Remove SandboxBPF's dependency on CompatibilityPolicy

Remove SandboxBPF's dependency on CompatibilityPolicy

SandboxBPF users are now required to always provide a SandboxBPFPolicy
instead of a SyscallEvaluator.  CompatibilityPolicy can't be removed
just yet though because it's still used by the deprecated BPF_TEST
macros.

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=271607

[mdempsky, 2014-05-16 22:38:07]

Building on top of the previous CL to simplify SandboxBPF policies, this one
converts some of the legacy SyscallEvaluator implementations to use a
SandboxBPFPolicy instead.  (Still more to do, but this seemed like a good
intermediate point for review.)

[jln (very slow on Chromium), 2014-05-19 20:57:19]

lgtm, thanks for the lceanup!

https://chromiumcodereview.appspot.com/290223002/diff/20001/sandbox/linux/sec...
File sandbox/linux/seccomp-bpf/sandbox_bpf.cc (right):

https://chromiumcodereview.appspot.com/290223002/diff/20001/sandbox/linux/sec...
sandbox/linux/seccomp-bpf/sandbox_bpf.cc:264: // would be needed to support
threads.
It's still true that we can enable forking with threads with only minor changes
if needed (for Android) here, just wanted to call your attention to this.

But I think we might want to just completely remove this function. It used to be
that old glibc would still use vsyscall and not work with seccomp-bpf and there
were a few other things to test for. But I think we may be mature enough to
simply detect if seccomp-bpf is present in the kernel and expect it to work.

[mdempsky, 2014-05-19 21:01:28]

https://chromiumcodereview.appspot.com/290223002/diff/20001/sandbox/linux/sec...
File sandbox/linux/seccomp-bpf/sandbox_bpf.cc (right):

https://chromiumcodereview.appspot.com/290223002/diff/20001/sandbox/linux/sec...
sandbox/linux/seccomp-bpf/sandbox_bpf.cc:264: // would be needed to support
threads.
On 2014/05/19 20:57:19, jln wrote:
> It's still true that we can enable forking with threads with only minor
changes
> if needed (for Android) here, just wanted to call your attention to this.
> 
> But I think we might want to just completely remove this function. It used to
be
> that old glibc would still use vsyscall and not work with seccomp-bpf and
there
> were a few other things to test for. But I think we may be mature enough to
> simply detect if seccomp-bpf is present in the kernel and expect it to work.

Noted.  I'll look into this as a followup cleanup.

[mdempsky, 2014-05-19 21:01:35]

The CQ bit was checked by mdempsky@chromium.org

[commit-bot: I haz the power, 2014-05-19 21:02:20]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/mdempsky@chromium.org/290223002/20001

[mdempsky, 2014-05-20 00:13:17]

The CQ bit was unchecked by mdempsky@chromium.org

[mdempsky, 2014-05-20 00:21:17]

The CQ bit was checked by mdempsky@chromium.org

[mdempsky, 2014-05-20 00:21:48]

On 2014/05/20 00:21:17, mdempsky wrote:
> The CQ bit was checked by mailto:mdempsky@chromium.org

FYI, I added DISALLOW_COPY_AND_ASSIGN to each of the new policies per
chromium-dev@ discussion.

[commit-bot: I haz the power, 2014-05-20 00:21:53]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/mdempsky@chromium.org/290223002/30001

[mdempsky, 2014-05-20 00:35:16]

The CQ bit was checked by mdempsky@chromium.org

[commit-bot: I haz the power, 2014-05-20 00:36:23]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/mdempsky@chromium.org/290223002/50001

[mdempsky, 2014-05-20 00:40:35]

The CQ bit was unchecked by mdempsky@chromium.org

[mdempsky, 2014-05-20 06:50:48]

The CQ bit was checked by mdempsky@chromium.org

[commit-bot: I haz the power, 2014-05-20 06:51:48]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/mdempsky@chromium.org/290223002/70001

[commit-bot: I haz the power, 2014-05-20 07:25:13]

Change committed as 271607

[Ilya Sherman, 2014-05-20 11:15:41]

A revert of this CL has been created in
https://codereview.chromium.org/293993006/ by isherman@chromium.org.

The reason for reverting is: Broke VerboseAPITesting in sandbox_linux_unittests
on Linux:
http://build.chromium.org/p/chromium.linux/buildstatus?builder=Linux%20Tests%....

[earthdok, 2014-05-20 12:12:45]

On 2014/05/20 11:15:41, Ilya Sherman wrote:
> A revert of this CL has been created in
> https://codereview.chromium.org/293993006/ by mailto:isherman@chromium.org.
> 
> The reason for reverting is: Broke VerboseAPITesting in
sandbox_linux_unittests
> on Linux:
>
http://build.chromium.org/p/chromium.linux/buildstatus?builder=Linux%20Tests%....

FWIW MSan reported an uninit in that test:

http://build.chromium.org/p/chromium.fyi/builders/Chromium%20Linux%20MSan/bui...