* How to find an unused TAP device with kernel 3.8 ? @ 2013-01-26 10:29 Toralf Förster 2013-01-28 9:38 ` Jason Wang 0 siblings, 1 reply; 7+ messages in thread From: Toralf Förster @ 2013-01-26 10:29 UTC (permalink / raw) To: netdev; +Cc: Linux Kernel With kernel 3.8. I just realized that all pre-defined TAP devices are RUNNING even if no virtual linux system is using it. Now I'm wondering whether this is a new feature of the upcoming kernel, a bug of the old - and how I can avoid that. Background : When I start a user mode linux instance, I currently grep for the next free tap device in this way : Code: for t in $(ifconfig | grep "^tap" | cut -f1 -d:) do ethtool $t | grep -q 'Link detected: no' if [[ $? -eq 0 ]]; then NET="tuntap,$t" break fi done but this doesn't work now anymore. Here is the diff in dmesg $ diff 3.7.1 3.8.0-rc1+ | grep UP < br0: flags=4355<UP,BROADCAST,PROMISC,MULTICAST> mtu 1500 > br0: flags=4419<UP,BROADCAST,RUNNING,PROMISC,MULTICAST> mtu 1500 < tap0: flags=4099<UP,BROADCAST,MULTICAST> mtu 1500 > tap0: flags=4163<UP,BROADCAST,RUNNING,MULTICAST> mtu 1500 more details : http://forums.gentoo.org/viewtopic-p-7211498.html#7211498 -- MfG/Sincerely Toralf Förster pgp finger print: 7B1A 07F4 EC82 0F90 D4C2 8936 872A E508 7DB6 9DA3 ^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: How to find an unused TAP device with kernel 3.8 ? 2013-01-26 10:29 How to find an unused TAP device with kernel 3.8 ? Toralf Förster @ 2013-01-28 9:38 ` Jason Wang 2013-01-28 10:11 ` [PATCH] tun: fix carrier on/off status Michael S. Tsirkin 0 siblings, 1 reply; 7+ messages in thread From: Jason Wang @ 2013-01-28 9:38 UTC (permalink / raw) To: Toralf Förster; +Cc: netdev, Linux Kernel On Saturday, January 26, 2013 11:29:49 AM Toralf Förster wrote: > With kernel 3.8. I just realized that all pre-defined TAP devices are > RUNNING even if no virtual linux system is using it. Now I'm wondering > whether this is a new feature of the upcoming kernel, a bug of the old - > and how I can avoid that. > > Background : When I start a user mode linux instance, I currently grep > for the next free tap device in this way : > > Code: > for t in $(ifconfig | grep "^tap" | cut -f1 -d:) > do > ethtool $t | grep -q 'Link detected: no' > if [[ $? -eq 0 ]]; then > NET="tuntap,$t" > break > fi > done > > but this doesn't work now anymore. > > Here is the diff in dmesg > > $ diff 3.7.1 3.8.0-rc1+ | grep UP > > < br0: flags=4355<UP,BROADCAST,PROMISC,MULTICAST> mtu 1500 > > > br0: flags=4419<UP,BROADCAST,RUNNING,PROMISC,MULTICAST> mtu 1500 > > < tap0: flags=4099<UP,BROADCAST,MULTICAST> mtu 1500 > > > tap0: flags=4163<UP,BROADCAST,RUNNING,MULTICAST> mtu 1500 > > more details : http://forums.gentoo.org/viewtopic-p-7211498.html#7211498 Looks like something wrong with the carrier detection of tap. Please try the following patch to see if it helps. Thanks. diff --git a/drivers/net/tun.c b/drivers/net/tun.c index cc09b67..d8a53c3 100644 --- a/drivers/net/tun.c +++ b/drivers/net/tun.c @@ -438,6 +438,9 @@ static void __tun_detach(struct tun_file *tfile, bool clean) sock_put(&tfile->sk); } + if (tun && tun->numqueues == 0) + netif_carrier_off(tun->dev); + if (clean) { if (tun && tun->numqueues == 0 && tun->numdisabled == 0 && !