Skip to contentSkip to conversation

vhost: remove deferred shadow update

PATCH In review Marvin Liu 3 participants 9 messages Apr 1, 2020 – Apr 17, 2020

lore
Share & tools
lore

Marvin Liu

author
#0 lore raw

PATCH 1/1 vhost: remove deferred shadow update

Defer shadow ring update will help overall throughput when frontend much slower than backend. But that is not all the cases we faced now. In case like ovs-dpdk + dpdk virtio user, frontend will much faster than backend. Frontend may not be able to collect available descs when shadow update is deferred. Thus will harm RFC2544 performance.

Solution is just remove deferred shadow update, which will help RFC2544 and fix potential issue with virtio net driver.

lib/librte_vhost/virtio_net.c +2 -44
index 37c47c7dc..2ba0575a7 100644
@@ -382,25 +382,6 @@ vhost_shadow_enqueue_single_packed(struct virtio_net *dev,
 	}
 }
 
-static __rte_always_inline void
-vhost_flush_dequeue_packed(struct virtio_net *dev,
-			   struct vhost_virtqueue *vq)
-{
-	int shadow_count;
-	if (!vq->shadow_used_idx)
-		return;
-
-	shadow_count = vq->last_used_idx - vq->shadow_last_used_idx;
-	if (shadow_count <= 0)
-		shadow_count += vq->size;
-
-	if ((uint32_t)shadow_count >= (vq->size - MAX_PKT_BURST)) {
-		do_data_copy_dequeue(vq);
-		vhost_flush_dequeue_shadow_packed(dev, vq);
-		vhost_vring_call_packed(dev, vq);
-	}
-}
-
 /* avoid write operation when necessary, to lessen cache issues */
 #define ASSIGN_UNLESS_EQUAL(var, val) do {	\
 	if ((var) != (val))			\
@@ -2133,20 +2114,6 @@ virtio_dev_tx_packed_zmbuf(struct virtio_net *dev,
 	return pkt_idx;
 }
 
-static __rte_always_inline bool
-next_desc_is_avail(const struct vhost_virtqueue *vq)
-{
-	bool wrap_counter = vq->avail_wrap_counter;
-	uint16_t next_used_idx = vq->last_used_idx + 1;
-
-	if (next_used_idx >= vq->size) {
-		next_used_idx -= vq->size;
-		wrap_counter ^= 1;
-	}
-
-	return desc_is_avail(&vq->desc_packed[next_used_idx], wrap_counter);
-}
-
 static __rte_noinline uint16_t
 virtio_dev_tx_packed(struct virtio_net *dev,
 		     struct vhost_virtqueue *vq,
@@ -2163,7 +2130,6 @@ virtio_dev_tx_packed(struct virtio_net *dev,
 		if (remained >= PACKED_BATCH_SIZE) {
 			if (!virtio_dev_tx_batch_packed(dev, vq, mbuf_pool,
 							&pkts[pkt_idx])) {
-				vhost_flush_dequeue_packed(dev, vq);
 				pkt_idx += PACKED_BATCH_SIZE;
 				remained -= PACKED_BATCH_SIZE;
 				continue;
@@ -2173,7 +2139,6 @@ virtio_dev_tx_packed(struct virtio_net *dev,
 		if (virtio_dev_tx_single_packed(dev, vq, mbuf_pool,
 						&pkts[pkt_idx]))
 			break;
-		vhost_flush_dequeue_packed(dev, vq);
 		pkt_idx++;
 		remained--;
 
@@ -2182,15 +2147,8 @@ virtio_dev_tx_packed(struct virtio_net *dev,
 	if (vq->shadow_used_idx) {
 		do_data_copy_dequeue(vq);
 
-		if (remained && !next_desc_is_avail(vq)) {
-			/*
-			 * The guest may be waiting to TX some buffers to
-			 * enqueue more to avoid bufferfloat, so we try to
-			 * reduce latency here.
-			 */
-			vhost_flush_dequeue_shadow_packed(dev, vq);
-			vhost_vring_call_packed(dev, vq);
-		}
+		vhost_flush_dequeue_shadow_packed(dev, vq);
+		vhost_vring_call_packed(dev, vq);
 	}
 
 	return pkt_idx;
Signature · 1 line
2.17.1
Tested-by on v1 1/1
#1 lore raw
Quoted 108 lines

-----Original Message-----
From: dev <[email protected]> On Behalf Of Marvin Liu
Sent: 2020年4月2日 5:29
To: [email protected]; Ye, Xiaolong <[email protected]>; Wang, Zhihong <[email protected]>; [email protected]
Cc: [email protected]; Liu, Yong <[email protected]>
Subject: [dpdk-dev] [PATCH] vhost: remove deferred shadow update

Defer shadow ring update will help overall throughput when frontend much slower than backend. But that is not all the cases we faced now. In case like ovs-dpdk + dpdk virtio user, frontend will much faster than backend. Frontend may not be able to collect available descs when shadow update is deferred. Thus will harm RFC2544 performance.

