Andre Hedrick wrote:
>
> On Fri, 25 Feb 2000, Andrea Arcangeli wrote:
>
> > here the fix for the tulip deadlock at against 2.3.48pre1:
>
> > @@ -1009,6 +1003,9 @@
> > printk(KERN_INFO "%s: %s rev %d at %#3lx,",
> > dev->name, tulip_tbl[chip_idx].chip_name, chip_rev, ioaddr);
> >
> > + /* initalize before accessing the chipsets */
> > + spin_lock_init(&tp->tx_lock);
> > +
>
> Little overkill, prettier than my kludge, and looks like a decenct
> standard to setup "spin_lock_init(....);"
See my patch posted under subject "Tulip driver overhaul" which was also
sent to Linus [separately]
Short term, what should really be done is something like
outl_csr6() {
if (chip != xircom)
outl(data)
return
...ugly xircom-specific locking and such...
}
The lock should not even be held for most adapters. (Linus pointed out
that even my version, as posted to lkml and sent to him, is still ugly
because of said locking)
However, the REAL solution is to completely audit this driver for SMP.
eepro100 should be a good model, and 8139too after 2.3.48-pre2 (or
whatever) is released.
-- Jeff Garzik | Building 1024 | Viva la open source! MandrakeSoft, Inc. |- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Tue Feb 29 2000 - 21:00:11 EST