vaapi_encode: Refactor initialisation

This allows better checking of capabilities and will make it easier
to add more functionality later.

It also commonises some duplicated code around rate control setup
and adds more comments explaining the internals.

(cherry picked from commit 80a5d051)

libavcodec/vaapi_encode.h

@@ -48,15 +48,6 @@ enum {
     PICTURE_TYPE_B   = 3,
 };
 
-enum {
-    // All encode operations are done independently.
-    ISSUE_MODE_SERIALISE_EVERYTHING = 0,
-    // Overlap as many operations as possible.
-    ISSUE_MODE_MAXIMISE_THROUGHPUT,
-    // Overlap operations only when satisfying parallel dependencies.
-    ISSUE_MODE_MINIMISE_LATENCY,
-};
-
 typedef struct VAAPIEncodeSlice {
     void           *priv_data;
     void           *codec_slice_params;
@@ -102,43 +93,65 @@ typedef struct VAAPIEncodeContext {
     // Codec-specific hooks.
     const struct VAAPIEncodeType *codec;
 
+    // Encoding profile (VAProfileXXX).
+    VAProfile       va_profile;
+    // Encoding entrypoint (usually VAEntryointEncSlice).
+    VAEntrypoint    va_entrypoint;
+    // Surface colour/sampling format (usually VA_RT_FORMAT_YUV420).
+    unsigned int    va_rt_format;
+    // Rate control mode.
+    unsigned int    va_rc_mode;
+
+    // The required size of surfaces.  This is probably the input
+    // size (AVCodecContext.width|height) aligned up to whatever
+    // block size is required by the codec.
+    int             surface_width;
+    int             surface_height;
+
+    // Everything above this point must be set before calling
+    // ff_vaapi_encode_init().
+
     // Codec-specific state.
     void *priv_data;
 
-    VAProfile       va_profile;
-    VAEntrypoint    va_entrypoint;
+    // Configuration attributes to use when creating va_config.
+    VAConfigAttrib  config_attributes[MAX_CONFIG_ATTRIBUTES];
+    int          nb_config_attributes;
+
     VAConfigID      va_config;
     VAContextID     va_context;
 
-    int             va_rc_mode;
-
     AVBufferRef    *device_ref;
     AVHWDeviceContext *device;
     AVVAAPIDeviceContext *hwctx;
 
+    // The hardware frame context containing the input frames.
     AVBufferRef    *input_frames_ref;
     AVHWFramesContext *input_frames;
 
-    // Input size, set from input frames.
-    int             input_width;
-    int             input_height;
-    // Aligned size, set by codec init, becomes hwframe size.
-    int             aligned_width;
-    int             aligned_height;
-
-    int          nb_recon_frames;
+    // The hardware frame context containing the reconstructed frames.
     AVBufferRef    *recon_frames_ref;
     AVHWFramesContext *recon_frames;
 
+    // Pool of (reusable) bitstream output buffers.
     AVBufferPool   *output_buffer_pool;
 
-    VAConfigAttrib  config_attributes[MAX_CONFIG_ATTRIBUTES];
-    int          nb_config_attributes;
-
+    // Global parameters which will be applied at the start of the
+    // sequence (includes rate control parameters below).
     VAEncMiscParameterBuffer *global_params[MAX_GLOBAL_PARAMS];
     size_t          global_params_size[MAX_GLOBAL_PARAMS];
     int          nb_global_params;
 
+    // Rate control parameters.
+    struct {
+        VAEncMiscParameterBuffer misc;
+        VAEncMiscParameterRateControl rc;
+    } rc_params;
+    struct {
+        VAEncMiscParameterBuffer misc;
+        VAEncMiscParameterHRD hrd;
+    } hrd_params;
+
     // Per-sequence parameter structure (VAEncSequenceParameterBuffer*).
     void           *codec_sequence_params;
 
