Seccomp-bpf: more policy cleanup.

Seccomp-bpf: more policy cleanup.

We make sure no system call is commented out anymore.
We change the policy in a minor way.

BUG=


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=151321

[jln (very slow on Chromium), 2012-08-10 20:37:17]

- We move system calls away from comments.
- We do some tiny policy changes. Those policy changes are clearly commented in
the review tool.

https://chromiumcodereview.appspot.com/10829286/diff/6001/content/common/sand...
File content/common/sandbox_seccomp_bpf_linux.cc (right):

https://chromiumcodereview.appspot.com/10829286/diff/6001/content/common/sand...
content/common/sandbox_seccomp_bpf_linux.cc:187: case __NR_getcwd:
These are now getting gracefully denied.

https://chromiumcodereview.appspot.com/10829286/diff/6001/content/common/sand...
content/common/sandbox_seccomp_bpf_linux.cc:198: case __NR_umask:
Getting newly gracefully denied via EPERM.

https://chromiumcodereview.appspot.com/10829286/diff/6001/content/common/sand...
content/common/sandbox_seccomp_bpf_linux.cc:289: case __NR_capget:
Newly allowed for consistency: should be relatively harmless

https://chromiumcodereview.appspot.com/10829286/diff/6001/content/common/sand...
content/common/sandbox_seccomp_bpf_linux.cc:406: case __NR_get_robust_list:
It's unfortunate that we have to allow futex at all at the moment, but as long
as we allow set_robust_list, there is really little reason to denying
get_robust_list.

Now we have consistent and fully working futex support.

https://chromiumcodereview.appspot.com/10829286/diff/6001/content/common/sand...
content/common/sandbox_seccomp_bpf_linux.cc:417: case __NR_epoll_create1:
create1 uses harmless flags.

https://chromiumcodereview.appspot.com/10829286/diff/6001/content/common/sand...
content/common/sandbox_seccomp_bpf_linux.cc:432: case __NR_pipe2:
pipe2 is pipe + harmless flags.

[Markus (顧孟勤), 2012-08-10 22:48:56]

lgtm

The small number of policy changes all look sane and seem quite reasonable.

The bulk of the CL involves some rearranging of code and renaming of functions.
All of those are good and make the code more readable.

There still are some significant issues with some of these policies, but we know
that; and they are outside the scope of this CL.

[Jorge Lucangeli Obes, 2012-08-10 23:19:35]

lgtm

https://chromiumcodereview.appspot.com/10829286/diff/6001/content/common/sand...
File content/common/sandbox_seccomp_bpf_linux.cc (right):

https://chromiumcodereview.appspot.com/10829286/diff/6001/content/common/sand...
content/common/sandbox_seccomp_bpf_linux.cc:205: // System calls that directly
access the file system. They might aquire
acquire... mind adding a small typo fix to your CL?

https://chromiumcodereview.appspot.com/10829286/diff/6001/content/common/sand...
content/common/sandbox_seccomp_bpf_linux.cc:279: case __NR_getdents64:       //
EPERM not a valid errno.
Why the switch?

https://chromiumcodereview.appspot.com/10829286/diff/6001/content/common/sand...
content/common/sandbox_seccomp_bpf_linux.cc:351: case __NR_signalfd4:
Why the switch?

[jln (very slow on Chromium), 2012-08-10 23:23:37]

Thanks!

https://chromiumcodereview.appspot.com/10829286/diff/6001/content/common/sand...
File content/common/sandbox_seccomp_bpf_linux.cc (right):

https://chromiumcodereview.appspot.com/10829286/diff/6001/content/common/sand...
content/common/sandbox_seccomp_bpf_linux.cc:205: // System calls that directly
access the file system. They might aquire
On 2012/08/10 23:19:35, Jorge Lucangeli Obes wrote:
> acquire... mind adding a small typo fix to your CL?

Done.

https://chromiumcodereview.appspot.com/10829286/diff/6001/content/common/sand...
content/common/sandbox_seccomp_bpf_linux.cc:279: case __NR_getdents64:       //
EPERM not a valid errno.
On 2012/08/10 23:19:35, Jorge Lucangeli Obes wrote:
> Why the switch?

Oops.

https://chromiumcodereview.appspot.com/10829286/diff/6001/content/common/sand...
content/common/sandbox_seccomp_bpf_linux.cc:351: case __NR_signalfd4:
On 2012/08/10 23:19:35, Jorge Lucangeli Obes wrote:
> Why the switch?

Oops.

[commit-bot: I haz the power, 2012-08-13 17:34:20]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/jln@chromium.org/10829286/3002

[commit-bot: I haz the power, 2012-08-13 18:46:49]

Change committed as 151321