Message ID | 20110516131757.ED57E13A6A@rere.qmqm.pl |
---|---|
State | Accepted, archived |
Delegated to: | David Miller |
Headers | show |
From: Michał Mirosław <mirq-linux@rere.qmqm.pl> Date: Mon, 16 May 2011 15:17:57 +0200 (CEST) > Those reduced to DEBUG can possibly be triggered by unprivileged processes > and are nothing exceptional. Illegal checksum combinations can only be > caused by driver bug, so promote those messages to WARN. > > Since GSO without SG will now only cause DEBUG message from > netdev_fix_features(), remove the workaround from register_netdevice(). > > Signed-off-by: Michał Mirosław <mirq-linux@rere.qmqm.pl> Applied, thanks. -- 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 Mon, May 16, 2011 at 03:14:34PM -0400, David Miller wrote: > From: Michał Mirosław <mirq-linux@rere.qmqm.pl> > Date: Mon, 16 May 2011 15:17:57 +0200 (CEST) > > > Those reduced to DEBUG can possibly be triggered by unprivileged processes > > and are nothing exceptional. Illegal checksum combinations can only be > > caused by driver bug, so promote those messages to WARN. > > > > Since GSO without SG will now only cause DEBUG message from > > netdev_fix_features(), remove the workaround from register_netdevice(). > > > > Signed-off-by: Michał Mirosław <mirq-linux@rere.qmqm.pl> > > Applied, thanks. I think this patch is not a good idea. The only problem here is the tun driver which should just fix up the feature bits silently (we need to do that anyway as right now it doesn't verify the feature bits at all, those warnings are only trigger on registering the tun device). With this patch future buggy drivers will instead trigger hard- to-debug warnings further down the stack in the TSO code path, like they have before (or silent packet drops if those warnings aren't there anymore). Cheers,
On Tue, May 17, 2011 at 09:48:16AM +1000, Herbert Xu wrote: > On Mon, May 16, 2011 at 03:14:34PM -0400, David Miller wrote: > > From: Michał Mirosław <mirq-linux@rere.qmqm.pl> > > Date: Mon, 16 May 2011 15:17:57 +0200 (CEST) > > > > > Those reduced to DEBUG can possibly be triggered by unprivileged processes > > > and are nothing exceptional. Illegal checksum combinations can only be > > > caused by driver bug, so promote those messages to WARN. > > > > > > Since GSO without SG will now only cause DEBUG message from > > > netdev_fix_features(), remove the workaround from register_netdevice(). > > > > > > Signed-off-by: Michał Mirosław <mirq-linux@rere.qmqm.pl> > > Applied, thanks. > I think this patch is not a good idea. The only problem here > is the tun driver which should just fix up the feature bits > silently (we need to do that anyway as right now it doesn't > verify the feature bits at all, those warnings are only trigger > on registering the tun device). > > With this patch future buggy drivers will instead trigger hard- > to-debug warnings further down the stack in the TSO code path, > like they have before (or silent packet drops if those warnings > aren't there anymore). If core TSO code really depends on checksumming features being turned on then this dependency should be checked in netdev_fix_features. There will be no possibility for a driver to cause bugs then. Network core is making up for devices which can't checksum just before calling ndo_hard_start_xmit. It's a possibility that packet to be TSOed will be unnecessarily checksummed in dev_hard_start_xmit() if HW checksum is off while TSO is on (I deduce this based on the code, can't test as TSO in my test-box is busted). Best Regards, Michał Mirosław -- 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 9200944..b624fe4 100644 --- a/net/core/dev.c +++ b/net/core/dev.c @@ -5186,27 +5186,27 @@ u32 netdev_fix_features(struct net_device *dev, u32 features) /* Fix illegal checksum combinations */ if ((features & NETIF_F_HW_CSUM) && (features & (NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM))) { - netdev_info(dev, "mixed HW and IP checksum settings.\n"); + netdev_warn(dev, "mixed HW and IP checksum settings.\n"); features &= ~(NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM); } if ((features & NETIF_F_NO_CSUM) && (features & (NETIF_F_HW_CSUM|NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM))) { - netdev_info(dev, "mixed no checksumming and other settings.\n"); + netdev_warn(dev, "mixed no checksumming and other settings.\n"); features &= ~(NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM|NETIF_F_HW_CSUM); } /* Fix illegal SG+CSUM combinations. */ if ((features & NETIF_F_SG) && !(features & NETIF_F_ALL_CSUM)) { - netdev_info(dev, - "Dropping NETIF_F_SG since no checksum feature.\n"); + netdev_dbg(dev, + "Dropping NETIF_F_SG since no checksum feature.\n"); features &= ~NETIF_F_SG; } /* TSO requires that SG is present as well. */ if ((features & NETIF_F_ALL_TSO) && !(features & NETIF_F_SG)) { - netdev_info(dev, "Dropping TSO features since no SG feature.\n"); + netdev_dbg(dev, "Dropping TSO features since no SG feature.\n"); features &= ~NETIF_F_ALL_TSO; } @@ -5216,7 +5216,7 @@ u32 netdev_fix_features(struct net_device *dev, u32 features) /* Software GSO depends on SG. */ if ((features & NETIF_F_GSO) && !(features & NETIF_F_SG)) { - netdev_info(dev, "Dropping NETIF_F_GSO since no SG feature.\n"); + netdev_dbg(dev, "Dropping NETIF_F_GSO since no SG feature.\n"); features &= ~NETIF_F_GSO; } @@ -5226,13 +5226,13 @@ u32 netdev_fix_features(struct net_device *dev, u32 features) if (!((features & NETIF_F_GEN_CSUM) || (features & (NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM)) == (NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM))) { - netdev_info(dev, + netdev_dbg(dev, "Dropping NETIF_F_UFO since no checksum offload features.\n"); features &= ~NETIF_F_UFO; } if (!(features & NETIF_F_SG)) { - netdev_info(dev, + netdev_dbg(dev, "Dropping NETIF_F_UFO since no NETIF_F_SG feature.\n"); features &= ~NETIF_F_UFO; } @@ -5414,12 +5414,6 @@ int register_netdevice(struct net_device *dev) dev->features |= NETIF_F_SOFT_FEATURES; dev->wanted_features = dev->features & dev->hw_features; - /* Avoid warning from netdev_fix_features() for GSO without SG */ - if (!(dev->wanted_features & NETIF_F_SG)) { - dev->wanted_features &= ~NETIF_F_GSO; - dev->features &= ~NETIF_F_GSO; - } - /* Enable GRO and NETIF_F_HIGHDMA for vlans by default, * vlan_dev_init() will do the dev->features check, so these features * are enabled only if supported by underlying device.
Those reduced to DEBUG can possibly be triggered by unprivileged processes and are nothing exceptional. Illegal checksum combinations can only be caused by driver bug, so promote those messages to WARN. Since GSO without SG will now only cause DEBUG message from netdev_fix_features(), remove the workaround from register_netdevice(). Signed-off-by: Michał Mirosław <mirq-linux@rere.qmqm.pl> --- This is for net tree, not net-next. net/core/dev.c | 22 ++++++++-------------- 1 files changed, 8 insertions(+), 14 deletions(-) -- 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