(tun->flags & TUN_PERSIST)) @@ -473,6 +476,7 @@ static void tun_detach_all(struct net_device *dev) BUG_ON(tun->numqueues != 0); synchronize_net(); + netif_carrier_off(dev); for (i = 0; i < n; i++) { tfile = rtnl_dereference(tun->tfiles[i]); /* Drop read queue */ @@ -531,6 +535,7 @@ static int tun_attach(struct tun_struct *tun, struct file *file) sock_hold(&tfile->sk); tun_set_real_num_queues(tun); + netif_carrier_on(tun->dev); /* device is allowed to go away first, so no need to hold extra * refcnt. -- 1.7.1 ^ permalink raw reply related [flat|nested] 7+ messages in thread
* [PATCH] tun: fix carrier on/off status 2013-01-28 9:38 ` Jason Wang @ 2013-01-28 10:11 ` Michael S. Tsirkin 2013-01-28 10:38 ` [PATCHv2] " Michael S. Tsirkin 0 siblings, 1 reply; 7+ messages in thread From: Michael S. Tsirkin @ 2013-01-28 10:11 UTC (permalink / raw) To: Jason Wang; +Cc: Toralf Förster, netdev, Linux Kernel, David Miller Commit c8d68e6be1c3b242f1c598595830890b65cea64a removed carrier off call from tun_detach since it's now called on queue disable and not only on tun close. This confuses userspace which used this flag to detect a free tun. To fix, move on/off calls to setiff/close respectively. Signed-off-by: Michael S. Tsirkin <mst@redhat.com> --- On Mon, Jan 28, 2013 at 05:38:24PM +0800, Jason Wang wrote: > On Saturday, January 26, 2013 11:29:49 AM Toralf Förster wrote: > > Re: How to find an unused TAP device with kernel 3.8 ? > > With kernel 3.8. I just realized that all pre-defined TAP devices are > > RUNNING even if no virtual linux system is using it. Now I'm wondering > > whether this is a new feature of the upcoming kernel, a bug of the old - > > and how I can avoid that. > > > > Background : When I start a user mode linux instance, I currently grep > > for the next free tap device in this way : > > > > Code: > > for t in $(ifconfig | grep "^tap" | cut -f1 -d:) > > do > > ethtool $t | grep -q 'Link detected: no' > > if [[ $? -eq 0 ]]; then > > NET="tuntap,$t" > > break > > fi > > done > > > > but this doesn't work now anymore. > > > > Here is the diff in dmesg > > > > $ diff 3.7.1 3.8.0-rc1+ | grep UP > > > > < br0: flags=4355<UP,BROADCAST,PROMISC,MULTICAST> mtu 1500 > > > > > br0: flags=4419<UP,BROADCAST,RUNNING,PROMISC,MULTICAST> mtu 1500 > > > > < tap0: flags=4099<UP,BROADCAST,MULTICAST> mtu 1500 > > > > > tap0: flags=4163<UP,BROADCAST,RUNNING,MULTICAST> mtu 1500 > > > > more details : http://forums.gentoo.org/viewtopic-p-7211498.html#7211498 > > Looks like something wrong with the carrier detection of tap. > > Please try the following patch to see if it helps. > > Thanks. > > diff --git a/drivers/net/tun.c b/drivers/net/tun.c > index cc09b67..d8a53c3 100644 > --- a/drivers/net/tun.c > +++ b/drivers/net/tun.c > @@ -438,6 +438,9 @@ static void __tun_detach(struct tun_file *tfile, bool clean) > sock_put(&tfile->sk); > } > > + if (tun && tun->numqueues == 0) > + netif_carrier_off(tun->dev); > + > if (clean) { > if (tun && tun->numqueues == 0 && tun->numdisabled == 0 && > !