lavu: VAAPI hwcontext implementation

Signed-off-by: Anton Khirnov <anton@khirnov.net>

9 files changed, 948 insertions, 1 deletions

diff --git a/configure b/configure
index 9ffb67903d..5376e672f4 100755
--- a/configure
+++ b/configure
@@ -4681,6 +4681,10 @@ if enabled x11grab; then
     require Xfixes X11/extensions/Xfixes.h XFixesGetCursorImage -lXfixes
 fi
 
+enabled vaapi &&
+    check_code cc "va/va.h" "vaCreateSurfaces(0, 0, 0, 0, 0, 0, 0, 0)" ||
+    disable vaapi
+
 enabled vaapi && enabled xlib &&
     check_lib2 "va/va.h va/va_x11.h" vaGetDisplay -lva -lva-x11 &&
     enable vaapi_x11
diff --git a/doc/APIchanges b/doc/APIchanges
index 4effe0a1f6..bce85049b9 100644
--- a/doc/APIchanges
+++ b/doc/APIchanges
@@ -13,6 +13,9 @@ libavutil:     2015-08-28
 
 API changes, most recent first:
 
+2016-xx-xx - xxxxxxx - lavu 55.9.0 - hwcontext_vaapi.h
+  Add new installed header with VAAPI-specific hwcontext definitions.
+
 2016-xx-xx - xxxxxxx - lavu 55.8.0 - pixfmt.h
   Deprecate all AV_PIX_FMT_VAAPI_* formats.
   Replaced by AV_PIX_FMT_VAAPI.
diff --git a/libavutil/Makefile b/libavutil/Makefile
index a095f0dd6e..0956703ce8 100644
--- a/libavutil/Makefile
+++ b/libavutil/Makefile
@@ -25,6 +25,7 @@ HEADERS = adler32.h                                                     \
           hmac.h                                                        \
           hwcontext.h                                                   \
           hwcontext_cuda.h                                              \
+          hwcontext_vaapi.h                                             \
           hwcontext_vdpau.h                                             \
           imgutils.h                                                    \
           intfloat.h                                                    \
@@ -108,11 +109,13 @@ OBJS = adler32.o                                                        \
 
 OBJS-$(CONFIG_LZO)                      += lzo.o
 OBJS-$(CONFIG_CUDA)                     += hwcontext_cuda.o
+OBJS-$(CONFIG_VAAPI)                    += hwcontext_vaapi.o
 OBJS-$(CONFIG_VDPAU)                    += hwcontext_vdpau.o
 
 OBJS += $(COMPAT_OBJS:%=../compat/%)
 
 SKIPHEADERS-$(CONFIG_CUDA)             += hwcontext_cuda.h
+SKIPHEADERS-$(CONFIG_VAAPI)            += hwcontext_vaapi.h
 SKIPHEADERS-$(CONFIG_VDPAU)            += hwcontext_vdpau.h
 SKIPHEADERS-$(HAVE_ATOMICS_GCC)        += atomic_gcc.h
 SKIPHEADERS-$(HAVE_ATOMICS_SUNCC)      += atomic_suncc.h
diff --git a/libavutil/hwcontext.c b/libavutil/hwcontext.c
index 53e11b8eae..ac1e2c9d67 100644
--- a/libavutil/hwcontext.c
+++ b/libavutil/hwcontext.c
@@ -32,6 +32,9 @@ static const HWContextType *hw_table[] = {
 #if CONFIG_CUDA
     &ff_hwcontext_type_cuda,
 #endif
+#if CONFIG_VAAPI
+    &ff_hwcontext_type_vaapi,
+#endif
 #if CONFIG_VDPAU
     &ff_hwcontext_type_vdpau,
 #endif
diff --git a/libavutil/hwcontext.h b/libavutil/hwcontext.h
index 681b55566c..8502342409 100644
--- a/libavutil/hwcontext.h
+++ b/libavutil/hwcontext.h
@@ -27,6 +27,7 @@
 enum AVHWDeviceType {
     AV_HWDEVICE_TYPE_VDPAU,
     AV_HWDEVICE_TYPE_CUDA,
+    AV_HWDEVICE_TYPE_VAAPI,
 };
 
 typedef struct AVHWDeviceInternal AVHWDeviceInternal;
diff --git a/libavutil/hwcontext_internal.h b/libavutil/hwcontext_internal.h
index 27de1f9e62..b7828c0350 100644
--- a/libavutil/hwcontext_internal.h
+++ b/libavutil/hwcontext_internal.h
@@ -97,6 +97,7 @@ struct AVHWFramesInternal {
 };
 
 extern const HWContextType ff_hwcontext_type_cuda;
+extern const HWContextType ff_hwcontext_type_vaapi;
 extern const HWContextType ff_hwcontext_type_vdpau;
 
 #endif /* AVUTIL_HWCONTEXT_INTERNAL_H */
diff --git a/libavutil/hwcontext_vaapi.c b/libavutil/hwcontext_vaapi.c
new file mode 100644
index 0000000000..1a385ba514
--- /dev/null
+++ b/libavutil/hwcontext_vaapi.c
@@ -0,0 +1,850 @@
+/*
+ * This file is part of Libav.
