[PATCH 0/2] pxa3xx: Data buffer dynamic allocation

[Ezequiel Garcia]

On Wed, Oct 09, 2013 at 02:00:28PM +0200, Daniel Mack wrote:
> On 09.10.2013 13:48, Ezequiel Garcia wrote:
> > On Fri, Oct 04, 2013 at 03:30:36PM -0300, Ezequiel Garcia wrote:
> >> As part of my work to support the NAND controllers in Armada 370/XP SoC
> >> here's a tiny set of patches to replace the currently data buffer allocation
> >> based in a hardcoded buffer size, into a dynamic scheme.
> >>
> >> We first use kmalloc to request a 256 bytes (big enough for the ONFI paramater
> >> page) and then re-allocate (either with kmalloc or using DMA allocators) using
> >> the detected page size.
> >>
> >> The first patch is required to perform the initial NAND operations (READID
> >> and friends) using programmable I/O since the DMA buffers are not ready
> >> at such early stage.
> >>
> >> The second patch performs to actual dynamic allocation and it's exactly
> >> the same patch I sent a few weeks ago:
> >>
> >>   http://patchwork.ozlabs.org/patch/275982/
> >>
> >> Daniel: Can you test these two If they look fine we can add them now,
> >> being an improvement to the driver pxa3xx beyond the Armada 370/XP effort.
> > 
> > Daniel: sorry to bother, any luck testing this?
> 
> Sorry for the delay, got other things to do lately ...
> 

No problem, thanks for testing.

> I tested these two patches successfully on my board with both use_dma=0
> and =1. Nice :) You can take my
> 
> 	Tested-by: Daniel Mack <zonque at gmail.com>
> 

Good news! Brian: all yours :)

> I hope I can catch up with my pxa DMA cleanups very soon.

Indeed. BTW: I had some feedback on why the Armada 370/XP SoC can't
do DMA on the NAND, I'll reply to the older mail with the information.
-- 
Ezequiel García, Free Electrons
Embedded Linux, Kernel and Android Engineering
http://free-electrons.com