The macro is defined as
#define WEBRTC_SPL_MUL_16_16_RSFT(a, b, c) \
(WEBRTC_SPL_MUL_16_16(a, b) >> (c))
where the latter macro is in C defined as
(For definitions on ARMv7 and MIPS, see common_audio/signal_processing/include/spl_inl_{armv7,mips}.h)
The replacement consists of
- avoiding casts to int16_t if inputs already are int16_t
- adding explicit cast to <type> if result is assigned to <type> (other than int or int32_t)
- minor cleanups like remove of unnecessary parantheses and style changes
BUG=3348, 3353
TESTED=locally on Mac and trybots
R=kwiberg@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/39139004
Cr-Commit-Position: refs/heads/master@{#8544}
git-svn-id: http://webrtc.googlecode.com/svn/trunk@8544 4adac7df-926f-26a2-2b94-8c16560cd09d
> Ensure only temporary IPv6 address is selected as the best IP.
>
> The current logic of IPv6 selection could still have a small chance for non-temporary address to be selected for candidate. The scenario is that when there is no non-deprecated temporary IP, the global ones could be selected.
>
> Global ones don't necessarily carry MAC. However, instead of comparing whether it has the MAC in it (sometimes 5 out of 6 elements from a MAC are the same, only one diffs), we should just err on the safe side.
>
> BUG=4348
> R=juberti@webrtc.org, pthatcher@webrtc.org
>
> Review URL: https://webrtc-codereview.appspot.com/38289004TBR=guoweis@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/38319004
Cr-Commit-Position: refs/heads/master@{#8534}
git-svn-id: http://webrtc.googlecode.com/svn/trunk@8534 4adac7df-926f-26a2-2b94-8c16560cd09d
The current logic of IPv6 selection could still have a small chance for non-temporary address to be selected for candidate. The scenario is that when there is no non-deprecated temporary IP, the global ones could be selected.
Global ones don't necessarily carry MAC. However, instead of comparing whether it has the MAC in it (sometimes 5 out of 6 elements from a MAC are the same, only one diffs), we should just err on the safe side.
BUG=4348
R=juberti@webrtc.org, pthatcher@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/38289004
Cr-Commit-Position: refs/heads/master@{#8532}
git-svn-id: http://webrtc.googlecode.com/svn/trunk@8532 4adac7df-926f-26a2-2b94-8c16560cd09d
After looking at how this class is used I discovered that the threading model isn't completely crazy and that we can remove a lock from a method that's causing the most contention in getStats right now (and probably many other places). IsFrameCallbackRegistered() looks innocent enough, but it contented with the frame encoder, which can make it wait for tens of milliseconds, unnecessarily.
In addition to removing the lock, I:
* Documented the threading model
* Added checks to guard against regressions
* Reduced lock scope in the Deregister function (not calling out to the notification while holding the lock).
* made virtual functions non-virtual (since they're not really being overriden) and removed unnecessary method that, if called by other threads than the ctor thread, returned a value that wasn't necessarily correct.
BUG=2822
R=mflodman@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/40859004
Cr-Commit-Position: refs/heads/master@{#8531}
git-svn-id: http://webrtc.googlecode.com/svn/trunk@8531 4adac7df-926f-26a2-2b94-8c16560cd09d
This reverts commit r8434.
Reason for revert: Introduced a race condition. If ViECaptureProcess() -> SwapCapturedAndDeliverFrameIfAvailable() is called twice without a call to OnIncomingCapturedFrame() in between (with both captured_frame_ and deliver_frame_ populated), an old frame will be delivered again, since captured_frame_->IsZeroSize() will never be true.
BUG=4352
TBR=perkj@webrtc.org, stefan@webrtc.org, tommi@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/40129004
Cr-Commit-Position: refs/heads/master@{#8530}
git-svn-id: http://webrtc.googlecode.com/svn/trunk@8530 4adac7df-926f-26a2-2b94-8c16560cd09d
* Added a way to notify a Module that it's been attached to a ProcessThread.
The benefit of this is to give the module a way to wake up the thread
when it needs work to happen on the worker thread, immediately.
Today, module instances are typically registered with a process thread
outside the control of the modules themselves. I.e. they typically
don't know about the process thread they're attached to.
* Improve ProcessThread's WakeUp algorithm to not call TimeUntilNextProcess
when a WakeUp call is requested. This is an optimization for the above
case which avoids the module having to acquire a lock or do an interlocked
operation before calling WakeUp(), which would ensure the module's
TimeUntilNextProcess() implementation would return 0.
