Date: Mon, 23 Apr 2001 19:00:23 +0930 From: Greg Lehey <grog@lemis.com> To: Bruce Evans <bde@zeta.org.au> Cc: cvs-committers@FreeBSD.org, cvs-all@FreeBSD.org Subject: Re: cvs commit: src/sys/isofs/cd9660 cd9660_bmap.c cd9660_lookup.c cd9660_node.c cd9660_rrip.c cd9660_util.c cd9660_vfsops.c cd9660_vnops.c Message-ID: <20010423190023.D37321@wantadilla.lemis.com> In-Reply-To: <Pine.BSF.4.21.0104231911490.4712-100000@besplex.bde.org>; from bde@zeta.org.au on Mon, Apr 23, 2001 at 07:25:09PM %2B1000 References: <20010423183839.C37321@wantadilla.lemis.com> <Pine.BSF.4.21.0104231911490.4712-100000@besplex.bde.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Monday, 23 April 2001 at 19:25:09 +1000, Bruce Evans wrote: > On Mon, 23 Apr 2001, Greg Lehey wrote: > >> On Monday, 23 April 2001 at 18:57:53 +1000, Bruce Evans wrote: >>> Arggh. This is the start of breaking the includes in approx. 133 files to >>> work with recent breakage in <sys/mount.h>. Approx. 133 files that have >>> nothing to do with either networking or radixes now need to include >>> <net/radix.h>. >> >> Well, it doesn't exactly seem to be correct to redefine AF_MAX in >> sys/mount.h to get around it. But pass me the pointy hat: I checked >> with make world and forgot LINT. > > Defining AF_MAX in 2 places is easier than having a whole header just for > defining AF_MAX. The definitions won't get out of sync, because building > kernels will give warnings about them being different, and committed code > doesn't cause warnings ;-). > > The include of <net/radix.h> is unrelated to this. <net/radix.h> is already > close to being a secondary header that is included in primary headers like > <sys/mount.h> and <net/route.h>. It is only included directly in a few > networking .c files, and half of those includes are bogus anyway, since > the files also include <net/route.h>. (<net/route.h> doesn't include > <net/radix.h> if RNF_NORMAL is defined, but I don't see how defining > RNF_NORMAL could actually work in the kernel. It is apparently intended > for permitting an alternative to <net/radix.h>, but since other parts of > the kernel just include <net/radix.h>, there would probably be at best > namespace conflicts with any alternative.) Good analysis. You didn't answer the second part of my message: > So what would you suggest? At the moment, GENERIC builds. LINT doesn't. I have the fixes to LINT ready to commit, but there's not much point if you think I should back all the crap out. > BTW, 133 is from "grep -c sys/mount.h LINT/.depend". Yes, I thought as much. Greg -- Finger grog@lemis.com for PGP public key See complete headers for address and phone numbers To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe cvs-all" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20010423190023.D37321>