* Re: [E1000-devel] [PATCH 0/4] i40e: Neatening and object size reductions
From: Joe Perches @ 2013-09-04 1:30 UTC (permalink / raw)
To: Nelson, Shannon
Cc: Kirsher, Jeffrey T, e1000-devel@lists.sourceforge.net,
netdev@vger.kernel.org, linux-kernel@vger.kernel.org
In-Reply-To: <FC41C24E35F18A40888AACA1A36F3E416C61C0E5@FMSMSX102.amr.corp.intel.com>
On Wed, 2013-09-04 at 01:00 +0000, Nelson, Shannon wrote:
Hi Shannon.
> > -----Original Message----- > From: Joe Perches
> [mailto:joe@perches.com] > Sent: Friday, August 30, 2013 4:06 PM > >
> Just some potential cleanings...
>
> > i40e: Whitespace cleaning
>
> Hmmm, we hadn't noticed the new experimental "--fix" option before.
> There are a lot of good suggestions there, but obviously it needs a lot
> of reading and tweaking before it can be used. There are cases here
> where function call parameters are adjusted to line up with the opening
> '(' but that pushes the parameter(s) beyond 80 columns - we're trying
> to stay within the 80 column line and checkpatch clean. Also, there
> are several where the first continued parameter line indent is changed
> but the next line or two are not.
>
> We'll spend time going through these and try to take care of what makes
> sense.
Swell. All these are your choice to fix as you want.
Exceeding 80 columns doesn't bother me much.
Keeping alignment appropriate for multi-line statements
needs work inside checkpatch. I played with it a bit
but it's unfortunately complicated by intermixed
insertions and deletions.
> > i40e: Add and use pf_<level>
>
> We had considered this kind of macro awhile ago, but nixed it for a few
> different reasons, but primarily because it seems like
> yet-another-print-macro and not necessarily worth the effort.
>
> > i40e: pf_<level> remove "%s: " ... __func__
>
> We're beginning to remove many of the __func__ uses, so these prints
> are no longer all doing the __func__ thing. We originally had them
> there for early development and debugging and are currently removing
> them from the normal path messages.
Fine by me. I think __func__ is nearly always pretty
useless myself.
> > i40e: Convert pf_<level> macros to functions
>
> Doesn't this create a problem with polluting the kernel namespace?
> These don't apply to any other driver. I suppose we could lessen the
> namespace problem with i40e_ prefix, but I'm still not sold on it. I
> suspect we can still get much of the text savings replacing the
> __func__ with __builtin_return_address(0) where needed, and remove them
> where no longer needed. Does that work for you?
I think you could just as soon whatever combinations of the
other standard logging mechanisms without using pf_<level>
wiphy_<level>
netif_<level>
netdev_<level>
dev_<level>
pr_<level>
as appropriate. I did that only because there was ~10K
of what I think of as not too useful function names out
of a defconfig size of 140k.
> > i40e: Fix 32 bit shift compilation warnings
>
> Sure.
I think you should use the kernel.h standard macros
for lower_32_bits and upper_32_bits instead.
cheers, Joe
^ permalink raw reply
* RE: [E1000-devel] [PATCH 0/4] i40e: Neatening and object size reductions
From: Nelson, Shannon @ 2013-09-04 1:00 UTC (permalink / raw)
To: Joe Perches, Kirsher, Jeffrey T
Cc: e1000-devel@lists.sourceforge.net, netdev@vger.kernel.org,
linux-kernel@vger.kernel.org
In-Reply-To: <cover.1377903831.git.joe@perches.com>
> -----Original Message-----
> From: Joe Perches [mailto:joe@perches.com]
> Sent: Friday, August 30, 2013 4:06 PM
>
> Just some potential cleanings...
> i40e: Whitespace cleaning
Hmmm, we hadn't noticed the new experimental "--fix" option before. There are a lot of good suggestions there, but obviously it needs a lot of reading and tweaking before it can be used. There are cases here where function call parameters are adjusted to line up with the opening '(' but that pushes the parameter(s) beyond 80 columns - we're trying to stay within the 80 column line and checkpatch clean. Also, there are several where the first continued parameter line indent is changed but the next line or two are not.
We'll spend time going through these and try to take care of what makes sense.
> i40e: Add and use pf_<level>
We had considered this kind of macro awhile ago, but nixed it for a few different reasons, but primarily because it seems like yet-another-print-macro and not necessarily worth the effort.
> i40e: pf_<level> remove "%s: " ... __func__
We're beginning to remove many of the __func__ uses, so these prints are no longer all doing the __func__ thing. We originally had them there for early development and debugging and are currently removing them from the normal path messages.
> i40e: Convert pf_<level> macros to functions
Doesn't this create a problem with polluting the kernel namespace? These don't apply to any other driver. I suppose we could lessen the namespace problem with i40e_ prefix, but I'm still not sold on it. I suspect we can still get much of the text savings replacing the __func__ with __builtin_return_address(0) where needed, and remove them where no longer needed. Does that work for you?
> i40e: Fix 32 bit shift compilation warnings
Sure.
sln
^ permalink raw reply
* RE: [net-next v3 1/8] i40e: main driver core
From: Nelson, Shannon @ 2013-09-04 0:59 UTC (permalink / raw)
To: Joe Perches, Kirsher, Jeffrey T
Cc: davem@davemloft.net, Brandeburg, Jesse, netdev@vger.kernel.org,
gospo@redhat.com, sassmann@redhat.com, Waskiewicz Jr, Peter P,
e1000-devel@lists.sourceforge.net
In-Reply-To: <1377880368.2070.26.camel@joe-AO722>
> -----Original Message-----
> From: Joe Perches [mailto:joe@perches.com]
> Sent: Friday, August 30, 2013 9:33 AM
>
Hi Joe, thanks for your comments, and we're working through another version to be posted Real Soon. In the meantime I thought I'd respond to a couple specific comments.
> > +#define DRV_KERN "-k"
> > +
> > +#define DRV_VERSION_MAJOR 0
> > +#define DRV_VERSION_MINOR 3
> > +#define DRV_VERSION_BUILD 8
> > +#define DRV_VERSION __stringify(DRV_VERSION_MAJOR) "." \
> > + __stringify(DRV_VERSION_MINOR) "." \
> > + __stringify(DRV_VERSION_BUILD) DRV_KERN
>
> does "-k" add/signify anything useful?
The "-k" helps us to quickly spot that a customer is asking us about the upstream version of our driver and not an out-of-tree (e.g. SourceForge) version.
[...]
>
> > +/**
> > + * i40e_get_netdev_stats_struct - Get statistics for netdev interface
> > + * @netdev: network interface device structure
> > + *
> > + * Returns the address of the device statistics structure.
> > + * The statistics are actually updated from the service task.
> > + **/
> > +static struct rtnl_link_stats64 *i40e_get_netdev_stats_struct(
> > + struct net_device *netdev,
> > + struct rtnl_link_stats64 *storage)
>
> An alternative bsd declaration style might be
> more readable for these very long types and names.
>
> static struct rtnl_link_stats64 *
> i40e_get_netdev_stats_struct(struct net_device *netdev,
> struct rtnl_link_stats64 *storage)
We're trying to stay consistent in kernel function styles, stay within standard checkpatch restrictions, and keep the code readable and maintainable, which all together make for some difficult choices. No matter which way we go on some of these decisions, we run into the painfully gray opinion areas of how someone else might have solved the problems.
sln
^ permalink raw reply
* Re: [PATCH net] net: ipv6: tcp: fix potential use after free in tcp_v6_do_rcv
From: Jean Sacren @ 2013-09-04 0:51 UTC (permalink / raw)
To: Eric Dumazet; +Cc: Daniel Borkmann, davem, netdev, yoshfuji, Jiri Benc
In-Reply-To: <1378247352.7360.50.camel@edumazet-glaptop>
From: Eric Dumazet <eric.dumazet@gmail.com>
Date: Tue, 03 Sep 2013 15:29:12 -0700
>
> On Tue, 2013-09-03 at 15:59 -0600, Jean Sacren wrote:
> > From: Eric Dumazet <eric.dumazet@gmail.com>
> > Date: Tue, 03 Sep 2013 13:35:17 -0700
> > >
> > > How did you get your conclusion ?
> >
> > If one changes one line of code, doesn't one own that line?
> >
>
> 'Owning' like he is the guy who can sell it ? OK I understand now
> so many people send 'cleanups' ;)
'Owning' like he is the guy who is responsible for the change. Whether
you understand it or not, that's your own business.
The idea was by taking one example to determine who is really to blame
for: Is the one passing on the bug responsible? Thanks to Daniel
Borkmann for the answer.
--
Jean Sacren
^ permalink raw reply
* Re: ipv4: warnings on sk_wmem_queued
From: Eric Wong @ 2013-09-04 0:21 UTC (permalink / raw)
To: dormando; +Cc: netdev, linux-kernel
In-Reply-To: <alpine.DEB.2.02.1308311229280.3532@dtop>
dormando <dormando@rydia.net> wrote:
> > I noticed these warnings on stock 3.10.9 running stress tests on
> > cmogstored.git (git://bogomips.org/cmogstored.git) doing standard
> > HTTP server stuff between lo and tmpfs:
> >
> [...]
