825485 - Cannot play AAC music content in the browser & music app

Status: RESOLVED FIXED

(Core :: Audio/Video, defect)

Description

[Jason Smith [:jsmith]]

Attachment: Logcat

Build: B2G 18 12/29/2012
Device: Unagi

Steps:

1. Go to http://songza.com/listen/acoustic-bedtime-songza/ in the browser

Expected:

Music should eventually start playing from the phone.

Actual:

Music fails to start. This works fine on FF Android with the FF Android user agent and FF OS user agent, so it's definitely not a user agent issue.

Looking at the attached logcat, it appears we may be bombing on some aac decoding possibly:

12-30 00:18:51.893: E/OMXCodec(2347): Attempting to allocate OMX node 'OMX.qcom.audio.decoder.aac'
12-30 00:18:51.893: E/QC_AACDEC(2347): COmxDecAac::component_init role[OMX.qcom.audio.decoder.aac]
12-30 00:18:51.893: E/QC_BASE(2347): Dev Open Failed[-1] errno[13]
12-30 00:18:51.893: E/QC_BASE(2347): OMX AAC component destroyed
12-30 00:18:51.893: E/OMXCodec(2347): Attempting to allocate OMX node 'OMX.TI.AAC.decode'
12-30 00:18:51.893: E/OMXCodec(2347): Attempting to allocate OMX node 'OMX.google.aac.decoder'
12-30 00:18:51.893: E/OMXCodec(2347): Successfully allocated OMX node 'OMX.google.aac.decoder'
12-30 00:18:51.953: W/AudioFlinger(115): Thread AudioOut_1 cannot connect to the power manager service
12-30 00:18:51.973: E/AudioHardwareMSM76XXA(115): Parsing error in AudioFilter.csv.

Comment 1

[Jason Smith [:jsmith]]

Chris - Any ideas?

Comment 2

[[e-mail] [:onecyrenus]]

This is a regression between Dec 18th -> Dec 29th.. Most likely can get there bissecting builds, but a review of what changed at a driver level will probably provide some insight as well.

Comment 3

[Chris Lee [:clee]]

This should be marked blocking if we broke AAC playback.

Comment 4

[cajbir (:cajbir)]

Has AAC playback ever worked in the browser? The browser can't access the hardware decoders, for privilege reasons, which is why this fails:

> 12-30 00:18:51.893: E/OMXCodec(2347): Attempting to allocate OMX node
> 'OMX.qcom.audio.decoder.aac'
> 12-30 00:18:51.893: E/QC_AACDEC(2347): COmxDecAac::component_init
> role[OMX.qcom.audio.decoder.aac]

It does try with the software decoder:

> 12-30 00:18:51.893: E/OMXCodec(2347): Successfully allocated OMX node
> 'OMX.google.aac.decoder'

I'm not sure what the capabilities of that are. I'm on leave till next week and can't investigate until then.

Comment 5

[Andrew Overholt [:overholt]]

It's unclear if AAC ever worked in the browser to bb- unless that's incorrect.

Comment 6

[Jason Smith [:jsmith]]

(In reply to Andrew Overholt [:overholt] from comment #5)
> It's unclear if AAC ever worked in the browser to bb- unless that's
> incorrect.

comment 2 implies this is a regression. Renoming.

dclarke - can you confirm this was working previously on Dec 18th?

Comment 7

[Andrew Overholt [:overholt]]

Let's let Chris Double investigate a bit more when he gets back from PTO.

Comment 8

[[e-mail] [:onecyrenus]]

I can confirm that playback on songza was working on December 18th. 
I have also confirmed that the AAC codec playback seemed to work. 

E/OMXCodec( 3224): Attempting to allocate OMX node 'OMX.qcom.audio.decoder.aac'
E/QC_AACDEC( 3224): COmxDecAac::component_init role[OMX.qcom.audio.decoder.aac]
E/QC_BASE ( 3224): Dev Open Failed[-1] errno[13]
E/QC_BASE ( 3224): OMX AAC component destroyed
E/OMXCodec( 3224): Attempting to allocate OMX node 'OMX.TI.AAC.decode'
E/OMXCodec( 3224): Attempting to allocate OMX node 'OMX.google.aac.decoder'
E/OMXCodec( 3224): Successfully allocated OMX node 'OMX.google.aac.decoder'

Comment 9

[Andrew Overholt [:overholt]]

Thanks for confirming, David.

