From ac95330b96376550ae7a533d1396272d675adfa2 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?St=C3=A9phan=20Rafin?= Date: Fri, 4 Nov 2016 00:53:56 +0100 Subject: clk: sunxi: Fix M factor computation for APB1 MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit commit cfa636886033 ("clk: sunxi: factors: Consolidate get_factors parameters into a struct") introduced a regression for m factor computation in sun4i_get_apb1_factors function. The old code reassigned the "parent_rate" parameter to the targeted divisor value and was buggy for the returned frequency but not for the computed factors. Now, returned frequency is good but m factor is incorrectly computed (its max value 31 is always set resulting in a significantly slower frequency than the requested one...) This patch simply restores the original proper computation for m while keeping the good changes for returned rate. Fixes: cfa636886033 ("clk: sunxi: factors: Consolidate get_factors parameters into a struct") Signed-off-by: Stéphan Rafin Signed-off-by: Maxime Ripard diff --git a/drivers/clk/sunxi/clk-sunxi.c b/drivers/clk/sunxi/clk-sunxi.c index 838b22a..f2c9274 100644 --- a/drivers/clk/sunxi/clk-sunxi.c +++ b/drivers/clk/sunxi/clk-sunxi.c @@ -373,7 +373,7 @@ static void sun4i_get_apb1_factors(struct factors_request *req) else calcp = 3; - calcm = (req->parent_rate >> calcp) - 1; + calcm = (div >> calcp) - 1; req->rate = (req->parent_rate >> calcp) / (calcm + 1); req->m = calcm; -- cgit v0.10.2