Another phone with the OMX.SEC.mp3.dec bug: SM-G386T

[DonnKey]

Issue description

The bug with not finishing/repeating a file because of a bug in the OMX.SEC.mp3.dec codec repeats with the SM-G386T phone. Another instance of #4519.

Reproduction steps

Play (some) files with any app when playing an audiobook consisting of multiple files; only the first file plays. Hardware dependent (of course).

Link to test content

http://etc.usf.edu/lit2go/1/alices-adventures-in-wonderland/
or (same content)
http://freeclassicaudiobooks.com/audiobooks/Alice/mp3/

You'll need at least chapters 1 and 2.

I've seen this on other audiobooks as well. I don't have a big enough sample to be sure, but "older" recordings look suspicious.

Version of ExoPlayer being used

dev-v2 (and obviously anything earlier) pulled 12/18/18 around 7PM PST.

Device(s) and version(s) of Android being used

SM-G386T Android 4.2.2 (highest available for that phone).

A full bug report captured from the device

Not worth the trouble, since the obvious one-line fix of adding that phone to the list of blacklisted phones near line 325 of MediaCodecUtil.java fixes the problem for my phone.

Some additional notes.

The list of phones with the problem is getting long. Here's some additional info that may help fixing the "real" bug.

1. Replacing chapter 1 with a (very short) snippet of the same file created by Audacity (simply Export a trimmed version) causes it to behave correctly. (I can try with a full-legth Audacity copy if you would like.)
2. The phone's marketing name is Samsung Galaxy Avant, which was sold at Costco for a while, so it's probably fairly common (if old).
3. I haven't seen this line mentioned as being emitted in the debugger just as the failed file handoff should occur:
   E/AudioTrack: Discontinuity detected [expected 631954259, got 0]
   In my repro both the behavior and the above line are solid (100% of the time). (And the one-line change takes it to "never fail").
4. My guess is that the "T" in the phone model number means "T-Mobile". A truncated string compare to pick up other carrier's versions might be wise.

Since I have the hardware (at least currently - it needs to be handed off soon) I can try a few things if that would help.

[ojw28]

Closing this as a duplicate of the original issue, which I've re-opened. We'll add SM-G386T to the blacklist, and adjust the blacklist to also cover variants of the listed devices. I'm not sure if "T" does stand for "T-Mobile", but it is true that there are other SM-G386X variants (for different values of X), and that they probably all have the same issue. Similar rules likely apply for the other listed devices too.