1027180 - [Camera][hamachi][v1.3] Video freezes when trying to record a video

Status: RESOLVED INVALID

(Firefox OS Graveyard :: Gaia::Camera, defect)

Description

[Naoki Hirata :nhirata (please use needinfo instead of cc)]

Attachment: logcat.txt

1. launch camera app
2. switch to video
3. try to record a video

Expected: video app works and starts recording
Actual: preview freezes, no video is recorded, odd messages in logcat

Gaia      7f361e67aa65e40adb2233ab247ebd0f142cfbf1
Gecko     https://hg.mozilla.org/releases/mozilla-b2g28_v1_3/rev/b1a1c8c5d440
BuildID   20140618024023
Version   28.0
ro.build.version.incremental=eng.tclxa.20131223.163538
ro.build.date=Mon Dec 23 16:36:04 CST 2013
Buri

Comment 1

[Naoki Hirata :nhirata (please use needinfo instead of cc)]

tested 1.4 build and that seems fine.

Gaia      3bdd037ec1a11abebe16a5d7f6ff0d863e80bc07
Gecko     https://hg.mozilla.org/releases/mozilla-b2g30_v1_4/rev/523491fa3339
BuildID   20140618000203
Version   30.0
ro.build.version.incremental=eng.tclxa.20131223.163538
ro.build.date=Mon Dec 23 16:36:04 CST 2013
Buri

Comment 2

[Naoki Hirata :nhirata (please use needinfo instead of cc)]

Note: only reason why I found this was because I use 1.3 to record videos on the Buri for bugs.

Comment 3

[Mike Habicher [:mikeh] (high bugzilla latency)]

Odd messages in logcat?

Comment 4

[Mike Habicher [:mikeh] (high bugzilla latency)]

Interesting:

06-18 13:30:43.279: W/AudioFlinger(140): RecordThread: buffer overflow
06-18 13:30:43.339: E/OMXNodeInstance(140): !!! Observer died. Quickly, do something, ... anything...
06-18 13:30:43.339: E/QualcommCamera(140): Qvoid android::disable_msg_type(camera_device*, int32_t): E
06-18 13:30:43.339: I/QCameraHWI(140): disableMsgType: E
06-18 13:30:43.339: I/QCameraHWI(140): disableMsgType: X, msgType =0xffff, mMsgEnabled=0x0
06-18 13:30:43.339: E/QualcommCamera(140): Qvoid android::stop_preview(camera_device*): E
06-18 13:30:43.339: I/QCameraHWI(140): void android::QCameraHardwareInterface::stopPreview(): stopPreview: E
06-18 13:30:43.339: I/QCameraHWI(140): stopRecordingInternal: E
06-18 13:30:43.339: E/QCameraHWI_Record(140): virtual void android::QCameraStream_record::stop(): BEGIN
06-18 13:30:43.339: E/SurfaceTextureClient(140): queueBuffer: error queuing buffer to SurfaceTexture, -32
06-18 13:30:43.339: E/QCameraHWI_Preview(140): android::status_t android::QCameraStream_preview::processPreviewFrameWithDisplay(mm_camera_ch_data_buf_t*): enqueue_buffer failed, err = -32
06-18 13:30:43.339: E/QCameraHWI_Preview(140): android::status_t android::QCameraStream_preview::processPreviewFrameWithDisplay(mm_camera_ch_data_buf_t*): error in dequeue_buffer, enqueue_buffer idx = 5, no free buffer now
06-18 13:30:43.339: E/SurfaceTextureClient(140): queueBuffer: error queuing buffer to SurfaceTexture, -32
06-18 13:30:43.339: E/QCameraHWI_Preview(140): android::status_t android::QCameraStream_preview::processPreviewFrameWithDisplay(mm_camera_ch_data_buf_t*): enqueue_buffer failed, err = -32
06-18 13:30:43.339: E/QCameraHWI_Preview(140): android::status_t android::QCameraStream_preview::processPreviewFrameWithDisplay(mm_camera_ch_data_buf_t*): error in dequeue_buffer, enqueue_buffer idx = 6, no free buffer now
06-18 13:30:43.339: E/VENC_OMX(140): OMX_VENC_MSG_HIGH process_state_change::3676 Attempt to go to OMX_StateIdle
06-18 13:30:43.349: V/QCameraHWI_Record(140): virtual void android::QCameraStream_record::stop(): END
06-18 13:30:43.349: E/VENC_ENC(140): VENC_MSG_HIGH venc_stop::1422 Received command VENC_CMD_STOP
06-18 13:30:43.349: I/QCameraHWI(140): stopRecordingInternal: X

