On Tue, Jun 9, 2009 at 7:33 AM, Dan Kegel<[email protected]> wrote:
> Yes.  I tried leaving the initial process as browser, and making the
> zygote its first child,
> but it was hard to get all the UI tests passing that way (the process
> filtering in
> chrome_process_utils.cc in particular didn't work that way).

I noticed this too, but we'll have to switch it back. 90% of the
reason for having a zygote process is to get different ASLR for the
renderers, otherwise we would just fork from the browser process and
rewind the heap.

I'm doing a refactoring pass at the moment to:
  * Remove the list of magic fds from base (since that's a Chrome
specific thing)
  * Remove the lock file
  * Remove the ability for renderers to screw around with message order
  * Make sure the renderers have a different ASLR from the browser
(although all renderers will have the same ASL).

I was going to merge the SUID sandbox stuff with this as well, but
then face palmed as I realised that it breaks the SUID sandbox anyway.



AGL

--~--~---------~--~----~------------~-------~--~----~
Chromium Developers mailing list: [email protected] 
View archives, change email options, or unsubscribe: 
    http://groups.google.com/group/chromium-dev
-~----------~----~----~----~------~----~------~--~---

Reply via email to