Re: [PATCH] of/address: Return an error when no valid dma-ranges are found

[Mark Brown]

On Fri, Jan 27, 2023 at 12:37:35PM -0600, Rob Herring wrote:

> Looks to me like we are leaking 'r' with this change.

Oh, probably now that you mention it.  Usually the OF code keeps
track of more things than I expect...

> Wouldn't this change work:

> diff --git a/drivers/of/address.c b/drivers/of/address.c
> index c34ac33b7338..f43311f01c32 100644
> --- a/drivers/of/address.c
> +++ b/drivers/of/address.c
> @@ -968,6 +968,11 @@ int of_dma_get_range(struct device_node *np,
> const struct bus_dma_region **map)
>         for_each_of_range(&parser, &range)
>                 num_ranges++;
> 
> +       if (!num_ranges) {
> +               ret = -EINVAL;
> +               goto out;
> +       }
> +

Not as-is, there is a range counted by that first loop but it's
then rejected by the check in the second loop for cpu_addr ==
OF_BAD_ADDR.  We'd need to add a similar check in the first loop.
It should work otherwise though and avoids doing the allocation
in this case.

Attachment: signature.asc
Description: PGP signature