Messages in this thread | | | Date | Fri, 26 Apr 2019 23:06:51 +0200 | From | Alexandre Belloni <> | Subject | Re: [PATCH v3 7/7] clk: at91: add sam9x60 pmc driver |
| |
On 25/04/2019 14:20:41-0700, Stephen Boyd wrote: > Quoting Alexandre Belloni (2019-04-25 14:10:09) > > On 25/04/2019 13:58:10-0700, Stephen Boyd wrote: > > > > IIRC two clocks are used by the clocksource and having it as a platform > > > > device make the clocksource driver fail with -EPROBE_DEFER. > > > > > > > > > > Ok. I'll add a comment in the driver. It needs to be > > > CLK_OF_DECLARE_DRIVER() because it's used somewhere else too? Or it can > > > be CLK_OF_DECLARE() instead? > > > > It is coming from that comment: > > https://lore.kernel.org/lkml/153573753043.93865.2420370848457480370@swboyd.mtv.corp.google.com/ > > > > Maybe I misinterpreted it. > > > > Hmm I was just confused it seems. If you don't have two devices probing > the same DT node with this compatible string then I think it can just be > CLK_OF_DECLARE(). I was thinking that you were running into some sort of > issue where the other driver using this same compatible couldn't probe. >
I tried separating it into two initializations and do as much as possible in the platform driver but because the TCB clocks are peripheral clocks that depend on most of the other ones, I didn't find it convenient. This can safely be changed to CLK_OF_DECLARE. I can send v4 if you want. This should probably be changed for the other drivers too.
-- Alexandre Belloni, Bootlin Embedded Linux and Kernel engineering https://bootlin.com
| |