@@ -158,7 +171,15 @@ typedef struct VAAPIEncodeContext {
     // Next output order index (encode order).
     int64_t         output_order;
 
-    int             issue_mode;
+    enum {
+        // All encode operations are done independently (synchronise
+        // immediately after every operation).
+        ISSUE_MODE_SERIALISE_EVERYTHING = 0,
+        // Overlap as many operations as possible.
+        ISSUE_MODE_MAXIMISE_THROUGHPUT,
+        // Overlap operations only when satisfying parallel dependencies.
+        ISSUE_MODE_MINIMISE_LATENCY,
+    } issue_mode;
 
     // Timestamp handling.
     int64_t         first_pts;
@@ -185,13 +206,18 @@ typedef struct VAAPIEncodeContext {
 typedef struct VAAPIEncodeType {
     size_t priv_data_size;
 
-    int  (*init)(AVCodecContext *avctx);
-    int (*close)(AVCodecContext *avctx);
+    // Perform any extra codec-specific configuration after the
+    // codec context is initialised (set up the private data and
+    // add any necessary global parameters).
+    int (*configure)(AVCodecContext *avctx);
 
+    // The size of the parameter structures:
+    // sizeof(VAEnc{type}ParameterBuffer{codec}).
     size_t sequence_params_size;
     size_t picture_params_size;
     size_t slice_params_size;
 
+    // Fill the parameter structures.
     int  (*init_sequence_params)(AVCodecContext *avctx);
     int   (*init_picture_params)(AVCodecContext *avctx,
                                  VAAPIEncodePicture *pic);
@@ -199,10 +225,13 @@ typedef struct VAAPIEncodeType {
                                  VAAPIEncodePicture *pic,
                                  VAAPIEncodeSlice *slice);
 
+    // The type used by the packed header: this should look like
+    // VAEncPackedHeader{something}.
     int sequence_header_type;
     int picture_header_type;
     int slice_header_type;
 
+    // Write the packed header data to the provided buffer.
     int (*write_sequence_header)(AVCodecContext *avctx,
                                  char *data, size_t *data_len);
     int  (*write_picture_header)(AVCodecContext *avctx,
@@ -213,10 +242,18 @@ typedef struct VAAPIEncodeType {
                                  VAAPIEncodeSlice *slice,
                                  char *data, size_t *data_len);
 
+    // Fill an extra parameter structure, which will then be
+    // passed to vaRenderPicture().  Will be called repeatedly
+    // with increasing index argument until AVERROR_EOF is
+    // returned.
     int    (*write_extra_buffer)(AVCodecContext *avctx,
                                  VAAPIEncodePicture *pic,
                                  int index, int *type,
                                  char *data, size_t *data_len);
+
+    // Write an extra packed header.  Will be called repeatedly
+    // with increasing index argument until AVERROR_EOF is
+    // returned.
     int    (*write_extra_header)(AVCodecContext *avctx,
                                  VAAPIEncodePicture *pic,
                                  int index, int *type,
@@ -227,8 +264,7 @@ typedef struct VAAPIEncodeType {
 int ff_vaapi_encode2(AVCodecContext *avctx, AVPacket *pkt,
                      const AVFrame *input_image, int *got_packet);
 
-int ff_vaapi_encode_init(AVCodecContext *avctx,
-                         const VAAPIEncodeType *type);
+int ff_vaapi_encode_init(AVCodecContext *avctx);
 int ff_vaapi_encode_close(AVCodecContext *avctx);
 
 #endif /* AVCODEC_VAAPI_ENCODE_H */

libavcodec/vaapi_encode_h265.c

@@ -798,8 +798,8 @@ static int vaapi_encode_h265_init_sequence_params(AVCodecContext *avctx)
         vseq->intra_idr_period = 0;
         vseq->ip_period = 0;
 
