RE: [PATCH net v4 1/1] can: can_skb_set_owner(): fix ref counting if socket was closed before setting skb ownership
> -Original Message-
> From: Joakim Zhang
> Sent: 2021年3月1日 18:57
> To: Oleksij Rempel ; m...@pengutronix.de; David S.
> Miller ; Jakub Kicinski ; Oliver
> Hartkopp ; Robin van der Gracht
>
> Cc: Andre Naujoks ; Eric Dumazet
> ; ker...@pengutronix.de; linux-...@vger.kernel.org;
> net...@vger.kernel.org; linux-kernel@vger.kernel.org
> Subject: RE: [PATCH net v4 1/1] can: can_skb_set_owner(): fix ref counting if
> socket was closed before setting skb ownership
>
>
> > -Original Message-
> > From: Oleksij Rempel
> > Sent: 2021年2月26日 17:25
> > To: m...@pengutronix.de; David S. Miller ; Jakub
> > Kicinski ; Oliver Hartkopp ;
> > Robin van der Gracht
> > Cc: Oleksij Rempel ; Andre Naujoks
> > ; Eric Dumazet ;
> > ker...@pengutronix.de; linux-...@vger.kernel.org;
> > net...@vger.kernel.org; linux-kernel@vger.kernel.org
> > Subject: [PATCH net v4 1/1] can: can_skb_set_owner(): fix ref counting
> > if socket was closed before setting skb ownership
> >
> > There are two ref count variables controlling the free()ing of a socket:
> > - struct sock::sk_refcnt - which is changed by sock_hold()/sock_put()
> > - struct sock::sk_wmem_alloc - which accounts the memory allocated by
> > the skbs in the send path.
> >
> > In case there are still TX skbs on the fly and the socket() is closed,
> > the struct sock::sk_refcnt reaches 0. In the TX-path the CAN stack
> > clones an "echo" skb, calls sock_hold() on the original socket and
> > references it. This produces the following back trace:
> >
> > | WARNING: CPU: 0 PID: 280 at lib/refcount.c:25
> > | refcount_warn_saturate+0x114/0x134
> > | refcount_t: addition on 0; use-after-free.
> > | Modules linked in: coda_vpu(E) v4l2_jpeg(E) videobuf2_vmalloc(E)
> > imx_vdoa(E)
> > | CPU: 0 PID: 280 Comm: test_can.sh Tainted: GE
> > 5.11.0-04577-gf8ff6603c617 #203
> > | Hardware name: Freescale i.MX6 Quad/DualLite (Device Tree)
> > | Backtrace:
> > | [<80bafea4>] (dump_backtrace) from [<80bb0280>]
> > | (show_stack+0x20/0x24)
> > | r7: r6:600f0113 r5: r4:81441220 [<80bb0260>]
> > | (show_stack) from [<80bb593c>] (dump_stack+0xa0/0xc8) [<80bb589c>]
> > | (dump_stack) from [<8012b268>] (__warn+0xd4/0x114) r9:0019
> > | r8:80f4a8c2 r7:83e4150c r6: r5:0009 r4:80528f90
> > | [<8012b194>] (__warn) from [<80bb09c4>]
> > | (warn_slowpath_fmt+0x88/0xc8)
> > | r9:83f26400 r8:80f4a8d1 r7:0009 r6:80528f90 r5:0019
> > | r4:80f4a8c2 [<80bb0940>] (warn_slowpath_fmt) from [<80528f90>]
> > | (refcount_warn_saturate+0x114/0x134) r8: r7:
> > | r6:82b44000 r5:834e5600 r4:83f4d540 [<80528e7c>]
> > | (refcount_warn_saturate) from [<8079a4c8>]
> > | (__refcount_add.constprop.0+0x4c/0x50)
> > | [<8079a47c>] (__refcount_add.constprop.0) from [<8079a57c>]
> > | (can_put_echo_skb+0xb0/0x13c) [<8079a4cc>] (can_put_echo_skb) from
> > | [<8079ba98>] (flexcan_start_xmit+0x1c4/0x230) r9:0010
> > | r8:83f48610
> > | r7:0fdc r6:0c08 r5:82b44000 r4:834e5600 [<8079b8d4>]
> > | (flexcan_start_xmit) from [<80969078>] (netdev_start_xmit+0x44/0x70)
> > | r9:814c0ba0 r8:80c8790c r7: r6:834e5600 r5:82b44000
> > | r4:82ab1f00 [<80969034>] (netdev_start_xmit) from [<809725a4>]
> > | (dev_hard_start_xmit+0x19c/0x318) r9:814c0ba0 r8:
> > | r7:82ab1f00
> > | r6:82b44000 r5: r4:834e5600 [<80972408>]
> > | (dev_hard_start_xmit) from [<809c6584>] (sch_direct_xmit+0xcc/0x264)
> > | r10:834e5600
> > | r9: r8: r7:82b44000 r6:82ab1f00 r5:834e5600
> > | r4:83f27400 [<809c64b8>] (sch_direct_xmit) from [<809c6c0c>]
> > | (__qdisc_run+0x4f0/0x534)
> >
> > To fix this problem, only set skb ownership to sockets which have
> > still a ref count > 0.
