mbox series

[v4,0/3] virtio support cache indirect desc

Message ID 20211108114951.92862-1-xuanzhuo@linux.alibaba.com (mailing list archive)
Headers show
Series virtio support cache indirect desc | expand

Message

Xuan Zhuo Nov. 8, 2021, 11:49 a.m. UTC
If the VIRTIO_RING_F_INDIRECT_DESC negotiation succeeds, and the number
of sgs used for sending packets is greater than 1. We must constantly
call __kmalloc/kfree to allocate/release desc.

In the case of extremely fast package delivery, the overhead cannot be
ignored:

  27.46%  [kernel]  [k] virtqueue_add
  16.66%  [kernel]  [k] detach_buf_split
  16.51%  [kernel]  [k] virtnet_xsk_xmit
  14.04%  [kernel]  [k] virtqueue_add_outbuf
   5.18%  [kernel]  [k] __kmalloc
   4.08%  [kernel]  [k] kfree
   2.80%  [kernel]  [k] virtqueue_get_buf_ctx
   2.22%  [kernel]  [k] xsk_tx_peek_desc
   2.08%  [kernel]  [k] memset_erms
   0.83%  [kernel]  [k] virtqueue_kick_prepare
   0.76%  [kernel]  [k] virtnet_xsk_run
   0.62%  [kernel]  [k] __free_old_xmit_ptr
   0.60%  [kernel]  [k] vring_map_one_sg
   0.53%  [kernel]  [k] native_apic_mem_write
   0.46%  [kernel]  [k] sg_next
   0.43%  [kernel]  [k] sg_init_table
   0.41%  [kernel]  [k] kmalloc_slab

This patch adds a cache function to virtio to cache these allocated indirect
desc instead of constantly allocating and releasing desc.

v4:
    1. Only allow desc cache when VIRTIO_RING_F_INDIRECT_DESC negotiation is successful
    2. The desc cache threshold can be set for each virtqueue

v3:
  pre-allocate per buffer indirect descriptors array

v2:
  use struct list_head to cache the desc

Xuan Zhuo (3):
  virtio: cache indirect desc for split
  virtio: cache indirect desc for packed
  virtio-net: enable virtio desc cache

 drivers/net/virtio_net.c     |  12 ++-
 drivers/virtio/virtio_ring.c | 152 +++++++++++++++++++++++++++++++----
 include/linux/virtio.h       |  17 ++++
 3 files changed, 163 insertions(+), 18 deletions(-)

--
2.31.0

Comments

Michael S. Tsirkin Nov. 8, 2021, 1:49 p.m. UTC | #1
On Mon, Nov 08, 2021 at 07:49:48PM +0800, Xuan Zhuo wrote:
> If the VIRTIO_RING_F_INDIRECT_DESC negotiation succeeds, and the number
> of sgs used for sending packets is greater than 1. We must constantly
> call __kmalloc/kfree to allocate/release desc.
> 
> In the case of extremely fast package delivery, the overhead cannot be
> ignored:
> 
>   27.46%  [kernel]  [k] virtqueue_add
>   16.66%  [kernel]  [k] detach_buf_split
>   16.51%  [kernel]  [k] virtnet_xsk_xmit
>   14.04%  [kernel]  [k] virtqueue_add_outbuf
>    5.18%  [kernel]  [k] __kmalloc
>    4.08%  [kernel]  [k] kfree
>    2.80%  [kernel]  [k] virtqueue_get_buf_ctx
>    2.22%  [kernel]  [k] xsk_tx_peek_desc
>    2.08%  [kernel]  [k] memset_erms
>    0.83%  [kernel]  [k] virtqueue_kick_prepare
>    0.76%  [kernel]  [k] virtnet_xsk_run
>    0.62%  [kernel]  [k] __free_old_xmit_ptr
>    0.60%  [kernel]  [k] vring_map_one_sg
>    0.53%  [kernel]  [k] native_apic_mem_write
>    0.46%  [kernel]  [k] sg_next
>    0.43%  [kernel]  [k] sg_init_table
>    0.41%  [kernel]  [k] kmalloc_slab
> 
> This patch adds a cache function to virtio to cache these allocated indirect
> desc instead of constantly allocating and releasing desc.

Hmm a bunch of comments got ignored. See e.g.
https://lore.kernel.org/r/20211027043851-mutt-send-email-mst%40kernel.org
if they aren't relevant add code comments or commit log text explaining the
design choice please.


