Patchwork IXP4xx: Add Gateworks Cambria support.

login
register
mail settings
Submitter Krzysztof =?utf-8?Q?Ha=C5=82asa?=
Date Sept. 18, 2013, 7:13 a.m.
Message ID <m3vc1ya9fw.fsf@t19.piap.pl>
Download mbox | patch
Permalink /patch/275610/
State New
Headers show

Comments

Krzysztof =?utf-8?Q?Ha=C5=82asa?= - Sept. 18, 2013, 7:13 a.m.
Signed-off-by: Krzysztof Hałasa <khalasa@piap.pl>
Luka Perkov - Sept. 19, 2013, 4:03 p.m.
Hi Krzysztof,

it's great that you have decided to push upstream support for this
board.

I was unable to bring up serial console with your patch. I was using
latest OpenWrt trunk for rootfs, did you use the same?

drivers/rtc/hctosys.c: unable to open rtc device (rtc0)
IP-Config: Failed to open eth0
IP-Config: Failed to open eth1
IP-Config: No network devices available
Warning: unable to open an initial console.
Freeing unused kernel memory: 4124K (c0427000 - c082e000)
procd: Console is alive

On Wed, Sep 18, 2013 at 09:13:55AM +0200, Krzysztof Hałasa wrote:
> --- /dev/null
> +++ b/arch/arm/mach-ixp4xx/cambria.c
> @@ -0,0 +1,245 @@
> +/*
> + * Gateworks Cambria GW2350 platform support
> + * Portions based on OpenWRT.
> + */

I'm no expert on licenses but shouldn't here at least be GPLv2 since the
file originated from OpenWrt support for this board which is GPLv2.

> +static struct i2c_board_info __initdata cambria_i2c_board_info[] = {
> +	{
> +		I2C_BOARD_INFO("ad7418", 0x28),
> +	},
> +	{
> +		I2C_BOARD_INFO("24c08", 0x50),
> +		.platform_data	= &cambria_eeprom_info,
> +	},
> +};

I couldn't test this because of the console issue but in OpenWrt sources
the value is 0x51 and not 0x50. Did you change it by accident?

Also, shouldn't arch/arm/configs/ixp4xx_defconfig be updated too?

Luka
Krzysztof Halasa - Sept. 19, 2013, 6:27 p.m.
Hi Luka,

Luka Perkov <luka@openwrt.org> writes:

> I was unable to bring up serial console with your patch. I was using
> latest OpenWrt trunk for rootfs, did you use the same?

No, I use a different, custom initramfs.

> drivers/rtc/hctosys.c: unable to open rtc device (rtc0)
> IP-Config: Failed to open eth0
> IP-Config: Failed to open eth1
> IP-Config: No network devices available
> Warning: unable to open an initial console.

Perhaps it wants the network console instead?
Are you using "console=ttyS0,115200" or something alike?
Or a /dev filesystem problems maybe?

> I'm no expert on licenses but shouldn't here at least be GPLv2 since the
> file originated from OpenWrt support for this board which is GPLv2.

Of course the license is GPLv2, every Linux kernel file (except maybe
some firmware and non-copyrightable code) is so licensed. See the file
"COPYING"

>> +static struct i2c_board_info __initdata cambria_i2c_board_info[] = {
>> +	{
>> +		I2C_BOARD_INFO("ad7418", 0x28),
>> +	},
>> +	{
>> +		I2C_BOARD_INFO("24c08", 0x50),
>> +		.platform_data	= &cambria_eeprom_info,
>> +	},
>> +};
>
> I couldn't test this because of the console issue but in OpenWrt sources
> the value is 0x51 and not 0x50. Did you change it by accident?

Not really, the board simply has 24C08 at 0x50. OpenWRT does a trick with
offsets here, you can't really have a 24C08 at 0x51 (possible addresses
for most chips are 0x50 and 0x54). I just did it the straight way and it
works fine.

> Also, shouldn't arch/arm/configs/ixp4xx_defconfig be updated too?

Well, probably.
Imre Kaloz - Sept. 19, 2013, 8:52 p.m.
On Wed, 18 Sep 2013 09:13:55 +0200, Krzysztof Hałasa <khalasa@piap.pl> wrote:

> Signed-off-by: Krzysztof Hałasa <khalasa@piap.pl>

<snip>

This file seems to miss both the copyright from the code it's based on as well as most of the additional support that one has.
So what's the point?


