.
However, ctx->dc_bios->integrated_info is dereferenced without checking
for NULL in multiple clk_mgr_construct() implementations. On some
platforms, BIOS may not provide this information, and accessing it
directly can cause a NULL pointer dereference during initialization.
Fix this by adding a NULL check before accessing integrated_info.
If integrated_info is not available, the driver safely falls back to
default watermark tables.
Fixes:
../dcn21/rn_clk_mgr.c:775 rn_clk_mgr_construct() warn: variable dereferenced before check 'ctx->dc_bios->integrated_info' (see line 743)
../dcn301/vg_clk_mgr.c:750 vg_clk_mgr_construct() warn: variable dereferenced before check 'ctx->dc_bios->integrated_info' (see line 736)
../dcn31/dcn31_clk_mgr.c:789 dcn31_clk_mgr_construct() warn: variable dereferenced before check 'ctx->dc_bios->integrated_info' (see line 728)
../dcn314/dcn314_clk_mgr.c:906 dcn314_clk_mgr_construct() warn: variable dereferenced before check 'ctx->dc_bios->integrated_info' (see line 845)
../dcn315/dcn315_clk_mgr.c:716 dcn315_clk_mgr_construct() warn: variable dereferenced before check 'ctx->dc_bios->integrated_info' (see line 655)
../dcn316/dcn316_clk_mgr.c:660 dcn316_clk_mgr_construct() warn: variable dereferenced before check 'ctx->dc_bios->integrated_info' (see line 639)
../dcn35/dcn35_clk_mgr.c:1540 dcn35_clk_mgr_construct() warn: variable dereferenced before check 'ctx->dc_bios->integrated_info' (see line 1467)
Fixes: 25879d7b49 ("drm/amd/display: Clean FPGA code in dc")
Cc: Roman Li <roman.li@amd.com>
Cc: Alex Hung <alex.hung@amd.com>
Cc: Tom Chung <chiahsuan.chung@amd.com>
Cc: Dan Carpenter <dan.carpenter@linaro.org>
Cc: Aurabindo Pillai <aurabindo.pillai@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Reviewed-by: Alex Hung <alex.hung@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
In dc_dmub_srv_log_diagnostic_data() and
dc_dmub_srv_enable_dpia_trace().
Both functions check:
if (!dc_dmub_srv || !dc_dmub_srv->dmub)
and then call DC_LOG_ERROR() inside that block.
DC_LOG_ERROR() uses dc_dmub_srv->ctx internally. So if
dc_dmub_srv is NULL, the logging itself can dereference a
NULL pointer and cause a crash.
Fix this by splitting the checks.
First check if dc_dmub_srv is NULL and return immediately.
Then check dc_dmub_srv->dmub and log the error only when
dc_dmub_srv is valid.
Fixes the below:
../display/dc/dc_dmub_srv.c:962 dc_dmub_srv_log_diagnostic_data() error: we previously assumed 'dc_dmub_srv' could be null (see line 961)
../display/dc/dc_dmub_srv.c:1167 dc_dmub_srv_enable_dpia_trace() error: we previously assumed 'dc_dmub_srv' could be null (see line 1166)
Fixes: 2631ac1ac3 ("drm/amd/display: add DMUB registers to crash dump diagnostic data.")
Fixes: 71ba6b577a ("drm/amd/display: Add interface to enable DPIA trace")
Cc: Roman Li <roman.li@amd.com>
Cc: Alex Hung <alex.hung@amd.com>
Cc: Tom Chung <chiahsuan.chung@amd.com>
Cc: Dan Carpenter <dan.carpenter@linaro.org>
Cc: Aurabindo Pillai <aurabindo.pillai@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Reviewed-by: Alex Hung <alex.hung@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
Multiple locations in dml2_0 used num_clk_values-1 as array index
without checking if num_clk_values > 0. When num_clk_values is 0,
this results in accessing array index -1, which wraps to 255 for
unsigned types, causing out-of-bounds memory access and potential
crashes.
