2019-05-19 15:08:20 +03:00
// SPDX-License-Identifier: GPL-2.0-only
2016-10-21 13:03:44 +03:00
# include <linux/module.h>
# include <linux/inet_diag.h>
# include <linux/sock_diag.h>
2016-10-25 18:53:22 +03:00
# include <net/inet_sock.h>
2016-10-21 13:03:44 +03:00
# include <net/raw.h>
# include <net/rawv6.h>
# ifdef pr_fmt
# undef pr_fmt
# endif
# define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
static struct raw_hashinfo *
raw_get_hashinfo ( const struct inet_diag_req_v2 * r )
{
if ( r - > sdiag_family = = AF_INET ) {
return & raw_v4_hashinfo ;
# if IS_ENABLED(CONFIG_IPV6)
} else if ( r - > sdiag_family = = AF_INET6 ) {
return & raw_v6_hashinfo ;
# endif
} else {
return ERR_PTR ( - EINVAL ) ;
}
}
/*
* Due to requirement of not breaking user API we can ' t simply
* rename @ pad field in inet_diag_req_v2 structure , instead
* use helper to figure it out .
*/
2023-03-16 18:32:02 +03:00
static bool raw_lookup ( struct net * net , const struct sock * sk ,
2022-06-18 06:47:04 +03:00
const struct inet_diag_req_v2 * req )
2016-10-21 13:03:44 +03:00
{
struct inet_diag_req_raw * r = ( void * ) req ;
if ( r - > sdiag_family = = AF_INET )
2022-06-18 06:47:04 +03:00
return raw_v4_match ( net , sk , r - > sdiag_raw_protocol ,
r - > id . idiag_dst [ 0 ] ,
r - > id . idiag_src [ 0 ] ,
r - > id . idiag_if , 0 ) ;
2016-10-21 13:03:44 +03:00
# if IS_ENABLED(CONFIG_IPV6)
else
2022-06-18 06:47:04 +03:00
return raw_v6_match ( net , sk , r - > sdiag_raw_protocol ,
( const struct in6_addr * ) r - > id . idiag_src ,
( const struct in6_addr * ) r - > id . idiag_dst ,
r - > id . idiag_if , 0 ) ;
2016-10-21 13:03:44 +03:00
# endif
2022-06-18 06:47:04 +03:00
return false ;
2016-10-21 13:03:44 +03:00
}
static struct sock * raw_sock_get ( struct net * net , const struct inet_diag_req_v2 * r )
{
struct raw_hashinfo * hashinfo = raw_get_hashinfo ( r ) ;
raw: Fix NULL deref in raw_get_next().
Dae R. Jeong reported a NULL deref in raw_get_next() [0].
It seems that the repro was running these sequences in parallel so
that one thread was iterating on a socket that was being freed in
another netns.
unshare(0x40060200)
r0 = syz_open_procfs(0x0, &(0x7f0000002080)='net/raw\x00')
socket$inet_icmp_raw(0x2, 0x3, 0x1)
pread64(r0, &(0x7f0000000000)=""/10, 0xa, 0x10000000007f)
After commit 0daf07e52709 ("raw: convert raw sockets to RCU"), we
use RCU and hlist_nulls_for_each_entry() to iterate over SOCK_RAW
sockets. However, we should use spinlock for slow paths to avoid
the NULL deref.
Also, SOCK_RAW does not use SLAB_TYPESAFE_BY_RCU, and the slab object
is not reused during iteration in the grace period. In fact, the
lockless readers do not check the nulls marker with get_nulls_value().
So, SOCK_RAW should use hlist instead of hlist_nulls.
Instead of adding an unnecessary barrier by sk_nulls_for_each_rcu(),
let's convert hlist_nulls to hlist and use sk_for_each_rcu() for
fast paths and sk_for_each() and spinlock for /proc/net/raw.