> >
> > Cc: Oliver Hartkopp
> > Cc: Andre Naujoks
> > Suggested-by: Eric Dumazet
> > Fixes: 0ae89beb283a ("can: add destructor for self generated skbs")
> > Signed-off-by: Oleksij Rempel
>
> I will give out a test result tomorrow when the board is available.
I also met this issue in the past and this patch indeed fix it. Thanks Oleksij
Rempe.
Tested-by: Joakim Zhang
Best Regards,
Joakim Zhang
> Best Regards,
> Joakim Zhang
> > ---
> > include/linux/can/skb.h | 8 ++--
> > 1 file changed, 6 insertions(+), 2 deletions(-)
> >
> > diff --git a/include/linux/can/sk
RE: [PATCH net v4 1/1] can: can_skb_set_owner(): fix ref counting if socket was closed before setting skb ownership
> -Original Message-
> From: Oleksij Rempel
> Sent: 2021年2月26日 17:25
> To: m...@pengutronix.de; David S. Miller ; Jakub
> Kicinski ; Oliver Hartkopp ;
> Robin van der Gracht
> Cc: Oleksij Rempel ; Andre Naujoks
> ; Eric Dumazet ;
> ker...@pengutronix.de; linux-...@vger.kernel.org; net...@vger.kernel.org;
> linux-kernel@vger.kernel.org
> Subject: [PATCH net v4 1/1] can: can_skb_set_owner(): fix ref counting if
> socket
> was closed before setting skb ownership
>
> There are two ref count variables controlling the free()ing of a socket:
> - struct sock::sk_refcnt - which is changed by sock_hold()/sock_put()
> - struct sock::sk_wmem_alloc - which accounts the memory allocated by
> the skbs in the send path.
>
> In case there are still TX skbs on the fly and the socket() is closed, the
> struct
> sock::sk_refcnt reaches 0. In the TX-path the CAN stack clones an "echo" skb,
> calls sock_hold() on the original socket and references it. This produces the
> following back trace:
>
> | WARNING: CPU: 0 PID: 280 at lib/refcount.c:25
> | refcount_warn_saturate+0x114/0x134
> | refcount_t: addition on 0; use-after-free.
