From mboxrd@z Thu Jan  1 00:00:00 1970
From: Eric Dumazet <dada1@cosmosbay.com>
Subject: Re: [PATCH v2] Receive Packet Steering
Date: Mon, 04 May 2009 08:10:56 +0200
Message-ID: <49FE86F0.4070206@cosmosbay.com>
References: <65634d660905032103h614225dbg9911e290f5537fbf@mail.gmail.com> <49FE7D63.6050102@cosmosbay.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: QUOTED-PRINTABLE
Cc: netdev@vger.kernel.org, David Miller <davem@davemloft.net>
To: Tom Herbert <therbert@google.com>
Return-path: <netdev-owner@vger.kernel.org>
Received: from gw1.cosmosbay.com ([212.99.114.194]:44618 "EHLO
	gw1.cosmosbay.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org
	with ESMTP id S1751860AbZEDGLQ convert rfc822-to-8bit (ORCPT
	<rfc822;netdev@vger.kernel.org>); Mon, 4 May 2009 02:11:16 -0400
In-Reply-To: <49FE7D63.6050102@cosmosbay.com>
Sender: netdev-owner@vger.kernel.org
List-ID: <netdev.vger.kernel.org>

Eric Dumazet a =E9crit :
> Tom Herbert a =E9crit :
>> This is an update of the receive packet steering patch (RPS) based o=
n received
>> comments (thanks for all the comments).  Improvements are:
>>
>> 1) Removed config option for the feature.
>> 2) Made scheduling of backlog NAPI devices between CPUs lockless and=
 much
>> simpler.
>> 3) Added new softirq to do defer sending IPIs for coalescing.
>> 4) Imported hash from simple_rx_hash.  Eliminates modulo operation t=
o convert
>> hash to index.
>> 5) If no cpu is found for packet steering, then netif_receive_skb pr=
ocesses
>> packet inline as before without queueing.  In paritcular if RPS is n=
ot
>> configured on a device the receive path is unchanged from current fo=
r
>> NAPI devices (one additional conditional).
>>
>> Tom
>=20

another error :

>> diff --git a/net/core/net-sysfs.c b/net/core/net-sysfs.c
>> index 2da59a0..073a407 100644
>> --- a/net/core/net-sysfs.c
>> +++ b/net/core/net-sysfs.c
>> @@ -211,6 +211,63 @@ static ssize_t store_tx_queue_len(struct device=
 *dev,
>>  	return netdev_store(dev, attr, buf, len, change_tx_queue_len);
>>  }
>>
>> +static ssize_t store_rps_cpus(struct device *dev,
>> +			      struct device_attribute *attr,
>> +			      const char *buf, size_t len)
>> +{
>> +	struct net_device *net =3D to_net_dev(dev);
>> +	cpumask_t mask;
>> +	int err, cpu;
>> +	int i =3D 0;
>> +
>> +	if (!capable(CAP_NET_ADMIN))
>> +		return -EPERM;
>> +
>> +	err =3D bitmap_parse(buf, len, cpumask_bits(&mask), nr_cpumask_bit=
s);
>> +	if (err)
>> +		return err;
>> +
>> +	rtnl_lock();
>> +	if (dev_isalive(net)) {
>> +		if (!net->rps_map) {
>> +			net->rps_map =3D kzalloc(sizeof(u16) *
>> +			    num_possible_cpus(), GFP_KERNEL);
>=20
> num_possible_cpus() is not the max index of a cpu, but the number of =
possible cpus.
> it can be for example 2, but cpu0 being index 0, and second cpu at in=
dex 511
>=20
> So I believe you want nr_cpu_ids here
>=20
> (num_possible_cpus() <=3D nr_cpu_ids), not necessarly equal.=20
>=20
>=20
>> +			if (!net->rps_map)

lacks a rtnl_unlock() here, before return

>> +				return -ENOMEM;
>> +		}
>> +		cpus_and(mask, mask, cpu_online_map);
>> +		for_each_cpu_mask(cpu, mask)
>> +			net->rps_map[i++] =3D  cpu;
>> +		net->rps_map_len =3D i;
>> +	}
>> +	rtnl_unlock();
>> +
>> +	return len;
>> +}
>> +