[PATCH v3 net-next 1/2] net: Avoid unnessary loop when master_idx is invalid in inet_select_addr

[PATCH v3 net-next 1/2] net: Avoid unnessary loop when master_idx is invalid in inet_select_addr

[fgao]

From: Gao Feng <fgao@ikuai8.com>

When master_idx is invalid, it is zero. It is unnecessary to iterate
all netdevs. Because l3mdev_master_ifindex_rcu(dev) != master_idx must
be true.
Now put this loop into the condition block when master_idx is valid.

Signed-off-by: Gao Feng <fgao@ikuai8.com>
---
 v3: Add the cover letter, per David
 v2: Correct the comit log and remove useless braces, per Sergei
 v1: Initial Version

 net/ipv4/devinet.c | 68 +++++++++++++++++++++++++++++-------------------------
 1 file changed, 36 insertions(+), 32 deletions(-)

diff --git a/net/ipv4/devinet.c b/net/ipv4/devinet.c
index 5d367b7..1a9e550 100644
--- a/net/ipv4/devinet.c
+++ b/net/ipv4/devinet.c
@@ -1219,42 +1219,46 @@ __be32 inet_select_addr(const struct net_device *dev, __be32 dst, int scope)
 no_in_dev:
 	master_idx = l3mdev_master_ifindex_rcu(dev);
 
-	/* For VRFs, the VRF device takes the place of the loopback device,
-	 * with addresses on it being preferred.  Note in such cases the
-	 * loopback device will be among the devices that fail the master_idx
-	 * equality check in the loop below.
-	 */
-	if (master_idx &&
-	    (dev = dev_get_by_index_rcu(net, master_idx)) &&
-	    (in_dev = __in_dev_get_rcu(dev))) {
-		for_primary_ifa(in_dev) {
-			if (ifa->ifa_scope != RT_SCOPE_LINK &&
-			    ifa->ifa_scope <= scope) {
-				addr = ifa->ifa_local;
-				goto out_unlock;
+	if (master_idx) {
+		/* For VRFs, the VRF device takes the place of the loopback device,
+		 * with addresses on it being preferred.  Note in such cases the
+		 * loopback device will be among the devices that fail the master_idx
+		 * equality check in the loop below.
+		 */
+		dev = dev_get_by_index_rcu(net, master_idx);
+		if (dev) {
+			in_dev = __in_dev_get_rcu(dev);
+			if (in_dev) {
+				for_primary_ifa(in_dev) {
+					if (ifa->ifa_scope != RT_SCOPE_LINK &&
+					    ifa->ifa_scope <= scope) {
+						addr = ifa->ifa_local;
+						goto out_unlock;
+					}
+				} endfor_ifa(in_dev);
 			}
-		} endfor_ifa(in_dev);
-	}
+		}
 
-	/* Not loopback addresses on loopback should be preferred
-	   in this case. It is important that lo is the first interface
-	   in dev_base list.
-	 */
-	for_each_netdev_rcu(net, dev) {
-		if (l3mdev_master_ifindex_rcu(dev) != master_idx)
-			continue;
+		/* Not loopback addresses on loopback should be preferred
+		   in this case. It is important that lo is the first interface
+		   in dev_base list.
+		 */
+		for_each_netdev_rcu(net, dev) {
+			if (l3mdev_master_ifindex_rcu(dev) != master_idx)
+				continue;
 
-		in_dev = __in_dev_get_rcu(dev);
-		if (!in_dev)
-			continue;
+			in_dev = __in_dev_get_rcu(dev);
+			if (!in_dev)
+				continue;
 
-		for_primary_ifa(in_dev) {
-			if (ifa->ifa_scope != RT_SCOPE_LINK &&
-			    ifa->ifa_scope <= scope) {
-				addr = ifa->ifa_local;
-				goto out_unlock;
-			}
-		} endfor_ifa(in_dev);
+			for_primary_ifa(in_dev) {
+				if (ifa->ifa_scope != RT_SCOPE_LINK &&
+				    ifa->ifa_scope <= scope) {
+					addr = ifa->ifa_local;
+					goto out_unlock;
+				}
+			} endfor_ifa(in_dev);
+		}
 	}
 out_unlock:
 	rcu_read_unlock();
-- 
1.9.1

Re: [PATCH v3 net-next 1/2] net: Avoid unnessary loop when master_idx is invalid in inet_select_addr

[David Ahern]

On 3/9/17 7:52 PM, fgao@ikuai8.com wrote:
> From: Gao Feng <fgao@ikuai8.com>
> 
> When master_idx is invalid, it is zero. It is unnecessary to iterate
> all netdevs. Because l3mdev_master_ifindex_rcu(dev) != master_idx must
> be true.
> Now put this loop into the condition block when master_idx is valid.

you are significantly changing how this loop works, not just the
master_idx == 0 case. Basically, if dev does not have an address, you
are not going to look at the other interfaces.

The
   if (l3mdev_master_ifindex_rcu(dev) != master_idx)

is actually relevant.

If dev is enslaved to an L3 device, only consider devices in that
domain. Conversely, if dev is NOT enslaved to an L3 device, do NOT
consider devices that are enslaved to one. Both of those are equally
important.

Re: [PATCH v3 net-next 1/2] net: Avoid unnessary loop when master_idx is invalid in inet_select_addr

[Feng Gao]

Hi David,

On Fri, Mar 10, 2017 at 11:58 AM, David Ahern <dsa@cumulusnetworks.com> wrote:
> On 3/9/17 7:52 PM, fgao@ikuai8.com wrote:
>> From: Gao Feng <fgao@ikuai8.com>
>>
>> When master_idx is invalid, it is zero. It is unnecessary to iterate
>> all netdevs. Because l3mdev_master_ifindex_rcu(dev) != master_idx must
>> be true.
>> Now put this loop into the condition block when master_idx is valid.
>
> you are significantly changing how this loop works, not just the
> master_idx == 0 case. Basically, if dev does not have an address, you
> are not going to look at the other interfaces.
>
> The
>    if (l3mdev_master_ifindex_rcu(dev) != master_idx)
>
> is actually relevant.
>
> If dev is enslaved to an L3 device, only consider devices in that
> domain. Conversely, if dev is NOT enslaved to an L3 device, do NOT
> consider devices that are enslaved to one. Both of those are equally
> important.

Thanks. I didn't consider about the case that dev is enslaved to an l3
device you pointed.

Regards
Feng