Regards,


Imre
Krzysztof =?utf-8?Q?Ha=C5=82asa?= - Sept. 20, 2013, 8:40 a.m.
"Imre Kaloz" <kaloz@openwrt.org> writes:

> This file seems to miss both the copyright from the code it's based on
> as well as most of the additional support that one has.
> So what's the point?

I have the smaller Cambria (GW2350), not the bigger one which is also
supported by OpenWRT. I'm not going to blindly pick up the OpenWRT code
for which I don't have the hw and which I can't really test nor
maintain.
The file contains information that it's partially based on OpenWRT, is it
now a bad idea to pick up fragments from one GPLv2 project (OpenWRT) and
include them in another GPLv2 project (Linux kernel)? Isn't it what
OpenWRT does (in the opposite direction)? Aren't all the boards (in this
particular case) based on Avila and IXDP425, the devel platform?

The code doesn't carry any copyright headers maybe because in my country
they aren't required, maybe because it's a trivial board support code
which is mostly a list of devices and I didn't bother, and it doesn't
have OpenWRT copyright because it's not a version from OpenWRT (while
its parts are, as prominently stated). For precisely the same reason I
list myself as the platform maintainer, I don't feel authorized to put
somebody else here.

But I have an idea. I would be pleased if you submit and maintain (in
the official Linux, not OpenWRT) your own version instead. It would be
great if you also include Ubiquity RouterStation Pro (ATH79) and
Gateworks Laguna (CNS3xxx). Those boards are supported for a long time
by OpenWRT (at least there is support code there) and I'm personally
using them with (stock) Linux.
Imre Kaloz - Sept. 20, 2013, 9:13 a.m.
On Fri, 20 Sep 2013 10:40:10 +0200, Krzysztof Hałasa <khalasa@piap.pl> wrote:

> "Imre Kaloz" <kaloz@openwrt.org> writes:
>
>> This file seems to miss both the copyright from the code it's based on
>> as well as most of the additional support that one has.
>> So what's the point?
>
> I have the smaller Cambria (GW2350), not the bigger one which is also
> supported by OpenWRT. I'm not going to blindly pick up the OpenWRT code
> for which I don't have the hw and which I can't really test nor
> maintain.

So you delete portions of the code because you don't want to ask someone with the board to test your changes if needed, but given quite some people use the code for quite some time, you can also assume it works for them.

> The file contains information that it's partially based on OpenWRT, is it
> now a bad idea to pick up fragments from one GPLv2 project (OpenWRT) and
> include them in another GPLv2 project (Linux kernel)? Isn't it what
> OpenWRT does (in the opposite direction)? Aren't all the boards (in this
> particular case) based on Avila and IXDP425, the devel platform?

OpenWrt (yeah, we spell it this way for about 10 years now) is a Linux distribution, you do mix things up a bit here.

> The code doesn't carry any copyright headers maybe because in my country
> they aren't required, maybe because it's a trivial board support code
> which is mostly a list of devices and I didn't bother, and it doesn't
> have OpenWRT copyright because it's not a version from OpenWRT (while
> its parts are, as prominently stated). For precisely the same reason I
> list myself as the platform maintainer, I don't feel authorized to put
> somebody else here.

It's up to you if you add your copyright, but removing others' is clearly a different story legally. And I bet that's the same in your country, too.

> But I have an idea. I would be pleased if you submit and maintain (in
> the official Linux, not OpenWRT) your own version instead. It would be
> great if you also include Ubiquity RouterStation Pro (ATH79) and
> Gateworks Laguna (CNS3xxx). Those boards are supported for a long time
> by OpenWRT (at least there is support code there) and I'm personally
> using them with (stock) Linux.

I'll do when I'll get around it. Both others and we've pushed a lot of code upstream. In the meantime, anyone is free to do so, including you - dumbing down the code could be justifiable in some cases (I don't think this one applies), but dropping others' copyright -- hardly.


Imre
Krzysztof =?utf-8?Q?Ha=C5=82asa?= - Sept. 25, 2013, 7:24 a.m.
"Imre Kaloz" <kaloz@openwrt.org> writes:

> So you delete portions of the code because you don't want to ask
> someone with the board to test your changes if needed, but given quite
> some people use the code for quite some time, you can also assume it
> works for them.

