diff options
| author | Huang Rui <ray.huang@amd.com> | 2020-03-09 14:52:06 -0400 |
|---|---|---|
| committer | Alex Deucher <alexander.deucher@amd.com> | 2020-04-28 16:20:30 -0400 |
| commit | f77c9aff85caa9e9b0c2b6fc5be1fc2c76fe02b2 (patch) | |
| tree | 52f0e76d85c682da0d83af922bf910ecb801a35c /drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c | |
| parent | b71a564e2509e1000044a9873cbee6d6a6a5ab90 (diff) | |
| download | linux-f77c9aff85caa9e9b0c2b6fc5be1fc2c76fe02b2.tar.bz2 | |
drm/amdgpu: Fix per-IB secure flag GFX hang
Since commit "Move to a per-IB secure flag (TMZ)",
we've been seeing hangs in GFX. We need to send
FRAME CONTROL stop/start back-to-back, every time
we flip the TMZ flag. That is, when we transition
from TMZ to non-TMZ we have to send a stop with
TMZ followed by a start with non-TMZ, and
similarly for transitioning from non-TMZ into TMZ.
This patch implements this, thus fixing the GFX
hang.
v1 -> v2:
As suggested by Luben, and accept part of implemetation from this patch:
- Put "secure" closed to the loop and use optimization
- Change "secure" to bool again, and move "secure == -1" out of loop.
v3: Small fixes/optimizations.
Reported-and-Tested-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Signed-off-by: Huang Rui <ray.huang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Luben Tuikov <luben.tuikov@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Diffstat (limited to 'drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c')
| -rw-r--r-- | drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c | 15 |
1 files changed, 8 insertions, 7 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c index 404c6d470515..9fe20b782e88 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c @@ -3037,7 +3037,7 @@ static int gfx_v10_0_rlc_backdoor_autoload_enable(struct amdgpu_device *adev); static int gfx_v10_0_wait_for_rlc_autoload_complete(struct amdgpu_device *adev); static void gfx_v10_0_ring_emit_ce_meta(struct amdgpu_ring *ring, bool resume); static void gfx_v10_0_ring_emit_de_meta(struct amdgpu_ring *ring, bool resume); -static void gfx_v10_0_ring_emit_tmz(struct amdgpu_ring *ring, bool start); +static void gfx_v10_0_ring_emit_frame_cntl(struct amdgpu_ring *ring, bool start, bool secure); static void gfx10_kiq_set_resources(struct amdgpu_ring *kiq_ring, uint64_t queue_mask) { @@ -7599,12 +7599,13 @@ static void gfx_v10_0_ring_emit_de_meta(struct amdgpu_ring *ring, bool resume) sizeof(de_payload) >> 2); } -static void gfx_v10_0_ring_emit_tmz(struct amdgpu_ring *ring, bool start) +static void gfx_v10_0_ring_emit_frame_cntl(struct amdgpu_ring *ring, bool start, + bool secure) { - if (amdgpu_is_tmz(ring->adev)) { - amdgpu_ring_write(ring, PACKET3(PACKET3_FRAME_CONTROL, 0)); - amdgpu_ring_write(ring, FRAME_TMZ | FRAME_CMD(start ? 0 : 1)); - } + uint32_t v = secure ? FRAME_TMZ : 0; + + amdgpu_ring_write(ring, PACKET3(PACKET3_FRAME_CONTROL, 0)); + amdgpu_ring_write(ring, v | FRAME_CMD(start ? 0 : 1)); } static void gfx_v10_0_ring_emit_rreg(struct amdgpu_ring *ring, uint32_t reg, @@ -8058,7 +8059,7 @@ static const struct amdgpu_ring_funcs gfx_v10_0_ring_funcs_gfx = { .init_cond_exec = gfx_v10_0_ring_emit_init_cond_exec, .patch_cond_exec = gfx_v10_0_ring_emit_patch_cond_exec, .preempt_ib = gfx_v10_0_ring_preempt_ib, - .emit_tmz = gfx_v10_0_ring_emit_tmz, + .emit_frame_cntl = gfx_v10_0_ring_emit_frame_cntl, .emit_wreg = gfx_v10_0_ring_emit_wreg, .emit_reg_wait = gfx_v10_0_ring_emit_reg_wait, .emit_reg_write_reg_wait = gfx_v10_0_ring_emit_reg_write_reg_wait, |