[PATCH net] af_packet: fix vlan_get_tci() vs MSG_PEEK

[PATCH net] af_packet: fix vlan_get_tci() vs MSG_PEEK

[Eric Dumazet]

Blamed commit forgot MSG_PEEK case, allowing a crash [1] as found
by syzbot.

Rework vlan_get_tci() to not touch skb at all,
so that it can be used from many cpus on the same skb.

Add a const qualifier to skb argument.

[1]
skbuff: skb_under_panic: text:ffffffff8a8da482 len:32 put:14 head:ffff88807a1d5800 data:ffff88807a1d5810 tail:0x14 end:0x140 dev:<NULL>
------------[ cut here ]------------
 kernel BUG at net/core/skbuff.c:206 !
Oops: invalid opcode: 0000 [#1] PREEMPT SMP KASAN PTI
CPU: 0 UID: 0 PID: 5880 Comm: syz-executor172 Not tainted 6.13.0-rc3-syzkaller-00762-g9268abe611b0 #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 09/13/2024
 RIP: 0010:skb_panic net/core/skbuff.c:206 [inline]
 RIP: 0010:skb_under_panic+0x14b/0x150 net/core/skbuff.c:216
Code: 0b 8d 48 c7 c6 9e 6c 26 8e 48 8b 54 24 08 8b 0c 24 44 8b 44 24 04 4d 89 e9 50 41 54 41 57 41 56 e8 3a 5a 79 f7 48 83 c4 20 90 <0f> 0b 0f 1f 00 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 f3
RSP: 0018:ffffc90003baf5b8 EFLAGS: 00010286
RAX: 0000000000000087 RBX: dffffc0000000000 RCX: 8565c1eec37aa000
RDX: 0000000000000000 RSI: 0000000080000000 RDI: 0000000000000000
RBP: ffff88802616fb50 R08: ffffffff817f0a4c R09: 1ffff92000775e50
R10: dffffc0000000000 R11: fffff52000775e51 R12: 0000000000000140
R13: ffff88807a1d5800 R14: ffff88807a1d5810 R15: 0000000000000014
FS:  00007fa03261f6c0(0000) GS:ffff8880b8600000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007ffd65753000 CR3: 0000000031720000 CR4: 00000000003526f0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
Call Trace:
 <TASK>
  skb_push+0xe5/0x100 net/core/skbuff.c:2636
  vlan_get_tci+0x272/0x550 net/packet/af_packet.c:565
  packet_recvmsg+0x13c9/0x1ef0 net/packet/af_packet.c:3616
  sock_recvmsg_nosec net/socket.c:1044 [inline]
  sock_recvmsg+0x22f/0x280 net/socket.c:1066
  ____sys_recvmsg+0x1c6/0x480 net/socket.c:2814
  ___sys_recvmsg net/socket.c:2856 [inline]
  do_recvmmsg+0x426/0xab0 net/socket.c:2951
  __sys_recvmmsg net/socket.c:3025 [inline]
  __do_sys_recvmmsg net/socket.c:3048 [inline]
  __se_sys_recvmmsg net/socket.c:3041 [inline]
  __x64_sys_recvmmsg+0x199/0x250 net/socket.c:3041
  do_syscall_x64 arch/x86/entry/common.c:52 [inline]
  do_syscall_64+0xf3/0x230 arch/x86/entry/common.c:83

