[PATCH 0/3] riscv: sophgo: allow DMA multiplexer set channel number for DMA controller

[PATCH 0/3] riscv: sophgo: allow DMA multiplexer set channel number for DMA controller

[Inochi Amaoto]

As the DMA controller on Sophgo CV1800 series SoC only has 8 channels,
the SoC provides a dma multiplexer to reuse the DMA channel. However,
the dma multiplexer also controlls the DMA interrupt multiplexer, which
means that the dma multiplexer needs to know the channel number.

Inochi Amaoto (3):
  dt-bindings: dma: snps,dw-axi-dmac: Add CV1800B compatible
  dmaengine: dw-axi-dmac: Add support for CV1800B DMA
  riscv: dts: sophgo: cv180x: Allow the DMA multiplexer to set channel
    number for DMA controller

 .../bindings/dma/snps,dw-axi-dmac.yaml        |  1 +
 arch/riscv/boot/dts/sophgo/cv180x.dtsi        |  2 +-
 .../dma/dw-axi-dmac/dw-axi-dmac-platform.c    | 23 ++++++++++++++++---
 drivers/dma/dw-axi-dmac/dw-axi-dmac.h         |  1 +
 4 files changed, 23 insertions(+), 4 deletions(-)

--
2.52.0

[PATCH 1/3] dt-bindings: dma: snps,dw-axi-dmac: Add CV1800B compatible

[Inochi Amaoto]

The DMA controller on CV1800B needs to use the DMA phandle args
as the channel number instead of hardware handshake number, so
add a new compatible for the DMA controller on CV1800B.

Change the DMA phandle args parsing logic so it can use handshake
number as channel number if necessary

Signed-off-by: Inochi Amaoto <inochiama@gmail.com>
---
 Documentation/devicetree/bindings/dma/snps,dw-axi-dmac.yaml | 1 +
 1 file changed, 1 insertion(+)

diff --git a/Documentation/devicetree/bindings/dma/snps,dw-axi-dmac.yaml b/Documentation/devicetree/bindings/dma/snps,dw-axi-dmac.yaml
index a393a33c8908..0b5c8314e25e 100644
--- a/Documentation/devicetree/bindings/dma/snps,dw-axi-dmac.yaml
+++ b/Documentation/devicetree/bindings/dma/snps,dw-axi-dmac.yaml
@@ -20,6 +20,7 @@ properties:
     enum:
       - snps,axi-dma-1.01a
       - intel,kmb-axi-dma
+      - sophgo,cv1800b-axi-dma
       - starfive,jh7110-axi-dma
       - starfive,jh8100-axi-dma

--
2.52.0

Re: [PATCH 1/3] dt-bindings: dma: snps,dw-axi-dmac: Add CV1800B compatible

[Anton Stavinsky]

> The DMA controller on CV1800B needs to use the DMA phandle args
> as the channel number instead of hardware handshake number, so
> add a new compatible for the DMA controller on CV1800B.

Thanks a lot,  Inochi. I've tested on my Milk Duo 256M board. 
Seems to be working with the I2S driver, which I'm working on right now.
No issues with DMA interrupts anymore, DMA router used right channel. 

Re: [PATCH 1/3] dt-bindings: dma: snps,dw-axi-dmac: Add CV1800B compatible

[Chen Wang]

On 12/13/2025 2:48 AM, Anton Stavinsky wrote:
>> The DMA controller on CV1800B needs to use the DMA phandle args
>> as the channel number instead of hardware handshake number, so
>> add a new compatible for the DMA controller on CV1800B.
> Thanks a lot,  Inochi. I've tested on my Milk Duo 256M board.
> Seems to be working with the I2S driver, which I'm working on right now.
> No issues with DMA interrupts anymore, DMA router used right channel.

Thank you, Anton. Could you also add a "Tested-by" signature?

Chen.

Re: [PATCH 1/3] dt-bindings: dma: snps,dw-axi-dmac: Add CV1800B compatible

[Anton D. Stavinskii]

On Sat, Dec 13, 2025 at 08:53:34AM +0400, Chen Wang wrote:
> 
> On 12/13/2025 2:48 AM, Anton Stavinsky wrote:
> > > The DMA controller on CV1800B needs to use the DMA phandle args
> > > as the channel number instead of hardware handshake number, so
> > > add a new compatible for the DMA controller on CV1800B.
> > Thanks a lot,  Inochi. I've tested on my Milk Duo 256M board.
> > Seems to be working with the I2S driver, which I'm working on right now.
> > No issues with DMA interrupts anymore, DMA router used right channel.
> 
> Thank you, Anton. Could you also add a "Tested-by" signature?
I've tested the v2 https://lore.kernel.org/all/aUF4w9sO5lmU9T6v@anton.local/ Hope this will help. 
Thanks for pointing me. 
> 
> Chen.
> 
> 