+ *
+ * Libav is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * Libav is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with Libav; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#include "avassert.h"
+#include "buffer.h"
+#include "common.h"
+#include "hwcontext.h"
+#include "hwcontext_internal.h"
+#include "hwcontext_vaapi.h"
+#include "mem.h"
+#include "pixdesc.h"
+#include "pixfmt.h"
+
+typedef struct VAAPISurfaceFormat {
+    enum AVPixelFormat pix_fmt;
+    VAImageFormat image_format;
+} VAAPISurfaceFormat;
+
+typedef struct VAAPIDeviceContext {
+    // Surface formats which can be used with this device.
+    VAAPISurfaceFormat *formats;
+    int              nb_formats;
+} VAAPIDeviceContext;
+
+typedef struct VAAPIFramesContext {
+    // Surface attributes set at create time.
+    VASurfaceAttrib *attributes;
+    int           nb_attributes;
+    // RT format of the underlying surface (Intel driver ignores this anyway).
+    unsigned int rt_format;
+    // Whether vaDeriveImage works.
+    int derive_works;
+} VAAPIFramesContext;
+
+enum {
+    VAAPI_MAP_READ   = 0x01,
+    VAAPI_MAP_WRITE  = 0x02,
+    VAAPI_MAP_DIRECT = 0x04,
+};
+
+typedef struct VAAPISurfaceMap {
+    // The source hardware frame of this mapping (with hw_frames_ctx set).
+    const AVFrame *source;
+    // VAAPI_MAP_* flags which apply to this mapping.
+    int flags;
+    // Handle to the derived or copied image which is mapped.
+    VAImage image;
+} VAAPISurfaceMap;
+
+#define MAP(va, rt, av) { \
+        VA_FOURCC_ ## va, \
+        VA_RT_FORMAT_ ## rt, \
+        AV_PIX_FMT_ ## av \
+    }
+// The map fourcc <-> pix_fmt isn't bijective because of the annoying U/V
+// plane swap cases.  The frame handling below tries to hide these.
+static struct {
+    unsigned int fourcc;
+    unsigned int rt_format;
+    enum AVPixelFormat pix_fmt;
+} vaapi_format_map[] = {
+    MAP(NV12, YUV420,  NV12),
+    MAP(YV12, YUV420,  YUV420P), // With U/V planes swapped.
+    MAP(IYUV, YUV420,  YUV420P),
+  //MAP(I420, YUV420,  YUV420P), // Not in libva but used by Intel driver.
+#ifdef VA_FOURCC_YV16
+    MAP(YV16, YUV422,  YUV422P), // With U/V planes swapped.
+#endif
+    MAP(422H, YUV422,  YUV422P),
+    MAP(UYVY, YUV422,  UYVY422),
+    MAP(YUY2, YUV422,  YUYV422),
+    MAP(Y800, YUV400,  GRAY8),
+#ifdef VA_FOURCC_P010
+  //MAP(P010, YUV420_10BPP, P010),
+#endif
+    MAP(BGRA, RGB32,   BGRA),
+  //MAP(BGRX, RGB32,   BGR0),
+    MAP(RGBA, RGB32,   RGBA),
+  //MAP(RGBX, RGB32,   RGB0),
+    MAP(ABGR, RGB32,   ABGR),
+  //MAP(XBGR, RGB32,   0BGR),
+    MAP(ARGB, RGB32,   ARGB),
+  //MAP(XRGB, RGB32,   0RGB),
+};
+#undef MAP
+
+static enum AVPixelFormat vaapi_pix_fmt_from_fourcc(unsigned int fourcc)
+{
+    int i;
+    for (i = 0; i < FF_ARRAY_ELEMS(vaapi_format_map); i++)
+        if (vaapi_format_map[i].fourcc == fourcc)
+            return vaapi_format_map[i].pix_fmt;
+    return AV_PIX_FMT_NONE;
+}
+
+static int vaapi_get_image_format(AVHWDeviceContext *hwdev,
+                                  enum AVPixelFormat pix_fmt,
+                                  VAImageFormat **image_format)
+{
+    VAAPIDeviceContext *ctx = hwdev->internal->priv;
+    int i;
+
+    for (i = 0; i < ctx->nb_formats; i++) {
+        if (ctx->formats[i].pix_fmt == pix_fmt) {
+            *image_format = &ctx->formats[i].image_format;
+            return 0;
+        }
+    }
+    return AVERROR(EINVAL);
+}
+
+static int vaapi_frames_get_constraints(AVHWDeviceContext *hwdev,
+                                        const void *hwconfig,
+                                        AVHWFramesConstraints *constraints)
+{
+    AVVAAPIDeviceContext *hwctx = hwdev->hwctx;
+    const AVVAAPIHWConfig *config = hwconfig;
+    AVVAAPIHWConfig *tmp_config;
+    VASurfaceAttrib *attr_list = NULL;
+    VAStatus vas;
+    enum AVPixelFormat pix_fmt;
+    unsigned int fourcc;
+    int err, i, j, attr_count, pix_fmt_count;
+
+    if (!hwconfig) {
+        // No configuration was provided, so we create a temporary pipeline
+        // configuration in order to query all supported image formats.
+
+        tmp_config = av_mallocz(sizeof(*config));
+        if (!tmp_config)
+            return AVERROR(ENOMEM);
+
+        vas = vaCreateConfig(hwctx->display,
+                             VAProfileNone, VAEntrypointVideoProc,
+                             NULL, 0, &tmp_config->config_id);
+        if (vas != VA_STATUS_SUCCESS) {
+            // No vpp.  We might still be able to do something useful if
+            // codecs are supported, so try to make the most-commonly
+            // supported decoder configuration we can to query instead.
