Re: kernel panic: uvm_fault

[Robert Swindells <rjs%fdy2.co.uk@localhost>]

Thomas Klausner wrote:
>On Mon, Dec 22, 2014 at 03:29:58PM +0000, Robert Swindells wrote:
>> 
>> Thomas Klausner wrote:
>> >I had a kernel panic today with a amd64/7.99.3 kernel from Dec 20. The
>> >last activity I had started was downloading a file from network to an
>> >NFS directory mounted from a Synology.

>> It looks the same as the panic you had back in September to me:
>> 
>> <http://mail-index.netbsd.org/current-users/2014/09/13/msg025777.html>
>> 
>> Can you turn on HW checksumming on this machine ?

>I've just done that ("ifconfig wm0 tcp4csum ip4csum udp4csum"):
>
>wm0: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> mtu 1500
>	capabilities=7ff80<TSO4,IP4CSUM_Rx,IP4CSUM_Tx,TCP4CSUM_Rx>
>	capabilities=7ff80<TCP4CSUM_Tx,UDP4CSUM_Rx,UDP4CSUM_Tx,TCP6CSUM_Rx>
>	capabilities=7ff80<TCP6CSUM_Tx,UDP6CSUM_Rx,UDP6CSUM_Tx,TSO6>
>	enabled=3f00<IP4CSUM_Rx,IP4CSUM_Tx,TCP4CSUM_Rx,TCP4CSUM_Tx>
>	enabled=3f00<UDP4CSUM_Rx,UDP4CSUM_Tx>
>	ec_capabilities=7<VLAN_MTU,VLAN_HWTAGGING,JUMBO_MTU>
>	ec_enabled=0
>...
>
>I had in the back of my memory that hardware checksumming was usually
>the cause of bugs, not when it's turned off. Am I misremembering?

Depends on the network controller, wm works well for me with
everything enabled.

wm0: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> mtu 1500
        capabilities=7ff80<TSO4,IP4CSUM_Rx,IP4CSUM_Tx,TCP4CSUM_Rx>
        capabilities=7ff80<TCP4CSUM_Tx,UDP4CSUM_Rx,UDP4CSUM_Tx,TCP6CSUM_Rx>
        capabilities=7ff80<TCP6CSUM_Tx,UDP6CSUM_Rx,UDP6CSUM_Tx,TSO6>
        enabled=7ff80<TSO4,IP4CSUM_Rx,IP4CSUM_Tx,TCP4CSUM_Rx>
        enabled=7ff80<TCP4CSUM_Tx,UDP4CSUM_Rx,UDP4CSUM_Tx,TCP6CSUM_Rx>
        enabled=7ff80<TCP6CSUM_Tx,UDP6CSUM_Rx,UDP6CSUM_Tx,TSO6>
        ec_capabilities=7<VLAN_MTU,VLAN_HWTAGGING,JUMBO_MTU>
        ec_enabled=0

I have had similar crashes to you when using sw checksumming on
amd64, never seen it on i386 or arm.

There was also this:

<http://mail-index.netbsd.org/port-sparc64/2014/11/29/msg002298.html>

I guess we need to add some more KASSERT() checks.

Robert Swindells