Re: [PATCH 2/3] dmaengine: Add support for immediate commands in the client API

From: Frank Li

Date: Wed Nov 18 2026 - 15:03:41 EST


On Sun, Nov 15, 2026 at 05:02:44PM +0300, Nickolay Goppen wrote:
> [You don't often get email from setotau@xxxxxxxxxxxxxx. Learn why this is important at https://aka.ms/LearnAboutSenderIdentification ]
>
> From: Sireesh Kodali <sireeshkodali1@xxxxxxxxx>
>
> Immediate commands are needed by the IPA network driver, so that it can
> send commands via BAM to the IPA microcontroller.
>
> Signed-off-by: Sireesh Kodali <sireeshkodali1@xxxxxxxxx>
> Signed-off-by: Nickolay Goppen <setotau@xxxxxxxxxxxxxx>
> ---
> include/linux/dmaengine.h | 4 ++++
> 1 file changed, 4 insertions(+)
>
> diff --git a/include/linux/dmaengine.h b/include/linux/dmaengine.h
> index 805668fbe437..56c49254ed04 100644
> --- a/include/linux/dmaengine.h
> +++ b/include/linux/dmaengine.h
> @@ -200,6 +200,9 @@ struct dma_vec {
> * transaction is marked with DMA_PREP_REPEAT will cause the new transaction
> * to never be processed and stay in the issued queue forever. The flag is
> * ignored if the previous transaction is not a repeated transaction.
> + * @DMA_PREP_IMM_CMD: tell the driver that the data passed to the DMA API is
> + * immediate command data and the descriptor should be in a different format
> + * from the normal data and descriptor

does dmaengine_desc_attach_metadata() work for your case?

Frank

> */
> enum dma_ctrl_flags {
> DMA_PREP_INTERRUPT = (1 << 0),
> @@ -212,6 +215,7 @@ enum dma_ctrl_flags {
> DMA_PREP_CMD = (1 << 7),
> DMA_PREP_REPEAT = (1 << 8),
> DMA_PREP_LOAD_EOT = (1 << 9),
> + DMA_PREP_IMM_CMD = (1 << 10),
> };
>
> /**
>
> --
> 2.56.1