net: bonding: fix use-after-free in bond_xmit_broadcast()
Subsystems (3)
PATCH 1/1 net: bonding: fix use-after-free in bond_xmit_broadcast()
commit 2884bf72fb8f03409e423397319205de48adca16 upstream.
bond_xmit_broadcast() reuses the original skb for the last slave (determined by bond_is_last_slave()) and clones it for others. Concurrent slave enslave/release can mutate the slave list during RCU-protected iteration, changing which slave is "last" mid-loop. This causes the original skb to be double-consumed (double-freed).
Replace the racy bond_is_last_slave() check with a simple index comparison (i + 1 == slaves_count) against the pre-snapshot slave count taken via READ_ONCE() before the loop. This preserves the zero-copy optimization for the last slave while making the "last" determination stable against concurrent list mutations.
The UAF can trigger the following crash:
==================================================================
BUG: KASAN: slab-use-after-free in skb_clone Read of size 8 at addr ffff888100ef8d40 by task exploit/147 CPU: 1 UID: 0 PID: 147 Comm: exploit Not tainted 7.0.0-rc3+ #4 PREEMPTLAZY Call Trace: <TASK> dump_stack_lvl (lib/dump_stack.c:123) print_report (mm/kasan/report.c:379 mm/kasan/report.c:482) kasan_report (mm/kasan/report.c:597) skb_clone (include/linux/skbuff.h:1724 include/linux/skbuff.h:1792 include/linux/skbuff.h:3396 net/core/skbuff.c:2108) bond_xmit_broadcast (drivers/net/bonding/bond_main.c:5334) bond_start_xmit (drivers/net/bonding/bond_main.c:5567 drivers/net/bonding/bond_main.c:5593) dev_hard_start_xmit (include/linux/netdevice.h:5325 include/linux/netdevice.h:5334 net/core/dev.c:3871 net/core/dev.c:3887) __dev_queue_xmit (include/linux/netdevice.h:3601 net/core/dev.c:4838) ip6_finish_output2 (include/net/neighbour.h:540 include/net/neighbour.h:554 net/ipv6/ip6_output.c:136) ip6_finish_output (net/ipv6/ip6_output.c:208 net/ipv6/ip6_output.c:219) ip6_output (net/ipv6/ip6_output.c:250) ip6_send_skb (net/ipv6/ip6_output.c:1985) udp_v6_send_skb (net/ipv6/udp.c:1442) udpv6_sendmsg (net/ipv6/udp.c:1733) __sys_sendto (net/socket.c:730 net/socket.c:742 net/socket.c:2206) __x64_sys_sendto (net/socket.c:2209) do_syscall_64 (arch/x86/entry/syscall_64.c:63 arch/x86/entry/syscall_64.c:94) entry_SYSCALL_64_after_hwframe (arch/x86/entry/entry_64.S:130) </TASK> Allocated by task 147: Freed by task 147: The buggy address belongs to the object at ffff888100ef8c80 which belongs to the cache skbuff_head_cache of size 224
The buggy address is located 192 bytes inside of
freed 224-byte region [ffff888100ef8c80, ffff888100ef8d60)
Memory state around the buggy address:
ffff888100ef8c00: fb fb fb fb fc fc fc fc fc fc fc fc fc fc fc fc
ffff888100ef8c80: fa fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
>ffff888100ef8d00: fb fb fb fb fb fb fb fb fb fb fb fb fc fc fc fc
^
ffff888100ef8d80: fc fc fc fc fc fc fc fc fa fb fb fb fb fb fb fb
ffff888100ef8e00: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
==================================================================[Kevin Berry <[email protected]>: fixed merge conflicts and adapted to 6.12 struct]
--- drivers/net/bonding/bond_main.c | 24 +++++++++++++++++------- 1 file changed, 17 insertions(+), 7 deletions(-)
drivers/net/bonding/bond_main.c +17 -7
