[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#209301: modconf: on module installation failure, suggest isa-pnp.o to the user


Thank you for the nicer reply:

On Mon, Sep 15, 2003 at 10:20:09AM +0200, Eduard Bloch wrote:
> |After a module fails initialization, the suggestion should be made that
> |the user install the isa-pnp module if the hardware in question is an
> |ISA-PnP card.
> |
> |"The module failed initialization.  Check the hardware resources (I/O,
> |IRQ) and if the card is a ISA PnP card, ensure that you have installed
> |the isa-pnp module first."
> And I have just told you that loading isa-pnp is not enough.

You're right, the process is not as simple as I made it look.  I wrote
that in the early morning after a long tiresome installation for a user.

> The only practicable think I can imagine is extending this message to
> a micro-Howto about how to use isapnptools or the isa-pnp's proc
> interface. Both things that user can learn from RTFM before going to
> install.

Yeah, but how often does the user really RTFM?  Unfortunately many users
rely on previous experience and intuition to guide them; if we really
want the installer to be useful to the most people, we can give the
hints to help the non-RTFM people along as long as we don't get in the
way of the people who do RTFM.

> Another idea may be silent loading of isa-pnp when available and
> including the list of the devices in the message, suggesting how to
> continue.

This seems logical; at least, is there any case in which you would
_not_ want isa-pnp loaded on i386?

> > My contention is that without isa-pnp loaded, the user could not use
> > modules that do not support ISA-PnP configuration themselves when the
> > card is not in a configured state.  In the case of a new user, they see
> isa-pnp does not configure them when the module is just loaded.

Right.  It is a starting point, from there the user can google and find
the kernel documentation for it.

> > only a failure when inserting the module for their hardware which is
> > clearly installed in the system, and have no idea where to proceed.
> The general hint is _okay_ but the advisory to load isa-pnp is not okay
> in the suggested form, and will confuse the user more to the end.

Yes, I agree, sorry for confusion:

> > > Just for the case that you missed it, we do already include
> > > isapnptools for pre-2.4 kernels. And pnpdump&isapnp are a way more
> > > user-friendly and better documented than the proc interface.
> > 
> > The user was installing from bf2.4.  Is isapnptools included in that
> > case?  If so, then this "wishlist" bug is in error and you may ignore
> It is not included. For the simple reason, I could not find a driver for
> any _essential (for installation)_ ISA PnP card that is not supported by
> isapnp manegement in 2.4.18.

eepro is the card that caused the problem.  There must be others, eepro
is not exactly an uncommon card to find.  Examining the eepro source,
I find no code that configures the PNP component.  Please correct me if

> > it.  If it is not, then I stand by my statement that there must be a
> > documented way for the user to configure the ISA-PnP cards, or there is
> > a good chance of significant confusion occurring during an installation.
> I am not the only modconf maintainer but I won't do much more on that
> because of the reasons described above and because I think that there is
> enough documentation around. But patches are welcome, there are surely
> ways to improve the failure message.

Fine, I'll look further into it before bothering the maintainers about it


Ryan Underwood, <nemesis at icequake.net>, icq=10317253

Reply to: