nx6310 cardbus (was: 4.99.37 i386 regression ?)

[Frank Wille <frank%phoenix.owl.de@localhost>]

Chavdar Ivanov wrote:

> I have a Combo Cardbus USB 2.0 / Firewire card which used to work
> without any problems for quite some time on my nx6310 laptop. With
> -current from two days ago I no longer can use it. The following is
> taken from screen:
>
> ohci0 at cardbus0 function 0: Acer Labs M5237 USB 1.1 Host Controller (rev. 
> 0x03)
> NMI ... going to debugger
> [...]

I can confirm the strange NMI on my hp-nx6310 as well! It happens here with a
Vodafone UMTS card:

---8<---
ohci0 at cardbus0 function 0: Opti 82C861 (rev. 0x10)
NMI ... going to debugger
---8<---

The backtrace shows me that the NMI happened during inl(). But I can
continue the program:

---8<---
> cont
ohci0: interrupting at 10
ohci0: OHCI version 1.0, legacy support
usb5 at ohci0: USB revision 1.0
uhub5 at usb5
uhub5: Opti OHCI root hub, class 9/0, rev 1.00/1.00, addr 1
uhub5: 2 ports with 2 removable, self powered
ugen0 at uhub5 port 1
ugen0: Vodafone Vodafone Mobile Connect Card - 3G, rev 2.00/0.00, addr 2
uhub5: at usb5 (addr 1) disconnected
ugen0: at uhub5 port 1 (addr 2) disconnected
ugen0 detached
uhub5 detached
usb5 detached
ohci0 detached
---8<---

I tried this with another Laptop (a Fujitsu-Siemens Amilo), and I can insert
the card without an NMI! So it must have something to do with attaching ohci
to the nx6310's cardbus.

The kernels which I tried were 4.99.50 and 4.99.54 (latest, from yesterday).
And some time ago it did work - with 4.99.5:

---8<--
ohci0 at cardbus0 function 0: Opti 82C861 (rev. 0x10)
cardbus_mapreg_map called: cardbus0 0
cardbus_mapreg_map: physaddr 40000000
ohci0: interrupting at 10
ohci0: OHCI version 1.0, legacy support
usb5 at ohci0: USB revision 1.0
uhub5 at usb5
uhub5: Opti OHCI root hub, class 9/0, rev 1.00/1.00, addr 1
uhub5: 2 ports with 2 removable, self powered
ubsa0 at uhub5 port 1
ubsa0: Vodafone Vodafone Mobile Connect Card - 3G, rev 2.00/0.00, addr 2
ucom0 at ubsa0
---8<---

It even attaches an ubsa, and not ugen. Something bad must have been
introduced between 4.99.5 and 4.99.37.

-- 
    _  Frank Wille (frank%phoenix.owl.de@localhost)
 _ //  http://sun.hasenbraten.de/~frank/
 \X/   Phx @ #AmigaGer