Make sandbox/linux/seccomp-bpf/ follow the style guide.

Make sandbox/linux/seccomp-bpf/ follow the style guide.

This CL has mostly be generated with clang-format and should not
introduce any code change other than reformatting.

BUG=316486
R=markus@chromium.org, rsesek@chromium.org

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=234013

[jln (very slow on Chromium), 2013-11-08 20:14:08]

This CL is just reformatting code, mostly with clang-format.

I don't think manual review would be a good use of time, but feel free to take a
look or blindly approve!

[Markus (顧孟勤), 2013-11-08 20:48:11]

I am sure this is fine, if you automatically reformatted and if the unittests
still pass. But for some reason the review tool is only showing me about half of
the CL. For the rest it keeps giving errors.

Is that something we can fix? Maybe, re-upload?

[jln (very slow on Chromium), 2013-11-08 20:53:05]

On 2013/11/08 20:48:11, Markus (顧孟勤) wrote:
> I am sure this is fine, if you automatically reformatted and if the unittests
> still pass. But for some reason the review tool is only showing me about half
of
> the CL. For the rest it keeps giving errors.
> 
> Is that something we can fix? Maybe, re-upload?

I've just re-uploaded, PTAL!

I did some manual work in some files, such as sandbox_bpf_unittest.cc or
codegen.cc

Giant Cond() structures are better manually formatted.

[Robert Sesek, 2013-11-08 21:14:33]

Overall LGTM. I did go through it, but it was mostly just skimming.

https://codereview.chromium.org/66723007/diff/150001/sandbox/linux/seccomp-bp...
File sandbox/linux/seccomp-bpf/die.cc (right):

https://codereview.chromium.org/66723007/diff/150001/sandbox/linux/seccomp-bp...
sandbox/linux/seccomp-bpf/die.cc:83: if (HANDLE_EINTR(SandboxSyscall(__NR_write,
2, s.c_str(), s.length()))) {
This is an if() with an empty body.

https://codereview.chromium.org/66723007/diff/150001/sandbox/linux/seccomp-bp...
File sandbox/linux/seccomp-bpf/linux_seccomp.h (right):

https://codereview.chromium.org/66723007/diff/150001/sandbox/linux/seccomp-bp...
sandbox/linux/seccomp-bpf/linux_seccomp.h:5: #ifndef
SANDBOX_LINUX_SECCOMP_BPF_LINUX_SECCOMP_H__
I think this file was more readable before.

https://codereview.chromium.org/66723007/diff/150001/sandbox/linux/seccomp-bp...
File sandbox/linux/seccomp-bpf/verifier.cc (right):

https://codereview.chromium.org/66723007/diff/150001/sandbox/linux/seccomp-bp...
sandbox/linux/seccomp-bpf/verifier.cc:50: ? uint32_t(data.args[code.argno()])
For another CL: nested ternary is unreadable.

https://codereview.chromium.org/66723007/diff/150001/sandbox/linux/tests/unit...
File sandbox/linux/tests/unit_tests.cc (right):

https://codereview.chromium.org/66723007/diff/150001/sandbox/linux/tests/unit...
sandbox/linux/tests/unit_tests.cc:73: if (write(2, failure_message,
sizeof(failure_message) - 1) < 0) {
Another empty if()

[jln (very slow on Chromium), 2013-11-08 21:22:36]

Thanks Robert!

https://codereview.chromium.org/66723007/diff/150001/sandbox/linux/seccomp-bp...
File sandbox/linux/seccomp-bpf/die.cc (right):

https://codereview.chromium.org/66723007/diff/150001/sandbox/linux/seccomp-bp...
sandbox/linux/seccomp-bpf/die.cc:83: if (HANDLE_EINTR(SandboxSyscall(__NR_write,
2, s.c_str(), s.length()))) {
On 2013/11/08 21:14:33, rsesek wrote:
> This is an if() with an empty body.

This is to tame compilers that insist on checking the return value of
HANDLE_EINTR().

We can discuss this, but in another CL, I really don't want to touch anything
but formatting here.

https://codereview.chromium.org/66723007/diff/150001/sandbox/linux/seccomp-bp...
File sandbox/linux/seccomp-bpf/linux_seccomp.h (right):

https://codereview.chromium.org/66723007/diff/150001/sandbox/linux/seccomp-bp...
sandbox/linux/seccomp-bpf/linux_seccomp.h:5: #ifndef
SANDBOX_LINUX_SECCOMP_BPF_LINUX_SECCOMP_H__
On 2013/11/08 21:14:33, rsesek wrote:
> I think this file was more readable before.

Agreed, reverted.

https://codereview.chromium.org/66723007/diff/150001/sandbox/linux/seccomp-bp...
File sandbox/linux/seccomp-bpf/verifier.cc (right):

https://codereview.chromium.org/66723007/diff/150001/sandbox/linux/seccomp-bp...
sandbox/linux/seccomp-bpf/verifier.cc:50: ? uint32_t(data.args[code.argno()])
On 2013/11/08 21:14:33, rsesek wrote:
> For another CL: nested ternary is unreadable.

Agreed!

https://codereview.chromium.org/66723007/diff/150001/sandbox/linux/tests/unit...
File sandbox/linux/tests/unit_tests.cc (right):

https://codereview.chromium.org/66723007/diff/150001/sandbox/linux/tests/unit...
sandbox/linux/tests/unit_tests.cc:73: if (write(2, failure_message,
sizeof(failure_message) - 1) < 0) {
On 2013/11/08 21:14:33, rsesek wrote:
> Another empty if()

ack.

[Markus (顧孟勤), 2013-11-08 21:24:18]

lgtm

https://codereview.chromium.org/66723007/diff/150001/sandbox/linux/seccomp-bp...
File sandbox/linux/seccomp-bpf/die.cc (right):

https://codereview.chromium.org/66723007/diff/150001/sandbox/linux/seccomp-bp...
sandbox/linux/seccomp-bpf/die.cc:83: if (HANDLE_EINTR(SandboxSyscall(__NR_write,
2, s.c_str(), s.length()))) {
On 2013/11/08 21:14:33, rsesek wrote:
> This is an if() with an empty body.

Yes. That's as-intended. It makes the compiler shut up about us not using the
return value of HANDLE_EINTR(). After discussing with Jeffrey Yasskin, we
decided that this is the most standards-compliant way to do that.

[jln (very slow on Chromium), 2013-11-08 21:36:05]

https://codereview.chromium.org/66723007/diff/150001/sandbox/linux/seccomp-bp...
File sandbox/linux/seccomp-bpf/die.cc (right):

https://codereview.chromium.org/66723007/diff/150001/sandbox/linux/seccomp-bp...
sandbox/linux/seccomp-bpf/die.cc:83: if (HANDLE_EINTR(SandboxSyscall(__NR_write,
2, s.c_str(), s.length()))) {
On 2013/11/08 21:24:18, Markus (顧孟勤) wrote:
> On 2013/11/08 21:14:33, rsesek wrote:
> > This is an if() with an empty body.
> 
> Yes. That's as-intended. It makes the compiler shut up about us not using the
> return value of HANDLE_EINTR(). After discussing with Jeffrey Yasskin, we
> decided that this is the most standards-compliant way to do that.

Now that we can use base/, we are saved: Robert pointed out to me that there is
a ignore_result() ready.

[jln (very slow on Chromium), 2013-11-08 21:58:10]

Committed patchset #4 manually as r234013 (presubmit successful).