* [PATCH net] net: use a bounce buffer for copying skb->mark
@ 2023-02-13 16:00 Eric Dumazet
2023-02-14 17:45 ` Alexander Lobakin
2023-02-15 4:40 ` patchwork-bot+netdevbpf
0 siblings, 2 replies; 3+ messages in thread
From: Eric Dumazet @ 2023-02-13 16:00 UTC (permalink / raw)
To: David S . Miller, Jakub Kicinski, Paolo Abeni
Cc: David Ahern, netdev, eric.dumazet, Eric Dumazet, syzbot, Erin MacNeil
syzbot found arm64 builds would crash in sock_recv_mark()
when CONFIG_HARDENED_USERCOPY=y
x86 and powerpc are not detecting the issue because
they define user_access_begin.
This will be handled in a different patch,
because a check_object_size() is missing.
Only data from skb->cb[] can be copied directly to/from user space,
as explained in commit 79a8a642bf05 ("net: Whitelist
the skbuff_head_cache "cb" field")
syzbot report was:
usercopy: Kernel memory exposure attempt detected from SLUB object 'skbuff_head_cache' (offset 168, size 4)!
------------[ cut here ]------------
kernel BUG at mm/usercopy.c:102 !
Internal error: Oops - BUG: 00000000f2000800 [#1] PREEMPT SMP
Modules linked in:
CPU: 0 PID: 4410 Comm: syz-executor533 Not tainted 6.2.0-rc7-syzkaller-17907-g2d3827b3f393 #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/21/2023
pstate: 60400005 (nZCv daif +PAN -UAO -TCO -DIT -SSBS BTYPE=--)
pc : usercopy_abort+0x90/0x94 mm/usercopy.c:90
lr : usercopy_abort+0x90/0x94 mm/usercopy.c:90
sp : ffff80000fb9b9a0
x29: ffff80000fb9b9b0 x28: ffff0000c6073400 x27: 0000000020001a00
x26: 0000000000000014 x25: ffff80000cf52000 x24: fffffc0000000000
x23: 05ffc00000000200 x22: fffffc000324bf80 x21: ffff0000c92fe1a8
x20: 0000000000000001 x19: 0000000000000004 x18: 0000000000000000
x17: 656a626f2042554c x16: ffff0000c6073dd0 x15: ffff80000dbd2118
x14: ffff0000c6073400 x13: 00000000ffffffff x12: ffff0000c6073400
x11: ff808000081bbb4c x10: 0000000000000000 x9 : 7b0572d7cc0ccf00
x8 : 7b0572d7cc0ccf00 x7 : ffff80000bf650d4 x6 : 0000000000000000
x5 : 0000000000000001 x4 : 0000000000000001 x3 : 0000000000000000
x2 : ffff0001fefbff08 x1 : 0000000100000000 x0 : 000000000000006c
Call trace:
usercopy_abort+0x90/0x94 mm/usercopy.c:90
__check_heap_object+0xa8/0x100 mm/slub.c:4761
check_heap_object mm/usercopy.c:196 [inline]
__check_object_size+0x208/0x6b8 mm/usercopy.c:251
check_object_size include/linux/thread_info.h:199 [inline]
__copy_to_user include/linux/uaccess.h:115 [inline]
put_cmsg+0x408/0x464 net/core/scm.c:238
sock_recv_mark net/socket.c:975 [inline]
__sock_recv_cmsgs+0x1fc/0x248 net/socket.c:984
sock_recv_cmsgs include/net/sock.h:2728 [inline]
packet_recvmsg+0x2d8/0x678 net/packet/af_packet.c:3482
____sys_recvmsg+0x110/0x3a0
___sys_recvmsg net/socket.c:2737 [inline]
__sys_recvmsg+0x194/0x210 net/socket.c:2767
__do_sys_recvmsg net/socket.c:2777 [inline]
__se_sys_recvmsg net/socket.c:2774 [inline]
__arm64_sys_recvmsg+0x2c/0x3c net/socket.c:2774
__invoke_syscall arch/arm64/kernel/syscall.c:38 [inline]
invoke_syscall+0x64/0x178 arch/arm64/kernel/syscall.c:52
el0_svc_common+0xbc/0x180 arch/arm64/kernel/syscall.c:142
do_el0_svc+0x48/0x110 arch/arm64/kernel/syscall.c:193
el0_svc+0x58/0x14c arch/arm64/kernel/entry-common.c:637
el0t_64_sync_handler+0x84/0xf0 arch/arm64/kernel/entry-common.c:655
el0t_64_sync+0x190/0x194 arch/arm64/kernel/entry.S:591
Code: 91388800 aa0903e1 f90003e8 94e6d752 (d4210000)
Fixes: 6fd1d51cfa25 ("net: SO_RCVMARK socket option for SO_MARK with recvmsg()")
Reported-by: syzbot <syzkaller@googlegroups.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Erin MacNeil <lnx.erin@gmail.com>
