[PATCH] dmaengine: nbpfaxi: Add missing check after DMA map

Thomas Fourier posted 1 patch 3 months ago
drivers/dma/nbpfaxi.c | 13 +++++++++++++
1 file changed, 13 insertions(+)
[PATCH] dmaengine: nbpfaxi: Add missing check after DMA map
Posted by Thomas Fourier 3 months ago
The DMA map functions can fail and should be tested for errors.
If the mapping fails, unmap and return an error.

Fixes: b45b262cefd5 ("dmaengine: add a driver for AMBA AXI NBPF DMAC IP cores")
Signed-off-by: Thomas Fourier <fourier.thomas@gmail.com>
---
 drivers/dma/nbpfaxi.c | 13 +++++++++++++
 1 file changed, 13 insertions(+)

diff --git a/drivers/dma/nbpfaxi.c b/drivers/dma/nbpfaxi.c
index 0d6324c4e2be..0b75bb122898 100644
--- a/drivers/dma/nbpfaxi.c
+++ b/drivers/dma/nbpfaxi.c
@@ -711,6 +711,9 @@ static int nbpf_desc_page_alloc(struct nbpf_channel *chan)
 		list_add_tail(&ldesc->node, &lhead);
 		ldesc->hwdesc_dma_addr = dma_map_single(dchan->device->dev,
 					hwdesc, sizeof(*hwdesc), DMA_TO_DEVICE);
+		if (dma_mapping_error(dchan->device->dev,
+				      ldesc->hwdesc_dma_addr))
+			goto unmap_error;
 
 		dev_dbg(dev, "%s(): mapped 0x%p to %pad\n", __func__,
 			hwdesc, &ldesc->hwdesc_dma_addr);
@@ -737,6 +740,16 @@ static int nbpf_desc_page_alloc(struct nbpf_channel *chan)
 	spin_unlock_irq(&chan->lock);
 
 	return ARRAY_SIZE(dpage->desc);
+
+unmap_error:
+	while (i--) {
+		ldesc--; hwdesc--;
+
+		dma_unmap_single(dchan->device->dev, ldesc->hwdesc_dma_addr,
+				 sizeof(hwdesc), DMA_TO_DEVICE);
+	}
+
+	return -ENOMEM;
 }
 
 static void nbpf_desc_put(struct nbpf_desc *desc)
-- 
2.43.0
Re: [PATCH] dmaengine: nbpfaxi: Add missing check after DMA map
Posted by Vinod Koul 2 months, 2 weeks ago
On Mon, 07 Jul 2025 09:57:16 +0200, Thomas Fourier wrote:
> The DMA map functions can fail and should be tested for errors.
> If the mapping fails, unmap and return an error.
> 
> 

Applied, thanks!

[1/1] dmaengine: nbpfaxi: Add missing check after DMA map
      commit: c6ee78fc8f3e653bec427cfd06fec7877ee782bd

Best regards,
-- 
~Vinod
Re: [PATCH] dmaengine: nbpfaxi: Add missing check after DMA map
Posted by Vinod Koul 2 months, 2 weeks ago
On 07-07-25, 09:57, Thomas Fourier wrote:
> The DMA map functions can fail and should be tested for errors.
> If the mapping fails, unmap and return an error.

Pls change this to dmaengine: xxx that is the subsystem name as show in
the Fixes tag that you picked up below

Meanwhile this does not apply for me, pls rebase

> 
> Fixes: b45b262cefd5 ("dmaengine: add a driver for AMBA AXI NBPF DMAC IP cores")
> Signed-off-by: Thomas Fourier <fourier.thomas@gmail.com>
> ---
>  drivers/dma/nbpfaxi.c | 13 +++++++++++++
>  1 file changed, 13 insertions(+)
> 
> diff --git a/drivers/dma/nbpfaxi.c b/drivers/dma/nbpfaxi.c
> index 0d6324c4e2be..0b75bb122898 100644
> --- a/drivers/dma/nbpfaxi.c
> +++ b/drivers/dma/nbpfaxi.c
> @@ -711,6 +711,9 @@ static int nbpf_desc_page_alloc(struct nbpf_channel *chan)
>  		list_add_tail(&ldesc->node, &lhead);
>  		ldesc->hwdesc_dma_addr = dma_map_single(dchan->device->dev,
>  					hwdesc, sizeof(*hwdesc), DMA_TO_DEVICE);
> +		if (dma_mapping_error(dchan->device->dev,
> +				      ldesc->hwdesc_dma_addr))
> +			goto unmap_error;
>  
>  		dev_dbg(dev, "%s(): mapped 0x%p to %pad\n", __func__,
>  			hwdesc, &ldesc->hwdesc_dma_addr);
> @@ -737,6 +740,16 @@ static int nbpf_desc_page_alloc(struct nbpf_channel *chan)
>  	spin_unlock_irq(&chan->lock);
>  
>  	return ARRAY_SIZE(dpage->desc);
> +
> +unmap_error:
> +	while (i--) {
> +		ldesc--; hwdesc--;
> +
> +		dma_unmap_single(dchan->device->dev, ldesc->hwdesc_dma_addr,
> +				 sizeof(hwdesc), DMA_TO_DEVICE);
> +	}
> +
> +	return -ENOMEM;
>  }
>  
>  static void nbpf_desc_put(struct nbpf_desc *desc)
> -- 
> 2.43.0

-- 
~Vinod
Re: [PATCH] dmaengine: nbpfaxi: Add missing check after DMA map
Posted by Vinod Koul 2 months, 2 weeks ago
On 23-07-25, 17:36, Vinod Koul wrote:
> On 07-07-25, 09:57, Thomas Fourier wrote:
> > The DMA map functions can fail and should be tested for errors.
> > If the mapping fails, unmap and return an error.
> 
> Pls change this to dmaengine: xxx that is the subsystem name as show in
> the Fixes tag that you picked up below

Ignore this one pls, that was for another patch of your which I fixed up
while applying

-- 
~Vinod