The cur_*auth_type variables were set before the http_connect call
prior to 6a463e7fb - their sole purpose is to record the
authentication type used to do the latest request, since parsing
the http response sets the new type in the auth state.
CC: libav-stable@libav.org Signed-off-by: Martin Storsjö <martin@martin.st>
Andrew Stone [Tue, 12 Aug 2014 21:03:55 +0000 (17:03 -0400)]
ogg: update event_flags with STREAM_/METADATA_UPDATED whenever metadata changes.
Originally, AVFormatContext and a metadata dict were provided to ff_vorbis_comment(),
but this presented issues if an AVStream was being updated or the metadata on
AVFormatContext wasn't actually being updated. To remedy this, ff_vorbis_stream_comment()
explicitly updates a stream's metadata and sets any necessary flags.
ff_vorbis_comment() does not modify any flags, and any calls to it that update
AVFormatContext's metadata (just a single call) must also update
AVFormatContext.event_flags after detecting any metadata changes to the provided
dictionary, as signaled by a positive return value.
Andrew Stone [Mon, 11 Aug 2014 17:35:09 +0000 (13:35 -0400)]
Expose metadata found in onCuePoint events in .flv files.
Currently, only onMetaData is used, but some providers (wrongly)
put metadata into onCuePoint events, and it's still nice to be
able to use that data.
onCuePoint events also present metadata slightly differently than
onMetaData events: all metadata is found inside an object called
"parameters". In order to extract this metadata, it's easiest to
recurse through the object tree and pull out anything found in
child objects and put it in the top-level metadata.
Andrew Stone [Fri, 8 Aug 2014 17:09:23 +0000 (13:09 -0400)]
lavf: add AVFormatContext/AVStream fields for signaling to the user when events happen.
The only flags, for now, indicate if metadata was updated and are set after each call to
av_read_frame(). This comes with the caveat that, on stream start, it might not be set properly
as packets might be buffered in AVFormatContext.packet_buffer before being given to the user
in av_read_frame().
Martin Storsjö [Mon, 11 Aug 2014 12:02:28 +0000 (15:02 +0300)]
configure: Check for nanosleep in headers as well, not only in libs
On mingw64 with c++11 support, the link libraries do contain a
nanosleep function, while it isn't exposed via the headers. Using
check_func_headers instead of a plain check_func fixes this
misdetection.
Martin Storsjö [Wed, 6 Aug 2014 06:52:38 +0000 (09:52 +0300)]
w32threads: Use newer thread synchronization functions when targeting Vista
When explicitly targeting Vista or newer (which only happens if the
caller explicitly sets _WIN32_WINNT to a high enough value via the
extra cflags option - otherwise configure script sets
-D_WIN32_WINNT=0x0502), we already unconditionally link to the
ConditionVariable functions, since 4622f11f9.
Similarly use the newer -Ex versions of CreateEvent, CreateSemaphore,
InitializeCriticalSection and WaitForSingleObject, that all appeared
in Vista. When building Windows Store applications, the older versions
of these functions aren't available, only the -Ex functions. When
doing such a build, the user can set -D_WIN32_WINNT=0x0600 to
forcibly use the newer functions instead.
Martin Storsjö [Mon, 11 Aug 2014 07:18:28 +0000 (10:18 +0300)]
http: Stop reading after receiving the whole file for non-chunked transfers
Previously this logic was only used if the server didn't
respond with Connection: close, but use it even for that case,
if the server response is non-chunked.
Originally the http code has relied on Connection: close to close
the socket when the file/stream is received - the http protocol
code just kept reading from the socket until the socket was closed.
In f240ed18 we added a check for the file size, because some
http servers didn't respond with Connection: close (and wouldn't
close the socket) even though we requested it, which meant that the
http protocol blocked for a long time at the end of files, waiting
for a socket level timeout.
When reading over tls, trying to read at the end of the connection,
when the peer has closed the connection, can produce spurious (but
harmless) warnings. Therefore always voluntarily stop reading when
the specified file size has been received, if not using a chunked
transfer encoding. (For chunked transfers, we already return 0
as soon as we get the chunk header indicating end of stream.)
Martin Storsjö [Sun, 10 Aug 2014 11:40:09 +0000 (14:40 +0300)]
configure: Don't pass MSVC compiler options -M[TD] to armasm
The -MD option (for enabling a dynamically linked crt) gets interpreted
as a cpp option for generating dependency information (into a file named
'-.d', when preprocessing to a pipe). We shouldn't be passing
any and all C compiler flags to armasm (which is a plain assembler,
only with cpp bolted on via gas-preprocessor), but these are the
main conflicting ones.
Anton Khirnov [Sun, 27 Jul 2014 19:40:20 +0000 (19:40 +0000)]
hevc_filter: avoid excessive calls to ff_hevc_get_ref_list()
1) each of the loops run within a single CTB, so the relevant reference
list is constant
2) when that CTB is, or lies on the same slice as, the current one, we
can use a simple access instead of a relatively expensive call to
ff_hevc_get_ref_list()
Diego Biurrun [Mon, 4 Aug 2014 17:28:10 +0000 (10:28 -0700)]
avcodec: Undeprecate reordered_opaque
It allows attaching other external, opaque data to the frame and passing it
through the reordering process, for cases when the caller wants other data
than just the plain packet pts. There is no way to cleanly achieve this
without the field.