BUG=2822
R=stefan@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/39239004
Cr-Commit-Position: refs/heads/master@{#8527}
git-svn-id: http://webrtc.googlecode.com/svn/trunk@8527 4adac7df-926f-26a2-2b94-8c16560cd09d
Mostly, it's about moving constructors and descructors to the .cc
files, so that they won't be inlined everywhere.
The reason this CL is so big is that a lot of code was using
common_types.h without declaring a dependency on webrtc_common, which
broke the build once common_types.h started to depend on
common_types.cc.
BUG=163
R=kjellander@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/26089004
Cr-Commit-Position: refs/heads/master@{#8516}
git-svn-id: http://webrtc.googlecode.com/svn/trunk@8516 4adac7df-926f-26a2-2b94-8c16560cd09d
Add some const safety by DCHECK(HasOneRef()) in non-const GetYPlane. This CL also replaces all incorrect non-const calls with const calls for pixel data access in cricket::VideoFrame. It's easy to call the non-const version of e.g. GetYPlane by mistake, even if only const-access is needed. For example:
const scoped_ptr<cricket::VideoFrame> foo;
const uint8_t* y = foo->GetYPlane();
will actually call the non-const version of GetYPlane.
R=mflodman@webrtc.org, perkj@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/39079004
Cr-Commit-Position: refs/heads/master@{#8507}
git-svn-id: http://webrtc.googlecode.com/svn/trunk@8507 4adac7df-926f-26a2-2b94-8c16560cd09d
TestScaler::ComputeAvgSequencePSNR is currently a complex NOP, that always returns kPerfectPSNR. Two frames are read from files into arrays, and then converted into I420VideoFrames. However, the incorrect function ConvertFromI420 is used instead of ConvertToI420, resulting in two empty I420VideoFrames. I420PSNR on empty frames returns kPerfectPSNR.
This CL replaces ConvertFromI420 with ConvertToI420 and actually measures the PSNR. Unfortunately, some tests do not pass when we use the real psnr. The tests that fail are the ones that scale back and forth to a different aspect ratio. webrtc::Scaler has been changed to preserve aspect ratio, and this means that we will end up with a cropped frame if scale and rescale to a different target aspect ratio. I simply removed those tests to make it pass. Having some working tests instead of a lot of dummy tests seems like a win.
R=stefan@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/35339004
Cr-Commit-Position: refs/heads/master@{#8506}
git-svn-id: http://webrtc.googlecode.com/svn/trunk@8506 4adac7df-926f-26a2-2b94-8c16560cd09d
Call to AudioEncoder::Encode fails only if fed bad input, so instead of handling failure, we can just CHECK.
There is also no need to handle case where size of encoded data is larger than allowed maximum, so we just CHECK.
R=kwiberg@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/42099005
Cr-Commit-Position: refs/heads/master@{#8504}
git-svn-id: http://webrtc.googlecode.com/svn/trunk@8504 4adac7df-926f-26a2-2b94-8c16560cd09d
This should be safe to land now that issue 4143 was resolved (in r8492).
This change effectively reverts 8488.
TBR=kwiberg@webrtc.org
Original commit message:
This CL changes the way the decoder sample rate is set and updated. In
practice, it only concerns the iSAC (float) codec.
One single iSAC decoder instance is used for both wideband and
super-wideband decoding, and the instance must be told to switch
output frequency if the payload type changes. This used to be done
through a call to UpdateDecoderSampleRate, but is now instead done in
the Decode call as an extra parameter.
Review URL: https://webrtc-codereview.appspot.com/39289004
Cr-Commit-Position: refs/heads/master@{#8496}
git-svn-id: http://webrtc.googlecode.com/svn/trunk@8496 4adac7df-926f-26a2-2b94-8c16560cd09d
Without this patch, Valgrind's Memcheck was complaining that the test
for whether we should return -1 following the call to
WebRtcIsac_DecodeLb made a conditional branch or move based on the
value of numSamplesLB, which was uninitialized if WebRtcIsac_DecodeLb
failed.