> > I was going to reboot into 3.10.10 before I looked at dmesg. These
> > warnings happened after ~8 hours of stress tests, and those stress tests
> > are still running.
>
> I had a kernel panic this morning on a production machine, also running
> 3.10.9. I only got a small part of the end of the trace, but it matches:
I haven't managed to hit any kernel panic nor my original warning again
after all these days of constant 20+ GB/s lo traffic. I'll finally
reboot into 3.10.10 now. dmesg has been completely quiet since the warning
and the machine is still running well.
> (also: hi Eric!)
Hi dormando!
^ permalink raw reply
* [PATCH net-next 2/2] bridge: apply multicast snooping to IPv6 link-local, too
From: Linus Lüssing @ 2013-09-04 0:13 UTC (permalink / raw)
To: netdev
Cc: bridge, Stephen Hemminger, David S. Miller, linux-kernel,
Herbert Xu, Cong Wang, Adam Baker, Linus Lüssing
In-Reply-To: <1378253619-23918-1-git-send-email-linus.luessing@web.de>
The multicast snooping code should have matured enough to be safely
applicable to IPv6 link-local multicast addresses (excluding the
link-local all nodes address, ff02::1), too.
Signed-off-by: Linus Lüssing <linus.luessing@web.de>
---
net/bridge/br_mdb.c | 3 ++-
net/bridge/br_multicast.c | 7 ++++---
net/bridge/br_private.h | 10 ----------
3 files changed, 6 insertions(+), 14 deletions(-)
diff --git a/net/bridge/br_mdb.c b/net/bridge/br_mdb.c
index a7c6cd0..85a09bb 100644
--- a/net/bridge/br_mdb.c
+++ b/net/bridge/br_mdb.c
@@ -9,6 +9,7 @@
#include <net/netlink.h>
#if IS_ENABLED(CONFIG_IPV6)
#include <net/ipv6.h>
+#include <net/addrconf.h>
#endif
#include "br_private.h"
@@ -254,7 +255,7 @@ static bool is_valid_mdb_entry(struct br_mdb_entry *entry)
return false;
#if IS_ENABLED(CONFIG_IPV6)
} else if (entry->addr.proto == htons(ETH_P_IPV6)) {
- if (!ipv6_is_transient_multicast(&entry->addr.u.ip6))
+ if (ipv6_addr_is_ll_all_nodes(&entry->addr.u.ip6))
return false;
#endif
} else
diff --git a/net/bridge/br_multicast.c b/net/bridge/br_multicast.c
index 662ba7b..3b0ed99 100644
--- a/net/bridge/br_multicast.c
+++ b/net/bridge/br_multicast.c
@@ -29,6 +29,7 @@
#include <net/ipv6.h>
#include <net/mld.h>
#include <net/ip6_checksum.h>
+#include <net/addrconf.h>
#endif
#include "br_private.h"
@@ -724,7 +725,7 @@ static int br_ip6_multicast_add_group(struct net_bridge *br,
{
struct br_ip br_group;
- if (!ipv6_is_transient_multicast(group))
+ if (ipv6_addr_is_ll_all_nodes(group))
return 0;
br_group.u.ip6 = *group;
@@ -1410,7 +1411,7 @@ static void br_ip6_multicast_leave_group(struct net_bridge *br,
&br->ip6_query;
- if (!ipv6_is_transient_multicast(group))
+ if (ipv6_addr_is_ll_all_nodes(group))
return;
br_group.u.ip6 = *group;
@@ -1551,7 +1552,7 @@ static int br_multicast_ipv6_rcv(struct net_bridge *br,
return 0;
/* Prevent flooding this packet if there is no listener present */
- if (ipv6_is_transient_multicast(&ip6h->daddr))
+ if (!ipv6_addr_is_ll_all_nodes(&ip6h->daddr))
BR_INPUT_SKB_CB(skb)->mrouters_only = 1;
if (ip6h->nexthdr != IPPROTO_HOPOPTS ||
diff --git a/net/bridge/br_private.h b/net/bridge/br_private.h
index f225fb6..598cb0b 100644
--- a/net/bridge/br_private.h
+++ b/net/bridge/br_private.h
@@ -494,16 +494,6 @@ extern void br_mdb_notify(struct net_device *dev, struct net_bridge_port *port,
#define mlock_dereference(X, br) \
rcu_dereference_protected(X, lockdep_is_held(&br->multicast_lock))
-#if IS_ENABLED(CONFIG_IPV6)
-#include <net/addrconf.h>
-static inline int ipv6_is_transient_multicast(const struct in6_addr *addr)
-{
- if (ipv6_addr_is_multicast(addr) && IPV6_ADDR_MC_FLAG_TRANSIENT(addr))
- return 1;
- return 0;
-}
-#endif
-
static inline bool br_multicast_is_router(struct net_bridge *br)
{
return br->multicast_router == 2 ||
--
1.7.10.4
^ permalink raw reply related
* [PATCH net-next 1/2] bridge: prevent flooding IPv6 packets that do not have a listener
From: Linus Lüssing @ 2013-09-04 0:13 UTC (permalink / raw)
To: netdev
Cc: bridge, Stephen Hemminger, David S. Miller, linux-kernel,
Herbert Xu, Cong Wang, Adam Baker, Linus Lüssing
In-Reply-To: <1378253619-23918-1-git-send-email-linus.luessing@web.de>
Currently if there is no listener for a certain group then IPv6 packets
for that group are flooded on all ports, even though there might be no
host and router interested in it on a port.
With this commit they are only forwarded to ports with a multicast
router.
Just like commit bd4265fe36 ("bridge: Only flood unregistered groups
to routers") did for IPv4, let's do the same for IPv6 with the same
reasoning.
Signed-off-by: Linus Lüssing <linus.luessing@web.de>
---
net/bridge/br_multicast.c | 10 ++++++++--
1 file changed, 8 insertions(+), 2 deletions(-)
diff --git a/net/bridge/br_multicast.c b/net/bridge/br_multicast.c
index bbcb435..662ba7b 100644
--- a/net/bridge/br_multicast.c
+++ b/net/bridge/br_multicast.c
@@ -1547,8 +1547,14 @@ static int br_multicast_ipv6_rcv(struct net_bridge *br,
* - MLD has always Router Alert hop-by-hop option
* - But we do not support jumbrograms.
*/
- if (ip6h->version != 6 ||
- ip6h->nexthdr != IPPROTO_HOPOPTS ||
+ if (ip6h->version != 6)
+ return 0;
+
+ /* Prevent flooding this packet if there is no listener present */
+ if (ipv6_is_transient_multicast(&ip6h->daddr))
+ BR_INPUT_SKB_CB(skb)->mrouters_only = 1;
+
+ if (ip6h->nexthdr != IPPROTO_HOPOPTS ||
ip6h->payload_len == 0)
return 0;
--
1.7.10.4
^ permalink raw reply related
* bride: IPv6 multicast snooping enhancements
From: Linus Lüssing @ 2013-09-04 0:13 UTC (permalink / raw)
To: netdev
Cc: Herbert Xu, bridge, linux-kernel, Adam Baker, Stephen Hemminger,
Linus Lüssing, David S. Miller, Cong Wang
Hi,
Here are two, small feature changes I would like to submit to increase
the usefulness of the multicast snooping of the bridge code.
The first patch is an unaltered one I had submitted before, but since it
got no feedback I'm resubmitting it here for net-next. With the recently
added patch to disable snooping if there is no querier (b00589af + 248ba8ec05
+ 8d50af4fb), it should be a safe choice now (without these, patch 1/2 would
have introduced another potential for lost IPv6 multicast packets).
Both conceptually and also with some testing and fuzzing, I couldn't spot
any more causes for potential packet loss. And since the multicast snooping
code has now been tried by various people, I think it should be a safe
choice to apply the multicast snooping not only for IPv6 multicast packets
with a scope greater than link-local, but also for packets of exactly this
scope. The IPv6 standard mandates MLD reports for link-local multicast, too,
so we can safely snoop them as well (in contrast to IPv4 link-local).
Cheers, Linus
^ permalink raw reply
* Re: [PATCH net-next v2 4/8] net: ipv6: mld: implement RFC3810 MLDv2 mode only
From: Hannes Frederic Sowa @ 2013-09-03 23:01 UTC (permalink / raw)
To: Daniel Borkmann; +Cc: davem, netdev
In-Reply-To: <1378246784-21067-5-git-send-email-dborkman@redhat.com>
On Wed, Sep 04, 2013 at 12:19:40AM +0200, Daniel Borkmann wrote:
> RFC3810, 10. Security Considerations says under subsection 10.1.
> Query Message:
>
> A forged Version 1 Query message will put MLDv2 listeners on that
> link in MLDv1 Host Compatibility Mode. This scenario can be avoided
> by providing MLDv2 hosts with a configuration option to ignore
> Version 1 messages completely.
>
> Hence, implement a MLDv2-only mode that will ignore MLDv1 traffic:
>
> echo 2 > /proc/sys/net/ipv6/conf/ethX/force_mld_version or
> echo 2 > /proc/sys/net/ipv6/conf/all/force_mld_version
>
> Note that <all> device has a higher precedence as it was previously
> also the case in the macro MLD_V1_SEEN() that would "short-circuit"
> if condition on <all> case.
