drm/amdgpu: drop AMDGPU_FENCE_SIGNALED_SEQ

It's causing issues with VMID handling and comparing the
fence value two times actually doesn't make handling faster.

Port of radeon commit "d6d5c5b8364bcc4d52cddc68bcb0a330d2af20f3".

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Monk Liu <monk.liu@amd.com>

drivers/gpu/drm/amd/amdgpu/amdgpu.h

@@ -84,9 +84,6 @@ extern int amdgpu_vm_block_size;
 #define AMDGPUFB_CONN_LIMIT			4
 #define AMDGPU_BIOS_NUM_SCRATCH			8
 
-/* fence seq are set to this number when signaled */
-#define AMDGPU_FENCE_SIGNALED_SEQ		0LL
-
 /* max number of rings */
 #define AMDGPU_MAX_RINGS			16
 #define AMDGPU_MAX_GFX_RINGS			1

drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c

@@ -463,11 +463,7 @@ bool amdgpu_fence_signaled(struct amdgpu_fence *fence)
 	if (!fence)
 		return true;
 
-	if (fence->seq == AMDGPU_FENCE_SIGNALED_SEQ)
-		return true;
-
 	if (amdgpu_fence_seq_signaled(fence->ring, fence->seq)) {
-		fence->seq = AMDGPU_FENCE_SIGNALED_SEQ;
 		if (!fence_signal(&fence->base))
 			FENCE_TRACE(&fence->base, "signaled from amdgpu_fence_signaled\n");
 		return true;
@@ -637,15 +633,11 @@ int amdgpu_fence_wait(struct amdgpu_fence *fence, bool intr)
 	long r;
 
 	seq[fence->ring->idx] = fence->seq;
-	if (seq[fence->ring->idx] == AMDGPU_FENCE_SIGNALED_SEQ)
-		return 0;
-
 	r = amdgpu_fence_wait_seq_timeout(fence->ring->adev, seq, intr, MAX_SCHEDULE_TIMEOUT);
 	if (r < 0) {
 		return r;
 	}
 
-	fence->seq = AMDGPU_FENCE_SIGNALED_SEQ;
 	r = fence_signal(&fence->base);
 	if (!r)
 		FENCE_TRACE(&fence->base, "signaled from fence_wait\n");
@@ -682,10 +674,6 @@ int amdgpu_fence_wait_any(struct amdgpu_device *adev,
 
 		seq[i] = fences[i]->seq;
 		++num_rings;
-
-		/* test if something was allready signaled */
-		if (seq[i] == AMDGPU_FENCE_SIGNALED_SEQ)
-			return 0;
 	}
 
 	/* nothing to wait for ? */