drm/amdgpu: Fix cond_exec handling in amdgpu_ib_schedule()

The EXEC_COUNT field must be > 0.  In the gfx shadow
handling we always emit a cond_exec packet after the gfx_shadow
packet, but the EXEC_COUNT never gets patched.  This leads
to a hang when we try and reset queues on gfx11 APUs.

Fixes: c68cbbfd54c6 ("drm/amdgpu: cleanup conditional execution")
Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/4789
Reviewed-by: Jesse Zhang <Jesse.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit ba205ac3d6e83f56c4f824f23f1b4522cb844ff3)
Cc: stable@vger.kernel.org

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
index 72ec455fa932ca4e983b33b7558b01e7dc3e5d36..44f230d67da242d626facfd3b8afe89f88fbdc61 100644 (file)
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
@@ -235,7 +235,7 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned int num_ibs,
 
        amdgpu_ring_ib_begin(ring);
 
-       if (ring->funcs->emit_gfx_shadow)
+       if (ring->funcs->emit_gfx_shadow && adev->gfx.cp_gfx_shadow)
                amdgpu_ring_emit_gfx_shadow(ring, shadow_va, csa_va, gds_va,
                                            init_shadow, vmid);
 
@@ -291,7 +291,8 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned int num_ibs,
                                       fence_flags | AMDGPU_FENCE_FLAG_64BIT);
        }
 
-       if (ring->funcs->emit_gfx_shadow && ring->funcs->init_cond_exec) {
+       if (ring->funcs->emit_gfx_shadow && ring->funcs->init_cond_exec &&
+           adev->gfx.cp_gfx_shadow) {
                amdgpu_ring_emit_gfx_shadow(ring, 0, 0, 0, false, 0);
                amdgpu_ring_init_cond_exec(ring, ring->cond_exe_gpu_addr);
        }