(Encountered this while trying to reproduce bug 870564.) Seen on hamachi with build: - gecko: b2g18:b87559cef261 - gaia: 7d0ca6a07fc3d4817f80f4ebc24314b6cbcf049a STR: 0. BRANCH=v1.0.1 ./config.sh && B2G_DEBUG=1 ./build.sh && ./flash.sh 1. complete FTU, associating with WiFi 2. open Browser app 3. go to https://www.youtube.com/watch?v=V1IBH1bz5-c Expected: Youtube video should load and play. Observed: network traffic is indicated in the status bar, screen goes black and appears to be prepping to play the video, browser reappears, back on the original page. No video played, no audio is heard. (Note that in order to get to this failure, the assertion failures in bug 841206 and bug 873573 are circumvented by commenting out those lines.) logcat shows: E( 476:0x1ee) Successfully allocated OMX node 'OMX.qcom.audio.decoder.aac' I( 476:0x1ee) [Child 476] WARNING: stride not available, assuming width: file /home/mikeh/dev/mozilla/m-c/b2g18/content/media/omx/OmxDecoder.cpp, line 417 I( 476:0x1ee) [Child 476] WARNING: slice height not available, assuming height: file /home/mikeh/dev/mozilla/m-c/b2g18/content/media/omx/OmxDecoder.cpp, line 422 I( 476:0x1ee) [Child 476] WARNING: rotation not available, assuming 0: file /home/mikeh/dev/mozilla/m-c/b2g18/content/media/omx/OmxDecoder.cpp, line 427 ... E( 139:0x8b) SetParameter, Setting display format to Yamato E( 139:0x1f0) SPS: Decode: old WxH = 640x368,new WxH = 640x368 E( 133:0x12f) VirtualSensor:set accel for virtual sensor! E( 133:0x12f) VirtualSensor: readEvents start!!!count=15,mPendingMask=4 E( 133:0x12f) VirtualSensor: submit event:47eff660 E( 139:0x1f0) registering the PMEM buffers for index 0 fd 19 and base 0x414be000 E( 139:0x1f0) registering the PMEM buffers for index 1 fd 22 and base 0x4181c000 E( 139:0x1f0) registering the PMEM buffers for index 2 fd 25 and base 0x41949000 E( 139:0x1f0) registering the PMEM buffers for index 3 fd 28 and base 0x41b07000 E( 139:0x1f0) registering the PMEM buffers for index 4 fd 31 and base 0x41ce3000 E( 139:0x1f0) registering the PMEM buffers for index 5 fd 34 and base 0x41edb000 E( 139:0x1f0) registering the PMEM buffers for index 6 fd 37 and base 0x42076000 E( 139:0x1f0) registering the PMEM buffers for index 7 fd 40 and base 0x42214000 E( 139:0x1f0) registering the PMEM buffers for index 8 fd 43 and base 0x42426000 E( 139:0x1f0) registering the PMEM buffers for index 9 fd 46 and base 0x425ef000 E( 139:0x1f0) registering the PMEM buffers for index 10 fd 49 and base 0x42713000 E( 139:0x1f0) registering the PMEM buffers for index 11 fd 52 and base 0x4284a000 E( 139:0x1f0) registering the PMEM buffers for index 12 fd 55 and base 0x429c8000 E( 139:0x1f0) registering the PMEM buffers for index 13 fd 58 and base 0x42d31000 E( 139:0x1f0) pmem_alloc1 called for input Buffers E( 139:0x1f0) col_hist_buffers: Out of memory E( 139:0x1f0) error - H264Decoder::InitializeDecInternal() failed!! E( 139:0x1f0) E( 139:0x1f0) **** Frame Fatal Error **** E( 476:0x1f9) [OMX.qcom.video.decoder.avc] ERROR(0x8000100a, 0) F( 476:0x1f9) Fatal signal 11 (SIGSEGV) at 0x5a5a5a5e (code=1) D( 133:0xfa) Layer doesn't have a gralloc buffer D( 133:0xfa) Render aborted. Nothing was drawn to the screen

With B2G_DEBUG=0, logcat shows this: E( 139:0x1fe) pmem_alloc1 called for input Buffers E( 139:0x1fe) col_hist_buffers: Out of memory E( 139:0x1fe) error - H264Decoder::InitializeDecInternal() failed!! E( 139:0x1fe) E( 139:0x1fe) **** Frame Fatal Error **** E( 467:0x20e) [OMX.qcom.video.decoder.avc] ERROR(0x8000100a, 0) E( 467:0x20e) [OMX.qcom.video.decoder.avc] ERROR(0x8000101c, 0) E( 467:0x20e) [OMX.qcom.video.decoder.avc] ERROR(0x8000101c, 0) ...but the SIGSEGV doesn't happen. Instead, I hear some audio from the YouTube video, but there is no picture.

Roc, is this a dupe of one of the existing v1.0.1 YouTube issues?

> col_hist_buffers: Out of memory It says pmen is out of memory. It seems that pmem is already used by others.

(In reply to Andrew Overholt [:overholt] from comment #2) > Roc, is this a dupe of one of the existing v1.0.1 YouTube issues? I don't know. It could be related to bug 877461. (In reply to Sotaro Ikeda [:sotaro] from comment #3) > It says pmen is out of memory. It seems that pmem is already used by others. How can we tell whether pmem is used by others?

(In reply to Robert O'Callahan (:roc) (Mozilla Corporation) from comment #4) > > (In reply to Sotaro Ikeda [:sotaro] from comment #3) > > It says pmen is out of memory. It seems that pmem is already used by others. > > How can we tell whether pmem is used by others? Just the assumption from log and the situation. I saw this only first time start up after flash in the past. And first time start up runs FTU. video codec should alway use same amount of pmem for same video playback. And the log says pmem is out of memory. Fragmentation of pmem is also possible.

You can dump the following information to see pmem is used by someone. Because I saw video only ran once in attached log. If you are able to reproduce this issue, could you attach the full logcat and the following pmem result for analysis? I will try to reproduce this issue on my side too. peter@pchang-desktop:~/b2g_central_hg$ adb shell lsof|grep pmem|grep media mediaserv 152 media 30 ??? ??? ??? ??? /dev/pmem_adsp mediaserv 152 media 32 ??? ??? ??? ??? /dev/pmem_adsp mediaserv 152 media 35 ??? ??? ??? ??? /dev/pmem_adsp mediaserv 152 media 38 ??? ??? ??? ??? /dev/pmem_adsp mediaserv 152 media 41 ??? ??? ??? ??? /dev/pmem_adsp

So from comment 0, in order to get this failure, you need to comment out an assertion here: http://mxr.mozilla.org/mozilla-b2g18/source/dom/ipc/TabChild.cpp#507 Isn't it possible that this bug depends on bug 873573? I believe the assertion is there for a reason. Btw, I fail to load this youtube clip only at the first launch, after a flash.

We blocked on this originally thinking it was a pervasive issue. Please renominate if that's still the case.

Firefox OS is not being worked on