[PATCH net v4] ipv6: mld: fix add_grhead skb_over_panic for devs with large MTUs

[PATCH net v4] ipv6: mld: fix add_grhead skb_over_panic for devs with large MTUs

[Daniel Borkmann]

It has been reported that generating an MLD listener report on
devices with large MTUs (e.g. 9000) and a high number of IPv6
addresses can trigger a skb_over_panic():

skbuff: skb_over_panic: text:ffffffff80612a5d len:3776 put:20
head:ffff88046d751000 data:ffff88046d751010 tail:0xed0 end:0xec0
dev:port1
 ------------[ cut here ]------------
kernel BUG at net/core/skbuff.c:100!
invalid opcode: 0000 [#1] SMP
Modules linked in: ixgbe(O)
CPU: 3 PID: 0 Comm: swapper/3 Tainted: G O 3.14.23+ #4
[...]
Call Trace:
 <IRQ>
 [<ffffffff80578226>] ? skb_put+0x3a/0x3b
 [<ffffffff80612a5d>] ? add_grhead+0x45/0x8e
 [<ffffffff80612e3a>] ? add_grec+0x394/0x3d4
 [<ffffffff80613222>] ? mld_ifc_timer_expire+0x195/0x20d
 [<ffffffff8061308d>] ? mld_dad_timer_expire+0x45/0x45
 [<ffffffff80255b5d>] ? call_timer_fn.isra.29+0x12/0x68
 [<ffffffff80255d16>] ? run_timer_softirq+0x163/0x182
 [<ffffffff80250e6f>] ? __do_softirq+0xe0/0x21d
 [<ffffffff8025112b>] ? irq_exit+0x4e/0xd3
 [<ffffffff802214bb>] ? smp_apic_timer_interrupt+0x3b/0x46
 [<ffffffff8063f10a>] ? apic_timer_interrupt+0x6a/0x70

mld_newpack() skb allocations are usually requested with dev->mtu
in size, since commit 72e09ad107e7 ("ipv6: avoid high order allocations")
we have changed the limit in order to be less likely to fail.

However, in MLD/IGMP code, we have some rather ugly AVAILABLE(skb)
macros, which determine if we may end up doing an skb_put() for
adding another record. To avoid possible fragmentation, we check
the skb's tailroom as skb->dev->mtu - skb->len, which is a wrong
assumption as the actual max allocation size can be much smaller.

The IGMP case doesn't have this issue as commit 57e1ab6eaddc
("igmp: refine skb allocations") stores the allocation size in
the cb[].

Set a reserved_tailroom to make it fit into the MTU and use
skb_availroom() helper instead. This also allows to get rid of
igmp_skb_size().

Reported-by: Wei Liu <lw1a2.jing@gmail.com>
Fixes: 72e09ad107e7 ("ipv6: avoid high order allocations")
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Cc: Eric Dumazet <edumazet@google.com>
Cc: Hannes Frederic Sowa <hannes@stressinduktion.org>
Cc: David L Stevens <david.stevens@oracle.com>
---
 v3->v4:
  - Reduced noise from previous one as discussed
 v2->v3:
  - Still had a discussion w/ Hannes and improved the code a bit as
    suggested to make it more clear to read
 v1->v2:
  - Don't introduce skb_nofrag_tailroom(), but reuse skb_availroom()
    as suggested by Eric

 net/ipv4/igmp.c  | 11 +++++------
 net/ipv6/mcast.c |  9 +++++----
 2 files changed, 10 insertions(+), 10 deletions(-)

diff --git a/net/ipv4/igmp.c b/net/ipv4/igmp.c
index fb70e3e..bb15d0e 100644
--- a/net/ipv4/igmp.c
+++ b/net/ipv4/igmp.c
@@ -318,9 +318,7 @@ igmp_scount(struct ip_mc_list *pmc, int type, int gdeleted, int sdeleted)
 	return scount;
 }
 