@@ -5346,23 +5346,33 @@ static netdev_tx_t bond_3ad_xor_xmit(struct sk_buff *skb,return bond_tx_drop (dev, skb); } -/* in broadcast mode, we send everything to all usable interfaces. */ +/* in broadcast mode, we send everything to all or usable slave interfaces. +* under rcu_read_lock when this function is called. +*/ static netdev_tx_t bond_xmit_broadcast (struct sk_buff *skb, -struct net_device *bond_dev) +struct net_device *bond_dev, +bool all_slaves) {struct bonding *bond =netdev_priv (bond_dev); -struct slave *slave =NULL ; -struct list_head *iter; +struct bond_up_slave *slaves;bool xmit_suc = false;bool skb_used = false; +int slaves_count, i; -bond_for_each_slave_rcu (bond, slave, iter) { +if (all_slaves) + slaves =rcu_dereference (bond->all_slaves); +else + slaves =rcu_dereference (bond->usable_slaves); + + slaves_count = slaves ?READ_ONCE (slaves->count) :0 ; +for (i =0 ; i < slaves_count; i++) { +struct slave *slave = slaves->arr[i];struct sk_buff *skb2;if (!(bond_slave_is_up (slave) && slave->link ==BOND_LINK_UP ))continue ; -if (bond_is_last_slave (bond, slave)) { +if (i +1 == slaves_count) { skb2 = skb; skb_used = true; }else {
@@ -5597,7 +5607,7 @@ static netdev_tx_t __bond_start_xmit(struct sk_buff *skb, struct net_device *devcase BOND_MODE_XOR :return bond_3ad_xor_xmit (skb, dev);case BOND_MODE_BROADCAST : -return bond_xmit_broadcast (skb, dev); +return bond_xmit_broadcast (skb, dev, true);case BOND_MODE_ALB :return bond_alb_xmit (skb, dev);case BOND_MODE_TLB :
Signature · 1 line
2.54.0.rc0.605.g598a273b03-goog
Hi Chenglong,
I’m not entirely clear on how PATCH 6.12.y works and had a question regarding this patch.
Could you please clarify the need for the parameter bool all_slaves? As this feature (was introduced in ce7a381697cb) is not introduced in v6.12, I was wondering if it might be sufficient to apply my original patch instead. I’ve also checked against 6.12.81 and didn’t encounter any conflicts.
Thanks,
Xiang
Quoted 131 lines · Chenglong Tang
commit
2884bf72fb8f03409e423397319205de48adca16upstream.
bond_xmit_broadcast()reuses the original skb for the last slave (determined bybond_is_last_slave()) and clones it for others. Concurrent slave enslave/release can mutate the slave list during RCU-protected iteration, changing which slave is "last" mid-loop. This causes the original skb to be double-consumed (double-freed).Replace the racy
bond_is_last_slave()check with a simple index comparison (i + 1 ==slaves_count) against the pre-snapshot slave count taken viaREAD_ONCE()before the loop. This preserves the zero-copy optimization for the last slave while making the "last" determination stable against concurrent list mutations.The UAF can trigger the following crash:
==================================================================
BUG: KASAN: slab-use-after-free in skb_clone Read of size 8 at addr ffff888100ef8d40 by task exploit/147 CPU: 1 UID: 0 PID: 147 Comm: exploit Not tainted 7.0.0-rc3+ #4 PREEMPTLAZY Call Trace: <TASK> dump_stack_lvl (lib/dump_stack.c:123) print_report (mm/kasan/report.c:379 mm/kasan/report.c:482) kasan_report (mm/kasan/report.c:597) skb_clone (include/linux/skbuff.h:1724 include/linux/skbuff.h:1792 include/linux/skbuff.h:3396 net/core/skbuff.c:2108) bond_xmit_broadcast (drivers/net/bonding/bond_main.c:5334) bond_start_xmit (drivers/net/bonding/bond_main.c:5567 drivers/net/bonding/bond_main.c:5593) dev_hard_start_xmit (include/linux/netdevice.h:5325 include/linux/netdevice.h:5334 net/core/dev.c:3871 net/core/dev.c:3887) __dev_queue_xmit (include/linux/netdevice.h:3601 net/core/dev.c:4838) ip6_finish_output2 (include/net/neighbour.h:540 include/net/neighbour.h:554 