On Thu, Mar 01, 2007 at 06:54:47PM -0800, Thomas Tutone wrote: > > I get errors running the glibc-2.3.6 testsuite in > section 6.9 of Version 6.2 of the LFS book: > [snip a series of nptl failures, version-check showing a recent kernel, and relevant archive postings ] > > So, this suggests that my host kernel is too recent. > It also suggests that I shouldn't ignore these > errors and move on.
I disagree about "I shouldn't ignore". The nptl testsuite can be very affected by the running kernel. But, as a desktop user, I've not noticed any problems in the resulting system, so I think that you *can* ignore the errors you posted. OK, I haven't built 2.3.6 for a few months, but I've done a fair amount of building against glibc-2.4 and 2.5 with current and prerelease kernels, and seen various nptl failures. > A couple questions: > > 1. Any suggestions on how to proceed short of > building a 2.6.16.27 kernel for the host? Don't do that. Please. Adrian Bunk is maintaining 2.6.16 at the moment, if you are going to stay with 2.6.16 then please use a current version - most of what goes in is security fixes. > > 2. If I do take that route (building a 2.6.16.27 > kernel in order to get past these errors), do > I need to go back and rebuild the entire temporary > system while booted up with the 2.6.16.27 kernel? If you *are* going with a recent 2.6.16, once you have booted the kernel just start from glibc in chroot - nothing should have been *built* differently (you are using the same headers as the book, so the differences in the nptl tests are a runtime variation). ĸen -- das eine Mal als Tragödie, das andere Mal als Farce -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page