In the fixed point version of the Noise Suppression. At one place we subtract a value in the wrong Q-domain, which later may cause a divide by zero. Going through the floating point code that particular variable should be zero if this happens, which is what the old code tried to accomplish, but in an awkward way.
The bug has been there since development, so the likelihood of actually get a divide by zero is very small.
BUG=chromium:407812
R=ljubomir.papuga@gmail.com, tina.legrand@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/17229004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@7035 4adac7df-926f-26a2-2b94-8c16560cd09d
Most of these changes were done in r7014, but a few targets
were missed. This should make these tests run better
(but they might still be failing due to webrtc:3764).
BUG=webrtc:3741
TESTED=Local compilation using:
GYP_DEFINES="OS=android component=static_library fastbuild=1 target_arch=arm" webrtc/build/gyp_webrtc
ninja -C out/Debug
R=phoglund@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/24369004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@7019 4adac7df-926f-26a2-2b94-8c16560cd09d
Restructure how the Android APK tests are compiled now
that we have a Chromium checkout available (since r6938).
This removes the need of several hacks that were needed when
building these targets from inside a Chromium checkout.
By creating a symlink to Chromium's base we can compile the required
targets. This also removes the need of the previously precompiled
binaries we keep in /deps/tools/android at Google code.
All the user needs to do is to add the target_os = ["android"]
entry to his .gclient as described at
https://code.google.com/p/chromium/wiki/AndroidBuildInstructions
Before committing this CL, the Android APK buildbots will need
to be updated.
This also solves http://crbug.com/402594 since the apply_svn_patch.py
usage will be similar to the other standalone bots.
It also solves http://crbug.com/399297
BUG=chromium:399297, chromium:402594
TESTED=Locally compiled all APK targets by running:
GYP_DEFINES="OS=android include_tests=1 enable_tracing=1" gclient runhooks
ninja -C out/Release
checkdeps
R=henrike@webrtc.org, tommi@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/22149004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@7014 4adac7df-926f-26a2-2b94-8c16560cd09d
The GN files are based upon the GYP files as of r7009.
BUG=3441
TESTED=Passing builds with:
gn gen out/Default --args="build_with_chromium=false" && ninja -C out/Default
gn gen out/Default --args="build_with_chromium=false is_debug=true" && ninja -C out/Default
gn gen out/Default --args="build_with_chromium=false os=\"android\" cpu_arch=\"arm\" is_clang=false" && ninja -C out/Default
gn gen out/Default --args="build_with_chromium=false os=\"android\" cpu_arch=\"arm\" arm_version=7 is_clang=false" && ninja -C out/Default
Compilation of Chromium's 'all' target with src/third_party/webrtc
symlinked to the WebRTC checkout with this CL applied, both
with the default GN settings and using
--args="is_debug=false os=\"android\" cpu_arch=\"arm\""
R=brettw@chromium.org
Review URL: https://webrtc-codereview.appspot.com/14259004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@7013 4adac7df-926f-26a2-2b94-8c16560cd09d
NOTICE: Assembly offsets generation for audio processing will
not be ported to GN and the process of removing them is tracked
in https://code.google.com/p/webrtc/issues/detail?id=3580.
The GN files are based upon the GYP files as of r7009.
BUG=3441
TESTED=Passing builds with:
gn gen out/Default --args="build_with_chromium=false" && ninja -C out/Default
gn gen out/Default --args="build_with_chromium=false is_debug=true" && ninja -C out/Default
gn gen out/Default --args="build_with_chromium=false aec_debug_dump=true" && ninja -C out/Default
gn gen out/Default --args="build_with_chromium=false aec_untrusted_delay_for_testing=true" && ninja -C out/Default
gn gen out/Default --args="build_with_chromium=false prefer_fixed_point=true" && ninja -C out/Default
gn gen out/Default --args="build_with_chromium=false os=\"android\" cpu_arch=\"arm\" is_clang=false" && ninja -C out/Default
gn gen out/Default --args="build_with_chromium=false os=\"android\" cpu_arch=\"arm\" arm_version=7 is_clang=false" && ninja -C out/Default
I don't know how to setup the mips toolchain to test the following, but it's out of scope for the GN effort for now:
gn gen out/Default --args="build_with_chromium=false cpu_arch=\"mipsel\" mips_dsp_rev=0" && ninja -C out/Default
gn gen out/Default --args="build_with_chromium=false cpu_arch=\"mipsel\" mips_dsp_rev=1" && ninja -C out/Default
gn gen out/Default --args="build_with_chromium=false cpu_arch=\"mipsel\" mips_dsp_rev=2" && ninja -C out/Default
Compilation of Chromium's 'all' target with src/third_party/webrtc
symlinked to the WebRTC checkout with this CL applied, both
with the default GN settings and using
--args="is_debug=false os=\"android\" cpu_arch=\"arm\""
R=andrew@webrtc.org, brettw@chromium.org
Review URL: https://webrtc-codereview.appspot.com/15999004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@7012 4adac7df-926f-26a2-2b94-8c16560cd09d
Replace FATAL_ERROR_IF with the more familiar (to Chromium developers)
CHECK and DCHECK. The full Chromium implementation is fairly elaborate
but I copied enough to get us most of the benefits. I believe the main
missing component is a more advanced stack dump. For this bit I relied
on the V8 implementation.