Fixes: 79eecf631c14 ("af_packet: Handle outgoing VLAN packets without hardware offloading")
Reported-by: syzbot+8400677f3fd43f37d3bc@syzkaller.appspotmail.com
Closes: https://lore.kernel.org/netdev/6772c485.050a0220.2f3838.04c6.GAE@google.com/T/#u
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Chengen Du <chengen.du@canonical.com>
Cc: Willem de Bruijn <willemb@google.com>
---
 net/packet/af_packet.c | 12 +++---------
 1 file changed, 3 insertions(+), 9 deletions(-)

diff --git a/net/packet/af_packet.c b/net/packet/af_packet.c
index 886c0dd47b66210e4bbaf8e78f1778c2d25b896e..e2e34a49e98dedfed855ccb004ca57ef18626ddf 100644
--- a/net/packet/af_packet.c
+++ b/net/packet/af_packet.c
@@ -538,10 +538,8 @@ static void *packet_current_frame(struct packet_sock *po,
 	return packet_lookup_frame(po, rb, rb->head, status);
 }
 
-static u16 vlan_get_tci(struct sk_buff *skb, struct net_device *dev)
+static u16 vlan_get_tci(const struct sk_buff *skb, struct net_device *dev)
 {
-	u8 *skb_orig_data = skb->data;
-	int skb_orig_len = skb->len;
 	struct vlan_hdr vhdr, *vh;
 	unsigned int header_len;
 
@@ -562,12 +560,8 @@ static u16 vlan_get_tci(struct sk_buff *skb, struct net_device *dev)
 	else
 		return 0;
 
-	skb_push(skb, skb->data - skb_mac_header(skb));
-	vh = skb_header_pointer(skb, header_len, sizeof(vhdr), &vhdr);
-	if (skb_orig_data != skb->data) {
-		skb->data = skb_orig_data;
-		skb->len = skb_orig_len;
-	}
+	vh = skb_header_pointer(skb, skb_mac_offset(skb) + header_len,
+				sizeof(vhdr), &vhdr);
 	if (unlikely(!vh))
 		return 0;
 
-- 
2.47.1.613.gc27f4b7a9f-goog

[PATCH net] af_packet: fix vlan_get_protocol_dgram() vs MSG_PEEK

[Eric Dumazet]

Blamed commit forgot MSG_PEEK case, allowing a crash [1] as found
by syzbot.

Rework vlan_get_protocol_dgram() to not touch skb at all,
so that it can be used from many cpus on the same skb.

Add a const qualifier to skb argument.

[1]
skbuff: skb_under_panic: text:ffffffff8a8ccd05 len:29 put:14 head:ffff88807fc8e400 data:ffff88807fc8e3f4 tail:0x11 end:0x140 dev:<NULL>
------------[ cut here ]------------
 kernel BUG at net/core/skbuff.c:206 !
Oops: invalid opcode: 0000 [#1] PREEMPT SMP KASAN PTI
CPU: 1 UID: 0 PID: 5892 Comm: syz-executor883 Not tainted 6.13.0-rc4-syzkaller-00054-gd6ef8b40d075 #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 09/13/2024
 RIP: 0010:skb_panic net/core/skbuff.c:206 [inline]
 RIP: 0010:skb_under_panic+0x14b/0x150 net/core/skbuff.c:216
Code: 0b 8d 48 c7 c6 86 d5 25 8e 48 8b 54 24 08 8b 0c 24 44 8b 44 24 04 4d 89 e9 50 41 54 41 57 41 56 e8 5a 69 79 f7 48 83 c4 20 90 <0f> 0b 0f 1f 00 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 f3
RSP: 0018:ffffc900038d7638 EFLAGS: 00010282
RAX: 0000000000000087 RBX: dffffc0000000000 RCX: 609ffd18ea660600
RDX: 0000000000000000 RSI: 0000000080000000 RDI: 0000000000000000
RBP: ffff88802483c8d0 R08: ffffffff817f0a8c R09: 1ffff9200071ae60
R10: dffffc0000000000 R11: fffff5200071ae61 R12: 0000000000000140
R13: ffff88807fc8e400 R14: ffff88807fc8e3f4 R15: 0000000000000011
FS:  00007fbac5e006c0(0000) GS:ffff8880b8700000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007fbac5e00d58 CR3: 000000001238e000 CR4: 00000000003526f0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
Call Trace:
 <TASK>
  skb_push+0xe5/0x100 net/core/skbuff.c:2636
  vlan_get_protocol_dgram+0x165/0x290 net/packet/af_packet.c:585
  packet_recvmsg+0x948/0x1ef0 net/packet/af_packet.c:3552
  sock_recvmsg_nosec net/socket.c:1033 [inline]
  sock_recvmsg+0x22f/0x280 net/socket.c:1055
  ____sys_recvmsg+0x1c6/0x480 net/socket.c:2803
  ___sys_recvmsg net/socket.c:2845 [inline]
  do_recvmmsg+0x426/0xab0 net/socket.c:2940
  __sys_recvmmsg net/socket.c:3014 [inline]
  __do_sys_recvmmsg net/socket.c:3037 [inline]
  __se_sys_recvmmsg net/socket.c:3030 [inline]
  __x64_sys_recvmmsg+0x199/0x250 net/socket.c:3030
  do_syscall_x64 arch/x86/entry/common.c:52 [inline]
  do_syscall_64+0xf3/0x230 arch/x86/entry/common.c:83
 entry_SYSCALL_64_after_hwframe+0x77/0x7f

Fixes: 79eecf631c14 ("af_packet: Handle outgoing VLAN packets without hardware offloading")
Reported-by: syzbot+74f70bb1cb968bf09e4f@syzkaller.appspotmail.com
Closes: https://lore.kernel.org/netdev/6772c485.050a0220.2f3838.04c5.GAE@google.com/T/#u
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Chengen Du <chengen.du@canonical.com>
Cc: Willem de Bruijn <willemb@google.com>
---
 include/linux/if_vlan.h | 16 +++++++++++++---
 net/packet/af_packet.c  | 16 ++++------------
 2 files changed, 17 insertions(+), 15 deletions(-)

diff --git a/include/linux/if_vlan.h b/include/linux/if_vlan.h
index c1645c86eed9693454398f566a446b6dc6fa37f4..d65b5d71b93bf8e051d361ab4abc78d2a48d7904 100644
--- a/include/linux/if_vlan.h
+++ b/include/linux/if_vlan.h
@@ -585,13 +585,16 @@ static inline int vlan_get_tag(const struct sk_buff *skb, u16 *vlan_tci)
  * vlan_get_protocol - get protocol EtherType.
  * @skb: skbuff to query
  * @type: first vlan protocol
+ * @mac_offset: MAC offset
  * @depth: buffer to store length of eth and vlan tags in bytes
  *
  * Returns the EtherType of the packet, regardless of whether it is
  * vlan encapsulated (normal or hardware accelerated) or not.
  */
-static inline __be16 __vlan_get_protocol(const struct sk_buff *skb, __be16 type,
-					 int *depth)
+static inline __be16 __vlan_get_protocol_offset(const struct sk_buff *skb,
+						__be16 type,
+						int mac_offset,
+						int *depth)
 {
 	unsigned int vlan_depth = skb->mac_len, parse_depth = VLAN_MAX_DEPTH;
 
@@ -610,7 +613,8 @@ static inline __be16 __vlan_get_protocol(const struct sk_buff *skb, __be16 type,
 		do {
 			struct vlan_hdr vhdr, *vh;
 
-			vh = skb_header_pointer(skb, vlan_depth, sizeof(vhdr), &vhdr);
+			vh = skb_header_pointer(skb, mac_offset + vlan_depth,
+						sizeof(vhdr), &vhdr);
 			if (unlikely(!vh || !--parse_depth))
 				return 0;
 
@@ -625,6 +629,12 @@ static inline __be16 __vlan_get_protocol(const struct sk_buff *skb, __be16 type,
 	return type;
 }
 
