Re: refill msk(4) rx ring from a timeout when there's no mbufs

2017-10-24 Thread Mark Kettenis
> Date: Tue, 24 Oct 2017 09:37:38 +1000 > From: David Gwynne > > theo pointed out the timeout should be cancelled when the interface > goes down. this adds a timeout_del in msk_stop. Hmm, you didn't attach a new diff... Commenting on the previous diff; do we really need to

Re: refill msk(4) rx ring from a timeout when there's no mbufs

2017-10-23 Thread David Gwynne
theo pointed out the timeout should be cancelled when the interface goes down. this adds a timeout_del in msk_stop. some other drivers suffer this problem, ill have a look at them once this is in. On Mon, Oct 23, 2017 at 11:35:09AM +1000, David Gwynne wrote: > if msk runs out of mbufs, the rx

refill msk(4) rx ring from a timeout when there's no mbufs

2017-10-22 Thread David Gwynne
if msk runs out of mbufs, the rx ring remains empty and there's nothing except an ifconfig down and up to get it going again. this adds a timeout to refill the ring. it's largely copied from other drivers (vr in this case). tests? ok? Index: if_mskvar.h