RE: [RFC 1/2] of: reserved_mem: check return value of_dma_configure

[Peng Fan]

> -----Original Message-----
> From: Peng Fan
> Sent: Sunday, November 26, 2017 9:14 PM
> To: hch@xxxxxx; m.szyprowski@xxxxxxxxxxx; robin.murphy@xxxxxxx;
> gregkh@xxxxxxxxxxxxxxxxxxx; robh+dt@xxxxxxxxxx; frowand.list@xxxxxxxxx
> Cc: iommu@xxxxxxxxxxxxxxxxxxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx;
> devicetree@xxxxxxxxxxxxxxx; van.freenix@xxxxxxxxx; Peng Fan
> <peng.fan@xxxxxxx>
> Subject: [RFC 1/2] of: reserved_mem: check return value of_dma_configure
> 
> In commit <7b07cbefb6>("iommu: of: Handle IOMMU lookup failure with
> deferred probing or error"), there is possibility that of_dma_configure may fail.
> So in of_reserved_mem_device_init_by_idx,
> also need to propagate the return value of_dma_configure to caller, when
> need to use reserved memory for a device which needs iommu.

Seems my understanding is wrong, with iommu enabled, reserved memory will not be used
for devices with memory-region. Please still help review RFC 2/2.

Thanks,
Peng.

> 
> Signed-off-by: Peng Fan <peng.fan@xxxxxxx>
> ---
>  drivers/of/of_reserved_mem.c | 9 ++++++---
>  1 file changed, 6 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/of/of_reserved_mem.c b/drivers/of/of_reserved_mem.c
> index 22b75c82e377..61523819b50e 100644
> --- a/drivers/of/of_reserved_mem.c
> +++ b/drivers/of/of_reserved_mem.c
> @@ -357,9 +357,12 @@ int of_reserved_mem_device_init_by_idx(struct
> device *dev,
>  		/* ensure that dma_ops is set for virtual devices
>  		 * using reserved memory
>  		 */
> -		of_dma_configure(dev, np);
> -
> -		dev_info(dev, "assigned reserved memory node %s\n", rmem-
> >name);
> +		ret = of_dma_configure(dev, np);
> +		if (ret)
> +			of_reserved_mem_device_release(dev);
> +		else
> +			dev_info(dev, "assigned reserved memory node %s\n",
> +				 rmem->name);
>  	} else {
>  		kfree(rd);
>  	}
> --
> 2.14.1