Re: How to map high memory for block io

[Pierre Ossman]

Russell King wrote:
> On Thu, Mar 02, 2006 at 10:52:04AM +0100, Pierre Ossman wrote:
>   
>> Russell King wrote:
>>     
>>> I think you're asking Jens that question - I know of no way to tell
>>> the block layer that clustering is fine for normal but not highmem.
>>>       
>> That wasn't what I meant. What I was referring to was disabling highmem
>> altogether, the way that is done now through looking at the dma mask.
>>     
>
> You need to set your struct device's dma_mask appropriately:
>
>         u64 limit = BLK_BOUNCE_HIGH;
>
>         if (host->dev->dma_mask && *host->dev->dma_mask)
>                 limit = *host->dev->dma_mask;
>
>         blk_queue_bounce_limit(mq->queue, limit);
>
> Hence, if dma_mask is a NULL pointer or zero, highmem will be bounced.
>   

This I know. My beef is the readability of:

if (do_dma)
    *dev->dma_mask = DEVICE_DMA_MASK;
else
    *dev->dma_mask = 0;

I.e. we use dma_mask even though we don't do DMA.

> Neither PNP nor your platform device sets dma_mask, so highmem will
> always be bounced in the case of wbsd - which from what you write above
> is what you require anyway.
>
>   

wbsd isn't the issue, sdhci is. PCI sets a 32-bit mask by default so I
end up with highmem pages even when I'm not doing DMA.

Rgds
Pierre

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/