Anton Khirnov [Sun, 3 Aug 2014 08:14:48 +0000 (10:14 +0200)]
svq1: do not modify the input packet
The input data must remain constant, make a copy instead. This is in
theory a performance hit, but since I failed to find any samples
using this feature, this should not matter in practice.
Also, check the size of the header, avoiding invalid reads on truncated
data.
Martin Storsjö [Mon, 28 Jul 2014 08:51:29 +0000 (11:51 +0300)]
configure: Include the armcc build number in the compiler identification
This tries to find the most expressive part of the output of
armcc --vsn to include, giving a compiler identification of
"ARM Compiler 5.04 update 2 (build 82)" instead of just
"ARM Compiler 5.04" for armcc 5.0.
4.x versions of armcc output the following, for "armcc --vsn":
ARM C/C++ Compiler, RVCT4.0 [Build 925]
For evaluation purposes only
Software supplied by: ARM Limited
ARM C/C++ Compiler, 4.1 [Build 894]
For evaluation purposes only
Software supplied by: ARM Limited
5.0 versions output this:
Product: ARM Compiler 5.04
Component: ARM Compiler 5.04 update 2 (build 82)
Tool: armcc [5040081]
For evaluation purposes only
Software supplied by: ARM Limited
Janne Grunau [Tue, 5 Aug 2014 10:08:09 +0000 (12:08 +0200)]
configure: check $as first before using $gas as GNU as
llvm's integrated assembler supports the AArch64 asm on darwin since
August 2014. So check $as first before using gas-preprocessor.pl via
$gas. Makes the checks specific for that the architecture specific asm
needs. PPC Altivec and AArch64 needs on ':vararg' for macro arguments.
Arm needs in addition the '.altmacro' directive.
Ben Avison [Mon, 21 Jul 2014 13:53:09 +0000 (14:53 +0100)]
vc-1: Optimise parser (with special attention to ARM)
The previous implementation of the parser made four passes over each input
buffer (reduced to two if the container format already guaranteed the input
buffer corresponded to frames, such as with MKV). But these buffers are
often 200K in size, certainly enough to flush the data out of L1 cache, and
for many CPUs, all the way out to main memory. The passes were:
1) locate frame boundaries (not needed for MKV etc)
2) copy the data into a contiguous block (not needed for MKV etc)
3) locate the start codes within each frame
4) unescape the data between start codes
After this, the unescaped data was parsed to extract certain header fields,
but because the unescape operation was so large, this was usually also
effectively operating on uncached memory. Most of the unescaped data was
simply thrown away and never processed further. Only step 2 - because it
used memcpy - was using prefetch, making things even worse.
This patch reorganises these steps so that, aside from the copying, the
operations are performed in parallel, maximising cache utilisation. No more
than the worst-case number of bytes needed for header parsing is unescaped.
Most of the data is, in practice, only read in order to search for a start
code, for which optimised implementations already existed in the H264 codec
(notably the ARM version uses prefetch, so we end up doing both remaining
passes at maximum speed). For MKV files, we know when we've found the last
start code of interest in a given frame, so we are able to avoid doing even
that one remaining pass for most of the buffer.
In some use-cases (such as the Raspberry Pi) video decode is handled by the
GPU, but the entire elementary stream is still fed through the parser to
pick out certain elements of the header which are necessary to manage the
decode process. As you might expect, in these cases, the performance of the
parser is significant.
To measure parser performance, I used the same VC-1 elementary stream in
either an MPEG-2 transport stream or a MKV file, and fed it through avconv
with -c:v copy -c:a copy -f null. These are the gperftools counts for
those streams, both filtered to only include vc1_parse() and its callees,
and unfiltered (to include the whole binary). Lower numbers are better:
Before After
File Filtered Mean StdDev Mean StdDev Confidence Change
M2TS No 861.7 8.2 650.5 8.1 100.0% +32.5%
MKV No 868.9 7.4 731.7 9.0 100.0% +18.8%
M2TS Yes 250.0 11.2 27.2 3.4 100.0% +817.9%
MKV Yes 149.0 12.8 1.7 0.8 100.0% +8526.3%
Yes, that last case shows vc1_parse() running 86 times faster! The M2TS
case does show a larger absolute improvement though, since it was worse
to begin with.
This patch has been tested with the FATE suite (albeit on x86 for speed).
The rationale is that you have a packed format in form
<greyscale sample> <alpha sample> <greyscale sample> <alpha sample>
and shortening greyscale to 'G' might make one thing about Greenscale instead.
An alias pixel format and color space name are provided for compatibility.
Janne Grunau [Mon, 4 Aug 2014 08:04:08 +0000 (10:04 +0200)]
fate: explicitly set the default THREADS value
This makes the default of '1' more explicit than defaulting to '1' in
fate-run.sh and regression-funcs.sh if THREADS is not set.
Fixes the reported thread count in fate-cpu if THREADS is not set.
Diego Biurrun [Sun, 3 Aug 2014 19:19:10 +0000 (12:19 -0700)]
huffyuv: Check and propagate function return values
Bug-Id: CVE-2013-0868
inspired by a patch from Michael Niedermayer <michaelni@gmx.at> Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind Signed-off-by: Diego Biurrun <diego@biurrun.de> CC: libav-stable@libav.org