-#define igmp_skb_size(skb) (*(unsigned int *)((skb)->cb))
-
-static struct sk_buff *igmpv3_newpack(struct net_device *dev, int size)
+static struct sk_buff *igmpv3_newpack(struct net_device *dev, unsigned int mtu)
 {
 	struct sk_buff *skb;
 	struct rtable *rt;
@@ -330,6 +328,7 @@ static struct sk_buff *igmpv3_newpack(struct net_device *dev, int size)
 	struct flowi4 fl4;
 	int hlen = LL_RESERVED_SPACE(dev);
 	int tlen = dev->needed_tailroom;
+	unsigned int size = mtu;
 
 	while (1) {
 		skb = alloc_skb(size + hlen + tlen,
@@ -341,7 +340,6 @@ static struct sk_buff *igmpv3_newpack(struct net_device *dev, int size)
 			return NULL;
 	}
 	skb->priority = TC_PRIO_CONTROL;
-	igmp_skb_size(skb) = size;
 
 	rt = ip_route_output_ports(net, &fl4, NULL, IGMPV3_ALL_MCR, 0,
 				   0, 0,
@@ -354,6 +352,8 @@ static struct sk_buff *igmpv3_newpack(struct net_device *dev, int size)
 	skb_dst_set(skb, &rt->dst);
 	skb->dev = dev;
 
+	skb->reserved_tailroom = skb_end_offset(skb) -
+				 min(mtu, skb_end_offset(skb));
 	skb_reserve(skb, hlen);
 
 	skb_reset_network_header(skb);
@@ -423,8 +423,7 @@ static struct sk_buff *add_grhead(struct sk_buff *skb, struct ip_mc_list *pmc,
 	return skb;
 }
 
-#define AVAILABLE(skb) ((skb) ? ((skb)->dev ? igmp_skb_size(skb) - (skb)->len : \
-	skb_tailroom(skb)) : 0)
+#define AVAILABLE(skb)	((skb) ? skb_availroom(skb) : 0)
 
 static struct sk_buff *add_grec(struct sk_buff *skb, struct ip_mc_list *pmc,
 	int type, int gdeleted, int sdeleted)
diff --git a/net/ipv6/mcast.c b/net/ipv6/mcast.c
index 9648de2..ed2c4e4 100644
--- a/net/ipv6/mcast.c
+++ b/net/ipv6/mcast.c
@@ -1550,7 +1550,7 @@ static void ip6_mc_hdr(struct sock *sk, struct sk_buff *skb,
 	hdr->daddr = *daddr;
 }
 
-static struct sk_buff *mld_newpack(struct inet6_dev *idev, int size)
+static struct sk_buff *mld_newpack(struct inet6_dev *idev, unsigned int mtu)
 {
 	struct net_device *dev = idev->dev;
 	struct net *net = dev_net(dev);
@@ -1561,13 +1561,13 @@ static struct sk_buff *mld_newpack(struct inet6_dev *idev, int size)
 	const struct in6_addr *saddr;
 	int hlen = LL_RESERVED_SPACE(dev);
 	int tlen = dev->needed_tailroom;
+	unsigned int size = mtu + hlen + tlen;
 	int err;
 	u8 ra[8] = { IPPROTO_ICMPV6, 0,
 		     IPV6_TLV_ROUTERALERT, 2, 0, 0,
 		     IPV6_TLV_PADN, 0 };
 
 	/* we assume size > sizeof(ra) here */
-	size += hlen + tlen;
 	/* limit our allocations to order-0 page */
 	size = min_t(int, size, SKB_MAX_ORDER(0, 0));
 	skb = sock_alloc_send_skb(sk, size, 1, &err);
@@ -1576,6 +1576,8 @@ static struct sk_buff *mld_newpack(struct inet6_dev *idev, int size)
 		return NULL;
 
 	skb->priority = TC_PRIO_CONTROL;
+	skb->reserved_tailroom = skb_end_offset(skb) -
+				 min(mtu, skb_end_offset(skb));
 	skb_reserve(skb, hlen);
 
 	if (__ipv6_get_lladdr(idev, &addr_buf, IFA_F_TENTATIVE)) {
@@ -1690,8 +1692,7 @@ static struct sk_buff *add_grhead(struct sk_buff *skb, struct ifmcaddr6 *pmc,
 	return skb;
 }
 
