Non-SFI NaCl: Allow CLOCK_SYSTEM_TRACE on Chrome OS

sandbox/linux/services/linux_syscalls.h

Index: sandbox/linux/services/linux_syscalls.h
diff --git a/sandbox/linux/services/linux_syscalls.h b/sandbox/linux/services/linux_syscalls.h
index 77c1be8b82a0fe4aca308ca40547ebf7a008d24a..62fe35ed1f8d59e5ef2c782b8a06de0a7a4b16a6 100644
--- a/sandbox/linux/services/linux_syscalls.h
+++ b/sandbox/linux/services/linux_syscalls.h
@@ -21,5 +21,8 @@
 #include "sandbox/linux/services/arm_linux_syscalls.h"
 #endif
 
-#endif  // SANDBOX_LINUX_SERVICES_LINUX_SYSCALLS_H_
+// A Chrome OS specific clock ID for clock_getres, clock_gettime, and
+// clock_settime. See base/time/time_posix.cc for detail.
+#define CLOCK_SYSTEM_TRACE 11

[hamaji, 2014/05/28 07:36:46]

Not sure this is the appropriate location for this constant. I put this here as
Julien said we might want to do the same check for clock_gettime even from the
base policy.

[Mark Seaborn, 2014/05/28 15:24:03]

Would it make sense to change base/time/time_posix.cc to share a common
definition of this constant?

Where can I learn what CLOCK_SYSTEM_TRACE does?  time_posix.cc doesn't have any
pointers.  git annotate takes me to
https://code.google.com/p/chromium/issues/detail?id=128650, which is not
particularly informative.

I did find this:
http://code.google.com/p/chromium/issues/detail?id=324025
"ChromeOS 3.10 tracing CLOCK_SYSTEM_TRACE conflicts with CLOCK_TAI"

"The version of the CLOCK_SCHED_TRACE patch in chromeos-3.10 has
CLOCK_SYSTEM_TRACE defined at 12 instead of 11..."

-- This suggests that having a single #define would be important!

[hamaji, 2014/05/28 16:21:47]

Sounds like a good idea. Done.
I have added sleffler as a reviewer. Though it seems he is not working for
Chromium anymore, I hope he will give us a pointer.

 
+#endif  // SANDBOX_LINUX_SERVICES_LINUX_SYSCALLS_H_