>
> Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
> Cc: Hannes Frederic Sowa <hannes@stressinduktion.org>
Even better...
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Thanks!
^ permalink raw reply
* Re: [PATCH v2] ipv6: Don't depend on per socket memory for neighbour discovery messages
From: Hannes Frederic Sowa @ 2013-09-03 22:43 UTC (permalink / raw)
To: Thomas Graf; +Cc: Stephen Warren, Eric Dumazet, davem, netdev, Fabio Estevam
In-Reply-To: <20130903182348.GB4166@casper.infradead.org>
On Tue, Sep 03, 2013 at 07:23:48PM +0100, Thomas Graf wrote:
> I'm confused why lockdep would only trigger after my patch, the
> deadlock is unrelated.
Either we had patch caf92bc ("ipv6: do not call ndisc_send_rs() with write
lock") already applied or both our testing was flawed. I reverted caf92bc and
tested v1 again and could also reproduce this problem.
Because of the sk_mc_loop logic I do think it is better to set skb->sk, so:
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
If this should still go to stable, the above mentioned commit should also hit
the queue.
Greetings,
Hannes
^ permalink raw reply
* [PATCH] sh_eth: fix napi_{en|dis}able() calls racing against interrupts
From: Sergei Shtylyov @ 2013-09-03 22:41 UTC (permalink / raw)
To: netdev; +Cc: nobuhiro.iwamatsu.yj, linux-sh
While implementing NAPI for the driver, I overlooked the race conditions where
interrupt handler might have called napi_schedule_prep() before napi_enable()
was called or after napi_disable() was called. If RX interrupt happens, this
would cause the endless interrupts and messages like:
sh-eth eth0: ignoring interrupt, status 0x00040000, mask 0x01ff009f.
The interrupt wouldn't even be masked by the kernel eventually since the handler
would return IRQ_HANDLED all the time.
As a fix, move napi_enable() call before request_irq() call and napi_disable()
call after free_irq() call.
Signed-off-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>
---
The patch is against Dave's 'net.git' repo. It should be applied to 3.11.y too.
drivers/net/ethernet/renesas/sh_eth.c | 12 +++++++-----
1 file changed, 7 insertions(+), 5 deletions(-)
Index: net/drivers/net/ethernet/renesas/sh_eth.c
===================================================================
--- net.orig/drivers/net/ethernet/renesas/sh_eth.c
+++ net/drivers/net/ethernet/renesas/sh_eth.c
@@ -1857,11 +1857,13 @@ static int sh_eth_open(struct net_device
pm_runtime_get_sync(&mdp->pdev->dev);
+ napi_enable(&mdp->napi);
+
ret = request_irq(ndev->irq, sh_eth_interrupt,
mdp->cd->irq_flags, ndev->name, ndev);
if (ret) {
dev_err(&ndev->dev, "Can not assign IRQ number\n");
- return ret;
+ goto out_napi_off;
}
/* Descriptor set */
@@ -1879,12 +1881,12 @@ static int sh_eth_open(struct net_device
if (ret)
goto out_free_irq;
- napi_enable(&mdp->napi);
-
return ret;
out_free_irq:
free_irq(ndev->irq, ndev);
+out_napi_off:
+ napi_disable(&mdp->napi);
pm_runtime_put_sync(&mdp->pdev->dev);
return ret;
}
@@ -1976,8 +1978,6 @@ static int sh_eth_close(struct net_devic
{
struct sh_eth_private *mdp = netdev_priv(ndev);
- napi_disable(&mdp->napi);
-
netif_stop_queue(ndev);
/* Disable interrupts by clearing the interrupt mask. */
@@ -1995,6 +1995,8 @@ static int sh_eth_close(struct net_devic
free_irq(ndev->irq, ndev);
+ napi_disable(&mdp->napi);
+
/* Free all the skbuffs in the Rx queue. */
sh_eth_ring_free(ndev);
^ permalink raw reply
* Re: [PATCH net] net: ipv6: tcp: fix potential use after free in tcp_v6_do_rcv
From: Eric Dumazet @ 2013-09-03 22:29 UTC (permalink / raw)
To: Jean Sacren; +Cc: Daniel Borkmann, davem, netdev, yoshfuji, Jiri Benc
In-Reply-To: <20130903215915.GF8262@mail.gmail.com>
On Tue, 2013-09-03 at 15:59 -0600, Jean Sacren wrote:
> From: Eric Dumazet <eric.dumazet@gmail.com>
> Date: Tue, 03 Sep 2013 13:35:17 -0700
> >
> > How did you get your conclusion ?
>
> If one changes one line of code, doesn't one own that line?
>
'Owning' like he is the guy who can sell it ? OK I understand now
so many people send 'cleanups' ;)
Someone adding a space is not responsible for the real bug.
You'll have to be very careful to do proper attributions.
This is _critical_ for proper stable submissions.
In this case we want the fix to reach linux-3.4 and further versions.
The commit you pointed out was for linux-3.9, and added no bug.
^ permalink raw reply
* Re: [PATCH net] net: ipv6: tcp: fix potential use after free in tcp_v6_do_rcv
From: Daniel Borkmann @ 2013-09-03 22:25 UTC (permalink / raw)
To: Jean Sacren; +Cc: Eric Dumazet, davem, netdev, yoshfuji, Jiri Benc
In-Reply-To: <20130903215915.GF8262@mail.gmail.com>
On 09/03/2013 11:59 PM, Jean Sacren wrote:
> From: Eric Dumazet <eric.dumazet@gmail.com>
> Date: Tue, 03 Sep 2013 13:35:17 -0700
>>
>> How did you get your conclusion ?
>
> If one changes one line of code, doesn't one own that line?
The question is not about who currently "owns" the line, but who
first *introduced* an error.
^ permalink raw reply
* [PATCH net-next v2 7/8] net: ipv6: mld: introduce mld_{gq,ifc,dad}_stop_timer functions
From: Daniel Borkmann @ 2013-09-03 22:19 UTC (permalink / raw)
To: davem; +Cc: netdev, Hannes Frederic Sowa
In-Reply-To: <1378246784-21067-1-git-send-email-dborkman@redhat.com>
We already have mld_{gq,ifc,dad}_start_timer() functions, so introduce
mld_{gq,ifc,dad}_stop_timer() functions to reduce code size and make it
more readable.
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Cc: Hannes Frederic Sowa <hannes@stressinduktion.org>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
---
net/ipv6/mcast.c | 41 +++++++++++++++++++++++++----------------
1 file changed, 25 insertions(+), 16 deletions(-)
diff --git a/net/ipv6/mcast.c b/net/ipv6/mcast.c
index 3be4fe0..ffae869 100644
--- a/net/ipv6/mcast.c
+++ b/net/ipv6/mcast.c
@@ -1006,6 +1006,13 @@ static void mld_gq_start_timer(struct inet6_dev *idev)
in6_dev_hold(idev);
}
+static void mld_gq_stop_timer(struct inet6_dev *idev)
+{
+ idev->mc_gq_running = 0;
+ if (del_timer(&idev->mc_gq_timer))
+ __in6_dev_put(idev);
+}
+
static void mld_ifc_start_timer(struct inet6_dev *idev, unsigned long delay)
{
unsigned long tv = net_random() % delay;
@@ -1014,6 +1021,13 @@ static void mld_ifc_start_timer(struct inet6_dev *idev, unsigned long delay)
in6_dev_hold(idev);
}
+static void mld_ifc_stop_timer(struct inet6_dev *idev)
+{
+ idev->mc_ifc_count = 0;
+ if (del_timer(&idev->mc_ifc_timer))
+ __in6_dev_put(idev);
+}
+
static void mld_dad_start_timer(struct inet6_dev *idev, unsigned long delay)
{
unsigned long tv = net_random() % delay;
@@ -1022,6 +1036,12 @@ static void mld_dad_start_timer(struct inet6_dev *idev, unsigned long delay)
in6_dev_hold(idev);
}
+static void mld_dad_stop_timer(struct inet6_dev *idev)
+{
+ if (del_timer(&idev->mc_dad_timer))
+ __in6_dev_put(idev);
+}
+
/*
* IGMP handling (alias multicast ICMPv6 messages)
*/
@@ -1234,15 +1254,9 @@ static int mld_process_v1(struct inet6_dev *idev, struct mld_msg *mld,
mld_set_v1_mode(idev);
/* cancel MLDv2 report timer */
- idev->mc_gq_running = 0;
- if (del_timer(&idev->mc_gq_timer))
- __in6_dev_put(idev);
-
+ mld_gq_stop_timer(idev);
/* cancel the interface change timer */
- idev->mc_ifc_count = 0;
- if (del_timer(&idev->mc_ifc_timer))
- __in6_dev_put(idev);
-
+ mld_ifc_stop_timer(idev);
/* clear deleted report items */
mld_clear_delrec(idev);
@@ -2434,14 +2448,9 @@ void ipv6_mc_down(struct inet6_dev *idev)
/* Withdraw multicast list */
read_lock_bh(&idev->lock);
- idev->mc_ifc_count = 0;
- if (del_timer(&idev->mc_ifc_timer))
- __in6_dev_put(idev);
- idev->mc_gq_running = 0;
- if (del_timer(&idev->mc_gq_timer))
- __in6_dev_put(idev);
- if (del_timer(&idev->mc_dad_timer))
- __in6_dev_put(idev);
+ mld_ifc_stop_timer(idev);
+ mld_gq_stop_timer(idev);
+ mld_dad_stop_timer(idev);
for (i = idev->mc_list; i; i=i->next)
igmp6_group_dropped(i);
--
1.7.11.7
^ permalink raw reply related
* [PATCH net-next v2 8/8] net: ipv6: mld: document force_mld_version in ip-sysctl.txt
From: Daniel Borkmann @ 2013-09-03 22:19 UTC (permalink / raw)
To: davem; +Cc: netdev, Hannes Frederic Sowa
In-Reply-To: <1378246784-21067-1-git-send-email-dborkman@redhat.com>
Document force_mld_version parameter in ip-sysctl.txt.
