ohci1394 issues ( 2.6.5-rc2 and beyond )

ohci1394 issues ( 2.6.5-rc2 and beyond )

[Michael Baehr]

I'm running a linux 2.6.5-rc2-mm5 kernel atm... motherboard is an Abit 
NF7-s with an onboard firewire controller (I assume Nforce2 chipset).

I get this loveliness everytime I modprobe ohci1394:

ohci1394: fw-host0: OHCI-1394 1.1 (PCI): IRQ=[5]  
MMIO=[e0084000-e00847ff]  Max
Packet=[2048]
Badness in get_phy_reg at drivers/ieee1394/ohci1394.c:238
Call Trace:
  [<f8bb710e>] get_phy_reg+0x10e/0x120 [ohci1394]
  [<f8bb7f3c>] ohci_devctl+0x5c/0x5d0 [ohci1394]
  [<c03f23c0>] common_interrupt+0x18/0x20
  [<c0113d54>] delay_tsc+0x14/0x20
  [<f8bba058>] ohci_irq_handler+0x588/0x830 [ohci1394]
  [<c012643f>] do_timer+0xdf/0xf0
  [<c010a8ca>] handle_IRQ_event+0x3a/0x70
  [<c010ac85>] do_IRQ+0xd5/0x1b0
  =======================
  [<c03f23c0>] common_interrupt+0x18/0x20
  [<f8bb7193>] set_phy_reg+0x73/0x120 [ohci1394]
  [<f8bd5d40>] delayed_reset_bus+0x0/0xd0 [ieee1394]
  [<f8bb7f56>] ohci_devctl+0x76/0x5d0 [ohci1394]
  [<f8bde3b7>] csr1212_fill_cache+0x97/0x100 [ieee1394]
  [<f8bd5d40>] delayed_reset_bus+0x0/0xd0 [ieee1394]
  [<f8bd31a5>] hpsb_reset_bus+0x35/0x40 [ieee1394]
  [<f8bd5df4>] delayed_reset_bus+0xb4/0xd0 [ieee1394]
  [<c0126254>] run_timer_softirq+0xd4/0x1d0
  [<c0121e7d>] __do_softirq+0x7d/0x80
  [<c010b57e>] do_softirq+0x4e/0x60
  =======================
  [<c010ad0b>] do_IRQ+0x15b/0x1b0
  [<c03f23c0>] common_interrupt+0x18/0x20
  [<c0107053>] default_idle+0x23/0x30
  [<c01070bc>] cpu_idle+0x2c/0x40
  [<c04fc897>] start_kernel+0x197/0x1e0
  [<c04fc450>] unknown_bootoption+0x0/0x120

Badness in set_phy_reg at drivers/ieee1394/ohci1394.c:267
Call Trace:
  [<f8bb7230>] set_phy_reg+0x110/0x120 [ohci1394]
  [<f8bb7f56>] ohci_devctl+0x76/0x5d0 [ohci1394]
  [<c0113d54>] delay_tsc+0x14/0x20
  [<f8bba058>] ohci_irq_handler+0x588/0x830 [ohci1394]
  [<c012643f>] do_timer+0xdf/0xf0
  [<c010a8ca>] handle_IRQ_event+0x3a/0x70
  [<c010ac85>] do_IRQ+0xd5/0x1b0
  =======================
  [<c03f23c0>] common_interrupt+0x18/0x20
  [<f8bb7193>] set_phy_reg+0x73/0x120 [ohci1394]
  [<f8bd5d40>] delayed_reset_bus+0x0/0xd0 [ieee1394]
  [<f8bb7f56>] ohci_devctl+0x76/0x5d0 [ohci1394]
  [<f8bde3b7>] csr1212_fill_cache+0x97/0x100 [ieee1394]
  [<f8bd5d40>] delayed_reset_bus+0x0/0xd0 [ieee1394]
  [<f8bd31a5>] hpsb_reset_bus+0x35/0x40 [ieee1394]
  [<f8bd5df4>] delayed_reset_bus+0xb4/0xd0 [ieee1394]
  [<c0126254>] run_timer_softirq+0xd4/0x1d0
  [<c0121e7d>] __do_softirq+0x7d/0x80
  [<c010b57e>] do_softirq+0x4e/0x60
  =======================
  [<c010ad0b>] do_IRQ+0x15b/0x1b0
  [<c03f23c0>] common_interrupt+0x18/0x20
  [<c0107053>] default_idle+0x23/0x30
  [<c01070bc>] cpu_idle+0x2c/0x40
  [<c04fc897>] start_kernel+0x197/0x1e0
  [<c04fc450>] unknown_bootoption+0x0/0x120

ohci1394: fw-host0: SelfID received outside of bus reset sequence
ieee1394: Host added: ID:BUS[0-00:1023]  GUID[000000508dfea614]

Evidently, it still loads although the error about SelfID indicates 
already that something is not quite right.

Further fun continues when plugging in my firewire hard drive (it's a 
Maxtor external 160 gb, pretty generic as far as external firewire 
hard drives go).  About one out of five times I will get a hard drive 
mountable via /dev/sdc... other times, the kernel's SCSI module will 
backtrace and not work, or it will give me a series of errors from 
sbp2 (can't login to device, device timed out, could not read config 
rom, etc).

As far as I can remember, I never had these issues with 2.6.4 kernel, 
so it's something that's been added in the devel tree.

Also, this is my first post on LKML so if anybody feels like 
nitpicking any aspect of it, please do so (preferably without ccing 
the list, as that'll just add noise).  I'm all about netiquette :)


- Mike

Re: ohci1394 issues ( 2.6.5-rc2 and beyond )

[Andrew Morton]

Michael Baehr <tehdely_lkml@metawire.org> wrote:
>
> I'm running a linux 2.6.5-rc2-mm5 kernel atm... motherboard is an Abit 
> NF7-s with an onboard firewire controller (I assume Nforce2 chipset).
> 
> I get this loveliness everytime I modprobe ohci1394:
> 
> ohci1394: fw-host0: OHCI-1394 1.1 (PCI): IRQ=[5]  
> MMIO=[e0084000-e00847ff]  Max
> Packet=[2048]
> Badness in get_phy_reg at drivers/ieee1394/ohci1394.c:238
> Call Trace:
>   [<f8bb710e>] get_phy_reg+0x10e/0x120 [ohci1394]
>   [<f8bb7f3c>] ohci_devctl+0x5c/0x5d0 [ohci1394]
>   [<c03f23c0>] common_interrupt+0x18/0x20
>   [<c0113d54>] delay_tsc+0x14/0x20
>   [<f8bba058>] ohci_irq_handler+0x588/0x830 [ohci1394]
>   [<c012643f>] do_timer+0xdf/0xf0
>   [<c010a8ca>] handle_IRQ_event+0x3a/0x70
>   [<c010ac85>] do_IRQ+0xd5/0x1b0

That's due to a debug patch in -mm which catches out code which calls
mdelay() from hard IRQ context.   It's considered to be unsociable.

But there's nothing very wrong here.