Some stm32-dma consumers [1] rather use dma_get_slave_caps() to get
max_sg_burst of their DMA channel as dma_get_max_seg_size() is specific to
the DMA controller.
All stm32-dma channels have the same features so, don't need to implement
device_caps ops. Let dma_get_slave_caps() relies on dma_device
configuration.
That's why this patch sets dma_device max_sg_burst to the maximum segment
size, which is the maximum of data items that can be transferred without
software intervention.

[1] https://lore.kernel.org/lkml/20220110103739.118426-1-alain.volmat@foss.st.com/
    "media: stm32: dcmi: create a dma scatterlist based on DMA max_sg_burst value"

Signed-off-by: Amelie Delaunay <amelie.delaunay@foss.st.com>
---
 drivers/dma/stm32-dma.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/drivers/dma/stm32-dma.c b/drivers/dma/stm32-dma.c
index 83a37a6955a3..d2365fab1b7a 100644
--- a/drivers/dma/stm32-dma.c
+++ b/drivers/dma/stm32-dma.c
@@ -1389,6 +1389,7 @@ static int stm32_dma_probe(struct platform_device *pdev)
 	dd->residue_granularity = DMA_RESIDUE_GRANULARITY_BURST;
 	dd->copy_align = DMAENGINE_ALIGN_32_BYTES;
 	dd->max_burst = STM32_DMA_MAX_BURST;
+	dd->max_sg_burst = STM32_DMA_ALIGNED_MAX_DATA_ITEMS;
 	dd->descriptor_reuse = true;
 	dd->dev = &pdev->dev;
 	INIT_LIST_HEAD(&dd->channels);
-- 
2.25.1

Hi,

thanks Amelie.

Tested-by: Alain Volmat <alain.volmat@foss.st.com>

Alain

On Mon, Jan 17, 2022 at 10:17:40AM +0100, Amelie Delaunay wrote:
> Some stm32-dma consumers [1] rather use dma_get_slave_caps() to get
> max_sg_burst of their DMA channel as dma_get_max_seg_size() is specific to
> the DMA controller.
> All stm32-dma channels have the same features so, don't need to implement
> device_caps ops. Let dma_get_slave_caps() relies on dma_device
> configuration.
> That's why this patch sets dma_device max_sg_burst to the maximum segment
> size, which is the maximum of data items that can be transferred without
> software intervention.
> 
> [1] https://lore.kernel.org/lkml/20220110103739.118426-1-alain.volmat@foss.st.com/
>     "media: stm32: dcmi: create a dma scatterlist based on DMA max_sg_burst value"
> 
> Signed-off-by: Amelie Delaunay <amelie.delaunay@foss.st.com>
> ---
>  drivers/dma/stm32-dma.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/drivers/dma/stm32-dma.c b/drivers/dma/stm32-dma.c
> index 83a37a6955a3..d2365fab1b7a 100644
> --- a/drivers/dma/stm32-dma.c
> +++ b/drivers/dma/stm32-dma.c
> @@ -1389,6 +1389,7 @@ static int stm32_dma_probe(struct platform_device *pdev)
>  	dd->residue_granularity = DMA_RESIDUE_GRANULARITY_BURST;
>  	dd->copy_align = DMAENGINE_ALIGN_32_BYTES;
>  	dd->max_burst = STM32_DMA_MAX_BURST;
> +	dd->max_sg_burst = STM32_DMA_ALIGNED_MAX_DATA_ITEMS;
>  	dd->descriptor_reuse = true;
>  	dd->dev = &pdev->dev;
>  	INIT_LIST_HEAD(&dd->channels);
> -- 
> 2.25.1
> 

On 17-01-22, 10:17, Amelie Delaunay wrote:
> Some stm32-dma consumers [1] rather use dma_get_slave_caps() to get
> max_sg_burst of their DMA channel as dma_get_max_seg_size() is specific to
> the DMA controller.
> All stm32-dma channels have the same features so, don't need to implement
> device_caps ops. Let dma_get_slave_caps() relies on dma_device
> configuration.
> That's why this patch sets dma_device max_sg_burst to the maximum segment
> size, which is the maximum of data items that can be transferred without
> software intervention.

Applied, thanks

-- 
~Vinod