Re: A DMA patch for linux 2.2.9

H.J. Lu (hjl@varesearch.com)
Thu, 8 Jul 1999 14:51:40 -0700 (PDT)

> "H.J."
> I also tested Jaroslav's patch and made the following comments:
> - the hard limit needs at least a kernel load time
> command line option (absolute or % of available
> RAM were suggested by Dan Hollis)
> - on a GFP_DMA fetch if the special DMA pool runs dry
> then the main pool is not checked for candidates
> - what about architectures that don't have this 16 MB
> level (or something akin) limit??
>
> Your patch addresses the first 2 points. Does it address
> the 3rd?

No. We don't have any hardwares like that here. However, it is trivial
to add a global variable, say

int dma_memory_limit = 1;

to indicate if the architecture has the DMA memory limit and each
architecture can turn it off at the runt-time. We then can do

if (!dma_memory_limit)
RMQUEUE_ANY(order);
else {
if (gfp_mask & __GFP_DMA)
RMQUEUE_DMA(order);
else {
RMQUEUE_NODMA(order);
RMQUEUE_ANY(order);
}
}

>
> Doug Gilbert
>
> PS I don't think page_alloc.c changed between 2.2.9 and
> 2.2.10-ac8. Haven't looked at "ac9" yet. So your patch
> should work on more than just 2.2.9 .
>

I also use it on 2.2.7.

-- 
H.J. Lu (hjl@gnu.org)


-
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/