[0]:
general protection fault, probably for non-canonical address 0xdffffc0000000005: 0000 [#1] PREEMPT SMP KASAN
KASAN: null-ptr-deref in range [0x0000000000000028-0x000000000000002f]
CPU: 2 PID: 20952 Comm: syz-executor.0 Not tainted 6.2.0-g048ec869bafd-dirty #7
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.14.0-0-g155821a1990b-prebuilt.qemu.org 04/01/2014
RIP: 0010:read_pnet include/net/net_namespace.h:383 [inline]
RIP: 0010:sock_net include/net/sock.h:649 [inline]
RIP: 0010:raw_get_next net/ipv4/raw.c:974 [inline]
RIP: 0010:raw_get_idx net/ipv4/raw.c:986 [inline]
RIP: 0010:raw_seq_start+0x431/0x800 net/ipv4/raw.c:995
Code: ef e8 33 3d 94 f7 49 8b 6d 00 4c 89 ef e8 b7 65 5f f7 49 89 ed 49 83 c5 98 0f 84 9a 00 00 00 48 83 c5 c8 48 89 e8 48 c1 e8 03 <42> 80 3c 30 00 74 08 48 89 ef e8 00 3d 94 f7 4c 8b 7d 00 48 89 ef
RSP: 0018:ffffc9001154f9b0 EFLAGS: 00010206
RAX: 0000000000000005 RBX: 1ffff1100302c8fd RCX: 0000000000000000
RDX: 0000000000000028 RSI: ffffc9001154f988 RDI: ffffc9000f77a338
RBP: 0000000000000029 R08: ffffffff8a50ffb4 R09: fffffbfff24b6bd9
R10: fffffbfff24b6bd9 R11: 0000000000000000 R12: ffff88801db73b78
R13: fffffffffffffff9 R14: dffffc0000000000 R15: 0000000000000030
FS: 00007f843ae8e700(0000) GS:ffff888063700000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 000055bb9614b35f CR3: 000000003c672000 CR4: 00000000003506e0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
Call Trace:
<TASK>
seq_read_iter+0x4c6/0x10f0 fs/seq_file.c:225
seq_read+0x224/0x320 fs/seq_file.c:162
pde_read fs/proc/inode.c:316 [inline]
proc_reg_read+0x23f/0x330 fs/proc/inode.c:328
vfs_read+0x31e/0xd30 fs/read_write.c:468
ksys_pread64 fs/read_write.c:665 [inline]
__do_sys_pread64 fs/read_write.c:675 [inline]
__se_sys_pread64 fs/read_write.c:672 [inline]
__x64_sys_pread64+0x1e9/0x280 fs/read_write.c:672
do_syscall_x64 arch/x86/entry/common.c:51 [inline]
do_syscall_64+0x4e/0xa0 arch/x86/entry/common.c:82
entry_SYSCALL_64_after_hwframe+0x63/0xcd
RIP: 0033:0x478d29
Code: f7 d8 64 89 02 b8 ff ff ff ff c3 66 0f 1f 44 00 00 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 bc ff ff ff f7 d8 64 89 01 48
RSP: 002b:00007f843ae8dbe8 EFLAGS: 00000246 ORIG_RAX: 0000000000000011
RAX: ffffffffffffffda RBX: 0000000000791408 RCX: 0000000000478d29
RDX: 000000000000000a RSI: 0000000020000000 RDI: 0000000000000003
RBP: 00000000f477909a R08: 0000000000000000 R09: 0000000000000000
R10: 000010000000007f R11: 0000000000000246 R12: 0000000000791740
R13: 0000000000791414 R14: 0000000000791408 R15: 00007ffc2eb48a50
</TASK>
Modules linked in:
---[ end trace 0000000000000000 ]---
RIP: 0010:read_pnet include/net/net_namespace.h:383 [inline]
RIP: 0010:sock_net include/net/sock.h:649 [inline]
RIP: 0010:raw_get_next net/ipv4/raw.c:974 [inline]
RIP: 0010:raw_get_idx net/ipv4/raw.c:986 [inline]
RIP: 0010:raw_seq_start+0x431/0x800 net/ipv4/raw.c:995
Code: ef e8 33 3d 94 f7 49 8b 6d 00 4c 89 ef e8 b7 65 5f f7 49 89 ed 49 83 c5 98 0f 84 9a 00 00 00 48 83 c5 c8 48 89 e8 48 c1 e8 03 <42> 80 3c 30 00 74 08 48 89 ef e8 00 3d 94 f7 4c 8b 7d 00 48 89 ef
RSP: 0018:ffffc9001154f9b0 EFLAGS: 00010206
RAX: 0000000000000005 RBX: 1ffff1100302c8fd RCX: 0000000000000000
RDX: 0000000000000028 RSI: ffffc9001154f988 RDI: ffffc9000f77a338
RBP: 0000000000000029 R08: ffffffff8a50ffb4 R09: fffffbfff24b6bd9
R10: fffffbfff24b6bd9 R11: 0000000000000000 R12: ffff88801db73b78
R13: fffffffffffffff9 R14: dffffc0000000000 R15: 0000000000000030
FS: 00007f843ae8e700(0000) GS:ffff888063700000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007f92ff166000 CR3: 000000003c672000 CR4: 00000000003506e0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
Fixes: 0daf07e52709 ("raw: convert raw sockets to RCU")
Reported-by: syzbot <syzkaller@googlegroups.com>
Reported-by: Dae R. Jeong <threeearcat@gmail.com>
Link: https://lore.kernel.org/netdev/ZCA2mGV_cmq7lIfV@dragonet/
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2023-04-03 22:49:58 +03:00
struct hlist_head * hlist ;
2022-06-18 06:47:04 +03:00
struct sock * sk ;
2016-10-21 13:03:44 +03:00
int slot ;
if ( IS_ERR ( hashinfo ) )
return ERR_CAST ( hashinfo ) ;
2022-06-18 06:47:05 +03:00
rcu_read_lock ( ) ;
2016-10-21 13:03:44 +03:00
for ( slot = 0 ; slot < RAW_HTABLE_SIZE ; slot + + ) {
2022-06-18 06:47:05 +03:00
hlist = & hashinfo - > ht [ slot ] ;
raw: Fix NULL deref in raw_get_next().