There are a few minor modifications from the Chromium original:
- The FatalMessage class is specialized for logging fatal error
messages and aborting. Chromium uses the general LogMessage class,
which we could consider moving towards in the future.
- NOTIMPLEMENTED() and NOTREACHED() have been removed, partly because
I don't want to rely on our logging.h until base/ and system_wrappers/
are consolidated.
- FATAL() replaces LOG(FATAL).
Minor modifications from V8's stack dump:
- If parsing of a stack trace symbol fails, just print the unparsed
symbol. (I noticed this happened on Mac.)
- Use __GLIBCXX__ and __UCLIBC__. This is from examining the backtrace
use in Chromium.
UNREACHABLE() has been removed because its behavior is different than
Chromium's NOTREACHED(), which is bound to cause confusion. The few uses
were replaced with FATAL(), matching the previous behavior.
Add a NO_RETURN macro, allowing us to remove unreachable return
statements following a CHECK/FATAL.
TESTED=the addition of dummy CHECK, DCHECK, CHECK_EQ and FATAL did the
did the right things. Stack traces work on Mac, but I don't get symbols
on Linux.
R=henrik.lundin@webrtc.org, kwiberg@webrtc.org, tommi@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/22449004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@7003 4adac7df-926f-26a2-2b94-8c16560cd09d
Also add more from common.gypi to webrtc.gni.
These GN configs are based on GYP files in r6997.
BUG=3441
TEST=Trybots and local compile using:
gn gen out/Default --args="build_with_chromium=false" && ninja -C out/Default
gn gen out/Default --args="build_with_chromium=false is_debug=true" && ninja -C out/Default
Passed compile from a Chromium checkout with src/third_party/webrtc linked to the webrtc/ dir of a checkout with this patch applied.
R=brettw@chromium.org, glaznev@webrtc.org, mflodman@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/18709004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@6999 4adac7df-926f-26a2-2b94-8c16560cd09d
Now that WebRTC has rolled the chromium_revision past
http://crrev.com/284372 in r6784, clang has become the
default compiler. Since WebRTC standalone code doesn't
yet compile the Chromium Clang plugins enabled, this CL
disables them for the parts of the code that doesn't yet pass
compilation with them enabled.
The buildbots are using Goma which is not yet switched
over to Clang by default. That's why they're not red yet.
BUG=163
TEST=Passing compile locally on Linux using:
gn gen out/Debug --args="build_with_chromium=false is_debug=true" && ninja
-C out/Debug
gn gen out/Release --args="build_with_chromium=false is_debug=false" && ninja
-C out/Release
gn gen out/Default --args="build_with_chromium=false os=\"android\" cpu_arch=\"arm\" arm_version=7" && ninja -C out/Default
R=brettw@chromium.org
Review URL: https://webrtc-codereview.appspot.com/16279004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@6966 4adac7df-926f-26a2-2b94-8c16560cd09d
This test verifies bit exactness for the send-side of ACM. The test
setup is a chain of three different test classes:
test::AcmSendTest -> AcmSenderBitExactness -> test::AcmReceiveTest
The receiver side is driving the test by requesting new packets from
AcmSenderBitExactness::NextPacket(). This method, in turn, asks for the
packet from test::AcmSendTest::NextPacket, which inserts audio from the
input file until one packet is produced. (The input file loops
indefinitely.) Before passing the packet to the receiver, the
AcmSenderBitExactness class verifies the packet header and updates a
payload checksum with the new payload. The decoded output from the
receiver is also verified with a (separate) checksum.
The current CL only adds tests for 30 ms and 60 ms iSAC. More codecs
will be added in coming changes.
BUG=3521
R=kwiberg@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/20179004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@6949 4adac7df-926f-26a2-2b94-8c16560cd09d
This test has been failing every now and then. This is likely due to the
random input that was used. With this change, the input is now read from
an audio file, making it identical on each run.
