vo_gpu: hwdec_cuda: Implement interop for placebo

This change updates the vulkan interop code to work with the
libplacebo based ra_vk, but also introduces direct VkImage
sharing to avoid the use of the intermediate buffer.

It is also necessary and desirable to introduce explicit
semaphore bsed synchronisation for operations on the shared
images.

Synchronisation means we can safely reuse the same VkImage for every
mapped frame, by ensuring the frame is copied to the VkImage before
mapping the next frame.

This functionality requires a 417.xx or newer nvidia driver, due to
bugs in the VkImage interop in the earlier 411 and 415 drivers.

It's definitely worth the effort, as the raw throughput is about
twice that of implementation using an intermediate buffer.

video/out/opengl/hwdec_cuda.c

@@ -39,11 +39,7 @@
 #include "ra_gl.h"
 #endif
 #if HAVE_VULKAN
-/*
+#include "video/out/placebo/ra_pl.h"
-#include "video/out/vulkan/formats.h"
-#include "video/out/vulkan/ra_vk.h"
-#include "video/out/vulkan/utils.h"
-*/
 #endif
 
 #if HAVE_WIN32_DESKTOP
@@ -58,16 +54,20 @@ struct priv_owner {
 
     bool is_gl;
     bool is_vk;
+
+    enum pl_handle_type handle_type;
 };
 
-struct ext_buf {
+struct ext_vk {
-#if HAVE_WIN32_DESKTOP
-    HANDLE handle;
-#else
-    int fd;
-#endif
     CUexternalMemory mem;
-    CUdeviceptr buf;
+    CUmipmappedArray mma;
+
+    const struct pl_tex *pltex;
+
+    const struct pl_sync *sync;
+
+    CUexternalSemaphore ss;
+    CUexternalSemaphore ws;
 };
 
 struct priv {
@@ -77,8 +77,7 @@ struct priv {
 
     CUcontext display_ctx;
 
-    struct ra_buf_params buf_params[4];
+    struct ext_vk evk[4];
-    struct ra_buf_pool buf_pool[4];
 };
 
 static int check_cu(struct ra_hwdec *hw, CUresult err, const char *func)
@@ -127,17 +126,26 @@ static int cuda_init(struct ra_hwdec *hw)
 #endif
 
 #if HAVE_VULKAN
-    return -1; // TODO: reimplement
+    p->handle_type =
-    /*
+#if HAVE_WIN32_DESKTOP
-    p->is_vk = ra_vk_get(hw->ra) != NULL;
+        IsWindows8OrGreater() ? PL_HANDLE_WIN32 : PL_HANDLE_WIN32_KMT;
+#else
+        PL_HANDLE_FD;
+#endif
+
+    const struct pl_gpu *gpu = ra_pl_get(hw->ra);
+    p->is_vk = gpu != NULL;
     if (p->is_vk) {
-        if (!ra_vk_get(hw->ra)->has_ext_external_memory_export) {
+        if (!(gpu->export_caps.tex & p->handle_type)) {
-            MP_VERBOSE(hw, "CUDA hwdec with Vulkan requires the %s extension\n",
+            MP_VERBOSE(hw, "CUDA hwdec with Vulkan requires exportable texture memory of type 0x%X.\n",
-                       MP_VK_EXTERNAL_MEMORY_EXPORT_EXTENSION_NAME);
+                       p->handle_type);
+            return -1;
+        } else if (!(gpu->export_caps.sync & p->handle_type)) {
+            MP_VERBOSE(hw, "CUDA hwdec with Vulkan requires exportable semaphores of type 0x%X.\n",
+                       p->handle_type);
             return -1;
         }
     }
-    */
 #endif
 
     if (!p->is_gl && !p->is_vk) {
@@ -202,12 +210,6 @@ static int cuda_init(struct ra_hwdec *hw)
         }
     } else if (p->is_vk) {
 #if HAVE_VULKAN
-        /*
-        uint8_t vk_uuid[VK_UUID_SIZE];
-        struct mpvk_ctx *vk = ra_vk_get(hw->ra);
-
-        mpvk_get_phys_device_uuid(vk, vk_uuid);
-
         int count;
         ret = CHECK_CU(cu->cuDeviceGetCount(&count));
         if (ret < 0)
@@ -225,7 +227,7 @@ static int cuda_init(struct ra_hwdec *hw)
             if (ret < 0)
                 continue;
 
