re: kern/57694 (rge(4) hang)

["matthew green via gnats" <gnats-admin%NetBSD.org@localhost>]

The following reply was made to PR kern/57694; it has been noted by GNATS.

From: matthew green <mrg%eterna23.net@localhost>
To: Rin Okuyama <rokuyama.rk%gmail.com@localhost>
Cc: gnats-bugs%netbsd.org@localhost, kern-bug-people%netbsd.org@localhost,
    gnats-admin%netbsd.org@localhost, netbsd-bugs%netbsd.org@localhost,
    Thomas Klausner <wiz%NetBSD.org@localhost>, tech-net%netbsd.org@localhost
Subject: re: kern/57694 (rge(4) hang)
Date: Sat, 09 Nov 2024 06:16:08 +1100

 one of my rge(4)'s has soft-hung twice recently.  after a power
 outage, it lost connectivity soon after coming back online (on
 the order of 10s of minutes less than a hour), and ifconfig =
 
 down/up sequence restored connectivity.  then, about 2 days
 later, the same thing happened.
 
 i wrote a stupid script to notice and the down/up for me.
 
 i also started looking at rge_rxeof(), and i wonder if there's
 a bad case we get into and can't get out of without manually
 resetting the descriptors (eg, the down/up sequence):
 
    1249         for (i =3D sc->rge_ldata.rge_rxq_considx; ; i =3D RGE_NEXT=
 _RX_DESC(i)) {
 ...
    1255                 cur_rx =3D &sc->rge_ldata.rge_rx_list[i];
    1256
    1257                 if (RGE_OWN(cur_rx))
    1258                         break;
 ...
    1361         }
    1362
    1363         sc->rge_ldata.rge_rxq_considx =3D i;
 
 if for some reason rge_rxq_considx ends up pointing to a desc
 that is owned, then i is never changed from rge_rxq_considx and
 the assignment on L1363 is a no-op.  could it be being not
 filled in _ever_ for some other reason , this loop will never
 find other filled in descriptors.  can that happen?  i am not
 nearly familiar enough with ethernet drivers/hardware...
 
 i guess it might be useful to get a dump of all the descriptors
 at the hang-time, to see their status.
 
 i'll see about doing that, but this is the only machine i've
 seen this hang on and i'm loathe to reboot it ever as it does
 not reboot without power cycle, and recently, it took about 5
 power cycles for it to get past whatever problem it has.
 
 
 .mrg.