On Tue, Oct 13, 2026 at 8:46 PM Willian Oliveira <[email protected]> wrote: > > gfx_v8_0_sw_fini() already releases the RLC objects used by gfx8 > (clear_state_obj and cp_table_obj). The generic > amdgpu_gfx_rlc_fini() also frees save_restore_obj, which is not used > by gfx8. > > Signed-off-by: Willian Oliveira <[email protected]> > --- > drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c | 1 - > 1 file changed, 1 deletion(-) > > diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c > b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c > index 7f91186ef1d1..088428e41f44 100644 > --- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c > +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c > @@ -2059,7 +2059,6 @@ static int gfx_v8_0_sw_fini(struct amdgpu_ip_block > *ip_block) > amdgpu_gfx_kiq_fini(adev, 0); > > gfx_v8_0_mec_fini(adev); > - amdgpu_gfx_rlc_fini(adev);
I'd suggest dropping the explicit free calls and just use amdgpu_gfx_rlc_fini() for symmetry with other gfx IPs and the rlc init calls. Alex > amdgpu_bo_free_kernel(&adev->gfx.rlc.clear_state_obj, > &adev->gfx.rlc.clear_state_gpu_addr, > (void **)&adev->gfx.rlc.cs_ptr); > -- > 2.47.3 >
