Re: [RFC] support for bidirectional transfers and variable length CDBs

[Matthew Wilcox <matthew@wil.cx>]

On Sat, Oct 28, 2006 at 05:36:09PM +0100, Christoph Hellwig wrote:
> This should be an entirely separate patch.  (As should your updates to
> the list of scsi device types and the typo fix in a comment in your
> patch)

Actually, the update to the list of device types is already in Linus' tree
http://git.kernel.org/git/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=4ff36718ede26ee2da73f2dae94d71e2b06845fc

> I haven't looked at this code in detail but it looks rather
> messy already ;-)  I'd prefer if we could postponed implementing this
> properly until we've landed bidi support.

I think we should encapsulate:

-	int write = (data_direction == DMA_TO_DEVICE);
+	int write = ((data_direction == DMA_TO_DEVICE) || (data_direction == DMA_BIDIRECTIONAL));

into something like:

-	int write = (data_direction == DMA_TO_DEVICE);
+	int write = is_dma_write(data_direction);