On Sun, Oct 03, 2021 at 09:28:30PM +0200, Mark Kettenis wrote:
> Apparently some athn(4) variants are buggy and may hand us corrupted
> packets. Linux has some workarounds for this and the diff below is a
> (partial) addaptation of those workarounds. It seems the idea is that
> the error bits in the status word should not be set even if the frame
> is marked as ok. So if those bits are set, ignore the frame and bump
> the input error count. This doesn't filter out all corrupted frames,
> but it does seem to drop most of them and seems to prevent filling up
> the node cache on my access point that uses athn(4):
>
> athn0 at pci1 dev 0 function 0 "Atheros AR9281" rev 0x01: intx
> athn0: AR9280 rev 2 (2T2R), ROM rev 16, address xx:xx:xx:xx:xx:xx
>
> ok?
Yes! Thank you for tracking this down.
> Index: dev/ic/ar5008.c
> ===================================================================
> RCS file: /cvs/src/sys/dev/ic/ar5008.c,v
> retrieving revision 1.67
> diff -u -p -r1.67 ar5008.c
> --- dev/ic/ar5008.c 1 Jul 2021 11:51:55 -0000 1.67
> +++ dev/ic/ar5008.c 3 Oct 2021 19:21:11 -0000
> @@ -921,6 +921,12 @@ ar5008_rx_process(struct athn_softc *sc,
> ifp->if_ierrors++;
> goto skip;
> }
> + } else {
> + if (ds->ds_status8 & (AR_RXS8_CRC_ERR | AR_RXS8_PHY_ERR |
> + AR_RXS8_DECRYPT_CRC_ERR | AR_RXS8_MICHAEL_ERR)) {
> + ifp->if_ierrors++;
> + goto skip;
> + }
> }
>
> len = MS(ds->ds_status1, AR_RXS1_DATA_LEN);
>
>
