mbox

[GIT,PULL,1/4] ARM: exynos: SoC/Mach for v4.9

Message ID 1472548739-20050-5-git-send-email-k.kozlowski@samsung.com
State New
Headers show

Pull-request

git://git.kernel.org/pub/scm/linux/kernel/git/krzk/linux.git tags/samsung-soc-4.9

Message

Krzysztof Kozlowski Aug. 30, 2016, 9:18 a.m. UTC
The following changes since commit 29b4817d4018df78086157ea3a55c1d9424a7cfc:

  Linux 4.8-rc1 (2016-08-07 18:18:00 -0700)

are available in the git repository at:

  git://git.kernel.org/pub/scm/linux/kernel/git/krzk/linux.git tags/samsung-soc-4.9

for you to fetch changes up to a36289756212a79a1b32c02e66ddf11ba33ec7db:

  ARM: EXYNOS: Remove unused DMC and CMU offsets and their mappings (2016-08-24 06:42:27 +0200)

----------------------------------------------------------------
Samsung mach/soc update for v4.9:
1. Fix for DMA on S3C24xx. This was probably broken for long time, nobody runs
   this code... till now.
2. After fixes from Matthew Leach and Ben Dooks, most of our mach code
   and drivers is now endian-safe. Mark the platform as supporting big endian.
3. Cleanups.

----------------------------------------------------------------
Ben Dooks (1):
      ARM: EXYNOS: Enable ARCH_SUPPORTS_BIG_ENDIAN explicitly

Markus Elfring (1):
      ARM: s3c64xx: Delete unnecessary assignment for the field "owner"

Pankaj Dubey (1):
      ARM: EXYNOS: Remove unused DMC and CMU offsets and their mappings

Sylwester Nawrocki (2):
      ARM: S3C24XX: Add dma_mask assignments for DMA devices
      ARM: S3C24XX: Add missing DMA device for Mini2440 board

 arch/arm/mach-exynos/Kconfig                 |  1 +
 arch/arm/mach-exynos/exynos.c                | 15 ---------------
 arch/arm/mach-exynos/include/mach/map.h      |  5 -----
 arch/arm/mach-s3c24xx/common.c               | 20 +++++++++++++++-----
 arch/arm/mach-s3c24xx/mach-mini2440.c        |  1 +
 arch/arm/mach-s3c64xx/mach-crag6410-module.c |  3 +--
 arch/arm/plat-samsung/include/plat/map-s5p.h |  4 ----
 7 files changed, 18 insertions(+), 31 deletions(-)

Comments

Arnd Bergmann Sept. 2, 2016, 4:26 p.m. UTC | #1
On Tuesday, August 30, 2016 11:18:59 AM CEST Krzysztof Kozlowski wrote:
> Samsung mach/soc update for v4.9:
> 1. Fix for DMA on S3C24xx. This was probably broken for long time, nobody runs
>    this code... till now.

Good to know. This reminds me that the s3c24xx DMA code is one of those
that need to be converted to dma_slave_map at some point. Maybe if someone
is now running that code, they could give that a try as well?

It shouldn't be hard to convert, let me know if you need more information.

> 2. After fixes from Matthew Leach and Ben Dooks, most of our mach code
>    and drivers is now endian-safe. Mark the platform as supporting big endian.
> 3. Cleanups.
> 
> 

Pulled into next/soc, thanks!

	Arnd
Krzysztof Kozlowski Sept. 2, 2016, 4:32 p.m. UTC | #2
On Fri, Sep 02, 2016 at 06:26:42PM +0200, Arnd Bergmann wrote:
> On Tuesday, August 30, 2016 11:18:59 AM CEST Krzysztof Kozlowski wrote:
> > Samsung mach/soc update for v4.9:
> > 1. Fix for DMA on S3C24xx. This was probably broken for long time, nobody runs
> >    this code... till now.
> 
> Good to know. This reminds me that the s3c24xx DMA code is one of those
> that need to be converted to dma_slave_map at some point. Maybe if someone
> is now running that code, they could give that a try as well?
> 
> It shouldn't be hard to convert, let me know if you need more information.