[How]
Add proper bounds checking using ternary operators to guard all
num_clk_values-1 array accesses. When num_clk_values is 0, return 0
as fallback value instead of accessing invalid memory. This prevents
buffer overflows while maintaining backward compatibility and provides
sensible default behavior for empty clock tables.
Reviewed-by: Dillon Varone <dillon.varone@amd.com>
Signed-off-by: Gabe Teeger <gabe.teeger@amd.com>
Signed-off-by: Chuanyu Tseng <chuanyu.tseng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
The OTG's virtual pixel clock source for DVI comes from the PHY.
If the signal type is DVI then the OTG can become stuck on pre DCN401
ASIC when DPMS off occurs because the OTG remains running but the
PHY transmitter is disabled.
[How]
There exists logic to keep track of the OTG running refcount on the
link to determine if the link needs to go to PLL_EN instead of TX_EN
but the logic only checks for HDMI TMDS on older ASIC.
DVI is still a TMDS signal type so the constraint should also apply.
Replace the checks for dc_is_hdmi_tmds_signal with dc_is_tmds_signal to
cover both HDMI and DVI for the symclk refcount workaround.
Reviewed-by: Dillon Varone <dillon.varone@amd.com>
Reviewed-by: Charlene Liu <charlene.liu@amd.com>
Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Signed-off-by: Chuanyu Tseng <chuanyu.tseng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
Under DCN21, observe flip_done timeout issue while
running 3D benchmark under MPO case. Timeout is caused
by driver fails validate_bandwidth() during
atomic_commit_tail but passes atomic_check.
Under further analysis, indicates the delta of
atomic_check and atomic_commit_tail are
dc->current_state->bw_ctx.dml.soc.sr_exit_time_us and
dc->current_state->bw_ctx.dml.soc.sr_enter_plus_exit_time_us.
We set validate_mode as DC_VALIDATE_MODE_ONLY while calling
dc_validate_global_state() at atomic_check, but set mode as
DC_VALIDATE_MODE_AND_PROGRAMMING during atomic_commit_tail.
If dc_validate_mode set as DC_VALIDATE_MODE_ONLY,
validate_bandwidth() will skip the wm and dlg calculation.
During commit_tail, validate_bandwidth() is called with
dc_validate_mode set as DC_VALIDATE_MODE_AND_PROGRAMMING and
dc_state->bw_ctx.dml.soc.sr_exit_time_us might get modified
after the wm_calculation and stored into dc->current_state.
Which means dc->current_state->bw_ctx.dml.soc.sr_exit_time_us
might not aligned with the one stored in dm_state->context.
That causes duplicated dm_state->context not aligned with
dc->current_state, and might have bandwidth validation pass
in atomic_check and fail in commit_tail later.
[How]
When the issue occurs, it fails dml_get_voltage_level() with
the condition dm_allow_self_refresh_and_mclk_switch but pass
with the condition dm_allow_self_refresh. However, we should
support p-state. So we should not pass validate_bandwidth by
allowing self refresh only. Change the policy under DCN21.
Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Signed-off-by: Wayne Lin <Wayne.Lin@amd.com>
Signed-off-by: Chuanyu Tseng <chuanyu.tseng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
get_gpio_i2c_info() computes the number of GPIO I2C assignment records
present in the BIOS table and then uses bfI2C_LineMux as an array index
into header->asGPIO_Info[]. The current check only rejects values
strictly larger than the record count, so an index equal to count still
falls through and reaches the fixed table one element past the end.
Reject indices at or above the number of available records before using
them as an array index.
Signed-off-by: Pengpeng Hou <pengpeng@iscas.ac.cn>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Currently, AMDGPU_VA_RESERVED_TRAP_SIZE is hardcoded to 8KB, while
KFD_CWSR_TBA_TMA_SIZE is defined as 2 * PAGE_SIZE. On systems with
4K pages, both values match (8KB), so allocation and reserved space
are consistent.
