TBD:
* look at deferring rx frames rather than discarding (as per tulip)
* handle tx ring full as per tulip
- * performace test to tune rx_copybreak
+ * performance test to tune rx_copybreak
Most of my modifications relate to the braindead big-endian
implementation by Intel. When the i596 is operating in
if (!dev->irq) {
printk(KERN_ERR "%s: IRQ not found for i82596 at 0x%lx\n",
- __FILE__, dev->hpa.start);
+ __FILE__, (unsigned long)dev->hpa.start);
return -ENODEV;
}
- printk(KERN_INFO "Found i82596 at 0x%lx, IRQ %d\n", dev->hpa.start,
- dev->irq);
+ printk(KERN_INFO "Found i82596 at 0x%lx, IRQ %d\n",
+ (unsigned long)dev->hpa.start, dev->irq);
netdevice = alloc_etherdev(sizeof(struct i596_private));
if (!netdevice)