> v4:
>     1. Only allow desc cache when VIRTIO_RING_F_INDIRECT_DESC negotiation is successful
>     2. The desc cache threshold can be set for each virtqueue
> 
> v3:
>   pre-allocate per buffer indirect descriptors array
> 
> v2:
>   use struct list_head to cache the desc
> 
> Xuan Zhuo (3):
>   virtio: cache indirect desc for split
>   virtio: cache indirect desc for packed
>   virtio-net: enable virtio desc cache
> 
>  drivers/net/virtio_net.c     |  12 ++-
>  drivers/virtio/virtio_ring.c | 152 +++++++++++++++++++++++++++++++----
>  include/linux/virtio.h       |  17 ++++
>  3 files changed, 163 insertions(+), 18 deletions(-)
> 
> --
> 2.31.0
Michael S. Tsirkin Nov. 9, 2021, 1:03 p.m. UTC | #2
On Mon, Nov 08, 2021 at 07:49:48PM +0800, Xuan Zhuo wrote:
> If the VIRTIO_RING_F_INDIRECT_DESC negotiation succeeds, and the number
> of sgs used for sending packets is greater than 1. We must constantly
> call __kmalloc/kfree to allocate/release desc.
> 
> In the case of extremely fast package delivery, the overhead cannot be
> ignored:
> 
>   27.46%  [kernel]  [k] virtqueue_add
>   16.66%  [kernel]  [k] detach_buf_split
>   16.51%  [kernel]  [k] virtnet_xsk_xmit
>   14.04%  [kernel]  [k] virtqueue_add_outbuf
>    5.18%  [kernel]  [k] __kmalloc
>    4.08%  [kernel]  [k] kfree
>    2.80%  [kernel]  [k] virtqueue_get_buf_ctx
>    2.22%  [kernel]  [k] xsk_tx_peek_desc
>    2.08%  [kernel]  [k] memset_erms
>    0.83%  [kernel]  [k] virtqueue_kick_prepare
>    0.76%  [kernel]  [k] virtnet_xsk_run
>    0.62%  [kernel]  [k] __free_old_xmit_ptr
>    0.60%  [kernel]  [k] vring_map_one_sg
>    0.53%  [kernel]  [k] native_apic_mem_write
>    0.46%  [kernel]  [k] sg_next
>    0.43%  [kernel]  [k] sg_init_table
>    0.41%  [kernel]  [k] kmalloc_slab
> 
> This patch adds a cache function to virtio to cache these allocated indirect
> desc instead of constantly allocating and releasing desc.
> 
> v4:
>     1. Only allow desc cache when VIRTIO_RING_F_INDIRECT_DESC negotiation is successful
>     2. The desc cache threshold can be set for each virtqueue
> 
> v3:
>   pre-allocate per buffer indirect descriptors array

So I'm not sure why we are doing that. Did it improve anything?


> v2:
>   use struct list_head to cache the desc
> 
> Xuan Zhuo (3):
>   virtio: cache indirect desc for split
>   virtio: cache indirect desc for packed
>   virtio-net: enable virtio desc cache
> 
>  drivers/net/virtio_net.c     |  12 ++-
>  drivers/virtio/virtio_ring.c | 152 +++++++++++++++++++++++++++++++----
>  include/linux/virtio.h       |  17 ++++
>  3 files changed, 163 insertions(+), 18 deletions(-)
> 
> --
> 2.31.0
Michael S. Tsirkin Nov. 10, 2021, 12:53 p.m. UTC | #3
On Mon, Nov 08, 2021 at 10:47:40PM +0800, Xuan Zhuo wrote:
> On Mon, 8 Nov 2021 08:49:27 -0500, Michael S. Tsirkin <mst@redhat.com> wrote:
> >
> > Hmm a bunch of comments got ignored. See e.g.
> > https://lore.kernel.org/r/20211027043851-mutt-send-email-mst%40kernel.org
> > if they aren't relevant add code comments or commit log text explaining the
> > design choice please.
> 
> I should have responded to related questions, I am guessing whether some emails
> have been lost.
> 
> I have sorted out the following 6 questions, if there are any missing questions,
> please let me know.
> 
> 1. use list_head
>   In the earliest version, I used pointers directly. You suggest that I use
>   llist_head, but considering that llist_head has atomic operations. There is no
>   competition problem here, so I used list_head.
> 
>   In fact, I did not increase the allocated space for list_head.
> 
>   use as desc array: | vring_desc | vring_desc | vring_desc | vring_desc |
>   use as queue item: | list_head ........................................|

the concern is that you touch many cache lines when removing an entry.

