What steps will reproduce the problem?
If possible, include a link to a program on play.golang.org.
1. get tip: hg pull -u
2. cd src
3. ./all.bash

What is the expected output?

Everything compiles and tests okay.

What do you see instead?

A kernel panic.

Which compiler are you using (5g, 6g, 8g, gccgo)?

6g.

Which operating system are you using?

Kernel 2.6.26-2-xen-amd64 on a Debian 5.0 system (yes, I know it's obsolete).

Which version are you using?  (run 'go version')

go version devel +bddf2a72f5e1 Wed Nov 27 19:32:15 2013 +1100 linux/amd64

Please provide any additional information below.

This is a little hard to track down properly, since I'm compiling go on a VPS and cannot
see the console kernel messages.  I have two interfaces -- eth0, and the loopback
interface.  There are a few issues here.

#1: if an IPv6 address is assigned to eth0, TestDialMultiFDLeak in
src/pkg/net/dial_test.go causes the system to panic.

I used "ip -6 addr del ... " to delete the IPv6 addresses on eth0 (global,
link-local).  I got past that test.

#2: if the loopback interface has ::1/128, the system panics at seemingly random places.
 I've seen it crash during these tests: TestUnixListenerSpecificMethods,
TestUnixConnSpecificMethods, TestStreamConnServer, TestUDPListenClose

#3: with ::1/128 removed from loopback, there are no crashes.  However, it fails one
test:

=== RUN TestStreamConnServer
--- FAIL: TestStreamConnServer (0.00 seconds)
        server_test.go:218: Dial("tcp", "[::1]:52386") failed: dial tcp [::1]:52386: network is unreachable

I don't expect to use IPv6, so I don't mind disabling it on this server.  Plus, this is
probably a Linux issue, likely fixed in a newer kernel; it doesn't seem to be a problem
with go (except for #3 -- that test should be fixed).  The same version (commit
bddf2a72f5e1) works fine on a 3.2.0-55-virtual system.