[PATCHv2] crypto: cesa: check for sram_dma error

[PATCHv2] crypto: cesa: check for sram_dma error

[Rosen Penev]

dma_map_resource() might fail. In such a case, don't call
dma_unmap_resource()

Fixes: 37d728f76c41 ("crypto: marvell/cesa - Fix DMA API misuse")
Signed-off-by: Rosen Penev <rosenp@gmail.com>
---
 v2: use dma_mapping_error(). Fix Fixes tag.
 drivers/crypto/marvell/cesa/cesa.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/crypto/marvell/cesa/cesa.c b/drivers/crypto/marvell/cesa/cesa.c
index 75d8ba23d9a2..92e977027c8a 100644
--- a/drivers/crypto/marvell/cesa/cesa.c
+++ b/drivers/crypto/marvell/cesa/cesa.c
@@ -406,7 +406,8 @@ static void mv_cesa_put_sram(struct platform_device *pdev, int idx)
 	if (engine->pool)
 		gen_pool_free(engine->pool, (unsigned long)engine->sram_pool,
 			      cesa->sram_size);
-	else
+	else if (engine->sram &&
+		 !dma_mapping_error(cesa->dev, engine->sram_dma))
 		dma_unmap_resource(cesa->dev, engine->sram_dma,
 				   cesa->sram_size, DMA_BIDIRECTIONAL, 0);
 }
-- 
2.55.0

Re: [PATCHv2] crypto: cesa: check for sram_dma error

[Herbert Xu]

On Fri, Jul 17, 2026 at 03:58:31PM -0700, Rosen Penev wrote:
> dma_map_resource() might fail. In such a case, don't call
> dma_unmap_resource()
> 
> Fixes: 37d728f76c41 ("crypto: marvell/cesa - Fix DMA API misuse")
> Signed-off-by: Rosen Penev <rosenp@gmail.com>
> ---
>  v2: use dma_mapping_error(). Fix Fixes tag.
>  drivers/crypto/marvell/cesa/cesa.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)

I think it would be better to fix mv_cesa_probe to not call the
cleanup functions on uninitialised engines.

err_cleanup:
	for (i = 0; i < XXX; i++)
		mv_cesa_put_sram(pdev, i);

So replace XXX here with one past the last index that was successfully
initialised.

Cheers,
-- 
Email: Herbert Xu <herbert@gondor.apana.org.au>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt

Re: [PATCHv2] crypto: cesa: check for sram_dma error

[Robin Murphy]

On 17/07/2026 11:58 pm, Rosen Penev wrote:
> dma_map_resource() might fail. In such a case, don't call
> dma_unmap_resource()
> 
> Fixes: 37d728f76c41 ("crypto: marvell/cesa - Fix DMA API misuse")
> Signed-off-by: Rosen Penev <rosenp@gmail.com>
> ---
>   v2: use dma_mapping_error(). Fix Fixes tag.
>   drivers/crypto/marvell/cesa/cesa.c | 3 ++-
>   1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/crypto/marvell/cesa/cesa.c b/drivers/crypto/marvell/cesa/cesa.c
> index 75d8ba23d9a2..92e977027c8a 100644
> --- a/drivers/crypto/marvell/cesa/cesa.c
> +++ b/drivers/crypto/marvell/cesa/cesa.c
> @@ -406,7 +406,8 @@ static void mv_cesa_put_sram(struct platform_device *pdev, int idx)
>   	if (engine->pool)
>   		gen_pool_free(engine->pool, (unsigned long)engine->sram_pool,
>   			      cesa->sram_size);
> -	else
> +	else if (engine->sram &&

This is still not quite right if the 
devm_platform_get_and_ioremap_resource() call fails, since engine->sram 
will then be non-NULL (an ERR_PTR value), but engine->sram_dma is still 
untouched, and dma_mapping_error(0) is (usually) false.

Thanks,
Robin.

> +		 !dma_mapping_error(cesa->dev, engine->sram_dma))
>   		dma_unmap_resource(cesa->dev, engine->sram_dma,
>   				   cesa->sram_size, DMA_BIDIRECTIONAL, 0);
>   }