Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1532957 > unrolled thread

[PATCH net 1/2] tun: handle ubuf refcount correctly when meet erros

Back to article view | Back to linux.kernel

Contents

  [PATCH net 1/2] tun: handle ubuf refcount correctly when meet erros Jason Wang <jasowang@redhat.com> - 2016-11-30 06:20 +0100
    [PATCH net 2/2] macvtap: handle ubuf refcount correctly when meet erros Jason Wang <jasowang@redhat.com> - 2016-11-30 06:20 +0100
      Re: [PATCH net 2/2] macvtap: handle ubuf refcount correctly when  meet erros "Michael S. Tsirkin" <mst@redhat.com> - 2016-11-30 15:00 +0100
    Re: [PATCH net 1/2] tun: handle ubuf refcount correctly when meet  erros "Michael S. Tsirkin" <mst@redhat.com> - 2016-11-30 15:00 +0100

#1532957 — [PATCH net 1/2] tun: handle ubuf refcount correctly when meet erros

We trigger uarg->callback() immediately after we decide do datacopy
even if caller want to do zerocopy. This will cause the callback
(vhost_net_zerocopy_callback) decrease the refcount. But when we meet
an error afterwards, the error handling in vhost handle_tx() will try
to decrease it again. This is wrong and fix this by delay the
uarg->callback() until we're sure there's no errors.

Reported-by: wangyunjian <wangyunjian@huawei.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
---
The patch is needed for -stable.
---
 drivers/net/tun.c | 10 ++++------
 1 file changed, 4 insertions(+), 6 deletions(-)

diff --git a/drivers/net/tun.c b/drivers/net/tun.c
index 8093e39..db6acec 100644
--- a/drivers/net/tun.c
+++ b/drivers/net/tun.c
@@ -1246,13 +1246,8 @@ static ssize_t tun_get_user(struct tun_struct *tun, struct tun_file *tfile,
 
 	if (zerocopy)
 		err = zerocopy_sg_from_iter(skb, from);
-	else {
+	else
 		err = skb_copy_datagram_from_iter(skb, 0, from, len);
-		if (!err && msg_control) {
-			struct ubuf_info *uarg = msg_control;
-			uarg->callback(uarg, false);
-		}
-	}
 
 	if (err) {
 		this_cpu_inc(tun->pcpu_stats->rx_dropped);
@@ -1298,6 +1293,9 @@ static ssize_t tun_get_user(struct tun_struct *tun, struct tun_file *tfile,
 		skb_shinfo(skb)->destructor_arg = msg_control;
 		skb_shinfo(skb)->tx_flags |= SKBTX_DEV_ZEROCOPY;
 		skb_shinfo(skb)->tx_flags |= SKBTX_SHARED_FRAG;
+	} else if (msg_control) {
+		struct ubuf_info *uarg = msg_control;
+		uarg->callback(uarg, false);
 	}
 
 	skb_reset_network_header(skb);
-- 
2.7.4

[toc] | [next] | [standalone]

#1532960 — [PATCH net 2/2] macvtap: handle ubuf refcount correctly when meet erros

We trigger uarg->callback() immediately after we decide do datacopy
even if caller want to do zerocopy. This will cause the callback
(vhost_net_zerocopy_callback) decrease the refcount. But when we meet
an error afterwards, the error handling in vhost handle_tx() will try
to decrease it again. This is wrong and fix this by delay the
uarg->callback() until we're sure there's no errors.

Signed-off-by: Jason Wang <jasowang@redhat.com>
---
The patch is needed for -stable.
---
 drivers/net/macvtap.c | 11 +++++------
 1 file changed, 5 insertions(+), 6 deletions(-)

diff --git a/drivers/net/macvtap.c b/drivers/net/macvtap.c
index bceca28..7869b06 100644
--- a/drivers/net/macvtap.c
+++ b/drivers/net/macvtap.c
@@ -742,13 +742,8 @@ static ssize_t macvtap_get_user(struct macvtap_queue *q, struct msghdr *m,
 
 	if (zerocopy)
 		err = zerocopy_sg_from_iter(skb, from);
-	else {
+	else
 		err = skb_copy_datagram_from_iter(skb, 0, from, len);
-		if (!err && m && m->msg_control) {
-			struct ubuf_info *uarg = m->msg_control;
-			uarg->callback(uarg, false);
-		}
-	}
 
 	if (err)
 		goto err_kfree;
@@ -779,7 +774,11 @@ static ssize_t macvtap_get_user(struct macvtap_queue *q, struct msghdr *m,
 		skb_shinfo(skb)->destructor_arg = m->msg_control;
 		skb_shinfo(skb)->tx_flags |= SKBTX_DEV_ZEROCOPY;
 		skb_shinfo(skb)->tx_flags |= SKBTX_SHARED_FRAG;
+	} else if (m && m->msg_control) {
+		struct ubuf_info *uarg = m->msg_control;
+		uarg->callback(uarg, false);
 	}
+
 	if (vlan) {
 		skb->dev = vlan->dev;
 		dev_queue_xmit(skb);
-- 
2.7.4

