drm/amdgpu: add insert_nop ring func and default implementation

The insert_nop function is added to amdgpu_ring_funcs structure as
well as the default implementation

Signed-off-by: Jammy Zhou <Jammy.Zhou@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>

12 files changed, 36 insertions, 3 deletions

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
index 057e7ef5c0d0..668939a14206 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
@@ -343,6 +343,8 @@ struct amdgpu_ring_funcs {
         int (*test_ring)(struct amdgpu_ring *ring);
         int (*test_ib)(struct amdgpu_ring *ring);
         bool (*is_lockup)(struct amdgpu_ring *ring);
+        /* insert NOP packets */
+        void (*insert_nop)(struct amdgpu_ring *ring, uint32_t count);
 };
 
 /*
@@ -1217,6 +1219,7 @@ int amdgpu_ib_ring_tests(struct amdgpu_device *adev);
 void amdgpu_ring_free_size(struct amdgpu_ring *ring);
 int amdgpu_ring_alloc(struct amdgpu_ring *ring, unsigned ndw);
 int amdgpu_ring_lock(struct amdgpu_ring *ring, unsigned ndw);
+void amdgpu_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count);
 void amdgpu_ring_commit(struct amdgpu_ring *ring);
 void amdgpu_ring_unlock_commit(struct amdgpu_ring *ring);
 void amdgpu_ring_undo(struct amdgpu_ring *ring);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
index 7d442c51063e..9bec91484c24 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
@@ -131,6 +131,21 @@ int amdgpu_ring_lock(struct amdgpu_ring *ring, unsigned ndw)
         return 0;
 }
 
+/** amdgpu_ring_insert_nop - insert NOP packets
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ * @count: the number of NOP packets to insert
+ *
+ * This is the generic insert_nop function for rings except SDMA
+ */
+void amdgpu_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
+{
+        int i;
+
+        for (i = 0; i < count; i++)
+                amdgpu_ring_write(ring, ring->nop);
+}
+
 /**
  * amdgpu_ring_commit - tell the GPU to execute the new
  * commands on the ring buffer
@@ -143,10 +158,13 @@ int amdgpu_ring_lock(struct amdgpu_ring *ring, unsigned ndw)
  */
 void amdgpu_ring_commit(struct amdgpu_ring *ring)
 {
+        uint32_t count;
+
         /* We pad to match fetch size */
-        while (ring->wptr & ring->align_mask) {
-                amdgpu_ring_write(ring, ring->nop);
-        }
+        count = ring->align_mask + 1 - (ring->wptr & ring->align_mask);
+        count %= ring->align_mask + 1;
+        ring->funcs->insert_nop(ring, count);
+
         mb();
         amdgpu_ring_set_wptr(ring);
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/cik_sdma.c b/drivers/gpu/drm/amd/amdgpu/cik_sdma.c
index cc909c9cee63..f0661b269a63 100644
--- a/drivers/gpu/drm/amd/amdgpu/cik_sdma.c
+++ b/drivers/gpu/drm/amd/amdgpu/cik_sdma.c
@@ -1305,6 +1305,7 @@ static const struct amdgpu_ring_funcs cik_sdma_ring_funcs = {
         .test_ring = cik_sdma_ring_test_ring,
         .test_ib = cik_sdma_ring_test_ib,
         .is_lockup = cik_sdma_ring_is_lockup,
+        .insert_nop = amdgpu_ring_insert_nop,
 };
 
 static void cik_sdma_set_ring_funcs(struct amdgpu_device *adev)
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
index fab7b236f37f..517a68f82ec3 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
@@ -5598,6 +5598,7 @@ static const struct amdgpu_ring_funcs gfx_v7_0_ring_funcs_gfx = {
         .test_ring = gfx_v7_0_ring_test_ring,
         .test_ib = gfx_v7_0_ring_test_ib,
         .is_lockup = gfx_v7_0_ring_is_lockup,
+        .insert_nop = amdgpu_ring_insert_nop,
 };
 
 static const struct amdgpu_ring_funcs gfx_v7_0_ring_funcs_compute = {
@@ -5614,6 +5615,7 @@ static const struct amdgpu_ring_funcs gfx_v7_0_ring_funcs_compute = {
         .test_ring = gfx_v7_0_ring_test_ring,
         .test_ib = gfx_v7_0_ring_test_ib,
         .is_lockup = gfx_v7_0_ring_is_lockup,
+        .insert_nop = amdgpu_ring_insert_nop,
 };
 
 static void gfx_v7_0_set_ring_funcs(struct amdgpu_device *adev)
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
index 818edb37fa9c..956a35822f86 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
@@ -4378,6 +4378,7 @@ static const struct amdgpu_ring_funcs gfx_v8_0_ring_funcs_gfx = {
         .test_ring = gfx_v8_0_ring_test_ring,
         .test_ib = gfx_v8_0_ring_test_ib,
         .is_lockup = gfx_v8_0_ring_is_lockup,
+        .insert_nop = amdgpu_ring_insert_nop,
 };
 