-            if (memcmp(vk_uuid, uuid.bytes, VK_UUID_SIZE) == 0) {
+            if (memcmp(gpu->uuid, uuid.bytes, sizeof (gpu->uuid)) == 0) {
                 display_dev = dev;
                 break;
             }
@@ -242,7 +244,6 @@ static int cuda_init(struct ra_hwdec *hw)
             return -1;
 
         p->decode_ctx = p->display_ctx;
-        */
 #endif
     }
 
@@ -300,105 +301,188 @@ static void cuda_uninit(struct ra_hwdec *hw)
 #define CHECK_CU(x) check_cu((mapper)->owner, (x), #x)
 
 #if HAVE_VULKAN
-/*
+static bool cuda_ext_vk_init(struct ra_hwdec_mapper *mapper,
-static struct ra_buf *cuda_buf_pool_get(struct ra_hwdec_mapper *mapper, int n)
+                             const struct ra_format *format, int n)
 {
     struct priv_owner *p_owner = mapper->owner->priv;
     struct priv *p = mapper->priv;
     CudaFunctions *cu = p_owner->cu;
+    int mem_fd = -1, wait_fd = -1, signal_fd = -1;
     int ret = 0;
 
-    struct ra_buf_pool *pool = &p->buf_pool[n];
+    struct ext_vk *evk = &p->evk[n];
-    struct ra_buf *buf = ra_buf_pool_get(mapper->ra, pool, &p->buf_params[n]);
+
-    if (!buf) {
+    const struct pl_gpu *gpu = ra_pl_get(mapper->ra);
+
+    struct pl_tex_params tex_params = {
+        .w = mp_image_plane_w(&p->layout, n),
+        .h = mp_image_plane_h(&p->layout, n),
+        .d = 0,
+        .format = ra_pl_fmt_get(format),
+        .sampleable = true,
+        .sample_mode = format->linear_filter ? PL_TEX_SAMPLE_LINEAR
+                                             : PL_TEX_SAMPLE_NEAREST,
+        .export_handle = p_owner->handle_type,
+    };
+
+    evk->pltex = pl_tex_create(gpu, &tex_params);
+    if (!evk->pltex) {
         goto error;
     }
 
-    if (!ra_vk_buf_get_user_data(buf)) {
+    struct ra_tex *ratex = talloc_ptrtype(NULL, ratex);
-        struct ext_buf *ebuf = talloc_zero(NULL, struct ext_buf);
+    ret = mppl_wrap_tex(mapper->ra, evk->pltex, ratex);
-        struct vk_external_mem mem_info;
+    if (!ret) {
+        pl_tex_destroy(gpu, &evk->pltex);
+        talloc_free(ratex);
+        goto error;
+    }
+    mapper->tex[n] = ratex;
 