Solution is just remove deferred shadow update, which will help RFC2544 and fix potential issue with virtio net driver.

lib/librte_vhost/virtio_net.c index renamed from lib/librte_vhost/virtio_net.c +0 -0

37c47c7dc..2ba0575a7 100644

lib/librte_vhost/virtio_net.c +0 -19
@@ -382,25 +382,6 @@ vhost_shadow_enqueue_single_packed(struct
 virtio_net *dev,
 	}
 }
 
-static __rte_always_inline void
-vhost_flush_dequeue_packed(struct virtio_net *dev,
-			   struct vhost_virtqueue *vq)
-{
-	int shadow_count;
-	if (!vq->shadow_used_idx)
-		return;
-
-	shadow_count = vq->last_used_idx - vq->shadow_last_used_idx;
-	if (shadow_count <= 0)
-		shadow_count += vq->size;
-
-	if ((uint32_t)shadow_count >= (vq->size - MAX_PKT_BURST)) {
-		do_data_copy_dequeue(vq);
-		vhost_flush_dequeue_shadow_packed(dev, vq);
-		vhost_vring_call_packed(dev, vq);
-	}
-}
-
 /* avoid write operation when necessary, to lessen cache issues */
 #define ASSIGN_UNLESS_EQUAL(var, val) do {	\
lib/librte_vhost/virtio_net.c +2 -25
 	if ((var) != (val))			\
@@ -2133,20 +2114,6 @@ virtio_dev_tx_packed_zmbuf(struct virtio_net
 *dev,
 	return pkt_idx;
 }
 
-static __rte_always_inline bool
-next_desc_is_avail(const struct vhost_virtqueue *vq) -{
-	bool wrap_counter = vq->avail_wrap_counter;
-	uint16_t next_used_idx = vq->last_used_idx + 1;
-
-	if (next_used_idx >= vq->size) {
-		next_used_idx -= vq->size;
-		wrap_counter ^= 1;
-	}
-
-	return desc_is_avail(&vq->desc_packed[next_used_idx],
 wrap_counter);
-}
-
 static __rte_noinline uint16_t
 virtio_dev_tx_packed(struct virtio_net *dev,
 		     struct vhost_virtqueue *vq,
@@ -2163,7 +2130,6 @@ virtio_dev_tx_packed(struct virtio_net *dev,
 		if (remained >= PACKED_BATCH_SIZE) {
 			if (!virtio_dev_tx_batch_packed(dev, vq, mbuf_pool,
 							&pkts[pkt_idx])) {
-				vhost_flush_dequeue_packed(dev, vq);
 				pkt_idx += PACKED_BATCH_SIZE;
 				remained -= PACKED_BATCH_SIZE;
 				continue;
@@ -2173,7 +2139,6 @@ virtio_dev_tx_packed(struct virtio_net *dev,
 		if (virtio_dev_tx_single_packed(dev, vq, mbuf_pool,
 						&pkts[pkt_idx]))
 			break;
-		vhost_flush_dequeue_packed(dev, vq);
 		pkt_idx++;
 		remained--;
 
@@ -2182,15 +2147,8 @@ virtio_dev_tx_packed(struct virtio_net *dev,
 	if (vq->shadow_used_idx) {
 		do_data_copy_dequeue(vq);
 
-		if (remained && !next_desc_is_avail(vq)) {
-			/*
-			 * The guest may be waiting to TX some buffers to
-			 * enqueue more to avoid bufferfloat, so we try to
-			 * reduce latency here.
-			 */
-			vhost_flush_dequeue_shadow_packed(dev, vq);
-			vhost_vring_call_packed(dev, vq);
-		}
+		vhost_flush_dequeue_shadow_packed(dev, vq);
+		vhost_vring_call_packed(dev, vq);
 	}
 
 	return pkt_idx;
--

2.17.1

#2 lore raw
In reply to Marvin Liu
Quoted 5 lines · Marvin Liu

Defer shadow ring update will help overall throughput when frontend much slower than backend. But that is not all the cases we faced now. In case like ovs-dpdk + dpdk virtio user, frontend will much faster than backend. Frontend may not be able to collect available descs when shadow update is deferred. Thus will harm RFC2544 performance.

