On Wed, 2010-09-29 at 10:16 +0200, Michael S. Tsirkin wrote:

quoted

I compared several approaches for addressing the issue being raised

here

quoted

on how/when to update vhost_add_used_and_signal. The simple approach

I

quoted

have found is:

1. Adding completion field in struct virtqueue;
2. when it is a zero copy packet, put vhost thread wait for

completion

quoted

to update vhost_add_used_and_signal;
3. passing vq from vhost to macvtap as skb destruct_arg;
4. when skb is freed for the last reference, signal vq completion
The test results show same performance as the original patch. How do

you

quoted

think? If it sounds good to you. I will resubmit this reversion

patch.

quoted

The patch still keeps as simple as it was before. :)

Thanks
Shirley

If you look at dev_hard_start_xmit you will see a call
to skb_orphan_try which often calls the skb destructor.
So I suspect this is almost equivalent to your original patch,
and has the same correctness issue. 

If I didn't address this, then vhost net will wait for ever with wait
for completion :))


Thanks
Shirley