GNOME Bugzilla – Bug 549073
buffer_alloc function from new basetransform sometimes fails to alloc data
Last modified: 2008-08-25 09:48:10 UTC
While trying to write a test program to prove that bug #548764 exists... I wrote the attached program.. that causes a segfault on almost every try here (but works perfectly with 0.10.20). The bufferalloc function from the new basetransform stuff sometimes returns a GstBuffer with a NULL data pointer and size 0... It very much seems like random corruption... It may have the same cause as bug #548764 and #546494 ..
Created attachment 117249 [details] Test that crashes the new basetransform on every try
* gst/law/alaw-encode.c: (gst_alaw_enc_init), (gst_alaw_enc_chain): * gst/law/mulaw-conversion.c: * gst/law/mulaw-encode.c: (gst_mulawenc_init), (gst_mulawenc_chain): The encoder can't really renegotiate at the time they perform a pad-alloc so make the srcpads use fixed caps. Check the buffer size after a pad-alloc because the returned size might not be right when the downstream element does not know the size of the new buffer (capsfilter). Fixes #549073.