However, on 64K page-size systems, KFD_CWSR_TBA_TMA_SIZE becomes 128KB,
while the reserved trap area remains 8KB. This mismatch causes the
kernel to crash when running rocminfo or rccl unit tests.
Kernel attempted to read user page (2) - exploit attempt? (uid: 1001)
BUG: Kernel NULL pointer dereference on read at 0x00000002
Faulting instruction address: 0xc0000000002c8a64
Oops: Kernel access of bad area, sig: 11 [#1]
LE PAGE_SIZE=64K MMU=Radix SMP NR_CPUS=2048 NUMA pSeries
CPU: 34 UID: 1001 PID: 9379 Comm: rocminfo Tainted: G E
6.19.0-rc4-amdgpu-00320-gf23176405700 #56 VOLUNTARY
Tainted: [E]=UNSIGNED_MODULE
Hardware name: IBM,9105-42A POWER10 (architected) 0x800200 0xf000006
of:IBM,FW1060.30 (ML1060_896) hv:phyp pSeries
NIP: c0000000002c8a64 LR: c00000000125dbc8 CTR: c00000000125e730
REGS: c0000001e0957580 TRAP: 0300 Tainted: G E
MSR: 8000000000009033 <SF,EE,ME,IR,DR,RI,LE> CR: 24008268
XER: 00000036
CFAR: c00000000125dbc4 DAR: 0000000000000002 DSISR: 40000000
IRQMASK: 1
GPR00: c00000000125d908 c0000001e0957820 c0000000016e8100
c00000013d814540
GPR04: 0000000000000002 c00000013d814550 0000000000000045
0000000000000000
GPR08: c00000013444d000 c00000013d814538 c00000013d814538
0000000084002268
GPR12: c00000000125e730 c000007e2ffd5f00 ffffffffffffffff
0000000000020000
GPR16: 0000000000000000 0000000000000002 c00000015f653000
0000000000000000
GPR20: c000000138662400 c00000013d814540 0000000000000000
c00000013d814500
GPR24: 0000000000000000 0000000000000002 c0000001e0957888
c0000001e0957878
GPR28: c00000013d814548 0000000000000000 c00000013d814540
c0000001e0957888
NIP [c0000000002c8a64] __mutex_add_waiter+0x24/0xc0
LR [c00000000125dbc8] __mutex_lock.constprop.0+0x318/0xd00
Call Trace:
0xc0000001e0957890 (unreliable)
__mutex_lock.constprop.0+0x58/0xd00
amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu+0x6fc/0xb60 [amdgpu]
kfd_process_alloc_gpuvm+0x54/0x1f0 [amdgpu]
kfd_process_device_init_cwsr_dgpu+0xa4/0x1a0 [amdgpu]
kfd_process_device_init_vm+0xd8/0x2e0 [amdgpu]
kfd_ioctl_acquire_vm+0xd0/0x130 [amdgpu]
kfd_ioctl+0x514/0x670 [amdgpu]
sys_ioctl+0x134/0x180
system_call_exception+0x114/0x300
system_call_vectored_common+0x15c/0x2ec
This patch changes AMDGPU_VA_RESERVED_TRAP_SIZE to 64 KB and
KFD_CWSR_TBA_TMA_SIZE to the AMD GPU page size. This means we reserve
64 KB for the trap in the address space, but only allocate 8 KB within
it. With this approach, the allocation size never exceeds the reserved
area.
Fixes: 34a1de0f79 ("drm/amdkfd: Relocate TBA/TMA to opposite side of VM hole")
Reviewed-by: Christian König <christian.koenig@amd.com>
Suggested-by: Felix Kuehling <felix.kuehling@amd.com>
Suggested-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Donet Tom <donettom@linux.ibm.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Fix the code alignment for if condition and also provide
a line space between multiline if condition and next
statement.
Signed-off-by: Sunil Khatri <sunil.khatri@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
In mes_userq_mqd_create(), the memdup_user() allocations for
IP-specific MQD structs are not freed when subsequent VA validation
fails. The goto free_mqd label only cleans up the MQD BO object and
userq_props.
