clk: divider: fix selection of divider when rounding to closest

It's an invalid approach to assume that among two divider values
the one nearer the exact divider is the better one.

Assume a parent rate of 1000 Hz, a divider with CLK_DIVIDER_POWER_OF_TWO
and a target rate of 89 Hz. The exact divider is ~ 11.236 so 8 and 16
are the candidates to choose from yielding rates 125 Hz and 62.5 Hz
respectivly. While 8 is nearer to 11.236 than 16 is, the latter is still
the better divider as 62.5 is nearer to 89 than 125 is.

Fixes: 774b514 (clk: divider: Add round to closest divider)
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Acked-by: Sascha Hauer <s.hauer@pengutronix.de>
Acked-by: Maxime Coquelin <maxime.coquelin@st.com>
Signed-off-by: Michael Turquette <mturquette@linaro.org>

drivers/clk/clk-divider.c

@@ -220,6 +220,7 @@ static int _div_round_closest(const struct clk_div_table *table,
 			      unsigned long flags)
 {
 	int up, down, div;
+	unsigned long up_rate, down_rate;
 
 	up = down = div = DIV_ROUND_CLOSEST(parent_rate, rate);
 
@@ -231,7 +232,10 @@ static int _div_round_closest(const struct clk_div_table *table,
 		down = _round_down_table(table, div);
 	}
 
-	return (up - div) <= (div - down) ? up : down;
+	up_rate = DIV_ROUND_UP(parent_rate, up);
+	down_rate = DIV_ROUND_UP(parent_rate, down);
+
+	return (rate - up_rate) <= (down_rate - rate) ? up : down;
 }
 
 static int _div_round(const struct clk_div_table *table,