[toc] | [prev] | [next] | [standalone]

#1533315 — Re: [PATCH net 2/2] macvtap: handle ubuf refcount correctly when meet erros

On Wed, Nov 30, 2016 at 01:17:52PM +0800, Jason Wang wrote:
> We trigger uarg->callback() immediately after we decide do datacopy
> even if caller want to do zerocopy. This will cause the callback
> (vhost_net_zerocopy_callback) decrease the refcount. But when we meet
> an error afterwards, the error handling in vhost handle_tx() will try
> to decrease it again. This is wrong and fix this by delay the
> uarg->callback() until we're sure there's no errors.
> 
> Signed-off-by: Jason Wang <jasowang@redhat.com>

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

> ---
> The patch is needed for -stable.
> ---
>  drivers/net/macvtap.c | 11 +++++------
>  1 file changed, 5 insertions(+), 6 deletions(-)
> 
> diff --git a/drivers/net/macvtap.c b/drivers/net/macvtap.c
> index bceca28..7869b06 100644
> --- a/drivers/net/macvtap.c
> +++ b/drivers/net/macvtap.c
> @@ -742,13 +742,8 @@ static ssize_t macvtap_get_user(struct macvtap_queue *q, struct msghdr *m,
>  
>  	if (zerocopy)
>  		err = zerocopy_sg_from_iter(skb, from);
> -	else {
> +	else
>  		err = skb_copy_datagram_from_iter(skb, 0, from, len);
> -		if (!err && m && m->msg_control) {
> -			struct ubuf_info *uarg = m->msg_control;
> -			uarg->callback(uarg, false);
> -		}
> -	}
>  
>  	if (err)
>  		goto err_kfree;
> @@ -779,7 +774,11 @@ static ssize_t macvtap_get_user(struct macvtap_queue *q, struct msghdr *m,
>  		skb_shinfo(skb)->destructor_arg = m->msg_control;
>  		skb_shinfo(skb)->tx_flags |= SKBTX_DEV_ZEROCOPY;
>  		skb_shinfo(skb)->tx_flags |= SKBTX_SHARED_FRAG;
> +	} else if (m && m->msg_control) {
> +		struct ubuf_info *uarg = m->msg_control;
> +		uarg->callback(uarg, false);
>  	}
> +
>  	if (vlan) {
>  		skb->dev = vlan->dev;
>  		dev_queue_xmit(skb);
> -- 
> 2.7.4

[toc] | [prev] | [next] | [standalone]

#1533307 — Re: [PATCH net 1/2] tun: handle ubuf refcount correctly when meet erros

On Wed, Nov 30, 2016 at 01:17:51PM +0800, Jason Wang wrote:
> We trigger uarg->callback() immediately after we decide do datacopy
> even if caller want to do zerocopy. This will cause the callback
> (vhost_net_zerocopy_callback) decrease the refcount. But when we meet
> an error afterwards, the error handling in vhost handle_tx() will try
> to decrease it again. This is wrong and fix this by delay the
> uarg->callback() until we're sure there's no errors.
> 
> Reported-by: wangyunjian <wangyunjian@huawei.com>
> Signed-off-by: Jason Wang <jasowang@redhat.com>

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

> ---
> The patch is needed for -stable.
> ---
>  drivers/net/tun.c | 10 ++++------
>  1 file changed, 4 insertions(+), 6 deletions(-)
> 
> diff --git a/drivers/net/tun.c b/drivers/net/tun.c
> index 8093e39..db6acec 100644
> --- a/drivers/net/tun.c
> +++ b/drivers/net/tun.c
> @@ -1246,13 +1246,8 @@ static ssize_t tun_get_user(struct tun_struct *tun, struct tun_file *tfile,
>  
>  	if (zerocopy)
>  		err = zerocopy_sg_from_iter(skb, from);
> -	else {
> +	else
>  		err = skb_copy_datagram_from_iter(skb, 0, from, len);
> -		if (!err && msg_control) {
> -			struct ubuf_info *uarg = msg_control;
> -			uarg->callback(uarg, false);
> -		}
> -	}
>  
>  	if (err) {
>  		this_cpu_inc(tun->pcpu_stats->rx_dropped);
> @@ -1298,6 +1293,9 @@ static ssize_t tun_get_user(struct tun_struct *tun, struct tun_file *tfile,
>  		skb_shinfo(skb)->destructor_arg = msg_control;
>  		skb_shinfo(skb)->tx_flags |= SKBTX_DEV_ZEROCOPY;
>  		skb_shinfo(skb)->tx_flags |= SKBTX_SHARED_FRAG;
> +	} else if (msg_control) {
> +		struct ubuf_info *uarg = msg_control;
> +		uarg->callback(uarg, false);
>  	}
>  
>  	skb_reset_network_header(skb);
> -- 
> 2.7.4

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web