-        bool success = ra_vk_buf_get_external_info(mapper->ra, buf, &mem_info);
+#if !HAVE_WIN32_DESKTOP
-        if (!success) {
+    if (evk->pltex->params.export_handle == PL_HANDLE_FD) {
-            ret = -1;
+        mem_fd = dup(evk->pltex->shared_mem.handle.fd);
+        if (mem_fd < 0) {
             goto error;
         }
-
-#if HAVE_WIN32_DESKTOP
-        ebuf->handle = mem_info.mem_handle;
-        MP_DBG(mapper, "vk_external_info[%d][%d]: %p %zu %zu\n", n, pool->index, ebuf->handle, mem_info.size, mem_info.offset);
-#else
-        ebuf->fd = mem_info.mem_fd;
-        MP_DBG(mapper, "vk_external_info[%d][%d]: %d %zu %zu\n", n, pool->index, ebuf->fd, mem_info.size, mem_info.offset);
-#endif
-
-        CUDA_EXTERNAL_MEMORY_HANDLE_DESC ext_desc = {
-#if HAVE_WIN32_DESKTOP
-            .type = IsWindows8OrGreater()
-                ? CU_EXTERNAL_MEMORY_HANDLE_TYPE_OPAQUE_WIN32
-                : CU_EXTERNAL_MEMORY_HANDLE_TYPE_OPAQUE_WIN32_KMT,
-            .handle.win32.handle = ebuf->handle,
-#else
-            .type = CU_EXTERNAL_MEMORY_HANDLE_TYPE_OPAQUE_FD,
-            .handle.fd = ebuf->fd,
-#endif
-            .size = mem_info.mem_size,
-            .flags = 0,
-        };
-        ret = CHECK_CU(cu->cuImportExternalMemory(&ebuf->mem, &ext_desc));
-        if (ret < 0)
-            goto error;
-
-        CUDA_EXTERNAL_MEMORY_BUFFER_DESC buf_desc = {
-            .offset = mem_info.offset,
-            .size = mem_info.size,
-            .flags = 0,
-        };
-        ret = CHECK_CU(cu->cuExternalMemoryGetMappedBuffer(&ebuf->buf, ebuf->mem, &buf_desc));
-        if (ret < 0)
-            goto error;
-
-        ra_vk_buf_set_user_data(buf, ebuf);
     }
-    return buf;
+#endif
+
+    CUDA_EXTERNAL_MEMORY_HANDLE_DESC ext_desc = {
+#if HAVE_WIN32_DESKTOP
+        .type = IsWindows8OrGreater()
+            ? CU_EXTERNAL_MEMORY_HANDLE_TYPE_OPAQUE_WIN32
+            : CU_EXTERNAL_MEMORY_HANDLE_TYPE_OPAQUE_WIN32_KMT,
+        .handle.win32.handle = evk->pltex->shared_mem.handle.handle,
+#else
+        .type = CU_EXTERNAL_MEMORY_HANDLE_TYPE_OPAQUE_FD,
+        .handle.fd = mem_fd,
+#endif
+        .size = evk->pltex->shared_mem.size,
+        .flags = 0,
+    };
+    ret = CHECK_CU(cu->cuImportExternalMemory(&evk->mem, &ext_desc));
+    if (ret < 0)
+        goto error;
+    // CUDA takes ownership of imported memory
+    mem_fd = -1;
+
+    CUarray_format cufmt;
