[PATCH v2] kni: fix possible rx_q mbuf leak and speed up alloc_q release

[PATCH v2] kni: fix possible rx_q mbuf leak and speed up alloc_q release

[Yangchao Zhou]

>From 3293d014123150bcd0650ce1d09b19bbb9eae9c4 Mon Sep 17 00:00:00 2001
From: zhouyangchao <zhouyates@gmail.com>
Date: Tue, 10 Apr 2018 18:26:35 +0800
Subject: [PATCH] kni: fix possible rx_q mbuf leak and speed up alloc_q release

rx_q fifo can only be released by kernel thread. There may be
mbuf leaks in rx_q because  kernel threads are randomly stopped.

When the kni is released and netdev is unregisterd, convert the
physical address mbufs in rx_q to the virtual address in free_q.
By the way, alloc_q can be processed together to speed up the
release rate in userspace.

Signed-off-by: Yangchao Zhou <zhouyates@gmail.com>
Suggested-by: Ferruh Yigit <ferruh.yigit@intel.com>
---
 kernel/linux/kni/kni_dev.h  |  1 +
 kernel/linux/kni/kni_misc.c |  2 ++
 kernel/linux/kni/kni_net.c  | 40 ++++++++++++++++++++++++++++++++++++++++
 3 files changed, 43 insertions(+)

diff --git a/kernel/linux/kni/kni_dev.h b/kernel/linux/kni/kni_dev.h
index c9393d8..6275ef2 100644
--- a/kernel/linux/kni/kni_dev.h
+++ b/kernel/linux/kni/kni_dev.h
@@ -92,6 +92,7 @@ struct kni_dev {
 	void *alloc_va[MBUF_BURST_SZ];
 };
 
+void kni_net_release_fifo_phy(struct kni_dev *kni);
 void kni_net_rx(struct kni_dev *kni);
 void kni_net_init(struct net_device *dev);
 void kni_net_config_lo_mode(char *lo_str);
diff --git a/kernel/linux/kni/kni_misc.c b/kernel/linux/kni/kni_misc.c
index 01574ec..fa69f8e 100644
--- a/kernel/linux/kni/kni_misc.c
+++ b/kernel/linux/kni/kni_misc.c
@@ -192,6 +192,8 @@ kni_dev_remove(struct kni_dev *dev)
 		free_netdev(dev->net_dev);
 	}
 
+	kni_net_release_fifo_phy(dev);
+
 	return 0;
 }
 
diff --git a/kernel/linux/kni/kni_net.c b/kernel/linux/kni/kni_net.c
index 9f9b798..1d64d78 100644
--- a/kernel/linux/kni/kni_net.c
+++ b/kernel/linux/kni/kni_net.c
@@ -163,6 +163,46 @@ kni_net_release(struct net_device *dev)
 	return (ret == 0) ? req.result : ret;
 }
 
+static void
+kni_fifo_trans_pa2va(struct kni_dev *kni,
+	struct rte_kni_fifo *src_pa, struct rte_kni_fifo *dst_va)
+{
+	uint32_t ret, i, num_fq, num_rx;
+	void *kva;
+	do {
+		num_fq = kni_fifo_free_count(kni->free_q);
+		if (num_fq == 0)
+			return;
+
+		num_rx = min_t(uint32_t, num_fq, MBUF_BURST_SZ);
+
+		num_rx = kni_fifo_get(src_pa, kni->pa, num_rx);
+		if (num_rx == 0)
+			return;
+
+		for (i = 0; i < num_rx; i++) {
+			kva = pa2kva(kni->pa[i]);
+			kni->va[i] = pa2va(kni->pa[i], kva);
+		}
+
+		ret = kni_fifo_put(dst_va, kni->va, num_rx);
+		if (ret != num_rx) {
+			/* Failing should not happen */
+			pr_err("Fail to enqueue entries into dst_va\n");
+			return;
+		}
+	} while (1);
+}
+
+/* Try to release mbufs when kni release */
+void kni_net_release_fifo_phy(struct kni_dev *kni)
+{
+	/* release rx_q first, because it can't release in userspace */
+	kni_fifo_trans_pa2va(kni, kni->rx_q, kni->free_q);
+	/* release alloc_q for speeding up kni release in userspace */
+	kni_fifo_trans_pa2va(kni, kni->alloc_q, kni->free_q);
+}
+
 /*
  * Configuration changes (passed on by ifconfig)
  */
-- 
2.9.0.windows.1

[PATCH v3] kni: fix possible rx_q mbuf leaks and speed up alloc_q release

[Yangchao Zhou]

rx_q fifo can only be released by kernel thread. There may be
mbuf leaks in rx_q because  kernel threads are randomly stopped.

When the kni is released and netdev is unregisterd, convert the
physical address mbufs in rx_q to the virtual address in free_q.
By the way, alloc_q can be processed together to speed up the
release rate in userspace.

Signed-off-by: Yangchao Zhou <zhouyates@gmail.com>
Suggested-by: Ferruh Yigit <ferruh.yigit@intel.com>
---
 kernel/linux/kni/kni_dev.h  |    1 +
 kernel/linux/kni/kni_misc.c |    2 ++
 kernel/linux/kni/kni_net.c  |   40 ++++++++++++++++++++++++++++++++++++++++
 3 files changed, 43 insertions(+), 0 deletions(-)

diff --git a/kernel/linux/kni/kni_dev.h b/kernel/linux/kni/kni_dev.h
index c9393d8..6275ef2 100644
--- a/kernel/linux/kni/kni_dev.h
+++ b/kernel/linux/kni/kni_dev.h
@@ -92,6 +92,7 @@ struct kni_dev {
 	void *alloc_va[MBUF_BURST_SZ];
 };
 
+void kni_net_release_fifo_phy(struct kni_dev *kni);
 void kni_net_rx(struct kni_dev *kni);
 void kni_net_init(struct net_device *dev);
 void kni_net_config_lo_mode(char *lo_str);
diff --git a/kernel/linux/kni/kni_misc.c b/kernel/linux/kni/kni_misc.c
index 01574ec..fa69f8e 100644
--- a/kernel/linux/kni/kni_misc.c
+++ b/kernel/linux/kni/kni_misc.c
@@ -192,6 +192,8 @@ struct kni_net {
 		free_netdev(dev->net_dev);
 	}
 
+	kni_net_release_fifo_phy(dev);
+
 	return 0;
 }
 
diff --git a/kernel/linux/kni/kni_net.c b/kernel/linux/kni/kni_net.c
index 9f9b798..1d64d78 100644
--- a/kernel/linux/kni/kni_net.c
+++ b/kernel/linux/kni/kni_net.c
@@ -163,6 +163,46 @@
 	return (ret == 0) ? req.result : ret;
 }
 
+static void
+kni_fifo_trans_pa2va(struct kni_dev *kni,
+	struct rte_kni_fifo *src_pa, struct rte_kni_fifo *dst_va)
+{
+	uint32_t ret, i, num_fq, num_rx;
+	void *kva;
+	do {
+		num_fq = kni_fifo_free_count(kni->free_q);
+		if (num_fq == 0)
+			return;
+
+		num_rx = min_t(uint32_t, num_fq, MBUF_BURST_SZ);
+
+		num_rx = kni_fifo_get(src_pa, kni->pa, num_rx);
+		if (num_rx == 0)
+			return;
+
+		for (i = 0; i < num_rx; i++) {
+			kva = pa2kva(kni->pa[i]);
+			kni->va[i] = pa2va(kni->pa[i], kva);
+		}
+
+		ret = kni_fifo_put(dst_va, kni->va, num_rx);
+		if (ret != num_rx) {
+			/* Failing should not happen */
+			pr_err("Fail to enqueue entries into dst_va\n");
+			return;
+		}
+	} while (1);
+}
+
+/* Try to release mbufs when kni release */
+void kni_net_release_fifo_phy(struct kni_dev *kni)
+{
+	/* release rx_q first, because it can't release in userspace */
+	kni_fifo_trans_pa2va(kni, kni->rx_q, kni->free_q);
+	/* release alloc_q for speeding up kni release in userspace */
+	kni_fifo_trans_pa2va(kni, kni->alloc_q, kni->free_q);
+}
+
 /*
  * Configuration changes (passed on by ifconfig)
  */
-- 
1.7.1

Re: [PATCH v3] kni: fix possible rx_q mbuf leaks and speed up alloc_q release

[Ferruh Yigit]

On 4/17/2018 5:01 AM, Yangchao Zhou wrote:
> rx_q fifo can only be released by kernel thread. There may be
> mbuf leaks in rx_q because  kernel threads are randomly stopped.
> 
> When the kni is released and netdev is unregisterd, convert the
> physical address mbufs in rx_q to the virtual address in free_q.
> By the way, alloc_q can be processed together to speed up the
> release rate in userspace.

Overall looks good to me. A few comments below.

Do you observe a speed up in userspace related alloc_q free?

