[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#755249: IPv6 link non-functional after suspend



On Sun, 2014-07-20 at 15:39 +0200, Christoph Egger wrote:
> Hi!
> 
> Ben Hutchings <ben@decadent.org.uk> writes:
> > On Sat, 2014-07-19 at 08:39 +0200, Christoph Egger wrote:
> >> Package: src:linux
> >> Version: 3.14.12-1
> >> Severity: important
> >> 
> >>   After suspend/resume the computer is unable to perform any IPv6
> >> connection (no ping to fixed IP, no nc, no anything). tcpdump was
> >> unfortunately not helpfull in figuring out where exactly it breaks
> >> because, as soon as I start tcpdump, things work again. If I stop
> >> tcpdump after resume I can still reach all routed IPv6 addresses but
> >> only the local ones I have contacted while tcpdump running. Removing
> >> the interface and adding it again helps as well.
> > [...]
> >
> > IPv6 addresses are dropped when the interface is taken down.  Do you
> > have a script that does that on suspend?
> 
> I'm using pm-suspend for suspending:
> 
> | % cat /var/log/pm-suspend.log
> |            
> | % ls /etc/pm/*      
> | /etc/pm/config.d:
> | unload_module
> | 
> | /etc/pm/power.d:
> | 
> | /etc/pm/sleep.d:
> |                                                                                                   
> | % cat /etc/pm/config.d/unload_module 
> | SUSPEND_MODULES="xhci_hcd"
> 
> anywhere else to look? I haven't done anything manually. Also what's the
> current-good way to do suspend without any magic for debugging purposes?

echo mem > /sys/power/state

> The interface seems up after suspend (IPv4 works, even the autoconf address
> works after I remove the manually set IPv6 address).

It sounds a little bit like multicast RX is broken after resume (IPv6
uses multicast for neighbour discovery).  But, looking at the driver
code (r8169) it certainly appears to reprogram the hardware multicast
filter on resume.

Please send the output of 'ethtool -d eth0' before and after
suspend/resume.

> The problem either appeared with booting into the 3.13 kernel or with
> adding the br0.65 vlan interface -- I can try and see if removing the
> vlan interface changes anything as soon as I'm back home.

I don't see any changes to r8169 between 3.12 and 3.13 that could
explain this.

Ben.

-- 
Ben Hutchings
Kids!  Bringing about Armageddon can be dangerous.  Do not attempt it in
your own home. - Terry Pratchett and Neil Gaiman, `Good Omens'

Attachment: signature.asc
Description: This is a digitally signed message part


Reply to: