Date: Fri, 28 May 2004 11:44:14 -0600 From: Scott Long <scottl@freebsd.org> To: John Baldwin <jhb@freebsd.org> Cc: Marius Strobl <marius@alchemy.franken.de> Subject: Re: cvs commit: src/sys/sparc64/include atomic.h Message-ID: <40B77A6E.7000808@freebsd.org> In-Reply-To: <200405281340.47011.jhb@FreeBSD.org> References: <Pine.GSO.4.10.10405280430560.14569-100000@pcnet5.pcnet.com> <200405281340.47011.jhb@FreeBSD.org>
next in thread | previous in thread | raw e-mail | index | archive | help
John Baldwin wrote: > On Friday 28 May 2004 04:38 am, Daniel Eischen wrote: > >>On Fri, 28 May 2004, Mike Makonnen wrote: >> >>>On Sat, May 22, 2004 at 02:58:06AM +0200, Marius Strobl wrote: >>> >>>>On Fri, May 21, 2004 at 05:52:16PM -0700, Marius Strobl wrote: >>>> >>>>>marius 2004/05/21 17:52:16 PDT >>>>> >>>>> FreeBSD src repository >>>>> >>>>> Modified files: >>>>> sys/sparc64/include atomic.h >>>>> Log: >>>>> Use unsigned types for the arguments of the atomic(9) operations, >>>>> like described in the man page and done on all other architectures. >>>> >>>>This fixes the current build problems of libthr on sparc64 but the >>>>libthr code shouldn't rely on the integer-type of enums. >>> >>>Thanks. I'll change those enums to integers soon (others have suggested >>>it as well). It's just that I find them so convenient when debugging. >> >>And the use of atomic_cmpset_foo() isn't going to work on >>80386 since it doesn't have cmpxchgl. <machine/atomic.h> >>will try to use cli instead. > > > I'd say just punt and suggest that 80386 machines use libc_r. > I think that this is a reasonable suggestion. libc_r likely has a longer lifespan than 80386 at this point (hard to believe that the 80386 has been out for nearly 20 years!). Scott
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?40B77A6E.7000808>