+            vas = vaCreateConfig(hwctx->display,
+                                 VAProfileH264ConstrainedBaseline,
+                                 VAEntrypointVLD, NULL, 0,
+                                 &tmp_config->config_id);
+            if (vas != VA_STATUS_SUCCESS) {
+                av_freep(&tmp_config);
+                return AVERROR(ENOSYS);
+            }
+        }
+
+        config = tmp_config;
+    }
+
+    attr_count = 0;
+    vas = vaQuerySurfaceAttributes(hwctx->display, config->config_id,
+                                   0, &attr_count);
+    if (vas != VA_STATUS_SUCCESS) {
+        av_log(hwdev, AV_LOG_ERROR, "Failed to query surface attributes: "
+               "%d (%s).\n", vas, vaErrorStr(vas));
+        err = AVERROR(ENOSYS);
+        goto fail;
+    }
+
+    attr_list = av_malloc(attr_count * sizeof(*attr_list));
+    if (!attr_list) {
+        err = AVERROR(ENOMEM);
+        goto fail;
+    }
+
+    vas = vaQuerySurfaceAttributes(hwctx->display, config->config_id,
+                                   attr_list, &attr_count);
+    if (vas != VA_STATUS_SUCCESS) {
+        av_log(hwdev, AV_LOG_ERROR, "Failed to query surface attributes: "
+               "%d (%s).\n", vas, vaErrorStr(vas));
+        err = AVERROR(ENOSYS);
+        goto fail;
+    }
+
+    pix_fmt_count = 0;
+    for (i = 0; i < attr_count; i++) {
+        switch (attr_list[i].type) {
+        case VASurfaceAttribPixelFormat:
+            fourcc = attr_list[i].value.value.i;
+            pix_fmt = vaapi_pix_fmt_from_fourcc(fourcc);
+            if (pix_fmt != AV_PIX_FMT_NONE) {
+                ++pix_fmt_count;
+            } else {
+                // Something unsupported - ignore.
+            }
+            break;
+        case VASurfaceAttribMinWidth:
+            constraints->min_width  = attr_list[i].value.value.i;
+            break;
+        case VASurfaceAttribMinHeight:
+            constraints->min_height = attr_list[i].value.value.i;
+            break;
+        case VASurfaceAttribMaxWidth:
+            constraints->max_width  = attr_list[i].value.value.i;
+            break;
+        case VASurfaceAttribMaxHeight:
+            constraints->max_height = attr_list[i].value.value.i;
+            break;
+        }
+    }
+    if (pix_fmt_count == 0) {
+        // Nothing usable found.  Presumably there exists something which
+        // works, so leave the set null to indicate unknown.
+        constraints->valid_sw_formats = NULL;
+    } else {
+        constraints->valid_sw_formats = av_malloc_array(pix_fmt_count + 1,
+                                                        sizeof(pix_fmt));
+        if (!constraints->valid_sw_formats) {
+            err = AVERROR(ENOMEM);
+            goto fail;
+        }
+
+        for (i = j = 0; i < attr_count; i++) {
+            if (attr_list[i].type != VASurfaceAttribPixelFormat)
+                continue;
+            fourcc = attr_list[i].value.value.i;
+            pix_fmt = vaapi_pix_fmt_from_fourcc(fourcc);
+            if (pix_fmt != AV_PIX_FMT_NONE)
+                constraints->valid_sw_formats[j++] = pix_fmt;
+        }
+        av_assert0(j == pix_fmt_count);
+        constraints->valid_sw_formats[j] = AV_PIX_FMT_NONE;
+    }
+
+    constraints->valid_hw_formats = av_malloc_array(2, sizeof(pix_fmt));
+    if (!constraints->valid_hw_formats) {
+        err = AVERROR(ENOMEM);
+        goto fail;
+    }
+    constraints->valid_hw_formats[0] = AV_PIX_FMT_VAAPI;
+    constraints->valid_hw_formats[1] = AV_PIX_FMT_NONE;
+
+    err = 0;
+fail:
+    av_freep(&attr_list);
+    if (!hwconfig) {
+        vaDestroyConfig(hwctx->display, tmp_config->config_id);
+        av_freep(&tmp_config);
+    }
+    return err;
+}
+
+static int vaapi_device_init(AVHWDeviceContext *hwdev)
+{
+    VAAPIDeviceContext *ctx = hwdev->internal->priv;
+    AVVAAPIDeviceContext *hwctx = hwdev->hwctx;
+    AVHWFramesConstraints *constraints = NULL;
