LFS 6.2. Ch6.9 Glibc-2.3.6 error during make check (system halted)

Dan Nicholson dbn.lists at gmail.com
Sat Aug 19 07:35:34 PDT 2006


On 8/18/06, Danny <linux at danchi.nl> wrote:
>
> I'm actually building the system straight from a command line on an
> 'empty' machine. I fully understand that I'm building on an older
> machine, but didn't have this problem when building 6.1.1

What host are you using? You say this is an empty machine, so does
that mean you're using the LiveCD? This is extremely important when
you're running the Glibc test suite. It's very much tied to the
running system. What version of LInux are you running?

> First three errors are the same as with my 6.1.1 build and can be ignored:
>
> make[2]: *** [/sources/glibc-build/posix/bug-regex24.out] Error 1
> make[2]:          "           "      "  /tst-regex2.out] error 1
> make[2]:          "           "      "  /Error 1 (ignored)
>
> then it starts with:
>
> make[2]: *** [/sources/glibc-build/nptl/tst-mutex8.out] Error 1
> ...
> three more pages of errors in this directory nptl and it ends with:
> ...
> make[2]: *** [/sources/glibc-build/nptl/tst-fini1.out] Error 134
> make[2]: *** [/  "        "         "  /tst-execstack.out] Error 1
> make1]: [nptl/tests] Error 2

If there's three pages of NPTL errors, then that's a problem.

> make[2]: *** [/sources/glibc-build/rt/tst-mqueue8.out] Error 1
> make[2]: *** [/sources/glibc-build/rt/tst-mqueue8x.out] Error 127
> make[1]: [rt/tests] error 2
> make: *** [check] Error 2

Looking back at your first mail, you said the system was "PII, 350MHz,
96Mb". The processor seems OK. At least you shouldn't be running into
problems with i[3456]86. 96 Mb is tough. If you already built
LFS-6.1.1, you probably know that you're gonna have to open up a lot
of swap space.

That many errors in Glibc would usually signal to me that something
has gone wrong. But maybe you can run the tests again to see if you
get similar results. I would suggest shutting down any non-essential
services while the tests are running. I've noticed on both machines I
build on that the tests are much more likely to pass on an idle box.
So, go get a cup of coffee (or in your case, a pot of coffee) and let
the Glibc tests have the run of your machine. Resist the temptation to
do things that spike your processor.

If you run them again and NPTL is still hemorraghing badly, then you
might want to start over or dig further into the specific failures.

--
Dan



More information about the lfs-support mailing list