Comment 10

[cajbir (:cajbir)]

Sadly, visting the URL in comment 0 gives:

"We are very sorry to say that due to licensing constraints we cannot allow access to Songza for listeners located outside of the United States and Canada".

Is there an URL accessible from NZ?

Comment 11

[cajbir (:cajbir)]

I put an AAC audio file in an MP4 container at http://cd.pn/a3 and this does not play on b2g. In the absence of an working URL from NZ I'll track down the issue with this and hope it's the same problem.

Comment 12

[Jason Smith [:jsmith]]

(In reply to Chris Double (:doublec) from comment #10)
> Sadly, visting the URL in comment 0 gives:
> 
> "We are very sorry to say that due to licensing constraints we cannot allow
> access to Songza for listeners located outside of the United States and
> Canada".
> 
> Is there an URL accessible from NZ?

Ack. You are right. I'm in Germany right now and hitting the same thing.

(In reply to Chris Double (:doublec) from comment #11)
> I put an AAC audio file in an MP4 container at http://cd.pn/a3 and this does
> not play on b2g. In the absence of an working URL from NZ I'll track down
> the issue with this and hope it's the same problem.

Sounds like a good mitigation strategy. Are we getting the same/similar logcat output as above with that test case? If so, that would probably confirm it's close comparison to the problem.

Comment 13

[cajbir (:cajbir)]

(In reply to Jason Smith [:jsmith] from comment #12)
> (In reply to Chris Double (:doublec) from comment #10)
> Sounds like a good mitigation strategy. Are we getting the same/similar
> logcat output as above with that test case? If so, that would probably
> confirm it's close comparison to the problem.

Yes, similar logs. I reverted back to a December 18 build and that plays the AAC fine. I'm tracking down the regression now.

Comment 14

[David Flanagan [:djf]]

Is this the same bug as bug 825523?

Comment 15

[cajbir (:cajbir)]

A bisect shows the bad commit as:

dc040b91b7b6a616f2969e0568b080aa60dccdae is the first bad commit
commit dc040b91b7b6a616f2969e0568b080aa60dccdae
Author: Chris Jones <[e-mail]>
Date:   Mon Dec 24 15:43:10 2012 -0800

    Bug 824230 - Handle UNKNOWN_ERROR in OmxDecoder. r=doublec, a=blocking-basecamp

Which means it's probably the same as bug 825523 and maybe bug 826083.

Comment 17

[cajbir (:cajbir)]

Attachment: Fix

Comment 19

[Ed Morley [:emorley]]

https://hg.mozilla.org/integration/mozilla-inbound/rev/0037353015da

Presume this also wants to land on b2g18? Does it need additional approval? I can't find the post/wiki page that described the changes as we progressed through the milestones.

Comment 20

[Chris Jones [:cjones] inactive; ni?/f?/r? if you need me]

bb+ is still enough for uplift.  The patches only need to go to b2g18 now.

Comment 21

[Ed Morley [:emorley]]

Ok, thank you :-)

https://hg.mozilla.org/releases/mozilla-b2g18/rev/7ebd5d44afa1

Comment 22

[Chris Jones [:cjones] inactive; ni?/f?/r? if you need me]

Thanks! :)

Comment 23

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

Just wondering, is there a reason to have this bug still open?

Comment 24

[Jason Smith [:jsmith]]

(In reply to Naoki Hirata :nhirata from comment #23)
> Just wondering, is there a reason to have this bug still open?

Nope.

Comment 25

[Ed Morley [:emorley]]

(In reply to Naoki Hirata :nhirata from comment #23)
> Just wondering, is there a reason to have this bug still open?

Bugs stay open until the changeset has merged from mozilla-inbound to mozilla-central.

In this case I'll leave closed, since will be merging shortly and not worth the dependency bugspam of reopening and closing again.

Comment 27

[Ed Morley [:emorley]]

https://hg.mozilla.org/mozilla-central/rev/0037353015da