-        vseq->pic_width_in_luma_samples  = ctx->aligned_width;
-        vseq->pic_height_in_luma_samples = ctx->aligned_height;
+        vseq->pic_width_in_luma_samples  = ctx->surface_width;
+        vseq->pic_height_in_luma_samples = ctx->surface_height;
 
         vseq->seq_fields.bits.chroma_format_idc = 1; // 4:2:0.
         vseq->seq_fields.bits.separate_colour_plane_flag = 0;
@@ -911,15 +911,15 @@ static int vaapi_encode_h265_init_sequence_params(AVCodecContext *avctx)
         mseq->vps_poc_proportional_to_timing_flag = 1;
         mseq->vps_num_ticks_poc_diff_minus1 = 0;
 
-        if (ctx->input_width  != ctx->aligned_width ||
-            ctx->input_height != ctx->aligned_height) {
+        if (avctx->width  != ctx->surface_width ||
+            avctx->height != ctx->surface_height) {
             mseq->conformance_window_flag = 1;
             mseq->conf_win_left_offset   = 0;
             mseq->conf_win_right_offset  =
-                (ctx->aligned_width - ctx->input_width) / 2;
+                (ctx->surface_width - avctx->width) / 2;
             mseq->conf_win_top_offset    = 0;
             mseq->conf_win_bottom_offset =
-                (ctx->aligned_height - ctx->input_height) / 2;
+                (ctx->surface_height - avctx->height) / 2;
         } else {
             mseq->conformance_window_flag = 0;
         }
@@ -1154,68 +1154,20 @@ static int vaapi_encode_h265_init_slice_params(AVCodecContext *avctx,
     return 0;
 }
 
-static av_cold int vaapi_encode_h265_init_constant_bitrate(AVCodecContext *avctx)
+static av_cold int vaapi_encode_h265_configure(AVCodecContext *avctx)
 {
     VAAPIEncodeContext      *ctx = avctx->priv_data;
     VAAPIEncodeH265Context *priv = ctx->priv_data;
-    int hrd_buffer_size;
-    int hrd_initial_buffer_fullness;
-
-    if (avctx->bit_rate > INT32_MAX) {
-        av_log(avctx, AV_LOG_ERROR, "Target bitrate of 2^31 bps or "
-               "higher is not supported.\n");
-        return AVERROR(EINVAL);
-    }
-
-    if (avctx->rc_buffer_size)
-        hrd_buffer_size = avctx->rc_buffer_size;
-    else
-        hrd_buffer_size = avctx->bit_rate;
-    if (avctx->rc_initial_buffer_occupancy)
-        hrd_initial_buffer_fullness = avctx->rc_initial_buffer_occupancy;
-    else
-        hrd_initial_buffer_fullness = hrd_buffer_size * 3 / 4;
-
-    priv->rc_params.misc.type = VAEncMiscParameterTypeRateControl;
-    priv->rc_params.rc = (VAEncMiscParameterRateControl) {
-        .bits_per_second   = avctx->bit_rate,
-        .target_percentage = 66,
-        .window_size       = 1000,
-        .initial_qp        = (avctx->qmax >= 0 ? avctx->qmax : 40),
-        .min_qp            = (avctx->qmin >= 0 ? avctx->qmin : 20),
-        .basic_unit_size   = 0,
-    };
-    ctx->global_params[ctx->nb_global_params] =
-        &priv->rc_params.misc;
-    ctx->global_params_size[ctx->nb_global_params++] =
-        sizeof(priv->rc_params);
-
-    priv->hrd_params.misc.type = VAEncMiscParameterTypeHRD;
-    priv->hrd_params.hrd = (VAEncMiscParameterHRD) {
-        .initial_buffer_fullness = hrd_initial_buffer_fullness,
-        .buffer_size             = hrd_buffer_size,
-    };
-    ctx->global_params[ctx->nb_global_params] =
-        &priv->hrd_params.misc;
-    ctx->global_params_size[ctx->nb_global_params++] =
-        sizeof(priv->hrd_params);
-
-    // These still need to be  set for pic_init_qp/slice_qp_delta.
-    priv->fixed_qp_idr = 30;
-    priv->fixed_qp_p   = 30;
-    priv->fixed_qp_b   = 30;
+    VAAPIEncodeH265Options  *opt = ctx->codec_options;
 