+    VAImageFormat *image_list = NULL;
+    VAStatus vas;
+    int err, i, j, image_count;
+    enum AVPixelFormat pix_fmt;
+    unsigned int fourcc;
+
+    constraints = av_mallocz(sizeof(*constraints));
+    if (!constraints)
+        goto fail;
+
+    err = vaapi_frames_get_constraints(hwdev, NULL, constraints);
+    if (err < 0)
+        goto fail;
+
+    image_count = vaMaxNumImageFormats(hwctx->display);
+    if (image_count <= 0) {
+        err = AVERROR(EIO);
+        goto fail;
+    }
+    image_list = av_malloc(image_count * sizeof(*image_list));
+    if (!image_list) {
+        err = AVERROR(ENOMEM);
+        goto fail;
+    }
+    vas = vaQueryImageFormats(hwctx->display, image_list, &image_count);
+    if (vas != VA_STATUS_SUCCESS) {
+        err = AVERROR(EIO);
+        goto fail;
+    }
+
+    ctx->formats  = av_malloc(image_count * sizeof(*ctx->formats));
+    if (!ctx->formats) {
+        err = AVERROR(ENOMEM);
+        goto fail;
+    }
+    ctx->nb_formats = 0;
+    for (i = 0; i < image_count; i++) {
+        fourcc  = image_list[i].fourcc;
+        pix_fmt = vaapi_pix_fmt_from_fourcc(fourcc);
+        for (j = 0; constraints->valid_sw_formats[j] != AV_PIX_FMT_NONE; j++) {
+            if (pix_fmt == constraints->valid_sw_formats[j])
+                break;
+        }
+        if (constraints->valid_sw_formats[j] != AV_PIX_FMT_NONE) {
+            av_log(hwdev, AV_LOG_DEBUG, "Format %#x -> %s.\n",
+                   fourcc, av_get_pix_fmt_name(pix_fmt));
+            ctx->formats[ctx->nb_formats].pix_fmt      = pix_fmt;
+            ctx->formats[ctx->nb_formats].image_format = image_list[i];
+            ++ctx->nb_formats;
+        } else {
+            av_log(hwdev, AV_LOG_DEBUG, "Format %#x -> unknown.\n", fourcc);
+        }
+    }
+
+    av_free(image_list);
+    av_hwframe_constraints_free(&constraints);
+    return 0;
+fail:
+    av_freep(&ctx->formats);
+    av_free(image_list);
+    av_hwframe_constraints_free(&constraints);
+    return err;
+}
+
+static void vaapi_device_uninit(AVHWDeviceContext *hwdev)
+{
+    VAAPIDeviceContext *ctx = hwdev->internal->priv;
+
+    av_freep(&ctx->formats);
+}
+
+static void vaapi_buffer_free(void *opaque, uint8_t *data)
+{
+    AVHWFramesContext     *hwfc = opaque;
+    AVVAAPIDeviceContext *hwctx = hwfc->device_ctx->hwctx;
+    VASurfaceID surface_id;
+    VAStatus vas;
+
+    surface_id = (VASurfaceID)(uintptr_t)data;
+
+    vas = vaDestroySurfaces(hwctx->display, &surface_id, 1);
+    if (vas != VA_STATUS_SUCCESS) {
+        av_log(hwfc, AV_LOG_ERROR, "Failed to destroy surface %#x: "
+               "%d (%s).\n", surface_id, vas, vaErrorStr(vas));
+    }
+}
+
+static AVBufferRef *vaapi_pool_alloc(void *opaque, int size)
+{
+    AVHWFramesContext     *hwfc = opaque;
+    VAAPIFramesContext     *ctx = hwfc->internal->priv;
+    AVVAAPIDeviceContext *hwctx = hwfc->device_ctx->hwctx;
+    AVVAAPIFramesContext  *avfc = hwfc->hwctx;
+    VASurfaceID surface_id;
+    VAStatus vas;
+    AVBufferRef *ref;
+
+    vas = vaCreateSurfaces(hwctx->display, ctx->rt_format,
+                           hwfc->width, hwfc->height,
+                           &surface_id, 1,
+                           ctx->attributes, ctx->nb_attributes);
+    if (vas != VA_STATUS_SUCCESS) {
+        av_log(hwfc, AV_LOG_ERROR, "Failed to create surface: "
+               "%d (%s).\n", vas, vaErrorStr(vas));
+        return NULL;
+    }
+    av_log(hwfc, AV_LOG_DEBUG, "Created surface %#x.\n", surface_id);
+
+    ref = av_buffer_create((uint8_t*)(uintptr_t)surface_id,
+                           sizeof(surface_id), &vaapi_buffer_free,
+                           hwfc, AV_BUFFER_FLAG_READONLY);
+    if (!ref) {
+        vaDestroySurfaces(hwctx->display, &surface_id, 1);
+        return NULL;
+    }
+
+    if (hwfc->initial_pool_size > 0) {
+        // This is a fixed-size pool, so we must still be in the initial
+        // allocation sequence.
