[PATCH linux dev-4.13] ARM: dts: aspeed: Move random number device

Joel Stanley joel at jms.id.au
Thu May 10 12:35:01 AEST 2018


On 10 May 2018 at 11:32, Andrew Jeffery <andrew at aj.id.au> wrote:
> On Fri, 4 May 2018, at 12:13, Joel Stanley wrote:
>> Move the node out from under the syscon/simple-mfd.
>>
>> Being a child of this node causes the driver to fail to probe, as
>> platform_get_resource returns NULL due to dev->num_resources being zero.
>
> Ugh. I'm frustrated by the schism that simple-mfd/syscon introduces. It looks like such an attractive solution right up until its not, when you have existing drivers like this that you want to plug into MMIO space already covered by the syscon/simple-mfd.
>
> Do we have a way out of this kind of mess?
>
> What about adding regmap support to the timeriomem_rng driver?

I'm sure that they would take patches, especially if it retains the
current behaviour. Given how little it's reading from the one
register, there are no concerns with overhead/performance.

In general, it would be good to not have to rewrite drivers in order
to use them. Is it a matter of allowing more than one device to bind a
of resource?


More information about the openbmc mailing list