On Fri, 2009-12-18 at 15:20 +0100, Matthias Schniedermeyer wrote:
> On 18.12.2009 03:57, Ben Hutchings wrote:
> > On Thu, 2009-12-17 at 19:51 -0800, Steve Langasek wrote:
> > > On Fri, Dec 18, 2009 at 12:24:01AM -0000, Ben Hutchings wrote:
> > > > I substituted a much cleaner test in Debian (see the linked bug report)
> > > > and suggest you take that change.
> > >
> > > As near as I can tell, the Debian fix is incorrect because
> > > /sys/module/nfsd
> > > is only available if nfsd is built as a module. Anyone who has the code
> > > built into their kernel will not have this directory.
> >
> > Anything that *can* be built as a module still appears in /sys/modules
> > even if built-in.
>
> Then nfsd must be the exception that proves the rule, because that
> change just killed my nfs-kernel-server ("Not starting NFS kernel
> daemon: no support in current kernel.").
> I'm running 2.6.32 with nfsd compiled in and there is no
> "/sys/module/nfsd"
>
> Killing the check make nfsd work again.You're right. /sys/module/<name> is created for built-in code which has "module" parameters, but nfsd doesn't have any. Let's check for /proc/fs/nfs/exports instead. That will always be created on successful initialisation of nfsd. Ben. -- Ben Hutchings Humans are not rational beings; they are rationalising beings.
signature.asc
Description: This is a digitally signed message part