+        av_assert0(avfc->nb_surfaces < hwfc->initial_pool_size);
+        avfc->surface_ids[avfc->nb_surfaces] = surface_id;
+        ++avfc->nb_surfaces;
+    }
+
+    return ref;
+}
+
+static int vaapi_frames_init(AVHWFramesContext *hwfc)
+{
+    AVVAAPIFramesContext  *avfc = hwfc->hwctx;
+    VAAPIFramesContext     *ctx = hwfc->internal->priv;
+    AVVAAPIDeviceContext *hwctx = hwfc->device_ctx->hwctx;
+    VAImageFormat *expected_format;
+    AVBufferRef *test_surface = NULL;
+    VASurfaceID test_surface_id;
+    VAImage test_image;
+    VAStatus vas;
+    int err, i;
+    unsigned int fourcc, rt_format;
+
+    for (i = 0; i < FF_ARRAY_ELEMS(vaapi_format_map); i++) {
+        if (vaapi_format_map[i].pix_fmt == hwfc->sw_format) {
+            fourcc    = vaapi_format_map[i].fourcc;
+            rt_format = vaapi_format_map[i].rt_format;
+            break;
+        }
+    }
+    if (i >= FF_ARRAY_ELEMS(vaapi_format_map)) {
+        av_log(hwfc, AV_LOG_ERROR, "Unsupported format: %s.\n",
+               av_get_pix_fmt_name(hwfc->sw_format));
+        return AVERROR(EINVAL);
+    }
+
+    if (!hwfc->pool) {
+        int need_memory_type = 1, need_pixel_format = 1;
+        for (i = 0; i < avfc->nb_attributes; i++) {
+            if (ctx->attributes[i].type == VASurfaceAttribMemoryType)
+                need_memory_type  = 0;
+            if (ctx->attributes[i].type == VASurfaceAttribPixelFormat)
+                need_pixel_format = 0;
+        }
+        ctx->nb_attributes =
+            avfc->nb_attributes + need_memory_type + need_pixel_format;
+
+        ctx->attributes = av_malloc(ctx->nb_attributes *
+                                        sizeof(*ctx->attributes));
+        if (!ctx->attributes) {
+            err = AVERROR(ENOMEM);
+            goto fail;
+        }
+
+        for (i = 0; i < avfc->nb_attributes; i++)
+            ctx->attributes[i] = avfc->attributes[i];
+        if (need_memory_type) {
+            ctx->attributes[i++] = (VASurfaceAttrib) {
+                .type          = VASurfaceAttribMemoryType,
+                .flags         = VA_SURFACE_ATTRIB_SETTABLE,
+                .value.type    = VAGenericValueTypeInteger,
+                .value.value.i = VA_SURFACE_ATTRIB_MEM_TYPE_VA,
+            };
+        }
+        if (need_pixel_format) {
+            ctx->attributes[i++] = (VASurfaceAttrib) {
+                .type          = VASurfaceAttribPixelFormat,
+                .flags         = VA_SURFACE_ATTRIB_SETTABLE,
+                .value.type    = VAGenericValueTypeInteger,
+                .value.value.i = fourcc,
+            };
+        }
+        av_assert0(i == ctx->nb_attributes);
+
+        ctx->rt_format = rt_format;
+
+        if (hwfc->initial_pool_size > 0) {
+            // This pool will be usable as a render target, so we need to store
+            // all of the surface IDs somewhere that vaCreateContext() calls
+            // will be able to access them.
+            avfc->nb_surfaces = 0;
+            avfc->surface_ids = av_malloc(hwfc->initial_pool_size *
+                                          sizeof(*avfc->surface_ids));
+            if (!avfc->surface_ids) {
+                err = AVERROR(ENOMEM);
+                goto fail;
+            }
+        } else {
+            // This pool allows dynamic sizing, and will not be usable as a
+            // render target.
+            avfc->nb_surfaces = 0;
+            avfc->surface_ids = NULL;
+        }
+
+        hwfc->internal->pool_internal =
+            av_buffer_pool_init2(sizeof(VASurfaceID), hwfc,
+                                 &vaapi_pool_alloc, NULL);
+        if (!hwfc->internal->pool_internal) {
+            av_log(hwfc, AV_LOG_ERROR, "Failed to create VAAPI surface pool.\n");
+            err = AVERROR(ENOMEM);
+            goto fail;
+        }
+    }
+
+    // Allocate a single surface to test whether vaDeriveImage() is going
+    // to work for the specific configuration.
+    if (hwfc->pool) {
+        test_surface = av_buffer_pool_get(hwfc->pool);
+        if (!test_surface) {
+            av_log(hwfc, AV_LOG_ERROR, "Unable to allocate a surface from "
+                   "user-configured buffer pool.\n");
+            err = AVERROR(ENOMEM);
+            goto fail;
+        }
+    } else {
+        test_surface = av_buffer_pool_get(hwfc->internal->pool_internal);
+        if (!test_surface) {
+            av_log(hwfc, AV_LOG_ERROR, "Unable to allocate a surface from "
+                   "internal buffer pool.\n");
+            err = AVERROR(ENOMEM);
+            goto fail;
+        }
+    }
+    test_surface_id = (VASurfaceID)(uintptr_t)test_surface->data;
+
+    ctx->derive_works = 0;
+
+    err = vaapi_get_image_format(hwfc->device_ctx,
+                                 hwfc->sw_format, &expected_format);
+    if (err == 0) {
+        vas = vaDeriveImage(hwctx->display, test_surface_id, &test_image);
+        if (vas == VA_STATUS_SUCCESS) {
