lkml.org 
[lkml]   [1998]   [May]   [15]   [last100]   RSS Feed
Views: [more markup]  [less markup]  [headers]  [forward] 
 
Messages in this thread
/
SubjectRe: DMA from/to user-space memory
DateFri, 15 May 1998 19:20:48 -0600
FromStephen Williams <>
davem@dm.cobaltmicro.com said:
> On the 64-bit architectures I know of, and PCI as specified, is not
> limited to 64-bits of physical addressing. 

I think you meant 32bits? PCI *is* limited to 64bits, but not 32:-)

Well, the 32bit PCI supports 64bit addresses with the dual address cycle,
but that adds a clock cycle to the PCI transaction, which would best
be avoided. Longer bursts can amortize that cost away pretty quickly,
but it is there all the same.

Unfortunately, the GFP_DMA bit is getting to be overloaded. I'm going to
make the ugly suggestion that there be added a variant of get_free_pages
that can be compelled to allocate pages from specific regions. One can
make that extra parameter the address size (in bits) and this should work
well with a buddy-system allocator.

I can hope 64bit PCI will come of age in time to save us (I know that
my alpha has a couple 64bit PCI slots in it) but I'm afraid I know better:-)

-- 
Steve Williams                "The woods are lovely, dark and deep.
steve@icarus.com              But I have promises to keep,
steve@picturel.com            and lines to code before I sleep,
http://www.picturel.com       And lines to code before I sleep."



-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu

\
 
 \ /
  Last update: 2005-03-22 12:42    [from the cache]
©2003-2008