strange strange 2.6.2 bug - pcmcia sleeps until I move the mouse ?!

[Daniel Ritz]

hi

On Monday 16 February 2004 10:58, Guylhem Aznar wrote:
...
> Some information: Using the latest init.d/pcmcia on 2.4+i82365 , I do have the
> same bug I get with kernel 2.6
> I've been trying various stuff (boot time options, with/without bios "Os support
> PnP" option, etc) because I just couldn't make pcmcia in 2.4 work as it did.
> Looks like an update of the pcmcia-cs tools, most specifically the boot script,
> caused the problem.
> 
> Kernel messages with new (broken) pcmcia script:
> Linux PCMCIA Card Services 3.2.5
>   kernel build: 2.4.21-guylhem #6 Mon Sep 29 20:34:41 CEST 2003
>   options:  [pci] [cardbus] [apm] [pnp]
> PnP: PNP BIOS installation structure at 0xc00f6c00
> PnP: PNP BIOS version 1.0, entry at f0000:5b64, dseg at f0000
> Intel ISA/PCI/CardBus PCIC probe:
>   TI 1410 rev 01 PCI-to-CardBus at slot 00:05, mem 0x1e000000
>     host opts [0]: [serial pci & irq] [pci irq 5] [lat 168/176] [bus 2/5]
>     PCI irq 5 test failed
>     ISA irqs (scanned) = none!<6>    *NO* card interrupts, polling interval = 10
> 00 ms
> cs: memory probe 0xa0000000-0xa0ffffff: clean.
> orinoco.c 0.13e (David Gibson <hermes at gibson.dropbear.id.au> and others)
> orinoco_cs.c 0.13e (David Gibson <hermes at gibson.dropbear.id.au> and others)
> cs: IO port probe 0x0100-0x04ff: excluding 0x3c0-0x3e7
> cs: IO port probe 0x0800-0x08ff: clean.
> cs: IO port probe 0x0a00-0x0aff: clean.
> cs: IO port probe 0x0c00-0x0cff: clean.
> orinoco_cs: RequestIRQ: Resource in use
> 
> With old script:
> 
> Linux PCMCIA Card Services 3.2.5
>   kernel build: 2.4.21-guylhem #2 Tue Sep 9 15:02:11 CEST 2003
>   options:  [pci] [cardbus] [apm]
> Intel ISA/PCI/CardBus PCIC probe:
>   TI 1410 rev 01 PCI-to-CardBus at slot 00:05, mem 0x1e000000
>     host opts [0]: [pci only] [pci irq 5] [lat 168/176] [bus 2/5]
>     PCI card interrupts, PCI status changes
> cs: memory probe 0xa0000000-0xa0ffffff: clean.
> orinoco.c 0.13e (David Gibson <hermes at gibson.dropbear.id.au> and others)
> orinoco_cs.c 0.13e (David Gibson <hermes at gibson.dropbear.id.au> and others)
> cs: IO port probe 0x0100-0x04ff: excluding 0x180-0x187 0x290-0x297 0x3c0-0x3df 0
> x480-0x48f 0x4d0-0x4d7
> cs: IO port probe 0x0800-0x08ff: clean.
> cs: IO port probe 0x0a00-0x0aff: clean.
> cs: IO port probe 0x0c00-0x0cff: clean.
> eth1: Station identity 001f:0002:0002:0001
> eth1: Looks like a Symbol firmware version [V2.50-13] (parsing to 25013)
> eth1: Ad-hoc demo mode supported
> eth1: IEEE standard IBSS ad-hoc mode supported
> eth1: WEP supported, 104-bit key
> eth1: MAC address 00:02:B3:06:00:B8
> eth1: Station name "Prism  I"
> eth1: firmware ALLOC bug detected (old Symbol firmware?). Trying to work
> around... ok.
> eth1: ready
> eth1: index 0x01: Vcc 5.0, irq 5, io 0x0100-0x0147
> eth1: New link status: Connected (0001)
> 

ok, the main difference here:
new:
       host opts [0]: [serial pci & irq] [pci irq 5] [lat 168/176] [bus 2/5]
but with the old script:
       host opts [0]: [pci only] [pci irq 5] [lat 168/176] [bus 2/5]

so it messes around with some options. the main difference between old
and new script:

the old one uses /etc/default/pcmcia, the new one uses /etc/pcmcia.conf
for the options like PCIC, PCIC_OPTS, ...
so the difference should be there...
try 
  cp /etc/pcmcia.conf /etc/pcmcia.conf.old
  cp /etc/default/pcmcia /etc/pcmcia.conf

or even rename those file so they are not found by the init script


> > you have two probles here: for one, there are no ISA interrupts available...
> 
> Looks like the card is very happy with a PCI irq. I'd like to enable ISA irq for
> modems and other pcmcia cards however.
> 
> > second, you have a TI cardbus bridge that forgets to route the interrupts
> > from the card to the PCI bus...so your card only works when there are
> > interrupts on the same line as the interrupt handler of the card is
> > registered...
> 
> There should be *many* interrupts available. Uncommending the "excludes" in
> config.opts didn't help.
> 
> 2.6 interrupts
>            CPU0       
>   0:     932315          XT-PIC  timer
>   2:          0          XT-PIC  cascade
>   5:          0          XT-PIC  acpi
>   8:          4          XT-PIC  rtc
>   9:         47          XT-PIC  eth0
>  10:          0          XT-PIC  Trident Audio
>  11:      22866          XT-PIC  yenta, ohci_hcd, ohci_hcd, orinoco_cs
>  14:       7407          XT-PIC  ide0
>  15:         36          XT-PIC  ide1
> NMI:          0 
> LOC:     932216 
> ERR:          0
> MIS:          0
> 
> I guess it leaves 1 2 3 4 6 7 12 13

yenta probes for 3,4,5,6,7,9,10,11

> 
> 2.4 interrupts
>            CPU0       
>   0:     454008          XT-PIC  timer
>   1:          4          XT-PIC  keyboard
>   2:          0          XT-PIC  cascade
>   5:       1387          XT-PIC  acpi, i82365, orinoco_cs
>   8:          4          XT-PIC  rtc
>   9:         59          XT-PIC  eth0
>  10:        159          XT-PIC  Trident Audio
>  11:      18196          XT-PIC  usb-ohci, usb-ohci
>  14:      28940          XT-PIC  ide0
>  15:          2          XT-PIC  ide1
> NMI:          0 
> ERR:          0
> 
> I've no parallel port (disabled in the bios) and only 1 serial port which I can
> also disable. There are only 2 cards in the machine : a pcmcia-pci and a
> winmodem-pci (worthless but close a hole - I've lost the slot covert :-) The
> machine is a so-called "legacy free" so I can also remove PS/2 support  (I've a
> USB keyboard and mouse)
> 
> I'm now wondering why it can't use these free IRQ. On the PCMCIA-HOWTO I've read
> /proc/interrupts didn't help, but I guess at least IRQ 7 (disabled in the bios!)
> should be available !
> 

yenta probes the interrupts but none of them is working. this is because the
cardbus bridge doesn't know how to deliver the interrupts. why? it has some
registers wrongly programmed. the notebook manufacter screwed it up....

> Best regards,
> Guylhem
> 

rgds
-daniel