skpush:under kerne; panic

David Mansfield (david@cobite.com)
Mon, 24 Nov 1997 23:43:43 -0500 (EST)

After seeing that skput: over crash that's been all the buzz, I'll throw
my two cents, that is two lockups in 24 hours, in. My poor machine was up
over 300 days until the hard-drive crashed, and I upgraded to RedHat 4.2
one week ago. Now I got this on sunday and again monday morning. As you
can see, I have a Rocketport driver (I've sent mail to the maintainer).
If the address 00149aac is a reference to System.map (and it is, since I
found the panic statement :-) I think the "offending" call was in
tcp_do_retransmit but I don't know how to interpret ":14". Here are the
details (on the screen before I hit reset, copied by hand):

rp: WARNING: rp_handle_port called with info_flags&NOT_INIT
Kernel Panic: skpush: under: 00149aac:14
In Swapper task not syncing

Machine details:
Linux 2.0.30 Stock, from RedHat 4.2
Intel Pentium stepping 5
3c509 NIC, module
IDE Western Digital (no fancy hdparm settings)
Comtrol Rocketport PCI, drivers 1.14 then 1.12 on the subsequent crash

Do you want the kernel config?? Sorry, I don't have it, I think I deleted
the RedHat original stuff to put down a "pristine" source. Actually I
have 2.0.32 poised to go...

Is this the teardrop or IP fragmentation attack? I've also had some
mysterious inetd exiting problems (SIGPIPE) which Dave Holland (NetKit
maintainer) says are also evidence of an attack.

The above problem though, like the poster of the skput:over crash, is on a
production machine that may just be the death of me!

David Mansfield
david@cobite.com