However, as can be seen in the source, the control flow only depends
on the value of numSamplesLB if numDecodedBytesLB >= 0; i.e., if
WebRtcIsac_DecodeLb returned successfully, in which case numSamplesLB
is always initialized. The discrepancy is due to the fact that
Valgrind works on the generated machine code, which contains spurious
such dependencies. The generated code for this test:
if ((numDecodedBytesLB < 0) || (numDecodedBytesLB > lenEncodedLBBytes) ||
(numSamplesLB > MAX_FRAMESAMPLES)) {
looks like this:
95: 0f bf 45 d6 movswl -0x2a(%rbp),%eax
99: 3d c0 03 00 00 cmp $0x3c0,%eax
9e: 0f 8f 45 01 00 00 jg 1e9 <Decode+0x1e9>
a4: 44 89 f0 mov %r14d,%eax
a7: c1 e0 10 shl $0x10,%eax
aa: 0f 88 39 01 00 00 js 1e9 <Decode+0x1e9>
b0: 41 0f bf ce movswl %r14w,%ecx
b4: 89 8d 98 e1 ff ff mov %ecx,-0x1e68(%rbp)
ba: 41 0f bf c7 movswl %r15w,%eax
be: 39 c1 cmp %eax,%ecx
c0: 0f 8f 23 01 00 00 jg 1e9 <Decode+0x1e9>
Note how the compiler has seemingly ignored the C language's guarantee
that the arguments to || must be evaluated in left-to-right order, and
compares numSamplesLB (%eax) with MAX_FRAMESAMPLES (0x3c0, a.k.a. 960)
before the other two conditions! If the uninitialized value in
numSamplesLB happens to be greater than 960, we'll jump to
Decode+0x1e9 (where we'll return -1) without even looking at the other
two conditions. Has the compiler generated broken code?
Well, no. If numDecodedBytesLB is < 0 so that numSamplesLB is
uninitialized, we'll end up jumping to 1e9 whether that value is
greater than 960 or not; we'll just do it with different jump
instructions. This is entirely invisible as far as the C language is
concerned, but the dependency on the uninitialized value is visible at
the machine code level, which is why Memcheck complains.
This patch solves the problem by pragmatically initializing
numSamplesLB before the call even though it isn't necessary other than
for placating Memcheck.
BUG=4143
R=henrik.lundin@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/36309004
Cr-Commit-Position: refs/heads/master@{#8492}
git-svn-id: http://webrtc.googlecode.com/svn/trunk@8492 4adac7df-926f-26a2-2b94-8c16560cd09d
This CL changes the way the decoder sample rate is set and updated. In
practice, it only concerns the iSAC (float) codec.
One single iSAC decoder instance is used for both wideband and
super-wideband decoding, and the instance must be told to switch
output frequency if the payload type changes. This used to be done
through a call to UpdateDecoderSampleRate, but is now instead done in
the Decode call as an extra parameter.
R=kwiberg@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/34349004
Cr-Commit-Position: refs/heads/master@{#8476}
git-svn-id: http://webrtc.googlecode.com/svn/trunk@8476 4adac7df-926f-26a2-2b94-8c16560cd09d
Calls to WebRtcCng_Encode, AudioEncoder::Encode and Vad::VoiceActivity fail only if fed bad input, so instead of handling failure, we can just CHECK. This also makes it unnecessary for methods AudioEncoderCng::EncodePassive and AudioEncoderCng::EncodeActive to return a value, so we can make them void.
R=kwiberg@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/39279004
Cr-Commit-Position: refs/heads/master@{#8475}
git-svn-id: http://webrtc.googlecode.com/svn/trunk@8475 4adac7df-926f-26a2-2b94-8c16560cd09d
* Move constants into the files/functions that use them
* Declare variables in the narrowest scope possible
* Use correct (expected, actual) order for gtest macros
* Remove unused functions
* Untabify
* 80-column limit
* Avoid C-style casts
* Prefer true typed constants to "enum hack" constants
* Print size_t using the right format macro
* Shorten and simplify code
* Other random cleanup bits and style fixes
BUG=none
TEST=none
R=henrik.lundin@webrtc.org, tommi@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/36179004
Cr-Commit-Position: refs/heads/master@{#8467}
git-svn-id: http://webrtc.googlecode.com/svn/trunk@8467 4adac7df-926f-26a2-2b94-8c16560cd09d
ACMGenericCodecWrapper was the only remaining subclass of
ACMGenericCodec, and was the only class that was ever instantiated.
This CL merges the two, essentially keeping the function implementations
from ACMGenericCodecWrapper except where the base class's code was
invoked.
As it turns out, a lot of functions were never used, but in some cases
they were refernced in AudioCodingModuleImpl. In these cases, the
referencing code is commented out and marked FATAL(). This will be
further cleaned up in follow-up CLs.
BUG=4228
COAUTHOR=kwiberg@webrtc.orgR=minyue@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/38209004
Cr-Commit-Position: refs/heads/master@{#8463}
git-svn-id: http://webrtc.googlecode.com/svn/trunk@8463 4adac7df-926f-26a2-2b94-8c16560cd09d