+    switch (format->pixel_size / format->num_components) {
+    case 1:
+        cufmt = CU_AD_FORMAT_UNSIGNED_INT8;
+        break;
+    case 2:
+        cufmt = CU_AD_FORMAT_UNSIGNED_INT16;
+        break;
+    default:
+        ret = -1;
+        goto error;
+    }
+
+    CUDA_EXTERNAL_MEMORY_MIPMAPPED_ARRAY_DESC tex_desc = {
+        .offset = evk->pltex->shared_mem.offset,
+        .arrayDesc = {
+            .Width = mp_image_plane_w(&p->layout, n),
+            .Height = mp_image_plane_h(&p->layout, n),
+            .Depth = 0,
+            .Format = cufmt,
+            .NumChannels = format->num_components,
+            .Flags = 0,
+        },
+        .numLevels = 1,
+    };
+
+    ret = CHECK_CU(cu->cuExternalMemoryGetMappedMipmappedArray(&evk->mma, evk->mem, &tex_desc));
+    if (ret < 0)
+        goto error;
+
+    ret = CHECK_CU(cu->cuMipmappedArrayGetLevel(&p->cu_array[n], evk->mma, 0));
+    if (ret < 0)
+        goto error;
+
+    evk->sync = pl_sync_create(gpu, p_owner->handle_type);
+    if (!evk->sync) {
+        ret = -1;
+        goto error;
+    }
+
+#if !HAVE_WIN32_DESKTOP
+    if (evk->sync->handle_type == PL_HANDLE_FD) {
+        wait_fd = dup(evk->sync->wait_handle.fd);
+        signal_fd = dup(evk->sync->signal_handle.fd);
+    }
+#endif
+
+    CUDA_EXTERNAL_SEMAPHORE_HANDLE_DESC w_desc = {
+#if HAVE_WIN32_DESKTOP
+        .type = IsWindows8OrGreater()
+            ? CU_EXTERNAL_MEMORY_HANDLE_TYPE_OPAQUE_WIN32
+            : CU_EXTERNAL_MEMORY_HANDLE_TYPE_OPAQUE_WIN32_KMT,
+        .handle.win32.handle = evk->sync->wait_handle.handle,
+#else
+        .type = CU_EXTERNAL_MEMORY_HANDLE_TYPE_OPAQUE_FD,
+        .handle.fd = wait_fd,
+#endif
+    };
+    ret = CHECK_CU(cu->cuImportExternalSemaphore(&evk->ws, &w_desc));
+    if (ret < 0)
+        goto error;
+    wait_fd = -1;
+
+    CUDA_EXTERNAL_SEMAPHORE_HANDLE_DESC s_desc = {
+#if HAVE_WIN32_DESKTOP
+        .type = IsWindows8OrGreater()
+            ? CU_EXTERNAL_MEMORY_HANDLE_TYPE_OPAQUE_WIN32
+            : CU_EXTERNAL_MEMORY_HANDLE_TYPE_OPAQUE_WIN32_KMT,
+        .handle.win32.handle = evk->sync->signal_handle.handle,
+#else
+        .type = CU_EXTERNAL_MEMORY_HANDLE_TYPE_OPAQUE_FD,
+        .handle.fd = signal_fd,
+#endif
+    };
+
+    ret = CHECK_CU(cu->cuImportExternalSemaphore(&evk->ss, &s_desc));
+    if (ret < 0)
+        goto error;
+    signal_fd = -1;
+
+    return true;
 
