Hi,
On Mon, Jan 30, 2023 at 02:51:49PM +0100, Greg Kroah-Hartman wrote:
> From: Sriram Yagnaraman <sriram.yagnaraman@xxxxxxxx>
>
> [ Upstream commit 98ee0077452527f971567db01386de3c3d97ce13 ]
>
> skb_header_pointer() will return NULL if offset + sizeof(_sch) exceeds
> skb->len, so this offset < skb->len test is redundant.
>
> if sch->length == 0, this will end up in an infinite loop, add a check
> for sch->length > 0
Please, keep back this patch. for_each_sctp_chunk() is fine because
do_basic_checks() already make sure what sch->length is sane.
A revert is scheduled for the net.git tree, the revert should show up
in the next 6.2-rc release.
Sorry for the inconvenience.
> Fixes: 9fb9cbb1082d ("[NETFILTER]: Add nf_conntrack subsystem.")
> Suggested-by: Florian Westphal <fw@xxxxxxxxx>
> Signed-off-by: Sriram Yagnaraman <sriram.yagnaraman@xxxxxxxx>
> Signed-off-by: Pablo Neira Ayuso <pablo@xxxxxxxxxxxxx>
> Signed-off-by: Sasha Levin <sashal@xxxxxxxxxx>
> ---
> net/netfilter/nf_conntrack_proto_sctp.c | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/net/netfilter/nf_conntrack_proto_sctp.c b/net/netfilter/nf_conntrack_proto_sctp.c
> index 3704d1c7d3c2..ee317f9a22e5 100644
> --- a/net/netfilter/nf_conntrack_proto_sctp.c
> +++ b/net/netfilter/nf_conntrack_proto_sctp.c
> @@ -155,8 +155,8 @@ static void sctp_print_conntrack(struct seq_file *s, struct nf_conn *ct)
>
> #define for_each_sctp_chunk(skb, sch, _sch, offset, dataoff, count) \
> for ((offset) = (dataoff) + sizeof(struct sctphdr), (count) = 0; \
> - (offset) < (skb)->len && \
> - ((sch) = skb_header_pointer((skb), (offset), sizeof(_sch), &(_sch))); \
> + ((sch) = skb_header_pointer((skb), (offset), sizeof(_sch), &(_sch))) && \
> + (sch)->length; \
> (offset) += (ntohs((sch)->length) + 3) & ~3, (count)++)
>
> /* Some validity checks to make sure the chunks are fine */
> --
> 2.39.0
>
>
>