No, I didn't delete any code at all. Look at the patch again, it's only
adding stuff.
I have picked up parts of your code (parts of another project). GPL
doesn't allow a requirement that I have to take the whole.

> OpenWrt (yeah, we spell it this way for about 10 years now)

Sorry, didn't know, I don't use it.

> is a Linux distribution, you do mix things up a bit here.

Well, at least Cambria code is not a part of the official Linux then.

> It's up to you if you add your copyright, but removing others' is
> clearly a different story legally. And I bet that's the same in your
> country, too.

You are making things up. I didn't remove anything. I've created
a derivative work of (among others) your project. The copyright for the
base work doesn't apply for the derivative work, the creator of the
derivative work formally holds the copyright. The GPL explicitly allows
creating derivative works - actually, the whole point about GPL is
encouraging derivative works. OpenWrt is based on other works in
precisely the same way.

I didn't include my copyright because I didn't care a bit. The code in
question is mostly a trivial list of devices, and I guess in most parts
it's not even copyrightable. Remember, copyright is about the form, and
there aren't two ways to express e.g. that it has a serial port
(identical to other IXP4xx platforms), that it has a NOR flash
(identical to others) and so on.

> I'll do when I'll get around it. Both others and we've pushed a lot of
> code upstream. In the meantime, anyone is free to do so, including you
> - dumbing down the code could be justifiable in some cases (I don't
> think this one applies), but dropping others' copyright -- hardly.

Again, creating derivative works by picking up needed parts is not
dropping or dumbing anything. You are trying to force your view of what
is justifiable and what does apply. There is absolutely no legal or
other basis for this.

I have to say I'm disappointed. I didn't expect such reaction from you.
Actually I thought I was helping your project a bit (and my project, of
course). Since the former is not the case, I'm simply withdrawing my
patch.

Patch

diff --git a/arch/arm/mach-ixp4xx/Kconfig b/arch/arm/mach-ixp4xx/Kconfig
index 676e144..70c5cb9 100644
--- a/arch/arm/mach-ixp4xx/Kconfig
+++ b/arch/arm/mach-ixp4xx/Kconfig
@@ -25,6 +25,14 @@  config MACH_AVILA
 	  Avila Network Platform. For more information on this platform,
 	  see <file:Documentation/arm/IXP4xx>.
 
+config MACH_CAMBRIA
+	bool "Cambria"
+	select PCI
+	help
+	  Say 'Y' here if you want your kernel to support the Gateworks
+	  Cambria GW2350 board. For more information on this platform,
+	  see <file:Documentation/arm/IXP4xx>.
+
 config MACH_LOFT
     bool "Loft"
     depends on MACH_AVILA
@@ -158,7 +166,7 @@  config CPU_IXP46X
 
 config CPU_IXP43X
 	bool
-	depends on MACH_KIXRP435
+	depends on MACH_KIXRP435 || MACH_CAMBRIA
 	default y
 
 config MACH_GTWX5715
diff --git a/arch/arm/mach-ixp4xx/Makefile b/arch/arm/mach-ixp4xx/Makefile
index eded94c..a7aa6a5 100644
--- a/arch/arm/mach-ixp4xx/Makefile
+++ b/arch/arm/mach-ixp4xx/Makefile
@@ -23,6 +23,7 @@  obj-y	+= common.o
 