-#define AVAILABLE(skb) ((skb) ? ((skb)->dev ? (skb)->dev->mtu - (skb)->len : \
-	skb_tailroom(skb)) : 0)
+#define AVAILABLE(skb)	((skb) ? skb_availroom(skb) : 0)
 
 static struct sk_buff *add_grec(struct sk_buff *skb, struct ifmcaddr6 *pmc,
 	int type, int gdeleted, int sdeleted, int crsend)
-- 
1.7.11.7

Re: [PATCH net v4] ipv6: mld: fix add_grhead skb_over_panic for devs with large MTUs

[Eric Dumazet]

On Wed, 2014-11-05 at 20:27 +0100, Daniel Borkmann wrote:
> It has been reported that generating an MLD listener report on
> devices with large MTUs (e.g. 9000) and a high number of IPv6
> addresses can trigger a skb_over_panic():
...
> 
> Reported-by: Wei Liu <lw1a2.jing@gmail.com>
> Fixes: 72e09ad107e7 ("ipv6: avoid high order allocations")
> Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
> Cc: Eric Dumazet <edumazet@google.com>
> Cc: Hannes Frederic Sowa <hannes@stressinduktion.org>
> Cc: David L Stevens <david.stevens@oracle.com>
> ---
>  v3->v4:
>   - Reduced noise from previous one as discussed

Thanks for your patience ;)

Acked-by: Eric Dumazet <edumazet@google.com>

Re: [PATCH net v4] ipv6: mld: fix add_grhead skb_over_panic for devs with large MTUs

[Daniel Borkmann]

On 11/05/2014 09:47 PM, Eric Dumazet wrote:
...
> Thanks for your patience ;)

No problem, thanks for the feedback, Eric!

Re: [PATCH net v4] ipv6: mld: fix add_grhead skb_over_panic for devs with large MTUs

[Cong Wang]

On Wed, Nov 5, 2014 at 11:27 AM, Daniel Borkmann <dborkman@redhat.com> wrote:
> -static struct sk_buff *mld_newpack(struct inet6_dev *idev, int size)
> +static struct sk_buff *mld_newpack(struct inet6_dev *idev, unsigned int mtu)

For net-next, you probably want to get rid of the 'mtu' parameter,
since all callers use dev->mtu. :)

Re: [PATCH net v4] ipv6: mld: fix add_grhead skb_over_panic for devs with large MTUs

[Daniel Borkmann]

On 11/05/2014 11:11 PM, Cong Wang wrote:
> On Wed, Nov 5, 2014 at 11:27 AM, Daniel Borkmann <dborkman@redhat.com> wrote:
>> -static struct sk_buff *mld_newpack(struct inet6_dev *idev, int size)
>> +static struct sk_buff *mld_newpack(struct inet6_dev *idev, unsigned int mtu)
>
> For net-next, you probably want to get rid of the 'mtu' parameter,
> since all callers use dev->mtu. :)

Yeah, feel free. ;) Probably for the longer term it might make sense to
look into ways to refactor and unify some of the more generic portions of
the IGMP/MLD code.

Re: [PATCH net v4] ipv6: mld: fix add_grhead skb_over_panic for devs with large MTUs

[Hannes Frederic Sowa]