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Cc: Hannes Frederic Sowa <hannes@stressinduktion.org>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
---
Documentation/networking/ip-sysctl.txt | 5 +++++
1 file changed, 5 insertions(+)
diff --git a/Documentation/networking/ip-sysctl.txt b/Documentation/networking/ip-sysctl.txt
index 1cb3aeb..a46d785 100644
--- a/Documentation/networking/ip-sysctl.txt
+++ b/Documentation/networking/ip-sysctl.txt
@@ -1358,6 +1358,11 @@ mldv2_unsolicited_report_interval - INTEGER
MLDv2 report retransmit will take place.
Default: 1000 (1 second)
+force_mld_version - INTEGER
+ 0 - (default) No enforcement of a MLD version, MLDv1 fallback allowed
+ 1 - Enforce to use MLD version 1
+ 2 - Enforce to use MLD version 2
+
suppress_frag_ndisc - INTEGER
Control RFC 6980 (Security Implications of IPv6 Fragmentation
with IPv6 Neighbor Discovery) behavior:
--
1.7.11.7
^ permalink raw reply related
* [PATCH net-next v2 6/8] net: ipv6: mld: refactor query processing into v1/v2 functions
From: Daniel Borkmann @ 2013-09-03 22:19 UTC (permalink / raw)
To: davem; +Cc: netdev, Hannes Frederic Sowa
In-Reply-To: <1378246784-21067-1-git-send-email-dborkman@redhat.com>
Make igmp6_event_query() a bit easier to read by refactoring code
parts into mld_process_v1() and mld_process_v2().
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Cc: Hannes Frederic Sowa <hannes@stressinduktion.org>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
---
net/ipv6/mcast.c | 89 +++++++++++++++++++++++++++++++++++---------------------
1 file changed, 56 insertions(+), 33 deletions(-)
diff --git a/net/ipv6/mcast.c b/net/ipv6/mcast.c
index d7e7c58..3be4fe0 100644
--- a/net/ipv6/mcast.c
+++ b/net/ipv6/mcast.c
@@ -1218,6 +1218,55 @@ static void mld_update_qri(struct inet6_dev *idev,
idev->mc_qri = msecs_to_jiffies(mldv2_mrc(mlh2));
}
+static int mld_process_v1(struct inet6_dev *idev, struct mld_msg *mld,
+ unsigned long *max_delay)
+{
+ unsigned long mldv1_md;
+
+ /* Ignore v1 queries */
+ if (mld_in_v2_mode_only(idev))
+ return -EINVAL;
+
+ /* MLDv1 router present */
+ mldv1_md = ntohs(mld->mld_maxdelay);
+ *max_delay = max(msecs_to_jiffies(mldv1_md), 1UL);
+
+ mld_set_v1_mode(idev);
+
+ /* cancel MLDv2 report timer */
+ idev->mc_gq_running = 0;
+ if (del_timer(&idev->mc_gq_timer))
+ __in6_dev_put(idev);
+
+ /* cancel the interface change timer */
+ idev->mc_ifc_count = 0;
+ if (del_timer(&idev->mc_ifc_timer))
+ __in6_dev_put(idev);
+
+ /* clear deleted report items */
+ mld_clear_delrec(idev);
+
+ return 0;
+}
+
+static int mld_process_v2(struct inet6_dev *idev, struct mld2_query *mld,
+ unsigned long *max_delay)
+{
+ /* hosts need to stay in MLDv1 mode, discard MLDv2 queries */
+ if (mld_in_v1_mode(idev))
+ return -EINVAL;
+
+ *max_delay = max(msecs_to_jiffies(mldv2_mrc(mld)), 1UL);
+
+ mld_update_qrv(idev, mld);
+ mld_update_qi(idev, mld);
+ mld_update_qri(idev, mld);
+
+ idev->mc_maxdelay = *max_delay;
+
+ return 0;
+}
+
/* called with rcu_read_lock() */
int igmp6_event_query(struct sk_buff *skb)
{
@@ -1229,7 +1278,7 @@ int igmp6_event_query(struct sk_buff *skb)
struct mld_msg *mld;
int group_type;
int mark = 0;
- int len;
+ int len, err;
if (!pskb_may_pull(skb, sizeof(struct in6_addr)))
return -EINVAL;
@@ -1255,47 +1304,21 @@ int igmp6_event_query(struct sk_buff *skb)
return -EINVAL;
if (len == MLD_V1_QUERY_LEN) {
- unsigned long mldv1_md;
-
- /* Ignore v1 queries */
- if (mld_in_v2_mode_only(idev))
- return 0;
-
- /* MLDv1 router present */
- mldv1_md = ntohs(mld->mld_maxdelay);
- max_delay = max(msecs_to_jiffies(mldv1_md), 1UL);
-
- mld_set_v1_mode(idev);
-
- /* cancel MLDv2 report timer */
- idev->mc_gq_running = 0;
- if (del_timer(&idev->mc_gq_timer))
- __in6_dev_put(idev);
-
- /* cancel the interface change timer */
- idev->mc_ifc_count = 0;
- if (del_timer(&idev->mc_ifc_timer))
- __in6_dev_put(idev);
- /* clear deleted report items */
- mld_clear_delrec(idev);
+ err = mld_process_v1(idev, mld, &max_delay);
+ if (err < 0)
+ return err;
} else if (len >= MLD_V2_QUERY_LEN_MIN) {
int srcs_offset = sizeof(struct mld2_query) -
sizeof(struct icmp6hdr);
- /* hosts need to stay in MLDv1 mode, discard MLDv2 queries */
- if (mld_in_v1_mode(idev))
- return 0;
if (!pskb_may_pull(skb, srcs_offset))
return -EINVAL;
mlh2 = (struct mld2_query *)skb_transport_header(skb);
- max_delay = max(msecs_to_jiffies(mldv2_mrc(mlh2)), 1UL);
- idev->mc_maxdelay = max_delay;
-
- mld_update_qrv(idev, mlh2);
- mld_update_qi(idev, mlh2);
- mld_update_qri(idev, mlh2);
+ err = mld_process_v2(idev, mlh2, &max_delay);
+ if (err < 0)
+ return err;
if (group_type == IPV6_ADDR_ANY) { /* general query */
if (mlh2->mld2q_nsrcs)
--
1.7.11.7
^ permalink raw reply related
* [PATCH net-next v2 5/8] net: ipv6: mld: similarly to MLDv2 have min max_delay of 1
From: Daniel Borkmann @ 2013-09-03 22:19 UTC (permalink / raw)
To: davem; +Cc: netdev, Hannes Frederic Sowa
In-Reply-To: <1378246784-21067-1-git-send-email-dborkman@redhat.com>
Similarly as we do in MLDv2 queries, set a forged MLDv1 query with
0 ms mld_maxdelay to minimum timer shot time of 1 jiffies. This is
eventually done in igmp6_group_queried() anyway, so we can simplify
a check there.
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Cc: Hannes Frederic Sowa <hannes@stressinduktion.org>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
---
net/ipv6/mcast.c | 14 +++++++-------
1 file changed, 7 insertions(+), 7 deletions(-)
diff --git a/net/ipv6/mcast.c b/net/ipv6/mcast.c
index 24fdfbb..d7e7c58 100644
--- a/net/ipv6/mcast.c
+++ b/net/ipv6/mcast.c
@@ -1040,12 +1040,9 @@ static void igmp6_group_queried(struct ifmcaddr6 *ma, unsigned long resptime)
delay = ma->mca_timer.expires - jiffies;
}
- if (delay >= resptime) {
- if (resptime)
- delay = net_random() % resptime;
- else
- delay = 1;
- }
+ if (delay >= resptime)
+ delay = net_random() % resptime;
+
ma->mca_timer.expires = jiffies + delay;
if (!mod_timer(&ma->mca_timer, jiffies + delay))
atomic_inc(&ma->mca_refcnt);
@@ -1258,12 +1255,15 @@ int igmp6_event_query(struct sk_buff *skb)
return -EINVAL;
if (len == MLD_V1_QUERY_LEN) {
+ unsigned long mldv1_md;
+
/* Ignore v1 queries */
if (mld_in_v2_mode_only(idev))
return 0;
/* MLDv1 router present */
- max_delay = msecs_to_jiffies(ntohs(mld->mld_maxdelay));
+ mldv1_md = ntohs(mld->mld_maxdelay);
+ max_delay = max(msecs_to_jiffies(mldv1_md), 1UL);
mld_set_v1_mode(idev);
--
1.7.11.7
^ permalink raw reply related
* [PATCH net-next v2 4/8] net: ipv6: mld: implement RFC3810 MLDv2 mode only
From: Daniel Borkmann @ 2013-09-03 22:19 UTC (permalink / raw)
To: davem; +Cc: netdev, Hannes Frederic Sowa
In-Reply-To: <1378246784-21067-1-git-send-email-dborkman@redhat.com>
RFC3810, 10. Security Considerations says under subsection 10.1.
