[Date Prev][Date Next] [Thread Prev][Thread Next]
[Thread Index]
[Date Index]
[Author Index]
Re: First cut of PowerPC support in NPTL
- From: Ian Wienand <ianw gelato unsw edu au>
- To: phil-list redhat com
- Cc: sjmunroe us ibm com, paulus au1 ibm com
- Subject: Re: First cut of PowerPC support in NPTL
- Date: Fri, 7 Mar 2003 13:08:02 +1100
On Fri, Mar 07, 2003 at 11:20:16AM +1100, Paul Mackerras wrote:
> I have been working on adding support for PowerPC in NPTL. The patch
> below shows where I am up to with it. I am sending it out at this
> stage in the hope of getting some useful comments and criticism, even
> though it is not finished or working yet.
Hi,
I have an IA64 port that is a few inches futher along (just some
tricky problems when I clone) that I haven't been brave enough to
release (it's also not against the latest release).
I think before either of these goes too much futher it might be good
to abstract lowlevellock.h more. All that is really system dependent
is an atomic_increment / atomic_decrement and a compare and swap for
the trylock. I would leave the ia32 (optimised asm) and s390 ports
(uses compare and swap to do inc/dec) alone, but we should both use
generic versions, as we're basically doing the same thing.
If there's no objections over the next few days I might update my
non-complete ia64 version to the latest code and move as much as i can
into generic, then we can work on tweaking it? My implementation of
the locking is described in
http://www.gelato.unsw.edu.au/IA64wiki/lowlevellock.h
if I have anything wrong, please edit the document.
-i
ianw gelato unsw edu au
http://www.gelato.unsw.edu.au
[Date Prev][Date Next] [Thread Prev][Thread Next]
[Thread Index]
[Date Index]
[Author Index]