Non-SFI NaCl: Allow openat system call to gracefully fail on BareMetal seccomp-bpf.

Non-SFI NaCl: Allow openat system call to gracefully fail on BareMetal seccomp-bpf.

sysconf via get_nprocs_conf glibc function calls opendir() which calls
openat() syscall, make them gracefully fail with EPERM instead of killing with SIGSYS.

This fixes crash trying to start chrome tracing when running
application with BareMetal with recent glibc.

BUG=407516
TESTED=launched BareMetal NaCl app and started chrome tracing and doesn't crash.

Committed: https://crrev.com/8348200de4dbd124b18dd2ecbf6cd44368083012
Cr-Commit-Position: refs/heads/master@{#292287}

[Junichi Uekawa, 2014-08-27 05:44:30]

uekawa@chromium.org changed reviewers:
+ hamaji@chromium.org, mazda@chromium.org, mseaborn@chromium.org

[Junichi Uekawa, 2014-08-27 05:44:30]

Hi Mark, PTAL.

FYI mazda, hamaji.

[hamaji, 2014-08-27 05:59:55]

Could you add "Non-SFI NaCl:" prefix to the subject of this change?

https://codereview.chromium.org/509783003/diff/1/components/nacl/loader/nonsf...
File components/nacl/loader/nonsfi/nonsfi_sandbox_unittest.cc (right):

https://codereview.chromium.org/509783003/diff/1/components/nacl/loader/nonsf...
components/nacl/loader/nonsfi/nonsfi_sandbox_unittest.cc:517:
RESTRICT_SYSCALL_EPERM_TEST(openat);
I think you should remove

RESTRICT_SYSCALL_DEATH_TEST(openat);

in nonsfi_sandbox_sigsys_unittest

[Junichi Uekawa, 2014-08-27 06:21:24]

https://codereview.chromium.org/509783003/diff/1/components/nacl/loader/nonsf...
File components/nacl/loader/nonsfi/nonsfi_sandbox_unittest.cc (right):

https://codereview.chromium.org/509783003/diff/1/components/nacl/loader/nonsf...
components/nacl/loader/nonsfi/nonsfi_sandbox_unittest.cc:517:
RESTRICT_SYSCALL_EPERM_TEST(openat);
On 2014/08/27 05:59:55, hamaji wrote:
> I think you should remove
> 
> RESTRICT_SYSCALL_DEATH_TEST(openat);
> 
> in nonsfi_sandbox_sigsys_unittest

Done.

[Junichi Uekawa, 2014-08-27 22:48:32]

On 2014/08/27 05:59:55, hamaji wrote:
> Could you add "Non-SFI NaCl:" prefix to the subject of this change?


done

[jln (very slow on Chromium), 2014-08-27 22:53:59]

lgtm

Modern libcs also tend to replace open with openat and open doesn't even exist
in the newest system call interfaces (like ARM64).

[Mark Seaborn, 2014-08-27 23:08:20]

LGTM

[Junichi Uekawa, 2014-08-27 23:38:07]

The CQ bit was checked by uekawa@chromium.org

[commit-bot: I haz the power, 2014-08-27 23:39:46]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/uekawa@chromium.org/509783003/20001

[commit-bot: I haz the power, 2014-08-28 00:48:11]

Committed patchset #2 (id:20001) as c01b19c52cf60eb92eaa7c4a84acf0ccd3051681

[commit-bot: I haz the power, 2014-09-10 02:56:17]

Patchset 2 (id:??) landed as
https://crrev.com/8348200de4dbd124b18dd2ecbf6cd44368083012
Cr-Commit-Position: refs/heads/master@{#292287}