(tun->flags & TUN_PERSIST)) > @@ -473,6 +476,7 @@ static void tun_detach_all(struct net_device *dev) > BUG_ON(tun->numqueues != 0); > > synchronize_net(); > + netif_carrier_off(dev); > for (i = 0; i < n; i++) { > tfile = rtnl_dereference(tun->tfiles[i]); > /* Drop read queue */ > @@ -531,6 +535,7 @@ static int tun_attach(struct tun_struct *tun, struct file *file) > sock_hold(&tfile->sk); > > tun_set_real_num_queues(tun); > + netif_carrier_on(tun->dev); > > /* device is allowed to go away first, so no need to hold extra > * refcnt. This will set carrier off if all queues are disabled, which I think is a problem as it would confuse the script. How about the following instead: diff --git a/drivers/net/tun.c b/drivers/net/tun.c index af372d0..00dfdad 100644 --- a/drivers/net/tun.c +++ b/drivers/net/tun.c @@ -1644,10 +1644,10 @@ static int tun_set_iff(struct net *net, struct file *file, struct ifreq *ifr) device_create_file(&tun->dev->dev, &dev_attr_owner) || device_create_file(&tun->dev->dev, &dev_attr_group)) pr_err("Failed to create tun sysfs files\n"); - - netif_carrier_on(tun->dev); } + netif_carrier_on(tun->dev); + tun_debug(KERN_INFO, tun, "tun_set_iff\n"); if (ifr->ifr_flags & IFF_NO_PI) @@ -2124,6 +2124,8 @@ static int tun_chr_close(struct inode *inode, struct file *file) struct tun_file *tfile = file->private_data; struct net *net = tfile->net; + netif_carrier_off(tun->dev); + tun_detach(tfile, true); put_net(net); -- MST ^ permalink raw reply related [flat|nested] 7+ messages in thread
* [PATCHv2] tun: fix carrier on/off status 2013-01-28 10:11 ` [PATCH] tun: fix carrier on/off status Michael S. Tsirkin @ 2013-01-28 10:38 ` Michael S. Tsirkin 2013-01-28 11:00 ` Jason Wang ` (2 more replies) 0 siblings, 3 replies; 7+ messages in thread From: Michael S. Tsirkin @ 2013-01-28 10:38 UTC (permalink / raw) To: Jason Wang; +Cc: Toralf Förster, netdev, Linux Kernel, David Miller Commit c8d68e6be1c3b242f1c598595830890b65cea64a removed carrier off call from tun_detach since it's now called on queue disable and not only on tun close. This confuses userspace which used this flag to detect a free tun. To fix, put this back but under if (clean). Signed-off-by: Michael S. Tsirkin <mst@redhat.com> --- Changes from v1: Don't set carrier off unless all queues are closed. Note: didn't test in MQ mode - Jason, care checking this does the right thing there? diff --git a/drivers/net/tun.c b/drivers/net/tun.c index af372d0..06b2723 100644 --- a/drivers/net/tun.c +++ b/drivers/net/tun.c @@ -434,10 +434,13 @@ static void __tun_detach(struct tun_file *tfile, bool clean) } if (clean) { - if (tun && tun->numqueues == 0 && tun->numdisabled == 0 && - !(tun->flags & TUN_PERSIST)) - if (tun->dev->reg_state == NETREG_REGISTERED) + if (tun && tun->numqueues == 0 && tun->numdisabled == 0) { + netif_carrier_off(tun->dev); + + if (!(tun->flags & TUN_PERSIST) && + tun->dev->reg_state == NETREG_REGISTERED) unregister_netdevice(tun->dev); + } BUG_ON(!test_bit(SOCK_EXTERNALLY_ALLOCATED, &tfile->socket.flags)); @@ -1644,10 +1647,10 @@ static int tun_set_iff(struct net *net, struct file *file, struct ifreq *ifr) device_create_file(&tun->dev->dev, &dev_attr_owner) || device_create_file(&tun->dev->dev, &dev_attr_group)) pr_err("Failed to create tun sysfs files\n"); - - netif_carrier_on(tun->dev); } + netif_carrier_on(tun->dev); + tun_debug(KERN_INFO, tun, "tun_set_iff\n"); if (ifr->ifr_flags & IFF_NO_PI) ^ permalink raw reply related [flat|nested] 7+ messages in thread