Query Message:
A forged Version 1 Query message will put MLDv2 listeners on that
link in MLDv1 Host Compatibility Mode. This scenario can be avoided
by providing MLDv2 hosts with a configuration option to ignore
Version 1 messages completely.
Hence, implement a MLDv2-only mode that will ignore MLDv1 traffic:
echo 2 > /proc/sys/net/ipv6/conf/ethX/force_mld_version or
echo 2 > /proc/sys/net/ipv6/conf/all/force_mld_version
Note that <all> device has a higher precedence as it was previously
also the case in the macro MLD_V1_SEEN() that would "short-circuit"
if condition on <all> case.
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Cc: Hannes Frederic Sowa <hannes@stressinduktion.org>
---
net/ipv6/mcast.c | 34 ++++++++++++++++++++++++++++++----
1 file changed, 30 insertions(+), 4 deletions(-)
diff --git a/net/ipv6/mcast.c b/net/ipv6/mcast.c
index 005b22f..24fdfbb 100644
--- a/net/ipv6/mcast.c
+++ b/net/ipv6/mcast.c
@@ -1112,11 +1112,34 @@ static bool mld_marksources(struct ifmcaddr6 *pmc, int nsrcs,
return true;
}
+static int mld_force_mld_version(const struct inet6_dev *idev)
+{
+ /* Normally, both are 0 here. If enforcement to a particular is
+ * being used, individual device enforcement will have a lower
+ * precedence over 'all' device (.../conf/all/force_mld_version).
+ */
+
+ if (dev_net(idev->dev)->ipv6.devconf_all->force_mld_version != 0)
+ return dev_net(idev->dev)->ipv6.devconf_all->force_mld_version;
+ else
+ return idev->cnf.force_mld_version;
+}
+
+static bool mld_in_v2_mode_only(const struct inet6_dev *idev)
+{
+ return mld_force_mld_version(idev) == 2;
+}
+
+static bool mld_in_v1_mode_only(const struct inet6_dev *idev)
+{
+ return mld_force_mld_version(idev) == 1;
+}
+
static bool mld_in_v1_mode(const struct inet6_dev *idev)
{
- if (dev_net(idev->dev)->ipv6.devconf_all->force_mld_version == 1)
- return true;
- if (idev->cnf.force_mld_version == 1)
+ if (mld_in_v2_mode_only(idev))
+ return false;
+ if (mld_in_v1_mode_only(idev))
return true;
if (idev->mc_v1_seen && time_before(jiffies, idev->mc_v1_seen))
return true;
@@ -1223,7 +1246,6 @@ int igmp6_event_query(struct sk_buff *skb)
return -EINVAL;
idev = __in6_dev_get(skb->dev);
-
if (idev == NULL)
return 0;
@@ -1236,6 +1258,10 @@ int igmp6_event_query(struct sk_buff *skb)
return -EINVAL;
if (len == MLD_V1_QUERY_LEN) {
+ /* Ignore v1 queries */
+ if (mld_in_v2_mode_only(idev))
+ return 0;
+
/* MLDv1 router present */
max_delay = msecs_to_jiffies(ntohs(mld->mld_maxdelay));
--
1.7.11.7
^ permalink raw reply related
* [PATCH net-next v2 3/8] net: ipv6: mld: get rid of MLDV2_MRC and simplify calculation
From: Daniel Borkmann @ 2013-09-03 22:19 UTC (permalink / raw)
To: davem; +Cc: netdev, Hannes Frederic Sowa
In-Reply-To: <1378246784-21067-1-git-send-email-dborkman@redhat.com>
Get rid of MLDV2_MRC and use our new macros for mantisse and
exponent to calculate Maximum Response Delay out of the Maximum
Response Code.
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Cc: Hannes Frederic Sowa <hannes@stressinduktion.org>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
---
This will give a merge conflict in net/bridge/br_multicast.c with
net tree. You can resolve it by simply taking this version as we
already do msecs_to_jiffies() here.
include/net/mld.h | 28 +++++++++++++++++++---------
net/bridge/br_multicast.c | 3 ++-
net/ipv6/mcast.c | 18 ++----------------
3 files changed, 23 insertions(+), 26 deletions(-)
diff --git a/include/net/mld.h b/include/net/mld.h
index 2b5421f..faa1d16 100644
--- a/include/net/mld.h
+++ b/include/net/mld.h
@@ -63,15 +63,6 @@ struct mld2_query {
#define mld2q_mrc mld2q_hdr.icmp6_maxdelay
#define mld2q_resv1 mld2q_hdr.icmp6_dataun.un_data16[1]
-/* Max Response Code, TODO: transform this to use the below */
-#define MLDV2_MASK(value, nb) ((nb)>=32 ? (value) : ((1<<(nb))-1) & (value))
-#define MLDV2_EXP(thresh, nbmant, nbexp, value) \
- ((value) < (thresh) ? (value) : \
- ((MLDV2_MASK(value, nbmant) | (1<<(nbmant))) << \
- (MLDV2_MASK((value) >> (nbmant), nbexp) + (nbexp))))
-
-#define MLDV2_MRC(value) MLDV2_EXP(0x8000, 12, 3, value)
-
/* RFC3810, 5.1.3. Maximum Response Code:
*
* If Maximum Response Code >= 32768, Maximum Response Code represents a
@@ -97,4 +88,23 @@ struct mld2_query {
#define MLDV2_QQIC_EXP(value) (((value) >> 4) & 0x07)
#define MLDV2_QQIC_MAN(value) ((value) & 0x0f)
+static inline unsigned long mldv2_mrc(const struct mld2_query *mlh2)
+{
+ /* RFC3810, 5.1.3. Maximum Response Code */
+ unsigned long ret, mc_mrc = ntohs(mlh2->mld2q_mrc);
+
+ if (mc_mrc < 32768) {
+ ret = mc_mrc;
+ } else {
+ unsigned long mc_man, mc_exp;
+
+ mc_exp = MLDV2_MRC_EXP(mc_mrc);
+ mc_man = MLDV2_MRC_MAN(mc_mrc);
+
+ ret = (mc_man | 0x1000) << (mc_exp + 3);
+ }
+
+ return ret;
+}
+
#endif
diff --git a/net/bridge/br_multicast.c b/net/bridge/br_multicast.c
index 08e576a..4accd0d 100644
--- a/net/bridge/br_multicast.c
+++ b/net/bridge/br_multicast.c
@@ -1203,7 +1203,8 @@ static int br_ip6_multicast_query(struct net_bridge *br,
mld2q = (struct mld2_query *)icmp6_hdr(skb);
if (!mld2q->mld2q_nsrcs)
group = &mld2q->mld2q_mca;
- max_delay = mld2q->mld2q_mrc ? MLDV2_MRC(ntohs(mld2q->mld2q_mrc)) : 1;
+
+ max_delay = max(msecs_to_jiffies(mldv2_mrc(mld2q)), 1UL);
}
br_multicast_query_received(br, port, !ipv6_addr_any(&ip6h->saddr),
diff --git a/net/ipv6/mcast.c b/net/ipv6/mcast.c
index f86e26b..005b22f 100644
--- a/net/ipv6/mcast.c
+++ b/net/ipv6/mcast.c
@@ -1195,20 +1195,7 @@ static void mld_update_qri(struct inet6_dev *idev,
* - 5.1.3. Maximum Response Code
* - 9.3. Query Response Interval
*/
- unsigned long mc_qri, mc_mrc = ntohs(mlh2->mld2q_mrc);
-
- if (mc_mrc < 32768) {
- mc_qri = mc_mrc;
- } else {
- unsigned long mc_man, mc_exp;
-
- mc_exp = MLDV2_MRC_EXP(mc_mrc);
- mc_man = MLDV2_MRC_MAN(mc_mrc);
-
- mc_qri = (mc_man | 0x1000) << (mc_exp + 3);
- }
-
- idev->mc_qri = msecs_to_jiffies(mc_qri);
+ idev->mc_qri = msecs_to_jiffies(mldv2_mrc(mlh2));
}
/* called with rcu_read_lock() */
@@ -1277,8 +1264,7 @@ int igmp6_event_query(struct sk_buff *skb)
mlh2 = (struct mld2_query *)skb_transport_header(skb);
- max_delay = max(msecs_to_jiffies(MLDV2_MRC(ntohs(mlh2->mld2q_mrc))), 1UL);
-
+ max_delay = max(msecs_to_jiffies(mldv2_mrc(mlh2)), 1UL);
idev->mc_maxdelay = max_delay;
mld_update_qrv(idev, mlh2);
--
1.7.11.7
^ permalink raw reply related
* [PATCH net-next v2 2/8] net: ipv6: mld: clean up MLD_V1_SEEN macro
From: Daniel Borkmann @ 2013-09-03 22:19 UTC (permalink / raw)
To: davem; +Cc: netdev, Hannes Frederic Sowa
In-Reply-To: <1378246784-21067-1-git-send-email-dborkman@redhat.com>
Replace the macro with a function to make it more readable. GCC will
eventually decide whether to inline this or not (also, that's not
fast-path anyway).
