svn commit: r346441 - in head/sys/modules: em fusefs iavf

[Warner Losh]

On Sat, Apr 20, 2019 at 2:03 PM John Baldwin <jhb at freebsd.org> wrote:

> On 4/20/19 10:21 AM, Warner Losh wrote:
> > On Sat, Apr 20, 2019, 10:44 AM John Baldwin <jhb at freebsd.org> wrote:
> >
> >> On 4/20/19 6:23 AM, Justin Hibbits wrote:
> >>> On Sat, Apr 20, 2019, 08:21 Alan Somers <asomers at freebsd.org> wrote:
> >>>
> >>>> On Sat, Apr 20, 2019 at 6:58 AM Justin Hibbits <chmeeedalf at gmail.com>
> >>>> wrote:
> >>>>>
> >>>>>
> >>>>>
> >>>>> On Sat, Apr 20, 2019, 07:51 Alan Somers <asomers at freebsd.org> wrote:
> >>>>>>
> >>>>>> Author: asomers
> >>>>>> Date: Sat Apr 20 12:51:05 2019
> >>>>>> New Revision: 346441
> >>>>>> URL: https://svnweb.freebsd.org/changeset/base/346441
> >>>>>>
> >>>>>> Log:
> >>>>>>   Use symlinks for kernel modules rather than hardlinks
> >>>>>>
> >>>>>>   When aliasing a kernel module to a different name (ie if_igb for
> >>>> if_em),
> >>>>>>   it's better to use symlinks than hard links. kldxref will omit
> >>>> entries for
> >>>>>>   the links, ensuring that the loaded module has the correct name.
> >>>>>>
> >>>>>>
> >>>>>
> >>>>>
> >>>>> Thanks! This should fix installkernel on my POWER9.
> >>>>>
> >>>>> - Justin
> >>>>
> >>>> What's the problem with your POWER9?  Is that one of those msdosfs
> >>>> /boot systems?  If so, I don't think this will fix it.  msdosfs
> >>>> doesn't support either symlinks or hardlinks.  Or is there some other
> >>>> problem?
> >>>> -Alan
> >>>>
> >>>
> >>> Yes it is.  Well that's a bummer then. I thought we faked symlinks on
> >>> msdosfs, but on second thought not sure how well would do that.
> >>
> >> You could just use cp instead of a link?
> >>
> >
> > I don't think this solves the original issue.
>
> FWIW, I think igb and em still use separate driver_t's with separate
> probe tables, etc. to preserve POLA of existing device names, so in
> theory there could still be two modules, it's just by choice that they
> are both in the same C file and same kld.
>

Yea, I sometimes wish they'd have if_em.ko that's just this table and a
dependency to igb.ko plus another if_igb.ko with the same dependency. That
would be cleaner than the hacky thing we're doing which forces us to have
other hacks in kldload, etc.

Warner