-    av_log(avctx, AV_LOG_DEBUG, "Using constant-bitrate = %"PRId64" bps.\n",
-           avctx->bit_rate);
-    return 0;
-}
+    priv->ctu_width     = FFALIGN(ctx->surface_width,  32) / 32;
+    priv->ctu_height    = FFALIGN(ctx->surface_height, 32) / 32;
 
-static av_cold int vaapi_encode_h265_init_fixed_qp(AVCodecContext *avctx)
-{
-    VAAPIEncodeContext      *ctx = avctx->priv_data;
-    VAAPIEncodeH265Context *priv = ctx->priv_data;
-    VAAPIEncodeH265Options  *opt = ctx->codec_options;
+    av_log(avctx, AV_LOG_VERBOSE, "Input %ux%u -> Surface %ux%u -> CTU %ux%u.\n",
+           avctx->width, avctx->height, ctx->surface_width,
+           ctx->surface_height, priv->ctu_width, priv->ctu_height);
 
+    if (ctx->va_rc_mode == VA_RC_CQP) {
         priv->fixed_qp_p = opt->qp;
         if (avctx->i_quant_factor > 0.0)
             priv->fixed_qp_idr = (int)((priv->fixed_qp_p * avctx->i_quant_factor +
@@ -1231,79 +1183,27 @@ static av_cold int vaapi_encode_h265_init_fixed_qp(AVCodecContext *avctx)
         av_log(avctx, AV_LOG_DEBUG, "Using fixed QP = "
                "%d / %d / %d for IDR- / P- / B-frames.\n",
                priv->fixed_qp_idr, priv->fixed_qp_p, priv->fixed_qp_b);
-    return 0;
-}
 
-static av_cold int vaapi_encode_h265_init_internal(AVCodecContext *avctx)
-{
-    static const VAConfigAttrib default_config_attributes[] = {
-        { .type  = VAConfigAttribRTFormat,
-          .value = VA_RT_FORMAT_YUV420 },
-        { .type  = VAConfigAttribEncPackedHeaders,
-          .value = (VA_ENC_PACKED_HEADER_SEQUENCE |
-                    VA_ENC_PACKED_HEADER_SLICE) },
-    };
-
-    VAAPIEncodeContext      *ctx = avctx->priv_data;
-    VAAPIEncodeH265Context *priv = ctx->priv_data;
-    int i, err;
-
-    switch (avctx->profile) {
-    case FF_PROFILE_HEVC_MAIN:
-    case FF_PROFILE_UNKNOWN:
-        ctx->va_profile = VAProfileHEVCMain;
-        break;
-    case FF_PROFILE_HEVC_MAIN_10:
-        av_log(avctx, AV_LOG_ERROR, "H.265 main 10-bit profile "
-               "is not supported.\n");
-        return AVERROR_PATCHWELCOME;
-    default:
-        av_log(avctx, AV_LOG_ERROR, "Unknown H.265 profile %d.\n",
-               avctx->profile);
-        return AVERROR(EINVAL);
-    }
-    ctx->va_entrypoint = VAEntrypointEncSlice;
-
-    ctx->input_width    = avctx->width;
-    ctx->input_height   = avctx->height;
-    ctx->aligned_width  = FFALIGN(ctx->input_width,  16);
-    ctx->aligned_height = FFALIGN(ctx->input_height, 16);
-    priv->ctu_width     = FFALIGN(ctx->aligned_width,  32) / 32;
-    priv->ctu_height    = FFALIGN(ctx->aligned_height, 32) / 32;
-
-    av_log(avctx, AV_LOG_VERBOSE, "Input %ux%u -> Aligned %ux%u -> CTU %ux%u.\n",
-           ctx->input_width, ctx->input_height, ctx->aligned_width,
-           ctx->aligned_height, priv->ctu_width, priv->ctu_height);
+    } else if (ctx->va_rc_mode == VA_RC_CBR) {
+        // These still need to be  set for pic_init_qp/slice_qp_delta.
+        priv->fixed_qp_idr = 30;
+        priv->fixed_qp_p   = 30;
+        priv->fixed_qp_b   = 30;
 