Comment 5

[Jason Smith [:jsmith]]

Just to make sure other people can see - can someone else confirm they can see this bug on Buri 1.3?

Comment 6

[Joshua Mitchell (Inactive)]

I have NOT been able to Repro this bug - 

The latest build we had in our tinder-box folder is a bit dated and this bug does NOT repro there
Device: Buri 1.3
Build ID: 20140502004836
Gaia: 667539f1ed4becc45b182a5f1046221d3eeb9e7c
Gecko: 084f8c2adae5
Version: 28.0 (1.3) 
Firmware Version: v1.2device.cfg

So I downloaded the exact build this was initially seen in from the pvt nightly site and was also NOT able to repro it out of 30 tries. The videos record fine and previews are watch-able.

Environmental Variables:
Device: Buri 1.3
Build ID: 20140618024023
Gaia: 7f361e67aa65e40adb2233ab247ebd0f142cfbf1
Gecko: b1a1c8c5d440
Version: 28.0 (1.3) 
Firmware Version: v1.2device.cfg

Comment 7

[Hema Koka [:hema]]

(In reply to Joshua Mitchell from comment #6)
> I have NOT been able to Repro this bug - 
> 
> The latest build we had in our tinder-box folder is a bit dated and this bug
> does NOT repro there
> Device: Buri 1.3
> Build ID: 20140502004836
> Gaia: 667539f1ed4becc45b182a5f1046221d3eeb9e7c
> Gecko: 084f8c2adae5
> Version: 28.0 (1.3) 
> Firmware Version: v1.2device.cfg
> 
> So I downloaded the exact build this was initially seen in from the pvt
> nightly site and was also NOT able to repro it out of 30 tries. The videos
> record fine and previews are watch-able.
> 
> Environmental Variables:
> Device: Buri 1.3
> Build ID: 20140618024023
> Gaia: 7f361e67aa65e40adb2233ab247ebd0f142cfbf1
> Gecko: b1a1c8c5d440
> Version: 28.0 (1.3) 
> Firmware Version: v1.2device.cfg

Naoki ,

Is this still an issue with the latest 1.3 build?

Comment 8

[Hema Koka [:hema]]

Re-nom if this is still a problem. If not, please close. Thanks!

Comment 9

[Naoki Hirata :nhirata (please use needinfo instead of cc)]

It's still an issue.

Ok, I guess it may be that this has to do with the OEM seccomp build that I'm using base.
Checking with jld to see if he's seeing this issue; if he's not then I guess it would be safe to close.

Jed, are you seeing an issue with video recording on your seccomp builds of hamachi/buri?

Comment 10

[Jed Davis [:jld] ⟨⏰|UTC-7⟩ ⟦he/him⟧]

I haven't been using my hamachi lately, so I wouldn't have seen or not-seen this.  But any eng build (after bug 912807, which includes the 1.3 build in comment #0) should have --enable-content-sandbox-reporter, so a sandbox failure would show up in logcat with the "Sandbox" tag.

I recall that the hamachi seccomp boot.img changes the kernel modversion string, so it needs a matching system.img or else modules won't load.  Could that be part of the problem?

And for what it's worth, there are known issues with seccomp on 1.3, at least when WebRTC is involved; see bug 988105.

Comment 11

[Naoki Hirata :nhirata (please use needinfo instead of cc)]

Well based on the last comment, that there are known issues, and we're not releasing seccomp for 1.3; I'll close this out as invalid.