* Re: [PATCH net] sunvnet: set queue mapping when doing packet copies
[not found] <54C90AED.7040404@oracle.com>
@ 2015-01-28 16:48 ` Sowmini Varadhan
2015-01-29 16:46 ` David L Stevens
2015-01-30 17:29 ` David L Stevens
2 siblings, 0 replies; 6+ messages in thread
From: Sowmini Varadhan @ 2015-01-28 16:48 UTC (permalink / raw)
To: David L Stevens; +Cc: David Miller, netdev
On (01/28/15 11:14), David L Stevens wrote:
>
> This patch fixes a bug where vnet_skb_shape() didn't set the already-selected
> queue mapping when a packet copy was required. This results in using the
> wrong queue index for stops/starts, hung tx queues and watchdog timeouts
> under heavy load.
Acked-by: Sowmini Varadhan <sowmini.varadhan@oracle.com>
^ permalink raw reply [flat|nested] 6+ messages in thread
* [PATCH net] sunvnet: set queue mapping when doing packet copies
[not found] <54C90AED.7040404@oracle.com>
2015-01-28 16:48 ` [PATCH net] sunvnet: set queue mapping when doing packet copies Sowmini Varadhan
@ 2015-01-29 16:46 ` David L Stevens
2015-01-29 18:12 ` Eric Dumazet
2015-01-30 17:29 ` David L Stevens
2 siblings, 1 reply; 6+ messages in thread
From: David L Stevens @ 2015-01-29 16:46 UTC (permalink / raw)
To: David Miller; +Cc: netdev, Sowmini Varadhan
[resending this since it never appeared on netdev yesterday]
This patch fixes a bug where vnet_skb_shape() didn't set the already-selected
queue mapping when a packet copy was required. This results in using the
wrong queue index for stops/starts, hung tx queues and watchdog timeouts
under heavy load.
Signed-off-by: David L Stevens <david.stevens@oracle.com>
Acked-by: Sowmini Varadhan <sowmini.varadhan@oracle.com>
---
drivers/net/ethernet/sun/sunvnet.c | 1 +
1 files changed, 1 insertions(+), 0 deletions(-)
diff --git a/drivers/net/ethernet/sun/sunvnet.c b/drivers/net/ethernet/sun/sunvnet.c
index 2b719cc..2b10b85 100644
--- a/drivers/net/ethernet/sun/sunvnet.c
+++ b/drivers/net/ethernet/sun/sunvnet.c
@@ -1123,6 +1123,7 @@ static inline struct sk_buff *vnet_skb_shape(struct sk_buff *skb, int ncookies)
skb_shinfo(nskb)->gso_size = skb_shinfo(skb)->gso_size;
skb_shinfo(nskb)->gso_type = skb_shinfo(skb)->gso_type;
}
+ nskb->queue_mapping = skb->queue_mapping;
dev_kfree_skb(skb);
skb = nskb;
}
^ permalink raw reply related [flat|nested] 6+ messages in thread
* Re: [PATCH net] sunvnet: set queue mapping when doing packet copies
2015-01-29 16:46 ` David L Stevens
@ 2015-01-29 18:12 ` Eric Dumazet
2015-01-29 18:19 ` Eric Dumazet
0 siblings, 1 reply; 6+ messages in thread
From: Eric Dumazet @ 2015-01-29 18:12 UTC (permalink / raw)
To: David L Stevens; +Cc: David Miller, netdev, Sowmini Varadhan
On Thu, 2015-01-29 at 11:46 -0500, David L Stevens wrote:
> [resending this since it never appeared on netdev yesterday]
>
> This patch fixes a bug where vnet_skb_shape() didn't set the already-selected
> queue mapping when a packet copy was required. This results in using the
> wrong queue index for stops/starts, hung tx queues and watchdog timeouts
> under heavy load.
>
> Signed-off-by: David L Stevens <david.stevens@oracle.com>
> Acked-by: Sowmini Varadhan <sowmini.varadhan@oracle.com>
> ---
> drivers/net/ethernet/sun/sunvnet.c | 1 +
> 1 files changed, 1 insertions(+), 0 deletions(-)
>
> diff --git a/drivers/net/ethernet/sun/sunvnet.c b/drivers/net/ethernet/sun/sunvnet.c
> index 2b719cc..2b10b85 100644
> --- a/drivers/net/ethernet/sun/sunvnet.c
> +++ b/drivers/net/ethernet/sun/sunvnet.c
> @@ -1123,6 +1123,7 @@ static inline struct sk_buff *vnet_skb_shape(struct sk_buff *skb, int ncookies)
> skb_shinfo(nskb)->gso_size = skb_shinfo(skb)->gso_size;
> skb_shinfo(nskb)->gso_type = skb_shinfo(skb)->gso_type;
> }
> + nskb->queue_mapping = skb->queue_mapping;
> dev_kfree_skb(skb);
> skb = nskb;
> }
Note that vnet_skb_shape() also drops the skb -> sk assocation.