I don't understand this comment. What is the difference in term of performance between Qemu + Virtio PMD and Virtio-User PMD, as the datapath is the same?

Quoted 2 lines

Solution is just remove deferred shadow update, which will help RFC2544 and fix potential issue with virtio net driver.

What is the potential issue?

Maxime

author
#3 lore raw
Quoted 19 lines

-----Original Message-----
From: Maxime Coquelin <[email protected]>
Sent: Wednesday, April 15, 2020 10:16 PM
To: Liu, Yong <[email protected]>; Ye, Xiaolong <[email protected]>; Wang, Zhihong <[email protected]>; [email protected]
Cc: [email protected]
Subject: Re: [PATCH] vhost: remove deferred shadow update

Quoted 5 lines · Marvin Liu

Defer shadow ring update will help overall throughput when frontend much slower than backend. But that is not all the cases we faced now. In case like ovs-dpdk + dpdk virtio user, frontend will much faster than backend. Frontend may not be able to collect available descs when shadow update is deferred. Thus will harm RFC2544 performance.

Quoted 4 lines

I don't understand this comment. What is the difference in term of performance between Qemu + Virtio PMD and Virtio-User PMD, as the datapath is the same?

Hi Maxime,
The statement is for the different situations between virtio-net + vhost pmd and virtio-user + vhost pmd in ovs.
When combination is virtio-user + vhost pmd in ovs, frontend will be much faster than backend. Defer used ring update won't give benefit when requiring zero packet loss.

Regards,
Marvin

Quoted 6 lines
Quoted 2 lines

Solution is just remove deferred shadow update, which will help RFC2544 and fix potential issue with virtio net driver.

What is the potential issue?

Maxime

It is napi stops issue which has been fixed by Eugenio.

#4 lore raw

Hi Marvin,

Quoted 19 lines · Yong
Quoted 19 lines

-----Original Message-----
From: Maxime Coquelin <[email protected]>
Sent: Wednesday, April 15, 2020 10:16 PM
To: Liu, Yong <[email protected]>; Ye, Xiaolong <[email protected]>; Wang, Zhihong <[email protected]>; [email protected]
Cc: [email protected]
Subject: Re: [PATCH] vhost: remove deferred shadow update

Quoted 5 lines · Marvin Liu

Defer shadow ring update will help overall throughput when frontend much slower than backend. But that is not all the cases we faced now. In case like ovs-dpdk + dpdk virtio user, frontend will much faster than backend. Frontend may not be able to collect available descs when shadow update is deferred. Thus will harm RFC2544 performance.

Quoted 8 lines
Quoted 4 lines

I don't understand this comment. What is the difference in term of performance between Qemu + Virtio PMD and Virtio-User PMD, as the datapath is the same?

Hi Maxime,
The statement is for the different situations between virtio-net + vhost pmd and virtio-user + vhost pmd in ovs.
When combination is virtio-user + vhost pmd in ovs, frontend will be much faster than backend. Defer used ring update won't give benefit when requiring zero packet loss.

Ok, so you mean Virtio PMD vs. Virtio-net kernel driver.

Regarding who is faster between Virtio PMD and Vhost PMD, it actually depends on what the applications using them are doing.

If you have OVS on host + testpmd on guest doing IO fowarding, then of course the frontent is much faster.

But if you have testpmd IO forward on host + tespmd MACSWAP forward in guest, then the frontend could be slower.

That looks like a benchmark optimization only.

Quoted 11 lines

Regards,
Marvin

Quoted 6 lines
Quoted 2 lines

Solution is just remove deferred shadow update, which will help RFC2544 and fix potential issue with virtio net driver.

What is the potential issue?

Maxime

It is napi stops issue which has been fixed by Eugenio.

OK, then I would suggest to change the patch title to:
"vhost: fix shadow update"

Then explicit the commit message to point to Eugenio's bug, and tag it with the proper Fixes tag, so that the patch gets backported to 19.11 LTS.

