Re: compat_sys_times() bogus until jiffies >= 0.

[Andrew Morton <akpm@linux-foundation.org>]

> On Thu, 8 Nov 2007 16:36:08 +1100 Paul Mackerras <paulus@samba.org> wrote:
> Andrew Morton writes:
> 
> > Yup.  But userspace will already have a fit if either the start or end time
> > advanced into the glibc-thought-that-was-an-error range.
> 
> Not nearly as much of a fit.  The effect on x86 is that values between
> -4095 and -1 are reported as -1, so the end-start difference will be
> out by less than 41 seconds.  That's not nearly as dramatic as a
> difference of 21 million seconds (over 16 years). :)
> 
> I really think that wrapping at 0x7fffffff makes the situation worse,
> not better.
> 

Sure.

So we need to do what you say: never return an error from sys_times() and
change glibc to not perform error-interpretation on sys_times() return
values and recommend that people bypass libc and go direct to the syscall
so they'll work correctly on older glibc.   Lovely.

I wonder what happens with things like F_GETOWN, shmat() and lseek(/dev/mem)
on x86 (things which use force_successful_syscall_return()).  According
to the comment in include/linux/ptrace.h, glibc should be special-casing
these.