[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <ecb50c84-b412-ff0b-6c52-fd789b6c8a86@redhat.com>
Date: Wed, 16 May 2018 20:51:43 +0800
From: Jason Wang <jasowang@...hat.com>
To: Tiwei Bie <tiwei.bie@...el.com>
Cc: mst@...hat.com, virtualization@...ts.linux-foundation.org,
linux-kernel@...r.kernel.org, netdev@...r.kernel.org,
wexu@...hat.com, jfreimann@...hat.com
Subject: Re: [RFC v4 3/5] virtio_ring: add packed ring support
On 2018年05月16日 20:39, Tiwei Bie wrote:
> On Wed, May 16, 2018 at 07:50:16PM +0800, Jason Wang wrote:
>> On 2018年05月16日 16:37, Tiwei Bie wrote:
> [...]
>>> struct vring_virtqueue {
>>> @@ -116,6 +117,9 @@ struct vring_virtqueue {
>>> /* Last written value to driver->flags in
>>> * guest byte order. */
>>> u16 event_flags_shadow;
>>> +
>>> + /* ID allocation. */
>>> + struct idr buffer_id;
>> I'm not sure idr is fit for the performance critical case here. Need to
>> measure its performance impact, especially if we have few unused slots.
> I'm also not sure.. But fortunately, it should be quite easy
> to replace it with something else without changing other code.
> If it will really hurt the performance, I'll change it.
We may want to do some benchmarking/profiling to see.
>
>>> };
>>> };
> [...]
>>> +static void detach_buf_packed(struct vring_virtqueue *vq, unsigned int head,
>>> + unsigned int id, void **ctx)
>>> +{
>>> + struct vring_packed_desc *desc;
>>> + unsigned int i, j;
>>> +
>>> + /* Clear data ptr. */
>>> + vq->desc_state[id].data = NULL;
>>> +
>>> + i = head;
>>> +
>>> + for (j = 0; j < vq->desc_state[id].num; j++) {
>>> + desc = &vq->vring_packed.desc[i];
>>> + vring_unmap_one_packed(vq, desc);
>> As mentioned in previous discussion, this probably won't work for the case
>> of out of order completion since it depends on the information in the
>> descriptor ring. We probably need to extend ctx to record such information.
> Above code doesn't depend on the information in the descriptor
> ring. The vq->desc_state[] is the extended ctx.
>
> Best regards,
> Tiwei Bie
Yes, but desc is a pointer to descriptor ring I think so
vring_unmap_one_packed() still depends on the content of descriptor ring?
Thanks
Powered by blists - more mailing lists