I had a couple of these under 1.3.69 (back to back), but I had already
killed the kernel tree to compile 1.3.71. So now I have a new one:
double lock on socket at 00153bba
00153078 t _wait_for_tcp_connect
001530f8 t _wait_for_tcp_memory
001531b8 t _do_tcp_sendmsg
---> 00153b28 t _tcp_sendmsg
00153c38 T _tcp_read_wakeup
00153f08 t _tcp_recv_urg
001540c8 t _cleanup_rbuf
(When I say 'network connected', I mean I was rlogged into another
system across my PPP link)
I had 4 VC's open, -- 1 & 3 were idle but network connected, 2 was
connected to a host which I was reasonably active on, and 4 was in the
middle of a local 'make modules' for 1.3.71. I think I had just
switched to 4 to check the status, then back to 2 to keep going, and
it popped up.
That seems to be when the other ones hit -- reasonably heavy activity
with high memory turnover (8MB RAM), switching VC's from the window
causing the activity to another, network connected VC, and starting to
type.
Hmmm.... I just got it again at the same address. I switched to vc
#1, which had been idle for a good bit of time (20 minutes?), and
started logging out of the remote host. It hit as soon as I started
typing.