| Message ID | 1403016777-15121-6-git-send-email-wens@csie.org (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Tue, Jun 17, 2014 at 9:52 AM, Chen-Yu Tsai <wens@csie.org> wrote: > The PLLs on newer Allwinner SoC's, such as the A31 and A23, have a > N multiplier factor that starts from 1, not 0. > > This patch adds an option to the clock driver's config data structures > to define the difference. > > Signed-off-by: Chen-Yu Tsai <wens@csie.org> > Acked-by: Maxime Ripard <maxime.ripard@free-electrons.com> > --- > drivers/clk/sunxi/clk-factors.c | 5 ++++- > drivers/clk/sunxi/clk-factors.h | 1 + > 2 files changed, 5 insertions(+), 1 deletion(-) > > diff --git a/drivers/clk/sunxi/clk-factors.c b/drivers/clk/sunxi/clk-factors.c > index 3806d97..399cf4d 100644 > --- a/drivers/clk/sunxi/clk-factors.c > +++ b/drivers/clk/sunxi/clk-factors.c > @@ -62,7 +62,10 @@ static unsigned long clk_factors_recalc_rate(struct clk_hw *hw, > p = FACTOR_GET(config->pshift, config->pwidth, reg); > > /* Calculate the rate */ > - rate = (parent_rate * n * (k + 1) >> p) / (m + 1); > + if (config->n_from_one) > + rate = (parent_rate * (n + 1) * (k + 1) >> p) / (m + 1); > + else > + rate = (parent_rate * n * (k + 1) >> p) / (m + 1); This can be simplified and support any base with: rate = (parent_rate * (n + config->n_base) * (k + 1) >> p) / (m + 1);
On Tue, Jun 17, 2014 at 10:52:42PM +0800, Chen-Yu Tsai wrote: > The PLLs on newer Allwinner SoC's, such as the A31 and A23, have a > N multiplier factor that starts from 1, not 0. > > This patch adds an option to the clock driver's config data structures > to define the difference. > > Signed-off-by: Chen-Yu Tsai <wens@csie.org> > Acked-by: Maxime Ripard <maxime.ripard@free-electrons.com> > --- > drivers/clk/sunxi/clk-factors.c | 5 ++++- > drivers/clk/sunxi/clk-factors.h | 1 + > 2 files changed, 5 insertions(+), 1 deletion(-) > > diff --git a/drivers/clk/sunxi/clk-factors.c b/drivers/clk/sunxi/clk-factors.c > index 3806d97..399cf4d 100644 > --- a/drivers/clk/sunxi/clk-factors.c > +++ b/drivers/clk/sunxi/clk-factors.c > @@ -62,7 +62,10 @@ static unsigned long clk_factors_recalc_rate(struct clk_hw *hw, > p = FACTOR_GET(config->pshift, config->pwidth, reg); > > /* Calculate the rate */ > - rate = (parent_rate * n * (k + 1) >> p) / (m + 1); > + if (config->n_from_one) > + rate = (parent_rate * (n + 1) * (k + 1) >> p) / (m + 1); > + else > + rate = (parent_rate * n * (k + 1) >> p) / (m + 1); Thinking a bit more about this, I wonder wether it wouldn't be better to just have a n_start variable or something, and just use (n + n_start) instead. That would avoid having to declare twice the same function. Maxime > > return rate; > } > diff --git a/drivers/clk/sunxi/clk-factors.h b/drivers/clk/sunxi/clk-factors.h > index 02e1a43..0484a48 100644 > --- a/drivers/clk/sunxi/clk-factors.h > +++ b/drivers/clk/sunxi/clk-factors.h > @@ -15,6 +15,7 @@ struct clk_factors_config { > u8 mwidth; > u8 pshift; > u8 pwidth; > + u8 n_from_one; Especially when you declare it as an u8, and not a bool. Maxime
diff --git a/drivers/clk/sunxi/clk-factors.c b/drivers/clk/sunxi/clk-factors.c index 3806d97..399cf4d 100644 --- a/drivers/clk/sunxi/clk-factors.c +++ b/drivers/clk/sunxi/clk-factors.c @@ -62,7 +62,10 @@ static unsigned long clk_factors_recalc_rate(struct clk_hw *hw, p = FACTOR_GET(config->pshift, config->pwidth, reg); /* Calculate the rate */ - rate = (parent_rate * n * (k + 1) >> p) / (m + 1); + if (config->n_from_one) + rate = (parent_rate * (n + 1) * (k + 1) >> p) / (m + 1); + else + rate = (parent_rate * n * (k + 1) >> p) / (m + 1); return rate; } diff --git a/drivers/clk/sunxi/clk-factors.h b/drivers/clk/sunxi/clk-factors.h index 02e1a43..0484a48 100644 --- a/drivers/clk/sunxi/clk-factors.h +++ b/drivers/clk/sunxi/clk-factors.h @@ -15,6 +15,7 @@ struct clk_factors_config { u8 mwidth; u8 pshift; u8 pwidth; + u8 n_from_one; }; struct clk_factors {