[PATCH 2/3] dmaengine: dw-axi-dmac: Add support for CV1800B DMA

[Inochi Amaoto]

As the DMA controller on Sophgo CV1800 series SoC only has 8 channels,
the SoC provides a dma multiplexer to reuse the DMA channel. However,
the dma multiplexer also controlls the DMA interrupt multiplexer, which
means that the dma multiplexer needs to know the channel number.

Allow the driver to use DMA phandle args as the channel number, so the
DMA multiplexer can route the DMA interrupt correctly.

Signed-off-by: Inochi Amaoto <inochiama@gmail.com>
---
 .../dma/dw-axi-dmac/dw-axi-dmac-platform.c    | 23 ++++++++++++++++---
 drivers/dma/dw-axi-dmac/dw-axi-dmac.h         |  1 +
 2 files changed, 21 insertions(+), 3 deletions(-)

diff --git a/drivers/dma/dw-axi-dmac/dw-axi-dmac-platform.c b/drivers/dma/dw-axi-dmac/dw-axi-dmac-platform.c
index b23536645ff7..62bf0d0dc354 100644
--- a/drivers/dma/dw-axi-dmac/dw-axi-dmac-platform.c
+++ b/drivers/dma/dw-axi-dmac/dw-axi-dmac-platform.c
@@ -50,6 +50,7 @@
 #define AXI_DMA_FLAG_HAS_APB_REGS	BIT(0)
 #define AXI_DMA_FLAG_HAS_RESETS		BIT(1)
 #define AXI_DMA_FLAG_USE_CFG2		BIT(2)
+#define AXI_DMA_FLAG_HANDSHAKE_AS_CHAN	BIT(3)

 static inline void
 axi_dma_iowrite32(struct axi_dma_chip *chip, u32 reg, u32 val)
@@ -1361,15 +1362,26 @@ static struct dma_chan *dw_axi_dma_of_xlate(struct of_phandle_args *dma_spec,
 					    struct of_dma *ofdma)
 {
 	struct dw_axi_dma *dw = ofdma->of_dma_data;
+	unsigned int handshake = dma_spec->args[0];
 	struct axi_dma_chan *chan;
 	struct dma_chan *dchan;

-	dchan = dma_get_any_slave_channel(&dw->dma);
+	if (dw->hdata->use_handshake_as_channel_number) {
+		if (handshake >= dw->hdata->nr_channels)
+			return NULL;
+
+		chan = &dw->chan[handshake];
+		dchan = dma_get_slave_channel(&chan->vc.chan);
+	} else {
+		dchan = dma_get_any_slave_channel(&dw->dma);
+	}
+
 	if (!dchan)
 		return NULL;

-	chan = dchan_to_axi_dma_chan(dchan);
-	chan->hw_handshake_num = dma_spec->args[0];
+	if (!chan)
+		chan = dchan_to_axi_dma_chan(dchan);
+	chan->hw_handshake_num = handshake;
 	return dchan;
 }

@@ -1508,6 +1520,8 @@ static int dw_probe(struct platform_device *pdev)
 			return ret;
 	}

+	chip->dw->hdata->use_handshake_as_channel_number = !!(flags & AXI_DMA_FLAG_HANDSHAKE_AS_CHAN);
+
 	chip->dw->hdata->use_cfg2 = !!(flags & AXI_DMA_FLAG_USE_CFG2);

 	chip->core_clk = devm_clk_get(chip->dev, "core-clk");
@@ -1663,6 +1677,9 @@ static const struct of_device_id dw_dma_of_id_table[] = {
 	}, {
 		.compatible = "intel,kmb-axi-dma",
 		.data = (void *)AXI_DMA_FLAG_HAS_APB_REGS,
+	}, {
+		.compatible = "sophgo,cv1800b-axi-dma",
+		.data = (void *)AXI_DMA_FLAG_HANDSHAKE_AS_CHAN,
 	}, {
 		.compatible = "starfive,jh7110-axi-dma",
 		.data = (void *)(AXI_DMA_FLAG_HAS_RESETS | AXI_DMA_FLAG_USE_CFG2),
diff --git a/drivers/dma/dw-axi-dmac/dw-axi-dmac.h b/drivers/dma/dw-axi-dmac/dw-axi-dmac.h
index b842e6a8d90d..67cc199e24d1 100644
--- a/drivers/dma/dw-axi-dmac/dw-axi-dmac.h
+++ b/drivers/dma/dw-axi-dmac/dw-axi-dmac.h
@@ -34,6 +34,7 @@ struct dw_axi_dma_hcfg {
 	bool	reg_map_8_channels;
 	bool	restrict_axi_burst_len;
 	bool	use_cfg2;
+	bool	use_handshake_as_channel_number;
 };

