Fetching from platform ROM doesn't work with hybrid ROM images. For
default VGA devices also prefer ROM BAR.
Signed-off-by: Lijo Lazar <lijo.lazar@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
GART TLB is flushed after unmapping but not after mapping. Since
amdgpu_bo_create_kernel() does not zero-initialize the buffer, when a
single PTE is written the TLB may speculatively load other uninitialized
entries from the same cacheline. Those garbage entries can appear valid,
and a subsequent write to another PTE in the same cacheline may cause the
GPU to use a stale garbage PTE from the TLB.
Fix this by calling memset_io() to zero-initialize the GART table with
gart_pte_flags immediately after allocation.
Using AMDGPU_GEM_CREATE_VRAM_CLEARED, SDMA-based clear will not work
since SDMA needs GART to be initialized to work.
Suggested-by: Felix Kuehling <felix.kuehling@amd.com>
Signed-off-by: Philip Yang <Philip.Yang@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
sdma_v4_0_ring_emit_fence() contains two BUG_ON(addr & 0x3) assertions
that verify fence writeback addresses are dword-aligned. These
assertions can be reached from unprivileged userspace via crafted
DRM_IOCTL_AMDGPU_CS submissions, causing a fatal kernel panic in a
scheduler worker thread.
Replace both BUG_ON() calls with WARN_ON() to log the condition without
crashing the kernel. A misaligned fence address at this point indicates
a driver bug, but crashing the kernel is never the correct response when
the assertion is reachable from userspace.
The CS IOCTL path is the correct place to filter invalid submissions;
the ring emission callback is too late to do anything about it.
Fixes: 2130f89ced ("drm/amdgpu: add SDMA v4.0 implementation (v2)")
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: John B. Moore <jbmoore61@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Remove the BUG_ON(flags & AMDGPU_FENCE_FLAG_64BIT) assertion from
gfx_v9_0_ring_emit_fence_kiq(). The KIQ hardware supports 64-bit
fence writes; the 32-bit writeback address constraint is an
upper-layer convention, not a hardware limitation. The check serves
no purpose and should not be present.
Found by code inspection while investigating related BUG_ON
assertions in the GFX and compute ring emission paths.
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: John B. Moore <jbmoore61@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The ring preemption function is identical for both gfx_v11_0 and
gfx_v12_0. This patch refactors the code by moving the core logic
into a generic function inside amdgpu_gfx.c to reduce code
duplication and simplify future maintenance.
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Leonardo Cesar <leonardocesar@usp.br>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
With only one sequence number we cannot track the need for legacy vs
heavy-weight flushes reliably. Always use heavy-weight.
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>
Fix all kernel-doc warnings in amdgpu.h and amdgpu_reset.h:
- Use the struct keyword for kernel-doc struct comments.
- Use the correct enum names in enum amd_reset_method.
This eliminates these warnings:
Warning: drivers/gpu/drm/amd/amdgpu/amdgpu.h:477 cannot understand
function prototype: 'struct amdgpu_wb'
Warning: drivers/gpu/drm/amd/amdgpu/amdgpu.h:576 Enum value
'AMD_RESET_METHOD_LEGACY' not described in enum 'amd_reset_method'
Warning: drivers/gpu/drm/amd/amdgpu/amdgpu.h:576 Enum value
'AMD_RESET_METHOD_MODE0' not described in enum 'amd_reset_method'
Warning: drivers/gpu/drm/amd/amdgpu/amdgpu.h:576 Enum value
'AMD_RESET_METHOD_MODE1' not described in enum 'amd_reset_method'
Warning: drivers/gpu/drm/amd/amdgpu/amdgpu.h:576 Enum value
'AMD_RESET_METHOD_MODE2' not described in enum 'amd_reset_method'
Warning: drivers/gpu/drm/amd/amdgpu/amdgpu.h:576 Enum value
'AMD_RESET_METHOD_LINK' not described in enum 'amd_reset_method'
Warning: drivers/gpu/drm/amd/amdgpu/amdgpu.h:576 Enum value
'AMD_RESET_METHOD_BACO' not described in enum 'amd_reset_method'
Warning: drivers/gpu/drm/amd/amdgpu/amdgpu.h:576 Enum value
'AMD_RESET_METHOD_PCI' not described in enum 'amd_reset_method'
Warning: drivers/gpu/drm/amd/amdgpu/amdgpu.h:576 Enum value
'AMD_RESET_METHOD_ON_INIT' not described in enum 'amd_reset_method'
Warning: drivers/gpu/drm/amd/amdgpu/amdgpu.h:576 Excess enum value
'@AMD_RESET_LEGACY' description in 'amd_reset_method'
Warning: drivers/gpu/drm/amd/amdgpu/amdgpu.h:576 Excess enum value
'@AMD_RESET_MODE0' description in 'amd_reset_method'
Warning: drivers/gpu/drm/amd/amdgpu/amdgpu.h:576 Excess enum value
'@AMD_RESET_MODE1' description in 'amd_reset_method'
Warning: drivers/gpu/drm/amd/amdgpu/amdgpu.h:576 Excess enum value
'@AMD_RESET_MODE2' description in 'amd_reset_method'
Warning: drivers/gpu/drm/amd/amdgpu/amdgpu.h:576 Excess enum value
'@AMD_RESET_LINK' description in 'amd_reset_method'
Warning: drivers/gpu/drm/amd/amdgpu/amdgpu.h:576 Excess enum value
'@AMD_RESET_BACO' description in 'amd_reset_method'
Warning: drivers/gpu/drm/amd/amdgpu/amdgpu.h:576 Excess enum value
'@AMD_RESET_PCI' description in 'amd_reset_method'
Also move the enum to amdgpu_reset.h and eventually only forward declare
it in amdgpu.h. (Christian)
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
For enums that have a kernel-doc ("/**") comment block, add or correct
their kernel-doc to eliminate all kernel-doc enum warnings.
