Closed Bug 785276 Opened 12 years ago Closed 12 years ago

Emulator builds built on CI VM fail to boot since 32064546c8fe

Categories

(Core :: General, defect)

defect
Not set
critical

Tracking

()

RESOLVED WORKSFORME
blocking-basecamp +

People

(Reporter: jgriffin, Unassigned)

References

Details

All emulator builds that are compiled on the CI build VM fail to fully boot since https://hg.mozilla.org/mozilla-central/rev/32064546c8fe. Builds made against previous commits work fine. This is not a runtime problem - I can download such a build to my local machine and experience the same problem, but builds I make locally are fine. The CI build VM is an Amazon Fedora-based Linux distro, with all current updates. My local machine is Ubuntu 11.10. When such a build is launched, it stays stuck on the "mozilla" loading screen. There are no errors in logcat. I can provide a link to download one of these builds on request.
This blocks all WebAPI and Gaia tests on the CI. If this can't be fixed quickly, it's possible we could move the CI builds to an Ubuntu VM, which might fix the problem.
blocking-basecamp: --- → ?
Is the VM using a different toolchain than you are locally, by any chance?
There are a bunch of people seeing this now locally. I'm bisecting.
Severity: normal → critical
blocking-basecamp: ? → +
Product: Boot2Gecko → Core
Some of the errors here E/GeckoConsole( 7926): [JavaScript Error: "NS_ERROR_FAILURE: Denied" {file: "jar:file:///system/b2g/omni.ja!/components/Webapps.js" line: 381}] E/GeckoConsole( 7926): [JavaScript Error: "uncaught exception: 2147500033"] E/GeckoConsole( 7926): [JavaScript Error: "TypeError: power is null" {file: "app://system.gaiamobile.org/js/screen_manager.js" line: 69}] E/GeckoConsole( 7926): Content JS DEBUG at app://system.gaiamobile.org/js/pinlock/pinlock.js:19 in pl_init: No mozMobileConnection :(
It was Mounir's megapush to m-c :(.
The first bad revision is: changeset: 103228:8700d3b9054d user: Mounir Lamouri <mounir.lamouri@gmail.com> date: Thu Aug 23 11:47:55 2012 -0700 summary: Bug 777072 - 3/7 - Update nsHostTable and nsHostEntry to make them aware of new security model. r=sicking,jlebar
Status: NEW → RESOLVED
Closed: 12 years ago
Resolution: --- → FIXED
Unfortunately, this doesn't solve the original problem; it appears that the problem the above checkin fixed was coincidentally causing the same visible failure as reported in the original description. Even after this fix, builds made on the CI VM are broken. > Is the VM using a different toolchain than you are locally, by any chance? Locally I'm using gcc 4.6.1; the build VM is using gcc 4.4.6. I'll upgrade the build VM to see if this makes a difference.
Status: RESOLVED → REOPENED
Resolution: FIXED → ---
Upgrading gcc/g++ to 4.6.2 didn't solve the issue. I'm going to create a new Ubuntu 11.10 VM to act as builder, and hopefully that will resolve it.
This problem is also affecting the emulator nightly builds, which are built on a different machine entirely. The nightly builder is running RHEL 6.2.
This issue has mysteriously resolved itself with this merge from inbound: http://hg.mozilla.org/mozilla-central/pushloghtml?changeset=f077de66e52d. Builds made on the CI, and on the nightly builder, are both working fine again.
Status: REOPENED → RESOLVED
Closed: 12 years ago12 years ago
Resolution: --- → WORKSFORME
You need to log in before you can comment on or make changes to this bug.