 struct axi_dma_chan {
--
2.52.0

Re: [PATCH 2/3] dmaengine: dw-axi-dmac: Add support for CV1800B DMA

[Samuel Holland]

Hi Inochi,

On 2025-12-12 11:05 AM, Inochi Amaoto wrote:
> As the DMA controller on Sophgo CV1800 series SoC only has 8 channels,
> the SoC provides a dma multiplexer to reuse the DMA channel. However,
> the dma multiplexer also controlls the DMA interrupt multiplexer, which

typo: controls

> means that the dma multiplexer needs to know the channel number.
> 
> Allow the driver to use DMA phandle args as the channel number, so the
> DMA multiplexer can route the DMA interrupt correctly.
> 
> Signed-off-by: Inochi Amaoto <inochiama@gmail.com>
> ---
>  .../dma/dw-axi-dmac/dw-axi-dmac-platform.c    | 23 ++++++++++++++++---
>  drivers/dma/dw-axi-dmac/dw-axi-dmac.h         |  1 +
>  2 files changed, 21 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/dma/dw-axi-dmac/dw-axi-dmac-platform.c b/drivers/dma/dw-axi-dmac/dw-axi-dmac-platform.c
> index b23536645ff7..62bf0d0dc354 100644
> --- a/drivers/dma/dw-axi-dmac/dw-axi-dmac-platform.c
> +++ b/drivers/dma/dw-axi-dmac/dw-axi-dmac-platform.c
> @@ -50,6 +50,7 @@
>  #define AXI_DMA_FLAG_HAS_APB_REGS	BIT(0)
>  #define AXI_DMA_FLAG_HAS_RESETS		BIT(1)
>  #define AXI_DMA_FLAG_USE_CFG2		BIT(2)
> +#define AXI_DMA_FLAG_HANDSHAKE_AS_CHAN	BIT(3)
> 
>  static inline void
>  axi_dma_iowrite32(struct axi_dma_chip *chip, u32 reg, u32 val)
> @@ -1361,15 +1362,26 @@ static struct dma_chan *dw_axi_dma_of_xlate(struct of_phandle_args *dma_spec,
>  					    struct of_dma *ofdma)
>  {
>  	struct dw_axi_dma *dw = ofdma->of_dma_data;
> +	unsigned int handshake = dma_spec->args[0];
>  	struct axi_dma_chan *chan;
>  	struct dma_chan *dchan;
> 
> -	dchan = dma_get_any_slave_channel(&dw->dma);
> +	if (dw->hdata->use_handshake_as_channel_number) {
> +		if (handshake >= dw->hdata->nr_channels)
> +			return NULL;
> +
> +		chan = &dw->chan[handshake];
> +		dchan = dma_get_slave_channel(&chan->vc.chan);
> +	} else {
> +		dchan = dma_get_any_slave_channel(&dw->dma);
> +	}
> +
>  	if (!dchan)
>  		return NULL;
> 
> -	chan = dchan_to_axi_dma_chan(dchan);
> -	chan->hw_handshake_num = dma_spec->args[0];
> +	if (!chan)

When use_handshake_as_channel_number is false, chan is uninitialized here.