 error:
-    MP_ERR(mapper, "cuda_buf_pool_get failed\n");
+    MP_ERR(mapper, "cuda_ext_vk_init failed\n");
-    return NULL;
+    if (mem_fd > -1)
+        close(mem_fd);
+    if (wait_fd > -1)
+        close(wait_fd);
+    if (signal_fd > -1)
+        close(signal_fd);
+    return false;
 }
 
-static void cuda_buf_pool_uninit(struct ra_hwdec_mapper *mapper, int n)
+static void cuda_ext_vk_uninit(struct ra_hwdec_mapper *mapper, int n)
 {
     struct priv_owner *p_owner = mapper->owner->priv;
     struct priv *p = mapper->priv;
     CudaFunctions *cu = p_owner->cu;
 
-    struct ra_buf_pool *pool = &p->buf_pool[n];
+    struct ext_vk *evk = &p->evk[n];
-    for (int i = 0; i < pool->num_buffers; i++) {
+    if (evk) {
-        struct ra_buf *buf = pool->buffers[i];
+        if (evk->mma) {
-        struct ext_buf *ebuf = ra_vk_buf_get_user_data(buf);
+            CHECK_CU(cu->cuMipmappedArrayDestroy(evk->mma));
-        if (ebuf) {
-            if (ebuf->mem > 0) {
-                CHECK_CU(cu->cuDestroyExternalMemory(ebuf->mem));
-#if HAVE_WIN32_DESKTOP
-            }
-            if (ebuf->handle) {
-                // Handle must always be closed by us.
-                CloseHandle(ebuf->handle);
-            }
-#else
-            } else if (ebuf->fd > -1) {
-                // fd should only be closed if external memory was not imported
-                close(ebuf->fd);
-            }
-#endif
         }
-        talloc_free(ebuf);
+        if (evk->mem) {
-        ra_vk_buf_set_user_data(buf, NULL);
+            CHECK_CU(cu->cuDestroyExternalMemory(evk->mem));
+        }
+        if (evk->ss) {
+            CHECK_CU(cu->cuDestroyExternalSemaphore(evk->ss));
+        }
+        if (evk->ws) {
+            CHECK_CU(cu->cuDestroyExternalSemaphore(evk->ws));
+        }
+        pl_sync_destroy(ra_pl_get(mapper->ra), &evk->sync);
     }
-    ra_buf_pool_uninit(mapper->ra, pool);
 }
-*/
 #endif // HAVE_VULKAN
 
 static int mapper_init(struct ra_hwdec_mapper *mapper)
@@ -431,24 +515,24 @@ static int mapper_init(struct ra_hwdec_mapper *mapper)
     for (int n = 0; n < desc.num_planes; n++) {
         const struct ra_format *format = desc.planes[n];
 
-        struct ra_tex_params params = {
-            .dimensions = 2,
-            .w = mp_image_plane_w(&p->layout, n),
-            .h = mp_image_plane_h(&p->layout, n),
-            .d = 1,
-            .format = format,
-            .render_src = true,
-            .src_linear = format->linear_filter,
-        };
-
-        mapper->tex[n] = ra_tex_create(mapper->ra, &params);
-        if (!mapper->tex[n]) {
-            ret = -1;
-            goto error;
-        }
-
         if (p_owner->is_gl) {
 #if HAVE_GL
+            struct ra_tex_params params = {
+                .dimensions = 2,
+                .w = mp_image_plane_w(&p->layout, n),
+                .h = mp_image_plane_h(&p->layout, n),
+                .d = 1,
+                .format = format,
+                .render_src = true,
+                .src_linear = format->linear_filter,
+            };
+
+            mapper->tex[n] = ra_tex_create(mapper->ra, &params);
+            if (!mapper->tex[n]) {
+                ret = -1;
+                goto error;
+            }
+
             GLuint texture;
             GLenum target;
             ra_gl_get_raw_tex(mapper->ra, mapper->tex[n], &texture, &target);
@@ -472,13 +556,11 @@ static int mapper_init(struct ra_hwdec_mapper *mapper)
                 goto error;
 #endif
         } else if (p_owner->is_vk) {
-            struct ra_buf_params buf_params = {
+#if HAVE_VULKAN
-                .type = RA_BUF_TYPE_SHARED_MEMORY,
+            ret = cuda_ext_vk_init(mapper, format, n);
-                .size = mp_image_plane_h(&p->layout, n) *
+            if (ret < 0)
-                        mp_image_plane_w(&p->layout, n) *
+                goto error;
-                        mapper->tex[n]->params.format->pixel_size,
+#endif
-            };
-            p->buf_params[n] = buf_params;
         }
     }
 
@@ -503,11 +585,10 @@ static void mapper_uninit(struct ra_hwdec_mapper *mapper)
         if (p->cu_res[n] > 0)
             CHECK_CU(cu->cuGraphicsUnregisterResource(p->cu_res[n]));
         p->cu_res[n] = 0;
-        ra_tex_free(mapper->ra, &mapper->tex[n]);
-
 #if HAVE_VULKAN
-        //cuda_buf_pool_uninit(mapper, n);
+        cuda_ext_vk_uninit(mapper, n);
 #endif
+        ra_tex_free(mapper->ra, &mapper->tex[n]);
     }
     CHECK_CU(cu->cuCtxPopCurrent(&dummy));
 }
@@ -523,7 +604,6 @@ static int mapper_map(struct ra_hwdec_mapper *mapper)
     CudaFunctions *cu = p_owner->cu;
     CUcontext dummy;
     int ret = 0, eret = 0;
-    bool is_gl = p_owner->is_gl;
     bool is_vk = p_owner->is_vk;
 
