Linux brk() syscall returns the new program break

Linux brk() syscall returns the new program break

The behavior of the Linux brk() system call is different from the
behavior provided by the glibc wrapper. The actual system call
returns the new program break on success. On failure, the system
call returns the current break.
                
This change fixes the LSS brk() declaration to match the Linux system
call rather than that of the glibc wrapper.

BUG= none
TEST= none
R=mseaborn@chromium.org

Committed: https://code.google.com/p/linux-syscall-support/source/detail?r=22

[Mark Seaborn, 2013-08-14 18:10:01]

Please rerun "git cl upload":
https://codereview.chromium.org/22999007/diff/1/linux_syscall_support.h says
"error: old chunk mismatch".

[Petr Hosek, 2013-08-14 18:12:56]

On 2013/08/14 18:10:01, Mark Seaborn wrote:
> Please rerun "git cl upload":
> https://codereview.chromium.org/22999007/diff/1/linux_syscall_support.h says
> "error: old chunk mismatch".

Done, seems to be working fine now.

[Mark Seaborn, 2013-08-14 18:17:54]

LGTM.  Please set "BUG=none" and "TEST=none": don't just leave these fields
empty.

What is the motivation for the change, BTW?  How did you come to notice that
'int' wasn't the right return type?

[Petr Hosek, 2013-08-14 18:39:25]

On 2013/08/14 18:17:54, Mark Seaborn wrote:
> LGTM.  Please set "BUG=none" and "TEST=none": don't just leave these fields
> empty.
> 
> What is the motivation for the change, BTW?  How did you come to notice that
> 'int' wasn't the right return type?

I used LSS to implement Newlib system call stubs for x86 and x86-64 Linux in my
private project and while implementing the _sbrk_r stub I've noticed that the
LSS brk declaration is incorrect.

[Petr Hosek, 2013-08-16 17:31:56]

Committed patchset #2 manually as r22 (presubmit successful).