(No comment was entered for this change.)

(No comment was entered for this change.)

Reverted 97662 and booted. It leaves me with a device that infinite-loops trying to start up Chrome, just like James described off-line.

/var/log/chrome/chrome says

FATAL:netowkr_library.cc(3786)] Check failed: res.

Issue 19519 has been merged into this issue.

(No comment was entered for this change.)

I also see the crash loop on mario reverting r97662 from ToT.  If I reset to r97673 and revert r97662 and r97481 then I get a successful startup on seaboard and mario.

Then something is wrong in (97673, ToT). Steven will be doing a bisect.

Confirmed that r97673 -r97662 -r97481 reverted was good (- indicates a revert) as rhashimoto commented.

Confirmed that r97797 -r97662 -r97481 reverted was good.

Per stevenjb, 97836 -r97662 was bad.

Confirmed that r97819 -r97662 was bad (chrome keep restarting)

Confirmed that r97802 -r97662 was good (chrome booted)

Per stevenjb, 97812 -r97662 was good.

r97662 also breaks pyauto (test automation). When we have a resolution please try desktopui_UrlFetch test and verify that there are no error messages like in  issue 19532 . Or let me know and I can try it out as well. Thanks

Per stevenjb, 97815 was the culprit. We are now double-checking...

Confirmed that 97815 is the culprit, and we know why:
We are passing Value between libcros and Chrome. 97815 changed the size of Value for Chrome, but not libcros (which gets Value from libchrome).

We are discussing options to fix this now.

Created  issue 19577  to track the crash bug which needs to be fixed before we can revert 97662.

 Issue 19526  has been merged into this issue.

Why not immediately revert 97815 so that 97662 can be reverted, until the problem in  issue 19576  can be fixed?
97662 is blocking  issue 19532  due to which all test automation is broken, due to which chrome with the nasty 'webpage not rendering bug' ( issue 19468 ) cannot be uprev'ed.

I'd say it's not right to revert 97815 at this moment. It would've made sense to rever the patch if it broke chrome os build bots, but it didn't as 97662 masked the issue unveiled by 97815 (crosbug.com/19577).  Note that nothing is wrong with 97815 itself. The root cause of the problem is the way we use the Value class in in libcros.

Hopefully, the fix for this issue is now under code review: http://codereview.chromium.org/7714024/

Are you expecting the temp fix and the 97662-revert to land today?
If not, chromeos build tomorrow will be stuck with  issue 19468 , which blocks all R15 testing for another day.

Got approval from a base directory owner. I'm about to check this in on behalf of stevenjb.

I meant I was about to check in http://codereview.chromium.org/7714024/.

Afterward, 97662 should also be reverted. I'll give it a try...

Next R15 Dev RC  build is tonight, please update this bug if commit is landed in ToT before 1 am 

Done.

http://src.chromium.org/viewvc/chrome?view=rev&revision=97997 should fix 19577 (Value size issue)

http://src.chromium.org/viewvc/chrome?view=rev&revision=97999 should fix this (fake network library)

Thanks satorou!

(No comment was entered for this change.)

Make sure Fake* networks are not showing up in the network list.

Verified on :
Google Chrome	15.0.862.1 (Official Build 98372)
Platform	950.0 (Official Build) dev-channel x86
WebKit	        535.2 (@93705)
JavaScript	V8 3.5.8

Select a network menu shows all the available wifi network.

(No comment was entered for this change.)

(No comment was entered for this change.)