* Re: [PATCHv2] tun: fix carrier on/off status 2013-01-28 10:38 ` [PATCHv2] " Michael S. Tsirkin @ 2013-01-28 11:00 ` Jason Wang 2013-01-28 19:38 ` Toralf Förster 2013-01-29 20:45 ` David Miller 2 siblings, 0 replies; 7+ messages in thread From: Jason Wang @ 2013-01-28 11:00 UTC (permalink / raw) To: Michael S. Tsirkin Cc: Toralf Förster, netdev, Linux Kernel, David Miller On 01/28/2013 06:38 PM, Michael S. Tsirkin wrote: > Commit c8d68e6be1c3b242f1c598595830890b65cea64a removed carrier off call > from tun_detach since it's now called on queue disable and not only on > tun close. This confuses userspace which used this flag to detect a > free tun. To fix, put this back but under if (clean). > > Signed-off-by: Michael S. Tsirkin <mst@redhat.com> > > --- > > Changes from v1: > Don't set carrier off unless all queues are closed. > > Note: didn't test in MQ mode - Jason, care checking this does the > right thing there? Tested-by: Jason Wang <jasowang@redhat.com> Acked-by: Jason Wang <jasowang@redhat.com> > > diff --git a/drivers/net/tun.c b/drivers/net/tun.c > index af372d0..06b2723 100644 > --- a/drivers/net/tun.c > +++ b/drivers/net/tun.c > @@ -434,10 +434,13 @@ static void __tun_detach(struct tun_file *tfile, bool clean) > } > > if (clean) { > - if (tun && tun->numqueues == 0 && tun->numdisabled == 0 && > - !(tun->flags & TUN_PERSIST)) > - if (tun->dev->reg_state == NETREG_REGISTERED) > + if (tun && tun->numqueues == 0 && tun->numdisabled == 0) { > + netif_carrier_off(tun->dev); > + > + if (!(tun->flags & TUN_PERSIST) && > + tun->dev->reg_state == NETREG_REGISTERED) > unregister_netdevice(tun->dev); > + } > > BUG_ON(!test_bit(SOCK_EXTERNALLY_ALLOCATED, > &tfile->socket.flags)); > @@ -1644,10 +1647,10 @@ static int tun_set_iff(struct net *net, struct file *file, struct ifreq *ifr) > device_create_file(&tun->dev->dev, &dev_attr_owner) || > device_create_file(&tun->dev->dev, &dev_attr_group)) > pr_err("Failed to create tun sysfs files\n"); > - > - netif_carrier_on(tun->dev); > } > > + netif_carrier_on(tun->dev); > + > tun_debug(KERN_INFO, tun, "tun_set_iff\n"); > > if (ifr->ifr_flags & IFF_NO_PI) > -- > To unsubscribe from this list: send the line "unsubscribe linux-kernel" in > the body of a message to majordomo@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html > Please read the FAQ at http://www.tux.org/lkml/ ^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCHv2] tun: fix carrier on/off status 2013-01-28 10:38 ` [PATCHv2] " Michael S. Tsirkin 2013-01-28 11:00 ` Jason Wang @ 2013-01-28 19:38 ` Toralf Förster 2013-01-29 20:45 ` David Miller 2 siblings, 0 replies; 7+ messages in thread From: Toralf Förster @ 2013-01-28 19:38 UTC (permalink / raw) To: Michael S. Tsirkin; +Cc: Jason Wang, netdev, Linux Kernel, David Miller Tested-by: Toralf Förster <toralf.foerster@gmx.de> On 01/28/2013 11:38 AM, Michael S. Tsirkin wrote: > Commit