On 09/12/2016 07:54 PM, Wenyou Yang wrote:
In order to make clk->dev available in ops->of_xlate() to get the
clock ID from the 'reg' property of the clock node, assign the
clk->dev before calling ops->of_xlate().
It does seem reasonable to me to allow using the same of_xlate
implementation across multiple similar-but-different clock providers,
and this change is required to allow that. I note that the reset and
power domain uclasses already work this way. So, from that perspective,
you get an ack.
diff --git a/drivers/clk/clk-uclass.c b/drivers/clk/clk-uclass.c
@@ -80,6 +80,9 @@ int clk_get_by_index(struct udevice *dev, int index, struct
clk *clk)
__func__, ret);
return ret;
}
+
+ clk->dev = dev_clk;
That assignment also happens in clk_request() itself. I'm tempted to say
that we should modify clk_request() to remove the dev parameter, and
remove the assignment of clk->dev. clk_request()'s documentation says:
* @clock: A pointer to a clock struct to initialize. The caller must
* have already initialized any field in this struct which the
* clock provider uses to identify the clock.
... and since clk->dev is a field that is used to identify the clock, it
really should be set already, thus removing the need for this function
to set clk->dev.
However, I suppose we can clean that up later, so,
Acked-by: Stephen Warren <swar...@nvidia.com>
_______________________________________________
U-Boot mailing list
U-Boot@lists.denx.de
http://lists.denx.de/mailman/listinfo/u-boot