I suggest something like:

llist: add a non-atomic list_del_first

One has to know what one's doing, but if one has locked the list
preventing all accesses, then it's ok to just pop off an entry without
atomics.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

---

diff --git a/include/linux/llist.h b/include/linux/llist.h
index 24f207b0190b..13a47dddb12b 100644
--- a/include/linux/llist.h
+++ b/include/linux/llist.h
@@ -247,6 +247,17 @@ static inline struct llist_node *__llist_del_all(struct llist_head *head)
 
 extern struct llist_node *llist_del_first(struct llist_head *head);
 
+static inline struct llist_node *__llist_del_first(struct llist_head *head)
+{
+	struct llist_node *first = head->first;
+
+	if (!first)
+		return NULL;
+
+	head->first = first->next;
+	return first;
+}
+
 struct llist_node *llist_reverse_order(struct llist_node *head);
 
 #endif /* LLIST_H */


-----


> 2.
> > > +	if (vq->use_desc_cache && total_sg <= VIRT_QUEUE_CACHE_DESC_NUM) {
> > > +		if (vq->desc_cache_chain) {
> > > +			desc = vq->desc_cache_chain;
> > > +			vq->desc_cache_chain = (void *)desc->addr;
> > > +			goto got;
> > > +		}
> > > +		n = VIRT_QUEUE_CACHE_DESC_NUM;
> >
> > Hmm. This will allocate more entries than actually used. Why do it?
> 
> 
> This is because the size of each cache item is fixed, and the logic has been
> modified in the latest code. I think this problem no longer exists.
> 
> 
> 3.
> > What bothers me here is what happens if cache gets
> > filled on one numa node, then used on another?
> 
> I'm thinking about another question, how did the cross-numa appear here, and
> virtio desc queue also has the problem of cross-numa. So is it necessary for us
> to deal with the cross-numa scene?

It's true that desc queue might be cross numa, and people are looking
for ways to improve that. Not a reason to make things worse ...


> Indirect desc is used as virtio desc, so as long as it is in the same numa as
> virito desc. So we can allocate indirect desc cache at the same time when
> allocating virtio desc queue.

Using it from current node like we do now seems better.

> 4.
> > So e.g. for rx, we are wasting memory since indirect isn't used.
> 
> In the current version, desc cache is set up based on pre-queue.
> 
> So if the desc cache is not used, we don't need to set the desc cache.
> 
> For example, virtio-net, as long as the tx queue and the rx queue in big packet
> mode enable desc cache.


I liked how in older versions adding indrect enabled it implicitly
though without need to hack drivers.

> 5.
> > Would a better API be a cache size in bytes? This controls how much
> > memory is spent after all.
> 
> My design is to set a threshold. When total_sg is greater than this threshold,
> it will fall back to kmalloc/kfree. When total_sg is less than or equal to
> this threshold, use the allocated cache.
> 

I know. My question is this, do devices know what a good threshold is?
If yes how do they know?

> 6. kmem_cache_*
> 
> I have tested these, the performance is not as good as the method used in this
> patch.

Do you mean kmem_cache_alloc_bulk/kmem_cache_free_bulk?
You mentioned just kmem_cache_alloc previously.