net/ipv6/ip6_output.c:136) ip6_finish_output (net/ipv6/ip6_output.c:208 net/ipv6/ip6_output.c:219) ip6_output (net/ipv6/ip6_output.c:250) ip6_send_skb (net/ipv6/ip6_output.c:1985) udp_v6_send_skb (net/ipv6/udp.c:1442) udpv6_sendmsg (net/ipv6/udp.c:1733) __sys_sendto (net/socket.c:730 net/socket.c:742 net/socket.c:2206) __x64_sys_sendto (net/socket.c:2209) do_syscall_64 (arch/x86/entry/syscall_64.c:63 arch/x86/entry/syscall_64.c:94) entry_SYSCALL_64_after_hwframe (arch/x86/entry/entry_64.S:130) </TASK> Allocated by task 147: Freed by task 147: The buggy address belongs to the object at ffff888100ef8c80 which belongs to the cache skbuff_head_cache of size 224The buggy address is located 192 bytes inside of
freed 224-byte region [ffff888100ef8c80,ffff888100ef8d60)Memory state around the buggy address: ffff888100ef8c00: fb fb fb fb fc fc fc fc fc fc fc fc fc fc fc fc ffff888100ef8c80: fa fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb >ffff888100ef8d00: fb fb fb fb fb fb fb fb fb fb fb fb fc fc fc fc ^ ffff888100ef8d80: fc fc fc fc fc fc fc fc fa fb fb fb fb fb fb fb ffff888100ef8e00: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb ==================================================================Fixes 4e5bd03ae346 ("net: bonding: fix bond_xmit_broadcast return value error bug")Reported-by Weiming Shi <[email protected]>Change-Id I2349f4953b5760b7f4a12da583aa779c19c4b59cSigned-off-by Xiang Mei <[email protected]>Signed-off-by Paolo Abeni <[email protected]>[Kevin Berry <[email protected]>: fixed merge conflicts and adapted to 6.12 struct]
Signed-off-by Chenglong Tang <[email protected]>--- drivers/net/bonding/bond_main.c | 24 +++++++++++++++++------- 1 file changed, 17 insertions(+), 7 deletions(-)drivers/net/bonding/bond_main.c +17
-8@@ -5346,23 +5346,33 @@ static netdev_tx_t bond_3ad_xor_xmit(struct sk_buff *skb,return bond_tx_drop (dev, skb); } -/* in broadcast mode, we send everything to all usable interfaces. */ +/* in broadcast mode, we send everything to all or usable slave interfaces. +* under rcu_read_lock when this function is called. +*/ static netdev_tx_t bond_xmit_broadcast (struct sk_buff *skb, -struct net_device *bond_dev) +struct net_device *bond_dev, +bool all_slaves) {struct bonding *bond =netdev_priv (bond_dev); -struct slave *slave =NULL ; -struct list_head *iter; +struct bond_up_slave *slaves;bool xmit_suc = false;bool skb_used = false; +int slaves_count, i; -bond_for_each_slave_rcu (bond, slave, iter) { +if (all_slaves) + slaves =rcu_dereference (bond->all_slaves); +else + slaves =rcu_dereference (bond->usable_slaves); + + slaves_count = slaves ?READ_ONCE (slaves->count) :0 ; +for (i =0 ; i < slaves_count; i++) { +struct slave *slave = slaves->arr[i];struct sk_buff *skb2;if (!(bond_slave_is_up (slave) && slave->link ==BOND_LINK_UP ))continue ; -if (bond_is_last_slave (bond, slave)) { +if (i +1 == slaves_count) { skb2 = skb; skb_used = true; }else {@@ -5597,7 +5607,7 @@ static netdev_tx_t __bond_start_xmit(struct sk_buff *skb, struct net_device *devcase BOND_MODE_XOR :return bond_3ad_xor_xmit (skb, dev);case BOND_MODE_BROADCAST : -return bond_xmit_broadcast (skb, dev); +return bond_xmit_broadcast (skb, dev, true);case BOND_MODE_ALB :return bond_alb_xmit (skb, dev);case BOND_MODE_TLB : --2.54.0.rc0.605.g598a273b03-goog
Hi Xiang,
Quoted 3 lines
I was wondering if it might be sufficient to apply my original patch instead. I’ve also checked against 6.12.81 and didn’t encounter any conflicts.