Dae R. Jeong reported a NULL deref in raw_get_next() [0].
It seems that the repro was running these sequences in parallel so
that one thread was iterating on a socket that was being freed in
another netns.
unshare(0x40060200)
r0 = syz_open_procfs(0x0, &(0x7f0000002080)='net/raw\x00')
socket$inet_icmp_raw(0x2, 0x3, 0x1)
pread64(r0, &(0x7f0000000000)=""/10, 0xa, 0x10000000007f)
After commit 0daf07e52709 ("raw: convert raw sockets to RCU"), we
use RCU and hlist_nulls_for_each_entry() to iterate over SOCK_RAW
sockets. However, we should use spinlock for slow paths to avoid
the NULL deref.
Also, SOCK_RAW does not use SLAB_TYPESAFE_BY_RCU, and the slab object
is not reused during iteration in the grace period. In fact, the
lockless readers do not check the nulls marker with get_nulls_value().
So, SOCK_RAW should use hlist instead of hlist_nulls.
Instead of adding an unnecessary barrier by sk_nulls_for_each_rcu(),
let's convert hlist_nulls to hlist and use sk_for_each_rcu() for
fast paths and sk_for_each() and spinlock for /proc/net/raw.
[0]:
general protection fault, probably for non-canonical address 0xdffffc0000000005: 0000 [#1] PREEMPT SMP KASAN
KASAN: null-ptr-deref in range [0x0000000000000028-0x000000000000002f]
CPU: 2 PID: 20952 Comm: syz-executor.0 Not tainted 6.2.0-g048ec869bafd-dirty #7
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.14.0-0-g155821a1990b-prebuilt.qemu.org 04/01/2014
RIP: 0010:read_pnet include/net/net_namespace.h:383 [inline]
RIP: 0010:sock_net include/net/sock.h:649 [inline]
RIP: 0010:raw_get_next net/ipv4/raw.c:974 [inline]
RIP: 0010:raw_get_idx net/ipv4/raw.c:986 [inline]
RIP: 0010:raw_seq_start+0x431/0x800 net/ipv4/raw.c:995
Code: ef e8 33 3d 94 f7 49 8b 6d 00 4c 89 ef e8 b7 65 5f f7 49 89 ed 49 83 c5 98 0f 84 9a 00 00 00 48 83 c5 c8 48 89 e8 48 c1 e8 03 <42> 80 3c 30 00 74 08 48 89 ef e8 00 3d 94 f7 4c 8b 7d 00 48 89 ef
RSP: 0018:ffffc9001154f9b0 EFLAGS: 00010206
RAX: 0000000000000005 RBX: 1ffff1100302c8fd RCX: 0000000000000000
RDX: 0000000000000028 RSI: ffffc9001154f988 RDI: ffffc9000f77a338
RBP: 0000000000000029 R08: ffffffff8a50ffb4 R09: fffffbfff24b6bd9
R10: fffffbfff24b6bd9 R11: 0000000000000000 R12: ffff88801db73b78
R13: fffffffffffffff9 R14: dffffc0000000000 R15: 0000000000000030
FS: 00007f843ae8e700(0000) GS:ffff888063700000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 000055bb9614b35f CR3: 000000003c672000 CR4: 00000000003506e0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
Call Trace:
<TASK>
seq_read_iter+0x4c6/0x10f0 fs/seq_file.c:225
seq_read+0x224/0x320 fs/seq_file.c:162
pde_read fs/proc/inode.c:316 [inline]
proc_reg_read+0x23f/0x330 fs/proc/inode.c:328
vfs_read+0x31e/0xd30 fs/read_write.c:468
ksys_pread64 fs/read_write.c:665 [inline]
__do_sys_pread64 fs/read_write.c:675 [inline]
__se_sys_pread64 fs/read_write.c:672 [inline]
__x64_sys_pread64+0x1e9/0x280 fs/read_write.c:672
do_syscall_x64 arch/x86/entry/common.c:51 [inline]
do_syscall_64+0x4e/0xa0 arch/x86/entry/common.c:82
entry_SYSCALL_64_after_hwframe+0x63/0xcd
RIP: 0033:0x478d29
Code: f7 d8 64 89 02 b8 ff ff ff ff c3 66 0f 1f 44 00 00 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 bc ff ff ff f7 d8 64 89 01 48
RSP: 002b:00007f843ae8dbe8 EFLAGS: 00000246 ORIG_RAX: 0000000000000011
RAX: ffffffffffffffda RBX: 0000000000791408 RCX: 0000000000478d29
RDX: 000000000000000a RSI: 0000000020000000 RDI: 0000000000000003
RBP: 00000000f477909a R08: 0000000000000000 R09: 0000000000000000
R10: 000010000000007f R11: 0000000000000246 R12: 0000000000791740
R13: 0000000000791414 R14: 0000000000791408 R15: 00007ffc2eb48a50
</TASK>
Modules linked in:
---[ end trace 0000000000000000 ]---
RIP: 0010:read_pnet include/net/net_namespace.h:383 [inline]
RIP: 0010:sock_net include/net/sock.h:649 [inline]
RIP: 0010:raw_get_next net/ipv4/raw.c:974 [inline]
RIP: 0010:raw_get_idx net/ipv4/raw.c:986 [inline]
