From: Shaurya Rane <ssrane_b23@ee.vjti.ac.in>

Use pskb_may_pull() to ensure the CAN ID is accessible in the linear
data buffer before reading it. A simple skb->len check is insufficient
because it only verifies the total data length but does not guarantee
the data is present in skb->data (it could be in fragments).

pskb_may_pull() both validates the length and pulls fragmented data
into the linear buffer if necessary, making it safe to directly
access skb->data.

Reported-by: syzbot+5d8269a1e099279152bc@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=5d8269a1e099279152bc
Fixes: f057bbb6f9ed ("net: em_canid: Ematch rule to match CAN frames according to their identifiers")
Signed-off-by: Shaurya Rane <ssrane_b23@ee.vjti.ac.in>
---
v2: Use pskb_may_pull() instead of skb->len check to properly
    handle fragmented skbs (Eric Dumazet)
---
 net/sched/em_canid.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/net/sched/em_canid.c b/net/sched/em_canid.c
index 5337bc462755..2214b548fab8 100644
--- a/net/sched/em_canid.c
+++ b/net/sched/em_canid.c
@@ -99,6 +99,9 @@ static int em_canid_match(struct sk_buff *skb, struct tcf_ematch *m,
 	int i;
 	const struct can_filter *lp;
 
+	if (!pskb_may_pull(skb, sizeof(canid_t)))
+		return 0;
+
 	can_id = em_canid_get_id(skb);
 
 	if (can_id & CAN_EFF_FLAG) {
-- 
2.34.1

Hello Shaurya,

many thanks that you picked up this KMSAN issue!

On 26.11.25 08:06, ssrane_b23@ee.vjti.ac.in wrote:
> From: Shaurya Rane <ssrane_b23@ee.vjti.ac.in>
> 
> Use pskb_may_pull() to ensure the CAN ID is accessible in the linear
> data buffer before reading it. A simple skb->len check is insufficient
> because it only verifies the total data length but does not guarantee
> the data is present in skb->data (it could be in fragments).
> 
> pskb_may_pull() both validates the length and pulls fragmented data
> into the linear buffer if necessary, making it safe to directly
> access skb->data.
> 
> Reported-by: syzbot+5d8269a1e099279152bc@syzkaller.appspotmail.com
> Closes: https://syzkaller.appspot.com/bug?extid=5d8269a1e099279152bc
> Fixes: f057bbb6f9ed ("net: em_canid: Ematch rule to match CAN frames according to their identifiers")
> Signed-off-by: Shaurya Rane <ssrane_b23@ee.vjti.ac.in>
> ---
> v2: Use pskb_may_pull() instead of skb->len check to properly
>      handle fragmented skbs (Eric Dumazet)
> ---
>   net/sched/em_canid.c | 3 +++
>   1 file changed, 3 insertions(+)
> 
> diff --git a/net/sched/em_canid.c b/net/sched/em_canid.c
> index 5337bc462755..2214b548fab8 100644
> --- a/net/sched/em_canid.c
> +++ b/net/sched/em_canid.c
> @@ -99,6 +99,9 @@ static int em_canid_match(struct sk_buff *skb, struct tcf_ematch *m,
>   	int i;
>   	const struct can_filter *lp;
>   
> +	if (!pskb_may_pull(skb, sizeof(canid_t)))

The EM CANID handles struct CAN frames in skb->data.

https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/include/uapi/linux/can.h#n221

The smallest type of CAN frame that can be properly handled with EM 
CANID is a Classical CAN frame which has a length of 16 bytes.

Therefore I would suggest

	if (!pskb_may_pull(skb, CAN_MTU))

instead of only checking for the first element in struct can_frame.

Many thanks and best regards,
Oliver


> +		return 0;
> +
>   	can_id = em_canid_get_id(skb);
>   
>   	if (can_id & CAN_EFF_FLAG) {

Thanks Oliver,
That makes sense EM_CANID operates on a full struct can_frame, so
ensuring CAN_MTU bytes are available is the correct approach. I’ll
update the patch accordingly and send a v3.
Thanks for the clarification!
Shaurya

On Wed, Nov 26, 2025 at 1:33 PM Oliver Hartkopp <socketcan@hartkopp.net> wrote:
>
> Hello Shaurya,
>
> many thanks that you picked up this KMSAN issue!
>
> On 26.11.25 08:06, ssrane_b23@ee.vjti.ac.in wrote:
> > From: Shaurya Rane <ssrane_b23@ee.vjti.ac.in>
> >
> > Use pskb_may_pull() to ensure the CAN ID is accessible in the linear
> > data buffer before reading it. A simple skb->len check is insufficient
> > because it only verifies the total data length but does not guarantee
> > the data is present in skb->data (it could be in fragments).
> >
> > pskb_may_pull() both validates the length and pulls fragmented data
> > into the linear buffer if necessary, making it safe to directly
> > access skb->data.
> >
> > Reported-by: syzbot+5d8269a1e099279152bc@syzkaller.appspotmail.com
> > Closes: https://syzkaller.appspot.com/bug?extid=5d8269a1e099279152bc
> > Fixes: f057bbb6f9ed ("net: em_canid: Ematch rule to match CAN frames according to their identifiers")
> > Signed-off-by: Shaurya Rane <ssrane_b23@ee.vjti.ac.in>
> > ---
> > v2: Use pskb_may_pull() instead of skb->len check to properly
> >      handle fragmented skbs (Eric Dumazet)
> > ---
> >   net/sched/em_canid.c | 3 +++
> >   1 file changed, 3 insertions(+)
> >
> > diff --git a/net/sched/em_canid.c b/net/sched/em_canid.c
> > index 5337bc462755..2214b548fab8 100644
> > --- a/net/sched/em_canid.c
> > +++ b/net/sched/em_canid.c
> > @@ -99,6 +99,9 @@ static int em_canid_match(struct sk_buff *skb, struct tcf_ematch *m,
> >       int i;
> >       const struct can_filter *lp;
> >
> > +     if (!pskb_may_pull(skb, sizeof(canid_t)))
>
> The EM CANID handles struct CAN frames in skb->data.
>
> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/include/uapi/linux/can.h#n221
>
> The smallest type of CAN frame that can be properly handled with EM
> CANID is a Classical CAN frame which has a length of 16 bytes.
>
> Therefore I would suggest
>
>         if (!pskb_may_pull(skb, CAN_MTU))
>
> instead of only checking for the first element in struct can_frame.
>
> Many thanks and best regards,
> Oliver
>
>
> > +             return 0;
> > +
> >       can_id = em_canid_get_id(skb);
> >
> >       if (can_id & CAN_EFF_FLAG) {
>