Re: Proposal for new generic device API: dma_get_required_mask()

[James Bottomley <James.Bottomley@steeleye.com>]

On Thu, 2004-06-17 at 19:46, Krzysztof Halasa wrote:
> #define OUR_COST_32 = 4
> #define OUR_COST_39 = 8
> #define OUR_COST_64 = 10
> 
> int cost32 = check_dma_mask(32 bits);
> int cost39 = check_dma_mask(39 bits);
> int cost64 = check_dma_mask(64 bits);
> 
> if (!cost32  && !cost39 && !cost64)
> 	printk(KERN_ERR "64 bits aren't enough for RAM addressing?\n")
> else
> 	use_mode_with_minimal_cost(cost32 * OUR_COST_32,
> 				   cost39 * OUR_COST_39,
> 				   cost64 * OUR_COST_64);
> 
> This check_dma_mask() should be renamed + extended to cover different
> RAM access types:
> - coherent vs non-coherent memory
> - preallocated/initialized memory (such as skb->data passed to
>   hard_start_xmit()) vs uninitialized memory (such as returned by
>   kmalloc()).

Well, I did consider a similar API, but not for long.

It falls victim to the 95/5 rule---when you engineer an API, if 95% of
the complexity is dealing with the 5% of special cases, you're over
engineering.

So the original proposal is the remaining 5% that covers 95% of the use
cases (and will do better even on the remaining 5%).

James