Hi Jean-Christophe,

I think you are a bit late and the series got taken by Nicolas. Anyway, most of
your comments have been taken care of.

On 17/03/2014 at 12:40:18 +0100, Jean-Christophe PLAGNIOL-VILLARD wrote :

On 11:05 Mon 03 Mar     , Jean-Jacques Hiblot wrote:

quoted

+		usb0: ohci at 00500000 {
+			compatible = "atmel,at91rm9200-ohci", "usb-ohci";
+			reg = <0x00500000 0x100000>;
+			interrupts = <20 IRQ_TYPE_LEVEL_HIGH 2>;

for 5th time NACK

stop using interrupts une interrupts-extended

As there is only one interrupt controller, I don't see why we would
absolutely need to specify the interrupt controller here.

quoted

+			clocks = <&usb>, <&ohci_clk>, <&hclk0>, <&uhpck>;
+			clock-names = "usb_clk", "ohci_clk", "hclk", "uhpck";
+			status = "disabled";
+		};

diff --git a/arch/arm/mach-at91/at91sam9261.c b/arch/arm/mach-at91/at91sam9261.c
index 6276b4c..5c90581 100644
--- a/arch/arm/mach-at91/at91sam9261.c
+++ b/arch/arm/mach-at91/at91sam9261.c

@@ -189,6 +189,23 @@ static struct clk_lookup periph_clocks_lookups[] = {
 	CLKDEV_CON_ID("pioA", &pioA_clk),
 	CLKDEV_CON_ID("pioB", &pioB_clk),
 	CLKDEV_CON_ID("pioC", &pioC_clk),
+	/* more lookup table for DT entries */
+	CLKDEV_CON_DEV_ID("usart", "fffff200.serial", &mck),
+	CLKDEV_CON_DEV_ID("usart", "fffb0000.serial", &usart0_clk),
+	CLKDEV_CON_DEV_ID("usart", "ffffb400.serial", &usart1_clk),
+	CLKDEV_CON_DEV_ID("usart", "fff94000.serial", &usart2_clk),
+	CLKDEV_CON_DEV_ID("t0_clk", "fffa0000.timer", &tc0_clk),
+	CLKDEV_CON_DEV_ID("t1_clk", "fffa0000.timer", &tc1_clk),
+	CLKDEV_CON_DEV_ID("t2_clk", "fffa0000.timer", &tc2_clk),
+	CLKDEV_CON_DEV_ID("hclk", "500000.ohci", &hck0),
+	CLKDEV_CON_DEV_ID("hclk", "600000.fb", &hck1),
+	CLKDEV_CON_DEV_ID("spi_clk", "fffc8000.spi", &spi0_clk),
+	CLKDEV_CON_DEV_ID("spi_clk", "fffcc000.spi", &spi1_clk),
+	CLKDEV_CON_DEV_ID("mci_clk", "fffa8000.mmc", &mmc_clk),
+	CLKDEV_CON_DEV_ID(NULL, "fffac000.i2c", &twi_clk),
+	CLKDEV_CON_DEV_ID(NULL, "fffff400.gpio", &pioA_clk),
+	CLKDEV_CON_DEV_ID(NULL, "fffff600.gpio", &pioB_clk),
+	CLKDEV_CON_DEV_ID(NULL, "fffff800.gpio", &pioC_clk),

do we really need this?

Yes, until all the SoCs are switch to the CCF, we may build kernel
supporting multiple sam9 SoCs without CCF.

Best regards,

-- 
Alexandre Belloni, Free Electrons
Embedded Linux, Kernel and Android engineering
http://free-electrons.com