+static inline __be16 __vlan_get_protocol(const struct sk_buff *skb, __be16 type,
+					 int *depth)
+{
+	return __vlan_get_protocol_offset(skb, type, 0, depth);
+}
+
 /**
  * vlan_get_protocol - get protocol EtherType.
  * @skb: skbuff to query
diff --git a/net/packet/af_packet.c b/net/packet/af_packet.c
index e2e34a49e98dedfed855ccb004ca57ef18626ddf..2d73769d67f47c18db75cf43d67d8820350ffe74 100644
--- a/net/packet/af_packet.c
+++ b/net/packet/af_packet.c
@@ -568,21 +568,13 @@ static u16 vlan_get_tci(const struct sk_buff *skb, struct net_device *dev)
 	return ntohs(vh->h_vlan_TCI);
 }
 
-static __be16 vlan_get_protocol_dgram(struct sk_buff *skb)
+static __be16 vlan_get_protocol_dgram(const struct sk_buff *skb)
 {
 	__be16 proto = skb->protocol;
 
-	if (unlikely(eth_type_vlan(proto))) {
-		u8 *skb_orig_data = skb->data;
-		int skb_orig_len = skb->len;
-
-		skb_push(skb, skb->data - skb_mac_header(skb));
-		proto = __vlan_get_protocol(skb, proto, NULL);
-		if (skb_orig_data != skb->data) {
-			skb->data = skb_orig_data;
-			skb->len = skb_orig_len;
-		}
-	}
+	if (unlikely(eth_type_vlan(proto)))
+		proto = __vlan_get_protocol_offset(skb, proto,
+						   skb_mac_offset(skb), NULL);
 
 	return proto;
 }
-- 
2.47.1.613.gc27f4b7a9f-goog

Re: [PATCH net] af_packet: fix vlan_get_tci() vs MSG_PEEK

[Willem de Bruijn]

Eric Dumazet wrote:
> Blamed commit forgot MSG_PEEK case, allowing a crash [1] as found
> by syzbot.
> 
> Rework vlan_get_tci() to not touch skb at all,
> so that it can be used from many cpus on the same skb.
> 
> Add a const qualifier to skb argument.
> 
> [1]
> skbuff: skb_under_panic: text:ffffffff8a8da482 len:32 put:14 head:ffff88807a1d5800 data:ffff88807a1d5810 tail:0x14 end:0x140 dev:<NULL>
> ------------[ cut here ]------------
>  kernel BUG at net/core/skbuff.c:206 !
> Oops: invalid opcode: 0000 [#1] PREEMPT SMP KASAN PTI
> CPU: 0 UID: 0 PID: 5880 Comm: syz-executor172 Not tainted 6.13.0-rc3-syzkaller-00762-g9268abe611b0 #0
> Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 09/13/2024
>  RIP: 0010:skb_panic net/core/skbuff.c:206 [inline]
>  RIP: 0010:skb_under_panic+0x14b/0x150 net/core/skbuff.c:216
> Code: 0b 8d 48 c7 c6 9e 6c 26 8e 48 8b 54 24 08 8b 0c 24 44 8b 44 24 04 4d 89 e9 50 41 54 41 57 41 56 e8 3a 5a 79 f7 48 83 c4 20 90 <0f> 0b 0f 1f 00 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 f3
> RSP: 0018:ffffc90003baf5b8 EFLAGS: 00010286
> RAX: 0000000000000087 RBX: dffffc0000000000 RCX: 8565c1eec37aa000
> RDX: 0000000000000000 RSI: 0000000080000000 RDI: 0000000000000000
> RBP: ffff88802616fb50 R08: ffffffff817f0a4c R09: 1ffff92000775e50
> R10: dffffc0000000000 R11: fffff52000775e51 R12: 0000000000000140
> R13: ffff88807a1d5800 R14: ffff88807a1d5810 R15: 0000000000000014
> FS:  00007fa03261f6c0(0000) GS:ffff8880b8600000(0000) knlGS:0000000000000000
> CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> CR2: 00007ffd65753000 CR3: 0000000031720000 CR4: 00000000003526f0
> DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
> DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
> Call Trace:
>  <TASK>
>   skb_push+0xe5/0x100 net/core/skbuff.c:2636
>   vlan_get_tci+0x272/0x550 net/packet/af_packet.c:565
>   packet_recvmsg+0x13c9/0x1ef0 net/packet/af_packet.c:3616
>   sock_recvmsg_nosec net/socket.c:1044 [inline]
>   sock_recvmsg+0x22f/0x280 net/socket.c:1066
>   ____sys_recvmsg+0x1c6/0x480 net/socket.c:2814
>   ___sys_recvmsg net/socket.c:2856 [inline]
>   do_recvmmsg+0x426/0xab0 net/socket.c:2951
>   __sys_recvmmsg net/socket.c:3025 [inline]
>   __do_sys_recvmmsg net/socket.c:3048 [inline]
>   __se_sys_recvmmsg net/socket.c:3041 [inline]
>   __x64_sys_recvmmsg+0x199/0x250 net/socket.c:3041
>   do_syscall_x64 arch/x86/entry/common.c:52 [inline]
>   do_syscall_64+0xf3/0x230 arch/x86/entry/common.c:83
> 
> Fixes: 79eecf631c14 ("af_packet: Handle outgoing VLAN packets without hardware offloading")
> Reported-by: syzbot+8400677f3fd43f37d3bc@syzkaller.appspotmail.com
> Closes: https://lore.kernel.org/netdev/6772c485.050a0220.2f3838.04c6.GAE@google.com/T/#u
> Signed-off-by: Eric Dumazet <edumazet@google.com>
> Cc: Chengen Du <chengen.du@canonical.com>
> Cc: Willem de Bruijn <willemb@google.com>

