Garrett D'Amore wrote: > dnet.c:302: D_HOTPLUG is not needed anymore. (It is implicitly assumed.) > > Remove the cmn_err, and associated test at dnet.c 1912, that is not VLAN > safe. > > Increase rx_buf_size at dnet.c:237 by 4 to account for VLAN tagging. > > The screwy logic at dnet.c:1530 could be removed. Probably could use a > stat to track pullups, but it doesn't matter that much. I don't think > anyone uses that for debug anymore. > > The check at dnet.c:1625 needs to be ETHERMAX + 4, to be VLAN safe. > > At around dnet.c:2735, you need to change the code -- the err_summary > bit is set for VLAN packets -- probably need to look at the individual > bits. I'd check to see what mxfe does here. mxfe is very nearly an > identical clone to dnet. (Including having really crummy variations > between SYM and MII xcvrs.) > > There could be other VLAN related issues I've not uncovered yet, but > those are the ones that stand out the most to me. (Possibly a watchdog > or jabber timer...) Testing will tell if we found 'em all or not. > > I'll do more review when you have nuked the cmn_err's. > > -- Garrett >
Will do. I'll also get rid of _depends_on and modify usr/src/uts/intel/dnet/Makefile accordingly. Also, there is currently a prototype for dnet_devinfo (dnet.c:117), but no function definition - can this be safely removed as well? Steve -- Yet magic and hierarchy arise from the same source, and this source has a null pointer. Reference the NULL within NULL, it is the gateway to all wizardry. _______________________________________________ driver-discuss mailing list driver-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/driver-discuss
