MPEGPlayer: Try a few timestamps for all streams at the beginning, not just video, to help deal with anomalous start PTS values in a stream. Hopefully the hardcoded number gives it enough tolerance.

git-svn-id: svn://svn.rockbox.org/rockbox/trunk@28961 a1c6a512-1295-4272-9138-f99709370657

1 files changed, 5 insertions, 2 deletions

diff --git a/apps/plugins/mpegplayer/mpeg_parser.c b/apps/plugins/mpegplayer/mpeg_parser.c
index b683efe1c9..572bd1b13e 100644
--- a/apps/plugins/mpegplayer/mpeg_parser.c
+++ b/apps/plugins/mpegplayer/mpeg_parser.c
@@ -289,8 +289,11 @@ static void init_times(struct stream *str)
 
     str->start_pts = INVALID_TIMESTAMP;
 
-    /* Probe many for video because of B-frames */
-    for (i = STREAM_IS_VIDEO(str->id) ? 5 : 1; i > 0;)
+    /* Probe for many for the start because a stamp or two could be anomalous.
+     * Video also can also have things out of order. How many? There isn't any
+     * "right" value but just a few seems suffient to filter some bad cases.
+     * Too many and file loading could take too long. */
+    for (i = 5; i > 0;)
     {
         switch (parser_get_next_data(&tmp_str, STREAM_PM_RANDOM_ACCESS))
         {