Re: [PATCH 1/3 v3] dma: document dma_{un}map_{single|sg}_attrs() interface

[Grant Grundler <grundler@parisc-linux.org>]

On Tue, Mar 11, 2008 at 06:19:54PM -0700, akepner@sgi.com wrote:
> 
> I have a new proposal for the documentation portion of this 
> patchset. The code changes essentially amount to 
> s/DMA_ATTR_SYNC_ON_WRITE/DMA_ATTR_BARRIER/ so thought I'd send 
> just the doc change for comments now.
> 
> The description of the DMA_ATTR_BARRIER is now very short and 
> generic - no particular architecture is even mentioned. I can add 
> a sentence or two near the architecture-specific changes in 
> arch/ia64/sn/pci/pci_dma.c about why the implementation works 
> on ia64/sn, etc.

Acked-by: Grant Grundler <grundler@parisc-linux.org>

One part of me still wants some history in the doc. But I think
it's better to keep it short and let folks look for implementations
and make sure they are sufficiently well explained.

thanks,
grant

> diff --git a/Documentation/DMA-attributes.txt b/Documentation/DMA-attributes.txt
> index e69de29..a4106ec 100644
> --- a/Documentation/DMA-attributes.txt
> +++ b/Documentation/DMA-attributes.txt
> @@ -0,0 +1,24 @@
> +			DMA attributes
> +			==============
> +
> +This document describes the semantics of the DMA attributes that are 
> +defined in linux/dma-attrs.h. 
> +
> +DMA_ATTR_BARRIER
> +----------------
> +
> +DMA_ATTR_BARRIER is a barrier attribute for DMA.  DMA to a memory 
> +region with the DMA_ATTR_BARRIER attribute forces all pending DMA 
> +writes to complete, and thus provides a mechanism to strictly order 
> +DMA from a device across all intervening busses and bridges.  This 
> +barrier is not specific to a particular type of interconnect, it 
> +applies to the system as a whole, and so its implementation must 
> +account for the idiosyncracies of the system all the way from the 
> +DMA device to memory. 
> +
> +As an example of a situation where DMA_ATTR_BARRIER would be useful, 
> +suppose that a device does a DMA write to indicate that data is ready 
> +and available in memory.  The DMA of the "completion indication" could
> +race with data DMA.  Mapping the memory used for completion indications 
> +with DMA_ATTR_BARRIER would prevent the race.
> +