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Cc: Hannes Frederic Sowa <hannes@stressinduktion.org>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
---
net/ipv6/mcast.c | 34 +++++++++++++++++++++-------------
1 file changed, 21 insertions(+), 13 deletions(-)
diff --git a/net/ipv6/mcast.c b/net/ipv6/mcast.c
index 8992ff2..f86e26b 100644
--- a/net/ipv6/mcast.c
+++ b/net/ipv6/mcast.c
@@ -95,6 +95,7 @@ static void mld_ifc_event(struct inet6_dev *idev);
static void mld_add_delrec(struct inet6_dev *idev, struct ifmcaddr6 *pmc);
static void mld_del_delrec(struct inet6_dev *idev, const struct in6_addr *addr);
static void mld_clear_delrec(struct inet6_dev *idev);
+static bool mld_in_v1_mode(const struct inet6_dev *idev);
static int sf_setstate(struct ifmcaddr6 *pmc);
static void sf_markstate(struct ifmcaddr6 *pmc);
static void ip6_mc_clear_src(struct ifmcaddr6 *pmc);
@@ -117,11 +118,6 @@ static int ip6_mc_leave_src(struct sock *sk, struct ipv6_mc_socklist *iml,
#define MLD_V1_QUERY_LEN 24
#define MLD_V2_QUERY_LEN_MIN 28
-#define MLD_V1_SEEN(idev) (dev_net((idev)->dev)->ipv6.devconf_all->force_mld_version == 1 || \
- (idev)->cnf.force_mld_version == 1 || \
- ((idev)->mc_v1_seen && \
- time_before(jiffies, (idev)->mc_v1_seen)))
-
#define IPV6_MLD_MAX_MSF 64
int sysctl_mld_max_msf __read_mostly = IPV6_MLD_MAX_MSF;
@@ -139,7 +135,7 @@ static int unsolicited_report_interval(struct inet6_dev *idev)
{
int iv;
- if (MLD_V1_SEEN(idev))
+ if (mld_in_v1_mode(idev))
iv = idev->cnf.mldv1_unsolicited_report_interval;
else
iv = idev->cnf.mldv2_unsolicited_report_interval;
@@ -695,7 +691,7 @@ static void igmp6_group_added(struct ifmcaddr6 *mc)
if (!(dev->flags & IFF_UP) || (mc->mca_flags & MAF_NOREPORT))
return;
- if (MLD_V1_SEEN(mc->idev)) {
+ if (mld_in_v1_mode(mc->idev)) {
igmp6_join_group(mc);
return;
}
@@ -1116,6 +1112,18 @@ static bool mld_marksources(struct ifmcaddr6 *pmc, int nsrcs,
return true;
}
+static bool mld_in_v1_mode(const struct inet6_dev *idev)
+{
+ if (dev_net(idev->dev)->ipv6.devconf_all->force_mld_version == 1)
+ return true;
+ if (idev->cnf.force_mld_version == 1)
+ return true;
+ if (idev->mc_v1_seen && time_before(jiffies, idev->mc_v1_seen))
+ return true;
+
+ return false;
+}
+
static void mld_set_v1_mode(struct inet6_dev *idev)
{
/* RFC3810, relevant sections:
@@ -1262,7 +1270,7 @@ int igmp6_event_query(struct sk_buff *skb)
sizeof(struct icmp6hdr);
/* hosts need to stay in MLDv1 mode, discard MLDv2 queries */
- if (MLD_V1_SEEN(idev))
+ if (mld_in_v1_mode(idev))
return 0;
if (!pskb_may_pull(skb, srcs_offset))
return -EINVAL;
@@ -1942,7 +1950,7 @@ err_out:
static void mld_resend_report(struct inet6_dev *idev)
{
- if (MLD_V1_SEEN(idev)) {
+ if (mld_in_v1_mode(idev)) {
struct ifmcaddr6 *mcaddr;
read_lock_bh(&idev->lock);
for (mcaddr = idev->mc_list; mcaddr; mcaddr = mcaddr->next) {
@@ -2006,7 +2014,7 @@ static int ip6_mc_del1_src(struct ifmcaddr6 *pmc, int sfmode,
else
pmc->mca_sources = psf->sf_next;
if (psf->sf_oldin && !(pmc->mca_flags & MAF_NOREPORT) &&
- !MLD_V1_SEEN(idev)) {
+ !mld_in_v1_mode(idev)) {
psf->sf_crcount = idev->mc_qrv;
psf->sf_next = pmc->mca_tomb;
pmc->mca_tomb = psf;
@@ -2306,7 +2314,7 @@ static int ip6_mc_leave_src(struct sock *sk, struct ipv6_mc_socklist *iml,
static void igmp6_leave_group(struct ifmcaddr6 *ma)
{
- if (MLD_V1_SEEN(ma->idev)) {
+ if (mld_in_v1_mode(ma->idev)) {
if (ma->mca_flags & MAF_LAST_REPORTER)
igmp6_send(&ma->mca_addr, ma->idev->dev,
ICMPV6_MGM_REDUCTION);
@@ -2340,7 +2348,7 @@ static void mld_ifc_timer_expire(unsigned long data)
static void mld_ifc_event(struct inet6_dev *idev)
{
- if (MLD_V1_SEEN(idev))
+ if (mld_in_v1_mode(idev))
return;
idev->mc_ifc_count = idev->mc_qrv;
mld_ifc_start_timer(idev, 1);
@@ -2351,7 +2359,7 @@ static void igmp6_timer_handler(unsigned long data)
{
struct ifmcaddr6 *ma = (struct ifmcaddr6 *) data;
- if (MLD_V1_SEEN(ma->idev))
+ if (mld_in_v1_mode(ma->idev))
igmp6_send(&ma->mca_addr, ma->idev->dev, ICMPV6_MGM_REPORT);
else
mld_send_report(ma->idev, ma);
--
1.7.11.7
^ permalink raw reply related
* [PATCH net-next v2 1/8] net: ipv6: mld: fix v1/v2 switchback timeout to rfc3810, 9.12.
From: Daniel Borkmann @ 2013-09-03 22:19 UTC (permalink / raw)
To: davem; +Cc: netdev, David Stevens, Hannes Frederic Sowa
In-Reply-To: <1378246784-21067-1-git-send-email-dborkman@redhat.com>
i) RFC3810, 9.2. Query Interval [QI] says:
The Query Interval variable denotes the interval between General
Queries sent by the Querier. Default value: 125 seconds. [...]
ii) RFC3810, 9.3. Query Response Interval [QRI] says:
The Maximum Response Delay used to calculate the Maximum Response
Code inserted into the periodic General Queries. Default value:
10000 (10 seconds) [...] The number of seconds represented by the
[Query Response Interval] must be less than the [Query Interval].
iii) RFC3810, 9.12. Older Version Querier Present Timeout [OVQPT] says:
The Older Version Querier Present Timeout is the time-out for
transitioning a host back to MLDv2 Host Compatibility Mode. When an
MLDv1 query is received, MLDv2 hosts set their Older Version Querier
Present Timer to [Older Version Querier Present Timeout].
This value MUST be ([Robustness Variable] times (the [Query Interval]
in the last Query received)) plus ([Query Response Interval]).
Hence, on *default* the timeout results in:
[RV] = 2, [QI] = 125sec, [QRI] = 10sec
[OVQPT] = [RV] * [QI] + [QRI] = 260sec
Having that said, we currently calculate [OVQPT] (here given as 'switchback'
variable) as ...
switchback = (idev->mc_qrv + 1) * max_delay
RFC3810, 9.12. says "the [Query Interval] in the last Query received". In
section "9.14. Configuring timers", it is said:
This section is meant to provide advice to network administrators on
how to tune these settings to their network. Ambitious router
implementations might tune these settings dynamically based upon
changing characteristics of the network. [...]
iv) RFC38010, 9.14.2. Query Interval:
The overall level of periodic MLD traffic is inversely proportional
to the Query Interval. A longer Query Interval results in a lower
overall level of MLD traffic. The value of the Query Interval MUST
be equal to or greater than the Maximum Response Delay used to
calculate the Maximum Response Code inserted in General Query
messages.
I assume that was why switchback is calculated as is (3 * max_delay), although
this setting seems to be meant for routers only to configure their [QI]
interval for non-default intervals. So usage here like this is clearly wrong.
Concluding, the current behaviour in IPv6's multicast code is not conform
to the RFC as switch back is calculated wrongly. That is, it has a too small
value, so MLDv2 hosts switch back again to MLDv2 way too early, i.e. ~30secs
instead of ~260secs on default.
Hence, introduce necessary helper functions and fix this up properly as it
should be.