> 
> Signed-off-by: Yangchao Zhou <zhouyates@gmail.com>
> Suggested-by: Ferruh Yigit <ferruh.yigit@intel.com>
> ---
>  kernel/linux/kni/kni_dev.h  |    1 +
>  kernel/linux/kni/kni_misc.c |    2 ++
>  kernel/linux/kni/kni_net.c  |   40 ++++++++++++++++++++++++++++++++++++++++
>  3 files changed, 43 insertions(+), 0 deletions(-)
> 
> diff --git a/kernel/linux/kni/kni_dev.h b/kernel/linux/kni/kni_dev.h
> index c9393d8..6275ef2 100644
> --- a/kernel/linux/kni/kni_dev.h
> +++ b/kernel/linux/kni/kni_dev.h
> @@ -92,6 +92,7 @@ struct kni_dev {
>  	void *alloc_va[MBUF_BURST_SZ];
>  };
>  
> +void kni_net_release_fifo_phy(struct kni_dev *kni);
>  void kni_net_rx(struct kni_dev *kni);
>  void kni_net_init(struct net_device *dev);
>  void kni_net_config_lo_mode(char *lo_str);
> diff --git a/kernel/linux/kni/kni_misc.c b/kernel/linux/kni/kni_misc.c
> index 01574ec..fa69f8e 100644
> --- a/kernel/linux/kni/kni_misc.c
> +++ b/kernel/linux/kni/kni_misc.c
> @@ -192,6 +192,8 @@ struct kni_net {
>  		free_netdev(dev->net_dev);
>  	}
>  
> +	kni_net_release_fifo_phy(dev);
> +
>  	return 0;
>  }
>  
> diff --git a/kernel/linux/kni/kni_net.c b/kernel/linux/kni/kni_net.c
> index 9f9b798..1d64d78 100644
> --- a/kernel/linux/kni/kni_net.c
> +++ b/kernel/linux/kni/kni_net.c
> @@ -163,6 +163,46 @@
>  	return (ret == 0) ? req.result : ret;
>  }
>  
> +static void
> +kni_fifo_trans_pa2va(struct kni_dev *kni,
> +	struct rte_kni_fifo *src_pa, struct rte_kni_fifo *dst_va)
> +{
> +	uint32_t ret, i, num_fq, num_rx;
> +	void *kva;
> +	do {
> +		num_fq = kni_fifo_free_count(kni->free_q);

Since this function made more generic with src, dst arguments, it is better to
use dst_va here instead of kni->free_q

> +		if (num_fq == 0)

And this variable name can be updated from free queue to dst.

> +			return;
> +
> +		num_rx = min_t(uint32_t, num_fq, MBUF_BURST_SZ);
> +
> +		num_rx = kni_fifo_get(src_pa, kni->pa, num_rx);
> +		if (num_rx == 0)
> +			return;
> +
> +		for (i = 0; i < num_rx; i++) {
> +			kva = pa2kva(kni->pa[i]);
> +			kni->va[i] = pa2va(kni->pa[i], kva);
> +		}
> +
> +		ret = kni_fifo_put(dst_va, kni->va, num_rx);
> +		if (ret != num_rx) {
> +			/* Failing should not happen */
> +			pr_err("Fail to enqueue entries into dst_va\n");
> +			return;
> +		}
> +	} while (1);
> +}
> +
> +/* Try to release mbufs when kni release */
> +void kni_net_release_fifo_phy(struct kni_dev *kni)
> +{
> +	/* release rx_q first, because it can't release in userspace */
> +	kni_fifo_trans_pa2va(kni, kni->rx_q, kni->free_q);
> +	/* release alloc_q for speeding up kni release in userspace */
> +	kni_fifo_trans_pa2va(kni, kni->alloc_q, kni->free_q);
> +}
> +
>  /*
>   * Configuration changes (passed on by ifconfig)
>   */
> 

[PATCH v4] kni: fix possible rx_q mbuf leaks and speed up alloc_q release

[Yangchao Zhou]

rx_q fifo can only be released by kernel thread. There may be
mbuf leaks in rx_q because kernel threads are randomly stopped.

When the kni is released and netdev is unregisterd, convert the
physical address mbufs in rx_q to the virtual address in free_q.
By the way, alloc_q can be processed together to speed up the
release rate in userspace.

In my test, it is improved from 300-500ms with a mempool that has
 131072 mbufs to 10ms(regardless of the specifications).

Signed-off-by: Yangchao Zhou <zhouyates@gmail.com>
Suggested-by: Ferruh Yigit <ferruh.yigit@intel.com>
---
v4:
* Add improve performance description.
---
 kernel/linux/kni/kni_dev.h  |    1 +
 kernel/linux/kni/kni_misc.c |    2 ++
 kernel/linux/kni/kni_net.c  |   40 ++++++++++++++++++++++++++++++++++++++++
 3 files changed, 43 insertions(+), 0 deletions(-)

diff --git a/kernel/linux/kni/kni_dev.h b/kernel/linux/kni/kni_dev.h
index c9393d8..6275ef2 100644
--- a/kernel/linux/kni/kni_dev.h
+++ b/kernel/linux/kni/kni_dev.h
@@ -92,6 +92,7 @@ struct kni_dev {
 	void *alloc_va[MBUF_BURST_SZ];
 };
 