-    for (i = 0; i < FF_ARRAY_ELEMS(default_config_attributes); i++) {
-        ctx->config_attributes[ctx->nb_config_attributes++] =
-            default_config_attributes[i];
-    }
+        av_log(avctx, AV_LOG_DEBUG, "Using constant-bitrate = %d bps.\n",
+               avctx->bit_rate);
 
-    if (avctx->bit_rate > 0) {
-        ctx->va_rc_mode = VA_RC_CBR;
-        err = vaapi_encode_h265_init_constant_bitrate(avctx);
     } else {
-        ctx->va_rc_mode = VA_RC_CQP;
-        err = vaapi_encode_h265_init_fixed_qp(avctx);
+        av_assert0(0 && "Invalid RC mode.");
     }
-    if (err < 0)
-        return err;
-
-    ctx->config_attributes[ctx->nb_config_attributes++] = (VAConfigAttrib) {
-        .type  = VAConfigAttribRateControl,
-        .value = ctx->va_rc_mode,
-    };
-
-    ctx->nb_recon_frames = 20;
 
     return 0;
 }
 
-static VAAPIEncodeType vaapi_encode_type_h265 = {
+static const VAAPIEncodeType vaapi_encode_type_h265 = {
     .priv_data_size        = sizeof(VAAPIEncodeH265Context),
 
-    .init                  = &vaapi_encode_h265_init_internal,
+    .configure             = &vaapi_encode_h265_configure,
 
     .sequence_params_size  = sizeof(VAEncSequenceParameterBufferHEVC),
     .init_sequence_params  = &vaapi_encode_h265_init_sequence_params,
@@ -1323,7 +1223,39 @@ static VAAPIEncodeType vaapi_encode_type_h265 = {
 
 static av_cold int vaapi_encode_h265_init(AVCodecContext *avctx)
 {
-    return ff_vaapi_encode_init(avctx, &vaapi_encode_type_h265);
+    VAAPIEncodeContext *ctx = avctx->priv_data;
+
+    ctx->codec = &vaapi_encode_type_h265;
+
+    switch (avctx->profile) {
+    case FF_PROFILE_HEVC_MAIN:
+    case FF_PROFILE_UNKNOWN:
+        ctx->va_profile = VAProfileHEVCMain;
+        break;
+    case FF_PROFILE_HEVC_MAIN_10:
+        av_log(avctx, AV_LOG_ERROR, "H.265 main 10-bit profile "
+               "is not supported.\n");
+        return AVERROR_PATCHWELCOME;
+    default:
+        av_log(avctx, AV_LOG_ERROR, "Unknown H.265 profile %d.\n",
+               avctx->profile);
+        return AVERROR(EINVAL);
+    }
+    ctx->va_entrypoint = VAEntrypointEncSlice;
+
+    // This will be dependent on profile when 10-bit is supported.
+    ctx->va_rt_format = VA_RT_FORMAT_YUV420;
+
+    if (avctx->bit_rate > 0)
+        ctx->va_rc_mode = VA_RC_CBR;
+    else
+        ctx->va_rc_mode = VA_RC_CQP;
+
+
+    ctx->surface_width  = FFALIGN(avctx->width,  16);
+    ctx->surface_height = FFALIGN(avctx->height, 16);
+
+    return ff_vaapi_encode_init(avctx);
 }
 
