[LinuxBIOS] pci_level_irq: lower order bits are wrong: want 0x0, got 0x20

[Peter Stuge]

Since the first boot with LB I've always seen this in the output:

--8<--
pci_routing_fixup: dev is 000104c0
setting firewire
setting usb
Assigning IRQ 5 to 0:10.0
  Readback = 5
pci_level_irq: lower order bits are wrong: want 0x0, got 0x20
Assigning IRQ 9 to 0:10.1
  Readback = 9
pci_level_irq: lower order bits are wrong: want 0x0, got 0x20
Assigning IRQ 9 to 0:10.2
  Readback = 9
pci_level_irq: lower order bits are wrong: want 0x0, got 0x20
Assigning IRQ 5 to 0:10.3
  Readback = 5
pci_level_irq: lower order bits are wrong: want 0x0, got 0x20
setting vt8235
Assigning IRQ 5 to 0:11.1
  Readback = 5
pci_level_irq: lower order bits are wrong: want 0x0, got 0x20
Assigning IRQ 9 to 0:11.5
  Readback = 9
pci_level_irq: lower order bits are wrong: want 0x0, got 0x20
setting ethernet
Assigning IRQ 5 to 0:12.0
  Readback = 5
pci_level_irq: lower order bits are wrong: want 0x0, got 0x20
setting vga
Assigning IRQ 5 to 1:0.0
  Readback = 5
pci_level_irq: lower order bits are wrong: want 0x0, got 0x20
setting pci slot
setting cardbus slot
Assigning IRQ 5 to 0:a.0
  Readback = 5
pci_level_irq: lower order bits are wrong: want 0x0, got 0x20
Assigning IRQ 9 to 0:a.1
  Readback = 9
pci_level_irq: lower order bits are wrong: want 0x0, got 0x20
setting riser slot
PNP: 002e.0 init
PNP: 002e.1 init
PNP: 002e.2 init
PNP: 002e.3 init
PNP: 002e.b init
PCI: 00:11.1 init
Enabling VIA IDE.
ide_init: enabling compatibility IDE addresses
enables in reg 0x42 0x0
enables in reg 0x42 read back as 0x0
-->8--


I'm having a few problems with the CF card, but they are probably
related to pcmcia things in the kernel. I'm about to talk to
linux-pcmcia but wanted some input on those 0x20 messages first..

..because if I force Linux to look for the CF card (0:a.1) with
ide2=0x1e0,0x1ec,9 I get:

--8<--
hde: probing with STATUS(0x50) instead of ALTSTATUS(0x00)
hde: SAnDiSk, CFA DISK drive
ide4: I/O resource 0x1E0-0x1E7 not free.
ide4: ports already in use, skipping probe
ide2 at 0x1e0-0x1e7,0x1ec on irq 9
hde: max request size: 128KiB
hde: 985056 sectors (504 MB) w/1KiB Cache, CHS=1986/16/31
 hde:<4>hde: lost interrupt
hde: lost interrupt
hde: lost interrupt
-->8--

Note SAnDiSk vs. SanDisk - bit 5 (0x20) is jammed high on a few
bytes.. (plus this is a brand new Extreme III 1GB CF, not 504MB)

--8<--
FILO version 0.5 (stuge at carepad4) Sat Sep 16 18:00:42 CEST 2006                 
boot: hde1:/bzImage root=/dev/hde1 ro console=tty0 console=ttyS0,115200
hde: LBA 1025MB: SAnDiSk
Unknown filesystem type
boot: hde1:/bzImage root=/dev/hde1 ro console=tty0 console=ttyS0,115200
-->8--

At one point I did get "SanDisk" out of FILO (the CF that I thought I
busted turned out to work after all) but I didn't think much of it
until now and so don't recall the exact circumstances. I recall
needing to unplug the power cable and let the board sit for a while
and then only on the first boot would I get "SanDisk" properly.
I need to experiment more with this. (This was also before Nick's
recent changes to the rl5c476 IRQ fixes.)


Speaking about the 5c476, I've also gotten:
cs: pcmcia_socket0: unable to apply power

(as seen on
http://www.kernel.org/pub/linux/utils/kernel/pcmcia/powerbugs.html)

when inserting cards. This is on 2.6.18rc7, I've been experimenting
with 2.6.17.6 and 2.6.18rc6 as well. Some combination has worked.


I'm set for a day of testing to make a matrix of what works and what
doesn't. But if anyone can provide insight into the mysterious 0x20
bit I'd appreciate it. Best case it will even narrow my search.


//Peter