Regards,
Samuel

> +		chan = dchan_to_axi_dma_chan(dchan);
> +	chan->hw_handshake_num = handshake;
>  	return dchan;
>  }
> 
> @@ -1508,6 +1520,8 @@ static int dw_probe(struct platform_device *pdev)
>  			return ret;
>  	}
> 
> +	chip->dw->hdata->use_handshake_as_channel_number = !!(flags & AXI_DMA_FLAG_HANDSHAKE_AS_CHAN);
> +
>  	chip->dw->hdata->use_cfg2 = !!(flags & AXI_DMA_FLAG_USE_CFG2);
> 
>  	chip->core_clk = devm_clk_get(chip->dev, "core-clk");
> @@ -1663,6 +1677,9 @@ static const struct of_device_id dw_dma_of_id_table[] = {
>  	}, {
>  		.compatible = "intel,kmb-axi-dma",
>  		.data = (void *)AXI_DMA_FLAG_HAS_APB_REGS,
> +	}, {
> +		.compatible = "sophgo,cv1800b-axi-dma",
> +		.data = (void *)AXI_DMA_FLAG_HANDSHAKE_AS_CHAN,
>  	}, {
>  		.compatible = "starfive,jh7110-axi-dma",
>  		.data = (void *)(AXI_DMA_FLAG_HAS_RESETS | AXI_DMA_FLAG_USE_CFG2),
> diff --git a/drivers/dma/dw-axi-dmac/dw-axi-dmac.h b/drivers/dma/dw-axi-dmac/dw-axi-dmac.h
> index b842e6a8d90d..67cc199e24d1 100644
> --- a/drivers/dma/dw-axi-dmac/dw-axi-dmac.h
> +++ b/drivers/dma/dw-axi-dmac/dw-axi-dmac.h
> @@ -34,6 +34,7 @@ struct dw_axi_dma_hcfg {
>  	bool	reg_map_8_channels;
>  	bool	restrict_axi_burst_len;
>  	bool	use_cfg2;
> +	bool	use_handshake_as_channel_number;
>  };
> 
>  struct axi_dma_chan {
> --
> 2.52.0
> 
> 
> _______________________________________________
> linux-riscv mailing list
> linux-riscv@lists.infradead.org
> http://lists.infradead.org/mailman/listinfo/linux-riscv

Re: [PATCH 2/3] dmaengine: dw-axi-dmac: Add support for CV1800B DMA

[Inochi Amaoto]

On Sat, Dec 13, 2025 at 04:55:06PM +0900, Samuel Holland wrote:
> Hi Inochi,
> 
> On 2025-12-12 11:05 AM, Inochi Amaoto wrote:
> > As the DMA controller on Sophgo CV1800 series SoC only has 8 channels,
> > the SoC provides a dma multiplexer to reuse the DMA channel. However,
> > the dma multiplexer also controlls the DMA interrupt multiplexer, which
> 
> typo: controls
> 

Thanks.

> > means that the dma multiplexer needs to know the channel number.
> > 
> > Allow the driver to use DMA phandle args as the channel number, so the
> > DMA multiplexer can route the DMA interrupt correctly.
> > 
> > Signed-off-by: Inochi Amaoto <inochiama@gmail.com>
> > ---
> >  .../dma/dw-axi-dmac/dw-axi-dmac-platform.c    | 23 ++++++++++++++++---
> >  drivers/dma/dw-axi-dmac/dw-axi-dmac.h         |  1 +
> >  2 files changed, 21 insertions(+), 3 deletions(-)
> > 
> > diff --git a/drivers/dma/dw-axi-dmac/dw-axi-dmac-platform.c b/drivers/dma/dw-axi-dmac/dw-axi-dmac-platform.c
> > index b23536645ff7..62bf0d0dc354 100644
> > --- a/drivers/dma/dw-axi-dmac/dw-axi-dmac-platform.c
> > +++ b/drivers/dma/dw-axi-dmac/dw-axi-dmac-platform.c
> > @@ -50,6 +50,7 @@
> >  #define AXI_DMA_FLAG_HAS_APB_REGS	BIT(0)
> >  #define AXI_DMA_FLAG_HAS_RESETS		BIT(1)
> >  #define AXI_DMA_FLAG_USE_CFG2		BIT(2)
> > +#define AXI_DMA_FLAG_HANDSHAKE_AS_CHAN	BIT(3)
> > 
> >  static inline void
> >  axi_dma_iowrite32(struct axi_dma_chip *chip, u32 reg, u32 val)
> > @@ -1361,15 +1362,26 @@ static struct dma_chan *dw_axi_dma_of_xlate(struct of_phandle_args *dma_spec,
> >  					    struct of_dma *ofdma)
> >  {
> >  	struct dw_axi_dma *dw = ofdma->of_dma_data;
> > +	unsigned int handshake = dma_spec->args[0];
> >  	struct axi_dma_chan *chan;
> >  	struct dma_chan *dchan;
> > 
> > -	dchan = dma_get_any_slave_channel(&dw->dma);
> > +	if (dw->hdata->use_handshake_as_channel_number) {
> > +		if (handshake >= dw->hdata->nr_channels)
> > +			return NULL;
> > +
> > +		chan = &dw->chan[handshake];
> > +		dchan = dma_get_slave_channel(&chan->vc.chan);
> > +	} else {
> > +		dchan = dma_get_any_slave_channel(&dw->dma);
> > +	}
> > +
> >  	if (!dchan)
> >  		return NULL;
> > 
> > -	chan = dchan_to_axi_dma_chan(dchan);
> > -	chan->hw_handshake_num = dma_spec->args[0];
> > +	if (!chan)
> 
> When use_handshake_as_channel_number is false, chan is uninitialized here.
> 
> Regards,
> Samuel
> 

Thanks, I also noticed this, will fixed in the V2.