The encoding is moved to inside the main test loop, so that new data is
added with each packet. (Before this change, the same payload was added
over and over again; only the RTP header was updated.)
BUG=3715
R=turaj@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/19079004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@6948 4adac7df-926f-26a2-2b94-8c16560cd09d
There are currently a number of places in the code where we dump audio
data in various stages of processing for debug purposes. Currently
these all write raw, uncompressed PCM files, which isn't supported by
the most common audio players, and requires the user to supply
metadata such as sample rate, sample size and endianness, etc.
This patch adds a simple class that makes it easy to write WAV files
instead. WAV files still contain the same uncompressed PCM data, but
they have a small header that contains all the requisite metadata, and
are supported by virtually all audio players.
Since some of the debug code that will be writing WAV files is written
in plain C, a C API is included as well.
R=andrew@webrtc.org, bjornv@webrtc.org, henrike@webrtc.org, tommi@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/16809004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@6932 4adac7df-926f-26a2-2b94-8c16560cd09d
These global arrays are shared amongst all AEC instances, and were at
serious risk of data races. A Chromium TSAN bot recently caught this.
Also move the function pointer selection for optimization to
create-time. (Ideally this would only be done once.)
BUG=chromium:404133,1503
R=bjornv@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/19069004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@6922 4adac7df-926f-26a2-2b94-8c16560cd09d
When building WebRTC from a Chromium checkout (i.e. with
https://codereview.chromium.org/321313006/ applied) GN
cannot execute successfully.
This CL fixes:
- include path for video_processing module's SSE2 target.
- NSS/SSL targets
BUG=3441
TEST=
Passing WebRTC GN trybots.
Passing build from a Chromium checkout with https://codereview.chromium.org/321313006 applied and src/third_party/webrtc symlinked to the WebRTC checkout with this CL:
gn gen out/Default --args="clang_use_chrome_plugins=false" && ninja -C out/Default
gn gen out/Default --args="os=\"android\" cpu_arch=\"arm\" clang_use_chrome_plugins=false" && ninja -C out/Default
R=brettw@chromium.org
Review URL: https://webrtc-codereview.appspot.com/21179005
git-svn-id: http://webrtc.googlecode.com/svn/trunk@6921 4adac7df-926f-26a2-2b94-8c16560cd09d
The modification only uses the unique part of the spectrum (as is done for the C and asm code). It passes
byte to byte conformance test, and the single function performance
(if not specified, the code is compiled by GCC 4.6) on different
platforms:
| run 100k times | cortex-a7 | cortex-a9 | cortex-a15 |
| use C as the base on each | (1.2Ghz) | (1.0Ghz) | (1.7Ghz) |
| CPU target | | | |
|----------------------------+-----------+-----------+------------|
| C | 100% | 100% | 100% |
| Neon asm | 18% | 14% | 19% |
| Neon inline asm | 31% | 25% | 27% |
| Neon intrinsic (GCC 4.6) | 33% | 27% | 42% |
| Neon intrinscis (GCC 4.8) | 17% | 14% | 19% |
| Neon intrinsics (LLVM 3.3) | 15% | 13% | 18% |
BUG=
R=andrew@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/13739004
Patch from Joe Yu <joe.yu@arm.com>.
git-svn-id: http://webrtc.googlecode.com/svn/trunk@6920 4adac7df-926f-26a2-2b94-8c16560cd09d
Macros should in general be avoided. WEBRTC_SPL_UMUL_32_16_RSFT16 is only used in iSAC fixed point as part of multiplying with LSB and MSB. A better approach is to have one function for that complete operation in iSAC.
This CL removes the macro and replace the operation locally.
BUG=3148, 3353
TESTED=locally on Linux and trybots
R=tina.legrand@webrtc.org, turaj@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/16349004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@6907 4adac7df-926f-26a2-2b94-8c16560cd09d
Fixes issues where statistics only was reported for the first stream if
configured with simulcast, and in case of RTX the reported statistics was
depending on the order of the report blocks.
Also fixes issues with multiple report blocks in the SendStatisticsProxy and the
RtcpStatisticsCallback. SendStatisticsProxy is now aware of RTX ssrcs, and the
RTCPReceiver is calling the RtcpStatisticsCallback with the correct SSRCs, and
not only the primary stream SSRC.
R=mflodman@webrtc.org, sprang@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/20149004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@6903 4adac7df-926f-26a2-2b94-8c16560cd09d
