Unify the sandboxed vs. unsandboxed code paths more, for two reasons:

Unify the sandboxed vs. unsandboxed code paths more, for two reasons:
1) It seems dangerous to have developers and tests running different code
paths to production users.
2) It's needed for the new in-progress seccomp filter sandbox.
Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=132472

[Jorge Lucangeli Obes, 2012-04-16 19:59:28]

https://chromiumcodereview.appspot.com/10082040/diff/1/content/browser/zygote...
File content/browser/zygote_main_linux.cc (right):

https://chromiumcodereview.appspot.com/10082040/diff/1/content/browser/zygote...
content/browser/zygote_main_linux.cc:860: 
So this will do the pre-sandbox init even if the SUID sandbox is not enabled,
because we also need the pre-init for the new seccomp sandbox right?

Does SkiaFontConfigSetImplementation(new
FontConfigIPC(kMagicSandboxIPCDescriptor)) work when the SUID sandbox is not
enabled? Given that most tests in the trybots run without the SUID sandbox, the
green try run suggests it does, but no harm in testing explicitly.

[Jorge Lucangeli Obes, 2012-04-16 20:04:42]

On 2012/04/16 19:59:28, Jorge Lucangeli Obes wrote:
>
https://chromiumcodereview.appspot.com/10082040/diff/1/content/browser/zygote...
> File content/browser/zygote_main_linux.cc (right):
> 
>
https://chromiumcodereview.appspot.com/10082040/diff/1/content/browser/zygote...
> content/browser/zygote_main_linux.cc:860: 
> So this will do the pre-sandbox init even if the SUID sandbox is not enabled,
> because we also need the pre-init for the new seccomp sandbox right?
> 
> Does SkiaFontConfigSetImplementation(new
> FontConfigIPC(kMagicSandboxIPCDescriptor)) work when the SUID sandbox is not
> enabled? Given that most tests in the trybots run without the SUID sandbox,
the
> green try run suggests it does, but no harm in testing explicitly.

lgtm after discussing with cevans@ in person.

[jln (very slow on Chromium), 2012-04-16 21:42:23]

LGTM

I'm very much in favor of such unifications.