     ret = CHECK_CU(cu->cuCtxPushCurrent(p->display_ctx));
@@ -531,45 +611,44 @@ static int mapper_map(struct ra_hwdec_mapper *mapper)
         return ret;
 
     for (int n = 0; n < p->layout.num_planes; n++) {
-        struct ra_buf *buf = NULL;
+#if HAVE_VULKAN
+        if (is_vk) {
+            ret = pl_tex_export(ra_pl_get(mapper->ra),
+                                p->evk[n].pltex, p->evk[n].sync);
+            if (!ret)
+                goto error;
 
+            CUDA_EXTERNAL_SEMAPHORE_WAIT_PARAMS wp = { 0, };
+            ret = CHECK_CU(cu->cuWaitExternalSemaphoresAsync(&p->evk[n].ws,
+                                                             &wp, 1, 0));
+            if (ret < 0)
+                goto error;
+        }
+#endif
         CUDA_MEMCPY2D cpy = {
             .srcMemoryType = CU_MEMORYTYPE_DEVICE,
             .srcDevice     = (CUdeviceptr)mapper->src->planes[n],
             .srcPitch      = mapper->src->stride[n],
             .srcY          = 0,
+            .dstMemoryType = CU_MEMORYTYPE_ARRAY,
+            .dstArray      = p->cu_array[n],
             .WidthInBytes  = mp_image_plane_w(&p->layout, n) *
                              mapper->tex[n]->params.format->pixel_size,
             .Height        = mp_image_plane_h(&p->layout, n),
         };
 
-        if (is_gl) {
+        ret = CHECK_CU(cu->cuMemcpy2DAsync(&cpy, 0));
-            cpy.dstMemoryType = CU_MEMORYTYPE_ARRAY;
-            cpy.dstArray = p->cu_array[n];
-        } else if (is_vk) {
-#if HAVE_VULKAN
-            buf = cuda_buf_pool_get(mapper, n);
-            struct ext_buf *ebuf = ra_vk_buf_get_user_data(buf);
-
-            cpy.dstMemoryType = CU_MEMORYTYPE_DEVICE;
-            cpy.dstDevice = ebuf->buf;
-            cpy.dstPitch  = mp_image_plane_w(&p->layout, n) *
-                            mapper->tex[n]->params.format->pixel_size;
-#endif
-        }
-
-        ret = CHECK_CU(cu->cuMemcpy2D(&cpy));
         if (ret < 0)
             goto error;
-
+#if HAVE_VULKAN
         if (is_vk) {
-            struct ra_tex_upload_params params = {
+            CUDA_EXTERNAL_SEMAPHORE_SIGNAL_PARAMS sp = { 0, };
-                .tex = mapper->tex[n],
+            ret = CHECK_CU(cu->cuSignalExternalSemaphoresAsync(&p->evk[n].ss,
-                .invalidate = true,
+                                                               &sp, 1, 0));
-                .buf = buf,
+            if (ret < 0)
-            };
+                goto error;
-            mapper->ra->fns->tex_upload(mapper->ra, &params);
         }
+#endif
     }
 
  error:

wscript

@@ -852,7 +852,7 @@ hwaccel_features = [
     }, {
         'name': 'ffnvcodec',
         'desc': 'CUDA Headers and dynamic loader',
-        'func': check_pkg_config('ffnvcodec >= 8.2.15.3'),
+        'func': check_pkg_config('ffnvcodec >= 8.2.15.7'),
     }, {
         'name': '--cuda-hwaccel',
         'desc': 'CUDA hwaccel',