> 
> Thanks.
Michael S. Tsirkin Nov. 10, 2021, 12:54 p.m. UTC | #4
On Wed, Nov 10, 2021 at 07:53:49AM -0500, Michael S. Tsirkin wrote:
> On Mon, Nov 08, 2021 at 10:47:40PM +0800, Xuan Zhuo wrote:
> > On Mon, 8 Nov 2021 08:49:27 -0500, Michael S. Tsirkin <mst@redhat.com> wrote:
> > >
> > > Hmm a bunch of comments got ignored. See e.g.
> > > https://lore.kernel.org/r/20211027043851-mutt-send-email-mst%40kernel.org
> > > if they aren't relevant add code comments or commit log text explaining the
> > > design choice please.
> > 
> > I should have responded to related questions, I am guessing whether some emails
> > have been lost.
> > 
> > I have sorted out the following 6 questions, if there are any missing questions,
> > please let me know.
> > 
> > 1. use list_head
> >   In the earliest version, I used pointers directly. You suggest that I use
> >   llist_head, but considering that llist_head has atomic operations. There is no
> >   competition problem here, so I used list_head.
> > 
> >   In fact, I did not increase the allocated space for list_head.
> > 
> >   use as desc array: | vring_desc | vring_desc | vring_desc | vring_desc |
> >   use as queue item: | list_head ........................................|
> 
> the concern is that you touch many cache lines when removing an entry.
> 
> I suggest something like:
> 
> llist: add a non-atomic list_del_first
> 
> One has to know what one's doing, but if one has locked the list
> preventing all accesses, then it's ok to just pop off an entry without
> atomics.
> 
> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
> 
> ---
> 
> diff --git a/include/linux/llist.h b/include/linux/llist.h
> index 24f207b0190b..13a47dddb12b 100644
> --- a/include/linux/llist.h
> +++ b/include/linux/llist.h
> @@ -247,6 +247,17 @@ static inline struct llist_node *__llist_del_all(struct llist_head *head)
>  
>  extern struct llist_node *llist_del_first(struct llist_head *head);
>  
> +static inline struct llist_node *__llist_del_first(struct llist_head *head)
> +{
> +	struct llist_node *first = head->first;
> +
> +	if (!first)
> +		return NULL;
> +
> +	head->first = first->next;
> +	return first;
> +}
> +
>  struct llist_node *llist_reverse_order(struct llist_node *head);
>  
>  #endif /* LLIST_H */
> 
> 
> -----
> 
> 
> > 2.
> > > > +	if (vq->use_desc_cache && total_sg <= VIRT_QUEUE_CACHE_DESC_NUM) {
> > > > +		if (vq->desc_cache_chain) {
> > > > +			desc = vq->desc_cache_chain;
> > > > +			vq->desc_cache_chain = (void *)desc->addr;
> > > > +			goto got;
> > > > +		}
> > > > +		n = VIRT_QUEUE_CACHE_DESC_NUM;
> > >
> > > Hmm. This will allocate more entries than actually used. Why do it?
> > 
> > 
> > This is because the size of each cache item is fixed, and the logic has been
> > modified in the latest code. I think this problem no longer exists.
> > 
> > 
> > 3.
> > > What bothers me here is what happens if cache gets
> > > filled on one numa node, then used on another?
> > 
> > I'm thinking about another question, how did the cross-numa appear here, and
> > virtio desc queue also has the problem of cross-numa. So is it necessary for us
> > to deal with the cross-numa scene?
> 
> It's true that desc queue might be cross numa, and people are looking
> for ways to improve that. Not a reason to make things worse ...
> 

To add to that, given it's a concern, how about actually
testing performance for this config?

