> > From: Chris Staub <[email protected]> > To: LFS Developers Mailinglist <[email protected]> > Date: Sun, 13 Jun 2010 12:02:45 -0400 > Subject: Re: Inconsistency in glibc requirements between Chapters 5 and 6 > On 06/13/2010 10:36 AM, Guillem Pagčs Gassull wrote: > >> -----BEGIN PGP SIGNED MESSAGE----- >> Hash: SHA1 >> >> Hi, >> >> I was building a new LFS (using jhalfs; SVN version), using a (very old) >> LFS 6.3 as a host. I know, I should have read the host requirements list >> first; 6.3 just contains too many unsupported old versions. That is not >> the point. ;-) >> >> I was able to build through Chapter 5 without any problems, but when >> building glibc in Chapter 6, configure failed because the required >> kernel version is 2.6.22.5, and I was using an older kernel. >> In the temporary tools in chapter 5, the required version is 2.6.18, >> which did not cause me any problems. >> >> After letting the system run overnight (I am using a slow 1GHz VIA C7 >> computer), I woke up next morning with the mentioned error. >> Wouldn't it make sense to have the same 2.6.22.5 requirement for glibc >> in both chapter 5 and 6? Or is there a reason why chapter 5 admits an >> older kernel? >> >> It is an inconsistency and should be fixed, but I don't see how it could > result in any kind of error, as the Host System Requirements *do* list > 2.6.22.5 as the minimum kernel version. Also, LFS 6.3 should work fine, as > package versions for the Host Sys. Reqs. are in fact taken directly *from* > 6.3. If your host system has a kernel older than 2.6.22, it must not > actually be an LFS 6.3 system (or you deviated when building 6.3 and used an > older kernel). >
I might have been wrong about the 6.3 LFS; might have been older. I do have many versions installed on that system and the "lfs-release" file on etc might have been corrupted or i just looked at the wrong one. In any case I was using a kernel older than 2.6.22, and compiled with the old gcc 3.4.3. I already stated, that I did not read the sys reqs, so I admit it was my fault anyway; but fixing the inconsistency will save other people like me, who do not RTFM until it is too late, lots of time ;-) On the other hand, that is the way to teach us to actually read it :-D As an aside note, using a kernel newer than 2.6.22.5, but compiled with the old GCC 3.4.3 does not complain in the glibc configure, but it also does not build; it brings a link error. I was so upset about the time waste that I just deleted the partition and started again, so I don't have the log anymore :-( But if I recall correctly, it was an (actually about 5 or 6) undefined reference; in nss or nscd. Using a kernel compiled with a newer GCC, as the book explicitly says fixes the problem. But I somehow find it strange, that the error only occured in chapter 6 and not in chapter 5. Unless it is related to the inconsistency specified below. Using the right kernel and compiler, everything worked up to now; chapter 6 glibc (finally) build. I just wanted to say that in case anyone is interested / finds a similar problem. Let's see when my chapter 6 GCC finishes; 44 SBUs with an SBU over 11 Minutes is a looooooong time ;-) Guillem -- Nothing is ever a complete failure; it can always be used as a bad example
-- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
