Re: [PATCH] ARM: at91: add Acme Systems Aria G25 board
From: Douglas Gilbert
Date: Mon Mar 25 2013 - 11:24:15 EST
On 13-03-25 10:31 AM, Jean-Christophe PLAGNIOL-VILLARD wrote:
On 09:48 Mon 25 Mar , Douglas Gilbert wrote:
On 13-03-25 08:22 AM, Jean-Christophe PLAGNIOL-VILLARD wrote:
On 09:49 Mon 25 Mar , Nicolas Ferre wrote:
From: Douglas Gilbert <dgilbert@xxxxxxxxxxxx>
Signed-off-by: Douglas Gilbert <dgilbert@xxxxxxxxxxxx>
Signed-off-by: Nicolas Ferre <nicolas.ferre@xxxxxxxxx>
---
arch/arm/boot/dts/ariag25.dts | 168 ++++++++++++++++++++++++++++++++++++++++++
1 file changed, 168 insertions(+)
create mode 100644 arch/arm/boot/dts/ariag25.dts
diff --git a/arch/arm/boot/dts/ariag25.dts b/arch/arm/boot/dts/ariag25.dts
new file mode 100644
index 0000000..d18ef50
--- /dev/null
+++ b/arch/arm/boot/dts/ariag25.dts
@@ -0,0 +1,168 @@
+/*
+ * ariag25.dts - Device Tree file for Acme Systems Aria G25 (AT91SAM9G25 based)
+ *
+ * Copyright (C) 2013 Douglas Gilbert <dgilbert@xxxxxxxxxxxx>,
+ * Robert Nelson <robertcnelson@xxxxxxxxx>
+ *
+ * Licensed under GPLv2 or later.
+ */
+/dts-v1/;
+/include/ "at91sam9g25.dtsi"
+
+/ {
+ model = "Acme Systems Aria G25";
+ compatible = "acme,ariag25", "atmel,at91sam9g25ek", "atmel,at91sam9x5ek",
+ "atmel,at91sam9x5", "atmel,at91sam9";
I doube the code is compatible with the 9g25ek
specially when you do not include it
+
+ aliases {
+ serial4 = &usart3;
+ serial5 = &uart0;
+ };
you need to specify all
+
+ chosen {
+ bootargs = "console=ttyS0,115200 root=/dev/mmcblk0p2 rw rootwait";
+ };
+
+ memory {
+ /* 128 MB, change this for 256 MB revision */
+ reg = <0x20000000 0x8000000>;
+ };
+
+ clocks {
+ #address-cells = <1>;
+ #size-cells = <1>;
+ ranges;
+
+ main_clock: clock@0 {
+ compatible = "atmel,osc", "fixed-clock";
+ clock-frequency = <12000000>;
+ };
+ };
+
+ ahb {
+ apb {
+ mmc0: mmc@f0008000 {
+ /* N.B. Aria has no SD card detect (CD), assumed present */
+
+ pinctrl-0 = <
+ &pinctrl_mmc0_slot0_clk_cmd_dat0
+ &pinctrl_mmc0_slot0_dat1_3>;
+ status = "okay";
+ slot@0 {
+ reg = <0>;
+ bus-width = <4>;
+ };
+ };
+
+ i2c0: i2c@f8010000 {
+ status = "okay";
+ };
+
+ i2c1: i2c@f8014000 {
+ status = "okay";
+ };
+
+ /* TWD2+TCLK2 hidden behind ethernet, so no i2c2 */
+
+ usart0: serial@f801c000 {
+ pinctrl-0 = <&pinctrl_usart0
+ &pinctrl_usart0_rts
+ &pinctrl_usart0_cts>;
+ status = "okay";
+ };
+
+ usart1: serial@f8020000 {
+ pinctrl-0 = <&pinctrl_usart1
+ /* &pinctrl_usart1_rts */
+ /* &pinctrl_usart1_cts */
+ >;
+ status = "okay";
+ };
+
+ usart2: serial@f8024000 {
+ /* cannot activate RTS2+CTS2, clash with
+ * ethernet on PB0 and PB1 */
+ pinctrl-0 = <&pinctrl_usart2>;
+ status = "okay";
+ };
+
+ usart3: serial@f8028000 {
+ compatible = "atmel,at91sam9260-usart";
+ reg = <0xf8028000 0x200>;
+ interrupts = <8 4 5>;
+ pinctrl-names = "default";
+ pinctrl-0 = <&pinctrl_usart3
+ /* &pinctrl_usart3_rts */
+ /* &pinctrl_usart3_cts */
+ >;
+ status = "okay";
+ };
+
+ macb0: ethernet@f802c000 {
+ phy-mode = "rmii";
+ /* following can be overwritten by uboot 'ftd set' command */
+ local-mac-address = [00 04 25 dd 10 01];
drop this, this is board specific
+ status = "okay";
Also the local-mac-address line is needed since it is edited
by u-boot via this line:
mod_dtb=fdt addr 0x27FF0000 ; fdt set ${eth0_dt_path}
local-mac-address [ ${my_ethaddr} ]
Interesting technique that: editing the dtb from u-boot prior to the
Linux kernel using it ...
So the u-boot environment file actually sets the MAC address. The Aria
G25 has no NAND, flash or switches for setting a quasi unique MAC
address. There seems to be no way to pass the MAC address on the
kernel boot line (there was in the past). That leaves the microSD card
and IMO u-boot is the best place to set that MAC address. My guess
is that the Aria G25 is not the only board that has this "setting the
MAC address" challenge.
Finally of course that line is "board specific", that is why the
file is called ariag25.dts :-)
The mac address is board (hw) uniq so no default in the dts, if none present
the kernel will generate one (private class).
Since when have MAC addresses been hw uniq? My guess is
a huge majority of MAC addresses are set in firmware.
Since many embedded systems with wired ethernet face
this problem, is it the job of Linux to make this more
difficult than it already is?
If the mac is specifued in the register the kernel re-use it.
Unless u-boot (or whatever) edits the dtb file before the
Linux kernel sees it.
This is the job of the booloader to set one. The dts in the kernel will be
used for every Aria not one board in particular
It is the job of someone to set it. A MAC address can be
compiled into u-boot but that is about as useful as
compiling it into Linux kernel.
if you keep the mac addres in the dts NACK
Seems strange to object to setting
local-mac-address = [00 00 00 00 00 00];
when if the kernel sees that line it will do exactly the
same thing as without that line? But the advantage of that
line is that it gives the u-boot environment file the
ability to edit the ariag25.dtb before the kernel sees it.
And if the u-boot environment file is properly configured
by the board manufacturer, then a quasi uniq, stable MAC
address is generated.
For more information about this technique see:
http://www.petalogix.com/support/kb/mac_addr_update
Doug Gilbert
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/