Regards,
Inochi

> > +		chan = dchan_to_axi_dma_chan(dchan);
> > +	chan->hw_handshake_num = handshake;
> >  	return dchan;
> >  }
> > 
> > @@ -1508,6 +1520,8 @@ static int dw_probe(struct platform_device *pdev)
> >  			return ret;
> >  	}
> > 
> > +	chip->dw->hdata->use_handshake_as_channel_number = !!(flags & AXI_DMA_FLAG_HANDSHAKE_AS_CHAN);
> > +
> >  	chip->dw->hdata->use_cfg2 = !!(flags & AXI_DMA_FLAG_USE_CFG2);
> > 
> >  	chip->core_clk = devm_clk_get(chip->dev, "core-clk");
> > @@ -1663,6 +1677,9 @@ static const struct of_device_id dw_dma_of_id_table[] = {
> >  	}, {
> >  		.compatible = "intel,kmb-axi-dma",
> >  		.data = (void *)AXI_DMA_FLAG_HAS_APB_REGS,
> > +	}, {
> > +		.compatible = "sophgo,cv1800b-axi-dma",
> > +		.data = (void *)AXI_DMA_FLAG_HANDSHAKE_AS_CHAN,
> >  	}, {
> >  		.compatible = "starfive,jh7110-axi-dma",
> >  		.data = (void *)(AXI_DMA_FLAG_HAS_RESETS | AXI_DMA_FLAG_USE_CFG2),
> > diff --git a/drivers/dma/dw-axi-dmac/dw-axi-dmac.h b/drivers/dma/dw-axi-dmac/dw-axi-dmac.h
> > index b842e6a8d90d..67cc199e24d1 100644
> > --- a/drivers/dma/dw-axi-dmac/dw-axi-dmac.h
> > +++ b/drivers/dma/dw-axi-dmac/dw-axi-dmac.h
> > @@ -34,6 +34,7 @@ struct dw_axi_dma_hcfg {
> >  	bool	reg_map_8_channels;
> >  	bool	restrict_axi_burst_len;
> >  	bool	use_cfg2;
> > +	bool	use_handshake_as_channel_number;
> >  };
> > 
> >  struct axi_dma_chan {
> > --
> > 2.52.0
> > 
> > 
> > _______________________________________________
> > linux-riscv mailing list
> > linux-riscv@lists.infradead.org
> > http://lists.infradead.org/mailman/listinfo/linux-riscv
> 

[PATCH 3/3] riscv: dts: sophgo: cv180x: Allow the DMA multiplexer to set channel number for DMA controller

[Inochi Amaoto]

Change the DMA controller compatible to the sophgo,cv1800b-axi-dma,
which supports setting DMA channel number in DMA phandle args.

Fixes: 514951a81a5e ("riscv: dts: sophgo: cv18xx: add DMA controller")
Reported-by: Anton D. Stavinskii <stavinsky@gmail.com>
Closes: https://github.com/sophgo/linux/issues/9
Signed-off-by: Inochi Amaoto <inochiama@gmail.com>
---
 arch/riscv/boot/dts/sophgo/cv180x.dtsi | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/riscv/boot/dts/sophgo/cv180x.dtsi b/arch/riscv/boot/dts/sophgo/cv180x.dtsi
index 1b2b1969a648..e1b515b46466 100644
--- a/arch/riscv/boot/dts/sophgo/cv180x.dtsi
+++ b/arch/riscv/boot/dts/sophgo/cv180x.dtsi
@@ -417,7 +417,7 @@ sdhci1: mmc@4320000 {
 		};
 
 		dmac: dma-controller@4330000 {
-			compatible = "snps,axi-dma-1.01a";
+			compatible = "sophgo,cv1800b-axi-dma";
 			reg = <0x04330000 0x1000>;
 			interrupts = <SOC_PERIPHERAL_IRQ(13) IRQ_TYPE_LEVEL_HIGH>;
 			clocks = <&clk CLK_SDMA_AXI>, <&clk CLK_SDMA_AXI>;
-- 
2.52.0