c8d68e6be1c3b242f1c598595830890b65cea64a removed carrier off call > from tun_detach since it's now called on queue disable and not only on > tun close. This confuses userspace which used this flag to detect a > free tun. To fix, put this back but under if (clean). > > Signed-off-by: Michael S. Tsirkin <mst@redhat.com> > > --- > > Changes from v1: > Don't set carrier off unless all queues are closed. > > Note: didn't test in MQ mode - Jason, care checking this does the > right thing there? > > diff --git a/drivers/net/tun.c b/drivers/net/tun.c > index af372d0..06b2723 100644 > --- a/drivers/net/tun.c > +++ b/drivers/net/tun.c > @@ -434,10 +434,13 @@ static void __tun_detach(struct tun_file *tfile, bool clean) > } > > if (clean) { > - if (tun && tun->numqueues == 0 && tun->numdisabled == 0 && > - !(tun->flags & TUN_PERSIST)) > - if (tun->dev->reg_state == NETREG_REGISTERED) > + if (tun && tun->numqueues == 0 && tun->numdisabled == 0) { > + netif_carrier_off(tun->dev); > + > + if (!(tun->flags & TUN_PERSIST) && > + tun->dev->reg_state == NETREG_REGISTERED) > unregister_netdevice(tun->dev); > + } > > BUG_ON(!test_bit(SOCK_EXTERNALLY_ALLOCATED, > &tfile->socket.flags)); > @@ -1644,10 +1647,10 @@ static int tun_set_iff(struct net *net, struct file *file, struct ifreq *ifr) > device_create_file(&tun->dev->dev, &dev_attr_owner) || > device_create_file(&tun->dev->dev, &dev_attr_group)) > pr_err("Failed to create tun sysfs files\n"); > - > - netif_carrier_on(tun->dev); > } > > + netif_carrier_on(tun->dev); > + > tun_debug(KERN_INFO, tun, "tun_set_iff\n"); > > if (ifr->ifr_flags & IFF_NO_PI) > -- MfG/Sincerely Toralf Förster pgp finger print: 7B1A 07F4 EC82 0F90 D4C2 8936 872A E508 7DB6 9DA3 ^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCHv2] tun: fix carrier on/off status 2013-01-28 10:38 ` [PATCHv2] " Michael S. Tsirkin 2013-01-28 11:00 ` Jason Wang 2013-01-28 19:38 ` Toralf Förster @ 2013-01-29 20:45 ` David Miller 2 siblings, 0 replies; 7+ messages in thread From: David Miller @ 2013-01-29 20:45 UTC (permalink / raw) To: mst; +Cc: jasowang, toralf.foerster, netdev, linux-kernel From: "Michael S. Tsirkin" <mst@redhat.com> Date: Mon, 28 Jan 2013 12:38:02 +0200 > Commit c8d68e6be1c3b242f1c598595830890b65cea64a removed carrier off call > from tun_detach since it's now called on queue disable and not only on > tun close. This confuses userspace which used this flag to detect a > free tun. To fix, put this back but under if (clean). > > Signed-off-by: Michael S. Tsirkin <mst@redhat.com> Applied. ^ permalink raw reply [flat|nested] 7+ messages in thread
end of thread, other threads:[~2013-01-29 20:45 UTC | newest] Thread overview: 7+ messages (download: mbox.gz / follow: Atom feed) -- links below jump to the message on this page -- 2013-01-26 10:29 How to find an unused TAP device with kernel 3.8 ? Toralf Förster 2013-01-28 9:38 ` Jason Wang 2013-01-28 10:11 ` [PATCH] tun: fix carrier on/off status Michael S. Tsirkin 2013-01-28 10:38 ` [PATCHv2] " Michael S. Tsirkin 2013-01-28 11:00 ` Jason Wang 2013-01-28 19:38 ` Toralf Förster 2013-01-29 20:45 ` David Miller
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for NNTP newsgroup(s).