Index | Thread | Search

From:
Claudio Jeker <cjeker@diehard.n-r-g.com>
Subject:
Re: boost sparc64 support for context
To:
Otto Moerbeek <otto@drijf.net>
Cc:
ports@openbsd.org, Rafael Sadowski <rsadowski@openbsd.org>, Brad Smith <brad@comstyle.com>
Date:
Tue, 13 Feb 2024 12:12:29 +0100

Download raw body.

Thread
On Tue, Feb 13, 2024 at 12:08:06PM +0100, Otto Moerbeek wrote:
> On Tue, Feb 13, 2024 at 09:03:49AM +0100, Claudio Jeker wrote:
> 
> > On Thu, Feb 08, 2024 at 06:50:55PM +0100, Claudio Jeker wrote:
> > > Since I did most of the work for php I decided to also implement the fiber
> > > / context ASM for sparc64 in boost.
> > > 
> > > The difference is that boost has a ontop_fcontext() function which is
> > > not really documented but in the end I figured it out.
> > > 
> > > I did test this against the tests in boost-context. Will do more tests
> > > tomorrow but maybe someone else wants to join the context party :)
> > 
> > I realized that I can save an instruction by using the builtin ADD
> > instruction in 'restore' to set up %o0. So a context switch on sparc64
> > is now 10 instructions :)
> > 
> > I ran test for context, fiber, coroutine and coroutine2 in boost and all
> > test pass. The powerdns_recoursor package built but I'm not sure how to
> > test this properly.
> 
> If it starts and does not crash you're good. My sparc64s are currently
> in storage, so I cannot test myself right now as our attic will be
> under construction soon. Will test when things are back to normal.
> 

It does start and even resolve properly. So I guess we're good here :)

-- 
:wq Claudio