---
net/socket.c | 9 ++++++---
1 file changed, 6 insertions(+), 3 deletions(-)
diff --git a/net/socket.c b/net/socket.c
index 77626e4d96900b946e340e59e6984e5bab672bbe..4080b4ba7daf35a3a0b88a160299f4abd3afc88f 100644
--- a/net/socket.c
+++ b/net/socket.c
@@ -982,9 +982,12 @@ static inline void sock_recv_drops(struct msghdr *msg, struct sock *sk,
static void sock_recv_mark(struct msghdr *msg, struct sock *sk,
struct sk_buff *skb)
{
- if (sock_flag(sk, SOCK_RCVMARK) && skb)
- put_cmsg(msg, SOL_SOCKET, SO_MARK, sizeof(__u32),
- &skb->mark);
+ if (sock_flag(sk, SOCK_RCVMARK) && skb) {
+ /* We must use a bounce buffer for CONFIG_HARDENED_USERCOPY=y */
+ __u32 mark = skb->mark;
+
+ put_cmsg(msg, SOL_SOCKET, SO_MARK, sizeof(__u32), &mark);
+ }
}
void __sock_recv_cmsgs(struct msghdr *msg, struct sock *sk,
--
2.39.1.581.gbfd45094c4-goog
^ permalink raw reply related [flat|nested] 3+ messages in thread
* Re: [PATCH net] net: use a bounce buffer for copying skb->mark
2023-02-13 16:00 [PATCH net] net: use a bounce buffer for copying skb->mark Eric Dumazet
@ 2023-02-14 17:45 ` Alexander Lobakin
2023-02-15 4:40 ` patchwork-bot+netdevbpf
1 sibling, 0 replies; 3+ messages in thread
From: Alexander Lobakin @ 2023-02-14 17:45 UTC (permalink / raw)
To: Eric Dumazet
Cc: David S . Miller, Jakub Kicinski, Paolo Abeni, David Ahern,
netdev, eric.dumazet, syzbot, Erin MacNeil
From: Eric Dumazet <edumazet@google.com>
Date: Mon, 13 Feb 2023 16:00:59 +0000
> syzbot found arm64 builds would crash in sock_recv_mark()
> when CONFIG_HARDENED_USERCOPY=y
>
> x86 and powerpc are not detecting the issue because
> they define user_access_begin.
> This will be handled in a different patch,
> because a check_object_size() is missing.
>
> Only data from skb->cb[] can be copied directly to/from user space,
> as explained in commit 79a8a642bf05 ("net: Whitelist
> the skbuff_head_cache "cb" field")
>
> syzbot report was:
> usercopy: Kernel memory exposure attempt detected from SLUB object 'skbuff_head_cache' (offset 168, size 4)!
> ------------[ cut here ]------------
> kernel BUG at mm/usercopy.c:102 !
> Internal error: Oops - BUG: 00000000f2000800 [#1] PREEMPT SMP
> Modules linked in:
> CPU: 0 PID: 4410 Comm: syz-executor533 Not tainted 6.2.0-rc7-syzkaller-17907-g2d3827b3f393 #0
> Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/21/2023
> pstate: 60400005 (nZCv daif +PAN -UAO -TCO -DIT -SSBS BTYPE=--)
> pc : usercopy_abort+0x90/0x94 mm/usercopy.c:90
> lr : usercopy_abort+0x90/0x94 mm/usercopy.c:90
> sp : ffff80000fb9b9a0
> x29: ffff80000fb9b9b0 x28: ffff0000c6073400 x27: 0000000020001a00
> x26: 0000000000000014 x25: ffff80000cf52000 x24: fffffc0000000000
> x23: 05ffc00000000200 x22: fffffc000324bf80 x21: ffff0000c92fe1a8
> x20: 0000000000000001 x19: 0000000000000004 x18: 0000000000000000
> x17: 656a626f2042554c x16: ffff0000c6073dd0 x15: ffff80000dbd2118
> x14: ffff0000c6073400 x13: 00000000ffffffff x12: ffff0000c6073400
> x11: ff808000081bbb4c x10: 0000000000000000 x9 : 7b0572d7cc0ccf00
> x8 : 7b0572d7cc0ccf00 x7 : ffff80000bf650d4 x6 : 0000000000000000
> x5 : 0000000000000001 x4 : 0000000000000001 x3 : 0000000000000000
> x2 : ffff0001fefbff08 x1 : 0000000100000000 x0 : 000000000000006c
> Call trace:
> usercopy_abort+0x90/0x94 mm/usercopy.c:90
> __check_heap_object+0xa8/0x100 mm/slub.c:4761
> check_heap_object mm/usercopy.c:196 [inline]
> __check_object_size+0x208/0x6b8 mm/usercopy.c:251
> check_object_size include/linux/thread_info.h:199 [inline]
> __copy_to_user include/linux/uaccess.h:115 [inline]
