> Hello, world!
>
> There's a scary problem with the newer kernels (tried 2.2.3 and 2.2.6) and
> flood ping. I was able to kill my computer with the following (as root):
> # (ping -f somehost) & (ping -f otherhost) &
> I would consider this as a serious bug.
I would too, although only the superuser can send ping -f. It would be a
much bigger problem if linux couldn't handle flood pings from multiple
hosts. Still, it shouldn't occur.
>
> The machine fell into the ever-interesting land of __delay, __udelay and
> __global_cli (as witnessed with SysRq-P), and was not able to respond
> outside pings. Unfortunately, SysRq-U (or -S) didn't work.
I tried to reproduce this quickly on my machine: Uniprocessor with 3c509
compiled into the kernel. I ran a flood ping (using Alexy Kuznetsov's
fixed ping program) from my machine to 2 other machines on a personal lan,
as well as a flood ping to the machine's own ethernet address, and a flood
ping on loopback. The machine was totally stable (if a bit slow) - it's a
P66.
Anyway, maybe the problem is some SMP locking issue not present on UP.
Could you try recompiling the kernel uniprocessor and run the exact same
test (just be careful you remake all of the modules uniprocessor).
Comments from networking gurus?
Greg Zornetzer - gaz+@andrew.cmu.edu
"Light shines brightest in the darkest night"
http://www.contrib.andrew.cmu.edu/~gaz
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/