+void kni_net_release_fifo_phy(struct kni_dev *kni);
 void kni_net_rx(struct kni_dev *kni);
 void kni_net_init(struct net_device *dev);
 void kni_net_config_lo_mode(char *lo_str);
diff --git a/kernel/linux/kni/kni_misc.c b/kernel/linux/kni/kni_misc.c
index 01574ec..fa69f8e 100644
--- a/kernel/linux/kni/kni_misc.c
+++ b/kernel/linux/kni/kni_misc.c
@@ -192,6 +192,8 @@ struct kni_net {
 		free_netdev(dev->net_dev);
 	}
 
+	kni_net_release_fifo_phy(dev);
+
 	return 0;
 }
 
diff --git a/kernel/linux/kni/kni_net.c b/kernel/linux/kni/kni_net.c
index 9f9b798..7fcfa10 100644
--- a/kernel/linux/kni/kni_net.c
+++ b/kernel/linux/kni/kni_net.c
@@ -163,6 +163,46 @@
 	return (ret == 0) ? req.result : ret;
 }
 
+static void
+kni_fifo_trans_pa2va(struct kni_dev *kni,
+	struct rte_kni_fifo *src_pa, struct rte_kni_fifo *dst_va)
+{
+	uint32_t ret, i, num_dst, num_rx;
+	void *kva;
+	do {
+		num_dst = kni_fifo_free_count(dst_va);
+		if (num_dst == 0)
+			return;
+
+		num_rx = min_t(uint32_t, num_dst, MBUF_BURST_SZ);
+
+		num_rx = kni_fifo_get(src_pa, kni->pa, num_rx);
+		if (num_rx == 0)
+			return;
+
+		for (i = 0; i < num_rx; i++) {
+			kva = pa2kva(kni->pa[i]);
+			kni->va[i] = pa2va(kni->pa[i], kva);
+		}
+
+		ret = kni_fifo_put(dst_va, kni->va, num_rx);
+		if (ret != num_rx) {
+			/* Failing should not happen */
+			pr_err("Fail to enqueue entries into dst_va\n");
+			return;
+		}
+	} while (1);
+}
+
+/* Try to release mbufs when kni release */
+void kni_net_release_fifo_phy(struct kni_dev *kni)
+{
+	/* release rx_q first, because it can't release in userspace */
+	kni_fifo_trans_pa2va(kni, kni->rx_q, kni->free_q);
+	/* release alloc_q for speeding up kni release in userspace */
+	kni_fifo_trans_pa2va(kni, kni->alloc_q, kni->free_q);
+}
+
 /*
  * Configuration changes (passed on by ifconfig)
  */
-- 
1.7.1

Re: [PATCH v4] kni: fix possible rx_q mbuf leaks and speed up alloc_q release

[Ferruh Yigit]

On 4/19/2018 4:12 AM, Yangchao Zhou wrote:
> rx_q fifo can only be released by kernel thread. There may be
> mbuf leaks in rx_q because kernel threads are randomly stopped.
> 
> When the kni is released and netdev is unregisterd, convert the
> physical address mbufs in rx_q to the virtual address in free_q.
> By the way, alloc_q can be processed together to speed up the
> release rate in userspace.
> 
> In my test, it is improved from 300-500ms with a mempool that has
>  131072 mbufs to 10ms(regardless of the specifications).
> 
> Signed-off-by: Yangchao Zhou <zhouyates@gmail.com>
> Suggested-by: Ferruh Yigit <ferruh.yigit@intel.com>
> ---
> v4:
> * Add improve performance description.

Acked-by: Ferruh Yigit <ferruh.yigit@intel.com>

Re: [dpdk-stable] [PATCH v4] kni: fix possible rx_q mbuf leaks and speed up alloc_q release

[Thomas Monjalon]

04/05/2018 14:46, Ferruh Yigit:
> On 4/19/2018 4:12 AM, Yangchao Zhou wrote:
> > rx_q fifo can only be released by kernel thread. There may be
> > mbuf leaks in rx_q because kernel threads are randomly stopped.
> > 
> > When the kni is released and netdev is unregisterd, convert the
> > physical address mbufs in rx_q to the virtual address in free_q.
> > By the way, alloc_q can be processed together to speed up the
> > release rate in userspace.
> > 
> > In my test, it is improved from 300-500ms with a mempool that has
> >  131072 mbufs to 10ms(regardless of the specifications).
> > 
> > Signed-off-by: Yangchao Zhou <zhouyates@gmail.com>
> > Suggested-by: Ferruh Yigit <ferruh.yigit@intel.com>
> > ---
> > v4:
> > * Add improve performance description.
> 
> Acked-by: Ferruh Yigit <ferruh.yigit@intel.com>

Applied, thanks