[net-next PATCH v2 0/2] A couple of minor clean-ups and optimizations

[net-next PATCH v2 0/2] A couple of minor clean-ups and optimizations

[Alexander Duyck]

This patch series is basically just a v2 of a couple patches I recently
submitted.

The two patches aren't technically related but there are just items I found
while cleaning up and prepping some further work to enable Tx checksums for
tunnels.

---

Alexander Duyck (2):
      gro: Defer clearing of flush bit in tunnel paths
      csum: Update csum_block_add to use rotate instead of byteswap


 drivers/net/geneve.c   |    7 ++-----
 drivers/net/vxlan.c    |    3 +--
 include/net/checksum.h |   12 ++++++------
 net/ipv4/fou.c         |    3 +--
 net/ipv4/gre_offload.c |    3 +--
 5 files changed, 11 insertions(+), 17 deletions(-)

[net-next PATCH v2 1/2] gro: Defer clearing of flush bit in tunnel paths

[Alexander Duyck]

This patch updates the GRO handlers for GRE, VXLAN, GENEVE, and FOU so that
we do not clear the flush bit until after we have called the next level GRO
handler.  Previously this was being cleared before parsing through the list
of frames, however this resulted in several paths where either the bit
needed to be reset but wasn't as in the case of FOU, or cases where it was
being set as in GENEVE.  By just deferring the clearing of the bit until
after the next level protocol has been parsed we can avoid any unnecessary
bit twiddling and avoid bugs.

Signed-off-by: Alexander Duyck <aduyck@mirantis.com>
---

v2: This was originally code for checking the flush_id in tunnels but since
    that was rejected I thought it best to at least submit the bit that can
    fix actual issues in the code.

 drivers/net/geneve.c   |    7 ++-----
 drivers/net/vxlan.c    |    3 +--
 net/ipv4/fou.c         |    3 +--
 net/ipv4/gre_offload.c |    3 +--
 4 files changed, 5 insertions(+), 11 deletions(-)

diff --git a/drivers/net/geneve.c b/drivers/net/geneve.c
index 6a0cbbe03e5d..aebb4a9289c1 100644
--- a/drivers/net/geneve.c
+++ b/drivers/net/geneve.c
@@ -462,8 +462,6 @@ static struct sk_buff **geneve_gro_receive(struct sk_buff **head,
 			goto out;
 	}
 
-	flush = 0;
-
 	for (p = *head; p; p = p->next) {
 		if (!NAPI_GRO_CB(p)->same_flow)
 			continue;
@@ -480,14 +478,13 @@ static struct sk_buff **geneve_gro_receive(struct sk_buff **head,
 
 	rcu_read_lock();
 	ptype = gro_find_receive_by_type(type);
-	if (!ptype) {
-		flush = 1;
+	if (!ptype)
 		goto out_unlock;
-	}
 
 	skb_gro_pull(skb, gh_len);
 	skb_gro_postpull_rcsum(skb, gh, gh_len);
 	pp = ptype->callbacks.gro_receive(head, skb);
+	flush = 0;
 
 out_unlock:
 	rcu_read_unlock();
diff --git a/drivers/net/vxlan.c b/drivers/net/vxlan.c
index 2399099e68cf..63929877619f 100644
--- a/drivers/net/vxlan.c
+++ b/drivers/net/vxlan.c
@@ -591,8 +591,6 @@ static struct sk_buff **vxlan_gro_receive(struct sk_buff **head,
 
 	skb_gro_pull(skb, sizeof(struct vxlanhdr)); /* pull vxlan header */
 
-	flush = 0;
-
 	for (p = *head; p; p = p->next) {
 		if (!NAPI_GRO_CB(p)->same_flow)
 			continue;
@@ -606,6 +604,7 @@ static struct sk_buff **vxlan_gro_receive(struct sk_buff **head,
 	}
 
 	pp = eth_gro_receive(head, skb);
+	flush = 0;
 
 out:
 	skb_gro_remcsum_cleanup(skb, &grc);
diff --git a/net/ipv4/fou.c b/net/ipv4/fou.c
index 88dab0c1670c..780484243e14 100644
--- a/net/ipv4/fou.c
+++ b/net/ipv4/fou.c
@@ -319,8 +319,6 @@ static struct sk_buff **gue_gro_receive(struct sk_buff **head,
 
 	skb_gro_pull(skb, hdrlen);
 
