Re: [PATCH v2] dmaengine: dw-edma: fix MSI data programming for multi-IRQ case

[Frank Li <Frank.li@nxp.com>]

On Thu, Jan 29, 2026 at 10:20:24AM +0800, Shenghui Shi wrote:
> When using MSI (not MSI-X) with multiple IRQs, the MSI data value
> must be unique per vector to ensure correct interrupt delivery.
> Currently, the driver fails to increment the MSI data per vector,
> causing interrupts to be misrouted.
>
> Fix this by caching the base MSI data and adjusting each vector's
> data accordingly during IRQ setup.
>
> This issue was reproduced and tested on:
> - Device: [20e0:2502] (rev 01)
> - Kernel: 6.8.0-90-generic
>
> Fixes: e63d79d1ff04 ("dmaengine: dw-edma: Add Synopsys DesignWare eDMA IP core driver")
>
> Signed-off-by: Shenghui Shi <brody.shi@m2semi.com>
> ---
>  drivers/dma/dw-edma/dw-edma-core.c | 25 +++++++++++++++++++++++--
>  1 file changed, 23 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/dma/dw-edma/dw-edma-core.c b/drivers/dma/dw-edma/dw-edma-core.c
> index 8e5f7defa..516770388 100644
> --- a/drivers/dma/dw-edma/dw-edma-core.c
> +++ b/drivers/dma/dw-edma/dw-edma-core.c
> @@ -844,11 +844,15 @@ static int dw_edma_irq_request(struct dw_edma *dw,
>  {
>  	struct dw_edma_chip *chip = dw->chip;
>  	struct device *dev = dw->chip->dev;
> +	struct msi_desc *msi_desc;
>  	u32 wr_mask = 1;
>  	u32 rd_mask = 1;
>  	int i, err = 0;
>  	u32 ch_cnt;
>  	int irq;
> +	u16 msi_base_data = 0;
> +	bool msi_base_valid = false;
> +	bool is_msix = false;
>
>  	ch_cnt = dw->wr_ch_cnt + dw->rd_ch_cnt;
>
> @@ -869,8 +873,15 @@ static int dw_edma_irq_request(struct dw_edma *dw,
>  			return err;
>  		}
>
> -		if (irq_get_msi_desc(irq))
> +		if (irq_get_msi_desc(irq)) {
>  			get_cached_msi_msg(irq, &dw->irq[0].msi);
> +			msi_desc = irq_get_msi_desc(irq);
> +			is_msix = msi_desc && msi_desc->pci.msi_attrib.is_msix;
> +			if (!is_msix) {
> +				msi_base_data = dw->irq[0].msi.data;
> +				msi_base_valid = true;
> +			}
> +		}

look like this dead code. Not touch dw->irq[i].msi.data at all. All local
variable will not affect else branch.

>
>  		dw->nr_irqs = 1;
>  	} else {
> @@ -896,8 +907,18 @@ static int dw_edma_irq_request(struct dw_edma *dw,
>  			if (err)
>  				goto err_irq_free;
>
> -			if (irq_get_msi_desc(irq))
> +			if (irq_get_msi_desc(irq)) {
>  				get_cached_msi_msg(irq, &dw->irq[i].msi);
> +				msi_desc = irq_get_msi_desc(irq);
> +				is_msix = msi_desc && msi_desc->pci.msi_attrib.is_msix;
> +				if (!is_msix) {
> +					if (!msi_base_valid) {
> +						msi_base_data = dw->irq[i].msi.data;
> +						msi_base_valid = true;
> +					}
> +					dw->irq[i].msi.data = (u16)(msi_base_data + i);

look like all dw->irq[i].msi.data is same for msi. (according to your
commit message).

Simple dw->irq[i].msi.data += i; should work. Needn't msi_base_data and

msi_base_valid

Frank

> +				}
> +			}
>  		}
>
>  		dw->nr_irqs = i;
> --
> 2.49.0.windows.1
>