+Cc Sylwester,

Sylwester recently dig into this platform. I think he managed to play
some audio.

Best regards,
Krzysztof
Sylwester Nawrocki Sept. 5, 2016, 8:06 a.m. UTC | #3
On 09/02/2016 06:32 PM, Krzysztof Kozlowski wrote:
> On Fri, Sep 02, 2016 at 06:26:42PM +0200, Arnd Bergmann wrote:
>> On Tuesday, August 30, 2016 11:18:59 AM CEST Krzysztof Kozlowski wrote:
>>> Samsung mach/soc update for v4.9:
>>> 1. Fix for DMA on S3C24xx. This was probably broken for long time, nobody runs
>>>    this code... till now.
>>
>> Good to know. This reminds me that the s3c24xx DMA code is one of those
>> that need to be converted to dma_slave_map at some point. Maybe if someone
>> is now running that code, they could give that a try as well?
>>
>> It shouldn't be hard to convert, let me know if you need more information.
> 
> +Cc Sylwester,
> 
> Sylwester recently dig into this platform. I think he managed to play
> some audio.

I revived this platform for testing ASoC changes.  I should find time
to do that dma_slave_map refactoring sooner or later.

Or if you would like to do it Sam please go ahead.
Sam Van Den Berge Sept. 8, 2016, 7:24 p.m. UTC | #4
On Mon, Sep 05, 2016 at 10:06:14AM +0200, Sylwester Nawrocki wrote:
> On 09/02/2016 06:32 PM, Krzysztof Kozlowski wrote:
> > On Fri, Sep 02, 2016 at 06:26:42PM +0200, Arnd Bergmann wrote:
> >> On Tuesday, August 30, 2016 11:18:59 AM CEST Krzysztof Kozlowski wrote:
> >>> Samsung mach/soc update for v4.9:
> >>> 1. Fix for DMA on S3C24xx. This was probably broken for long time, nobody runs
> >>>    this code... till now.
> >>
> >> Good to know. This reminds me that the s3c24xx DMA code is one of those
> >> that need to be converted to dma_slave_map at some point. Maybe if someone
> >> is now running that code, they could give that a try as well?
> >>
> >> It shouldn't be hard to convert, let me know if you need more information.
> > 
> > +Cc Sylwester,
> > 
> > Sylwester recently dig into this platform. I think he managed to play
> > some audio.
> 
> I revived this platform for testing ASoC changes.  I should find time
> to do that dma_slave_map refactoring sooner or later.
> 
> Or if you would like to do it Sam please go ahead.

Thanks for the suggestion and apologies for the late response. Last couple of 
days I've been figuring out what this would involve and I decided that I want
to give this a try. I found this patch series [1] that I'll use as an example.
If I understand it correctly, I'll need to create a dma_slave_map and
pass that via the platform data into the s3c24xx-dma driver. There I'll
need to fill in the filter_map of the dma slave device. Right?

[1] http://lists.infradead.org/pipermail/linux-arm-kernel/2015-December/393635.html

Kr,
Sam
Sylwester Nawrocki Sept. 12, 2016, 9:38 a.m. UTC | #5
On 09/08/2016 09:24 PM, Sam Van Den Berge wrote:
> Thanks for the suggestion and apologies for the late response. Last couple of 
> days I've been figuring out what this would involve and I decided that I want
> to give this a try. I found this patch series [1] that I'll use as an example.
> If I understand it correctly, I'll need to create a dma_slave_map and
> pass that via the platform data into the s3c24xx-dma driver. There I'll
> need to fill in the filter_map of the dma slave device. Right?

Yes, AFAIU that's what needs to be done.  Subsequently, related DMA
clients could be updated to not use dma_request_slave_channel_compat().