Introduced in 06da92283 ("[IPV6]: Add MLDv2 support."). Credits to Hannes
Frederic Sowa who also had a hand in this as well. Also thanks to Hangbin Liu
who did initial testing.
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Cc: David Stevens <dlstevens@us.ibm.com>
Cc: Hannes Frederic Sowa <hannes@stressinduktion.org>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
---
06da92283 is linux-history tree.
include/net/if_inet6.h | 9 +++-
include/net/mld.h | 27 +++++++++++-
net/ipv6/mcast.c | 116 ++++++++++++++++++++++++++++++++++++++++++++++---
3 files changed, 143 insertions(+), 9 deletions(-)
diff --git a/include/net/if_inet6.h b/include/net/if_inet6.h
index 736b5fb..02ef772 100644
--- a/include/net/if_inet6.h
+++ b/include/net/if_inet6.h
@@ -171,12 +171,17 @@ struct inet6_dev {
struct ifmcaddr6 *mc_list;
struct ifmcaddr6 *mc_tomb;
spinlock_t mc_lock;
- unsigned char mc_qrv;
+
+ unsigned char mc_qrv; /* Query Robustness Variable */
unsigned char mc_gq_running;
unsigned char mc_ifc_count;
unsigned char mc_dad_count;
- unsigned long mc_v1_seen;
+
+ unsigned long mc_v1_seen; /* Max time we stay in MLDv1 mode */
+ unsigned long mc_qi; /* Query Interval */
+ unsigned long mc_qri; /* Query Response Interval */
unsigned long mc_maxdelay;
+
struct timer_list mc_gq_timer; /* general query timer */
struct timer_list mc_ifc_timer; /* interface change timer */
struct timer_list mc_dad_timer; /* dad complete mc timer */
diff --git a/include/net/mld.h b/include/net/mld.h
index 467143c..2b5421f 100644
--- a/include/net/mld.h
+++ b/include/net/mld.h
@@ -63,7 +63,7 @@ struct mld2_query {
#define mld2q_mrc mld2q_hdr.icmp6_maxdelay
#define mld2q_resv1 mld2q_hdr.icmp6_dataun.un_data16[1]
-/* Max Response Code */
+/* Max Response Code, TODO: transform this to use the below */
#define MLDV2_MASK(value, nb) ((nb)>=32 ? (value) : ((1<<(nb))-1) & (value))
#define MLDV2_EXP(thresh, nbmant, nbexp, value) \
((value) < (thresh) ? (value) : \
@@ -72,4 +72,29 @@ struct mld2_query {
#define MLDV2_MRC(value) MLDV2_EXP(0x8000, 12, 3, value)
+/* RFC3810, 5.1.3. Maximum Response Code:
+ *
+ * If Maximum Response Code >= 32768, Maximum Response Code represents a
+ * floating-point value as follows:
+ *
+ * 0 1 2 3 4 5 6 7 8 9 A B C D E F
+ * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+ * |1| exp | mant |
+ * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+ */
+#define MLDV2_MRC_EXP(value) (((value) >> 12) & 0x0007)
+#define MLDV2_MRC_MAN(value) ((value) & 0x0fff)
+
+/* RFC3810, 5.1.9. QQIC (Querier's Query Interval Code):
+ *
+ * If QQIC >= 128, QQIC represents a floating-point value as follows:
+ *
+ * 0 1 2 3 4 5 6 7
+ * +-+-+-+-+-+-+-+-+
+ * |1| exp | mant |
+ * +-+-+-+-+-+-+-+-+
+ */
+#define MLDV2_QQIC_EXP(value) (((value) >> 4) & 0x07)
+#define MLDV2_QQIC_MAN(value) ((value) & 0x0f)
+
#endif
diff --git a/net/ipv6/mcast.c b/net/ipv6/mcast.c
index 98ead2b..8992ff2 100644
--- a/net/ipv6/mcast.c
+++ b/net/ipv6/mcast.c
@@ -108,6 +108,10 @@ static int ip6_mc_leave_src(struct sock *sk, struct ipv6_mc_socklist *iml,
struct inet6_dev *idev);
#define MLD_QRV_DEFAULT 2
+/* RFC3810, 9.2. Query Interval */
+#define MLD_QI_DEFAULT (125 * HZ)
+/* RFC3810, 9.3. Query Response Interval */
+#define MLD_QRI_DEFAULT (10 * HZ)
/* RFC3810, 8.1 Query Version Distinctions */
#define MLD_V1_QUERY_LEN 24
@@ -1112,6 +1116,93 @@ static bool mld_marksources(struct ifmcaddr6 *pmc, int nsrcs,
return true;
}
+static void mld_set_v1_mode(struct inet6_dev *idev)
+{
+ /* RFC3810, relevant sections:
+ * - 9.1. Robustness Variable
+ * - 9.2. Query Interval
+ * - 9.3. Query Response Interval
+ * - 9.12. Older Version Querier Present Timeout
+ */
+ unsigned long switchback;
+
+ switchback = (idev->mc_qrv * idev->mc_qi) + idev->mc_qri;
+
+ idev->mc_v1_seen = jiffies + switchback;
+}
+
+static void mld_update_qrv(struct inet6_dev *idev,
+ const struct mld2_query *mlh2)
+{
+ /* RFC3810, relevant sections:
+ * - 5.1.8. QRV (Querier's Robustness Variable)
+ * - 9.1. Robustness Variable
+ */
+
+ /* The value of the Robustness Variable MUST NOT be zero,
+ * and SHOULD NOT be one. Catch this here if we ever run
+ * into such a case in future.
+ */
+ WARN_ON(idev->mc_qrv == 0);
+
+ if (mlh2->mld2q_qrv > 0)
+ idev->mc_qrv = mlh2->mld2q_qrv;
+
+ if (unlikely(idev->mc_qrv < 2)) {
+ net_warn_ratelimited("IPv6: MLD: clamping QRV from %u to %u!\n",
+ idev->mc_qrv, MLD_QRV_DEFAULT);
+ idev->mc_qrv = MLD_QRV_DEFAULT;
+ }
+}
+
+static void mld_update_qi(struct inet6_dev *idev,
+ const struct mld2_query *mlh2)
+{
+ /* RFC3810, relevant sections:
+ * - 5.1.9. QQIC (Querier's Query Interval Code)
+ * - 9.2. Query Interval
+ * - 9.12. Older Version Querier Present Timeout
+ * (the [Query Interval] in the last Query received)
+ */
+ unsigned long mc_qqi;
+
+ if (mlh2->mld2q_qqic < 128) {
+ mc_qqi = mlh2->mld2q_qqic;
+ } else {
+ unsigned long mc_man, mc_exp;
+
+ mc_exp = MLDV2_QQIC_EXP(mlh2->mld2q_qqic);
+ mc_man = MLDV2_QQIC_MAN(mlh2->mld2q_qqic);
+
+ mc_qqi = (mc_man | 0x10) << (mc_exp + 3);
+ }
+
+ idev->mc_qi = mc_qqi * HZ;
+}
+
+static void mld_update_qri(struct inet6_dev *idev,
+ const struct mld2_query *mlh2)
+{
+ /* RFC3810, relevant sections:
+ * - 5.1.3. Maximum Response Code
+ * - 9.3. Query Response Interval
+ */
+ unsigned long mc_qri, mc_mrc = ntohs(mlh2->mld2q_mrc);
+
+ if (mc_mrc < 32768) {
+ mc_qri = mc_mrc;
+ } else {
+ unsigned long mc_man, mc_exp;
+
+ mc_exp = MLDV2_MRC_EXP(mc_mrc);
+ mc_man = MLDV2_MRC_MAN(mc_mrc);
+
+ mc_qri = (mc_man | 0x1000) << (mc_exp + 3);
+ }
+
+ idev->mc_qri = msecs_to_jiffies(mc_qri);
+}
+
/* called with rcu_read_lock() */
int igmp6_event_query(struct sk_buff *skb)
{
@@ -1150,12 +1241,15 @@ int igmp6_event_query(struct sk_buff *skb)
return -EINVAL;
if (len == MLD_V1_QUERY_LEN) {
- int switchback;
/* MLDv1 router present */
-
max_delay = msecs_to_jiffies(ntohs(mld->mld_maxdelay));
- switchback = (idev->mc_qrv + 1) * max_delay;
- idev->mc_v1_seen = jiffies + switchback;
+
+ mld_set_v1_mode(idev);
+
+ /* cancel MLDv2 report timer */
+ idev->mc_gq_running = 0;
+ if (del_timer(&idev->mc_gq_timer))
+ __in6_dev_put(idev);
/* cancel the interface change timer */
idev->mc_ifc_count = 0;
@@ -1166,6 +1260,10 @@ int igmp6_event_query(struct sk_buff *skb)
} else if (len >= MLD_V2_QUERY_LEN_MIN) {
int srcs_offset = sizeof(struct mld2_query) -
sizeof(struct icmp6hdr);
+
+ /* hosts need to stay in MLDv1 mode, discard MLDv2 queries */
+ if (MLD_V1_SEEN(idev))
+ return 0;
if (!pskb_may_pull(skb, srcs_offset))
return -EINVAL;
@@ -1175,8 +1273,10 @@ int igmp6_event_query(struct sk_buff *skb)
idev->mc_maxdelay = max_delay;
- if (mlh2->mld2q_qrv)
- idev->mc_qrv = mlh2->mld2q_qrv;
+ mld_update_qrv(idev, mlh2);
+ mld_update_qi(idev, mlh2);
+ mld_update_qri(idev, mlh2);
+
if (group_type == IPV6_ADDR_ANY) { /* general query */
if (mlh2->mld2q_nsrcs)
return -EINVAL; /* no sources allowed */
@@ -2337,7 +2437,11 @@ void ipv6_mc_init_dev(struct inet6_dev *idev)
(unsigned long)idev);
setup_timer(&idev->mc_dad_timer, mld_dad_timer_expire,
(unsigned long)idev);
+
idev->mc_qrv = MLD_QRV_DEFAULT;
+ idev->mc_qi = MLD_QI_DEFAULT;
+ idev->mc_qri = MLD_QRI_DEFAULT;
+
idev->mc_maxdelay = unsolicited_report_interval(idev);
idev->mc_v1_seen = 0;
write_unlock_bh(&idev->lock);
--
1.7.11.7
^ permalink raw reply related
* [PATCH net-next v2 0/8] IPv6 MLD updates
From: Daniel Borkmann @ 2013-09-03 22:19 UTC (permalink / raw)
To: davem; +Cc: netdev
This set contains the non-RFC version of the previous submission entitled
"[RFC PATCH net-next 0/7] IPv6 MLD updates". Most importantly it contains
a fix for MLDv1/v2 switchback timeout where hosts currently are switching
back from v1 compat mode to normal v2 too early (i.e. switchback time was
<= 30secs instead of >= 260secs on default), and the set also contains a
patch that allows for v2-only mode as per RFC recommendation. The rest is
related to cleanups that make the code more readable resp. maintainable.