> > Indirect desc is used as virtio desc, so as long as it is in the same numa as
> > virito desc. So we can allocate indirect desc cache at the same time when
> > allocating virtio desc queue.
> 
> Using it from current node like we do now seems better.
> 
> > 4.
> > > So e.g. for rx, we are wasting memory since indirect isn't used.
> > 
> > In the current version, desc cache is set up based on pre-queue.
> > 
> > So if the desc cache is not used, we don't need to set the desc cache.
> > 
> > For example, virtio-net, as long as the tx queue and the rx queue in big packet
> > mode enable desc cache.
> 
> 
> I liked how in older versions adding indrect enabled it implicitly
> though without need to hack drivers.
> 
> > 5.
> > > Would a better API be a cache size in bytes? This controls how much
> > > memory is spent after all.
> > 
> > My design is to set a threshold. When total_sg is greater than this threshold,
> > it will fall back to kmalloc/kfree. When total_sg is less than or equal to
> > this threshold, use the allocated cache.
> > 
> 
> I know. My question is this, do devices know what a good threshold is?
> If yes how do they know?
> 
> > 6. kmem_cache_*
> > 
> > I have tested these, the performance is not as good as the method used in this
> > patch.
> 
> Do you mean kmem_cache_alloc_bulk/kmem_cache_free_bulk?
> You mentioned just kmem_cache_alloc previously.
> 
> > 
> > Thanks.
Michael S. Tsirkin Nov. 11, 2021, 3:02 p.m. UTC | #5
On Thu, Nov 11, 2021 at 02:52:07PM +0800, Xuan Zhuo wrote:
> On Wed, 10 Nov 2021 07:53:44 -0500, Michael S. Tsirkin <mst@redhat.com> wrote:
> > On Mon, Nov 08, 2021 at 10:47:40PM +0800, Xuan Zhuo wrote:
> > > On Mon, 8 Nov 2021 08:49:27 -0500, Michael S. Tsirkin <mst@redhat.com> wrote:
> > > >
> > > > Hmm a bunch of comments got ignored. See e.g.
> > > > https://lore.kernel.org/r/20211027043851-mutt-send-email-mst%40kernel.org
> > > > if they aren't relevant add code comments or commit log text explaining the
> > > > design choice please.
> > >
> > > I should have responded to related questions, I am guessing whether some emails
> > > have been lost.
> > >
> > > I have sorted out the following 6 questions, if there are any missing questions,
> > > please let me know.
> > >
> > > 1. use list_head
> > >   In the earliest version, I used pointers directly. You suggest that I use
> > >   llist_head, but considering that llist_head has atomic operations. There is no
> > >   competition problem here, so I used list_head.
> > >
> > >   In fact, I did not increase the allocated space for list_head.
> > >
> > >   use as desc array: | vring_desc | vring_desc | vring_desc | vring_desc |
> > >   use as queue item: | list_head ........................................|
> >
> > the concern is that you touch many cache lines when removing an entry.
> >
> > I suggest something like:
> >
> > llist: add a non-atomic list_del_first
> >
> > One has to know what one's doing, but if one has locked the list
> > preventing all accesses, then it's ok to just pop off an entry without
> > atomics.
> >
> 
> Oh, great, but my way of solving the problem is too conservative.
> 
> > Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
> >
> > ---
> >
> > diff --git a/include/linux/llist.h b/include/linux/llist.h
> > index 24f207b0190b..13a47dddb12b 100644
> > --- a/include/linux/llist.h
> > +++ b/include/linux/llist.h
> > @@ -247,6 +247,17 @@ static inline struct llist_node *__llist_del_all(struct llist_head *head)
> >
> >  extern struct llist_node *llist_del_first(struct llist_head *head);
> >
> > +static inline struct llist_node *__llist_del_first(struct llist_head *head)
> > +{
> > +	struct llist_node *first = head->first;
> > +
> > +	if (!first)
> > +		return NULL;
> > +
> > +	head->first = first->next;
> > +	return first;
> > +}
> > +
> >  struct llist_node *llist_reverse_order(struct llist_node *head);
> >
> >  #endif /* LLIST_H */
> >
> >
> > -----
> >
> >
> > > 2.
> > > > > +	if (vq->use_desc_cache && total_sg <= VIRT_QUEUE_CACHE_DESC_NUM) {
> > > > > +		if (vq->desc_cache_chain) {
> > > > > +			desc = vq->desc_cache_chain;
> > > > > +			vq->desc_cache_chain = (void *)desc->addr;
> > > > > +			goto got;
> > > > > +		}
> > > > > +		n = VIRT_QUEUE_CACHE_DESC_NUM;
> > > >
> > > > Hmm. This will allocate more entries than actually used. Why do it?
> > >
> > >
> > > This is because the size of each cache item is fixed, and the logic has been
> > > modified in the latest code. I think this problem no longer exists.
> > >
> > >
> > > 3.
> > > > What bothers me here is what happens if cache gets
> > > > filled on one numa node, then used on another?
> > >
> > > I'm thinking about another question, how did the cross-numa appear here, and
> > > virtio desc queue also has the problem of cross-numa. So is it necessary for us
> > > to deal with the cross-numa scene?
> >
> > It's true that desc queue might be cross numa, and people are looking
> > for ways to improve that. Not a reason to make things worse ...
> >
> 
> I will test for it.
> 
> >
> > > Indirect desc is used as virtio desc, so as long as it is in the same numa as
> > > virito desc. So we can allocate indirect desc cache at the same time when
> > > allocating virtio desc queue.
> >
> > Using it from current node like we do now seems better.
> >
> > > 4.
> > > > So e.g. for rx, we are wasting memory since indirect isn't used.
> > >
> > > In the current version, desc cache is set up based on pre-queue.
> > >
> > > So if the desc cache is not used, we don't need to set the desc cache.
> > >
> > > For example, virtio-net, as long as the tx queue and the rx queue in big packet
> > > mode enable desc cache.
> >
> >
> > I liked how in older versions adding indrect enabled it implicitly
> > though without need to hack drivers.
> 
> I see.
> 
> >
> > > 5.
> > > > Would a better API be a cache size in bytes? This controls how much
> > > > memory is spent after all.
> > >
> > > My design is to set a threshold. When total_sg is greater than this threshold,
> > > it will fall back to kmalloc/kfree. When total_sg is less than or equal to
> > > this threshold, use the allocated cache.
> > >
> >
> > I know. My question is this, do devices know what a good threshold is?
> > If yes how do they know?
> 
> I think the driver knows the threshold, for example, MAX_SKB_FRAG + 2 is a
> suitable threshold for virtio-net.
> 

I guess... in that case I assume it's a good idea to have
virtio core round the size up to whole cache lines, right?

> >
> > > 6. kmem_cache_*
> > >
> > > I have tested these, the performance is not as good as the method used in this
> > > patch.
> >
> > Do you mean kmem_cache_alloc_bulk/kmem_cache_free_bulk?
> > You mentioned just kmem_cache_alloc previously.
> 
> 
> I will test for kmem_cache_alloc_bulk.
> 
> Thanks.
> 
> >
> > >
> > > Thanks.
> >