RIP: 0010:raw_seq_start+0x431/0x800 net/ipv4/raw.c:995
Code: ef e8 33 3d 94 f7 49 8b 6d 00 4c 89 ef e8 b7 65 5f f7 49 89 ed 49 83 c5 98 0f 84 9a 00 00 00 48 83 c5 c8 48 89 e8 48 c1 e8 03 <42> 80 3c 30 00 74 08 48 89 ef e8 00 3d 94 f7 4c 8b 7d 00 48 89 ef
RSP: 0018:ffffc9001154f9b0 EFLAGS: 00010206
RAX: 0000000000000005 RBX: 1ffff1100302c8fd RCX: 0000000000000000
RDX: 0000000000000028 RSI: ffffc9001154f988 RDI: ffffc9000f77a338
RBP: 0000000000000029 R08: ffffffff8a50ffb4 R09: fffffbfff24b6bd9
R10: fffffbfff24b6bd9 R11: 0000000000000000 R12: ffff88801db73b78
R13: fffffffffffffff9 R14: dffffc0000000000 R15: 0000000000000030
FS: 00007f843ae8e700(0000) GS:ffff888063700000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007f92ff166000 CR3: 000000003c672000 CR4: 00000000003506e0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
Fixes: 0daf07e52709 ("raw: convert raw sockets to RCU")
Reported-by: syzbot <syzkaller@googlegroups.com>
Reported-by: Dae R. Jeong <threeearcat@gmail.com>
Link: https://lore.kernel.org/netdev/ZCA2mGV_cmq7lIfV@dragonet/
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2023-04-03 22:49:58 +03:00
sk_for_each_rcu ( sk , hlist ) {
2022-06-18 06:47:04 +03:00
if ( raw_lookup ( net , sk , r ) ) {
2016-10-21 13:03:44 +03:00
/*
* Grab it and keep until we fill
2022-06-18 06:47:05 +03:00
* diag message to be reported , so
2016-10-21 13:03:44 +03:00
* caller should call sock_put then .
*/
2022-06-18 06:47:05 +03:00
if ( refcount_inc_not_zero ( & sk - > sk_refcnt ) )
goto out_unlock ;
2016-10-21 13:03:44 +03:00
}
}
}
2022-06-18 06:47:04 +03:00
sk = ERR_PTR ( - ENOENT ) ;
2016-11-02 15:36:32 +03:00
out_unlock :
2022-06-18 06:47:05 +03:00
rcu_read_unlock ( ) ;
2016-10-21 13:03:44 +03:00
2022-06-18 06:47:04 +03:00
return sk ;
2016-10-21 13:03:44 +03:00
}
2020-02-26 02:04:09 +03:00
static int raw_diag_dump_one ( struct netlink_callback * cb ,
2016-10-21 13:03:44 +03:00
const struct inet_diag_req_v2 * r )
{
2020-02-26 02:04:09 +03:00
struct sk_buff * in_skb = cb - > skb ;
2016-10-21 13:03:44 +03:00
struct sk_buff * rep ;
struct sock * sk ;
2020-02-26 02:04:09 +03:00
struct net * net ;
2016-10-21 13:03:44 +03:00
int err ;
2020-02-26 02:04:09 +03:00
net = sock_net ( in_skb - > sk ) ;
2016-10-21 13:03:44 +03:00
sk = raw_sock_get ( net , r ) ;
if ( IS_ERR ( sk ) )
return PTR_ERR ( sk ) ;
2020-03-05 15:33:12 +03:00
rep = nlmsg_new ( nla_total_size ( sizeof ( struct inet_diag_msg ) ) +
inet_diag_msg_attrs_size ( ) +
nla_total_size ( sizeof ( struct inet_diag_meminfo ) ) + 64 ,
2016-10-21 13:03:44 +03:00
GFP_KERNEL ) ;
if ( ! rep ) {
sock_put ( sk ) ;
return - ENOMEM ;
}
2020-02-26 02:04:09 +03:00
err = inet_sk_diag_fill ( sk , NULL , rep , cb , r , 0 ,
2016-10-21 13:03:44 +03:00
netlink_net_capable ( in_skb , CAP_NET_ADMIN ) ) ;
sock_put ( sk ) ;
if ( err < 0 ) {
kfree_skb ( rep ) ;
return err ;
}
2021-07-13 05:48:24 +03:00
err = nlmsg_unicast ( net - > diag_nlsk , rep , NETLINK_CB ( in_skb ) . portid ) ;
2016-10-21 13:03:44 +03:00
return err ;
}
static int sk_diag_dump ( struct sock * sk , struct sk_buff * skb ,
struct netlink_callback * cb ,
const struct inet_diag_req_v2 * r ,
struct nlattr * bc , bool net_admin )
{
if ( ! inet_diag_bc_sk ( bc , sk ) )
return 0 ;
2020-02-26 02:04:09 +03:00
return inet_sk_diag_fill ( sk , NULL , skb , cb , r , NLM_F_MULTI , net_admin ) ;
2016-10-21 13:03:44 +03:00
}
static void raw_diag_dump ( struct sk_buff * skb , struct netlink_callback * cb ,
2020-02-26 02:04:15 +03:00
const struct inet_diag_req_v2 * r )
2016-10-21 13:03:44 +03:00
{
bool net_admin = netlink_net_capable ( cb - > skb , CAP_NET_ADMIN ) ;
struct raw_hashinfo * hashinfo = raw_get_hashinfo ( r ) ;
struct net * net = sock_net ( skb - > sk ) ;
2020-02-26 02:04:15 +03:00
struct inet_diag_dump_data * cb_data ;
2016-10-21 13:03:44 +03:00
int num , s_num , slot , s_slot ;
raw: Fix NULL deref in raw_get_next().
