Date: Sun, 10 Sep 95 11:24:39 EDT From: jleppek@suw2k.ess.harris.com (James Leppek) To: freebsd-current@freefall.FreeBSD.org Subject: Re: Sig 11 and -current problems. Message-ID: <9509101524.AA01524@borg.ess.harris.com>
next in thread | raw e-mail | index | archive | help
august 29th kernel fails with fatal trap 26 page not present... and crash any attempt to continue the make after it fails with a sig11 or 10 fails again with another signal. I have had to reboot to continue the process. here are the steps I have followed and its still building: 1)hand walk the build to get compiler, libs , lkms and other critical components, build worlds will crash, just reboot and keep going 2)build a new kernel 3)try make world, it won't work but should get farther 4) try it again, mine is still compiling but it is the longest run yet Jim Leppek > From radha@ccnet.ccnet.com Sun Sep 10 10:42:51 1995 > To: jleppek@suw2k.ess.harris.com (James Leppek) > Subject: Re: Sig 11 and -current problems. > Date: Sun, 10 Sep 1995 07:36:22 +0000 > From: Radha Krishnan <radha@ccnet.ccnet.com> > > >well to add more fuel to the fire I still have the sig11's :-( > >I have tried to "hand walk" the process, and was able to do it > >once but then after that a make world died again > >my make worlds always(90% of the time) die during the installation > >of libcompat?? then everything is unstable. > I hope I'm not leading you off on a tangent. Yes, the make world does, > fall apart at various stages, if I attempt it on a "kernel" built with > the new supped sources ( Say 1 week old ). However, I've typically, > completed the "make world" process by simply "make" once the process fails. > This again does fail in various stages ( with sig 11s and core-dumps ). > But ensuring the process completes, ( by repeated "make" as opposed to > "make world")., and a reboot, seemed to cure my problem. ( I rebuilt the > kernel too later, just for completeness). > >I am now trying a 29 august kernel, then I will try a july kernel. > I am confident that Aug 29th kernel, would have no problems with > make world. The reason is that an old kernel of mine ( built Sep 1st ), also > has no problems. > > > >Later, > > > >Jim Leppek > > > >> From owner-freebsd-current@freefall.freebsd.org Sun Sep 10 06:27:23 1995 > >> From: John Hay <jhay@mikom.csir.co.za> > >> Subject: Re: Sig 11 and -current problems. > >> To: davidg@Root.COM > >> Date: Sun, 10 Sep 1995 12:19:46 +0200 (SAT) > >> Cc: FreeBSD-current@FreeBSD.ORG > >> X-Mailer: ELM [version 2.4 PL23] > >> Mime-Version: 1.0 > >> Content-Type> : > text/plain> ; > charset=US-ASCII> > >> Content-Transfer-Encoding: 7bit > >> Sender: current-owner@FreeBSD.ORG > >> > >> > > >> > >"Make world" again. Led to similar sig 11s of "make" mostly. > >> > >vii) Abandon make world, reboot with kernel built from Aug sources; > >> > >viii) Rebuild with "Make world". No sig 11s. Make completes with no > >> > >problems all the way and install the binaries, libs, lkms etc.. > >> > >ix) Rebuild a Kernel now, and reboot. All seems to be fine. > >> > > > >> > >So in a jist, I guess, the order of the making a "Kernel" with supped sources, > >> > >as opposed to "make world install" , followed by the kernel rebuild, seems to > >> > >result in a working machine once again. > >> > > >> > Hmmm, that's interesting. John told me earlier that he was having trouble > >> > with some binaries if they were built with -O2...I don't know why we didn't > >> > notice these earlier (-O2 is known broken in gcc 2.6.3). I don't know if this > >> > is at all related. > >> > Thanks for the info. > >> > > >> > -DG > >> > > >> Well I haven't got rid of al the sig 11s yet. :( The only way at the moment > >> for me is to run with a kernel that I compiled on 3 Sept. > >> > >> At first I had lots of sig 11s. Then I recompiled all the libraries, installed > >> them and recompiled and installed all the programs that sig 11ed. That left > >> me with just a few - rm, sed and login that still sig 11 when I reboot with > >> a new kernel. I then did a make world which I had to help a little because > >> ncrcontrol didn't compile. When I rebooted with a new kernel, login still > >> coredumped with a sig 11. I then recompiled login static and now it works. > >> Then when I started X, xterm coredumped with a sig 11. > >> > >> Well I don't have the diskspace to recompile X, so now I am using the old > >> kernel again. It seems to be working fine with the new libs and programs. > >> (Maybe we have forward compatability and not backward compatability? :-)) > >> > >> Something that I don't understand is why userlevel programs like sed, login > >> xterm etc. are so sensitive to the vm changes. I thought that they shouldn't > >> even be aware of the changes. They don't seem to care if being run on an old > >> kernel. > >> > >> This is on -current (getting it withctm) not -stable. > >> > >> -- > >> John Hay -- John.Hay@csir.co.za > >> > > >
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?9509101524.AA01524>