We tried applying your original patch first, but, while it does apply cleanly, it doesn't compile because the version of bond_xmit_broadcast in the 6.12 kernel doesn't have the slaves_count variable introduced in ce7a381697cb.
Quoted 1 line
Could you please clarify the need for the parameter bool
all_slaves?
We had included the all_slaves parameter to try to reduce the chance of conflicts with any other potential backports from later kernel versions, but I agree that omitting it is probably better.
I've sent an amended version of the patch which just adds slaves_count, uses an explicit loop, and applies your original change. Tested the patch by compiling it.
Thanks,
Kevin
PATCH 1/1 net: bonding: fix use-after-free in bond_xmit_broadcast()
From: Xiang Mei <[email protected]>
[ Upstream commit 2884bf72fb8f03409e423397319205de48adca16 ]
bond_xmit_broadcast() reuses the original skb for the last slave (determined by bond_is_last_slave()) and clones it for others. Concurrent slave enslave/release can mutate the slave list during RCU-protected iteration, changing which slave is "last" mid-loop. This causes the original skb to be double-consumed (double-freed).
Replace the racy bond_is_last_slave() check with a simple index comparison (i + 1 == slaves_count) against the pre-snapshot slave count taken via READ_ONCE() before the loop. This preserves the zero-copy optimization for the last slave while making the "last" determination stable against concurrent list mutations.
The UAF can trigger the following crash:
==================================================================
BUG: KASAN: slab-use-after-free in skb_clone Read of size 8 at addr ffff888100ef8d40 by task exploit/147 CPU: 1 UID: 0 PID: 147 Comm: exploit Not tainted 7.0.0-rc3+ #4 PREEMPTLAZY Call Trace: <TASK> dump_stack_lvl (lib/dump_stack.c:123) print_report (mm/kasan/report.c:379 mm/kasan/report.c:482) kasan_report (mm/kasan/report.c:597) skb_clone (include/linux/skbuff.h:1724 include/linux/skbuff.h:1792 include/linux/skbuff.h:3396 net/core/skbuff.c:2108) bond_xmit_broadcast (drivers/net/bonding/bond_main.c:5334) bond_start_xmit (drivers/net/bonding/bond_main.c:5567 drivers/net/bonding/bond_main.c:5593) dev_hard_start_xmit (include/linux/netdevice.h:5325 include/linux/netdevice.h:5334 net/core/dev.c:3871 net/core/dev.c:3887) __dev_queue_xmit (include/linux/netdevice.h:3601 net/core/dev.c:4838) ip6_finish_output2 (include/net/neighbour.h:540 include/net/neighbour.h:554 net/ipv6/ip6_output.c:136) ip6_finish_output (net/ipv6/ip6_output.c:208 net/ipv6/ip6_output.c:219) ip6_output (net/ipv6/ip6_output.c:250) ip6_send_skb (net/ipv6/ip6_output.c:1985) udp_v6_send_skb (net/ipv6/udp.c:1442) udpv6_sendmsg (net/ipv6/udp.c:1733) __sys_sendto (net/socket.c:730 net/socket.c:742 net/socket.c:2206) __x64_sys_sendto (net/socket.c:2209) do_syscall_64 (arch/x86/entry/syscall_64.c:63 arch/x86/entry/syscall_64.c:94) entry_SYSCALL_64_after_hwframe (arch/x86/entry/entry_64.S:130) </TASK> Allocated by task 147: Freed by task 147: The buggy address belongs to the object at ffff888100ef8c80 which belongs to the cache skbuff_head_cache of size 224