> | Modules linked in: coda_vpu(E) v4l2_jpeg(E) videobuf2_vmalloc(E)
> imx_vdoa(E)
> | CPU: 0 PID: 280 Comm: test_can.sh Tainted: GE
> 5.11.0-04577-gf8ff6603c617 #203
> | Hardware name: Freescale i.MX6 Quad/DualLite (Device Tree)
> | Backtrace:
> | [<80bafea4>] (dump_backtrace) from [<80bb0280>] (show_stack+0x20/0x24)
> | r7: r6:600f0113 r5: r4:81441220 [<80bb0260>]
> | (show_stack) from [<80bb593c>] (dump_stack+0xa0/0xc8) [<80bb589c>]
> | (dump_stack) from [<8012b268>] (__warn+0xd4/0x114) r9:0019
> | r8:80f4a8c2 r7:83e4150c r6: r5:0009 r4:80528f90
> | [<8012b194>] (__warn) from [<80bb09c4>] (warn_slowpath_fmt+0x88/0xc8)
> | r9:83f26400 r8:80f4a8d1 r7:0009 r6:80528f90 r5:0019
> | r4:80f4a8c2 [<80bb0940>] (warn_slowpath_fmt) from [<80528f90>]
> | (refcount_warn_saturate+0x114/0x134) r8: r7:
> | r6:82b44000 r5:834e5600 r4:83f4d540 [<80528e7c>]
> | (refcount_warn_saturate) from [<8079a4c8>]
> | (__refcount_add.constprop.0+0x4c/0x50)
> | [<8079a47c>] (__refcount_add.constprop.0) from [<8079a57c>]
> | (can_put_echo_skb+0xb0/0x13c) [<8079a4cc>] (can_put_echo_skb) from
> | [<8079ba98>] (flexcan_start_xmit+0x1c4/0x230) r9:0010 r8:83f48610
> | r7:0fdc r6:0c08 r5:82b44000 r4:834e5600 [<8079b8d4>]
> | (flexcan_start_xmit) from [<80969078>] (netdev_start_xmit+0x44/0x70)
> | r9:814c0ba0 r8:80c8790c r7: r6:834e5600 r5:82b44000
> | r4:82ab1f00 [<80969034>] (netdev_start_xmit) from [<809725a4>]
> | (dev_hard_start_xmit+0x19c/0x318) r9:814c0ba0 r8: r7:82ab1f00
> | r6:82b44000 r5: r4:834e5600 [<80972408>] (dev_hard_start_xmit)
> | from [<809c6584>] (sch_direct_xmit+0xcc/0x264) r10:834e5600
> | r9: r8: r7:82b44000 r6:82ab1f00 r5:834e5600
> | r4:83f27400 [<809c64b8>] (sch_direct_xmit) from [<809c6c0c>]
> | (__qdisc_run+0x4f0/0x534)
>
> To fix this problem, only set skb ownership to sockets which have still a ref
> count > 0.
>
> Cc: Oliver Hartkopp
> Cc: Andre Naujoks
> Suggested-by: Eric Dumazet
> Fixes: 0ae89beb283a ("can: add destructor for self generated skbs")
> Signed-off-by: Oleksij Rempel
I will give out a test result tomorrow when the board is available.
Best Regards,
Joakim Zhang
> ---
> include/linux/can/skb.h | 8 ++--
> 1 file changed, 6 insertions(+), 2 deletions(-)
>
> diff --git a/include/linux/can/skb.h b/include/linux/can/skb.h index
> 685f34cfba20..d82018cc0d0b 100644
> --- a/include/linux/can/skb.h
> +++ b/include/linux/can/skb.h
> @@ -65,8 +65,12 @@ static inline void can_skb_reserve(struct sk_buff *skb)
>
> static inline void can_skb_set_owner(struct sk_buff *skb, struct sock *sk) {
> - if (sk) {
> - sock_hold(sk);
> + /*
> + * If the socket has already been closed by user space, the refcount may
> + * already be 0 (and the socket will be freed after the last TX skb has
> + * been freed). So only increase socket refcount if the refcount is > 0.
> + */
> + if (sk && refcount_inc_not_zero(>sk_refcnt)) {
> skb->destructor = sock_efree;
> skb->sk = sk;
> }
> --
> 2.29.2
Re: [PATCH net v4 1/1] can: can_skb_set_owner(): fix ref counting if socket was closed before setting skb ownership
On 26.02.21 10:24, Oleksij Rempel wrote:
There are two ref count variables controlling the free()ing of a socket:
- struct sock::sk_refcnt - which is changed by sock_hold()/sock_put()
- struct sock::sk_wmem_alloc - which accounts the memory allocated by
the skbs in the send path.