Thanks,
Maxime

author
#5 lore raw
Quoted 51 lines

-----Original Message-----
From: Maxime Coquelin <[email protected]>
Sent: Wednesday, April 15, 2020 11:04 PM
To: Liu, Yong <[email protected]>; Ye, Xiaolong <[email protected]>; Wang, Zhihong <[email protected]>; [email protected]
Cc: [email protected]
Subject: Re: [PATCH] vhost: remove deferred shadow update

Hi Marvin,

Quoted 6 lines · Yong
Quoted 6 lines

-----Original Message-----
From: Maxime Coquelin <[email protected]>
Sent: Wednesday, April 15, 2020 10:16 PM
To: Liu, Yong <[email protected]>; Ye, Xiaolong

<[email protected]>;

Quoted 11 lines
Quoted 11 lines

Wang, Zhihong <[email protected]>; [email protected]
Cc: [email protected]
Subject: Re: [PATCH] vhost: remove deferred shadow update

Quoted 4 lines · Marvin Liu

Defer shadow ring update will help overall throughput when frontend much slower than backend. But that is not all the cases we faced now. In case like ovs-dpdk + dpdk virtio user, frontend will much faster than backend. Frontend may not be able to collect available descs

when

Quoted 9 lines
Quoted 6 lines
Quoted 1 line

shadow update is deferred. Thus will harm RFC2544 performance.

I don't understand this comment. What is the difference in term of performance between Qemu + Virtio PMD and Virtio-User PMD, as the datapath is the same?

Hi Maxime,
The statement is for the different situations between virtio-net + vhost

pmd and virtio-user + vhost pmd in ovs.

Quoted 1 line

When combination is virtio-user + vhost pmd in ovs, frontend will be

much faster than backend. Defer used ring update won't give benefit when requiring zero packet loss.

Ok, so you mean Virtio PMD vs. Virtio-net kernel driver.

Regarding who is faster between Virtio PMD and Vhost PMD, it actually depends on what the applications using them are doing.

Quoted 8 lines

If you have OVS on host + testpmd on guest doing IO fowarding, then of course the frontent is much faster.

But if you have testpmd IO forward on host + tespmd MACSWAP forward in guest, then the frontend could be slower.

That looks like a benchmark optimization only.

Maxime,
IMHO, it will be more like performance bug fix. Defer shadow ring update method brings performance issue in certain case.

Thanks,
Marvin

Quoted 11 lines
Quoted 4 lines

Regards,
Marvin

Quoted 1 line
Quoted 1 line

Solution is just remove deferred shadow update, which will help

RFC2544

Quoted 6 lines
Quoted 5 lines
Quoted 1 line

and fix potential issue with virtio net driver.

What is the potential issue?

Maxime

Quoted 9 lines
Quoted 1 line

It is napi stops issue which has been fixed by Eugenio.

OK, then I would suggest to change the patch title to:
"vhost: fix shadow update"

Then explicit the commit message to point to Eugenio's bug, and tag it with the proper Fixes tag, so that the patch gets backported to 19.11 LTS.

Thanks, will do it in next version.

Quoted 2 lines

Thanks,
Maxime

author
#6 lore raw

PATCH v2 1/1 vhost: fix shadow update

In reply to Marvin Liu

Defer shadow ring update introduces functional issue which has been described in Eugenio's fix patch.

The current implementation of vhost_net in packed vring tries to fill the shadow vector before send any actual changes to the guest. While this can be beneficial for the throughput, it conflicts with some bufferfloats methods like the linux kernel napi, that stops transmitting packets if there are too much bytes/buffers in the driver.

It also introduces performance issue when frontend run much faster than backend. Frontend may not be able to collect available descs when shadow update is deferred. That will harm RFC2544 throughput.

Appropriate choice is to remove deferred shadowed update method. Now shadowed used descs are flushed at the end of dequeue function.

Fixes 31d6c6a5b820 ("vhost: optimize packed ring dequeue")
lib/librte_vhost/virtio_net.c +2 -44
index 210415904..4a7531943 100644
@@ -382,25 +382,6 @@ vhost_shadow_enqueue_single_packed(struct virtio_net *dev,
 	}
 }
 
-static __rte_always_inline void
-vhost_flush_dequeue_packed(struct virtio_net *dev,
-			   struct vhost_virtqueue *vq)
-{
-	int shadow_count;
-	if (!vq->shadow_used_idx)
-		return;
-
-	shadow_count = vq->last_used_idx - vq->shadow_last_used_idx;
-	if (shadow_count <= 0)
-		shadow_count += vq->size;
-
-	if ((uint32_t)shadow_count >= (vq->size - MAX_PKT_BURST)) {
-		do_data_copy_dequeue(vq);
-		vhost_flush_dequeue_shadow_packed(dev, vq);
-		vhost_vring_call_packed(dev, vq);
-	}
-}
-
 /* avoid write operation when necessary, to lessen cache issues */
 #define ASSIGN_UNLESS_EQUAL(var, val) do {	\
 	if ((var) != (val))			\
@@ -2133,20 +2114,6 @@ virtio_dev_tx_packed_zmbuf(struct virtio_net *dev,
 	return pkt_idx;
 }
 
