Re: [PATCH 00/18] ARM: s3c64xx: Let amba-pl08x driver handle DMA

From: Tomasz Figa
Date: Thu Aug 15 2013 - 07:51:04 EST


[Ccing maintainers and other people responsible for spear, lpc32xx and
versatile platforms]

On Sunday 11 of August 2013 19:59:12 Tomasz Figa wrote:
> This is first non-RFC version of my patches extending support of
> amba-pl08x DMA engine driver to PL080S DMA engine (PL080 modified by
> Samsung) found in Samsung S3C64xx SoCs.
> Due to changes scattered across different areas of kernel, patches are
> based on merged 3 branches:
> - for-next of Kgene's Samsung tree,
> - clk-next of Mike's clock tree,
> - next of Vinod's slave DMA tree.
> To ease testing I have prepared a branch in my private tree for anyone
> willing to check the patches out:
> git:// v3.12-pl080
> Dependencies (already applied in my branch):
> - for patches 14 and 16 - CCF-based clock driver for s3c64xx.
> Some of the patches not related to the amba-pl08x driver itself
> can be likely applied into appropriate trees separately, namely:
> - 09/18 - ASoC: Samsung: Do not queue cyclic buffers multiple times,
> - 14/18 - clk: samsung: s3c64xx: Add aliases for DMA clocks.
> After patch 14/18, both old and new DMA drivers can be supported on
> S3C64xx, depending on Kconfig options. Patches 15-18 remove the old
> driver leaving support only for the generic pl08x driver. Feel free to
> drop those patches for now if we want more testing, but I don't suspect
> any problems.
> On S3C64xx-based Mini6410 and SMDK6410 boards, with I2S audio
> playback and capture (including full duplex operation) and also SPI
> using spidev:
> Tested-by: Tomasz Figa <tomasz.figa@xxxxxxxxx>

It would be nice if patches from this series touching the PL08x driver
(01-08) could be tested on other platforms that have this DMA controller
as well, to make sure I did not break anything. Unfortunately I do not
have any board based on any of them.

