[Clfs-dev] Weirdness.

[Ken Moffat]

On Sat, Dec 29, 2007 at 12:11:23AM +0000, Ken Moffat wrote:
> 
>  I've now built in ehci_hcd, ohci_hcd (both modules were loaded on
> the earlier system), usbkbd (actually, CONFIG_USB_KBD no longer
> exists as an option, so I just turned everything on), hid, and
> evdev.  So, it looks as if udev-118
> 
 Heh, I'm getting good at not finishing sentences ;)

 I have no view on the multitudinous failures in the gcc test suite,
but everything else is now tickety-boo.  Udev-118 is good.

 The problem is module-init-tools-3.4.  I'd noticed that it failed
to run the tests (needs the DOCBOOK2MAN override, which coincidentally
makes the manpages unusable as in 'empty'), but there is a bigger
problem with that.  Looking at my list of modules from a successful
boot, I noticed nfs.  Still thinking udev might be the problem, I
checked lsmod - no modules loaded - then tried

# modprobe nfs
MODTEST_OVERRIDE used but MODTEST_UNAME not set

 A look at the source shows those are for the testsuite.  A quick
build without running the tests showed ./modprobe worked.  So,
untar it again, list files, run the tests, list files again, make
distclean, list again, diff --side-by-side (for some reason, regular
invocation of diff wants me to believe all entries were deleted and
all the current entries are new) : all the objects, the binaries,
the empty manpages were still present.  Ran it through, proved this
was indeed the cause of hte error.

 I'll fix that in the book shortly.

ĸen
-- 
das eine Mal als Tragödie, das andere Mal als Farce