Date: Wed, 20 Oct 2004 17:30:25 -0400 From: Garance A Drosihn <drosih@rpi.edu> To: Scott Long <scottl@freebsd.org>, Ruslan Ermilov <ru@freebsd.org> Cc: freebsd-arch@freebsd.org Subject: Re: [Fwd: What do people think about not installing a stripped /kernel ?] Message-ID: <p0611040ebd9c7ef46715@[128.113.24.47]> In-Reply-To: <4176C0C8.4060408@freebsd.org> References: <41767CF1.2020005@FreeBSD.org> <20041020.105839.100358845.imp@bsdimp.com> <20041020170907.GA1216@orion.daedalusnetworks.priv> <200410201913.42879.max@love2party.net> <20041020194547.GD2195@ip.net.ua> <4176C0C8.4060408@freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
At 1:47 PM -0600 10/20/04, Scott Long wrote: >Ruslan Ermilov wrote: >>On Wed, Oct 20, 2004 at 07:13:35PM +0200, Max Laier wrote: >> >>>Why is this discussion ongoing? The consensus seems pretty >>>clear: "Implement it, but have a make.conf option to turn >>>it off." If there is concern with this, make it default to >>>off and have an option to turn it on. >> >>Implementing this is very easy, since it's already implemented, >>just not by default. >> >>What everyone seem to have forgotten is that we also have modules, >>and in the "config -g" case, we also build debug versions of the >>modules. And if we're also going to install modules with debug >>symbols, I think this puts the requirement for the root file >>system way beyond the rational limits. > >I tend to agree. What do you think of my proposal to have >installkernel (optionally or whatever) put unstriped binaries >somewhere outside of the root partition? A long time ago I had an update which allowed the person to set where the debug version of kernels and modules would go, based on some environment variable in make.conf. I am pretty sure I even posted it. But it was for 4-STABLE, and the feedback was that it should first go into 5-current. This made a lot of sense, of course, but *I* only needed it on my 4.x-stable system... There were also major changes in the build process between 4-stable and 5-current, so I never reworked that change. It was much easier to just create a larger root partition on my 5.x test system... Hmm. I might even have that disk still spinning around somewhere. Yes, I seem to have it. Frightening! What I seem to have is a patch to 4.x (as of Nov 20 2001), which adds support for a KERNSAVDBGDIR= environment variable. It modifies sys/conf/kmod.mk and sys/conf/Makefile.i386 . I am not sure how useful it would be, seeing that it's for the wrong branch and it is from so long ago. But if people are interested, I could look into that. In any case, we could also change this so that the kernel is not stripped by default, but still leave modules stripped by default. What I think is important is that the default, generic install will set up users with a kernel that has SOME symbols in it. As I say, right now we're in the situation where we install one thing (stripped kernel & modules), but as soon as anyone reports a problem we tell them to build a non-stripped kernel or we can not help them. If the "rational" root partition is too small for a non-stripped kernel, then users are screwed when we given them that advice. So we need to change our definition of a rational root partition, or we simply admit that we (as developers) are not rational. Changing the *default* kernel that we install does not effect the actual size of a kernel which is USEFUL for debugging. The only thing we are changing is whether users START OUT with a useful debugging kernel. -- Garance Alistair Drosehn = gad@gilead.netel.rpi.edu Senior Systems Programmer or gad@freebsd.org Rensselaer Polytechnic Institute or drosih@rpi.edu
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?p0611040ebd9c7ef46715>