If the the height is not a multiple of the macro block size then the memory
of the last line is reused for all extra lines. This is no problem if the
last line is duplicated properly. However, if the extra lines are not
initialized properly during encoding, then the last visible line is
overwritten with undefined data.
Use a extra buffer to avoid this problem.
Add property to ignore decoding errors. Default is to ignore a few
decoding errors if the input is packetized, but error out immediately
if the input is not packetized.
Ignoring errors for packetized input most likely doesn't work
properly yet, so don't do that for now.
https://bugzilla.gnome.org/show_bug.cgi?id=623063
When we can't decode directly into the output buffer, make our temp buffers
only as big as needed instead of allocating for the worst case scenario (well,
we still alloc more than strictly needed for some cases, but significantly
less than before).
Original commit message from CVS:
* ext/jpeg/gstjpegdec.c: (gst_jpeg_dec_finalize),
(gst_jpeg_dec_init), (gst_jpeg_dec_chain),
(gst_jpeg_dec_sink_event), (gst_jpeg_dec_change_state):
* ext/jpeg/gstjpegdec.h:
Clip outgoing buffers according to currently configured segment.
Original commit message from CVS:
* ext/jpeg/gstjpegdec.c: (gst_jpeg_dec_class_init),
(gst_jpeg_dec_chain), (gst_jpeg_dec_change_state):
* ext/jpeg/gstjpegdec.h:
Fix durations on outgoing buffers after seeking
in MJPEG files (#334083); some minor clean-ups.
Original commit message from CVS:
* ext/jpeg/gstjpegdec.c: (gst_jpeg_dec_init), (gst_jpeg_dec_chain),
(gst_jpeg_dec_change_state):
* ext/jpeg/gstjpegdec.h:
Cleanups. Don't create caps for every chain.
Original commit message from CVS:
* ext/jpeg/gstjpegdec.c: (gst_jpeg_dec_init), (gst_jpeg_dec_chain),
(gst_jpeg_dec_change_state):
* ext/jpeg/gstjpegdec.h:
Fix crashes/invalid memory access for pictures that have a height
that is not a multiple of 16 (or rather: v_samp_factor * DCTSIZE).
Also fix the state change function for downwards state changes
(need to chain up to parent before destroying our resources, to
make sure pads get deactivated and our chain function isn't
running and using those very same resources in another thread).
The jpeg line buffer only needs to be v_samp_factor*DCTSIZE lines
per plane, not picture_height lines; allocate that on the stack.
Original commit message from CVS:
* ext/jpeg/gstjpegdec.c: (gst_jpeg_dec_init),
(gst_jpeg_dec_setcaps), (gst_jpeg_dec_chain),
(gst_jpeg_dec_change_state):
* ext/jpeg/gstjpegdec.h:
Make mjpeg actually work and skip jpeg data parsing if we
know that the input is packetized (ie. each input buffer
is exactly one jpeg frame).
Original commit message from CVS:
New mimetypes gone into effect today - this commit changes all old mimetypes over to the new mimetypes spec as described in the previous commit's document. Note: some plugins will break, some pipelines will break, expect HEAD to be broken or at least not 100% working for a few days, but don't forget to report bugs
