Am Montag, 2. Juli 2007 schrieb David Brownell:
> On Monday 02 July 2007, Oliver Neukum wrote:
> > Am Montag, 2. Juli 2007 schrieb David Brownell:
> > > + Most drivers should *NOT* be using these primitives. On most systems
> > > + the memory returned will be uncached, so it's a bit more expensive to
> > > + access than what kmalloc() returns.
> >
> > This is somewhat hard to understand.
>
> Should that be two separate paragraphs then? "don't", then "this is
> probably going to be uncached." ?
Yes. In addition it needs to be more explicit.
And perhaps a bit about how to treat uncached memory.
> > If we use this memory as
> > a real working space it makes sense. But if we do a memcpy() into
> > the buffer, why do we care? The data has to hit RAM for DMA to work.
> > So why not do the work at once, at least in the write path?
> >
> > We are as an advantage sure to get memory we can do DMA to.
>
> The reason not to use those operations is that most drivers have
> no need for dma-coherent memory. The normal stuff does just fine.
Then you need to write exactly that. It is not stated as such.
And please explain why we can operate efficiently with kmalloced
memory although we don't evaluate the controller's dma mask.
Regards
Oliver
-------------------------------------------------------------------------
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and take
control of your XML. No limits. Just data. Click to get it now.
http://sourceforge.net/powerbar/db2/
_______________________________________________
linux-usb-devel@lists.sourceforge.net
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
