Re: [PATCH 28/33] dmaengine: at_hdmac: Use devm_request_irq()

[Vinod Koul]

On 20-08-22, 15:57, Tudor Ambarus wrote:
> IRQs requested with this function will be automatically freed on driver
> detach. Use devm_request_irq() and make the code cleaner.

That is not really good idea!

> 
> Signed-off-by: Tudor Ambarus <tudor.ambarus@xxxxxxxxxxxxx>
> ---
>  drivers/dma/at_hdmac.c | 11 ++++-------
>  1 file changed, 4 insertions(+), 7 deletions(-)
> 
> diff --git a/drivers/dma/at_hdmac.c b/drivers/dma/at_hdmac.c
> index 9c414f167b62..96b885f83374 100644
> --- a/drivers/dma/at_hdmac.c
> +++ b/drivers/dma/at_hdmac.c
> @@ -2241,6 +2241,10 @@ static int __init at_dma_probe(struct platform_device *pdev)
>  	irq = platform_get_irq(pdev, 0);
>  	if (irq < 0)
>  		return irq;
> +	err = devm_request_irq(&pdev->dev, irq, at_dma_interrupt, 0,
> +			       dev_name(&pdev->dev), atdma);
> +	if (err)
> +		return err;
>  
>  	/* discover transaction capabilities */
>  	atdma->dma_device.cap_mask = plat_dat->cap_mask;
> @@ -2257,10 +2261,6 @@ static int __init at_dma_probe(struct platform_device *pdev)
>  	/* force dma off, just in case */
>  	at_dma_off(atdma);
>  
> -	err = request_irq(irq, at_dma_interrupt, 0, "at_hdmac", atdma);
> -	if (err)
> -		goto err_irq;

here you are ready to receive the irq, but now you moved it early on..
it may cause issues..

> -
>  	platform_set_drvdata(pdev, atdma);
>  
>  	/* create a pool of consistent memory blocks for hardware descriptors */
> @@ -2377,8 +2377,6 @@ static int __init at_dma_probe(struct platform_device *pdev)
>  err_memset_pool_create:
>  	dma_pool_destroy(atdma->dma_desc_pool);
>  err_desc_pool_create:
> -	free_irq(platform_get_irq(pdev, 0), atdma);
> -err_irq:
>  	clk_disable_unprepare(atdma->clk);
>  err_clk_prepare:
>  	clk_put(atdma->clk);
> @@ -2397,7 +2395,6 @@ static int at_dma_remove(struct platform_device *pdev)
>  
>  	dma_pool_destroy(atdma->memset_pool);
>  	dma_pool_destroy(atdma->dma_desc_pool);
> -	free_irq(platform_get_irq(pdev, 0), atdma);

Now we have an active irq while we are unrolling device, which can be
fired and cause tasklet to also get scheduled...

You need to make sure now irq cant fire here and tasklet is killed..

>  
>  	list_for_each_entry_safe(chan, _chan, &atdma->dma_device.channels,
>  			device_node) {
> -- 
> 2.25.1

-- 
~Vinod