Re: [PATCH net v2] net: stmmac: Clear DMA errors for multi-channel interrupts

[Paolo Abeni]

On 11/7/26 11:01, Magnus Leatti wrote:
> > Should the Fixes: tag point at 1b9707e6f1a9 instead, or also include it?
> > As written, stable may backport this to 5.13+ trees that cannot hit these
> > storms.
>
> A data point for the RBU case from the YT6801, same 64-byte UDP RX
> flood, all on the same 7.0.13 kernel:
>
>    stmmac from 7.0.13 (has 6ccf12ae111e, not 1b9707e6f1a9):
>      no storm; the MAC interrupt never fires (4 streams for 10 s,
>      and 8 streams plus TX for 30 s)
>    stmmac from 7.2.11, built for 7.0.13:
>      storm, the machine hangs
>    the same, with only 1b9707e6f1a9 reverted:
>      no storm
>
> So on this hardware the RBU storm needs 1b9707e6f1a9. I did not test
> the RPS case.
>
> > Since RBU and RPS were enabled only for statistics, was masking them in
> > multi-MSI mode considered as an alternative?

AFAICS 'statistics' here means accounting of some corner cases errors. I
guess this could be a preferable route.

/P