Re: PCMCIA-related IDE problems on GA-7ZX motherboard

Re: PCMCIA-related IDE problems on GA-7ZX motherboard

[Charles Briscoe-Smith]

On Sat, Mar 02, 2002 at 08:26:45AM +0100, Ulrich Hahn wrote:
> Help!
> 
> You wrote in a linux mailing list about one year ago, describing mainly
> the same probelm I have when loading pcmcia-modules into a 2.4.14
> kernel: The ide0 reports "lost interrupt" and rien ne va plus.
> 
> >I'm having problems related to using a pcmcia bridge on a desktop PC.
> >The machine used to contain a TMC TI5-VG+ motherboard with a 400Mhz K6-II.
> >With the TMC motherboard, everything worked worked fine.  When I upgraded
> >the machine to a Gigabyte GA-7ZX with 800Mhz Athlon, the CD-writer and
> >LS-120 floppy on the secondary IDE channel stopped working.  By fiddling
> >around with the kernel configuration, I've finally narrowed it down to
> >the PC card drivers; the machine contains a Chase-AT "Duo" ISA-to-PCMCIA
> >bridge.
> 
> I read lost of questions like this - but mainly NO answer to it at all.
> Does it happen only to few people? Is it not relevant?
> 
> Personal question: did you find a solution?

Yes, I did, and it has since been documented in the PCMCIA HOWTO,
section 2.3, subsection "Card readers for desktop systems":

  For Chase CardPORT and Altec ISA card readers using the Cirrus PD6722
  ISA-to-PCMCIA bridge, the i82365 driver should be loaded with a
  ``has_ring=0'' parameter to prevent irq 15 conflicts.

I had been trying the option "has_ring=1", which I didn't know was
the default.

[ CC'ed linux-kernel so that this gets into its archives.  I am not on
linux-kernel so, if replying, please CC me (and, I presume, Ulrich). ]

-- 
Charles Briscoe-Smith             Hacking Free Software for fun and profit
Governing Law:
   This License Agreement shall be construed and governed in accordance
   with the laws of the State of Inebriation. 
                                    -- http://www.thalia.org/computer.html

Re: PCMCIA-related IDE problems - "hda: lost interrupt"

[Ulrich Hahn]

Hi there -

I was lucky today finding another solution that helped me out:

It seems the 2.4 kernel series produce the problem of messing with PCI
interrupts.
I had no problem with one and the same pcmcia_cs package with a 2.2.9 kernel
which locked up ALL of the 2.4 kernels I tried meanwile (I am on 2.4.18 now
using the kernel-owned pcmcia modules, yenta_socket, which does not seem to know
any parameters at all).

Today I found a hint on giving the kernel an irqmask on bootup in the lilo.conf
file:
append="pci=irqmask=0xafff"

This prevents the vital IRQs of ide or mouse from being taken by the PCI bridge
when the yenta_socket or i82365 module is loaded. (Unloading the module again
would  not give back interrupt control - so a reboot was the final step)

Charles Briscoe-Smith wrote:

> > Personal question: did you find a solution?
>
> Yes, I did, and it has since been documented in the PCMCIA HOWTO,
> section 2.3, subsection "Card readers for desktop systems":
>
>   For Chase CardPORT and Altec ISA card readers using the Cirrus PD6722
>   ISA-to-PCMCIA bridge, the i82365 driver should be loaded with a
>   ``has_ring=0'' parameter to prevent irq 15 conflicts.
>
> I had been trying the option "has_ring=1", which I didn't know was
> the default.
>
> [ CC'ed linux-kernel so that this gets into its archives.  I am not on
> linux-kernel so, if replying, please CC me (and, I presume, Ulrich). ]

Thanks for your hint!