Some of these needed "struct" changed to "enum" in the kernel-doc header.
For a few of them, I changed the "/**" comment to a plain "/*"
comment since there was no kernel-doc content there.
Example:
Warning: drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h:2673 Enum value 'FAMS2_ALLOW_DELAY_CHECK_NONE' not described in enum 'dmub_fams2_allow_delay_check_mode'
Warning: drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h:2673 Enum value 'FAMS2_ALLOW_DELAY_CHECK_FROM_START' not described in enum 'dmub_fams2_allow_delay_check_mode'
Warning: drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h:2673 Enum value 'FAMS2_ALLOW_DELAY_CHECK_FROM_PREPARE' not described in enum 'dmub_fams2_allow_delay_check_mode'
All .o files are the same with or without this patch.
Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Correct all typedef, struct, and union prototype warnings in
dmub_cmd.h by using the matching names or "typedef" keyword:
Warning: drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h:961 cannot understand function prototype: 'typedef uint32_t dmub_trace_code_t;'
Warning: drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h:1183 expecting prototype for union dmub_shared_state_ips_fw. Prototype was for union dmub_shared_state_ips_fw_signals instead
Warning: drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h:1203 expecting prototype for union dmub_shared_state_ips_signals. Prototype was for union dmub_shared_state_ips_driver_signals instead
Warning: drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h:1253 expecting prototype for struct dmub_shared_state_cursor_offload_v1. Prototype was for struct dmub_shared_state_cursor_offload_stream_v1 instead
Warning: drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h:1269 struct dmub_shared
_state_feature_common { uint32_t padding[62]; };: error: Cannot parse enum!
Warning: drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h:1278 struct dmub_shared
_state_feature_header { uint16_t id; uint16_t version; uint32_t reserved; };:
error: Cannot parse enum!
Warning: drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h:2001 expecting prototype for struct dmub_cmd_read_modify_write_sequence. Prototype was for struct dmub_rb_cmd_read_modify_write instead
Warning: drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h:2191 expecting prototype for struct dmub_rb_cmd_cab. Prototype was for struct dmub_rb_cmd_cab_for_ss instead
Warning: drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h:2875 expecting prototype for struct dmub_cmd_set_pixel_clock_data. Prototype was for struct dmub_rb_cmd_set_pixel_clock instead
Warning: drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h:3398 expecting prototype for union dpia_notify_data_type. Prototype was for union dpia_notification_data instead
Warning: drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h:3447 expecting prototype for struct dmub_rb_cmd_hpd_sense_notify. Prototype was for struct dmub_rb_cmd_hpd_sense_notify_data instead
Warning: drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h:6267 expecting prototype for struct dmub_cmd_cable_id_input. Prototype was for struct dmub_cmd_cable_id_output instead
Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Some board manufacturers hardcode the EDID for the embedded
panel in the VBIOS. This EDID should be used when the panel
doesn't have a DDC.
For reference, see the legacy non-DC display code:
amdgpu_atombios_encoder_get_lcd_info()
This is necessary to support embedded connectors without DDC.
Fixes: 4562236b3b ("drm/amd/dc: Add dc display driver (v2)")
Link: https://gitlab.freedesktop.org/drm/amd/-/work_items/5192
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Remove duplicate inclusion of dce/dmub_hw_lock_mgr.h in dcn42_hwseq.c to
clean up redundant code.
Signed-off-by: Chen Ni <nichen@iscas.ac.cn>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Gfx clock is fine grained on SMU v13.0.6/12 SOCs. Add the flag to report
clock frequencies correctly.
Fixes: 7380228401 ("drm/amd/pm: Use generic dpm table for SMUv13 SOCs")
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>
User mode compute queue/pipe reset was originally only supported in GC 9.
It is now also supported in GC 12.1. Report it in KFD topology.
Signed-off-by: Amber Lin <Amber.Lin@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
In a multi-XCC GPU, pass the master XCC's ID to amdgpu_mes_suspend,
amdgpu_mes_resume, and detect_and_reset_hung_queues so the command will be
sent to the matching master MES when the compute partition mode is not
SPX.
Signed-off-by: Amber Lin <Amber.Lin@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Refactor the NUMA-aware MTYPE override for VM page table entries:
Move the override_vm_pte_flags call from the centralized
amdgpu_vm_pte_update_flags() into the individual CPU and SDMA update
backends, enabling per-PTE MTYPE override including for scattered
pages (pages_addr path).
Move APU, IP version, and direct-mapped eligibility checks from
runtime (gmc_v9_0_override_vm_pte_flags) to init time
(gmc_v9_0_set_gmc_funcs), to avoid repeated runtime checks on
every PTE update.
Signed-off-by: Philip Yang <Philip.Yang@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
