New issue
Advanced search Search tips
Note: Color blocks (like or ) mean that a user may not be available. Tooltip shows the reason.

Issue 724448 link

Starred by 0 users

Issue metadata

Status: Started
Owner:
Last visit > 30 days ago
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 3
Type: Feature



Sign in to add a comment

Replace AudioFifo with a deque of AudioBus in AudioTrackRecorder::AudioEncoder

Project Member Reported by c.pa...@samsung.com, May 19 2017

Issue description

Currently, an AudioFifo is used in AudioTrackRecorder::AudioEncoder to store AudioBus elements that involves needless data copy.
This AudioFifo can be replaced with a deque of AudioBus to avoid this data copy.
 

Comment 1 by c.pa...@samsung.com, May 19 2017

The reland CL [1] too got reverted due to failures in the build cycles on https://findit-for-me.appspot.com/waterfall/culprit?key=ag9zfmZpbmRpdC1mb3ItbWVyRAsSDVdmU3VzcGVjdGVkQ0wiMWNocm9taXVtLzhlM2YwOTY3OGZkMjE5OTY0OWZkZmRlMWJiOWNjMGVjNTUyNDAzMDkM.

[1] https://codereview.chromium.org/2881713002/

Below are the two failures caused by this CL:
1. content_unittests
MediaRecorderHandlerTest.EncodeAudioFrames/2 fails on Windows i.e. it timed out.
[ RUN      ] MediaRecorderHandlerTest.EncodeAudioFrames/2
[4698/4698] MediaRecorderHandlerTest.EncodeAudioFrames/2 (TIMED OUT)

2. MSAN tests for content_browsertests are failing due to uninitialized memory access.

Need to investigate above two failures and fix them.


Comment 2 by c.pa...@samsung.com, May 23 2017

Created https://codereview.chromium.org/2901513002/ with the additional change being addition of audio_bus->Zero(); in AudioEncoder::ProvideInput(...) irrespective of the queue being empty or not. 

This fixes content_browsertests failures due to uninitialized memory access. Both linux_chromium_msan_rel_ng, linux_chromium_chromeos_msan_rel_ng trybots are green now.

miu@, mcasas@: The other issue related to time out is not reproducible locally on linux OS. Also, I am not able to find the trybot for Win7 Tests (dbg)(1) to trigger a dry run. How do I verify the same? Your inputs please on this?
What's the situation re. this bug?

Sign in to add a comment