Fix by adding kfree() before each goto free_mqd on VA validation
failure in the COMPUTE, GFX, and SDMA branches.
Fixes: 9e46b8bb05 ("drm/amdgpu: validate userq buffer virtual address and size")
Reported-by: Yuhao Jiang <danisjiang@gmail.com>
Signed-off-by: Junrui Luo <moonafterrain@outlook.com>
Reviewed-by: Prike Liang <Prike.Liang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
For gfxV9, due to a hardware bug ("based on the comments in the code
here [1]"), the control stack of a user-mode compute queue must be
allocated immediately after the page boundary of its regular MQD buffer.
To handle this, we allocate an enlarged MQD buffer where the first page
is used as the MQD and the remaining pages store the control stack.
Although these regions share the same BO, they require different memory
types: the MQD must be UC (uncached), while the control stack must be
NC (non-coherent), matching the behavior when the control stack is
allocated in user space.
This logic works correctly on systems where the CPU page size matches
the GPU page size (4K). However, the current implementation aligns both
the MQD and the control stack to the CPU PAGE_SIZE. On systems with a
larger CPU page size, the entire first CPU page is marked UC—even though
that page may contain multiple GPU pages. The GPU treats the second 4K
GPU page inside that CPU page as part of the control stack, but it is
incorrectly mapped as UC.
This patch fixes the issue by aligning both the MQD and control stack
sizes to the GPU page size (4K). The first 4K page is correctly marked
as UC for the MQD, and the remaining GPU pages are marked NC for the
control stack. This ensures proper memory type assignment on systems
with larger CPU page sizes.
[1]: https://elixir.bootlin.com/linux/v6.18/source/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v9.c#L118
Acked-by: Felix Kuehling <felix.kuehling@amd.com>
Signed-off-by: Donet Tom <donettom@linux.ibm.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The AQL queue size can be 4K, but the minimum buffer object (BO)
allocation size is PAGE_SIZE. On systems with a page size larger
than 4K, the expected queue size does not match the allocated BO
size, causing queue creation to fail.
Align the expected queue size to PAGE_SIZE so that it matches the
allocated BO size and allows queue creation to succeed.
Reviewed-by: Felix Kuehling <felix.kuehling@amd.com>
Signed-off-by: Donet Tom <donettom@linux.ibm.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
In amdgpu_device_fini_hw(), deferred coredump formatting work may still
be pending when hardware and IP components are being torn down. Since
the work may access device registers and memory that will be freed or
powered off, it must be completed before proceeding.
Add a flush_work() call for adev->coredump_work, guarded by
CONFIG_DEV_COREDUMP, to ensure any pending coredump work finishes
before the device enters the early IP fini stage.
This avoids potential use-after-free or accessing hardware resources
that are no longer available.
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Suggested-by: Lijo Lazar <lijo.lazar@amd.com>
Signed-off-by: Jesse Zhang <jesse.zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
amdgpu_userq_vm_validate function does not need userq_mutex and exec
lock is good enough to locking all bos and updating the eviction fence.
Also since we only need userq_mutex for amdgpu_userq_restore_all
so move the locks in the function itself.
Signed-off-by: Sunil Khatri <sunil.khatri@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
amdgpu_userq_get_doorbell_index() passes the user-provided
doorbell_offset to amdgpu_doorbell_index_on_bar() without bounds
checking. An arbitrarily large doorbell_offset can cause the
calculated doorbell index to fall outside the allocated doorbell BO,
potentially corrupting kernel doorbell space.
Validate that doorbell_offset falls within the doorbell BO before
computing the BAR index, using u64 arithmetic to prevent overflow.
Fixes: f09c1e6077 ("drm/amdgpu: generate doorbell index for userqueue")
Reported-by: Yuhao Jiang <danisjiang@gmail.com>
Signed-off-by: Junrui Luo <moonafterrain@outlook.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
