[net-next PATCH0/12] qlge: More features and cleanup.

[net-next PATCH0/12] qlge: More features and cleanup.

[Ron Mercer]

Attached are more qlge patches for adding features such as multi TXQ and
LRO.
I've also included 4 bug fixes, 3 of which I will also post to net-2.6
in a separate series tomorrow.

Thanks,
Ron Mercer

Re: [net-next PATCH0/12] qlge: More features and cleanup.

[David Miller]

From: Ron Mercer <ron.mercer@qlogic.com>
Date: Mon,  9 Mar 2009 13:59:16 -0700

> Attached are more qlge patches for adding features such as multi TXQ and
> LRO.
> I've also included 4 bug fixes, 3 of which I will also post to net-2.6
> in a separate series tomorrow.

All applied, thanks Ron, although one thing I think you should
get rid of:

> @@ -3354,6 +3351,7 @@ static int ql_configure_rings(struct ql_adapter *qdev)
>  	 * completion handler rx_rings.
>  	 */
>  	qdev->rx_ring_count = qdev->tx_ring_count + qdev->rss_ring_count + 1;
> +	netif_set_gso_max_size(qdev->ndev, 65536);
>  
>  	for (i = 0; i < qdev->tx_ring_count; i++) {
>  		tx_ring = &qdev->tx_ring[i];

65536 (defined by GSO_MAX_SIZE) is the default and it's really a good
idea to only use this override interface on chips that actaully have
restrictions and can't use the default.

I also don't see what this change had to do with adding TX
multiqueue support :-)  It wasn't mentioned in the commit log
message in any way either.  Please don't slip small things in
like this.

Re: [net-next PATCH0/12] qlge: More features and cleanup.

[Ron Mercer]

>> tatic int ql_configure_rings(struct ql_adapter *qdev)
>>  	 * completion handler rx_rings.
>>  	 */
>>  	qdev->rx_ring_count = qdev->tx_ring_count + qdev->rss_ring_count
>>  	+ 1;
>> +	netif_set_gso_max_size(qdev->ndev, 65536);
>>  
>>  	for (i = 0; i < qdev->tx_ring_count; i++) {
>>  		tx_ring = &qdev->tx_ring[i];
>
>65536 (defined by GSO_MAX_SIZE) is the default and it's really a good
>idea to only use this override interface on chips that actaully have
>restrictions and can't use the default.
>
>I also don't see what this change had to do with adding TX
>multiqueue support :-)  It wasn't mentioned in the commit log
>message in any way either.  Please don't slip small things in
>like this.

Dave,

Thanks for clarifying this for me.  I'll drop this line in my next
series. I should have sent it as a separate (rejectable) patch.
I'll be more careful about that.

Ron