On Mon, Jun 10, 2002 at 08:00:11PM +0200, Oliver Neukum wrote:
> Am Montag, 10. Juni 2002 19:46 schrieb Tom Rini:
> > On Mon, Jun 10, 2002 at 07:33:18PM +0200, Oliver Neukum wrote:
> > > this change set against 2.4 introduces a #define to be used by the
> > > driver who are concerned by this issue.
> >
> > Can we wait abit for this please?  Defining it to SMP_CACHE_BYTES is
> > counter-intuitive since these machines don't do SMP.  And this isn't
> > just a USB problem either, USB is just the current good example.
> 
> Hi,
> 
> well I doubt that Marcello will delay 2.4.20 long for a problem only
> a few minor architectures suffer. Yet I have a soft spot for those little
> machines ;-)

heh.. I don't think this should be fixed for 2.4.20 tho either.  I know
the machines Roland is working on aren't supported by kernel.org yet,
so..

> Secondly this is the reason I put a comment on the define. The definite
> name will eventually come from someone else, as it's arch dependent.
> A define to DMA_ALIGNED by David's suggestion seems very intuitive
> to me and a new name means that it can be quickly changed should that
> be needed.

DMA_ALIGNED makes more sense, yes.  So for 2.4.21-pre1 lets do something
like:
#ifndef DMA_ALIGN_BYTES
#define DMA_ALIGN_BYTES SMP_CACHE_BYTES
#endif

#define DMA_ALIGNED __atribute__(... DMA_ALIGN_BYTES ...)

-- 
Tom Rini (TR1265)
http://gate.crashing.org/~trini/

_______________________________________________________________

Don't miss the 2002 Sprint PCS Application Developer's Conference
August 25-28 in Las Vegas - 
http://devcon.sprintpcs.com/adp/index.cfm?source=osdntextlink

_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel

Reply via email to