libstagefright: Fix handling of 'skip' atom in MPEG4Extractor

As per the MPEG4 specs, the contents of the skip atom are irrelevant
and can be ignored. Currently, the parser attempted to parse the skip
atom which doesn't have any structure.

Attempting to parse the skip atom causes a problem if the bytes immediately
following the atom type (which denote the size for most other atoms)
form a extremely large number.  In that case we add this large number to
the offset.  Eventually, we get out of this atom because we can't parse
it.  But we fail the 'sanity check' in the previous recursive level
since '*offset != stop_offset' would not be true.

This change completely ignores the contents of the skip atom.

Change-Id: I837e8279aa1417f7c35ed66be80aa6d313a77722

1 files changed, 2 insertions, 1 deletions

diff --git a/media/libstagefright/MPEG4Extractor.cpp b/media/libstagefright/MPEG4Extractor.cpp
index 93d4765..8abcb9c 100755
--- a/media/libstagefright/MPEG4Extractor.cpp
+++ b/media/libstagefright/MPEG4Extractor.cpp
@@ -492,7 +492,6 @@ status_t MPEG4Extractor::parseChunk(off_t *offset, int depth) {
         case FOURCC('m', 'o', 'o', 'f'):
         case FOURCC('t', 'r', 'a', 'f'):
         case FOURCC('m', 'f', 'r', 'a'):
-        case FOURCC('s', 'k', 'i' ,'p'):
         case FOURCC('u', 'd', 't', 'a'):
         case FOURCC('i', 'l', 's', 't'):
         {
@@ -1087,6 +1086,8 @@ avcC_parse_fail: //do memory cleanup
             break;
         }
 
+        case FOURCC('s', 'k', 'i' ,'p'):
+        //Fall through; we can completely ignore this atom
         default:
         {
             *offset += chunk_size;