> On Sat, 27 Dec 1997, Tim Waugh wrote:
>
> > Here's the oops report (reconstructed by hand) I get when booting 2.1.76
> > (compiled with gcc-2.7.2.1) on my Alpha.
> >
> > Tim.
> > */
> >
> > Unable to handle kernel paging request at virtual address ffffbe1c0160918
> > swapper(1): Oops 1
> > pc = [<fffffc0000424678>] ra = [<fffffc000040341c>] ps = 0000
> > r0 = 0000000000000000 r1 = fffffc01c0000000 r2 = ffffffff0000b000
>
> (snip!)
>
> That's precisely what I was seeing with the DE45x ethernet driver. I'm
> currently running 2.1.76 with the tulip driver selected in kernel config,
> and Donald Becker's 0.83 beta code. The following little patch is
> required to build under newer kernels:
This didn't seem to apply cleanly, probably because I have version 0.86
not .83. I needed this patch:
--- tulip.c-086 Sat Jan 10 23:42:01 1998
+++ linux/drivers/net/tulip.c Sat Jan 10 23:28:43 1998
@@ -266,6 +266,8 @@
#ifndef PCI_VENDOR_ID_LITEON
#define PCI_VENDOR_ID_LITEON 0x11AD
+#endif
+#ifndef PCI_DEVICE_ID_PNIC
#define PCI_DEVICE_ID_PNIC 0x0002
#define PCI_DEVICE_ID_PNIC_X 0x0168
#endif
@@ -1774,7 +1776,6 @@
if (skb == NULL || skb->len <= 0) {
printk(KERN_ERR "%s: Obsolete driver layer request made: skbuff==NULL.\n",
dev->name);
- dev_tint(dev);
return 0;
}
#endif
> The driver is on Donald's ftp site at:
>
> ftp://cesdis.gsfc.nasa.gov/pub/linux/drivers/tulip-drivers/tulip.c-086
>
> Give it a try.
It compiled fine once I'd patched as above, although netdevice.h seems to
be alone in believing that dev_tint exists. Once I tried booting it, I
got an oops. The last thing I saw before it froze was "die_if_kernel
recursion detected". It happened during ifconfig, but the addresses look
like they're near schedule(). I couldn't write down the whole thing
because it didn't fit on screen (and the first oops was long gone anyhow).
TGA scrollback would have been nice...
Tim.
*/