*Environment: Phone - Otoro Gaia - 16fda982b02d7279abd1c15b66af25e692d502f3 Gecko - e070197f4c89 *How to reproduce: 1. Unlock the phone 2. Open bunch of apps and play around for awhile (keep all apps on the background running, do not close them) 3. Open the browser 4. Keep opening new tabs connecting to websites *Expected Result: You can continue using browser without any problems. *Actual Result: The screen suddenly turn black. Restarted the system with mozilla logo on the bottom-right. All the running program and browser are gone.

Hey walter, Can you reproduce this? Can you provider a logcat?

Hi, I reproduced the bug 3 times few minutes ago. Here is more detailed bug reproduction steps: 1. Open maybe 5~10 different programs (sometimes, it wouldn't matter, but this might help) 2. Open the browser 3. Enter "http://www.google.com/" at top 4. Quickly click ">" on the top-right, click "+", choose "http://www.google.com/" 5. repeat step4 again and again

I wasn't able to reproduce this by opening several apps (email, calendar, camera, settings, several installed apps) and opening 24 tabs in the browser with random sites. I'm using the 2012-11-06 stable build.

It's quite reproducible in aurora build, can you try aurora build?

I only have access to a dogfooding phone, which we've been asked not to flash outside of the OTA updates. I will try to reproduce again once I receive an update.

Hi, We got otoro here as for last week. Furthermore, we don't have OTA updates. Also, we just got unagi today (dogfooding is for unagi only).

(In reply to Lawrence Mandel [:lmandel] from comment #6) > I only have access to a dogfooding phone, which we've been asked not to > flash outside of the OTA updates. I will try to reproduce again once I > receive an update. I believe these are using an Aurora build. Sounds like we're OOMing and killing the main process.

After getting the unagi phone. reproduced in nightly build yesterday. *Environment: Phone - Unagi Gaia(git) - 139b4ff3d4e7cae1eb19922ca5e258b48cebd99e Gecko(git) - 43cff61762580cc1d870703ce40be8f3b011d80f *How to reproduce:(Step modified) 1. Unlock the phone 2. Open the browser 3. Keep opening new tabs. Maybe sometimes connect to a website. *Expected Result: You can continue using browser without any problems. *Actual Result: The screen suddenly turn black. Restarted the system with mozilla logo on the bottom-right. All the running program and browser are gone. *Side Note: 1. please be patient. if it didn't crash about 30~40 tabs, kill the browser and redo it. 2. sometimes it happens when u opened just 6~7 tabs. 3. browser sometimes shows "It's embarrassed... ... ..." meaning the browser is having some issues too

Please see the video for reference on both system restart and browser crashing: https://www.youtube.com/watch?v=X6yhqUWw4LA&feature=youtube_gdata_player

At some point you /will/ be able to OOM the phone, particularly since the browser runs in the main process. I don't think we should block release if the STR are "open a bajillion tabs in the browser". Walter, if you can show that this is /not/ an OOM crash, that would probably convince me that this is a more important bug.

Hi, Justin, That's not "bajillion" per say. In the video, I open only 6 tabs to crash it. (fresh fleshed, no other apps opened before I tested it) Sometimes, I can crash it with opening 4 tabs. You don't have to do it fast to crash it. At first, I didn't intent to cause a crash by quickly opened lots of tabs. I was trying to browse google map and google music by browser. And, it crashed. The methods that I documented here to show you people how to crash the OS is just a substitute way so that you people can reproduce it easily. Also, please explain OOM so that I can better understand your point of view. If random usage of browser crashed the OS doesn't count as a important bug, what should it be?

> Also, please explain OOM so that I can better understand your point of view. We're running out of memory. http://en.wikipedia.org/wiki/Out_of_memory It's a bug that we don't kill the browser process before we kill the main process; that's covered by bug 780437. It sounds like we sometimes kill the browser first, which is the right thing to do. I suggest we wait for bug 780437 to land and then see if you're still having problems here.

Ok. But, still, opening 6 browser tabs after a fresh started phone doesn't sound like an out of memory issue.

(In reply to Walter Chen from comment #14) > Ok. > But, still, opening 6 browser tabs after a fresh started phone doesn't sound > like an out of memory issue. Our phones have approximately 100mb of RAM available to all Gecko processes. Six tabs can totally OOM the phone; try opening six tabs in desktop Firefox and looking at the "resident" value in about:memory. Or attach the phone to your machine and run $ watch adb shell cat /proc/meminfo and see how the MemFree line changes.

basecamp+ as we need some answer for this issue. dougt suggestions that ben is working on some other bugs that may help with this issue.

Minusing this. This is likely due to wrongly prioritizing which process to kill on OOM, or that killing the child process ends up bringing down the parent process. We're doing work on both those issues in other bugs. Once I find the bug for the "bringing down parent" i'll mark it as blocking this.

It's already blocking; bug 780437.

I wasn't able to reproduce the main-process crash here, but I was able to cause high heap-unclassified in the main process. That high memory usage could be causing the crash cited here, but I don't want to conflate these bugs, so I've filed bug 820558.

I can't reproduce this main-process crash, so I'm going to close this bug as WFM. If you can reproduce and are willing and able to help investigate further, please re-open this bug.

Tested in unagi. It got fixed in some version. Works fine in 2012-12-16 build from https://releases.mozilla.com/b2g/