Changes from RFC to non-RFC:
We ignore v2 messages now when in v1 compat mode, otherwise report timers
are reset and triggered, also stop current v2 report timer in case it is
currently running and we received a v1 query; use WARN_ON instead of BUG_ON
for RV of 0 (patch1).
Two more patches have been added that makes to code more readable, that is
"net: ipv6: mld: refactor query processing into v1/v2 functions" and
"net: ipv6: mld: introduce mld_{gq,ifc,dad}_stop_timer functions", the
sysctl patch "net: ipv6: mld: restrict min/max of sysctl force_mld_version"
has been dropped as extra1 and extra2 vars are overwritten with idev and net
anyway when addrconf sysctl is registered, hence dropped for now. The rest is
unchanged, only adapted to take changes into account.
Changes from v1 to v2:
Minor update in "net: ipv6: mld: implement RFC3810 MLDv2 mode only" in MLD
version selection. Rest stays the same, added Acked-bys from Hannes in
unchanged patches.
Thanks!
Daniel Borkmann (8):
net: ipv6: mld: fix v1/v2 switchback timeout to rfc3810, 9.12.
net: ipv6: mld: clean up MLD_V1_SEEN macro
net: ipv6: mld: get rid of MLDV2_MRC and simplify calculation
net: ipv6: mld: implement RFC3810 MLDv2 mode only
net: ipv6: mld: similarly to MLDv2 have min max_delay of 1
net: ipv6: mld: refactor query processing into v1/v2 functions
net: ipv6: mld: introduce mld_{gq,ifc,dad}_stop_timer functions
net: ipv6: mld: document force_mld_version in ip-sysctl.txt
Documentation/networking/ip-sysctl.txt | 5 +
include/net/if_inet6.h | 9 +-
include/net/mld.h | 51 +++++--
net/bridge/br_multicast.c | 3 +-
net/ipv6/mcast.c | 248 +++++++++++++++++++++++++++------
5 files changed, 259 insertions(+), 57 deletions(-)
--
1.7.11.7
^ permalink raw reply
* Re: [PATCH 2/2] sh_eth: add device tree support
From: Kumar Gala @ 2013-09-03 22:08 UTC (permalink / raw)
To: Sergei Shtylyov
Cc: netdev, rob.herring, pawel.moll, mark.rutland, swarren,
ian.campbell, grant.likely, devicetree, nobuhiro.iwamatsu.yj,
linux-sh, rob, linux-doc
In-Reply-To: <201308310429.34856.sergei.shtylyov@cogentembedded.com>
On Aug 30, 2013, at 7:29 PM, Sergei Shtylyov wrote:
> Add support of the device tree probing for the Renesas SH-Mobile SoCs.
>
> This work is loosely based on an original patch by Nobuhiro Iwamatsu
> <nobuhiro.iwamatsu.yj@renesas.com>.
>
> Signed-off-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>
>
> ---
> This patch is against Dave's 'net-next.git' repo.
>
> Documentation/devicetree/bindings/net/sh_eth.txt | 40 +++++++++++++
> drivers/net/ethernet/renesas/sh_eth.c | 66 ++++++++++++++++++++++-
> 2 files changed, 105 insertions(+), 1 deletion(-)
>
> Index: net-next/Documentation/devicetree/bindings/net/sh_eth.txt
> ===================================================================
> --- /dev/null
> +++ net-next/Documentation/devicetree/bindings/net/sh_eth.txt
> @@ -0,0 +1,40 @@
> +* Renesas Electronics SH EtherMAC
> +
> +This file provides information on what the device node for the SH EtherMAC
> +interface contains.
> +
> +Required properties:
> +- compatible: "renesas,gether-r8a7740" if the device is a part of R8A7740 SoC.
> + "renesas,ether-r8a7779" if the device is a part of R8A7778/9 SoCs.
> + "renesas,ether-r8a7790" if the device is a part of R8A7790/1 SoCs.
Curious, 7779 (vs 7778) but 7790 (vs 7791).
> +- reg: offset and length of the register set for the device; if the device has
> + TSU registers, you need to specify two register sets here.
> +- interrupt-parent: the phandle for the interrupt controller that services
> + interrupts for this device.
> +- interrupts: interrupt mapping for the interrupt source.
> +- phy-mode: string, operation mode of the PHY interface (a string that
> + of_get_phy_mode() can understand).
> +- phy-handle: phandle of the PHY device.
> +
Should add something about requiring a phy subnode
> +Optional properties:
> +- local-mac-address: 6 bytes, MAC address.
What's the assumption if local-mac-address is not specified?
> +- renesas,no-ether-link: specify when a board does not provide a proper LINK
> + signal.
How is this different from the fixed link concept we have?
> +- renesas,ether-link-active-low: specify when the LINK signal is active-low.
> +
> +Example (Armadillo800EVA board):
> +
> + ethernet@e9a00000 {
> + compatible = "renesas,gether-r8a7740";
> + reg = <0xe9a00000 0x800>, <0xe9a01800 0x800>;
> + interrupt-parent = <&gic>;
> + interrupts = <0 142 0x4>;
> + phy-mode = "mii";
> + phy-handle = <&phy0>;
> + #address-cells = <1>;
> + #size-cells = <0>;
> +
> + phy0: ethernet-phy@0 {
> + reg = <0>;
> + };
> + };
- k
--
Employee of Qualcomm Innovation Center, Inc.
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, hosted by The Linux Foundation
^ permalink raw reply
* Re: [PATCH] openvswitch: fix sw_flow_key alignment
From: Jesse Gross @ 2013-09-03 22:06 UTC (permalink / raw)
To: David Miller; +Cc: Andy Zhou, dev@openvswitch.org, netdev
In-Reply-To: <20130831.001650.130549489746201651.davem@davemloft.net>
On Fri, Aug 30, 2013 at 9:16 PM, David Miller <davem@davemloft.net> wrote:
> From: Jesse Gross <jesse@nicira.com>
> Date: Fri, 30 Aug 2013 16:20:25 -0700
>
>> I looked through the struct definition and I think that the idea of
>> manually padding as Geert did in his patch will be difficult to
>> maintain over time (and actually there are a few that he missed) since
>> there are a number of different structs/unions contained in there.
>
> You have to be mindful of the gaps and wasted space for performance
> reasons anyways.
Yes, although the approaches for performance and correctness are not
necessarily the same. For example, we're taking about potentially
packing the struct, in which case we would still have the same
alignment needs even without any gaps. If the correctness is clearly
handled (through an explicit align and build assert) then it's easier
to pack/rearrange/whatever for performance since the whole thing isn't
fragile.
^ permalink raw reply
* Re: [PATCH net] net: ipv6: tcp: fix potential use after free in tcp_v6_do_rcv
From: Jean Sacren @ 2013-09-03 21:59 UTC (permalink / raw)
To: Eric Dumazet; +Cc: Daniel Borkmann, davem, netdev, yoshfuji, Jiri Benc
In-Reply-To: <1378240517.7360.41.camel@edumazet-glaptop>
From: Eric Dumazet <eric.dumazet@gmail.com>
Date: Tue, 03 Sep 2013 13:35:17 -0700
>
> How did you get your conclusion ?
If one changes one line of code, doesn't one own that line?
--
Jean Sacren
^ permalink raw reply
page: next (older) | prev (newer) | latest
- recent:[subjects (threaded)|topics (new)|topics (active)]
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox