[net PATCH] net: Fix Tx hash bound checking

[net PATCH] net: Fix Tx hash bound checking

[Amritha Nambiar]

Fixes: commit 1b837d489e06 ("net: Revoke export for __skb_tx_hash, update it to just be static skb_tx_hash")
Fixes: commit eadec877ce9c ("net: Add support for subordinate traffic classes to netdev_pick_tx")

Fixes the lower and upper bounds when there are multiple TCs and
traffic is on the the same TC on the same device.

The lower bound is represented by 'qoffset' and the upper limit for
hash value is 'qcount + qoffset'. This gives a clean Rx to Tx queue
mapping when there are multiple TCs, as the queue indices for upper TCs
will be offset by 'qoffset'.

Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com>
Reviewed-by: Alexander Duyck <alexander.h.duyck@linux.intel.com>
Reviewed-by: Sridhar Samudrala <sridhar.samudrala@intel.com>
---
 net/core/dev.c |    2 ++
 1 file changed, 2 insertions(+)

diff --git a/net/core/dev.c b/net/core/dev.c
index e10bd680dc03..c6c985fe7b1b 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -3076,6 +3076,8 @@ static u16 skb_tx_hash(const struct net_device *dev,
 
 	if (skb_rx_queue_recorded(skb)) {
 		hash = skb_get_rx_queue(skb);
+		if (hash >= qoffset)
+			hash -= qoffset;
 		while (unlikely(hash >= qcount))
 			hash -= qcount;
 		return hash + qoffset;

Re: [net PATCH] net: Fix Tx hash bound checking

[Sergei Shtylyov]

Hello!

On 22.02.2020 5:09, Amritha Nambiar wrote:

> Fixes: commit 1b837d489e06 ("net: Revoke export for __skb_tx_hash, update it to just be static skb_tx_hash")
> Fixes: commit eadec877ce9c ("net: Add support for subordinate traffic classes to netdev_pick_tx")

    No need to say "commit" here. And All tags should be together (below the 
patch description).

> Fixes the lower and upper bounds when there are multiple TCs and
> traffic is on the the same TC on the same device.
> 
> The lower bound is represented by 'qoffset' and the upper limit for
> hash value is 'qcount + qoffset'. This gives a clean Rx to Tx queue
> mapping when there are multiple TCs, as the queue indices for upper TCs
> will be offset by 'qoffset'.
> 
> Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com>
> Reviewed-by: Alexander Duyck <alexander.h.duyck@linux.intel.com>
> Reviewed-by: Sridhar Samudrala <sridhar.samudrala@intel.com>
[...]

MBR, Sergei

Re: [net PATCH] net: Fix Tx hash bound checking

[Nambiar, Amritha]

On 2/22/2020 12:07 AM, Sergei Shtylyov wrote:
> Hello!
> 
> On 22.02.2020 5:09, Amritha Nambiar wrote:
> 
>> Fixes: commit 1b837d489e06 ("net: Revoke export for __skb_tx_hash, update it to just be static skb_tx_hash")
>> Fixes: commit eadec877ce9c ("net: Add support for subordinate traffic classes to netdev_pick_tx")
> 
>     No need to say "commit" here. And All tags should be together (below the 
> patch description).
> 

Sure. Will send v2 with these changes.

-Amritha

>> Fixes the lower and upper bounds when there are multiple TCs and
>> traffic is on the the same TC on the same device.
>>
>> The lower bound is represented by 'qoffset' and the upper limit for
>> hash value is 'qcount + qoffset'. This gives a clean Rx to Tx queue
>> mapping when there are multiple TCs, as the queue indices for upper TCs
>> will be offset by 'qoffset'.
>>
>> Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com>
>> Reviewed-by: Alexander Duyck <alexander.h.duyck@linux.intel.com>
>> Reviewed-by: Sridhar Samudrala <sridhar.samudrala@intel.com>
> [...]
> 
> MBR, Sergei
>