> put_cmsg+0x408/0x464 net/core/scm.c:238
> sock_recv_mark net/socket.c:975 [inline]
> __sock_recv_cmsgs+0x1fc/0x248 net/socket.c:984
> sock_recv_cmsgs include/net/sock.h:2728 [inline]
> packet_recvmsg+0x2d8/0x678 net/packet/af_packet.c:3482
> ____sys_recvmsg+0x110/0x3a0
> ___sys_recvmsg net/socket.c:2737 [inline]
> __sys_recvmsg+0x194/0x210 net/socket.c:2767
> __do_sys_recvmsg net/socket.c:2777 [inline]
> __se_sys_recvmsg net/socket.c:2774 [inline]
> __arm64_sys_recvmsg+0x2c/0x3c net/socket.c:2774
> __invoke_syscall arch/arm64/kernel/syscall.c:38 [inline]
> invoke_syscall+0x64/0x178 arch/arm64/kernel/syscall.c:52
> el0_svc_common+0xbc/0x180 arch/arm64/kernel/syscall.c:142
> do_el0_svc+0x48/0x110 arch/arm64/kernel/syscall.c:193
> el0_svc+0x58/0x14c arch/arm64/kernel/entry-common.c:637
> el0t_64_sync_handler+0x84/0xf0 arch/arm64/kernel/entry-common.c:655
> el0t_64_sync+0x190/0x194 arch/arm64/kernel/entry.S:591
> Code: 91388800 aa0903e1 f90003e8 94e6d752 (d4210000)
>
> Fixes: 6fd1d51cfa25 ("net: SO_RCVMARK socket option for SO_MARK with recvmsg()")
> Reported-by: syzbot <syzkaller@googlegroups.com>
> Signed-off-by: Eric Dumazet <edumazet@google.com>
> Cc: Erin MacNeil <lnx.erin@gmail.com>
Reviewed-by: Alexander Lobakin <alexandr.lobakin@intel.com>
> ---
> net/socket.c | 9 ++++++---
> 1 file changed, 6 insertions(+), 3 deletions(-)
>
> diff --git a/net/socket.c b/net/socket.c
> index 77626e4d96900b946e340e59e6984e5bab672bbe..4080b4ba7daf35a3a0b88a160299f4abd3afc88f 100644
> --- a/net/socket.c
> +++ b/net/socket.c
> @@ -982,9 +982,12 @@ static inline void sock_recv_drops(struct msghdr *msg, struct sock *sk,
> static void sock_recv_mark(struct msghdr *msg, struct sock *sk,
> struct sk_buff *skb)
> {
> - if (sock_flag(sk, SOCK_RCVMARK) && skb)
> - put_cmsg(msg, SOL_SOCKET, SO_MARK, sizeof(__u32),
> - &skb->mark);
> + if (sock_flag(sk, SOCK_RCVMARK) && skb) {
> + /* We must use a bounce buffer for CONFIG_HARDENED_USERCOPY=y */
> + __u32 mark = skb->mark;
> +
> + put_cmsg(msg, SOL_SOCKET, SO_MARK, sizeof(__u32), &mark);
> + }
> }
>
> void __sock_recv_cmsgs(struct msghdr *msg, struct sock *sk,
Thanks,
Olek
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH net] net: use a bounce buffer for copying skb->mark
2023-02-13 16:00 [PATCH net] net: use a bounce buffer for copying skb->mark Eric Dumazet
2023-02-14 17:45 ` Alexander Lobakin
@ 2023-02-15 4:40 ` patchwork-bot+netdevbpf
1 sibling, 0 replies; 3+ messages in thread
From: patchwork-bot+netdevbpf @ 2023-02-15 4:40 UTC (permalink / raw)
To: Eric Dumazet
Cc: davem, kuba, pabeni, dsahern, netdev, eric.dumazet, syzkaller, lnx.erin
Hello:
This patch was applied to netdev/net.git (master)
by Jakub Kicinski <kuba@kernel.org>:
On Mon, 13 Feb 2023 16:00:59 +0000 you wrote:
> syzbot found arm64 builds would crash in sock_recv_mark()
> when CONFIG_HARDENED_USERCOPY=y
>
> x86 and powerpc are not detecting the issue because
> they define user_access_begin.
> This will be handled in a different patch,
> because a check_object_size() is missing.
>
> [...]
Here is the summary with links:
- [net] net: use a bounce buffer for copying skb->mark
https://git.kernel.org/netdev/net/c/2558b8039d05
You are awesome, thank you!
--
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2023-02-15 4:40 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2023-02-13 16:00 [PATCH net] net: use a bounce buffer for copying skb->mark Eric Dumazet
2023-02-14 17:45 ` Alexander Lobakin
2023-02-15 4:40 ` patchwork-bot+netdevbpf
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.