Re: [PATCH v2 0/8] R-Car Gen2 DMA Controller driver

[Laurent Pinchart <laurent.pinchart@ideasonboard.com>]

Hi Simon,

On Wednesday 23 July 2014 19:34:28 Simon Horman wrote:
> On Wed, Jul 23, 2014 at 10:47:11AM +0200, Laurent Pinchart wrote:
> > On Wednesday 23 July 2014 09:35:33 Simon Horman wrote:
> >> On Sat, Jul 19, 2014 at 01:50:17AM +0200, Laurent Pinchart wrote:
> >>> Hello,
> >>> 
> >>> This patch sets adds a new driver for the DMA controller found in the
> >>> R-Car Gen2 SoCs under the name "Direct Memory Access Controller for
> >>> System (SYS-DMAC)". Support for the "Realtime Direct Memory Access
> >>> Controller (RT-DMAC)" and "Direct Memory Access Controller for Audio
> >>> (Audio-DMAC)" will be added later.
> >>> 
> >>> For the rationale of why a new driver is needed, and performance
> >>> figures, please see the cover letter of v1 ("[PATCH 0/7] R-Car Gen2
> >>> DMA Controller driver") [1].
> >>> 
> >>> Support for hardware descriptors lists is under development and will
> >>> be posted in the near future. No change to the DT bindings will be
> >>> needed.
> >>> 
> >>> The first five patches should go through the DMA engine tree, while
> >>> the last three patches should go through the Renesas tree. If the
> >>> driver can make it to v3.17 merging the platform changes will be
> >>> easier.
> >>> 
> >>> Morimoto-san, could you please test if this version fixes the channel
> >>> filter issue you've experienced with the R-Car audio driver ? I would
> >>> also appreciate if you could tell me how to test cyclic DMA transfers
> >>> using the audio driver.
> >>> 
> >>> Changes since v1:
> >>> 
> >>> - Allocate IRQ name strings dynamically
> >>> - Only call the callback function if one is supplied
> >>> - Don't overallocate sg list entries
> >>> - Allocate sg list entries with GFP_KERNEL
> >>> - Don't manage function clock manually
> >>> - Make channel filter ignore unrelated devices
> >>> - Document why the cyclic sg list is kcalloc'ed
> >>> - Remove ch15 from interrupt names in DT
> >>> - Replace CONFIG_OF with OF in Kconfig
> >>> 
> >>> [1] http://www.spinics.net/lists/linux-sh/msg33768.html
> >>> 
> >>> Laurent Pinchart (8):
> >>>   dmaengine: sh: Rework Kconfig and Makefile
> >>>   dmaengine: shdma: Make channel filter ignore unrelated devices
> >>>   dmaengine: shdma: Allocate cyclic sg list dynamically
> >>>   dmaengine: rcar-dmac: Add device tree bindings documentation
> >>>   dmaengine: rcar-dmac: Add Renesas R-Car Gen2 DMA Controller (DMAC)
> >>>     driver
> >> 
> >> I have queued up the above three patches in the shdma-for-v3.17 branch
> >> of the renesas tree.
> > 
> > I suppose you meant five, not three.
> 
> There were three, then I decided to take the next two too.
> 
> > > I intend to send a pull-request for them once they have sat in next.
> > 
> > I was waiting for acks before sending the pull request, but that's fine.
> > Who will you send the pull request to, the dmaengine maintainer ?
> 
> Yes, I was planning to send a pull-request to Vinod.
> I have already sent two recently, one of which was even accepted.
> 
> Of course If you want to handle things then I'm happy to discuss that.
> My main aim is too try to streamline things for Vinod. And if I'm not
> achieving that then we need to talk :)

Your help is appreciated. I'm totally fine with you sending a pull request. 
The only issue I can think of with this unsynchronized process would be pull 
requests being sent too soon when the driver isn't ready yet. In this specific 
case I consider the driver as ready (even if it's obviously not perfect, more 
patches will come soon), so we can proceed.

> >>>   ARM: shmobile: r8a7790: Add DMAC clocks to DT
> >>>   ARM: shmobile: r8a7790: Add DMAC devices to DT
> >>>   ARM: shmobile: r8a7791: Add DMAC devices to DT
> >> 
> >> I am holding off on the above patches as there is still some chance
> >> the dmaengine subsystem maintainer may reject the new binding introduced
> >> by the 4th patch of this series.
> >> 
> >> Could you repost them once he has accepted my as-yet-unsent
> >> pull-request.
> > 
> > Sure. You could already apply "ARM: shmobile: r8a7790: Add DMAC clocks to
> > DT" as that's independent from the DMAC DT bindings, but that's up to
> > you.
>
> Yes, I realise that. But its also not very exciting without the rest of the
> series, so I think that I prefer to wait. If that creates a headache for
> you then I'm happy to reconsider.

That's fine with me.
 
> >>>  .../devicetree/bindings/dma/renesas,rcar-dmac.txt  |   98 ++
> >>>  arch/arm/boot/dts/r8a7790.dtsi                     |   66 +-
> >>>  arch/arm/boot/dts/r8a7791.dtsi                     |   60 +
> >>>  drivers/dma/sh/Kconfig                             |   30 +-
> >>>  drivers/dma/sh/Makefile                            |   15 +-
> >>>  drivers/dma/sh/rcar-dmac.c                         | 1525 +++++++++++++
> >>>  drivers/dma/sh/shdma-base.c                        |  101 +-
> >>>  7 files changed, 1845 insertions(+), 50 deletions(-)
> >>>  create mode 100644
> >>>  Documentation/devicetree/bindings/dma/renesas,rcar-dmac.txt
> >>>  create mode 100644 drivers/dma/sh/rcar-dmac.c

-- 
Regards,

Laurent Pinchart