Messages in this thread | | | Date | Fri, 29 Oct 2004 20:38:27 +0100 | From | Al Viro <> | Subject | Re: net: generic netdev_ioaddr |
| |
On Fri, Oct 29, 2004 at 09:18:18PM +0200, Krzysztof Halasa wrote: > "Pekka J Enberg" <penberg@cs.helsinki.fi> writes: > > > Yup, I thought about that after I sent the patch. However, as it > > stands now, many network drivers use netdev->base_addr for just that. > > Perhaps it should be nuked completely instead? > > I thinks so. With ifmap, SIOCSIFMAP, ifr_map, mem_end etc., > irq, if_port, dma.
SIOCSIFMAP is unfortunate, but legitimate - it passes more or less opaque structure to driver and lets driver interpret it.
SIOCGIFMAP, OTOH, is really bad - among other things, for many drivers it leaks ioremapped addresses to userland. And *that* is a LARTable offense - it's an information that makes no sense whatsoever for userland code and should never be exposed, just as with any kernel pointers.
What uses ->base_addr from the data returned by SIOCGIFMAP? - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
| |