 static const struct amdgpu_ring_funcs gfx_v8_0_ring_funcs_compute = {
@@ -4394,6 +4395,7 @@ static const struct amdgpu_ring_funcs gfx_v8_0_ring_funcs_compute = {
         .test_ring = gfx_v8_0_ring_test_ring,
         .test_ib = gfx_v8_0_ring_test_ib,
         .is_lockup = gfx_v8_0_ring_is_lockup,
+        .insert_nop = amdgpu_ring_insert_nop,
 };
 
 static void gfx_v8_0_set_ring_funcs(struct amdgpu_device *adev)
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c b/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c
index 2457bf3e3a1e..1b913bce2599 100644
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c
@@ -1316,6 +1316,7 @@ static const struct amdgpu_ring_funcs sdma_v2_4_ring_funcs = {
         .test_ring = sdma_v2_4_ring_test_ring,
         .test_ib = sdma_v2_4_ring_test_ib,
         .is_lockup = sdma_v2_4_ring_is_lockup,
+        .insert_nop = amdgpu_ring_insert_nop,
 };
 
 static void sdma_v2_4_set_ring_funcs(struct amdgpu_device *adev)
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c
index 70b2f03c5712..a7550a8f5d84 100644
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c
@@ -1440,6 +1440,7 @@ static const struct amdgpu_ring_funcs sdma_v3_0_ring_funcs = {
         .test_ring = sdma_v3_0_ring_test_ring,
         .test_ib = sdma_v3_0_ring_test_ib,
         .is_lockup = sdma_v3_0_ring_is_lockup,
+        .insert_nop = amdgpu_ring_insert_nop,
 };
 
 static void sdma_v3_0_set_ring_funcs(struct amdgpu_device *adev)
diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c b/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c
index 9ac383bc6c1f..5fac5da694f0 100644
--- a/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c
+++ b/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c
@@ -886,6 +886,7 @@ static const struct amdgpu_ring_funcs uvd_v4_2_ring_funcs = {
         .test_ring = uvd_v4_2_ring_test_ring,
         .test_ib = uvd_v4_2_ring_test_ib,
         .is_lockup = amdgpu_ring_test_lockup,
+        .insert_nop = amdgpu_ring_insert_nop,
 };
 
 static void uvd_v4_2_set_ring_funcs(struct amdgpu_device *adev)
diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c b/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c
index de4b3f57902d..2d5c59c318af 100644
--- a/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c
@@ -825,6 +825,7 @@ static const struct amdgpu_ring_funcs uvd_v5_0_ring_funcs = {
         .test_ring = uvd_v5_0_ring_test_ring,
         .test_ib = uvd_v5_0_ring_test_ib,
         .is_lockup = amdgpu_ring_test_lockup,
+        .insert_nop = amdgpu_ring_insert_nop,
 };
 
 static void uvd_v5_0_set_ring_funcs(struct amdgpu_device *adev)
diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c b/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c
index 66c975870e97..d9f553fce531 100644
--- a/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c
@@ -805,6 +805,7 @@ static const struct amdgpu_ring_funcs uvd_v6_0_ring_funcs = {
         .test_ring = uvd_v6_0_ring_test_ring,
         .test_ib = uvd_v6_0_ring_test_ib,
         .is_lockup = amdgpu_ring_test_lockup,
+        .insert_nop = amdgpu_ring_insert_nop,
 };
 
 static void uvd_v6_0_set_ring_funcs(struct amdgpu_device *adev)
diff --git a/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c b/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c
index 303d961d57bd..cd16df543f64 100644
--- a/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c
@@ -643,6 +643,7 @@ static const struct amdgpu_ring_funcs vce_v2_0_ring_funcs = {
         .test_ring = amdgpu_vce_ring_test_ring,
         .test_ib = amdgpu_vce_ring_test_ib,
         .is_lockup = amdgpu_ring_test_lockup,
+        .insert_nop = amdgpu_ring_insert_nop,
 };
 
 static void vce_v2_0_set_ring_funcs(struct amdgpu_device *adev)
diff --git a/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c b/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c
index 4349658081ff..5642b8eb92ad 100644
--- a/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c
@@ -608,6 +608,7 @@ static const struct amdgpu_ring_funcs vce_v3_0_ring_funcs = {
         .test_ring = amdgpu_vce_ring_test_ring,
         .test_ib = amdgpu_vce_ring_test_ib,
         .is_lockup = amdgpu_ring_test_lockup,
+        .insert_nop = amdgpu_ring_insert_nop,
 };
 
 static void vce_v3_0_set_ring_funcs(struct amdgpu_device *adev)