Date: Wed, 29 Oct 1997 23:07:10 +0000 (GMT) From: Terry Lambert <tlambert@primenet.com> To: asami@cs.berkeley.edu (Satoshi Asami) Cc: phk@critter.freebsd.dk, freebsd-current@freefall.FreeBSD.org Subject: Re: bad system call - world build Message-ID: <199710292307.QAA16732@usr02.primenet.com> In-Reply-To: <199710291108.DAA14491@silvia.HIP.Berkeley.EDU> from "Satoshi Asami" at Oct 29, 97 03:08:13 am
next in thread | previous in thread | raw e-mail | index | archive | help
> * I can't see any reason why make world should need to run any program > * that is linked to the newly compiled libc... And in that respect I > * think that most of your work is headed the wrong direction... > > That is because some of the tools required to build the world may need > functions that only exist in new libraries. It has always been that > way, I haven't changed it. I have to agree with Poul on this particular one. If you are depending on intermediate tools, then a cross-build is out of the question. Most likely, the intermediate tools should be targeted to the build environment, not the target environment. Better would be if there were no intermediate tools needed at all. > * I'm very willing to help, but only if we attack the correct problem: > * > * "How can I compile my FreeBSD sources on a foreign platform". > > Here you go again, throwing a smokescreen. I don't mind you planning > ahead, go ahead and think about it all you like. I will take that as > a "no", so please stop changing the subject. :( I agree with Poul on this one, too. This is the real problem; just because the architecture is the same, it's no less a cross-build. The getcwd() code already has to be able to fall back in the case that the data wasn't in cache in the kernel. The library wrapper should treat ENOSYS as this kind of failure as well, and the problem will go away. Now I'm kind of glad Poul rejected my suggestion to move the old code into the kernel... it leaves you an "out". But I still think that any code in your build tree should not be run unless it's installed. I had a bear of a time when the .mk file changes to use the "strip" feature of "ld" instead of just calling "strip". Now *that* was a gratuitous change, dwarfed only by assembly syntax changes (and those are GNU bugs, and are therefore out of our control). Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199710292307.QAA16732>