In case there are still TX skbs on the fly and the socket() is closed,
the struct sock::sk_refcnt reaches 0. In the TX-path the CAN stack
clones an "echo" skb, calls sock_hold() on the original socket and
references it. This produces the following back trace:
| WARNING: CPU: 0 PID: 280 at lib/refcount.c:25
refcount_warn_saturate+0x114/0x134
| refcount_t: addition on 0; use-after-free.
| Modules linked in: coda_vpu(E) v4l2_jpeg(E) videobuf2_vmalloc(E) imx_vdoa(E)
| CPU: 0 PID: 280 Comm: test_can.sh Tainted: GE
5.11.0-04577-gf8ff6603c617 #203
| Hardware name: Freescale i.MX6 Quad/DualLite (Device Tree)
| Backtrace:
| [<80bafea4>] (dump_backtrace) from [<80bb0280>] (show_stack+0x20/0x24)
r7: r6:600f0113 r5: r4:81441220
| [<80bb0260>] (show_stack) from [<80bb593c>] (dump_stack+0xa0/0xc8)
| [<80bb589c>] (dump_stack) from [<8012b268>] (__warn+0xd4/0x114) r9:0019
r8:80f4a8c2 r7:83e4150c r6: r5:0009 r4:80528f90
| [<8012b194>] (__warn) from [<80bb09c4>] (warn_slowpath_fmt+0x88/0xc8)
r9:83f26400 r8:80f4a8d1 r7:0009 r6:80528f90 r5:0019 r4:80f4a8c2
| [<80bb0940>] (warn_slowpath_fmt) from [<80528f90>]
(refcount_warn_saturate+0x114/0x134) r8: r7: r6:82b44000 r5:834e5600
r4:83f4d540
| [<80528e7c>] (refcount_warn_saturate) from [<8079a4c8>]
(__refcount_add.constprop.0+0x4c/0x50)
| [<8079a47c>] (__refcount_add.constprop.0) from [<8079a57c>]
(can_put_echo_skb+0xb0/0x13c)
| [<8079a4cc>] (can_put_echo_skb) from [<8079ba98>]
(flexcan_start_xmit+0x1c4/0x230) r9:0010 r8:83f48610 r7:0fdc r6:0c08
r5:82b44000 r4:834e5600
| [<8079b8d4>] (flexcan_start_xmit) from [<80969078>]
(netdev_start_xmit+0x44/0x70) r9:814c0ba0 r8:80c8790c r7: r6:834e5600 r5:82b44000
r4:82ab1f00
| [<80969034>] (netdev_start_xmit) from [<809725a4>]
(dev_hard_start_xmit+0x19c/0x318) r9:814c0ba0 r8: r7:82ab1f00 r6:82b44000
r5: r4:834e5600
| [<80972408>] (dev_hard_start_xmit) from [<809c6584>]
(sch_direct_xmit+0xcc/0x264) r10:834e5600 r9: r8: r7:82b44000 r6:82ab1f00
r5:834e5600 r4:83f27400
| [<809c64b8>] (sch_direct_xmit) from [<809c6c0c>] (__qdisc_run+0x4f0/0x534)
To fix this problem, only set skb ownership to sockets which have still
a ref count > 0.
Cc: Oliver Hartkopp
Cc: Andre Naujoks
Suggested-by: Eric Dumazet
Fixes: 0ae89beb283a ("can: add destructor for self generated skbs")
Signed-off-by: Oleksij Rempel
---
include/linux/can/skb.h | 8 ++--
1 file changed, 6 insertions(+), 2 deletions(-)
diff --git a/include/linux/can/skb.h b/include/linux/can/skb.h
index 685f34cfba20..d82018cc0d0b 100644
--- a/include/linux/can/skb.h
+++ b/include/linux/can/skb.h
@@ -65,8 +65,12 @@ static inline void can_skb_reserve(struct sk_buff *skb)
static inline void can_skb_set_owner(struct sk_buff *skb, struct sock *sk)
{
- if (sk) {
- sock_hold(sk);
+ /*
+* If the socket has already been closed by user space, the refcount may
+* already be 0 (and the socket will be freed after the last TX skb has
+* been freed). So only increase socket refcount if the refcount is > 0.
+*/
+ if (sk && refcount_inc_not_zero(>sk_refcnt)) {
skb->destructor = sock_efree;
skb->sk = sk;
}
Ah, now I got the problem. You are replacing sock_hold(). Maybe I was a
bit slow-witted here - but with the comment it is ok.
Reviewed-by: Oliver Hartkopp
Many thanks!
Oliver
