| Message ID | AANLkTinx6Ww4ZwRUTyYD_STPSRsSFck8O=W5yeJSbBs_@mail.gmail.com |
|---|---|
| State | Rejected, archived |
| Delegated to: | David Miller |
| Headers | show |
Le jeudi 18 novembre 2010 à 16:25 +0100, Daniel Turull a écrit : > When a new index is going to be assigned in register_netdevice, > the dev_new_index starts to search possible values from the last index > given to a device although there might be some free ifindex that has been > previously unregistered. This behaviour may create gap(s) in the ifindex list. > > This patch checks for unused values from 1 and gives to the new device the > first available value. This limits the maximum ifindex to a smaller value. > > The ifindex will still be unique since the old value is no longer in use. > > Reported-by: Voravit Tanyingyong <voravit@kth.se> > Signed-off-by: Daniel Turull <daniel.turull@gmail.com> > > --- > diff --git a/net/core/dev.c b/net/core/dev.c > index 381b8e2..a7babab 100644 > --- a/net/core/dev.c > +++ b/net/core/dev.c > @@ -4871,7 +4871,7 @@ int dev_ioctl(struct net *net, unsigned int cmd, > void __user *arg) > */ > static int dev_new_index(struct net *net) > { > - static int ifindex; > + int ifindex; > for (;;) { > if (++ifindex <= 0) > ifindex = 1; > -- NACK Two bugs 1) ifindex is not initialized : you'll be suprised of random values 2) ifindex should not be reused. You'll be surprised so applications can break. SNMP comes to mind. -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Ok, thanks you for you quick response. //Daniel On Thu, Nov 18, 2010 at 16:37, Eric Dumazet <eric.dumazet@gmail.com> wrote: > Le jeudi 18 novembre 2010 à 16:25 +0100, Daniel Turull a écrit : >> When a new index is going to be assigned in register_netdevice, >> the dev_new_index starts to search possible values from the last index >> given to a device although there might be some free ifindex that has been >> previously unregistered. This behaviour may create gap(s) in the ifindex list. >> >> This patch checks for unused values from 1 and gives to the new device the >> first available value. This limits the maximum ifindex to a smaller value. >> >> The ifindex will still be unique since the old value is no longer in use. >> >> Reported-by: Voravit Tanyingyong <voravit@kth.se> >> Signed-off-by: Daniel Turull <daniel.turull@gmail.com> >> >> --- >> diff --git a/net/core/dev.c b/net/core/dev.c >> index 381b8e2..a7babab 100644 >> --- a/net/core/dev.c >> +++ b/net/core/dev.c >> @@ -4871,7 +4871,7 @@ int dev_ioctl(struct net *net, unsigned int cmd, >> void __user *arg) >> */ >> static int dev_new_index(struct net *net) >> { >> - static int ifindex; >> + int ifindex; >> for (;;) { >> if (++ifindex <= 0) >> ifindex = 1; >> -- > > NACK > > Two bugs > > 1) ifindex is not initialized : you'll be suprised of random values > > 2) ifindex should not be reused. You'll be surprised so applications can > break. SNMP comes to mind. > > > > -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
From: Eric Dumazet <eric.dumazet@gmail.com> Date: Thu, 18 Nov 2010 16:37:08 +0100 > Two bugs > > 1) ifindex is not initialized : you'll be suprised of random values > > 2) ifindex should not be reused. You'll be surprised so applications can > break. SNMP comes to mind. Right, the current algorithm is intentionally trying to avoid new devices from using indexes that were used by another device in the past. -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Thu, 18 Nov 2010 08:01:39 -0800 (PST) David Miller <davem@davemloft.net> wrote: > From: Eric Dumazet <eric.dumazet@gmail.com> > Date: Thu, 18 Nov 2010 16:37:08 +0100 > > > Two bugs > > > > 1) ifindex is not initialized : you'll be suprised of random values > > > > 2) ifindex should not be reused. You'll be surprised so applications can > > break. SNMP comes to mind. > > Right, the current algorithm is intentionally trying to avoid new > devices from using indexes that were used by another device in the > past. Also think of the case of adding 10,000 VLAN's. If the search started from beginning on each insertion dev_new_index would go from O(N) to O(N^2) which really hurts. -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/net/core/dev.c b/net/core/dev.c index 381b8e2..a7babab 100644 --- a/net/core/dev.c +++ b/net/core/dev.c @@ -4871,7 +4871,7 @@ int dev_ioctl(struct net *net, unsigned int cmd, void __user *arg) */ static int dev_new_index(struct net *net) { - static int ifindex; + int ifindex; for (;;) { if (++ifindex <= 0)
When a new index is going to be assigned in register_netdevice, the dev_new_index starts to search possible values from the last index given to a device although there might be some free ifindex that has been previously unregistered. This behaviour may create gap(s) in the ifindex list. This patch checks for unused values from 1 and gives to the new device the first available value. This limits the maximum ifindex to a smaller value. The ifindex will still be unique since the old value is no longer in use. Reported-by: Voravit Tanyingyong <voravit@kth.se> Signed-off-by: Daniel Turull <daniel.turull@gmail.com> --- ifindex = 1; -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html