[PATCH 0/1] MR9757: mfsrcsnk: Use stream_map for index value on stream start.

[Brendan McGrath (＠redmcg)]

Fixes an issue where the queued samples for one stream can end up on the
other when performing a pause/play.

-- 
https://gitlab.winehq.org/wine/wine/-/merge_requests/9757

[PATCH 1/1] mfsrcsnk: Use stream_map for index value on stream start.

[Brendan McGrath]

From: Brendan McGrath <[e-mail]>

Fixes an issue where the queued samples for one stream can end up on the
other when performing a pause/play.
---
 dlls/mfsrcsnk/media_source.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/dlls/mfsrcsnk/media_source.c b/dlls/mfsrcsnk/media_source.c
index 16bf0c09204..b6f5654f8b0 100644
--- a/dlls/mfsrcsnk/media_source.c
+++ b/dlls/mfsrcsnk/media_source.c
@@ -493,7 +493,7 @@ static HRESULT media_source_start(struct media_source *source, IMFPresentationDe
     for (i = 0; i < source->stream_count; i++)
     {
         struct media_stream *stream = source->streams[i];
-        media_stream_start(stream, i, position);
+        media_stream_start(stream, source->stream_map[i], position);
     }
 
     if (position->vt == VT_I8 && (status = winedmo_demuxer_seek(source->winedmo_demuxer, position->hVal.QuadPart)))
-- 
GitLab

https://gitlab.winehq.org/wine/wine/-/merge_requests/9757

[Brendan McGrath (＠redmcg)]

Another way to fix this is by changing `media_stream_start` to accept either stream or index, and obtain the other value within `media_stream_start`. That way no one calling `media_stream_start` could pass mismatched values. But it's only called in one spot at the moment, so maybe just fixing it at the call site is good enough.

-- 
https://gitlab.winehq.org/wine/wine/-/merge_requests/9757#note_125500

[Rémi Bernon (＠rbernon)]

It looks good to me, but it probably doesn't count as regression fix though.

-- 
https://gitlab.winehq.org/wine/wine/-/merge_requests/9757#note_125530