+            if (expected_format->fourcc == test_image.format.fourcc) {
+                av_log(hwfc, AV_LOG_DEBUG, "Direct mapping possible.\n");
+                ctx->derive_works = 1;
+            } else {
+                av_log(hwfc, AV_LOG_DEBUG, "Direct mapping disabled: "
+                       "derived image format %08x does not match "
+                       "expected format %08x.\n",
+                       expected_format->fourcc, test_image.format.fourcc);
+            }
+            vaDestroyImage(hwctx->display, test_image.image_id);
+        } else {
+            av_log(hwfc, AV_LOG_DEBUG, "Direct mapping disabled: "
+                   "deriving image does not work: "
+                   "%d (%s).\n", vas, vaErrorStr(vas));
+        }
+    } else {
+        av_log(hwfc, AV_LOG_DEBUG, "Direct mapping disabled: "
+               "image format is not supported.\n");
+    }
+
+    av_buffer_unref(&test_surface);
+    return 0;
+
+fail:
+    av_buffer_unref(&test_surface);
+    av_freep(&avfc->surface_ids);
+    av_freep(&ctx->attributes);
+    return err;
+}
+
+static void vaapi_frames_uninit(AVHWFramesContext *hwfc)
+{
+    AVVAAPIFramesContext *avfc = hwfc->hwctx;
+    VAAPIFramesContext    *ctx = hwfc->internal->priv;
+
+    av_freep(&avfc->surface_ids);
+    av_freep(&ctx->attributes);
+}
+
+static int vaapi_get_buffer(AVHWFramesContext *hwfc, AVFrame *frame)
+{
+    frame->buf[0] = av_buffer_pool_get(hwfc->pool);
+    if (!frame->buf[0])
+        return AVERROR(ENOMEM);
+
+    frame->data[3] = frame->buf[0]->data;
+    frame->format  = AV_PIX_FMT_VAAPI;
+    frame->width   = hwfc->width;
+    frame->height  = hwfc->height;
+
+    return 0;
+}
+
+static int vaapi_transfer_get_formats(AVHWFramesContext *hwfc,
+                                      enum AVHWFrameTransferDirection dir,
+                                      enum AVPixelFormat **formats)
+{
+    VAAPIDeviceContext *ctx = hwfc->device_ctx->internal->priv;
+    enum AVPixelFormat *pix_fmts, preferred_format;
+    int i, k;
+
+    preferred_format = hwfc->sw_format;
+
+    pix_fmts = av_malloc((ctx->nb_formats + 1) * sizeof(*pix_fmts));
+    if (!pix_fmts)
+        return AVERROR(ENOMEM);
+
+    pix_fmts[0] = preferred_format;
+    k = 1;
+    for (i = 0; i < ctx->nb_formats; i++) {
+        if (ctx->formats[i].pix_fmt == preferred_format)
+            continue;
+        av_assert0(k < ctx->nb_formats);
+        pix_fmts[k++] = ctx->formats[i].pix_fmt;
+    }
+    av_assert0(k == ctx->nb_formats);
+    pix_fmts[k] = AV_PIX_FMT_NONE;
+
+    *formats = pix_fmts;
+    return 0;
+}
+
+static void vaapi_unmap_frame(void *opaque, uint8_t *data)
+{
+    AVHWFramesContext *hwfc = opaque;
+    AVVAAPIDeviceContext *hwctx = hwfc->device_ctx->hwctx;
+    VAAPISurfaceMap *map = (VAAPISurfaceMap*)data;
+    const AVFrame *src;
+    VASurfaceID surface_id;
+    VAStatus vas;
+
+    src = map->source;
+    surface_id = (VASurfaceID)(uintptr_t)src->data[3];
+    av_log(hwfc, AV_LOG_DEBUG, "Unmap surface %#x.\n", surface_id);
+
+    vas = vaUnmapBuffer(hwctx->display, map->image.buf);
+    if (vas != VA_STATUS_SUCCESS) {
+        av_log(hwfc, AV_LOG_ERROR, "Failed to unmap image from surface "
+               "%#x: %d (%s).\n", surface_id, vas, vaErrorStr(vas));
+    }
+
+    if ((map->flags & VAAPI_MAP_WRITE) &&
+        !(map->flags & VAAPI_MAP_DIRECT)) {
+        vas = vaPutImage(hwctx->display, surface_id, map->image.image_id,
+                         0, 0, hwfc->width, hwfc->height,
+                         0, 0, hwfc->width, hwfc->height);
+        if (vas != VA_STATUS_SUCCESS) {
+            av_log(hwfc, AV_LOG_ERROR, "Failed to write image to surface "
+                   "%#x: %d (%s).\n", surface_id, vas, vaErrorStr(vas));
+        }
+    }
+
+    vas = vaDestroyImage(hwctx->display, map->image.image_id);
+    if (vas != VA_STATUS_SUCCESS) {
+        av_log(hwfc, AV_LOG_ERROR, "Failed to destroy image from surface "
+               "%#x: %d (%s).\n", surface_id, vas, vaErrorStr(vas));
+    }
+
+    av_free(map);
+}
+
+static int vaapi_map_frame(AVHWFramesContext *hwfc,
+                           AVFrame *dst, const AVFrame *src, int flags)
+{
+    AVVAAPIDeviceContext *hwctx = hwfc->device_ctx->hwctx;
+    VAAPIFramesContext *ctx = hwfc->internal->priv;
+    VASurfaceID surface_id;
+    VAImageFormat *image_format;
+    VAAPISurfaceMap *map;
+    VAStatus vas;
+    void *address = NULL;
+    int err, i;
+
+    surface_id = (VASurfaceID)(uintptr_t)src->data[3];
+    av_log(hwfc, AV_LOG_DEBUG, "Map surface %#x.\n", surface_id);
+
+    if (!ctx->derive_works && (flags & VAAPI_MAP_DIRECT)) {
+        // Requested direct mapping but it is not possible.
+        return AVERROR(EINVAL);
+    }
+    if (dst->format == AV_PIX_FMT_NONE)
+        dst->format = hwfc->sw_format;
+    if (dst->format != hwfc->sw_format && (flags & VAAPI_MAP_DIRECT)) {
+        // Requested direct mapping but the formats do not match.
