On 09/04/2014 02:53 AM, Mike Turquette wrote: > Quoting Stephen Boyd (2014-09-03 16:39:37) >> On 09/03/14 08:33, Tomeu Vizoso wrote: >>> +int clk_set_ceiling_rate(struct clk *clk_user, unsigned long rate) >>> +{ >>> + struct clk_core *clk = clk_to_clk_core(clk_user); >>> + >>> + WARN(rate > 0 && rate < clk_user->floor_constraint, >>> + "clk %s dev %s con %s: new ceiling %lu lower than existing floor >>> %lu\n", >>> + __clk_get_name(clk), clk_user->dev_id, clk_user->con_id, rate, >>> + clk_user->floor_constraint); >>> + >>> + clk_user->ceiling_constraint = rate; >>> + return clk_provider_set_rate(clk, clk_provider_get_rate(clk)); >>> +} >>> +EXPORT_SYMBOL_GPL(clk_set_ceiling_rate); >> >> Maybe I'm late to this patch series given that Mike applied it, but I >> wonder why we wouldn't just have one API that takes a min and a max, >> i.e. clk_set_rate_range(clk, min, max)? Then clk_set_rate() is a small >> wrapper on top that just sets min and max to the same value. > > We certainly can have that. But being able to easily adjust a floor or > ceiling value seems like a good thing to me, and that is what these > functions do. > > If we decide to have a clk_set_rate_range (where we perhaps pass zero in > for a value that we do not wish to constrain) then I imagine that > clk_set_ceiling_rate and clk_set_floor_rate will simply become a wrapper > for that function. No harm having it both ways. If one way of doing > things falls out of favor we can always cull it and update all the > users.
I opted for separate functions because in the specific use cases I thought of, any user will be interested in setting either a floor or a ceiling constraint, but not both. Regards, Tomeu -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/