[ltp] NMIs, no wireless, broken consoles in new kernel

[David Abrahams]

[crossposted to ubuntu-users.  Background: this is a 2.6.15.7 kernel
built from the official dapper kernel sources with "make oldconfig",
the linux-phc patch and the machine architecture switched to
Pentium-M.]

David Abrahams <dave@boost-consulting.com> writes:

> Andrew Barr <andrew.james.barr@gmail.com> writes:
>
>> You should look through the output of 'dmesg' to see if there are
>> unresolved symbols or some other issue that is preventing the module
>> from actually loading. 
>
> @!*&^!*&~~~  Can't believe I didn't check this; I've seen it before.
>
>    ipw3945: Radio Frequency Kill Switch is On:
>    Kill switch must be turned off for wireless networking to work.
>
> There's a hardware kill switch on the front of the machine.  I thought
> I was going to like this feature but now I'm kinda sorry all the
> controls aren't in one software location that will give me a nice
> indicator, instead of showing the card as missing or unknown.
>
> Thanks for all your patience.

...of which I'm going to request more :)

The switch didn't solve the problem for the patched kernel, which I
even rebuilt again.  Also I'm having a problem with the consoles in
the patched kernel (they don't show up if I use vga=791 or splash as
boot arguments).

According to dmesg, the ipw3945 is detected without problems, but
other strange stuff is going on.  The most alarming thing is that in
both kernels' dmesgs (even the stock ubuntu dapper -686 kernel), I
see:

  Uhhuh. NMI received. Dazed and confused, but trying to continue
  You probably have a hardware problem with your RAM chips

I'm running Memtest86+ v1.65 right now and I'm at 75% with no errors,
so it seems unlikely that I'm having RAM problems... which makes me
suspect something more insidious.

I've sorted the dmesgs for both kernels and compared them.  The only
obviously interesting differences I can see in there are:

(in -stock but not in -phc kernel dmesgs):
  ADDRCONF(NETDEV_UP): eth1: link is not ready
  Console: switching to colour frame buffer device 80x25
  ipw3945: Detected geography ABG (11 802.11bg channels, 13 802.11a channels)

(in -phc kernel dmesgs twice but only in -stock kernel dmesgs once):
  eth0: no IPv6 routers present

PCI interrupt assignments are different; I can't tell whether that's
interesting or not:

--- dmesgs-stock-1	2006-08-23 19:23:00.156250000 -0400
+++ dmesgs-phc-1	2006-08-23 19:23:09.984375000 -040

-ACPI: PCI Interrupt 0000:00:1b.0[B] -> GSI 17 (level, low) -> IRQ 82
+ACPI: PCI Interrupt 0000:00:1b.0[B] -> GSI 17 (level, low) -> IRQ 74

-ACPI: PCI Interrupt 0000:00:1d.1[B] -> GSI 17 (level, low) -> IRQ 82
-ACPI: PCI Interrupt 0000:00:1d.2[C] -> GSI 18 (level, low) -> IRQ 90
-ACPI: PCI Interrupt 0000:00:1d.3[D] -> GSI 19 (level, low) -> IRQ 74
-ACPI: PCI Interrupt 0000:00:1d.7[D] -> GSI 19 (level, low) -> IRQ 74
+ACPI: PCI Interrupt 0000:00:1d.1[B] -> GSI 17 (level, low) -> IRQ 74
+ACPI: PCI Interrupt 0000:00:1d.2[C] -> GSI 18 (level, low) -> IRQ 82
+ACPI: PCI Interrupt 0000:00:1d.3[D] -> GSI 19 (level, low) -> IRQ 90
+ACPI: PCI Interrupt 0000:00:1d.7[D] -> GSI 19 (level, low) -> IRQ 90

-ACPI: PCI Interrupt 0000:03:00.0[A] -> GSI 17 (level, low) -> IRQ 82
+ACPI: PCI Interrupt 0000:03:00.0[A] -> GSI 17 (level, low) -> IRQ 74

-ehci_hcd 0000:00:1d.7: irq 74, io mem 0xee404000
+ehci_hcd 0000:00:1d.7: irq 90, io mem 0xee404000

-uhci_hcd 0000:00:1d.1: irq 82, io base 0x00001820
+uhci_hcd 0000:00:1d.1: irq 74, io base 0x00001820

-uhci_hcd 0000:00:1d.2: irq 90, io base 0x00001840
+uhci_hcd 0000:00:1d.2: irq 82, io base 0x00001840

-uhci_hcd 0000:00:1d.3: irq 74, io base 0x00001860
+uhci_hcd 0000:00:1d.3: irq 90, io base 0x00001860

Can anyone glean an insight from all this data?

Thanks!

-- 
Dave Abrahams
Boost Consulting
www.boost-consulting.com