 obj-$(CONFIG_ARCH_IXDP4XX)	+= ixdp425-setup.o
 obj-$(CONFIG_MACH_AVILA)	+= avila-setup.o
+obj-$(CONFIG_MACH_CAMBRIA)	+= cambria.o
 obj-$(CONFIG_MACH_IXDPG425)	+= coyote-setup.o
 obj-$(CONFIG_ARCH_ADI_COYOTE)	+= coyote-setup.o
 obj-$(CONFIG_MACH_GTWX5715)	+= gtwx5715-setup.o
diff --git a/arch/arm/mach-ixp4xx/cambria.c b/arch/arm/mach-ixp4xx/cambria.c
new file mode 100644
index 0000000..30ddfe0
--- /dev/null
+++ b/arch/arm/mach-ixp4xx/cambria.c
@@ -0,0 +1,245 @@ 
+/*
+ * Gateworks Cambria GW2350 platform support
+ * Portions based on OpenWRT.
+ */
+
+#include <linux/device.h>
+#include <linux/i2c.h>
+#include <linux/i2c-gpio.h>
+#include <linux/i2c/at24.h>
+#include <linux/if_ether.h>
+#include <linux/init.h>
+#include <linux/irq.h>
+#include <linux/kernel.h>
+#include <linux/leds.h>
+#include <linux/pci.h>
+#include <linux/serial_8250.h>
+#include <linux/types.h>
+#include <linux/usb/ehci_pdriver.h>
+#include <asm/mach-types.h>
+#include <asm/mach/arch.h>
+#include <asm/mach/flash.h>
+
+static struct platform_device cambria_flash = {
+	.name			= "IXP4XX-Flash",
+	.id			= 0,
+	.num_resources		= 1,
+	.resource		= &(struct resource) {
+		.start		= IXP4XX_EXP_BUS_BASE(0),
+		.end		= IXP4XX_EXP_BUS_BASE(0) + SZ_32M - 1,
+		.flags		= IORESOURCE_MEM,
+	},
+	.dev.platform_data	= &(struct flash_platform_data) {
+		.map_name	= "cfi_probe",
+		.width		= 2,
+	},
+};
+
+static struct platform_device cambria_i2c = {
+	.name			= "i2c-gpio",
+	.id			= 0,
+	.dev.platform_data	= &(struct i2c_gpio_platform_data) {
+		.sda_pin	= 7,
+		.scl_pin	= 6,
+	},
+};
+
+static struct eth_plat_info cambria_eth0_pdata = {
+	.phy			= 1,
+	.rxq			= 4,
+	.txreadyq		= 21,
+};
+
+static struct eth_plat_info cambria_eth1_pdata = {
+	.phy			= 2,
+	.rxq			= 2,
+	.txreadyq		= 19,
+};
+
+static struct platform_device cambria_eth0 = {
+	.name			= "ixp4xx_eth",
+	.id			= IXP4XX_ETH_NPEC,
+	.dev.platform_data	= &cambria_eth0_pdata,
+};
+
+static struct platform_device cambria_eth1 = {
+	.name			= "ixp4xx_eth",
+	.id			= IXP4XX_ETH_NPEA,
+	.dev.platform_data	= &cambria_eth1_pdata,
+};
+
+static struct platform_device cambria_uart = {
+	.name			= "serial8250",
+	.id			= PLAT8250_DEV_PLATFORM,
+	.num_resources		= 1,
+	.resource		= &(struct resource) {
+		.start		= IXP4XX_UART1_BASE_PHYS,
+		.end		= IXP4XX_UART1_BASE_PHYS + 0x0fff,
+		.flags		= IORESOURCE_MEM,
+	},
+	.dev.platform_data	= (struct plat_serial8250_port[]) {
+		{
+			.mapbase	= IXP4XX_UART1_BASE_PHYS,
+			.membase	= (char __iomem *)IXP4XX_UART1_BASE_VIRT + REG_OFFSET,
+			.irq		= IRQ_IXP4XX_UART1,
+			.flags		= UPF_BOOT_AUTOCONF | UPF_SKIP_TEST,
+			.iotype		= UPIO_MEM,
+			.regshift	= 2,
+			.uartclk	= IXP4XX_UART_XTAL,
+		},
+		{ },
+	}
+};
+
+static struct platform_device cambria_leds = {
+	.name		= "leds-gpio",
+	.id		= -1,
+	.dev.platform_data = &(struct gpio_led_platform_data) {
+		.num_leds	= 1,
+		.leds		= &(struct gpio_led) {
+			.name		= "user",
+			.gpio		= 5,
+			.active_low	= 1,
+		},
+	},
+};
+
+static struct usb_ehci_pdata cambria_usb_pdata = {
+	.caps_offset	= 0x100,
+	.has_tt		= 1,
+	.big_endian_desc = 1,
+	.big_endian_mmio = 1,
+};
+
+static struct platform_device cambria_usb0 = {
