Re: holes for ISA boards...
david parsons (o.r.c@p.e.l.l.p.o.r.t.l.a.n.d.o.r.u.s)
14 Jun 1999 11:55:03 -0700
In article <linux.kernel.19990613041154.32471.qmail@hog.ctrl-c.liu.se>,
Christer Weinigel <wingel@hog.ctrl-c.liu.se> wrote:
>Mark Jefferys <mjeffery@europa.com> wrote:
>
>>It was renamed to ioremap and moved to arch/<arch>/mm/ioremap.c .
>>
>>But this part of the patch looks bad to me anyway; it stops anyone from
>>mapping part of the hole to VM, but presumably anything that does this
>>knows what it's doing (probably a driver mapping memory straight from an
>>ISA card). Or am I missing something?
>
>The test is the other way around from what you expect. Normally
>vremap will return NULL for adresses below the top of memory (since it
>probably would be a mistake to allow physical RAM to be mapped by a
>driver), my patch excludes the memory hole from that test.
>
> if (MAP_NR(offset) < MAP_NR(high_memory)
> && (offset < memory_hole_start || offset+size > memory_hole_end))
> return NULL;
So in the general case you'd want to have a flag in the mmap that
says that memory foo can be used for a driver?
____
david parsons \bi/ Who's also got a patch for fancy e820 handling, but
\/ who doesn't have any hardware that might actually
take advantage of memory hole handling.
I suppose I should really port it to 2.2,
so people can test it with modern kernels.
-
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/