On 6/30/2004 4:44 PM, Paul Mackerras wrote:

> David S. Miller writes:
>> On Wed, 30 Jun 2004 12:03:43 -0700
>> Arun Sharma <arun.sharma@intel.com> wrote:
>> 
>> > This system call seems to be duplicated in many architectures. Here's
>> > a patch which moves it to kernel/compat.c
>> > 
>> > We didn't touch ppc64, because it seems to be different. We've tested
>> > this on ia64 with LTP.
>> 
>> PPC64 just wants some kind of callback invoked right after the
>> do_adjtimex() call, it should be possible to provide a "do-nothing"
>> macro in linux/compat.h that ppc can override in asm/compat.h
> 
> That code we have in ppc64 is about 10 times too clever to be allowed
> to live.  It was basically trying to avoid the overhead of the rwlock
> back in the days when xtime was protected by an rwlock.  Now that we
> have seqlocks we could (and should :) get rid of it.

ok, the function has been eliminated from ppc64 as well.

	-Arun