Re: [PATCH bpf] bpf: fix reading neigh ha in bpf_fib_lookup()

[Jiayuan Chen <jiayuan.chen@linux.dev>]

On 9/10/26 4:42 AM, Nikhil Ludder wrote:
> On Wed, Sep 9, 2026 at 1:18 PM Emil Tsalapatis <linux-lists@etsalapatis.com> wrote:
>> Bot is right wrt possible lockups, please adjust the seqlock
>> accordingly. Also please
>> add your full name in the SOB.
> Will fix the SOB in v2, thanks.
>
> On the lockup: I don't think it can happen here.  neigh->ha_lock has
> exactly one writer in the tree, __neigh_update() at
> net/core/neighbour.c:1495, and it already runs with softirqs disabled:
> write_lock_bh(&neigh->lock) at line 1384 covers it through
> write_unlock_bh() at line 1524.  A reader in softirq context therefore
> cannot preempt the writer on the same CPU and spin on the sequence
> count.


Yes, no deadlock.


BTW, if an IPoIB device can show up here, dmac is already truncated 
today and

the packet can't be forwarded anyway.

Shouldn't we just reject addr_len != ETH_ALEN instead of open-coding the 
copy?

Then you can use the native function instead.

> The same pattern is already used by the other lockless readers of
> neigh->ha on the tx path, neigh_resolve_output() and
> neigh_connected_output() (neighbour.c:1610 and :1639), and by the two
> recent fixes this patch follows, 57549ab90791 ("net: bridge: arp/nd
> proxy: fix reading neigh ha") and b824059a673b ("vxlan: fix reading
> neigh ha").
>
> Am I missing a writer path?  If you would still prefer a different
> form here, I am happy to respin.
>
> Thanks,
> Nikhil