I noticed that the aha2940 driver can kill your system regardless of
the kernel version. I tried with 1.2.13 and with 1.3.94 (I dont have 1.3.100
yet).
hardware layout is about:
+ asus sp3 board
+ i486/100
+ 16MB real
+ 20MB swap
+ aha2940u (but same problem with plain aha2940)
+ ne2000 compatible card
+ 1G quantum fireball EIDE drive OR
+ very old 80MB drive
+ SCSI CD-Rom - this is the only device on the SCSI-Bus !
= Thosiba Quad-Speed.
how to crash:
type "dd if=/dev/scd0 of=/dev/null"
switch to another VT, try to log in, do something like "ps" or "w" or "df"
again, switch to another VT, try to log in, .... -"-
[...]
when I used the 1G Quantum F.Ball, I got kernel panics, which I cannot
report right now (the havent been saved to the logfile anyway...)
after that, I switched to an old 80MB hd. with this drive I didn't get
kernel panics anymore, but instead messages like
"ide0 { irq timeout } "
"ide0 { Reset complete } "
and such. the system was in a slightly unusable state after that,
e.g. I found /etc/passwd to be a hardlink to /usr/bin/loadkeys :-)
my conclusion is that there is a) a misbehaviour in the aha2940 code,
which causes this "uncooperativeness" and b) a possible error in
the ide code as well. shouldn't a "Reset complete" result in a
continuing of the previous operations without confusing i-nodes ?
btw - I'm curious if someone else with a likely configuration can
reproduce this error.
best regards, herbert rosmanith
herp@wildsau.idv.uni-linz.ac.at
rosmanith@edvz.uni-linz.ac.at