Reviewed-by: Willem de Bruijn <willemb@google.com>

Re: [PATCH net] af_packet: fix vlan_get_protocol_dgram() vs MSG_PEEK

[Willem de Bruijn]

Eric Dumazet wrote:
> Blamed commit forgot MSG_PEEK case, allowing a crash [1] as found
> by syzbot.
> 
> Rework vlan_get_protocol_dgram() to not touch skb at all,
> so that it can be used from many cpus on the same skb.
> 
> Add a const qualifier to skb argument.
> 
> [1]
> skbuff: skb_under_panic: text:ffffffff8a8ccd05 len:29 put:14 head:ffff88807fc8e400 data:ffff88807fc8e3f4 tail:0x11 end:0x140 dev:<NULL>
> ------------[ cut here ]------------
>  kernel BUG at net/core/skbuff.c:206 !
> Oops: invalid opcode: 0000 [#1] PREEMPT SMP KASAN PTI
> CPU: 1 UID: 0 PID: 5892 Comm: syz-executor883 Not tainted 6.13.0-rc4-syzkaller-00054-gd6ef8b40d075 #0
> Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 09/13/2024
>  RIP: 0010:skb_panic net/core/skbuff.c:206 [inline]
>  RIP: 0010:skb_under_panic+0x14b/0x150 net/core/skbuff.c:216
> Code: 0b 8d 48 c7 c6 86 d5 25 8e 48 8b 54 24 08 8b 0c 24 44 8b 44 24 04 4d 89 e9 50 41 54 41 57 41 56 e8 5a 69 79 f7 48 83 c4 20 90 <0f> 0b 0f 1f 00 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 f3
> RSP: 0018:ffffc900038d7638 EFLAGS: 00010282
> RAX: 0000000000000087 RBX: dffffc0000000000 RCX: 609ffd18ea660600
> RDX: 0000000000000000 RSI: 0000000080000000 RDI: 0000000000000000
> RBP: ffff88802483c8d0 R08: ffffffff817f0a8c R09: 1ffff9200071ae60
> R10: dffffc0000000000 R11: fffff5200071ae61 R12: 0000000000000140
> R13: ffff88807fc8e400 R14: ffff88807fc8e3f4 R15: 0000000000000011
> FS:  00007fbac5e006c0(0000) GS:ffff8880b8700000(0000) knlGS:0000000000000000
> CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> CR2: 00007fbac5e00d58 CR3: 000000001238e000 CR4: 00000000003526f0
> DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
> DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
> Call Trace:
>  <TASK>
>   skb_push+0xe5/0x100 net/core/skbuff.c:2636
>   vlan_get_protocol_dgram+0x165/0x290 net/packet/af_packet.c:585
>   packet_recvmsg+0x948/0x1ef0 net/packet/af_packet.c:3552
>   sock_recvmsg_nosec net/socket.c:1033 [inline]
>   sock_recvmsg+0x22f/0x280 net/socket.c:1055
>   ____sys_recvmsg+0x1c6/0x480 net/socket.c:2803
>   ___sys_recvmsg net/socket.c:2845 [inline]
