Linux GPU sandbox: only allocate broker policy in the broker.

Linux GPU sandbox: only allocate broker policy in the broker.

The GPU broker policy was allocated in the main GPU process and then used in
the broker process. We switch the logic so that the broker policy is only ever
allocated in the broker process itself.

Besides fixing a small memory leak (in the GPU process), this makes sure that a
policy is only ever used in the process that allocated it. This will allow to
bind policies with properties such as "which processes does this policy allow
to send signal to".

BUG=367986
R=jorgelo@chromium.org

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=266726

[jln (very slow on Chromium), 2014-04-28 23:11:42]

Jorge, could you please take a look at this small CL?

[Jorge Lucangeli Obes, 2014-04-28 23:43:53]

lgtm

https://codereview.chromium.org/251183004/diff/1/content/common/sandbox_linux...
File content/common/sandbox_linux/bpf_gpu_policy_linux.h (right):

https://codereview.chromium.org/251183004/diff/1/content/common/sandbox_linux...
content/common/sandbox_linux/bpf_gpu_policy_linux.h:38:
sandbox::SandboxBPFPolicy* (*broker_sandboxer_allocator)(void),
It was nice to have base::Callback's instead of function pointers, they're
definitely easier to read. But I guess this way you can just pass the static
method directly.

[jln (very slow on Chromium), 2014-04-28 23:53:17]

Thanks Jorge!

https://codereview.chromium.org/251183004/diff/1/content/common/sandbox_linux...
File content/common/sandbox_linux/bpf_gpu_policy_linux.h (right):

https://codereview.chromium.org/251183004/diff/1/content/common/sandbox_linux...
content/common/sandbox_linux/bpf_gpu_policy_linux.h:38:
sandbox::SandboxBPFPolicy* (*broker_sandboxer_allocator)(void),
On 2014/04/28 23:43:53, Jorge Lucangeli Obes wrote:
> It was nice to have base::Callback's instead of function pointers, they're
> definitely easier to read. But I guess this way you can just pass the static
> method directly.

We could still make it a CallBack if we really wanted to, but the CallBack
taking a CallBack as a parameter that goes across fork() seemed too complicated
to follow.

[jln (very slow on Chromium), 2014-04-28 23:58:24]

The CQ bit was checked by jln@chromium.org

[commit-bot: I haz the power, 2014-04-29 00:00:16]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/jln@chromium.org/251183004/1

[commit-bot: I haz the power, 2014-04-29 00:46:23]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2014-04-29 00:46:24]

Try jobs failed on following builders:
  tryserver.chromium on chromium_presubmit
  tryserver.chromium on linux_chromium_rel

[jln (very slow on Chromium), 2014-04-29 01:16:01]

Committed patchset #1 manually as r266726 (presubmit successful).