Re: ISA IRQ mapping and problem on PCI-PCMCIA Bridge
Christoph Plattner <[email protected]> Thu, 14 Mar 2002 16:08:04 +0100
| Newsgroups | gmane.linux.pcmcia.devel |
|---|---|
| Organization | AAA |
| Message-ID | <[email protected]> |
Hello !
Ok, I used the work around using
PCIC_OPTS="irq_mode=0 pci_csc=0"
and the IDE (ide2) interface works, but when I try to remove the compact
flash, the system prints out (endless and with other interrupts locked):
ide2: unexpected interrupt, status=0xff, count=NNNNNN
(with increasing NNNN as loop counter).
I have to say, that I have not used the "cardctl eject" !!!
Is this really necessary, or is this a BUG ?
Again: linux-2.2.20, pcmcia-cs-3.1.33
Using CompactFlash (SAN DISK 32MB) und CF adapter
With friendly regards
Christoph Plattner
dhinds wrote:
>
> On Wed, Mar 13, 2002 at 04:21:22PM +0100, Christoph Plattner wrote:
> >
> > First problem: The driver seems not to find any ISA IRQs.
> > Why ? (In this setup I have "reserved" IRQ5 for ISA and
> > kept IRQ 9 free for PCI ("available")).
>
> The driver finds no ISA interrupts because this is a PCI adapter card:
> it has no connection to the ISA bus so it can only use PCI interrupts.
> This is as it should be; the adapter should use its PCI interrupt for
> everything.
>
> > In general: How should the ISA IRQ mapping work ?
>
> ISA interrupt delivery only works on laptops, where the PCMCIA bridge
> chip is on the system board, and is hooked up both to the PCI bus and
> to the ISA interrupt controller.
>
> > Does a PCI PCMCIA-Controller access 3 of them, one for the
> > controller itself, the other for the inserted devices (one
> > or two) ?
>
> In your case, INTA is used by the bridge itself, and is also used for
> the PCMCIA card.
>
> > When I insert the CF-Adapter, I see following log output (dmesg):
> >
> > hde: SanDisk SDCFB-32, ATA DISK drive
> > ide2: Disabled unable to get IRQ 9.
>
> This is an age-old bug in the IDE driver that I've been griping about
> to the IDE maintainer for YEARS. The driver does not understand that
> it needs to share its interrupt in this configuration. The problem is
> documented in the BUGS file for the pcmcia-cs package, along with a
> workaround.
>
> -- Dave
--
+--------V--------+ [email protected]
| A L C A T E L | -----------------------------
+-----------------+ Phone: +43 1 27722 3706
T A S Fax: +43 1 27722 3955
_______________________________________________
Pcmcia-cs-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/pcmcia-cs-devel