-> That removes flow control, a single socket can fill TX ring buffer.
TCP Small queue is also disabled.
I would try add before the dev_kfree_skb(skb) :
swap(nskb->sk, skb->sk);
swap(nskb->truesize, skb->truesize);
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH net] sunvnet: set queue mapping when doing packet copies
2015-01-29 18:12 ` Eric Dumazet
@ 2015-01-29 18:19 ` Eric Dumazet
0 siblings, 0 replies; 6+ messages in thread
From: Eric Dumazet @ 2015-01-29 18:19 UTC (permalink / raw)
To: David L Stevens; +Cc: David Miller, netdev, Sowmini Varadhan
On Thu, 2015-01-29 at 10:12 -0800, Eric Dumazet wrote:
> swap(nskb->sk, skb->sk);
> swap(nskb->truesize, skb->truesize);
and swap(nskb->destructor, skb->destructor);
(check skb_segment() for a similar path)
^ permalink raw reply [flat|nested] 6+ messages in thread
* [PATCH net] sunvnet: set queue mapping when doing packet copies
[not found] <54C90AED.7040404@oracle.com>
2015-01-28 16:48 ` [PATCH net] sunvnet: set queue mapping when doing packet copies Sowmini Varadhan
2015-01-29 16:46 ` David L Stevens
@ 2015-01-30 17:29 ` David L Stevens
2015-02-03 2:21 ` David Miller
2 siblings, 1 reply; 6+ messages in thread
From: David L Stevens @ 2015-01-30 17:29 UTC (permalink / raw)
To: David Miller; +Cc: netdev, Sowmini Varadhan, Eric Dumazet
This patch fixes a bug where vnet_skb_shape() didn't set the already-selected
queue mapping when a packet copy was required. This results in using the
wrong queue index for stops/starts, hung tx queues and watchdog timeouts
under heavy load.
Signed-off-by: David L Stevens <david.stevens@oracle.com>
Acked-by: Sowmini Varadhan <sowmini.varadhan@oracle.com>
--
I'm resubmitting this in its original form. Since skb_segment() doesn't do
TCP flow control either, due to an incorrect destructor check, and those
issues raised by Eric are a matter of fairness, while this missing queue_mapping
assignment results in crashes, I'd like to get this piece fixed separately.
---
drivers/net/ethernet/sun/sunvnet.c | 1 +
1 files changed, 1 insertions(+), 0 deletions(-)
diff --git a/drivers/net/ethernet/sun/sunvnet.c b/drivers/net/ethernet/sun/sunvnet.c
index 2b719cc..2b10b85 100644
--- a/drivers/net/ethernet/sun/sunvnet.c
+++ b/drivers/net/ethernet/sun/sunvnet.c
@@ -1123,6 +1123,7 @@ static inline struct sk_buff *vnet_skb_shape(struct sk_buff *skb, int ncookies)
skb_shinfo(nskb)->gso_size = skb_shinfo(skb)->gso_size;
skb_shinfo(nskb)->gso_type = skb_shinfo(skb)->gso_type;
}
+ nskb->queue_mapping = skb->queue_mapping;
dev_kfree_skb(skb);
skb = nskb;
}
--
1.7.1
^ permalink raw reply related [flat|nested] 6+ messages in thread
* Re: [PATCH net] sunvnet: set queue mapping when doing packet copies
2015-01-30 17:29 ` David L Stevens
@ 2015-02-03 2:21 ` David Miller
0 siblings, 0 replies; 6+ messages in thread
From: David Miller @ 2015-02-03 2:21 UTC (permalink / raw)
To: david.stevens; +Cc: netdev, sowmini.varadhan, eric.dumazet
From: David L Stevens <david.stevens@oracle.com>
Date: Fri, 30 Jan 2015 12:29:45 -0500
>
>
> This patch fixes a bug where vnet_skb_shape() didn't set the already-selected
> queue mapping when a packet copy was required. This results in using the
> wrong queue index for stops/starts, hung tx queues and watchdog timeouts
> under heavy load.
>
> Signed-off-by: David L Stevens <david.stevens@oracle.com>
> Acked-by: Sowmini Varadhan <sowmini.varadhan@oracle.com>
>
> --
> I'm resubmitting this in its original form. Since skb_segment() doesn't do
> TCP flow control either, due to an incorrect destructor check, and those
> issues raised by Eric are a matter of fairness, while this missing queue_mapping
> assignment results in crashes, I'd like to get this piece fixed separately.
Fair enough, applied.
Please sort out the callback issues with Eric, thanks.
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2015-02-03 2:21 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
[not found] <54C90AED.7040404@oracle.com>
2015-01-28 16:48 ` [PATCH net] sunvnet: set queue mapping when doing packet copies Sowmini Varadhan
2015-01-29 16:46 ` David L Stevens
2015-01-29 18:12 ` Eric Dumazet
2015-01-29 18:19 ` Eric Dumazet
2015-01-30 17:29 ` David L Stevens
2015-02-03 2:21 ` David Miller
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.