>These "total freeze and no sign of error at all" hangs in 2.0.33 kernels
>still exist in 2.0.34pre8. *sight*
>So I will finally try the print_eip patch port to 2.0.33. A question
>though, how much does it affect the performance? As this is in a running
>production enviroment I cant muck around which the system too much. These
>machines act as terminal servers and therefore have _alot_ of serial
>activity.
Having serial activities don' t make difference. The machine should be
sure more slow but I have no idea of how much. I think that if you have at
least a Pentium you should be fine also with the print_eip feature running
(at least it' s so on my hardware). The worse thing is that you can' t run
X. Note that when your computer will be locked you should be able to write
down the two columns on paper. If the two columns are not blocked when
your computer seems locked the print_eip patch can' t help so much. I
should implement a SysRQ that real lock the kenrel in that cases (with a
'cli(); for(;;);') to allow people to write down the EIP addresses hoping
that are the kernel was looping somewhere or so on. So if you can' t write
down the EIP addresses when your machine is locked you should at least try
to run AltGR+BlocScorr (but probably you just tried it) to try to
understand where the kernel was at that time.
Andrea[s] Arcangeli
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu