3.13.11-ckt16 -stable review patch. If anyone has any objections, please let me know.
------------------
From: Alex Deucher <alexander.deucher@xxxxxxx>
commit 3a01fd367e09ebf05d75a000407364e7ebe2b678 upstream.
We need to wait for the GPUVM flush to complete. There
was some confusion as to how this mechanism was supposed
to work. The operation is not atomic. For GPU initiated
invalidations you need to read back a VM register to
introduce enough latency for the update to complete.
v2: drop gart changes
v3: just read back rather than polling
Reviewed-by: Christian König <christian.koenig@xxxxxxx>
Signed-off-by: Alex Deucher <alexander.deucher@xxxxxxx>
Signed-off-by: Kamal Mostafa <kamal@xxxxxxxxxxxxx>
---
drivers/gpu/drm/radeon/cik.c | 11 +++++++++++
drivers/gpu/drm/radeon/cik_sdma.c | 9 +++++++++
2 files changed, 20 insertions(+)
diff --git a/drivers/gpu/drm/radeon/cik.c b/drivers/gpu/drm/radeon/cik.c
index e205912..de0652e 100644
--- a/drivers/gpu/drm/radeon/cik.c
+++ b/drivers/gpu/drm/radeon/cik.c
@@ -5397,6 +5397,17 @@ void cik_vm_flush(struct radeon_device *rdev, int ridx, struct radeon_vm *vm)
radeon_ring_write(ring, 0);
radeon_ring_write(ring, 1 << vm->id);
+ /* wait for the invalidate to complete */
+ radeon_ring_write(ring, PACKET3(PACKET3_WAIT_REG_MEM, 5));
+ radeon_ring_write(ring, (WAIT_REG_MEM_OPERATION(0) | /* wait */
+ WAIT_REG_MEM_FUNCTION(0) | /* always */
+ WAIT_REG_MEM_ENGINE(0))); /* me */
+ radeon_ring_write(ring, VM_INVALIDATE_REQUEST >> 2);
+ radeon_ring_write(ring, 0);
+ radeon_ring_write(ring, 0); /* ref */
+ radeon_ring_write(ring, 0); /* mask */
+ radeon_ring_write(ring, 0x20); /* poll interval */
+
/* compute doesn't have PFP */
if (usepfp) {
/* sync PFP to ME, otherwise we might get invalid PFP reads */
diff --git a/drivers/gpu/drm/radeon/cik_sdma.c b/drivers/gpu/drm/radeon/cik_sdma.c
index a1aeab0..cdcba5f 100644
--- a/drivers/gpu/drm/radeon/cik_sdma.c
+++ b/drivers/gpu/drm/radeon/cik_sdma.c
@@ -731,6 +731,8 @@ void cik_sdma_vm_set_page(struct radeon_device *rdev,
*/
void cik_dma_vm_flush(struct radeon_device *rdev, int ridx, struct radeon_vm *vm)
{
+ u32 extra_bits = (SDMA_POLL_REG_MEM_EXTRA_OP(0) |
+ SDMA_POLL_REG_MEM_EXTRA_FUNC(0)); /* always */
struct radeon_ring *ring = &rdev->ring[ridx];
if (vm == NULL)
@@ -776,5 +778,12 @@ void cik_dma_vm_flush(struct radeon_device *rdev, int ridx, struct radeon_vm *vm
radeon_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_SRBM_WRITE, 0, 0xf000));
radeon_ring_write(ring, VM_INVALIDATE_REQUEST >> 2);
radeon_ring_write(ring, 1 << vm->id);
+
+ radeon_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_POLL_REG_MEM, 0, extra_bits));
+ radeon_ring_write(ring, VM_INVALIDATE_REQUEST >> 2);
+ radeon_ring_write(ring, 0);
+ radeon_ring_write(ring, 0); /* reference */
+ radeon_ring_write(ring, 0); /* mask */
+ radeon_ring_write(ring, (0xfff << 16) | 10); /* retry count, poll interval */
}
--
1.9.1
--
To unsubscribe from this list: send the line "unsubscribe stable" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html