Quick-and-dirty hacks for getting policy working behind proxies.

Quick-and-dirty hacks for getting policy working behind proxies.

Note that this hard-wires policy OAuth fetching in login_utils.cc to the
authentication profile. I'm not entirely sure this is a good idea, but
required since the user profile isn't in a state to allow successful
requests at this point (the use-shared-proxies flag is off). Other stuff
in login_utils.cc might need the hack too (sync?).

BUG=
TEST=

[Mattias Nissler (ping if slow), 2012-01-31 21:05:19]

Hi folks,

this hack gets us back into the game. I've verified user policy fetching
succeeds (i.e. login doesn't block) for both enrolled and non-enrolled devices
when behind a proxy (not requiring auth). Zel probably has an opinion on the
authentication profile hack, see the CL description for more context.

[Mattias Nissler (ping if slow), 2012-01-31 21:06:46]

+pastarmovj

[kuan, 2012-01-31 21:13:38]

lgtm for proxy config service part; i'm also ok with login part, but zel would
know more abt that.

[zel, 2012-01-31 21:30:40]

lgtm

[xiyuan, 2012-02-01 02:25:31]

FYI, I've committed this CL on trunk.
https://src.chromium.org/viewvc/chrome?view=rev&revision=120013

[Nikita (slow), 2012-02-01 09:42:34]

https://chromiumcodereview.appspot.com/9315011/diff/1/chrome/browser/chromeos...
File chrome/browser/chromeos/login/login_utils.cc (right):

https://chromiumcodereview.appspot.com/9315011/diff/1/chrome/browser/chromeos...
chrome/browser/chromeos/login/login_utils.cc:792: // Make sure we flip every
profile to not share proxies if the user hasn't
Seems that if Chrome crashes inside user session, this setting would have
incorrect default value.

[pastarmovj, 2012-02-01 09:57:31]

https://chromiumcodereview.appspot.com/9315011/diff/1/chrome/browser/chromeos...
File chrome/browser/chromeos/login/login_utils.cc (right):

https://chromiumcodereview.appspot.com/9315011/diff/1/chrome/browser/chromeos...
chrome/browser/chromeos/login/login_utils.cc:792: // Make sure we flip every
profile to not share proxies if the user hasn't
On 2012/02/01 09:42:34, Nikita Kostylev wrote:
> Seems that if Chrome crashes inside user session, this setting would have
> incorrect default value.

Yes, theoretically if Chrome crashes right after this code and before the
preferences were written to disk it might lead to flipping back to the default
being now enabled, until the user logs out and back in. Then this code will get
re-executed and flip the bit again. However you are right that we might want to
have this code on the "crashed browser restore" code path too.