drm/amdgpu: request reg_val_offs each kiq read reg

According to the current kiq read register method,
there will be race condition when using KIQ to read
register if multiple clients want to read at same time
just like the expample below:
1. client-A start to read REG-0 throguh KIQ
2. client-A poll the seqno-0
3. client-B start to read REG-1 through KIQ
4. client-B poll the seqno-1
5. the kiq complete these two read operation
6. client-A to read the register at the wb buffer and
   get REG-1 value

Therefore, use amdgpu_device_wb_get() to request reg_val_offs
for each kiq read register.

v2: fix the error remove
v3: fix the print typo
v4: remove unused variables

Signed-off-by: Yintian Tao <yttao@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

1 files changed, 18 insertions, 9 deletions

diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
index f8273fd8191a..496205a8ee0c 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
@@ -4045,13 +4045,19 @@ static uint64_t gfx_v9_0_kiq_read_clock(struct amdgpu_device *adev)
 {
 	signed long r, cnt = 0;
 	unsigned long flags;
-	uint32_t seq;
+	uint32_t seq, reg_val_offs = 0;
+	uint64_t value = 0;
 	struct amdgpu_kiq *kiq = &adev->gfx.kiq;
 	struct amdgpu_ring *ring = &kiq->ring;
 
 	BUG_ON(!ring->funcs->emit_rreg);
 
 	spin_lock_irqsave(&kiq->ring_lock, flags);
+	if (amdgpu_device_wb_get(adev, &reg_val_offs)) {
+		spin_unlock_irqrestore(&kiq->ring_lock, flags);
+		pr_err("critical bug! too many kiq readers\n");
+		goto failed_kiq_read;
+	}
 	amdgpu_ring_alloc(ring, 32);
 	amdgpu_ring_write(ring, PACKET3(PACKET3_COPY_DATA, 4));
 	amdgpu_ring_write(ring, 9 |	/* src: register*/
@@ -4061,9 +4067,9 @@ static uint64_t gfx_v9_0_kiq_read_clock(struct amdgpu_device *adev)
 	amdgpu_ring_write(ring, 0);
 	amdgpu_ring_write(ring, 0);
 	amdgpu_ring_write(ring, lower_32_bits(adev->wb.gpu_addr +
-				kiq->reg_val_offs * 4));
+				reg_val_offs * 4));
 	amdgpu_ring_write(ring, upper_32_bits(adev->wb.gpu_addr +
-				kiq->reg_val_offs * 4));
+				reg_val_offs * 4));
 	amdgpu_fence_emit_polling(ring, &seq);
 	amdgpu_ring_commit(ring);
 	spin_unlock_irqrestore(&kiq->ring_lock, flags);
@@ -4090,8 +4096,11 @@ static uint64_t gfx_v9_0_kiq_read_clock(struct amdgpu_device *adev)
 	if (cnt > MAX_KIQ_REG_TRY)
 		goto failed_kiq_read;
 
-	return (uint64_t)adev->wb.wb[kiq->reg_val_offs] |
-		(uint64_t)adev->wb.wb[kiq->reg_val_offs + 1 ] << 32ULL;
+	mb();
+	value = (uint64_t)adev->wb.wb[reg_val_offs] |
+		(uint64_t)adev->wb.wb[reg_val_offs + 1 ] << 32ULL;
+	amdgpu_device_wb_free(adev, reg_val_offs);
+	return value;
 
 failed_kiq_read:
 	pr_err("failed to read gpu clock\n");
@@ -5490,10 +5499,10 @@ static void gfx_v9_0_ring_emit_patch_cond_exec(struct amdgpu_ring *ring, unsigne
 		ring->ring[offset] = (ring->ring_size>>2) - offset + cur;
 }
 
-static void gfx_v9_0_ring_emit_rreg(struct amdgpu_ring *ring, uint32_t reg)
+static void gfx_v9_0_ring_emit_rreg(struct amdgpu_ring *ring, uint32_t reg,
+				    uint32_t reg_val_offs)
 {
 	struct amdgpu_device *adev = ring->adev;
-	struct amdgpu_kiq *kiq = &adev->gfx.kiq;
 
 	amdgpu_ring_write(ring, PACKET3(PACKET3_COPY_DATA, 4));
 	amdgpu_ring_write(ring, 0 |	/* src: register*/
@@ -5502,9 +5511,9 @@ static void gfx_v9_0_ring_emit_rreg(struct amdgpu_ring *ring, uint32_t reg)
 	amdgpu_ring_write(ring, reg);
 	amdgpu_ring_write(ring, 0);
 	amdgpu_ring_write(ring, lower_32_bits(adev->wb.gpu_addr +
-				kiq->reg_val_offs * 4));
+				reg_val_offs * 4));
 	amdgpu_ring_write(ring, upper_32_bits(adev->wb.gpu_addr +
-				kiq->reg_val_offs * 4));
+				reg_val_offs * 4));
 }
 
 static void gfx_v9_0_ring_emit_wreg(struct amdgpu_ring *ring, uint32_t reg,