* Re: [PATCH net] net: xfrm: skip policies marked as dead while reinserting policies
2023-08-14 1:33 [PATCH net] net: xfrm: skip policies marked as dead while reinserting policies Dong Chenchen
@ 2023-08-14 7:03 ` Leon Romanovsky
2023-08-14 8:58 ` Dong Chenchen
2023-08-14 10:10 ` Leon Romanovsky
2023-08-14 10:10 ` Leon Romanovsky
2023-08-14 11:30 ` Florian Westphal
2 siblings, 2 replies; 7+ messages in thread
From: Leon Romanovsky @ 2023-08-14 7:03 UTC (permalink / raw)
To: Dong Chenchen
Cc: steffen.klassert, herbert, davem, edumazet, kuba, pabeni, fw,
timo.teras, yuehaibing, weiyongjun1, netdev, linux-kernel
On Mon, Aug 14, 2023 at 09:33:52AM +0800, Dong Chenchen wrote:
> BUG: KASAN: slab-use-after-free in xfrm_policy_inexact_list_reinsert+0xb6/0x430
> Read of size 1 at addr ffff8881051f3bf8 by task ip/668
>
> CPU: 2 PID: 668 Comm: ip Not tainted 6.5.0-rc5-00182-g25aa0bebba72-dirty #64
> Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.13 04/01/2014
> Call Trace:
> <TASK>
> dump_stack_lvl+0x72/0xa0
> print_report+0xd0/0x620
> kasan_report+0xb6/0xf0
> xfrm_policy_inexact_list_reinsert+0xb6/0x430
> xfrm_policy_inexact_insert_node.constprop.0+0x537/0x800
> xfrm_policy_inexact_alloc_chain+0x23f/0x320
> xfrm_policy_inexact_insert+0x6b/0x590
> xfrm_policy_insert+0x3b1/0x480
> xfrm_add_policy+0x23c/0x3c0
> xfrm_user_rcv_msg+0x2d0/0x510
> netlink_rcv_skb+0x10d/0x2d0
> xfrm_netlink_rcv+0x49/0x60
> netlink_unicast+0x3fe/0x540
> netlink_sendmsg+0x528/0x970
> sock_sendmsg+0x14a/0x160
> ____sys_sendmsg+0x4fc/0x580
> ___sys_sendmsg+0xef/0x160
> __sys_sendmsg+0xf7/0x1b0
> do_syscall_64+0x3f/0x90
> entry_SYSCALL_64_after_hwframe+0x73/0xdd
>
> The root cause is:
>
> cpu 0 cpu1
> xfrm_dump_policy
> xfrm_policy_walk
> list_move_tail
> xfrm_add_policy
> ... ...
> xfrm_policy_inexact_list_reinsert
> list_for_each_entry_reverse
> if (!policy->bydst_reinsert)
> //read non-existent policy
> xfrm_dump_policy_done
> xfrm_policy_walk_done
> list_del(&walk->walk.all);
>
> If dump_one_policy() returns err (triggered by netlink socket),
> xfrm_policy_walk() will move walk initialized by socket to list
> net->xfrm.policy_all. so this socket becomes visible in the global
> policy list. The head *walk can be traversed when users add policies
> with different prefixlen and trigger xfrm_policy node merge.
>
> It can be fixed by skip such "policies" with walk.dead set to 1.
But where in the xfrm_dump_policy() flow, these policies are becoming to
be walk.dead == 1?
Thanks
>
> Fixes: 9cf545ebd591 ("xfrm: policy: store inexact policies in a tree ordered by destination address")
> Fixes: 12a169e7d8f4 ("ipsec: Put dumpers on the dump list")
> Signed-off-by: Dong Chenchen <dongchenchen2@huawei.com>
> ---
> net/xfrm/xfrm_policy.c | 3 +++
> 1 file changed, 3 insertions(+)
>
> diff --git a/net/xfrm/xfrm_policy.c b/net/xfrm/xfrm_policy.c
> index d6b405782b63..5b56faad78e0 100644
> --- a/net/xfrm/xfrm_policy.c
> +++ b/net/xfrm/xfrm_policy.c
> @@ -848,6 +848,9 @@ static void xfrm_policy_inexact_list_reinsert(struct net *net,
> matched_d = 0;
>
> list_for_each_entry_reverse(policy, &net->xfrm.policy_all, walk.all) {
> + if (policy->walk.dead)
> + continue;
> +
> struct hlist_node *newpos = NULL;
> bool matches_s, matches_d;
>
> --
> 2.25.1
>
>
^ permalink raw reply [flat|nested] 7+ messages in thread* Re: [PATCH net] net: xfrm: skip policies marked as dead while reinserting policies
2023-08-14 7:03 ` Leon Romanovsky
@ 2023-08-14 8:58 ` Dong Chenchen
2023-08-14 10:10 ` Leon Romanovsky
1 sibling, 0 replies; 7+ messages in thread
From: Dong Chenchen @ 2023-08-14 8:58 UTC (permalink / raw)
To: leon
Cc: steffen.klassert, herbert, davem, edumazet, kuba, pabeni, fw,
timo.teras, yuehaibing, weiyongjun1, netdev, linux-kernel
On Mon, Aug 14, 2023 at 09:33:52AM +0800, Dong Chenchen wrote:
>> BUG: KASAN: slab-use-after-free in xfrm_policy_inexact_list_reinsert+0xb6/0x430
>> Read of size 1 at addr ffff8881051f3bf8 by task ip/668
>>
>> CPU: 2 PID: 668 Comm: ip Not tainted 6.5.0-rc5-00182-g25aa0bebba72-dirty #64
>> Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.13 04/01/2014
>> Call Trace:
>> <TASK>
>> dump_stack_lvl+0x72/0xa0
>> print_report+0xd0/0x620
>> kasan_report+0xb6/0xf0
>> xfrm_policy_inexact_list_reinsert+0xb6/0x430
>> xfrm_policy_inexact_insert_node.constprop.0+0x537/0x800
>> xfrm_policy_inexact_alloc_chain+0x23f/0x320
>> xfrm_policy_inexact_insert+0x6b/0x590
>> xfrm_policy_insert+0x3b1/0x480
>> xfrm_add_policy+0x23c/0x3c0
>> xfrm_user_rcv_msg+0x2d0/0x510
>> netlink_rcv_skb+0x10d/0x2d0
>> xfrm_netlink_rcv+0x49/0x60
>> netlink_unicast+0x3fe/0x540
>> netlink_sendmsg+0x528/0x970
>> sock_sendmsg+0x14a/0x160
>> ____sys_sendmsg+0x4fc/0x580
>> ___sys_sendmsg+0xef/0x160
>> __sys_sendmsg+0xf7/0x1b0
>> do_syscall_64+0x3f/0x90
>> entry_SYSCALL_64_after_hwframe+0x73/0xdd
>>
>> The root cause is:
>>
>> cpu 0 cpu1
>> xfrm_dump_policy
>> xfrm_policy_walk
>> list_move_tail
>> xfrm_add_policy
>> ... ...
>> xfrm_policy_inexact_list_reinsert
>> list_for_each_entry_reverse
>> if (!policy->bydst_reinsert)
>> //read non-existent policy
>> xfrm_dump_policy_done
>> xfrm_policy_walk_done
>> list_del(&walk->walk.all);
>>
>> If dump_one_policy() returns err (triggered by netlink socket),
>> xfrm_policy_walk() will move walk initialized by socket to list
>> net->xfrm.policy_all. so this socket becomes visible in the global
>> policy list. The head *walk can be traversed when users add policies
>> with different prefixlen and trigger xfrm_policy node merge.
>>
>> It can be fixed by skip such "policies" with walk.dead set to 1.
>
>But where in the xfrm_dump_policy() flow, these policies are becoming to
>be walk.dead == 1?
>
>Thanks
>
user will use xfrm_dispatch[XFRM_MSG_GETPOLICY] ops to get xfrm policy.
.start = xfrm_dump_policy_start
xfrm_dump_policy_start() will set walk.dead to 1 by call
xfrm_policy_walk_init().
Thanks
>>
>> Fixes: 9cf545ebd591 ("xfrm: policy: store inexact policies in a tree ordered by destination address")
>> Fixes: 12a169e7d8f4 ("ipsec: Put dumpers on the dump list")
>> Signed-off-by: Dong Chenchen <dongchenchen2@huawei.com>
>> ---
>> net/xfrm/xfrm_policy.c | 3 +++
>> 1 file changed, 3 insertions(+)
>>
>> diff --git a/net/xfrm/xfrm_policy.c b/net/xfrm/xfrm_policy.c
>> index d6b405782b63..5b56faad78e0 100644
>> --- a/net/xfrm/xfrm_policy.c
>> +++ b/net/xfrm/xfrm_policy.c
>> @@ -848,6 +848,9 @@ static void xfrm_policy_inexact_list_reinsert(struct net *net,
>> matched_d = 0;
>>
>> list_for_each_entry_reverse(policy, &net->xfrm.policy_all, walk.all) {
>> + if (policy->walk.dead)
>> + continue;
>> +
>> struct hlist_node *newpos = NULL;
>> bool matches_s, matches_d;
>>
>> --
>> 2.25.1
>>
>>
^ permalink raw reply [flat|nested] 7+ messages in thread* Re: [PATCH net] net: xfrm: skip policies marked as dead while reinserting policies
2023-08-14 7:03 ` Leon Romanovsky
2023-08-14 8:58 ` Dong Chenchen
@ 2023-08-14 10:10 ` Leon Romanovsky
1 sibling, 0 replies; 7+ messages in thread
From: Leon Romanovsky @ 2023-08-14 10:10 UTC (permalink / raw)
To: Dong Chenchen
Cc: steffen.klassert, herbert, davem, edumazet, kuba, pabeni, fw,
timo.teras, yuehaibing, weiyongjun1, netdev, linux-kernel
On Mon, Aug 14, 2023 at 04:58:46PM +0800, Dong Chenchen wrote:
> On Mon, Aug 14, 2023 at 09:33:52AM +0800, Dong Chenchen wrote:
> >> BUG: KASAN: slab-use-after-free in xfrm_policy_inexact_list_reinsert+0xb6/0x430
> >> Read of size 1 at addr ffff8881051f3bf8 by task ip/668
> >>
> >> CPU: 2 PID: 668 Comm: ip Not tainted 6.5.0-rc5-00182-g25aa0bebba72-dirty #64
> >> Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.13 04/01/2014
> >> Call Trace:
> >> <TASK>
> >> dump_stack_lvl+0x72/0xa0
> >> print_report+0xd0/0x620
> >> kasan_report+0xb6/0xf0
> >> xfrm_policy_inexact_list_reinsert+0xb6/0x430
> >> xfrm_policy_inexact_insert_node.constprop.0+0x537/0x800
> >> xfrm_policy_inexact_alloc_chain+0x23f/0x320
> >> xfrm_policy_inexact_insert+0x6b/0x590
> >> xfrm_policy_insert+0x3b1/0x480
> >> xfrm_add_policy+0x23c/0x3c0
> >> xfrm_user_rcv_msg+0x2d0/0x510
> >> netlink_rcv_skb+0x10d/0x2d0
> >> xfrm_netlink_rcv+0x49/0x60
> >> netlink_unicast+0x3fe/0x540
> >> netlink_sendmsg+0x528/0x970
> >> sock_sendmsg+0x14a/0x160
> >> ____sys_sendmsg+0x4fc/0x580
> >> ___sys_sendmsg+0xef/0x160
> >> __sys_sendmsg+0xf7/0x1b0
> >> do_syscall_64+0x3f/0x90
> >> entry_SYSCALL_64_after_hwframe+0x73/0xdd
> >>
> >> The root cause is:
> >>
> >> cpu 0 cpu1
> >> xfrm_dump_policy
> >> xfrm_policy_walk
> >> list_move_tail
> >> xfrm_add_policy
> >> ... ...
> >> xfrm_policy_inexact_list_reinsert
> >> list_for_each_entry_reverse
> >> if (!policy->bydst_reinsert)
> >> //read non-existent policy
> >> xfrm_dump_policy_done
> >> xfrm_policy_walk_done
> >> list_del(&walk->walk.all);
> >>
> >> If dump_one_policy() returns err (triggered by netlink socket),
> >> xfrm_policy_walk() will move walk initialized by socket to list
> >> net->xfrm.policy_all. so this socket becomes visible in the global
> >> policy list. The head *walk can be traversed when users add policies
> >> with different prefixlen and trigger xfrm_policy node merge.
> >>
> >> It can be fixed by skip such "policies" with walk.dead set to 1.
> >
> >But where in the xfrm_dump_policy() flow, these policies are becoming to
> >be walk.dead == 1?
> >
> >Thanks
> >
> user will use xfrm_dispatch[XFRM_MSG_GETPOLICY] ops to get xfrm policy.
>
> .start = xfrm_dump_policy_start
>
> xfrm_dump_policy_start() will set walk.dead to 1 by call
> xfrm_policy_walk_init().
Thanks
>
> Thanks
> >>
> >> Fixes: 9cf545ebd591 ("xfrm: policy: store inexact policies in a tree ordered by destination address")
> >> Fixes: 12a169e7d8f4 ("ipsec: Put dumpers on the dump list")
> >> Signed-off-by: Dong Chenchen <dongchenchen2@huawei.com>
> >> ---
> >> net/xfrm/xfrm_policy.c | 3 +++
> >> 1 file changed, 3 insertions(+)
> >>
> >> diff --git a/net/xfrm/xfrm_policy.c b/net/xfrm/xfrm_policy.c
> >> index d6b405782b63..5b56faad78e0 100644
> >> --- a/net/xfrm/xfrm_policy.c
> >> +++ b/net/xfrm/xfrm_policy.c
> >> @@ -848,6 +848,9 @@ static void xfrm_policy_inexact_list_reinsert(struct net *net,
> >> matched_d = 0;
> >>
> >> list_for_each_entry_reverse(policy, &net->xfrm.policy_all, walk.all) {
> >> + if (policy->walk.dead)
> >> + continue;
> >> +
> >> struct hlist_node *newpos = NULL;
> >> bool matches_s, matches_d;
> >>
> >> --
> >> 2.25.1
> >>
> >>
>
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH net] net: xfrm: skip policies marked as dead while reinserting policies
2023-08-14 1:33 [PATCH net] net: xfrm: skip policies marked as dead while reinserting policies Dong Chenchen
2023-08-14 7:03 ` Leon Romanovsky
@ 2023-08-14 10:10 ` Leon Romanovsky
2023-08-14 11:30 ` Florian Westphal
2 siblings, 0 replies; 7+ messages in thread
From: Leon Romanovsky @ 2023-08-14 10:10 UTC (permalink / raw)
To: Dong Chenchen
Cc: steffen.klassert, herbert, davem, edumazet, kuba, pabeni, fw,
timo.teras, yuehaibing, weiyongjun1, netdev, linux-kernel
On Mon, Aug 14, 2023 at 09:33:52AM +0800, Dong Chenchen wrote:
> BUG: KASAN: slab-use-after-free in xfrm_policy_inexact_list_reinsert+0xb6/0x430
> Read of size 1 at addr ffff8881051f3bf8 by task ip/668
>
> CPU: 2 PID: 668 Comm: ip Not tainted 6.5.0-rc5-00182-g25aa0bebba72-dirty #64
> Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.13 04/01/2014
> Call Trace:
> <TASK>
> dump_stack_lvl+0x72/0xa0
> print_report+0xd0/0x620
> kasan_report+0xb6/0xf0
> xfrm_policy_inexact_list_reinsert+0xb6/0x430
> xfrm_policy_inexact_insert_node.constprop.0+0x537/0x800
> xfrm_policy_inexact_alloc_chain+0x23f/0x320
> xfrm_policy_inexact_insert+0x6b/0x590
> xfrm_policy_insert+0x3b1/0x480
> xfrm_add_policy+0x23c/0x3c0
> xfrm_user_rcv_msg+0x2d0/0x510
> netlink_rcv_skb+0x10d/0x2d0
> xfrm_netlink_rcv+0x49/0x60
> netlink_unicast+0x3fe/0x540
> netlink_sendmsg+0x528/0x970
> sock_sendmsg+0x14a/0x160
> ____sys_sendmsg+0x4fc/0x580
> ___sys_sendmsg+0xef/0x160
> __sys_sendmsg+0xf7/0x1b0
> do_syscall_64+0x3f/0x90
> entry_SYSCALL_64_after_hwframe+0x73/0xdd
>
> The root cause is:
>
> cpu 0 cpu1
> xfrm_dump_policy
> xfrm_policy_walk
> list_move_tail
> xfrm_add_policy
> ... ...
> xfrm_policy_inexact_list_reinsert
> list_for_each_entry_reverse
> if (!policy->bydst_reinsert)
> //read non-existent policy
> xfrm_dump_policy_done
> xfrm_policy_walk_done
> list_del(&walk->walk.all);
>
> If dump_one_policy() returns err (triggered by netlink socket),
> xfrm_policy_walk() will move walk initialized by socket to list
> net->xfrm.policy_all. so this socket becomes visible in the global
> policy list. The head *walk can be traversed when users add policies
> with different prefixlen and trigger xfrm_policy node merge.
>
> It can be fixed by skip such "policies" with walk.dead set to 1.
>
> Fixes: 9cf545ebd591 ("xfrm: policy: store inexact policies in a tree ordered by destination address")
> Fixes: 12a169e7d8f4 ("ipsec: Put dumpers on the dump list")
> Signed-off-by: Dong Chenchen <dongchenchen2@huawei.com>
> ---
> net/xfrm/xfrm_policy.c | 3 +++
> 1 file changed, 3 insertions(+)
>
Thanks,
Reviewed-by: Leon Romanovsky <leonro@nvidia.com>
^ permalink raw reply [flat|nested] 7+ messages in thread* Re: [PATCH net] net: xfrm: skip policies marked as dead while reinserting policies
2023-08-14 1:33 [PATCH net] net: xfrm: skip policies marked as dead while reinserting policies Dong Chenchen
2023-08-14 7:03 ` Leon Romanovsky
2023-08-14 10:10 ` Leon Romanovsky
@ 2023-08-14 11:30 ` Florian Westphal
2023-08-14 12:58 ` Dong Chenchen
2 siblings, 1 reply; 7+ messages in thread
From: Florian Westphal @ 2023-08-14 11:30 UTC (permalink / raw)
To: Dong Chenchen
Cc: steffen.klassert, herbert, davem, edumazet, kuba, pabeni, fw,
timo.teras, yuehaibing, weiyongjun1, netdev, linux-kernel
Dong Chenchen <dongchenchen2@huawei.com> wrote:
> BUG: KASAN: slab-use-after-free in xfrm_policy_inexact_list_reinsert+0xb6/0x430
> Read of size 1 at addr ffff8881051f3bf8 by task ip/668
>
> CPU: 2 PID: 668 Comm: ip Not tainted 6.5.0-rc5-00182-g25aa0bebba72-dirty #64
> Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.13 04/01/2014
> Call Trace:
> <TASK>
> dump_stack_lvl+0x72/0xa0
> print_report+0xd0/0x620
> kasan_report+0xb6/0xf0
> xfrm_policy_inexact_list_reinsert+0xb6/0x430
> xfrm_policy_inexact_insert_node.constprop.0+0x537/0x800
> xfrm_policy_inexact_alloc_chain+0x23f/0x320
> xfrm_policy_inexact_insert+0x6b/0x590
> xfrm_policy_insert+0x3b1/0x480
> xfrm_add_policy+0x23c/0x3c0
> xfrm_user_rcv_msg+0x2d0/0x510
> netlink_rcv_skb+0x10d/0x2d0
> xfrm_netlink_rcv+0x49/0x60
> netlink_unicast+0x3fe/0x540
> netlink_sendmsg+0x528/0x970
> sock_sendmsg+0x14a/0x160
> ____sys_sendmsg+0x4fc/0x580
> ___sys_sendmsg+0xef/0x160
> __sys_sendmsg+0xf7/0x1b0
> do_syscall_64+0x3f/0x90
> entry_SYSCALL_64_after_hwframe+0x73/0xdd
>
> The root cause is:
>
> cpu 0 cpu1
> xfrm_dump_policy
> xfrm_policy_walk
> list_move_tail
> xfrm_add_policy
> ... ...
> xfrm_policy_inexact_list_reinsert
> list_for_each_entry_reverse
> if (!policy->bydst_reinsert)
> //read non-existent policy
> xfrm_dump_policy_done
> xfrm_policy_walk_done
> list_del(&walk->walk.all);
>
> If dump_one_policy() returns err (triggered by netlink socket),
> xfrm_policy_walk() will move walk initialized by socket to list
> net->xfrm.policy_all. so this socket becomes visible in the global
> policy list. The head *walk can be traversed when users add policies
> with different prefixlen and trigger xfrm_policy node merge.
>
> It can be fixed by skip such "policies" with walk.dead set to 1.
>
> Fixes: 9cf545ebd591 ("xfrm: policy: store inexact policies in a tree ordered by destination address")
> Fixes: 12a169e7d8f4 ("ipsec: Put dumpers on the dump list")
> Signed-off-by: Dong Chenchen <dongchenchen2@huawei.com>
> ---
> net/xfrm/xfrm_policy.c | 3 +++
> 1 file changed, 3 insertions(+)
>
> diff --git a/net/xfrm/xfrm_policy.c b/net/xfrm/xfrm_policy.c
> index d6b405782b63..5b56faad78e0 100644
> --- a/net/xfrm/xfrm_policy.c
> +++ b/net/xfrm/xfrm_policy.c
> @@ -848,6 +848,9 @@ static void xfrm_policy_inexact_list_reinsert(struct net *net,
> matched_d = 0;
>
> list_for_each_entry_reverse(policy, &net->xfrm.policy_all, walk.all) {
> + if (policy->walk.dead)
> + continue;
> +
Looks like we have other places that might trigger a splat, e.g.:
1816 int xfrm_policy_flush(struct net *net, u8 type, bool task_valid)
..
1827 again:
1828 list_for_each_entry(pol, &net->xfrm.policy_all, walk.all) {
1829 dir = xfrm_policy_id2dir(pol->index);
1830 if (pol->walk.dead ||
1831 dir >= XFRM_POLICY_MAX ||
'walker' has no pol->index, so I suspect this would trigger a kasan splat as well,
this needs to check walk.dead before access to pol->index.
^ permalink raw reply [flat|nested] 7+ messages in thread* Re: [PATCH net] net: xfrm: skip policies marked as dead while reinserting policies
2023-08-14 11:30 ` Florian Westphal
@ 2023-08-14 12:58 ` Dong Chenchen
0 siblings, 0 replies; 7+ messages in thread
From: Dong Chenchen @ 2023-08-14 12:58 UTC (permalink / raw)
To: fw
Cc: leon, steffen.klassert, herbert, davem, edumazet, kuba, pabeni,
timo.teras, yuehaibing, weiyongjun1, netdev, linux-kernel
Dong Chenchen <dongchenchen2@huawei.com> wrote:
>> BUG: KASAN: slab-use-after-free in xfrm_policy_inexact_list_reinsert+0xb6/0x430
>> Read of size 1 at addr ffff8881051f3bf8 by task ip/668
>>
>> CPU: 2 PID: 668 Comm: ip Not tainted 6.5.0-rc5-00182-g25aa0bebba72-dirty #64
>> Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.13 04/01/2014
>> Call Trace:
>> <TASK>
>> dump_stack_lvl+0x72/0xa0
>> print_report+0xd0/0x620
>> kasan_report+0xb6/0xf0
>> xfrm_policy_inexact_list_reinsert+0xb6/0x430
>> xfrm_policy_inexact_insert_node.constprop.0+0x537/0x800
>> xfrm_policy_inexact_alloc_chain+0x23f/0x320
>> xfrm_policy_inexact_insert+0x6b/0x590
>> xfrm_policy_insert+0x3b1/0x480
>> xfrm_add_policy+0x23c/0x3c0
>> xfrm_user_rcv_msg+0x2d0/0x510
>> netlink_rcv_skb+0x10d/0x2d0
>> xfrm_netlink_rcv+0x49/0x60
>> netlink_unicast+0x3fe/0x540
>> netlink_sendmsg+0x528/0x970
>> sock_sendmsg+0x14a/0x160
>> ____sys_sendmsg+0x4fc/0x580
>> ___sys_sendmsg+0xef/0x160
>> __sys_sendmsg+0xf7/0x1b0
>> do_syscall_64+0x3f/0x90
>> entry_SYSCALL_64_after_hwframe+0x73/0xdd
>>
>> The root cause is:
>>
>> cpu 0 cpu1
>> xfrm_dump_policy
>> xfrm_policy_walk
>> list_move_tail
>> xfrm_add_policy
>> ... ...
>> xfrm_policy_inexact_list_reinsert
>> list_for_each_entry_reverse
>> if (!policy->bydst_reinsert)
>> //read non-existent policy
>> xfrm_dump_policy_done
>> xfrm_policy_walk_done
>> list_del(&walk->walk.all);
>>
>> If dump_one_policy() returns err (triggered by netlink socket),
>> xfrm_policy_walk() will move walk initialized by socket to list
>> net->xfrm.policy_all. so this socket becomes visible in the global
>> policy list. The head *walk can be traversed when users add policies
>> with different prefixlen and trigger xfrm_policy node merge.
>>
>> It can be fixed by skip such "policies" with walk.dead set to 1.
>>
>> Fixes: 9cf545ebd591 ("xfrm: policy: store inexact policies in a tree ordered by destination address")
>> Fixes: 12a169e7d8f4 ("ipsec: Put dumpers on the dump list")
>> Signed-off-by: Dong Chenchen <dongchenchen2@huawei.com>
>> ---
>> net/xfrm/xfrm_policy.c | 3 +++
>> 1 file changed, 3 insertions(+)
>>
>> diff --git a/net/xfrm/xfrm_policy.c b/net/xfrm/xfrm_policy.c
>> index d6b405782b63..5b56faad78e0 100644
>> --- a/net/xfrm/xfrm_policy.c
>> +++ b/net/xfrm/xfrm_policy.c
>> @@ -848,6 +848,9 @@ static void xfrm_policy_inexact_list_reinsert(struct net *net,
>> matched_d = 0;
>>
>> list_for_each_entry_reverse(policy, &net->xfrm.policy_all, walk.all) {
>> + if (policy->walk.dead)
>> + continue;
>> +
>
>Looks like we have other places that might trigger a splat, e.g.:
>
>1816 int xfrm_policy_flush(struct net *net, u8 type, bool task_valid)
>..
>1827 again:
>1828 list_for_each_entry(pol, &net->xfrm.policy_all, walk.all) {
>1829 dir = xfrm_policy_id2dir(pol->index);
>1830 if (pol->walk.dead ||
>1831 dir >= XFRM_POLICY_MAX ||
>
>'walker' has no pol->index, so I suspect this would trigger a kasan splat as well,
>this needs to check walk.dead before access to pol->index.
Yes, you are right. xfrm_hash_rebuild(),xfrm_dev_policy_flush() might also trigger
similar bug. xfrm_policy_walk() add "walker" to list tail and these func use
list_for_each_entry() to traverse list.So it is unlikely to trigger the bug.
But there's still possible, i will fix it in v2.
Thanks!
^ permalink raw reply [flat|nested] 7+ messages in thread