+        return AVERROR(EINVAL);
+    }
+
+    err = vaapi_get_image_format(hwfc->device_ctx, dst->format, &image_format);
+    if (err < 0) {
+        // Requested format is not a valid output format.
+        return AVERROR(EINVAL);
+    }
+
+    map = av_malloc(sizeof(VAAPISurfaceMap));
+    if (!map)
+        return AVERROR(ENOMEM);
+
+    map->source         = src;
+    map->flags          = flags;
+    map->image.image_id = VA_INVALID_ID;
+
+    vas = vaSyncSurface(hwctx->display, surface_id);
+    if (vas != VA_STATUS_SUCCESS) {
+        av_log(hwfc, AV_LOG_ERROR, "Failed to sync surface "
+               "%#x: %d (%s).\n", surface_id, vas, vaErrorStr(vas));
+        err = AVERROR(EIO);
+        goto fail;
+    }
+
+    // The memory which we map using derive need not be connected to the CPU
+    // in a way conducive to fast access.  On Gen7-Gen9 Intel graphics, the
+    // memory is mappable but not cached, so normal memcpy()-like access is
+    // very slow to read it (but writing is ok).  It is possible to read much
+    // faster with a copy routine which is aware of the limitation, but we
+    // assume for now that the user is not aware of that and would therefore
+    // prefer not to be given direct-mapped memory if they request read access.
+    if (ctx->derive_works &&
+        ((flags & VAAPI_MAP_DIRECT) || !(flags & VAAPI_MAP_READ))) {
+        vas = vaDeriveImage(hwctx->display, surface_id, &map->image);
+        if (vas != VA_STATUS_SUCCESS) {
+            av_log(hwfc, AV_LOG_ERROR, "Failed to derive image from "
+                   "surface %#x: %d (%s).\n",
+                   surface_id, vas, vaErrorStr(vas));
+            err = AVERROR(EIO);
+            goto fail;
+        }
+        if (map->image.format.fourcc != image_format->fourcc) {
+            av_log(hwfc, AV_LOG_ERROR, "Derive image of surface %#x "
+                   "is in wrong format: expected %#08x, got %#08x.\n",
+                   surface_id, image_format->fourcc, map->image.format.fourcc);
+            err = AVERROR(EIO);
+            goto fail;
+        }
+        map->flags |= VAAPI_MAP_DIRECT;
+    } else {
+        vas = vaCreateImage(hwctx->display, image_format,
+                            hwfc->width, hwfc->height, &map->image);
+        if (vas != VA_STATUS_SUCCESS) {
+            av_log(hwfc, AV_LOG_ERROR, "Failed to create image for "
+                   "surface %#x: %d (%s).\n",
+                   surface_id, vas, vaErrorStr(vas));
+            err = AVERROR(EIO);
+            goto fail;
+        }
+        if (flags & VAAPI_MAP_READ) {
+            vas = vaGetImage(hwctx->display, surface_id, 0, 0,
+                             hwfc->width, hwfc->height, map->image.image_id);
+            if (vas != VA_STATUS_SUCCESS) {
+                av_log(hwfc, AV_LOG_ERROR, "Failed to read image from "
+                       "surface %#x: %d (%s).\n",
+                       surface_id, vas, vaErrorStr(vas));
+                err = AVERROR(EIO);
+                goto fail;
+            }
+        }
+    }
+
+    vas = vaMapBuffer(hwctx->display, map->image.buf, &address);
+    if (vas != VA_STATUS_SUCCESS) {
+        av_log(hwfc, AV_LOG_ERROR, "Failed to map image from surface "
+               "%#x: %d (%s).\n", surface_id, vas, vaErrorStr(vas));
+        err = AVERROR(EIO);
+        goto fail;
+    }
+
+    dst->width  = src->width;
+    dst->height = src->height;
+
+    for (i = 0; i < map->image.num_planes; i++) {
+        dst->data[i] = (uint8_t*)address + map->image.offsets[i];
+        dst->linesize[i] = map->image.pitches[i];
+    }
+    if (
+#ifdef VA_FOURCC_YV16
+        map->image.format.fourcc == VA_FOURCC_YV16 ||
+#endif
+        map->image.format.fourcc == VA_FOURCC_YV12) {
+        // Chroma planes are YVU rather than YUV, so swap them.
