libavcodec/videotoolbox: fix decoding of h264 streams with minor SPS changes

Previously the codec kept an entire copy of the SPS, and restarted the VT decoder
session whenever it changed. This fixed decoding errors in [1], as
described in 9519983c. On further inspection, that sample features an SPS change
from High/4.0 to High/3.2 while moving from one scene to another.

Yesterday I received [2], which contains minor SPS changes where the
profile and level do not change. These occur frequently and are not associated with
scene changes. After 9519983c, the VT decoder session is recreated unnecessarily when
these are encountered causing visual glitches.

This commit simplifies the state kept in the VTContext to include just the first three
bytes of the SPS, containing the profile and level details. This is populated initially
when the VT decoder session is created, and used to detect changes and force a restart.

This means minor SPS changes are fed directly into the existing decoder, whereas
profile/level changes force the decoder session to be recreated with the new parameters.

After this commit, both samples [1] and [2] playback as expected.

[1] https://s3.amazonaws.com/tmm1/videotoolbox/spschange.ts
[2] https://s3.amazonaws.com/tmm1/videotoolbox/spschange2.ts

Signed-off-by: Aman Gupta <aman@tmm1.net>

libavcodec/videotoolbox.c

@@ -90,6 +90,7 @@ int ff_videotoolbox_alloc_frame(AVCodecContext *avctx, AVFrame *frame)
 
 CFDataRef ff_videotoolbox_avcc_extradata_create(AVCodecContext *avctx)
 {
+    VTContext *vtctx = avctx->internal->hwaccel_priv_data;
     H264Context *h = avctx->priv_data;
     CFDataRef data = NULL;
     uint8_t *p;
@@ -116,6 +117,10 @@ CFDataRef ff_videotoolbox_avcc_extradata_create(AVCodecContext *avctx)
     p += 3 + h->ps.pps->data_size;
     av_assert0(p - vt_extradata == vt_extradata_size);
 
+    // save sps header (profile/level) used to create decoder session,
+    // so we can detect changes and recreate it.
+    memcpy(vtctx->sps, h->ps.sps->data + 1, 3);
+
     data = CFDataCreate(kCFAllocatorDefault, vt_extradata, vt_extradata_size);
     av_free(vt_extradata);
     return data;
@@ -320,16 +325,13 @@ static int videotoolbox_h264_decode_params(AVCodecContext *avctx,
     VTContext *vtctx = avctx->internal->hwaccel_priv_data;
 
     if (type == H264_NAL_SPS) {
-        if (!vtctx->sps || vtctx->sps_len != size || memcmp(buffer, vtctx->sps, size) != 0) {
-            vtctx->sps = av_fast_realloc(vtctx->sps, &vtctx->sps_capa, size);
-            if (vtctx->sps)
-                memcpy(vtctx->sps, buffer, size);
+        if (size > 4 && memcmp(vtctx->sps, buffer + 1, 3) != 0) {
             vtctx->reconfig_needed = true;
-            vtctx->sps_len = size;
+            memcpy(vtctx->sps, buffer + 1, 3);
         }
     }
 
-    // pass-through new PPS to the decoder
+    // pass-through SPS/PPS changes to the decoder
     return ff_videotoolbox_h264_decode_slice(avctx, buffer, size);
 }
 
@@ -365,7 +367,6 @@ int ff_videotoolbox_uninit(AVCodecContext *avctx)
     VTContext *vtctx = avctx->internal->hwaccel_priv_data;
     if (vtctx) {
         av_freep(&vtctx->bitstream);
-        av_freep(&vtctx->sps);
         if (vtctx->frame)
             CVPixelBufferRelease(vtctx->frame);
     }

libavcodec/vt_internal.h

@@ -40,9 +40,7 @@ typedef struct VTContext {
     struct AVVideotoolboxContext *vt_ctx;
 
     // Current H264 parameters (used to trigger decoder restart on SPS changes).
-    uint8_t                     *sps;
-    uint32_t                    sps_len;
-    unsigned int                sps_capa;
+    uint8_t                     sps[3];
     bool                        reconfig_needed;
 } VTContext;