Mesa (master): lavapipe: enable pipeline stats queries

[GitLab Mirror]

Module: Mesa
Branch: master
Commit: 87c70f1984fc6f9be6c32559ecfbde1946ff7360
URL:    http://cgit.freedesktop.org/mesa/mesa/commit/?id=87c70f1984fc6f9be6c32559ecfbde1946ff7360

Author: Dave Airlie <airlied at redhat.com>
Date:   Fri Nov 13 17:06:31 2020 +1000

lavapipe: enable pipeline stats queries

These pass CTS, but I think are missing some stuff CTS doesn't test.

This is one of the base Vulkan 1.0 features and I'd like to support
it for conformance.

Cc: "20.3" <mesa-stable at lists.freedesktop.org>
Reviewed-by: Adam Jackson <ajax at redhat.com>
Reviewed-by: Roland Scheidegger <sroland at vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7705>

---

 src/gallium/frontends/lavapipe/lvp_device.c  |  2 +-
 src/gallium/frontends/lavapipe/lvp_private.h |  1 +
 src/gallium/frontends/lavapipe/lvp_query.c   | 51 +++++++++++++++++++++++-----
 3 files changed, 44 insertions(+), 10 deletions(-)

diff --git a/src/gallium/frontends/lavapipe/lvp_device.c b/src/gallium/frontends/lavapipe/lvp_device.c
index 893c8a6af67..45734f95880 100644
--- a/src/gallium/frontends/lavapipe/lvp_device.c
+++ b/src/gallium/frontends/lavapipe/lvp_device.c
@@ -319,7 +319,7 @@ void lvp_GetPhysicalDeviceFeatures(
       .textureCompressionASTC_LDR               = false,
       .textureCompressionBC                     = true,
       .occlusionQueryPrecise                    = true,
-      .pipelineStatisticsQuery                  = false,
+      .pipelineStatisticsQuery                  = true,
       .vertexPipelineStoresAndAtomics           = (pdevice->pscreen->get_shader_param(pdevice->pscreen, PIPE_SHADER_VERTEX, PIPE_SHADER_CAP_MAX_SHADER_BUFFERS) != 0),
       .fragmentStoresAndAtomics                 = (pdevice->pscreen->get_shader_param(pdevice->pscreen, PIPE_SHADER_FRAGMENT, PIPE_SHADER_CAP_MAX_SHADER_BUFFERS) != 0),
       .shaderTessellationAndGeometryPointSize   = true,
diff --git a/src/gallium/frontends/lavapipe/lvp_private.h b/src/gallium/frontends/lavapipe/lvp_private.h
index e4300875bca..700875292d0 100644
--- a/src/gallium/frontends/lavapipe/lvp_private.h
+++ b/src/gallium/frontends/lavapipe/lvp_private.h
@@ -562,6 +562,7 @@ struct lvp_query_pool {
    struct vk_object_base base;
    VkQueryType type;
    uint32_t count;
+   VkQueryPipelineStatisticFlags pipeline_stats;
    enum pipe_query_type base_type;
    struct pipe_query *queries[0];
 };
diff --git a/src/gallium/frontends/lavapipe/lvp_query.c b/src/gallium/frontends/lavapipe/lvp_query.c
index f549fe1b89e..f807234ffb2 100644
--- a/src/gallium/frontends/lavapipe/lvp_query.c
+++ b/src/gallium/frontends/lavapipe/lvp_query.c
@@ -40,6 +40,9 @@ VkResult lvp_CreateQueryPool(
    case VK_QUERY_TYPE_TIMESTAMP:
       pipeq = PIPE_QUERY_TIMESTAMP;
       break;
+   case VK_QUERY_TYPE_PIPELINE_STATISTICS:
+      pipeq = PIPE_QUERY_PIPELINE_STATISTICS;
+      break;
    default:
       return VK_ERROR_FEATURE_NOT_PRESENT;
    }
@@ -57,6 +60,7 @@ VkResult lvp_CreateQueryPool(
    pool->type = pCreateInfo->queryType;
    pool->count = pCreateInfo->queryCount;
    pool->base_type = pipeq;
+   pool->pipeline_stats = pCreateInfo->pipelineStatistics;
 
    *pQueryPool = lvp_query_pool_to_handle(pool);
    return VK_SUCCESS;
@@ -112,17 +116,46 @@ VkResult lvp_GetQueryPoolResults(
       if (!ready && !(flags & VK_QUERY_RESULT_PARTIAL_BIT))
           vk_result = VK_NOT_READY;
       if (flags & VK_QUERY_RESULT_64_BIT) {
-         if (ready || (flags & VK_QUERY_RESULT_PARTIAL_BIT))
-            *(uint64_t *)dptr = result.u64;
-         dptr += 8;
+         if (ready || (flags & VK_QUERY_RESULT_PARTIAL_BIT)) {
+            if (pool->type == VK_QUERY_TYPE_PIPELINE_STATISTICS) {
+               uint32_t mask = pool->pipeline_stats;
+               uint64_t *pstats = (uint64_t *)&result.pipeline_statistics;
+               while (mask) {
+                  uint32_t i = u_bit_scan(&mask);
+
+                  *(uint64_t *)dptr = pstats[i];
+                  dptr += 8;
+               }
+            } else {
+               *(uint64_t *)dptr = result.u64;
+               dptr += 8;
+            }
+         } else
+            dptr += stride;
+
       } else {
          if (ready || (flags & VK_QUERY_RESULT_PARTIAL_BIT)) {
-            if (result.u64 > UINT32_MAX)
-               *(uint32_t *)dptr = UINT32_MAX;
-            else
-               *(uint32_t *)dptr = result.u32;
-         }
-         dptr += 4;
+            if (pool->type == VK_QUERY_TYPE_PIPELINE_STATISTICS) {
+               uint32_t mask = pool->pipeline_stats;
+               uint64_t *pstats = (uint64_t *)&result.pipeline_statistics;
+               while (mask) {
+                  uint32_t i = u_bit_scan(&mask);
+
+                  if (pstats[i] > UINT32_MAX)
+                     *(uint32_t *)dptr = UINT32_MAX;
+                  else
+                     *(uint32_t *)dptr = pstats[i];
+                  dptr += 4;
+               }
+            } else {
+               if (result.u64 > UINT32_MAX)
+                  *(uint32_t *)dptr = UINT32_MAX;
+               else
+                  *(uint32_t *)dptr = result.u32;
+               dptr += 4;
+            }
+         } else
+            dptr += stride;
       }
 
       if (flags & VK_QUERY_RESULT_WITH_AVAILABILITY_BIT) {