+        FFSWAP(uint8_t*, dst->data[1], dst->data[2]);
+    }
+
+    dst->buf[0] = av_buffer_create((uint8_t*)map, sizeof(*map),
+                                   &vaapi_unmap_frame, hwfc, 0);
+    if (!dst->buf[0]) {
+        err = AVERROR(ENOMEM);
+        goto fail;
+    }
+
+    return 0;
+
+fail:
+    if (map) {
+        if (address)
+            vaUnmapBuffer(hwctx->display, map->image.buf);
+        if (map->image.image_id != VA_INVALID_ID)
+            vaDestroyImage(hwctx->display, map->image.image_id);
+        av_free(map);
+    }
+    return err;
+}
+
+static int vaapi_transfer_data_from(AVHWFramesContext *hwfc,
+                                    AVFrame *dst, const AVFrame *src)
+{
+    AVFrame *map;
+    int err;
+
+    map = av_frame_alloc();
+    if (!map)
+        return AVERROR(ENOMEM);
+    map->format = dst->format;
+
+    err = vaapi_map_frame(hwfc, map, src, VAAPI_MAP_READ);
+    if (err)
+        goto fail;
+
+    err = av_frame_copy(dst, map);
+    if (err)
+        goto fail;
+
+    err = 0;
+fail:
+    av_frame_free(&map);
+    return err;
+}
+
+static int vaapi_transfer_data_to(AVHWFramesContext *hwfc,
+                                  AVFrame *dst, const AVFrame *src)
+{
+    AVFrame *map;
+    int err;
+
+    map = av_frame_alloc();
+    if (!map)
+        return AVERROR(ENOMEM);
+    map->format = src->format;
+
+    err = vaapi_map_frame(hwfc, map, dst, VAAPI_MAP_WRITE);
+    if (err)
+        goto fail;
+
+    err = av_frame_copy(map, src);
+    if (err)
+        goto fail;
+
+    err = 0;
+fail:
+    av_frame_free(&map);
+    return err;
+}
+
+const HWContextType ff_hwcontext_type_vaapi = {
+    .type                   = AV_HWDEVICE_TYPE_VAAPI,
+    .name                   = "VAAPI",
+
+    .device_hwctx_size      = sizeof(AVVAAPIDeviceContext),
+    .device_priv_size       = sizeof(VAAPIDeviceContext),
+    .device_hwconfig_size   = sizeof(AVVAAPIHWConfig),
+    .frames_hwctx_size      = sizeof(AVVAAPIFramesContext),
+    .frames_priv_size       = sizeof(VAAPIFramesContext),
+
+    .device_init            = &vaapi_device_init,
+    .device_uninit          = &vaapi_device_uninit,
+    .frames_get_constraints = &vaapi_frames_get_constraints,
+    .frames_init            = &vaapi_frames_init,
+    .frames_uninit          = &vaapi_frames_uninit,
+    .frames_get_buffer      = &vaapi_get_buffer,
+    .transfer_get_formats   = &vaapi_transfer_get_formats,
+    .transfer_data_to       = &vaapi_transfer_data_to,
+    .transfer_data_from     = &vaapi_transfer_data_from,
+
+    .pix_fmts = (const enum AVPixelFormat[]) {
+        AV_PIX_FMT_VAAPI,
+        AV_PIX_FMT_NONE
+    },
+};
diff --git a/libavutil/hwcontext_vaapi.h b/libavutil/hwcontext_vaapi.h
new file mode 100644
index 0000000000..1c87f5db5e
--- /dev/null
+++ b/libavutil/hwcontext_vaapi.h
@@ -0,0 +1,82 @@
+/*
+ * This file is part of Libav.
+ *
+ * Libav is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * Libav is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with Libav; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#ifndef AVUTIL_HWCONTEXT_VAAPI_H
+#define AVUTIL_HWCONTEXT_VAAPI_H
+
+#include <va/va.h>
+
+/**
+ * @file
+ * API-specific header for AV_HWDEVICE_TYPE_VAAPI.
+ *
+ * Dynamic frame pools are supported, but note that any pool used as a render
+ * target is required to be of fixed size in order to be be usable as an
+ * argument to vaCreateContext().
+ *
+ * For user-allocated pools, AVHWFramesContext.pool must return AVBufferRefs
+ * with the data pointer set to a VASurfaceID.
+ */
+
+/**
+ * VAAPI connection details.
+ *
+ * Allocated as AVHWDeviceContext.hwctx
+ */
+typedef struct AVVAAPIDeviceContext {
+    /**
+     * The VADisplay handle, to be filled by the user.
+     */
+    VADisplay display;
+} AVVAAPIDeviceContext;
+
+/**
+ * VAAPI-specific data associated with a frame pool.
+ *
+ * Allocated as AVHWFramesContext.hwctx.
+ */
+typedef struct AVVAAPIFramesContext {
+    /**
+     * Set by the user to apply surface attributes to all surfaces in
+     * the frame pool.  If null, default settings are used.
+     */
+    VASurfaceAttrib *attributes;
+    int           nb_attributes;
+    /**
+     * The surfaces IDs of all surfaces in the pool after creation.
+     * Only valid if AVHWFramesContext.initial_pool_size was positive.
+     * These are intended to be used as the render_targets arguments to
+     * vaCreateContext().
+     */
+    VASurfaceID     *surface_ids;
+    int           nb_surfaces;
+} AVVAAPIFramesContext;
+
+/**
+ * VAAPI hardware pipeline configuration details.
+ *
+ * Allocated with av_hwdevice_hwconfig_alloc().
+ */
+typedef struct AVVAAPIHWConfig {
+    /**
+     * ID of a VAAPI pipeline configuration.
+     */
+    VAConfigID config_id;
+} AVVAAPIHWConfig;
+
+#endif /* AVUTIL_HWCONTEXT_VAAPI_H */
diff --git a/libavutil/version.h b/libavutil/version.h
index 72eb5f4605..fdd27e3f0b 100644
--- a/libavutil/version.h
+++ b/libavutil/version.h
@@ -54,7 +54,7 @@
  */
 
 #define LIBAVUTIL_VERSION_MAJOR 55
-#define LIBAVUTIL_VERSION_MINOR  8
+#define LIBAVUTIL_VERSION_MINOR  9
 #define LIBAVUTIL_VERSION_MICRO  0
 
 #define LIBAVUTIL_VERSION_INT   AV_VERSION_INT(LIBAVUTIL_VERSION_MAJOR, \