On Tue, Jun 08, 2021 at 12:07:54PM +0200, Florian Westphal wrote:
> Pablo Neira Ayuso <pablo@xxxxxxxxxxxxx> wrote:
> > On Tue, Jun 08, 2021 at 11:40:57AM +0200, Phil Sutter wrote:
> > > While iterating through an SCTP packet's chunks, skb_header_pointer() is
> > > called for the minimum expected chunk header size. If (that part of) the
> > > skbuff is non-linear, the following memcpy() may read data past
> > > temporary buffer '_sch'. Use skb_copy_bits() instead which does the
> > > right thing in this situation.
> > >
> > > Fixes: 133dc203d77df ("netfilter: nft_exthdr: Support SCTP chunks")
> > > Suggested-by: Florian Westphal <fw@xxxxxxxxx>
> > > Signed-off-by: Phil Sutter <phil@xxxxxx>
> > > ---
> > > Changes since v1:
> > > - skb_copy_bits() call error handling added
> > > ---
> > > net/netfilter/nft_exthdr.c | 4 +++-
> > > 1 file changed, 3 insertions(+), 1 deletion(-)
> > >
> > > diff --git a/net/netfilter/nft_exthdr.c b/net/netfilter/nft_exthdr.c
> > > index 1b0579cb62d08..7f705b5c09de8 100644
> > > --- a/net/netfilter/nft_exthdr.c
> > > +++ b/net/netfilter/nft_exthdr.c
> > > @@ -327,7 +327,9 @@ static void nft_exthdr_sctp_eval(const struct nft_expr *expr,
> > > break;
> > >
> > > dest[priv->len / NFT_REG32_SIZE] = 0;
> > > - memcpy(dest, (char *)sch + priv->offset, priv->len);
> > > + if (skb_copy_bits(pkt->skb, offset + priv->offset,
> > > + dest, priv->len) < 0)
> > > + break;
> >
> > Hm, it looks like tcp exthdt matching has the same problem?
>
> Tcp exthdr either copies from skb linear area or a on-stack scratch space,
> it looks ok to me.
Indeed, I overlook the nft_tcp_header_pointer() call a few lines before.
