Re: [PATCH net v3 2/2] bonding: fix u32 overflow in compute_gap()

[Nikolay Aleksandrov]

On 19/08/2026 05:09, Hangbin Liu wrote:
> On Wed, Aug 19, 2026 at 10:07:22AM +0800, Hangbin Liu wrote:
> > > > > ... this here, as u64_stats_update_begin doesn't provide exclusive access, so writers
> > > > > must do that themselves, so you can't be sure what value will end up, the zeroing
> > > > > might not work at all and can get overwritten
> > > >
> > > > I meant - it doesn't improve on the current situation where it can also happen. :)
> >
> > Ah, yes. I forgot this. The reset_unbalanced_load() could be called on any
> > CPU, which conflicts with other writers.
> >
> > I re-checked the code. unbalanced_load is only called in two situations:
> > 1. To rebalance the load in bond_alb_monitor(), which only executes once
> >     every 10 seconds.
> > 2. !tx_slave in bond_do_alb_xmit(), which is only for multicast/broadcast
> >     traffic. This traffic shouldn't be significant.
> >
> > So looks using spin_lock here is acceptable. What do you think?
>
> I mean, drop the per-CPU design directly and use spin_lock to protect the data.
>
> Hangbin

hmm why don't you change the way the reset is done? *untested* but in theory
you could just record the values at a reset "moment" in reset unbalanced and
just use the delta, so it becomes a reader and there is only 1 writer left (tx).
Keep the counters only increasing (important), only record a snapshot at a reset
moment, count current total bytes (sum all per-cpu data), decrement the previous
total from it and use that as the "interval bytes" to div.

Cheers,
 Nik