On Wed, Nov 5, 2014, at 20:27, Daniel Borkmann wrote:
> It has been reported that generating an MLD listener report on
> devices with large MTUs (e.g. 9000) and a high number of IPv6
> addresses can trigger a skb_over_panic():
> 
> skbuff: skb_over_panic: text:ffffffff80612a5d len:3776 put:20
> head:ffff88046d751000 data:ffff88046d751010 tail:0xed0 end:0xec0
> dev:port1
>  ------------[ cut here ]------------
> kernel BUG at net/core/skbuff.c:100!
> invalid opcode: 0000 [#1] SMP
> Modules linked in: ixgbe(O)
> CPU: 3 PID: 0 Comm: swapper/3 Tainted: G O 3.14.23+ #4
> [...]
> Call Trace:
>  <IRQ>
>  [<ffffffff80578226>] ? skb_put+0x3a/0x3b
>  [<ffffffff80612a5d>] ? add_grhead+0x45/0x8e
>  [<ffffffff80612e3a>] ? add_grec+0x394/0x3d4
>  [<ffffffff80613222>] ? mld_ifc_timer_expire+0x195/0x20d
>  [<ffffffff8061308d>] ? mld_dad_timer_expire+0x45/0x45
>  [<ffffffff80255b5d>] ? call_timer_fn.isra.29+0x12/0x68
>  [<ffffffff80255d16>] ? run_timer_softirq+0x163/0x182
>  [<ffffffff80250e6f>] ? __do_softirq+0xe0/0x21d
>  [<ffffffff8025112b>] ? irq_exit+0x4e/0xd3
>  [<ffffffff802214bb>] ? smp_apic_timer_interrupt+0x3b/0x46
>  [<ffffffff8063f10a>] ? apic_timer_interrupt+0x6a/0x70
> 
> mld_newpack() skb allocations are usually requested with dev->mtu
> in size, since commit 72e09ad107e7 ("ipv6: avoid high order allocations")
> we have changed the limit in order to be less likely to fail.
> 
> However, in MLD/IGMP code, we have some rather ugly AVAILABLE(skb)
> macros, which determine if we may end up doing an skb_put() for
> adding another record. To avoid possible fragmentation, we check
> the skb's tailroom as skb->dev->mtu - skb->len, which is a wrong
> assumption as the actual max allocation size can be much smaller.
> 
> The IGMP case doesn't have this issue as commit 57e1ab6eaddc
> ("igmp: refine skb allocations") stores the allocation size in
> the cb[].
> 
> Set a reserved_tailroom to make it fit into the MTU and use
> skb_availroom() helper instead. This also allows to get rid of
> igmp_skb_size().
> 
> Reported-by: Wei Liu <lw1a2.jing@gmail.com>
> Fixes: 72e09ad107e7 ("ipv6: avoid high order allocations")
> Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
> Cc: Eric Dumazet <edumazet@google.com>
> Cc: Hannes Frederic Sowa <hannes@stressinduktion.org>
> Cc: David L Stevens <david.stevens@oracle.com>

Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>

Thanks and sorry for the back and forth, Daniel!

Re: [PATCH net v4] ipv6: mld: fix add_grhead skb_over_panic for devs with large MTUs

[David Miller]

From: Daniel Borkmann <dborkman@redhat.com>
Date: Wed,  5 Nov 2014 20:27:38 +0100

> It has been reported that generating an MLD listener report on
> devices with large MTUs (e.g. 9000) and a high number of IPv6
> addresses can trigger a skb_over_panic():
 ...
> mld_newpack() skb allocations are usually requested with dev->mtu
> in size, since commit 72e09ad107e7 ("ipv6: avoid high order allocations")
> we have changed the limit in order to be less likely to fail.
> 
> However, in MLD/IGMP code, we have some rather ugly AVAILABLE(skb)
> macros, which determine if we may end up doing an skb_put() for
> adding another record. To avoid possible fragmentation, we check
> the skb's tailroom as skb->dev->mtu - skb->len, which is a wrong
> assumption as the actual max allocation size can be much smaller.
> 
> The IGMP case doesn't have this issue as commit 57e1ab6eaddc
> ("igmp: refine skb allocations") stores the allocation size in
> the cb[].
> 
> Set a reserved_tailroom to make it fit into the MTU and use
> skb_availroom() helper instead. This also allows to get rid of
> igmp_skb_size().
> 
> Reported-by: Wei Liu <lw1a2.jing@gmail.com>
> Fixes: 72e09ad107e7 ("ipv6: avoid high order allocations")
> Signed-off-by: Daniel Borkmann <dborkman@redhat.com>

This has always been a tricky area, applied and queued up for
-stable, thanks everyone.