Date: Sat, 18 Mar 1995 00:42:36 -0500 From: Chris G Demetriou <Chris_G_Demetriou@LAGAVULIN.PDL.CS.CMU.EDU> To: "Rodney W. Grimes" <rgrimes@gndrsh.aac.dev.com> Cc: davidg@Root.COM, CVS-commiters@freefall.CDROM.COM, cvs-sys@freefall.CDROM.COM Subject: Re: cvs commit: src/sys/libkern locc.c random.c scanc.c skpc.c libkern.h Message-ID: <199503180543.VAA13038@freefall.cdrom.com> In-Reply-To: Your message of "Fri, 17 Mar 1995 20:16:04 PST." <199503180416.UAA20285@gndrsh.aac.dev.com>
next in thread | previous in thread | raw e-mail | index | archive | help
> I think we should just stop building libkern, and for machines that need > routines from there add them to sys/i386/files.i386. For people doing a > new port it is nice to have these generic C functions around, but other > than that I see libkern.a as an evil thing to have done. Among other things, it's damn near impossible to compile .o's in 'kernel libraries' with the correct flags, if you're trying to support multiple architectures. the files in libkern are supposed to be "machine-dependent", but there's certainly a need to compile them with different options on different architectures. (why? (1) the gcc optimizer should be considered "known broken" for new ports, until proved otherwise, (2) each different arch has a slightly different way to compile without FP instructions, (3) on architectures where it's natual to use a GP, you gnerally need to do something slightly different for the kernel, etc.) cgd
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199503180543.VAA13038>