>   do_recvmmsg+0x426/0xab0 net/socket.c:2940
>   __sys_recvmmsg net/socket.c:3014 [inline]
>   __do_sys_recvmmsg net/socket.c:3037 [inline]
>   __se_sys_recvmmsg net/socket.c:3030 [inline]
>   __x64_sys_recvmmsg+0x199/0x250 net/socket.c:3030
>   do_syscall_x64 arch/x86/entry/common.c:52 [inline]
>   do_syscall_64+0xf3/0x230 arch/x86/entry/common.c:83
>  entry_SYSCALL_64_after_hwframe+0x77/0x7f
> 
> Fixes: 79eecf631c14 ("af_packet: Handle outgoing VLAN packets without hardware offloading")
> Reported-by: syzbot+74f70bb1cb968bf09e4f@syzkaller.appspotmail.com
> Closes: https://lore.kernel.org/netdev/6772c485.050a0220.2f3838.04c5.GAE@google.com/T/#u
> Signed-off-by: Eric Dumazet <edumazet@google.com>
> Cc: Chengen Du <chengen.du@canonical.com>
> Cc: Willem de Bruijn <willemb@google.com>

Reviewed-by: Willem de Bruijn <willemb@google.com>

Re: [PATCH net] af_packet: fix vlan_get_tci() vs MSG_PEEK

[patchwork-bot+netdevbpf]

Hello:

This patch was applied to netdev/net.git (main)
by Jakub Kicinski <kuba@kernel.org>:

On Mon, 30 Dec 2024 16:10:03 +0000 you wrote:
> Blamed commit forgot MSG_PEEK case, allowing a crash [1] as found
> by syzbot.
> 
> Rework vlan_get_tci() to not touch skb at all,
> so that it can be used from many cpus on the same skb.
> 
> Add a const qualifier to skb argument.
> 
> [...]

Here is the summary with links:
  - [net] af_packet: fix vlan_get_tci() vs MSG_PEEK
    https://git.kernel.org/netdev/net/c/77ee7a6d16b6

You are awesome, thank you!
-- 
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html

Re: [PATCH net] af_packet: fix vlan_get_protocol_dgram() vs MSG_PEEK

[patchwork-bot+netdevbpf]

Hello:

This patch was applied to netdev/net.git (main)
by Jakub Kicinski <kuba@kernel.org>:

On Mon, 30 Dec 2024 16:10:04 +0000 you wrote:
> Blamed commit forgot MSG_PEEK case, allowing a crash [1] as found
> by syzbot.
> 
> Rework vlan_get_protocol_dgram() to not touch skb at all,
> so that it can be used from many cpus on the same skb.
> 
> Add a const qualifier to skb argument.
> 
> [...]

Here is the summary with links:
  - [net] af_packet: fix vlan_get_protocol_dgram() vs MSG_PEEK
    https://git.kernel.org/netdev/net/c/f91a5b808938

You are awesome, thank you!
-- 
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html