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>

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

@@ -533,7 +533,7 @@ static inline void amdgpu_set_ib_value(struct amdgpu_cs_parser *p,
 /*
  * Writeback
  */
-#define AMDGPU_MAX_WB 128	/* Reserve at most 128 WB slots for amdgpu-owned rings. */
+#define AMDGPU_MAX_WB 256	/* Reserve at most 256 WB slots for amdgpu-owned rings. */
 
 struct amdgpu_wb {
 	struct amdgpu_bo	*wb_obj;

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

@@ -304,10 +304,6 @@ int amdgpu_gfx_kiq_init_ring(struct amdgpu_device *adev,
 
 	spin_lock_init(&kiq->ring_lock);
 
-	r = amdgpu_device_wb_get(adev, &kiq->reg_val_offs);
-	if (r)
-		return r;
-
 	ring->adev = NULL;
 	ring->ring_obj = NULL;
 	ring->use_doorbell = true;
@@ -331,7 +327,6 @@ int amdgpu_gfx_kiq_init_ring(struct amdgpu_device *adev,
 
 void amdgpu_gfx_kiq_free_ring(struct amdgpu_ring *ring)
 {
-	amdgpu_device_wb_free(ring->adev, ring->adev->gfx.kiq.reg_val_offs);
 	amdgpu_ring_fini(ring);
 }
 
@@ -672,15 +667,20 @@ uint32_t amdgpu_kiq_rreg(struct amdgpu_device *adev, uint32_t reg)
 {
 	signed long r, cnt = 0;
 	unsigned long flags;
-	uint32_t seq;
+	uint32_t seq, reg_val_offs = 0, 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_emit_rreg(ring, reg);
+	amdgpu_ring_emit_rreg(ring, reg, reg_val_offs);
 	amdgpu_fence_emit_polling(ring, &seq);
 	amdgpu_ring_commit(ring);
 	spin_unlock_irqrestore(&kiq->ring_lock, flags);
@@ -707,7 +707,10 @@ uint32_t amdgpu_kiq_rreg(struct amdgpu_device *adev, uint32_t reg)
 	if (cnt > MAX_KIQ_REG_TRY)
 		goto failed_kiq_read;
 
-	return adev->wb.wb[kiq->reg_val_offs];
+	mb();
+	value = adev->wb.wb[reg_val_offs];
+	amdgpu_device_wb_free(adev, reg_val_offs);
+	return value;
 
 failed_kiq_read:
 	pr_err("failed to read reg:%x\n", reg);

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

@@ -103,7 +103,6 @@ struct amdgpu_kiq {
 	struct amdgpu_ring	ring;
 	struct amdgpu_irq_src	irq;
 	const struct kiq_pm4_funcs *pmf;
-	uint32_t			reg_val_offs;
 };
 
 /*

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

@@ -168,7 +168,8 @@ struct amdgpu_ring_funcs {
 	void (*end_use)(struct amdgpu_ring *ring);
 	void (*emit_switch_buffer) (struct amdgpu_ring *ring);
 	void (*emit_cntxcntl) (struct amdgpu_ring *ring, uint32_t flags);
-	void (*emit_rreg)(struct amdgpu_ring *ring, uint32_t reg);
+	void (*emit_rreg)(struct amdgpu_ring *ring, uint32_t reg,
+			  uint32_t reg_val_offs);
 	void (*emit_wreg)(struct amdgpu_ring *ring, uint32_t reg, uint32_t val);
 	void (*emit_reg_wait)(struct amdgpu_ring *ring, uint32_t reg,
 			      uint32_t val, uint32_t mask);
@@ -250,7 +251,7 @@ struct amdgpu_ring {
 #define amdgpu_ring_emit_hdp_flush(r) (r)->funcs->emit_hdp_flush((r))
 #define amdgpu_ring_emit_switch_buffer(r) (r)->funcs->emit_switch_buffer((r))
 #define amdgpu_ring_emit_cntxcntl(r, d) (r)->funcs->emit_cntxcntl((r), (d))
-#define amdgpu_ring_emit_rreg(r, d) (r)->funcs->emit_rreg((r), (d))
+#define amdgpu_ring_emit_rreg(r, d, o) (r)->funcs->emit_rreg((r), (d), (o))
 #define amdgpu_ring_emit_wreg(r, d, v) (r)->funcs->emit_wreg((r), (d), (v))
 #define amdgpu_ring_emit_reg_wait(r, d, v, m) (r)->funcs->emit_reg_wait((r), (d), (v), (m))
 #define amdgpu_ring_emit_reg_write_reg_wait(r, d0, d1, v, m) (r)->funcs->emit_reg_write_reg_wait((r), (d0), (d1), (v), (m))

drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c

@@ -7599,10 +7599,10 @@ static void gfx_v10_0_ring_emit_tmz(struct amdgpu_ring *ring, bool start)
 	amdgpu_ring_write(ring, FRAME_CMD(start ? 0 : 1)); /* frame_end */
 }
 
-static void gfx_v10_0_ring_emit_rreg(struct amdgpu_ring *ring, uint32_t reg)
+static void gfx_v10_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*/
@@ -7611,9 +7611,9 @@ static void gfx_v10_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_v10_0_ring_emit_wreg(struct amdgpu_ring *ring, uint32_t reg,

drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c

@@ -6389,10 +6389,10 @@ static void gfx_v8_0_ring_emit_patch_cond_exec(struct amdgpu_ring *ring, unsigne
 		ring->ring[offset] = (ring->ring_size >> 2) - offset + cur;
 }
 
-static void gfx_v8_0_ring_emit_rreg(struct amdgpu_ring *ring, uint32_t reg)
+static void gfx_v8_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*/
@@ -6401,9 +6401,9 @@ static void gfx_v8_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_v8_0_ring_emit_wreg(struct amdgpu_ring *ring, uint32_t reg,

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,