Unknown PYXIS Machine Checks and icewm
This is a rather odd problem...
At some point (possibly around Linux 2.2.10 or 2.2.11) I started getting a
flood of PYXIS machine checks of unknown type in my console logs. They got
so bad at one point that my /var partition filled up and I could not receive
mail. I ended up hacking the kernel to ignore these exceptions :) They
look like this:
Nov 2 14:43:06 blood-axp kernel: machine check type: unknown
Nov 2 14:43:07 blood-axp kernel: PYXIS machine check: vector=0x660
pc=0xfffffc0
0003a5c9c code=0x202
Nov 2 14:43:07 blood-axp kernel: machine check type: unknown
Nov 2 14:43:07 blood-axp kernel: PYXIS machine check: vector=0x660
pc=0xfffffc0
00044f388 code=0x202
Nov 2 14:43:07 blood-axp kernel: machine check type: unknown
Nov 2 14:43:07 blood-axp kernel: PYXIS machine check: vector=0x660
pc=0xfffffc0
00044f37c code=0x202
Anyhow I figured that it must be a certain driver in the kernel that is
causing them. Then I noticed that when I was not logged into my X session,
they did not occur.
I tried using a very plain X session - no window manager, just an xterm, and
they did not happen. I then noticed that if I used twm, I did not get them,
nor with any other window manager, but if I started icewm from this solitary
xterm, the flood of console messages started again.
I don't believe this is actually a bug in icewm - instead I imagine that GCC
is probably miscompiling part of it, or that the kernel is not anticipating
some sort of condition generated by the PYXIS chipset or the PALCode.
Can anyone give me a hint as to where I should look for the bug?
--
Everything should be clear by now, so you won't need any examples.
- from the CLC-INTERCAL manual (http://www.assurdo.com/INTERCAL/)
Reply to: