netdev - RE: [PATCH net-next] net/sched: fix a couple of splats in the error path of tfc_gate

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Message-ID: <VE1PR04MB6496BA8407706123819B9D31928F0@VE1PR04MB6496.eurprd04.prod.outlook.com>
Date:   Fri, 29 May 2020 02:43:23 +0000
From:   Po Liu <po.liu@....com>
To:     Davide Caratti <dcaratti@...hat.com>,
        Jamal Hadi Salim <jhs@...atatu.com>,
        "netdev@...r.kernel.org" <netdev@...r.kernel.org>,
        "David S. Miller" <davem@...emloft.net>
CC:     Ivan Vecera <ivecera@...hat.com>
Subject: RE: [PATCH net-next] net/sched: fix a couple of splats in the error
 path of tfc_gate_init()

Hi Davide,


> -----Original Message-----
> From: Davide Caratti <dcaratti@...hat.com>
> Sent: 2020年5月29日 6:06
> To: Jamal Hadi Salim <jhs@...atatu.com>; Po Liu <po.liu@....com>;
> netdev@...r.kernel.org; David S. Miller <davem@...emloft.net>
> Cc: Ivan Vecera <ivecera@...hat.com>
> Subject: [PATCH net-next] net/sched: fix a couple of splats in the
> error path of tfc_gate_init()
> 
> trying to configure TC 'act_gate' rules with invalid control actions, the
> following splat can be observed:
> 
>  general protection fault, probably for non-canonical address
> 0xdffffc0000000002: 0000 [#1] SMP KASAN NOPTI
>  KASAN: null-ptr-deref in range [0x0000000000000010-
> 0x0000000000000017]
>  CPU: 1 PID: 2143 Comm: tc Not tainted 5.7.0-rc6+ #168  Hardware name:
> Red Hat KVM, BIOS 1.11.1-4.module+el8.1.0+4066+0f1aadab 04/01/2014
>  RIP: 0010:hrtimer_active+0x56/0x290
>  [...]
>   Call Trace:
>   hrtimer_try_to_cancel+0x6d/0x330
>   hrtimer_cancel+0x11/0x20
>   tcf_gate_cleanup+0x15/0x30 [act_gate]
>   tcf_action_cleanup+0x58/0x170
>   __tcf_action_put+0xb0/0xe0
>   __tcf_idr_release+0x68/0x90
>   tcf_gate_init+0x7c7/0x19a0 [act_gate]
>   tcf_action_init_1+0x60f/0x960
>   tcf_action_init+0x157/0x2a0
>   tcf_action_add+0xd9/0x2f0
>   tc_ctl_action+0x2a3/0x39d
>   rtnetlink_rcv_msg+0x5f3/0x920
>   netlink_rcv_skb+0x121/0x350
>   netlink_unicast+0x439/0x630
>   netlink_sendmsg+0x714/0xbf0
>   sock_sendmsg+0xe2/0x110
>   ____sys_sendmsg+0x5b4/0x890
>   ___sys_sendmsg+0xe9/0x160
>   __sys_sendmsg+0xd3/0x170
>   do_syscall_64+0x9a/0x370
>   entry_SYSCALL_64_after_hwframe+0x44/0xa9
> 
> this is caused by hrtimer_cancel(), running before hrtimer_init(). Fix it
> ensuring to call hrtimer_cancel() only if clockid is valid, and the timer has

Can you share the test step? Clockid by default is set with CLOCK_TAI. And INIT_LIST_HEAD() also called in the init.
So I think maybe there is better method to avoid the duplicated code.

> been initialized. After fixing this splat, the same error path causes another
> problem:
> 
>  general protection fault, probably for non-canonical address
> 0xdffffc0000000000: 0000 [#1] SMP KASAN NOPTI
>  KASAN: null-ptr-deref in range [0x0000000000000000-
> 0x0000000000000007]
>  CPU: 1 PID: 980 Comm: tc Not tainted 5.7.0-rc6+ #168  Hardware name:
> Red Hat KVM, BIOS 1.11.1-4.module+el8.1.0+4066+0f1aadab 04/01/2014
>  RIP: 0010:release_entry_list+0x4a/0x240 [act_gate]  [...]  Call Trace:
>   tcf_action_cleanup+0x58/0x170
>   __tcf_action_put+0xb0/0xe0
>   __tcf_idr_release+0x68/0x90
>   tcf_gate_init+0x7ab/0x19a0 [act_gate]
>   tcf_action_init_1+0x60f/0x960
>   tcf_action_init+0x157/0x2a0
>   tcf_action_add+0xd9/0x2f0
>   tc_ctl_action+0x2a3/0x39d
>   rtnetlink_rcv_msg+0x5f3/0x920
>   netlink_rcv_skb+0x121/0x350
>   netlink_unicast+0x439/0x630
>   netlink_sendmsg+0x714/0xbf0
>   sock_sendmsg+0xe2/0x110
>   ____sys_sendmsg+0x5b4/0x890
>   ___sys_sendmsg+0xe9/0x160
>   __sys_sendmsg+0xd3/0x170
>   do_syscall_64+0x9a/0x370
>   entry_SYSCALL_64_after_hwframe+0x44/0xa9
> 
> the problem is similar: tcf_action_cleanup() was trying to release a list
> without initializing it first. Ensure that INIT_LIST_HEAD() is called for every
> newly created 'act_gate' action, same as what was done to 'act_ife'
> with commit 44c23d71599f ("net/sched: act_ife: initalize ife->metalist
> earlier").
> 
> Fixes: a51c328df310 ("net: qos: introduce a gate control flow action")
> CC: Ivan Vecera <ivecera@...hat.com>
> Signed-off-by: Davide Caratti <dcaratti@...hat.com>
> ---
>  net/sched/act_gate.c | 9 ++++++---
>  1 file changed, 6 insertions(+), 3 deletions(-)
> 
> diff --git a/net/sched/act_gate.c b/net/sched/act_gate.c index
> 35fc48795541c..9c628591f452c 100644
> --- a/net/sched/act_gate.c
> +++ b/net/sched/act_gate.c
> @@ -331,6 +331,10 @@ static int tcf_gate_init(struct net *net, struct
> nlattr *nla,
>                 tcf_idr_release(*a, bind);
>                 return -EEXIST;
>         }
> +       if (ret == ACT_P_CREATED) {
> +               to_gate(*a)->param.tcfg_clockid = -1;
> +               INIT_LIST_HEAD(&(to_gate(*a)->param.entries));
> +       }
> 
>         if (tb[TCA_GATE_PRIORITY])
>                 prio = nla_get_s32(tb[TCA_GATE_PRIORITY]);
> @@ -377,7 +381,6 @@ static int tcf_gate_init(struct net *net, struct nlattr
> *nla,
>                         goto chain_put;
>         }
> 
> -       INIT_LIST_HEAD(&p->entries);
>         if (tb[TCA_GATE_ENTRY_LIST]) {
>                 err = parse_gate_list(tb[TCA_GATE_ENTRY_LIST], p, extack);
>                 if (err < 0)
> @@ -449,9 +452,9 @@ static void tcf_gate_cleanup(struct tc_action *a)
>         struct tcf_gate *gact = to_gate(a);
>         struct tcf_gate_params *p;
> 
> -       hrtimer_cancel(&gact->hitimer);
> -
>         p = &gact->param;
> +       if (p->tcfg_clockid != -1)
> +               hrtimer_cancel(&gact->hitimer);
> 
>         release_entry_list(&p->entries);  }
> --
> 2.26.2

Thanks!

Br,
Po Liu