Fwd: Command-line switches for guest mode (ChromeOS)

[John Sheu]

For Chrome on ChromeOS, logging in as a guest user, ChromeOS restarts itself in guest mode through DBus.  There's a nice list of command-line switches that are propagated to the guest-mode Chrome session, in LoginUtils.impl::GetOffTheRecordCommandLine (src/chrome/browser/chromeos/login/login_utils.cc:1040).

Is there any particular criterion for adding a switch or leaving one off here?  I'm interested in the moment in adding disable-gpu-watchdog and no-sandbox to the list, and I'm wondering if there are other considerations at play here.

Thanks,
-John Sheu

[John Sheu]

Nobody is concerned about this?  Do not be unduly alarmed then when a CL lands and breaks everything.

-John Sheu

[Brett Wilson]

On Thu, Apr 19, 2012 at 8:12 PM, John Sheu <sh...@chromium.org> wrote:
> Nobody is concerned about this?  Do not be unduly alarmed then when a CL
> lands and breaks everything.

Your question is too specific and lacking context for anybody to
answer in a meaningful way. As with any other change, the process is
to do the right thing and get it reviewed by the right person. Give
what information you posted, the answer is obviously that you should
not disable the sandbox and GPU watchdog.

Brett

[John Sheu]

I'm not disabling the sandbox and/or the GPU watchdog.  As I noted before, what I'm doing is adding those switches to the list of switches propagated into guest mode.

As it happens now, if you set "--disable-gpu-watchdog" or "--no-sandbox" in /sbin/session_manager_setup.sh on a ChromeOS machine, those switches are only applied to the Chrome stack running the login prompt.  If you sign in as a guest, Chrome restarts itself for the guest session over DBus, and those switches are not propagated to the guest session, which makes (for example) debugging the GPU process non-trivial.

I'll ask again: are there any landmines associated with enabling the propagation that I should be aware of?

-John Sheu

[James Cook]

I'm not aware of any problems caused by adding switches there.  I don't know the history of why Chrome restarts itself for guest mode in this way.

I have been burned in the past by switches that were *not* propagated there, so I'm supportive of you adding the two you mention.  (In fact, I added a comment to one of the switches files warning people to look at this function when adding switches.)

James

--
Chromium OS Developers mailing list: chromiu...@chromium.org
View archives, change email options, or unsubscribe:
http://groups.google.com/a/chromium.org/group/chromium-os-dev?hl=en

[Chris Masone]

On Thu, Apr 19, 2012 at 9:53 PM, James Cook <jame...@google.com> wrote:

I'm not aware of any problems caused by adding switches there.  I don't know the history of why Chrome restarts itself for guest mode in this way.

We wish that it didn't, but in order to avoid doing so, there are a bunch of things that'd need to be disabled within Chrome after they've already been initialized.  We have a long-standing bug tracking fixing this, but no one's working on it :-/

http://code.google.com/p/chromium-os/issues/detail?id=6924

I would _love_ to see this get sorted out, but I'm not optimistic :-)

 

I have been burned in the past by switches that were *not* propagated there, so I'm supportive of you adding the two you mention.  (In fact, I added a comment to one of the switches files warning people to look at this function when adding switches.)

James

On Thu, Apr 19, 2012 at 8:42 PM, John Sheu <sh...@chromium.org> wrote:

I'm not disabling the sandbox and/or the GPU watchdog.  As I noted before, what I'm doing is adding those switches to the list of switches propagated into guest mode.

As it happens now, if you set "--disable-gpu-watchdog" or "--no-sandbox" in /sbin/session_manager_setup.sh on a ChromeOS machine, those switches are only applied to the Chrome stack running the login prompt.  If you sign in as a guest, Chrome restarts itself for the guest session over DBus, and those switches are not propagated to the guest session, which makes (for example) debugging the GPU process non-trivial.

I'll ask again: are there any landmines associated with enabling the propagation that I should be aware of?

-John Sheu

--
Chromium OS Developers mailing list: chromiu...@chromium.org
View archives, change email options, or unsubscribe:
http://groups.google.com/a/chromium.org/group/chromium-os-dev?hl=en

--
Chromium Developers mailing list: chromi...@chromium.org

View archives, change email options, or unsubscribe:

http://groups.google.com/a/chromium.org/group/chromium-dev

[John Sheu]

I'm not seeing any problems raised here, so I'm gonna start poking at reviewers.  Pretty please?

http://codereview.chromium.org/10228020/

Thanks,
-John Sheu