Messages in this thread | | | From | Masahiro Yamada <> | Date | Fri, 16 Sep 2016 16:41:44 +0900 | Subject | Re: [PATCH v6 1/2] clk: uniphier: add core support code for UniPhier clock driver |
| |
Hi Stephen,
2016-09-07 9:32 GMT+09:00 Stephen Boyd <sboyd@codeaurora.org>: > On 09/05, Masahiro Yamada wrote: >> 2016-08-30 3:22 GMT+09:00 Stephen Boyd <sboyd@codeaurora.org>: >> > On 08/29, Masahiro Yamada wrote: >> >> I tried this, but it did not work. >> >> >> >> To make dev_get_regmap() work, >> >> the parent device needs to call dev_regmap_init_mmio() beforehand. >> >> >> >> >> >> Since commit bdb0066df96e74a4002125467ebe459feff1ebef >> >> (mfd: syscon: Decouple syscon interface from platform devices), >> >> syscon_probe() is not called for platform devices, >> >> so that never happens. >> >> >> > >> > Ok. Is the syscon also a simple-mfd? >> > >> > It sounds like there's a device for the parent, but we've failed >> > to attach a regmap to it. Maybe the core DT code should assign >> > the regmap to the parent device when it creates it so that child >> > devices don't need to know this detail? It could look for >> > simple-mfd devices with compatible = "syscon" and then create the >> > regmap? Here's a totally untested patch for that. >> > >> >> >> I was not quire sure about this, >> but maybe worth submitting to DT subsystem? >> >> Anyway, I will go with syscon_node_to_regmap() for v7. >> Of course, I am happy to replace it with dev_get_regmap() >> when the issue is solved in the mainline. >> > > Ok that's fine. Did my patch fix dev_get_regmap() for you though? > That would be useful to know so that this patch can be merged in > parallel to yours.
Sorry for my late reply.
Yup, your patch worked fine!
And, I've posted v7 for my SoC clk driver.
-- Best Regards Masahiro Yamada
| |