On Tue, Jul 13, 2021 at 9:21 PM Florian Westphal <fw@xxxxxxxxx> wrote:
>
> Dongliang Mu <mudongliangabcd@xxxxxxxxx> wrote:
> > In nf_tables_commit, if nf_tables_commit_audit_alloc fails, it does not
> > free the adp variable.
> >
> > Fix this by freeing the linked list with head adl.
> >
> > backtrace:
> > kmalloc include/linux/slab.h:591 [inline]
> > kzalloc include/linux/slab.h:721 [inline]
> > nf_tables_commit_audit_alloc net/netfilter/nf_tables_api.c:8439 [inline]
> > nf_tables_commit+0x16e/0x1760 net/netfilter/nf_tables_api.c:8508
> > nfnetlink_rcv_batch+0x512/0xa80 net/netfilter/nfnetlink.c:562
> > nfnetlink_rcv_skb_batch net/netfilter/nfnetlink.c:634 [inline]
> > nfnetlink_rcv+0x1fa/0x220 net/netfilter/nfnetlink.c:652
> > netlink_unicast_kernel net/netlink/af_netlink.c:1314 [inline]
> > netlink_unicast+0x2c7/0x3e0 net/netlink/af_netlink.c:1340
> > netlink_sendmsg+0x36b/0x6b0 net/netlink/af_netlink.c:1929
> > sock_sendmsg_nosec net/socket.c:702 [inline]
> > sock_sendmsg+0x56/0x80 net/socket.c:722
> >
> > Reported-by: syzbot <syzkaller@xxxxxxxxxxxxxxxx>
> > Reported-by: kernel test robot <lkp@xxxxxxxxx>
> > Fixes: c520292f29b8 ("audit: log nftables configuration change events once per table")
> > Signed-off-by: Dongliang Mu <mudongliangabcd@xxxxxxxxx>
> > ---
> > v1->v2: fix the compile issue
> > net/netfilter/nf_tables_api.c | 12 ++++++++++++
> > 1 file changed, 12 insertions(+)
> >
> > diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c
> > index 390d4466567f..7f45b291be13 100644
> > --- a/net/netfilter/nf_tables_api.c
> > +++ b/net/netfilter/nf_tables_api.c
> > @@ -8444,6 +8444,16 @@ static int nf_tables_commit_audit_alloc(struct list_head *adl,
> > return 0;
> > }
> >
> > +static void nf_tables_commit_free(struct list_head *adl)
>
> nf_tables_commit_audit_free?
What do you mean? Modify the name of newly added function to
nf_tables_commit_audit_free?
>
> Aside from that, there should be a followup patch (for nf-next),
> adding empty inline functions in case of CONFIG_AUDITSYSCALL=n.
I see. I prefer to send them (two implementations of the newly added
function) in version v2.
>
> Right now it does pointless aggregation for the AUDIT=n case.
