mx31: correct csi_clk parent (v2)
changes since v1: we now check if the parent configuration bit was changed since reset and change the parent when needed. csi_clk parent was defined with ahb_clk. However, according to the m31 reference manual, it should be serial_pll_clk. Guennadi always used a 20 MHz clock that was by chance changed to a 45 MHz that fits in the mt9t031 spec. Now the clocks are computed and output correctly (measured on oscillo). Signed-off-by: Valentin Longchamp <valentin.longchamp@epfl.ch> Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
This commit is contained in:
parent
0b0ef44295
commit
0573cb5f45
|
@ -483,7 +483,7 @@ DEFINE_CLOCK(i2c3_clk, 2, MXC_CCM_CGR0, 30, NULL, NULL, &perclk_clk);
|
|||
DEFINE_CLOCK(mpeg4_clk, 0, MXC_CCM_CGR1, 0, NULL, NULL, &ahb_clk);
|
||||
DEFINE_CLOCK(mstick1_clk, 0, MXC_CCM_CGR1, 2, mstick1_get_rate, NULL, &usb_pll_clk);
|
||||
DEFINE_CLOCK(mstick2_clk, 1, MXC_CCM_CGR1, 4, mstick2_get_rate, NULL, &usb_pll_clk);
|
||||
DEFINE_CLOCK1(csi_clk, 0, MXC_CCM_CGR1, 6, csi, NULL, &ahb_clk);
|
||||
DEFINE_CLOCK1(csi_clk, 0, MXC_CCM_CGR1, 6, csi, NULL, &serial_pll_clk);
|
||||
DEFINE_CLOCK(rtc_clk, 0, MXC_CCM_CGR1, 8, NULL, NULL, &ipg_clk);
|
||||
DEFINE_CLOCK(wdog_clk, 0, MXC_CCM_CGR1, 10, NULL, NULL, &ipg_clk);
|
||||
DEFINE_CLOCK(pwm_clk, 0, MXC_CCM_CGR1, 12, NULL, NULL, &perclk_clk);
|
||||
|
@ -571,6 +571,13 @@ int __init mx31_clocks_init(unsigned long fref)
|
|||
for (i = 0; i < ARRAY_SIZE(lookups); i++)
|
||||
clkdev_add(&lookups[i]);
|
||||
|
||||
/* change the csi_clk parent if necessary */
|
||||
reg = __raw_readl(MXC_CCM_CCMR);
|
||||
if (!(reg & MXC_CCM_CCMR_CSCS))
|
||||
if (clk_set_parent(&csi_clk, &usb_pll_clk))
|
||||
pr_err("%s: error changing csi_clk parent\n", __func__);
|
||||
|
||||
|
||||
/* Turn off all possible clocks */
|
||||
__raw_writel((3 << 4), MXC_CCM_CGR0);
|
||||
__raw_writel(0, MXC_CCM_CGR1);
|
||||
|
|
Loading…
Reference in a new issue