Re: [PATCH] netlink: Deal with ESRCH error in nlmsg_notify()

July 19, 2021 10:47 PM, "Yonghong Song" [off-list ref] wrote:

On 7/18/21 10:18 PM, Yajun Deng wrote:

quoted

Yonghong Song report:
The bpf selftest tc_bpf failed with latest bpf-next.
The following is the command to run and the result:
$ ./test_progs -n 132
[ 40.947571] bpf_testmod: loading out-of-tree module taints kernel.
test_tc_bpf:PASS:test_tc_bpf__open_and_load 0 nsec
test_tc_bpf:PASS:bpf_tc_hook_create(BPF_TC_INGRESS) 0 nsec
test_tc_bpf:PASS:bpf_tc_hook_create invalid hook.attach_point 0 nsec
test_tc_bpf_basic:PASS:bpf_obj_get_info_by_fd 0 nsec
test_tc_bpf_basic:PASS:bpf_tc_attach 0 nsec
test_tc_bpf_basic:PASS:handle set 0 nsec
test_tc_bpf_basic:PASS:priority set 0 nsec
test_tc_bpf_basic:PASS:prog_id set 0 nsec
test_tc_bpf_basic:PASS:bpf_tc_attach replace mode 0 nsec
test_tc_bpf_basic:PASS:bpf_tc_query 0 nsec
test_tc_bpf_basic:PASS:handle set 0 nsec
test_tc_bpf_basic:PASS:priority set 0 nsec
test_tc_bpf_basic:PASS:prog_id set 0 nsec
libbpf: Kernel error message: Failed to send filter delete notification
test_tc_bpf_basic:FAIL:bpf_tc_detach unexpected error: -3 (errno 3)
test_tc_bpf:FAIL:test_tc_internal ingress unexpected error: -3 (errno 3)
The failure seems due to the commit
cfdf0d9ae75b ("rtnetlink: use nlmsg_notify() in rtnetlink_send()")
Deal with ESRCH error in nlmsg_notify() even the report variable is zero.
Reported-by: Yonghong Song <redacted>
Signed-off-by: Yajun Deng <redacted>

Thanks for quick fix. This does fix the bpf selftest issu.
But does this change have negative impacts on other
nlmsg_notify() callers, below 1-3 items?

0 net/core/rtnetlink.c rtnetlink_send 714 return nlmsg_notify(rtnl, skb, pid, group, echo,
GFP_KERNEL);

This is exactly what we need.

1 net/core/rtnetlink.c rtnl_notify 734 nlmsg_notify(rtnl, skb, pid, group, report, flags);

It doesn't matter because there is no return value.
 

2 net/netfilter/nfnetlink.c nfnetlink_send 176 return nlmsg_notify(nfnlnet->nfnl, skb, portid,
group, echo, flags);

It only ctnetlink_conntrack_event() use the return value when call nfnetlink_send() in
net/netfilter/nf_conntrack_netlink.c, but it doesn't matter when the return value is ESRCH or zero.

3 net/netlink/genetlink.c genl_notify 1506 nlmsg_notify(sk, skb, info->snd_portid, group, report,
flags);

It doesn't matter because there is no return value.

I think the caller for nlmsg_notify() doesn't need deal with the ESRCH. It also deal with ESRCH
when report variable is not zero.

quoted

---
net/netlink/af_netlink.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/net/netlink/af_netlink.c b/net/netlink/af_netlink.c
index 380f95aacdec..24b7cf447bc5 100644
--- a/net/netlink/af_netlink.c
+++ b/net/netlink/af_netlink.c

@@ -2545,13 +2545,15 @@ int nlmsg_notify(struct sock *sk, struct sk_buff *skb, u32 portid,

/* errors reported via destination sk->sk_err, but propagate
* delivery errors if NETLINK_BROADCAST_ERROR flag is set */
err = nlmsg_multicast(sk, skb, exclude_portid, group, flags);
+ if (err == -ESRCH)
+ err = 0;
}

quoted

if (report) {

int err2;

quoted

err2 = nlmsg_unicast(sk, skb, portid);

- if (!err || err == -ESRCH)
+ if (!err)
err = err2;
}

quoted