Closed Bug 875820 Opened 9 years ago Closed 9 years ago

Galaxy Nexus in Hax0r isn't working

Categories

(Testing Graveyard :: Eideticker, defect)

x86_64
Linux
defect
Not set
normal

Tracking

(Not tracked)

RESOLVED WORKSFORME

People

(Reporter: wlach, Unassigned)

Details

In the last week or so, it's gone down twice. Also, profiling doesn't seem to work (it freezes when trying to initialize the profiler thread). I think there's something seriously wrong with it; we probably either need to reflash or replace it.

There will be no eideticker numbers for the galaxy nexus until we resolve this issue. If we do get it working well, we can regenerate the missing data though.
Clint's going to look into this when he gets the chance. We think there's also some galaxy nexii in Toronto that we're not presently using for anything which we could use as a replacement.

<ctalbert> wlach: It would take some time to reflash the phone. I wish I had a backup phone I could put in there. Let me see what I can do
<ctalbert> wlach: I'm at home today, so not much I can do on it until Tuesday
Flags: needinfo?(ctalbert)
It's looking like it's a profiling issue and not a phone issue.  I'll have more details in a bit as I verify but essentially if we start with the profiling environment variable in place, we never actually render any content in fennec and the test times out.  The fennec browser will crash if we try to open a second tab when running in profile mode, but the UI is responsive so it doesn't appear to be a browser hang.  Running without the profiling turned on, the phone runs normally.
Flags: needinfo?(ctalbert)
So, yes, I ran the same build on both the Haxxor GN and my own personal GN. The build was yesterdays fennec nightly.

Running with the profiling environment variable set caused Fennec to come up but never finish loading any content. Running without the variable set, Fennec came up and loaded the same content just fine. (I used http://www.google.com).
0. Put fennec on device, start agent
1. Telnet to device port 20701
2. exec MOZ_PROFILER_STARTUP=true org.mozilla.fennec http://www.google.com
 --> fennec seems to hang, throbbing forever.

1. Telnet to device, port 20701
2. exec org.mozilla.fennec http://www.google.com
--> fennec comes up and renders google.

So the Haxxor GN may certainly be unstable, but specific instance of instability looks like a profiling issue.
It's funny, the profiler seems to work fine on my phone. Are you using the same version of Android I am? getprop ro.build.version.release from a shell should tell you (or exec getprop ro.build.version.release from the agent telnet prompt). I'm using 4.2.2.

Regardless of the profiling issue, there's something that's causing the GN in hax0r to go down frequently. It's been down since 1:32am today.

(now that I think about it, maybe it's starting fennec with a profiler that's causing the crash? I'll turn off the dashboard update script for now so we can verify that theory when we bring it back up)
So, Will, so how has it been running since we stopped running the profiling stuff? The GN in haxx0r is running 4.2.1, for whatever that's worth.
Flags: needinfo?(wlachance)
There have been no problems since we disabled the profiler, filed bug 879424 about that. I think I must have misdiagnosed the problem before.
Flags: needinfo?(wlachance)
Status: NEW → RESOLVED
Closed: 9 years ago
Resolution: --- → WORKSFORME
Product: Testing → Testing Graveyard
You need to log in before you can comment on or make changes to this bug.