The buggy address is located 192 bytes inside of
freed 224-byte region [ffff888100ef8c80, ffff888100ef8d60)
Memory state around the buggy address:
ffff888100ef8c00: fb fb fb fb fc fc fc fc fc fc fc fc fc fc fc fc
ffff888100ef8c80: fa fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
>ffff888100ef8d00: fb fb fb fb fb fb fb fb fb fb fb fb fc fc fc fc
^
ffff888100ef8d80: fc fc fc fc fc fc fc fc fa fb fb fb fb fb fb fb
ffff888100ef8e00: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
==================================================================--- drivers/net/bonding/bond_main.c | 12 ++++++++---- 1 file changed, 8 insertions(+), 4 deletions(-)
drivers/net/bonding/bond_main.c +8 -4
@@ -5322,18 +5322,22 @@ static netdev_tx_t bond_xmit_broadcast(struct sk_buff *skb, struct net_device *bond_dev) {struct bonding *bond =netdev_priv (bond_dev); -struct slave *slave =NULL ; -struct list_head *iter; +struct bond_up_slave *slaves;bool xmit_suc = false;bool skb_used = false; +int slaves_count, i; -bond_for_each_slave_rcu (bond, slave, iter) { + slaves =rcu_dereference (bond->all_slaves); + + slaves_count = slaves ?READ_ONCE (slaves->count) :0 ; +for (i =0 ; i < slaves_count; i++) { +struct slave *slave = slaves->arr[i];struct sk_buff *skb2;if (!(bond_slave_is_up (slave) && slave->link ==BOND_LINK_UP ))continue ; -if (bond_is_last_slave (bond, slave)) { +if (i +1 == slaves_count) { skb2 = skb; skb_used = true; }else {
base-commit: eefc95626b5cb02ea6268d1ae58237768004a60d
Signature · 1 line
2.54.0.545.g6539524ca2-goog
Hi Kevin,
Thanks for the explanation. I now understand that we are backporting the patch to 6.12 (I didn't realize it couldn't be backported).
I have one concern about your patch. On 6.12, bond->all_slaves is only set by bond_update_slave_arr(), which is gated by bond_mode_can_use_xmit_hash() (i.e., 802.3ad, XOR, TLB, ALB). This does not include BOND_MODE_BROADCAST, and nothing else initializes it. As a result, in bond_xmit_broadcast() on 6.12, rcu_dereference(bond->all_slaves) is NULL, slaves_count is 0, the loop never executes, the skb is freed, and NET_XMIT_DROP is returned, so it drops every packet.
This works in mainline because the array-based iteration was introduced alongside additional changes that ensure all_slaves is valid in broadcast mode. Those supporting changes are not present in 6.12.
To resolve this, we can either backport the prerequisite series or apply a smaller, self-contained fix like the one below:
```
--- a/drivers/net/bonding/bond_main.c
+++ b/drivers/net/bonding/bond_main.c
@@ -5326,14 +5326,21 @@ static netdev_tx_t bond_xmit_broadcast(struct
sk_buff *skb,
struct list_head *iter;
bool xmit_suc = false;
bool skb_used = false;
+ int slaves_count, i = 0;
+ slaves_count = READ_ONCE(bond->slave_cnt);
bond_for_each_slave_rcu(bond, slave, iter) {
struct sk_buff *skb2;
+ bool is_last;
+
+ if (++i > slaves_count)
+ break;
+ is_last = (i == slaves_count);
if (!(bond_slave_is_up(slave) && slave->link == BOND_LINK_UP))
continue;
- if (bond_is_last_slave(bond, slave)) {
+ if (is_last) {
skb2 = skb;
skb_used = true;
} else {
```Please let me know your thoughts.
