Re: [net-next RFC PATCH 4/7] tuntap: multiqueue support
flat view
From: Eric Dumazet <hidden>
Date: 2011-08-12 14:29:41
Also in:
kvm, lkml, qemu-devel
Le vendredi 12 août 2011 à 09:55 +0800, Jason Wang a écrit :
quoted hunk ↗ jump to hunk
+ rxq = skb_get_rxhash(skb); + if (rxq) { + tfile = rcu_dereference(tun->tfiles[rxq % numqueues]); + if (tfile) + goto out; + }
You can avoid an expensive divide with following trick : u32 idx = ((u64)rxq * numqueues) >> 32;
quoted hunk ↗ jump to hunk
-static struct tun_struct *tun_get(struct file *file) +static void tun_detach_all(struct net_device *dev) { - return __tun_get(file->private_data); + struct tun_struct *tun = netdev_priv(dev); + struct tun_file *tfile, *tfile_list[MAX_TAP_QUEUES]; + int i, j = 0; + + spin_lock(&tun_lock); + + for (i = 0; i < MAX_TAP_QUEUES && tun->numqueues; i++) { + tfile = rcu_dereference_protected(tun->tfiles[i], + lockdep_is_held(&tun_lock)); + if (tfile) { + wake_up_all(&tfile->wq.wait); + tfile_list[i++] = tfile;
typo here, you want tfile_list[j++] = tfile;
quoted hunk ↗ jump to hunk
+ rcu_assign_pointer(tun->tfiles[i], NULL); + rcu_assign_pointer(tfile->tun, NULL); + --tun->numqueues; + } + } + BUG_ON(tun->numqueues != 0); + spin_unlock(&tun_lock); + + synchronize_rcu(); + for(--j; j >= 0; j--) + sock_put(&tfile_list[j]->sk); }
Could you take a look at net/packet/af_packet.c, to check how David did the whole fanout thing ? __fanout_unlink() Trick is to not leave NULL entries in the tun->tfiles[] array. It makes things easier in hot path.