When decoding stream using hardware V4L2 decoder element, in any of the
currently supported formats, the decoding will fail once frame number
1000000 is reached. The reported error clearly indicates a wrap-around
occured, instead of receiving decoded frame 1000000, frame 0 is received
from the hardware V4L2 decoder driver.
The problem is actually not in the driver itself, but rather in gstreamer,
which uses `struct v4l2_buffer` member `.timestamp` in a special way. The
timestamp of buffers with encoded data added to the SINK (input) queue of
the driver is copied by the driver into matching buffers with decoded data
added to the SOURCE (output) queue of the driver. In fact, the timestamp
is not a timestamp at all, but rather in this special case, only part of
it is used as an incrementing frame counter.
The `.timestamp` is of type `struct timeval`, which is defined in
`sys/time.h` [1]. Only the `tv_usec` member of this structure is used
for the incrementing frame counter. However, suseconds_t tv_usec [2]
may be limited to range [-1, 1000000]:
"
[XSI] The type suseconds_t shall be a signed integer type capable of
storing values at least in the range [-1, 1000000].
"
Therefore, once frame 1000000 is reached, a rollover occurs and decoding
fails.
Fix this by using both `struct timeval` members, `.tv_sec` and `.tv_usec`
with matching modular arithmetic, this way the failure would occur again
just short of 2^84 frames, which should be plenty.
[1] https://pubs.opengroup.org/onlinepubs/9699919799/basedefs/sys_time.h.html
[2] https://pubs.opengroup.org/onlinepubs/9699919799/basedefs/sys_types.h.html
A test case using stateless hardware h264 decoder, the WARN/ERROR output
in gstreamer log indicates a failure occurred. With this change, that
error no longer occurs and the WARN/ERROR are not present:
```
pc$ gst-launch-1.0 videotestsrc num-buffers=1001001 pattern=6 ! \
video/x-raw,width=16,height=16,format=I420 ! \
x264enc ! filesink location=/tmp/test.h264
dut$ GST_DEBUG="*:3" gst-launch-1.0 filesrc location=/tmp/test.h264 ! \
h264parse ! v4l2slh264dec ! fakesink
...
0:03:51.393677606 12111 0x370df400 WARN \
v4l2codecs-decoder gstv4l2decoder.c:1157:gst_v4l2_request_set_done:<v4l2decoder2> \
Requested frame 1000000, but driver returned frame 0.
0:03:51.394140597 12111 0x370df400 WARN \
v4l2codecs-decoder gstv4l2decoder.c:1157:gst_v4l2_request_set_done:<v4l2decoder2> \
Requested frame 1000001, but driver returned frame 1.
0:03:51.394425216 12111 0x370df400 WARN \
v4l2codecs-decoder gstv4l2decoder.c:1157:gst_v4l2_request_set_done:<v4l2decoder2> \
Requested frame 1000002, but driver returned frame 2.
0:03:51.394665211 12111 0x370df400 WARN \
v4l2codecs-decoder gstv4l2decoder.c:1157:gst_v4l2_request_set_done:<v4l2decoder2> \
Requested frame 1000003, but driver returned frame 3.
0:03:51.394785833 12111 0x370df400 WARN \
v4l2codecs-h264dec gstv4l2codech264dec.c:1059:gst_v4l2_codec_h264_dec_output_picture:<v4l2slh264dec0> \
error: Failed to decode frame 1000000
ERROR: from element /GstPipeline:pipeline0/v4l2slh264dec:v4l2slh264dec0: Failed to decode frame 1000000
```
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5598>
While everything that's merged needs to have passed CI, the
testsuite result are not as consistent as we'd like and sometimes
regress.
It's nice to have have a pipeline hitory of the tests results
to make it easier to spot regressions that might have caused
racy/flaky tests.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5588>
This pair of elements, inspired from shmsink/shmsrc, send unix file
descriptors (e.g. memfd, dmabuf) from one sink to multiple source
elements in other processes.
The unixfdsink proposes a memfd/shm allocator, which causes for example
videotestsrc to write directly into memories that can be transfered to
other processes without copying.
Sponsored-by: Netflix Inc.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5328>
There is no guarantee that g_get_user_runtime_dir() is in a tmpfs. Using
an explicit shared memory API seems safer for all POSIX platforms.
Note that Android does not have shm_open() and only added memfd_create()
since API level 30 (Android 11).
Sponsored-by: Netflix Inc.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5328>
Detail a bit the intention behind GST_ALLOCATOR_FLAG_CUSTOM_ALLOC, even
if implementation does not currently fully follow that usage. Introduce
a new flag specifically for copying memories using the default system
allocator.
Sponsored-by: Netflix Inc.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5328>
The src caps of the libde265 is now fixed to I420, and so if the
stream is other format, such as 4:4:4 or 10 bits format, the pipeline
will crash because the dowstream element accesses the video buffer as
I420 format.
We now restrain the input caps to "main" profile, which only contains
4:2:0 8 bits stream.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5573>
While the minimum timeout duration is 5s, checking only every 5s means
that we would notice this 4.9s too late in the worst case.
Checking once a second reduces this considerably while keeping the
number of wakeups still low.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5520>
Otherwise it can happen that we regularly switch back and forth between
clocks under certain circumstances for no good reason.
Also remove redundant comparison when comparing the steps removed between two
clocks.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5520>
This can happen with the dummy "noopenh264" library that the freedesktop
flatpak runtime ships, and Fedora is planning on shipping as well. In
both cases the dummy implementation gets replaced with the actual
openh264 library that's downloaded directly from Cisco, but just to be
on safe side, this patch makes it careful to check the return values to
avoid crashing if the underlying library hasn't been swapped out yet.
The patch is taken from freedesktop-sdk and was originally written by
Valentin David <valentin.david@codethink.co.uk>.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5581>
- Fixes a crash in tutorial 5, which happened when going back from video playback to the 'library' view, due to
ui_delegate already being destroyed at that point.
- Updates layouts to avoid navigation bar overlapping play/pause buttons. Colours are now correctly updated
based on light/dark mode being enabled, overall look and feel is improved with bigger buttons and paddings.
New button types are used, so target version is now iOS 15.0.
- Disables debug log coloring, as the default terminal in XCode does not render that anyway, so logs are now
more readable there.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5473>
The local glib subproject doesn't exist so the glib/glib.supp file
cannot be included.
As it is needed for the do_lookup_by_name() function call, let's add the
system wide suppression file so that its version matches the installed glib
version.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5434>
Set up a test suite that runs fluster in a virtual machine using virtme.
This test only runs when a kernel image path is set in the new
`virtme_kernel_image` meson option.
The kernel iimage must have support for visl.
The suite contains 4 tests, 1 for each supported codec in visl:
- vp8
- vp9
- h.264
- hevc
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5434>
As the visl driver does not do actual decoding but outputs a test
pattern, the fluster reference md5sums have to be changed.
This is generated using fluster's reference command and stored here to
run the v4l2 stateless codecs test against.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5434>