Thanks,
Xiang
Quoted 111 lines · Kevin Berry
From: Xiang Mei <[email protected]>
[ Upstream commit
2884bf72fb8f03409e423397319205de48adca16]
bond_xmit_broadcast()reuses the original skb for the last slave (determined bybond_is_last_slave()) and clones it for others. Concurrent slave enslave/release can mutate the slave list during RCU-protected iteration, changing which slave is "last" mid-loop. This causes the original skb to be double-consumed (double-freed).Replace the racy
bond_is_last_slave()check with a simple index comparison (i + 1 ==slaves_count) against the pre-snapshot slave count taken viaREAD_ONCE()before the loop. This preserves the zero-copy optimization for the last slave while making the "last" determination stable against concurrent list mutations.The UAF can trigger the following crash:
==================================================================
BUG: KASAN: slab-use-after-free in skb_clone Read of size 8 at addr ffff888100ef8d40 by task exploit/147 CPU: 1 UID: 0 PID: 147 Comm: exploit Not tainted 7.0.0-rc3+ #4 PREEMPTLAZY Call Trace: <TASK> dump_stack_lvl (lib/dump_stack.c:123) print_report (mm/kasan/report.c:379 mm/kasan/report.c:482) kasan_report (mm/kasan/report.c:597) skb_clone (include/linux/skbuff.h:1724 include/linux/skbuff.h:1792 include/linux/skbuff.h:3396 net/core/skbuff.c:2108) bond_xmit_broadcast (drivers/net/bonding/bond_main.c:5334) bond_start_xmit (drivers/net/bonding/bond_main.c:5567 drivers/net/bonding/bond_main.c:5593) dev_hard_start_xmit (include/linux/netdevice.h:5325 include/linux/netdevice.h:5334 net/core/dev.c:3871 net/core/dev.c:3887) __dev_queue_xmit (include/linux/netdevice.h:3601 net/core/dev.c:4838) ip6_finish_output2 (include/net/neighbour.h:540 include/net/neighbour.h:554 net/ipv6/ip6_output.c:136) ip6_finish_output (net/ipv6/ip6_output.c:208 net/ipv6/ip6_output.c:219) ip6_output (net/ipv6/ip6_output.c:250) ip6_send_skb (net/ipv6/ip6_output.c:1985) udp_v6_send_skb (net/ipv6/udp.c:1442) udpv6_sendmsg (net/ipv6/udp.c:1733) __sys_sendto (net/socket.c:730 net/socket.c:742 net/socket.c:2206) __x64_sys_sendto (net/socket.c:2209) do_syscall_64 (arch/x86/entry/syscall_64.c:63 arch/x86/entry/syscall_64.c:94) entry_SYSCALL_64_after_hwframe (arch/x86/entry/entry_64.S:130) </TASK> Allocated by task 147: Freed by task 147: The buggy address belongs to the object at ffff888100ef8c80 which belongs to the cache skbuff_head_cache of size 224The buggy address is located 192 bytes inside of
freed 224-byte region [ffff888100ef8c80,ffff888100ef8d60)Memory state around the buggy address: ffff888100ef8c00: fb fb fb fb fc fc fc fc fc fc fc fc fc fc fc fc ffff888100ef8c80: fa fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb >ffff888100ef8d00: fb fb fb fb fb fb fb fb fb fb fb fb fc fc fc fc ^ ffff888100ef8d80: fc fc fc fc fc fc fc fc fa fb fb fb fb fb fb fb ffff888100ef8e00: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb ==================================================================Fixes 4e5bd03ae346 ("net: bonding: fix bond_xmit_broadcast return value error bug")Reported-by Weiming Shi <[email protected]>Signed-off-by Xiang Mei <[email protected]>Signed-off-by Paolo Abeni <[email protected]>Signed-off-by Sasha Levin <[email protected]>Signed-off-by Kevin Berry <[email protected]>--- drivers/net/bonding/bond_main.c | 12 ++++++++---- 1 file changed, 8 insertions(+), 4 deletions(-)drivers/net/bonding/bond_main.c +8
-4@@ -5322,18 +5322,22 @@ static netdev_tx_t bond_xmit_broadcast(struct sk_buff *skb, struct net_device *bond_dev) {struct bonding *bond =netdev_priv (bond_dev); -struct slave *slave =NULL ; -struct list_head *iter; +struct bond_up_slave *slaves;bool xmit_suc = false;bool skb_used = false; +int slaves_count, i; -bond_for_each_slave_rcu (bond, slave, iter) { + slaves =rcu_dereference (bond->all_slaves); + + slaves_count = slaves ?READ_ONCE (slaves->count) :0 ; +for (i =0 ; i < slaves_count; i++) { +struct slave *slave = slaves->arr[i];struct sk_buff *skb2;if (!(bond_slave_is_up (slave) && slave->link ==BOND_LINK_UP ))continue ; -if (bond_is_last_slave (bond, slave)) { +if (i +1 == slaves_count) { skb2 = skb; skb_used = true; }else {base-commit:
eefc95626b5cb02ea6268d1ae58237768004a60d--
2.54.0.545.g6539524ca2-goog
Hi Xiang,
Thanks for the explanation! I was worried that bond->all_slaves might have been uninitialized, thanks for confirming that.
Your patch looks good to me, minus the formatting.
Thanks,
Kevin