-static __rte_always_inline bool
-next_desc_is_avail(const struct vhost_virtqueue *vq)
-{
-	bool wrap_counter = vq->avail_wrap_counter;
-	uint16_t next_used_idx = vq->last_used_idx + 1;
-
-	if (next_used_idx >= vq->size) {
-		next_used_idx -= vq->size;
-		wrap_counter ^= 1;
-	}
-
-	return desc_is_avail(&vq->desc_packed[next_used_idx], wrap_counter);
-}
-
 static __rte_noinline uint16_t
 virtio_dev_tx_packed(struct virtio_net *dev,
 		     struct vhost_virtqueue *vq,
@@ -2163,7 +2130,6 @@ virtio_dev_tx_packed(struct virtio_net *dev,
 		if (remained >= PACKED_BATCH_SIZE) {
 			if (!virtio_dev_tx_batch_packed(dev, vq, mbuf_pool,
 							&pkts[pkt_idx])) {
-				vhost_flush_dequeue_packed(dev, vq);
 				pkt_idx += PACKED_BATCH_SIZE;
 				remained -= PACKED_BATCH_SIZE;
 				continue;
@@ -2173,7 +2139,6 @@ virtio_dev_tx_packed(struct virtio_net *dev,
 		if (virtio_dev_tx_single_packed(dev, vq, mbuf_pool,
 						&pkts[pkt_idx]))
 			break;
-		vhost_flush_dequeue_packed(dev, vq);
 		pkt_idx++;
 		remained--;
 
@@ -2182,15 +2147,8 @@ virtio_dev_tx_packed(struct virtio_net *dev,
 	if (vq->shadow_used_idx) {
 		do_data_copy_dequeue(vq);
 
-		if (remained && !next_desc_is_avail(vq)) {
-			/*
-			 * The guest may be waiting to TX some buffers to
-			 * enqueue more to avoid bufferfloat, so we try to
-			 * reduce latency here.
-			 */
-			vhost_flush_dequeue_shadow_packed(dev, vq);
-			vhost_vring_call_packed(dev, vq);
-		}
+		vhost_flush_dequeue_shadow_packed(dev, vq);
+		vhost_vring_call_packed(dev, vq);
 	}
 
 	return pkt_idx;
Signature · 1 line
2.17.1
Reviewed-by on v2 1/1
#7 lore raw

Re: [dpdk-dev] [PATCH v2] vhost: fix shadow update

Quoted 14 lines · Marvin Liu

Defer shadow ring update introduces functional issue which has been described in Eugenio's fix patch.

The current implementation of vhost_net in packed vring tries to fill the shadow vector before send any actual changes to the guest. While this can be beneficial for the throughput, it conflicts with some bufferfloats methods like the linux kernel napi, that stops transmitting packets if there are too much bytes/buffers in the driver.

It also introduces performance issue when frontend run much faster than backend. Frontend may not be able to collect available descs when shadow update is deferred. That will harm RFC2544 throughput.

Quoted 9 lines

Appropriate choice is to remove deferred shadowed update method. Now shadowed used descs are flushed at the end of dequeue function.

Fixes 31d6c6a5b820 ("vhost: optimize packed ring dequeue")

Thanks,
Maxime

#8 lore raw

Re: [dpdk-dev] [PATCH v2] vhost: fix shadow update

In reply to Marvin Liu
Quoted 14 lines · Marvin Liu

Defer shadow ring update introduces functional issue which has been described in Eugenio's fix patch.

The current implementation of vhost_net in packed vring tries to fill the shadow vector before send any actual changes to the guest. While this can be beneficial for the throughput, it conflicts with some bufferfloats methods like the linux kernel napi, that stops transmitting packets if there are too much bytes/buffers in the driver.

It also introduces performance issue when frontend run much faster than backend. Frontend may not be able to collect available descs when shadow update is deferred. That will harm RFC2544 throughput.

Quoted 8 lines

Appropriate choice is to remove deferred shadowed update method. Now shadowed used descs are flushed at the end of dequeue function.

Fixes 31d6c6a5b820 ("vhost: optimize packed ring dequeue")

Applied to dpdk-next-virtio/master

Thanks,
Maxime