I have a second parport installed as a PCI-card. In earlier Linux-versions
this would lock the machine completely if parport & Co where compiled as
modules (2.2.16 and 2.2.17). Compiled into the kernel however, everything
worked fine. I wrote about that to LK, but no solution was found. Now in
2.2.18 it is the other way round, modules work with the proper
initialization in modules.conf, but if compiled into the kernel, the
second parport vanishes completely.
lspci -v gives:
00:0d.0 Parallel controller: Timedia Technology Co Ltd: Unknown device
7268 (rev 01) (prog-if 02 [ECP])
Subsystem: Timedia Technology Co Ltd: Unknown device 0103
Flags: stepping, medium devsel, IRQ 9
I/O ports at 9000
I/O ports at 8800
/proc/parport however only knows about parport0
As I prefer the parport being modularized, it?s no problem for me. But
IMHO the other way it should work, too.
Cheers
Peter B
On Wed, Dec 13, 2000 at 09:01:56PM +0100, Peter Bornemann wrote:
> I have a second parport installed as a PCI-card. In earlier Linux-versions
> this would lock the machine completely if parport & Co where compiled as
> modules (2.2.16 and 2.2.17). Compiled into the kernel however, everything
> worked fine. I wrote about that to LK, but no solution was found. Now in
> 2.2.18 it is the other way round, modules work with the proper
> initialization in modules.conf, but if compiled into the kernel, the
> second parport vanishes completely.
The problem was that there are many more Timedia cards than I
realised, and they are only distinguishable from each other by their
subdevice ID (vendor and device isn't enough). So we were probably
using the wrong BARs, which probably caused the freezes you saw.
The 2.4.x parport_pc card table can deal with this; the 2.2.x
parport_pc card table can't. So rather than leave a potential freeze
in I pulled those cards from the table altogether. Sorry if I didn't
tell you this.
For 2.2.19 I'm hoping to backport the 2.4.x parport_pc card table
structure if Alan doesn't think that's too big a change.
The reason it works for you with modules is probably because you have
an options line in /etc/modules.conf that tells parport_pc which
addresses your PCI card is using at the moment.
Tim.
*/
On Wed, 13 Dec 2000, Tim Waugh wrote:
> The reason it works for you with modules is probably because you have
> an options line in /etc/modules.conf that tells parport_pc which
> addresses your PCI card is using at the moment.
Bad news: the old problem with lockups during insmoding parport is back! I
recompiled 2.2.18, because there is a problem with the 8138too-driver vs.
modutils-2.3.22. Since then, the old problem was back! Unfortunately I
did not save the old modules. So, for the moment I cannot use 2.2.18 at
all and I have switched to the 2.4 series and modutils-2.3.14 (version of
modprobe doesn?t matter for the lockups, however).
Any hint is welcome, for I would prefer a really stable kernel for this
machine.
Peter B
On Thu, Dec 14, 2000 at 08:43:18PM +0100, Peter Bornemann wrote:
> Any hint is welcome, for I would prefer a really stable kernel for this
> machine.
The problem isn't that the kernel is not stable, but that it doesn't
support your parallel port card. ;-)
I'll look at backporting the 2.4.x card table and keep you posted.
Tim.
*/