Best regards,

> Changes since RFC v2:
> - Added clkdev lookups to old clock driver.
> - Added patches removing the old DMA driver and any remaining code
> needed by it.
> - Fixed DMA support for SPI.
> - Added a word about PL080S to amba-pl08x.c file header.
> - Changed definition of LLI words from enums to macros.
> - Extended debugging messages to handle PL080S variant as well.
> - Little cleanup of LLI dumping code.
> - Added check for peripheral flow control, which is unsupported by
> PL080S to dma_set_runtime_config.
> - Corrected transfer size mask of PL080S.
> Changes since RFC v1:
> - Returned to original way of storing quirks as booleans, as suggested
> by Russell, Linus and Arnd.
> - Added reg_config field to pl08x_phy_chan struct, which stores
> variant-specific address of channel config register, as suggested
> by Russell.
> - Simplified handling of extended maximum transfer size of PL080S
> (no more conditional passing of 0 as length to pl08x_cctl_bits()).
> - Reworked LLI handling in the driver to stop casting arbitrary memory
> to a struct and allow different word count of LLI entry, as suggested
> by Linus.
> - Removed AMBA ID override from S3C64xx PL080 initialization code.
> - Fixed brokenness of Samsung DMA wrapper API, which caused cyclic
> buffers to be queued multiple times when DMA engine is used.
> - Included patch adding clock aliases for DMA engines (depends on
> Common Clock Framework driver for S3C64xx).
> - Fixed several minor stylistic issues.
> For reference, here is the original description of the series:
> One of the biggest roadblocks on the way of S3C64xx to DeviceTree
> support is its DMA driver, which is completely platform-specific and
> provides private API (s3c-dma), not even saying that its design is
> completely against multiplatform-awareness.
> The DMA controller present on this SoC series is a custom variant
> of ARM PrimeCell PL080 modified by Samsung to add some extra features.
> It is mostly compatible with original PL080, except:
> - CH_CONTROL2 register is added between CH_CONTROL and CH_CONFIG,
> - offset of CH_CONFIG register is different,
> - transfer size field is moved from CH_CONTROL to CH_CONTROL2,
> - transfer size field is extended to 24 bits, allowing much bigger
> single transfer,
> - LLI consists of one more word, to account for CH_CONTROL2 register.
> Since all the rest is fully compatible with standard PL080 there is no
> point in having separate driver just for this single variant, so I
> decided to look into adding support for it to the amba-pl08x driver.
> There was already some attempt to achieve this before, but this was
> before Russel's big rework of the driver to use virtual channels,
> making the old patches being not much of use.
> This RFC series is a proof of concept that I managed to make during last
> days of hacking. Except one patch adding clkdev lookup to clock driver
> (which is being replaced with a CCF-compliant driver ATM), this is
> enough to get memcpy and slave transfers to work on S3C64xx.
> I have tested this on Mini6410 and SMDK6410 boards using dmatest for
> memcpy and Samsung I2S with madplay/aplay for slave transfers.
> Unfortunately I do not have access to other platforms with PL08x so
> I could not test for any regressions introduced on them.
> Credits for two patches go to Alban Bedel, who made a series fixing this
> driver to make it usable with audio drivers. I rebased his patches on
> top of mine and corrected coding style a bit.
> OK, that's all. Any comments are welcome. Feel free to start throwing
> eggs and tomatoes if you find this awful, but I won't be upset if I get
> some Tested-by or Acked-by as well. ;)
> Alban Bedel (2):
> dmaengine: PL08x: Fix reading the byte count in cctl
> dmaengine: PL08x: Add cyclic transfer support
> Tomasz Figa (16):
> dmaengine: PL08x: Refactor pl08x_getbytes_chan() to lower indentation
> dmaengine: PL08x: Add support for different offset of CONFIG register
> dmaengine: PL08x: Rework LLI handling to be less fragile
> dmaengine: PL08x: Move LLI dumping code into separate function
> dmaengine: PL08x: Add support for PL080S variant
> dmaengine: PL08x: Add support for different maximum transfer size
> ASoC: Samsung: Do not queue cyclic buffers multiple times
> spi: s3c64xx: Do not require legacy DMA API in case of S3C64XX
> ASoC: Samsung: Do not require legacy DMA API in case of S3C64XX
> ARM: s3c64xx: Add support for DMA using generic amba-pl08x driver
> ARM: s3c64xx: clock: Add clkdev lookup for DMA clocks
> clk: samsung: s3c64xx: Add aliases for DMA clocks
> ARM: s3c64xx: Remove legacy DMA driver
> clk: samsung: s3c64xx: Remove clock aliases of old DMA driver
> spi: s3c64xx: Always select S3C64XX_PL080 when ARCH_S3C64XX is enabled
> ASoC: Samsung: Always select S3C64XX_PL080 when ARCH_S3C64XX is enabled
> arch/arm/Kconfig | 1 +
> arch/arm/mach-s3c64xx/Kconfig | 7 +-
> arch/arm/mach-s3c64xx/Makefile | 2 +-
> arch/arm/mach-s3c64xx/clock.c | 28 +-
> arch/arm/mach-s3c64xx/common.h | 5 +
> arch/arm/mach-s3c64xx/dma.c | 753
> -------------------------------
> arch/arm/mach-s3c64xx/include/mach/dma.h | 144 ++----
> arch/arm/mach-s3c64xx/pl080.c | 244 ++++++++++
> arch/arm/plat-samsung/devs.c | 6 +-
> arch/arm/plat-samsung/s3c-dma-ops.c | 13 +-
> drivers/clk/samsung/clk-s3c64xx.c | 4 +-
> drivers/dma/amba-pl08x.c | 501 ++++++++++++++------
> drivers/spi/Kconfig | 2 +-
> include/linux/amba/pl080.h | 1 +
> sound/soc/samsung/Kconfig | 2 +-
> sound/soc/samsung/dma.c | 7 +
> 16 files changed, 705 insertions(+), 1015 deletions(-)
> delete mode 100644 arch/arm/mach-s3c64xx/dma.c
> create mode 100644 arch/arm/mach-s3c64xx/pl080.c

Attachment: signature.asc
Description: This is a digitally signed message part.