Re: panic: assertion: _ifac->ifa_magic == IFA_CONTAINER_MAGIC in _IFAFREE

["Sepherosa Ziehau" <sepherosa@xxxxxxxxx>]

On Tue, Mar 18, 2008 at 5:53 AM, Matthew Dillon
<dillon@apollo.backplane.com> wrote:
>
>  :I have changed this test program a little bit.  Run it in following way:
>
> :./test_udp remote_ip
>  :
>  :If it paused, then on the other term:
>  :ifconfig iface_local down
>  :And kill test_udp, if you don't have TCP connection, the panic should
>  :happen immediately.
>
>     I'm not sure if I'm using the right patch set.  With your
>     rtfree_remote.diff and your latest test regimen I am getting
>     numerous "remote rtfree 1->0" messages on the console when I test
>     with a vkernel -n 2, but not getting a crash.

This patch is a workaround instead of real fix: it only makes sure
that route entry allocated on CPU0 is freed on CPU0, but it may still
be accessed by CPU1 before being freed.  The test program will not
cause panic with this patch in place.

There is one thing I want to get into cvs in this patch:
- Add rt_cpuid in rtentry
It at least could help debugging problems.  And we may want to assert
route entry is only accessed by owner CPU using this field.

Best Regards,
sephe

-- 
Live Free or Die