Ricky, do you want to take this?

Actually, this might be a good bug for thomasanderson@ to try ...

I attempted this as a first bug fix by removing the SetupSandbox() calls in browser_main_loop.cc and am currently recompiling and testing to see if it fixed the issue.

Tom, want to take a look at this now?

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/5ba7c75f39ad5e4f6bf67efb9f8d0b84355d4c96

commit 5ba7c75f39ad5e4f6bf67efb9f8d0b84355d4c96
Author: mdempsky <mdempsky@chromium.org>
Date: Sat Jun 11 03:39:45 2016

Fix logic for checking chrome-sandbox setuid binary

We only need to check for chrome-sandbox if we plan on launching
zygotes with the setuid sandbox or we need it to adjust OOM scores.
Also, the error only needs to be fatal when we want the setuid
sandbox.

BUG= 598454 

Review-Url: https://codereview.chromium.org/1976403002
Cr-Commit-Position: refs/heads/master@{#399361}

[modify] https://crrev.com/5ba7c75f39ad5e4f6bf67efb9f8d0b84355d4c96/content/browser/browser_main_loop.cc
[modify] https://crrev.com/5ba7c75f39ad5e4f6bf67efb9f8d0b84355d4c96/content/browser/zygote_host/zygote_communication_linux.cc
[modify] https://crrev.com/5ba7c75f39ad5e4f6bf67efb9f8d0b84355d4c96/content/browser/zygote_host/zygote_communication_linux.h
[modify] https://crrev.com/5ba7c75f39ad5e4f6bf67efb9f8d0b84355d4c96/content/browser/zygote_host/zygote_host_impl_linux.cc
[modify] https://crrev.com/5ba7c75f39ad5e4f6bf67efb9f8d0b84355d4c96/content/browser/zygote_host/zygote_host_impl_linux.h

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/5ba7c75f39ad5e4f6bf67efb9f8d0b84355d4c96

commit 5ba7c75f39ad5e4f6bf67efb9f8d0b84355d4c96
Author: mdempsky <mdempsky@chromium.org>
Date: Sat Jun 11 03:39:45 2016

Fix logic for checking chrome-sandbox setuid binary

We only need to check for chrome-sandbox if we plan on launching
zygotes with the setuid sandbox or we need it to adjust OOM scores.
Also, the error only needs to be fatal when we want the setuid
sandbox.

BUG= 598454 

Review-Url: https://codereview.chromium.org/1976403002
Cr-Commit-Position: refs/heads/master@{#399361}

[modify] https://crrev.com/5ba7c75f39ad5e4f6bf67efb9f8d0b84355d4c96/content/browser/browser_main_loop.cc
[modify] https://crrev.com/5ba7c75f39ad5e4f6bf67efb9f8d0b84355d4c96/content/browser/zygote_host/zygote_communication_linux.cc
[modify] https://crrev.com/5ba7c75f39ad5e4f6bf67efb9f8d0b84355d4c96/content/browser/zygote_host/zygote_communication_linux.h
[modify] https://crrev.com/5ba7c75f39ad5e4f6bf67efb9f8d0b84355d4c96/content/browser/zygote_host/zygote_host_impl_linux.cc
[modify] https://crrev.com/5ba7c75f39ad5e4f6bf67efb9f8d0b84355d4c96/content/browser/zygote_host/zygote_host_impl_linux.h

With this change there is only one use of kDisableSetuidSandbox remaining as far as I can see (apart from some forwarding)- and if control ever reaches the one point where it's checked and it is enabled then we LOG(FATAL).

kDisableSetuidSandbox therefore seems to be useless- should it be removed?

The LOG(FATAL) is the use. :)

The intention is if you want to run Chrome and only use the namespace sandbox, you can set --disable-setuid-sandbox.  But if you do so on a host without appropriate kernel support for the namespace sandbox, Chrome will loudly refuse to run.