-	flush = 0;
-
 	for (p = *head; p; p = p->next) {
 		const struct guehdr *guehdr2;
 
@@ -352,6 +350,7 @@ static struct sk_buff **gue_gro_receive(struct sk_buff **head,
 		goto out_unlock;
 
 	pp = ops->callbacks.gro_receive(head, skb);
+	flush = 0;
 
 out_unlock:
 	rcu_read_unlock();
diff --git a/net/ipv4/gre_offload.c b/net/ipv4/gre_offload.c
index 47f4c544c916..540866dbd27d 100644
--- a/net/ipv4/gre_offload.c
+++ b/net/ipv4/gre_offload.c
@@ -175,8 +175,6 @@ static struct sk_buff **gre_gro_receive(struct sk_buff **head,
 					     null_compute_pseudo);
 	}
 
-	flush = 0;
-
 	for (p = *head; p; p = p->next) {
 		const struct gre_base_hdr *greh2;
 
@@ -213,6 +211,7 @@ static struct sk_buff **gre_gro_receive(struct sk_buff **head,
 	skb_gro_postpull_rcsum(skb, greh, grehlen);
 
 	pp = ptype->callbacks.gro_receive(head, skb);
+	flush = 0;
 
 out_unlock:
 	rcu_read_unlock();

[net-next PATCH v2 2/2] csum: Update csum_block_add to use rotate instead of byteswap

[Alexander Duyck]

The code for csum_block_add was doing a funky byteswap to swap the even and
odd bytes of the checksum if the offset was odd.  Instead of doing this we
can save ourselves some trouble and just shift by 8 as this should have the
same effect in terms of the final checksum value and only requires one
instruction.

In addition we can update csum_block_sub to just use csum_block_add with a
inverse value for csum2.  This way we follow the same code path as
csum_block_add without having to duplicate it.

Signed-off-by: Alexander Duyck <aduyck@mirantis.com>
---

v2:  Replaced (x << 24) + (x >> 8) with ror32(x, 8) and added comment.

 include/net/checksum.h |   12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)

diff --git a/include/net/checksum.h b/include/net/checksum.h
index abffc64e7300..5c30891e84e5 100644
--- a/include/net/checksum.h
+++ b/include/net/checksum.h
@@ -88,8 +88,11 @@ static inline __wsum
 csum_block_add(__wsum csum, __wsum csum2, int offset)
 {
 	u32 sum = (__force u32)csum2;
-	if (offset&1)
-		sum = ((sum&0xFF00FF)<<8)+((sum>>8)&0xFF00FF);
+
+	/* rotate sum to align it with a 16b boundary */
+	if (offset & 1)
+		sum = ror32(sum, 8);
+
 	return csum_add(csum, (__force __wsum)sum);
 }
 
@@ -102,10 +105,7 @@ csum_block_add_ext(__wsum csum, __wsum csum2, int offset, int len)
 static inline __wsum
 csum_block_sub(__wsum csum, __wsum csum2, int offset)
 {
-	u32 sum = (__force u32)csum2;
-	if (offset&1)
-		sum = ((sum&0xFF00FF)<<8)+((sum>>8)&0xFF00FF);
-	return csum_sub(csum, (__force __wsum)sum);
+	return csum_block_add(csum, ~csum2, offset);
 }
 
 static inline __wsum csum_unfold(__sum16 n)

Re: [net-next PATCH v2 0/2] A couple of minor clean-ups and optimizations

[David Miller]

From: Alexander Duyck <aduyck@mirantis.com>
Date: Wed, 09 Mar 2016 09:22:48 -0800

> This patch series is basically just a v2 of a couple patches I recently
> submitted.
> 
> The two patches aren't technically related but there are just items I found
> while cleaning up and prepping some further work to enable Tx checksums for
> tunnels.

Series applied, thanks Alex.