On 5/9/2023 7:12 AM, Guchun Chen wrote:
Prevent further casting on chip MULLINS/KABINI/KAVERI when calling
amdgpu_dpm_is_overdrive_supported, this can avoid UBSAN complain
in init sequence.
Suggested-by: Evan Quan <evan.quan@xxxxxxx>
Signed-off-by: Guchun Chen <guchun.chen@xxxxxxx>
---
drivers/gpu/drm/amd/pm/amdgpu_dpm.c | 10 ++++++++--
1 file changed, 8 insertions(+), 2 deletions(-)
diff --git a/drivers/gpu/drm/amd/pm/amdgpu_dpm.c b/drivers/gpu/drm/amd/pm/amdgpu_dpm.c
index 86246f69dbe1..ccd3ea89eacf 100644
--- a/drivers/gpu/drm/amd/pm/amdgpu_dpm.c
+++ b/drivers/gpu/drm/amd/pm/amdgpu_dpm.c
@@ -1467,8 +1467,14 @@ int amdgpu_dpm_is_overdrive_supported(struct amdgpu_device *adev)
} else {
struct pp_hwmgr *hwmgr;
Instead of family check, what if you wrap it like -
is_legacy_dpm(adev)
return (adev->powerplay.pp_handle == adev)
Could be useful for legacy dpm checks.
Thanks,
Lijo
- /* SI asic does not carry od_enabled */
- if (adev->family == AMDGPU_FAMILY_SI)
+ /*
+ * SI asic and chip MULLINS/KABINI/KAVERI do not carry
+ * od_enabled, as its pp_handle is casted from adev.
+ */
+ if ((adev->family == AMDGPU_FAMILY_SI) ||
+ (adev->asic_type == CHIP_MULLINS) ||
+ (adev->asic_type == CHIP_KABINI) ||
+ (adev->asic_type == CHIP_KAVERI))
return false;
hwmgr = (struct pp_hwmgr *)adev->powerplay.pp_handle;