 #define OFFSET(x) (offsetof(VAAPIEncodeContext, codec_options_data) + \

libavcodec/vaapi_encode_mjpeg.c

@@ -277,8 +277,8 @@ static int vaapi_encode_mjpeg_init_picture_params(AVCodecContext *avctx,
     vpic->reconstructed_picture = pic->recon_surface;
     vpic->coded_buf = pic->output_buffer;
 
-    vpic->picture_width  = ctx->input_width;
-    vpic->picture_height = ctx->input_height;
+    vpic->picture_width  = avctx->width;
+    vpic->picture_height = avctx->height;
 
     vpic->pic_flags.bits.profile      = 0;
     vpic->pic_flags.bits.progressive  = 0;
@@ -333,31 +333,10 @@ static int vaapi_encode_mjpeg_init_slice_params(AVCodecContext *avctx,
     return 0;
 }
 
-static av_cold int vaapi_encode_mjpeg_init_internal(AVCodecContext *avctx)
+static av_cold int vaapi_encode_mjpeg_configure(AVCodecContext *avctx)
 {
-    static const VAConfigAttrib default_config_attributes[] = {
-        { .type  = VAConfigAttribRTFormat,
-          .value = VA_RT_FORMAT_YUV420 },
-        { .type  = VAConfigAttribEncPackedHeaders,
-          .value = VA_ENC_PACKED_HEADER_SEQUENCE },
-    };
-
     VAAPIEncodeContext       *ctx = avctx->priv_data;
     VAAPIEncodeMJPEGContext *priv = ctx->priv_data;
-    int i;
-
-    ctx->va_profile    = VAProfileJPEGBaseline;
-    ctx->va_entrypoint = VAEntrypointEncPicture;
-
-    ctx->input_width    = avctx->width;
-    ctx->input_height   = avctx->height;
-    ctx->aligned_width  = FFALIGN(ctx->input_width,  8);
-    ctx->aligned_height = FFALIGN(ctx->input_height, 8);
-
-    for (i = 0; i < FF_ARRAY_ELEMS(default_config_attributes); i++) {
-        ctx->config_attributes[ctx->nb_config_attributes++] =
-            default_config_attributes[i];
-    }
 
     priv->quality = avctx->global_quality;
     if (priv->quality < 1 || priv->quality > 100) {
@@ -374,7 +353,7 @@ static av_cold int vaapi_encode_mjpeg_init_internal(AVCodecContext *avctx)
 static VAAPIEncodeType vaapi_encode_type_mjpeg = {
     .priv_data_size        = sizeof(VAAPIEncodeMJPEGContext),
 
-    .init                  = &vaapi_encode_mjpeg_init_internal,
+    .configure             = &vaapi_encode_mjpeg_configure,
 
     .picture_params_size   = sizeof(VAEncPictureParameterBufferJPEG),
     .init_picture_params   = &vaapi_encode_mjpeg_init_picture_params,
@@ -390,7 +369,21 @@ static VAAPIEncodeType vaapi_encode_type_mjpeg = {
 
 static av_cold int vaapi_encode_mjpeg_init(AVCodecContext *avctx)
 {
-    return ff_vaapi_encode_init(avctx, &vaapi_encode_type_mjpeg);
+    VAAPIEncodeContext *ctx = avctx->priv_data;
+
+    ctx->codec = &vaapi_encode_type_mjpeg;
+
+    ctx->va_profile    = VAProfileJPEGBaseline;
+    ctx->va_entrypoint = VAEntrypointEncPicture;
+
+    ctx->va_rt_format = VA_RT_FORMAT_YUV420;
+
+    ctx->va_rc_mode = VA_RC_CQP;
+
+    ctx->surface_width  = FFALIGN(avctx->width,  8);
+    ctx->surface_height = FFALIGN(avctx->height, 8);
+
+    return ff_vaapi_encode_init(avctx);
 }
 
 static const AVCodecDefault vaapi_encode_mjpeg_defaults[] = {