Messages in this thread | | | Date | Fri, 6 Aug 1999 20:55:25 -0700 (PDT) | From | Linus Torvalds <> | Subject | Re: New resources - pls, explain :-( |
| |
On Sat, 7 Aug 1999, Martin Mares wrote: > > On all sane machines, all regions are automatically enabled by the BIOS.
Not necessarily true at all.
There are actually a number of BIOSes that have an option to enable _nothing_ but the actual devices needed for booting.
Now, I admit to not having tried that option, because it definitely doesn't work with the current state of Linux drivers.
> I think that requiring drivers to take care of the dirty work is > wrong. Also, the address space is large enough to hold all regions > (if it weren't, the BIOS setup would fail).
I'm not saying the driver should do the dirty work.
I'm saying the _helper_ function should do the dirty work, and I'm saying that the bootup sequence should _not_ necessarily do it.
Note that on non-PC machines this tends to make more sense. It also makes a whole lot more sense if you start thinking about hot-plug etc. You should let the driver do a simple "pci_enable_mem()", and that's the thing that enables the memory mapping.
The same way you should have the PCMCIA drivers do a "pcmcia_enable_mem()", for example.
Think of it as a part of a larger pattern. In general, the driver may be the only entity that knows which memory regions it needs enabled. It's not just about PCI, it's about the concept.
Linus
- 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/
| |