Dae R. Jeong reported a NULL deref in raw_get_next() [0].
It seems that the repro was running these sequences in parallel so
that one thread was iterating on a socket that was being freed in
another netns.
unshare(0x40060200)
r0 = syz_open_procfs(0x0, &(0x7f0000002080)='net/raw\x00')
socket$inet_icmp_raw(0x2, 0x3, 0x1)
pread64(r0, &(0x7f0000000000)=""/10, 0xa, 0x10000000007f)
After commit 0daf07e52709 ("raw: convert raw sockets to RCU"), we
use RCU and hlist_nulls_for_each_entry() to iterate over SOCK_RAW
sockets. However, we should use spinlock for slow paths to avoid
the NULL deref.
Also, SOCK_RAW does not use SLAB_TYPESAFE_BY_RCU, and the slab object
is not reused during iteration in the grace period. In fact, the
lockless readers do not check the nulls marker with get_nulls_value().
So, SOCK_RAW should use hlist instead of hlist_nulls.
Instead of adding an unnecessary barrier by sk_nulls_for_each_rcu(),
let's convert hlist_nulls to hlist and use sk_for_each_rcu() for
fast paths and sk_for_each() and spinlock for /proc/net/raw.
[0]:
general protection fault, probably for non-canonical address 0xdffffc0000000005: 0000 [#1] PREEMPT SMP KASAN
KASAN: null-ptr-deref in range [0x0000000000000028-0x000000000000002f]
CPU: 2 PID: 20952 Comm: syz-executor.0 Not tainted 6.2.0-g048ec869bafd-dirty #7
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.14.0-0-g155821a1990b-prebuilt.qemu.org 04/01/2014
RIP: 0010:read_pnet include/net/net_namespace.h:383 [inline]
RIP: 0010:sock_net include/net/sock.h:649 [inline]
RIP: 0010:raw_get_next net/ipv4/raw.c:974 [inline]
RIP: 0010:raw_get_idx net/ipv4/raw.c:986 [inline]
RIP: 0010:raw_seq_start+0x431/0x800 net/ipv4/raw.c:995
Code: ef e8 33 3d 94 f7 49 8b 6d 00 4c 89 ef e8 b7 65 5f f7 49 89 ed 49 83 c5 98 0f 84 9a 00 00 00 48 83 c5 c8 48 89 e8 48 c1 e8 03 <42> 80 3c 30 00 74 08 48 89 ef e8 00 3d 94 f7 4c 8b 7d 00 48 89 ef
RSP: 0018:ffffc9001154f9b0 EFLAGS: 00010206
RAX: 0000000000000005 RBX: 1ffff1100302c8fd RCX: 0000000000000000
RDX: 0000000000000028 RSI: ffffc9001154f988 RDI: ffffc9000f77a338
RBP: 0000000000000029 R08: ffffffff8a50ffb4 R09: fffffbfff24b6bd9
R10: fffffbfff24b6bd9 R11: 0000000000000000 R12: ffff88801db73b78
R13: fffffffffffffff9 R14: dffffc0000000000 R15: 0000000000000030
FS: 00007f843ae8e700(0000) GS:ffff888063700000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 000055bb9614b35f CR3: 000000003c672000 CR4: 00000000003506e0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
Call Trace:
<TASK>
seq_read_iter+0x4c6/0x10f0 fs/seq_file.c:225
seq_read+0x224/0x320 fs/seq_file.c:162
pde_read fs/proc/inode.c:316 [inline]
proc_reg_read+0x23f/0x330 fs/proc/inode.c:328
vfs_read+0x31e/0xd30 fs/read_write.c:468
ksys_pread64 fs/read_write.c:665 [inline]
__do_sys_pread64 fs/read_write.c:675 [inline]
__se_sys_pread64 fs/read_write.c:672 [inline]
__x64_sys_pread64+0x1e9/0x280 fs/read_write.c:672
do_syscall_x64 arch/x86/entry/common.c:51 [inline]
do_syscall_64+0x4e/0xa0 arch/x86/entry/common.c:82
entry_SYSCALL_64_after_hwframe+0x63/0xcd
RIP: 0033:0x478d29
Code: f7 d8 64 89 02 b8 ff ff ff ff c3 66 0f 1f 44 00 00 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 bc ff ff ff f7 d8 64 89 01 48
RSP: 002b:00007f843ae8dbe8 EFLAGS: 00000246 ORIG_RAX: 0000000000000011
RAX: ffffffffffffffda RBX: 0000000000791408 RCX: 0000000000478d29
RDX: 000000000000000a RSI: 0000000020000000 RDI: 0000000000000003
RBP: 00000000f477909a R08: 0000000000000000 R09: 0000000000000000
R10: 000010000000007f R11: 0000000000000246 R12: 0000000000791740
R13: 0000000000791414 R14: 0000000000791408 R15: 00007ffc2eb48a50
</TASK>
Modules linked in:
---[ end trace 0000000000000000 ]---
RIP: 0010:read_pnet include/net/net_namespace.h:383 [inline]
RIP: 0010:sock_net include/net/sock.h:649 [inline]
RIP: 0010:raw_get_next net/ipv4/raw.c:974 [inline]
RIP: 0010:raw_get_idx net/ipv4/raw.c:986 [inline]
RIP: 0010:raw_seq_start+0x431/0x800 net/ipv4/raw.c:995
Code: ef e8 33 3d 94 f7 49 8b 6d 00 4c 89 ef e8 b7 65 5f f7 49 89 ed 49 83 c5 98 0f 84 9a 00 00 00 48 83 c5 c8 48 89 e8 48 c1 e8 03 <42> 80 3c 30 00 74 08 48 89 ef e8 00 3d 94 f7 4c 8b 7d 00 48 89 ef
RSP: 0018:ffffc9001154f9b0 EFLAGS: 00010206
RAX: 0000000000000005 RBX: 1ffff1100302c8fd RCX: 0000000000000000
RDX: 0000000000000028 RSI: ffffc9001154f988 RDI: ffffc9000f77a338
RBP: 0000000000000029 R08: ffffffff8a50ffb4 R09: fffffbfff24b6bd9
R10: fffffbfff24b6bd9 R11: 0000000000000000 R12: ffff88801db73b78
R13: fffffffffffffff9 R14: dffffc0000000000 R15: 0000000000000030
FS: 00007f843ae8e700(0000) GS:ffff888063700000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007f92ff166000 CR3: 000000003c672000 CR4: 00000000003506e0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
Fixes: 0daf07e52709 ("raw: convert raw sockets to RCU")
Reported-by: syzbot <syzkaller@googlegroups.com>
Reported-by: Dae R. Jeong <threeearcat@gmail.com>
Link: https://lore.kernel.org/netdev/ZCA2mGV_cmq7lIfV@dragonet/
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2023-04-03 22:49:58 +03:00
struct hlist_head * hlist ;
2016-10-21 13:03:44 +03:00
struct sock * sk = NULL ;
2020-02-26 02:04:15 +03:00
struct nlattr * bc ;
2016-10-21 13:03:44 +03:00
if ( IS_ERR ( hashinfo ) )
return ;
2020-02-26 02:04:15 +03:00
cb_data = cb - > data ;
bc = cb_data - > inet_diag_nla_bc ;
2016-10-21 13:03:44 +03:00
s_slot = cb - > args [ 0 ] ;
num = s_num = cb - > args [ 1 ] ;
2022-06-20 13:05:09 +03:00
rcu_read_lock ( ) ;
2016-10-21 13:03:44 +03:00
for ( slot = s_slot ; slot < RAW_HTABLE_SIZE ; s_num = 0 , slot + + ) {
num = 0 ;
2022-06-18 06:47:05 +03:00
hlist = & hashinfo - > ht [ slot ] ;
raw: Fix NULL deref in raw_get_next().
Dae R. Jeong reported a NULL deref in raw_get_next() [0].
It seems that the repro was running these sequences in parallel so
that one thread was iterating on a socket that was being freed in
another netns.
unshare(0x40060200)
r0 = syz_open_procfs(0x0, &(0x7f0000002080)='net/raw\x00')
socket$inet_icmp_raw(0x2, 0x3, 0x1)
pread64(r0, &(0x7f0000000000)=""/10, 0xa, 0x10000000007f)
After commit 0daf07e52709 ("raw: convert raw sockets to RCU"), we
use RCU and hlist_nulls_for_each_entry() to iterate over SOCK_RAW
sockets. However, we should use spinlock for slow paths to avoid
the NULL deref.
Also, SOCK_RAW does not use SLAB_TYPESAFE_BY_RCU, and the slab object
is not reused during iteration in the grace period. In fact, the
lockless readers do not check the nulls marker with get_nulls_value().
So, SOCK_RAW should use hlist instead of hlist_nulls.
Instead of adding an unnecessary barrier by sk_nulls_for_each_rcu(),
let's convert hlist_nulls to hlist and use sk_for_each_rcu() for
fast paths and sk_for_each() and spinlock for /proc/net/raw.
[0]:
general protection fault, probably for non-canonical address 0xdffffc0000000005: 0000 [#1] PREEMPT SMP KASAN
KASAN: null-ptr-deref in range [0x0000000000000028-0x000000000000002f]
CPU: 2 PID: 20952 Comm: syz-executor.0 Not tainted 6.2.0-g048ec869bafd-dirty #7
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.14.0-0-g155821a1990b-prebuilt.qemu.org 04/01/2014
RIP: 0010:read_pnet include/net/net_namespace.h:383 [inline]
RIP: 0010:sock_net include/net/sock.h:649 [inline]
RIP: 0010:raw_get_next net/ipv4/raw.c:974 [inline]
RIP: 0010:raw_get_idx net/ipv4/raw.c:986 [inline]
RIP: 0010:raw_seq_start+0x431/0x800 net/ipv4/raw.c:995
Code: ef e8 33 3d 94 f7 49 8b 6d 00 4c 89 ef e8 b7 65 5f f7 49 89 ed 49 83 c5 98 0f 84 9a 00 00 00 48 83 c5 c8 48 89 e8 48 c1 e8 03 <42> 80 3c 30 00 74 08 48 89 ef e8 00 3d 94 f7 4c 8b 7d 00 48 89 ef
RSP: 0018:ffffc9001154f9b0 EFLAGS: 00010206
RAX: 0000000000000005 RBX: 1ffff1100302c8fd RCX: 0000000000000000
RDX: 0000000000000028 RSI: ffffc9001154f988 RDI: ffffc9000f77a338
RBP: 0000000000000029 R08: ffffffff8a50ffb4 R09: fffffbfff24b6bd9
R10: fffffbfff24b6bd9 R11: 0000000000000000 R12: ffff88801db73b78
R13: fffffffffffffff9 R14: dffffc0000000000 R15: 0000000000000030
FS: 00007f843ae8e700(0000) GS:ffff888063700000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 000055bb9614b35f CR3: 000000003c672000 CR4: 00000000003506e0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
Call Trace:
<TASK>
seq_read_iter+0x4c6/0x10f0 fs/seq_file.c:225
seq_read+0x224/0x320 fs/seq_file.c:162
pde_read fs/proc/inode.c:316 [inline]
proc_reg_read+0x23f/0x330 fs/proc/inode.c:328
vfs_read+0x31e/0xd30 fs/read_write.c:468
ksys_pread64 fs/read_write.c:665 [inline]
__do_sys_pread64 fs/read_write.c:675 [inline]
__se_sys_pread64 fs/read_write.c:672 [inline]
__x64_sys_pread64+0x1e9/0x280 fs/read_write.c:672
do_syscall_x64 arch/x86/entry/common.c:51 [inline]
do_syscall_64+0x4e/0xa0 arch/x86/entry/common.c:82
entry_SYSCALL_64_after_hwframe+0x63/0xcd
RIP: 0033:0x478d29
Code: f7 d8 64 89 02 b8 ff ff ff ff c3 66 0f 1f 44 00 00 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 bc ff ff ff f7 d8 64 89 01 48
RSP: 002b:00007f843ae8dbe8 EFLAGS: 00000246 ORIG_RAX: 0000000000000011
RAX: ffffffffffffffda RBX: 0000000000791408 RCX: 0000000000478d29
RDX: 000000000000000a RSI: 0000000020000000 RDI: 0000000000000003
RBP: 00000000f477909a R08: 0000000000000000 R09: 0000000000000000
R10: 000010000000007f R11: 0000000000000246 R12: 0000000000791740
R13: 0000000000791414 R14: 0000000000791408 R15: 00007ffc2eb48a50
</TASK>
Modules linked in:
---[ end trace 0000000000000000 ]---
RIP: 0010:read_pnet include/net/net_namespace.h:383 [inline]
RIP: 0010:sock_net include/net/sock.h:649 [inline]
RIP: 0010:raw_get_next net/ipv4/raw.c:974 [inline]
RIP: 0010:raw_get_idx net/ipv4/raw.c:986 [inline]
RIP: 0010:raw_seq_start+0x431/0x800 net/ipv4/raw.c:995
Code: ef e8 33 3d 94 f7 49 8b 6d 00 4c 89 ef e8 b7 65 5f f7 49 89 ed 49 83 c5 98 0f 84 9a 00 00 00 48 83 c5 c8 48 89 e8 48 c1 e8 03 <42> 80 3c 30 00 74 08 48 89 ef e8 00 3d 94 f7 4c 8b 7d 00 48 89 ef
RSP: 0018:ffffc9001154f9b0 EFLAGS: 00010206
RAX: 0000000000000005 RBX: 1ffff1100302c8fd RCX: 0000000000000000
RDX: 0000000000000028 RSI: ffffc9001154f988 RDI: ffffc9000f77a338
RBP: 0000000000000029 R08: ffffffff8a50ffb4 R09: fffffbfff24b6bd9
R10: fffffbfff24b6bd9 R11: 0000000000000000 R12: ffff88801db73b78
R13: fffffffffffffff9 R14: dffffc0000000000 R15: 0000000000000030
FS: 00007f843ae8e700(0000) GS:ffff888063700000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007f92ff166000 CR3: 000000003c672000 CR4: 00000000003506e0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
Fixes: 0daf07e52709 ("raw: convert raw sockets to RCU")
Reported-by: syzbot <syzkaller@googlegroups.com>
Reported-by: Dae R. Jeong <threeearcat@gmail.com>
Link: https://lore.kernel.org/netdev/ZCA2mGV_cmq7lIfV@dragonet/
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2023-04-03 22:49:58 +03:00
sk_for_each_rcu ( sk , hlist ) {
2016-10-21 13:03:44 +03:00
struct inet_sock * inet = inet_sk ( sk ) ;
if ( ! net_eq ( sock_net ( sk ) , net ) )
continue ;
if ( num < s_num )
goto next ;
if ( sk - > sk_family ! = r - > sdiag_family )
goto next ;
if ( r - > id . idiag_sport ! = inet - > inet_sport & &
r - > id . idiag_sport )
goto next ;
if ( r - > id . idiag_dport ! = inet - > inet_dport & &
r - > id . idiag_dport )
goto next ;
if ( sk_diag_dump ( sk , skb , cb , r , bc , net_admin ) < 0 )
goto out_unlock ;
next :
num + + ;
}
}
out_unlock :
2022-06-20 13:05:09 +03:00
rcu_read_unlock ( ) ;
2016-10-21 13:03:44 +03:00
cb - > args [ 0 ] = slot ;
cb - > args [ 1 ] = num ;
}
static void raw_diag_get_info ( struct sock * sk , struct inet_diag_msg * r ,
void * info )
{
r - > idiag_rqueue = sk_rmem_alloc_get ( sk ) ;
r - > idiag_wqueue = sk_wmem_alloc_get ( sk ) ;
}
# ifdef CONFIG_INET_DIAG_DESTROY
static int raw_diag_destroy ( struct sk_buff * in_skb ,
const struct inet_diag_req_v2 * r )
{
struct net * net = sock_net ( in_skb - > sk ) ;
struct sock * sk ;
2016-11-02 15:36:31 +03:00
int err ;
2016-10-21 13:03:44 +03:00
sk = raw_sock_get ( net , r ) ;
if ( IS_ERR ( sk ) )
return PTR_ERR ( sk ) ;
2016-11-02 15:36:31 +03:00
err = sock_diag_destroy ( sk , ECONNABORTED ) ;
sock_put ( sk ) ;
return err ;
2016-10-21 13:03:44 +03:00
}
# endif
static const struct inet_diag_handler raw_diag_handler = {
2024-01-22 14:25:57 +03:00
. owner = THIS_MODULE ,
2016-10-21 13:03:44 +03:00
. dump = raw_diag_dump ,
. dump_one = raw_diag_dump_one ,
. idiag_get_info = raw_diag_get_info ,
. idiag_type = IPPROTO_RAW ,
. idiag_info_size = 0 ,
# ifdef CONFIG_INET_DIAG_DESTROY
. destroy = raw_diag_destroy ,
# endif
} ;
static void __always_unused __check_inet_diag_req_raw ( void )
{
/*
* Make sure the two structures are identical ,
* except the @ pad field .
*/
# define __offset_mismatch(m1, m2) \
( offsetof ( struct inet_diag_req_v2 , m1 ) ! = \
offsetof ( struct inet_diag_req_raw , m2 ) )
BUILD_BUG_ON ( sizeof ( struct inet_diag_req_v2 ) ! =
sizeof ( struct inet_diag_req_raw ) ) ;
BUILD_BUG_ON ( __offset_mismatch ( sdiag_family , sdiag_family ) ) ;
BUILD_BUG_ON ( __offset_mismatch ( sdiag_protocol , sdiag_protocol ) ) ;
BUILD_BUG_ON ( __offset_mismatch ( idiag_ext , idiag_ext ) ) ;
BUILD_BUG_ON ( __offset_mismatch ( pad , sdiag_raw_protocol ) ) ;
BUILD_BUG_ON ( __offset_mismatch ( idiag_states , idiag_states ) ) ;
BUILD_BUG_ON ( __offset_mismatch ( id , id ) ) ;
# undef __offset_mismatch
}
static int __init raw_diag_init ( void )
{
return inet_diag_register ( & raw_diag_handler ) ;
}
static void __exit raw_diag_exit ( void )
{
inet_diag_unregister ( & raw_diag_handler ) ;
}
module_init ( raw_diag_init ) ;
module_exit ( raw_diag_exit ) ;
MODULE_LICENSE ( " GPL " ) ;
2023-11-19 06:30:06 +03:00
MODULE_DESCRIPTION ( " RAW socket monitoring via SOCK_DIAG " ) ;
2016-10-21 13:03:44 +03:00
MODULE_ALIAS_NET_PF_PROTO_TYPE ( PF_NETLINK , NETLINK_SOCK_DIAG , 2 - 255 /* AF_INET - IPPROTO_RAW */ ) ;
MODULE_ALIAS_NET_PF_PROTO_TYPE ( PF_NETLINK , NETLINK_SOCK_DIAG , 10 - 255 /* AF_INET6 - IPPROTO_RAW */ ) ;