+	.name		= "ehci-platform",
+	.id		= 0,
+	.resource	= (struct resource[]) {
+		{
+			.start	= IXP4XX_EHCI1_BASE_PHYS,
+			.end	= IXP4XX_EHCI1_BASE_PHYS + IXP4XX_EHCI_REGION_SIZE - 1,
+			.flags	= IORESOURCE_MEM,
+		},
+		{
+			.start	= 32,
+			.flags	= IORESOURCE_IRQ,
+		},
+	},
+	.num_resources	= 2,
+	.dev.platform_data = &cambria_usb_pdata,
+};
+
+static struct platform_device cambria_usb1 = {
+	.name		= "ehci-platform",
+	.id		= 1,
+	.resource	= (struct resource[]) {
+		{
+			.start	= IXP4XX_EHCI2_BASE_PHYS,
+			.end	= IXP4XX_EHCI2_BASE_PHYS + IXP4XX_EHCI_REGION_SIZE - 1,
+			.flags	= IORESOURCE_MEM,
+		},
+		{
+			.start	= 33,
+			.flags	= IORESOURCE_IRQ,
+		},
+	},
+	.num_resources	= 2,
+	.dev.platform_data = &cambria_usb_pdata,
+};
+
+static struct platform_device *cambria_devices[] __initdata = {
+	&cambria_flash,
+	&cambria_i2c,
+	&cambria_eth0,
+	&cambria_eth1,
+	&cambria_uart,
+	&cambria_leds,
+	&cambria_usb0,
+	&cambria_usb1,
+};
+
+static void at24_setup(struct memory_accessor *mem_acc, void *context)
+{
+	char mac_addr[ETH_ALEN];
+
+	/* Read MAC addresses */
+	if (mem_acc->read(mem_acc, mac_addr, 0x100, ETH_ALEN) == ETH_ALEN)
+		memcpy(&cambria_eth0_pdata.hwaddr, mac_addr, ETH_ALEN);
+
+	if (mem_acc->read(mem_acc, mac_addr, 0x106, ETH_ALEN) == ETH_ALEN)
+		memcpy(&cambria_eth1_pdata.hwaddr, mac_addr, ETH_ALEN);
+}
+
+static struct at24_platform_data cambria_eeprom_info = {
+	.byte_len	= 1024,
+	.page_size	= 16,
+	.flags		= AT24_FLAG_READONLY,
+	.setup		= at24_setup,
+};
+
+static struct i2c_board_info __initdata cambria_i2c_board_info[] = {
+	{
+		I2C_BOARD_INFO("ad7418", 0x28),
+	},
+	{
+		I2C_BOARD_INFO("24c08", 0x50),
+		.platform_data	= &cambria_eeprom_info,
+	},
+};
+
+static void __init cambria_init(void)
+{
+	ixp4xx_sys_init();
+	platform_add_devices(cambria_devices, ARRAY_SIZE(cambria_devices));
+	i2c_register_board_info(0, cambria_i2c_board_info,
+				ARRAY_SIZE(cambria_i2c_board_info));
+}
+
+#ifdef CONFIG_PCI
+void __init cambria_pci_preinit(void)
+{
+	irq_set_irq_type(IRQ_IXP4XX_GPIO11, IRQ_TYPE_LEVEL_LOW);
+	ixp4xx_pci_preinit();
+}
+
+static int __init cambria_map_irq(const struct pci_dev *dev, u8 slot, u8 pin)
+{
+	switch (slot) {
+	case 1:
+		return IRQ_IXP4XX_GPIO11;
+	default:
+		return -1;
+	}
+}
+
+struct hw_pci cambria_pci __initdata = {
+	.nr_controllers	= 1,
+	.ops		= &ixp4xx_ops,
+	.preinit	= cambria_pci_preinit,
+	.setup		= ixp4xx_setup,
+	.map_irq	= cambria_map_irq,
+};
+
+int __init cambria_pci_init(void)
+{
+	if (machine_is_cambria())
+		pci_common_init(&cambria_pci);
+	return 0;
+}
+
+subsys_initcall(cambria_pci_init);
+#endif /* CONFIG_PCI */
+
+MACHINE_START(CAMBRIA, "Gateworks Cambria")
+	/* Maintainer: Krzysztof Halasa */
+	.map_io		= ixp4xx_map_io,
+	.init_early	= ixp4xx_init_early,
+	.init_irq	= ixp4xx_init_irq,
+	.init_time	= ixp4xx_timer_init,
+	.atag_offset	= 0x0100,
+	.init_machine	= cambria_init,
+#ifdef CONFIG_PCI
+	.dma_zone_size	= SZ_64M,
+#endif
+	.restart	= ixp4xx_restart,
+MACHINE_END
diff --git a/arch/arm/mach-ixp4xx/include/mach/ixp4xx-regs.h b/arch/arm/mach-ixp4xx/include/mach/ixp4xx-regs.h
index c5bae9c..f6ced67 100644
--- a/arch/arm/mach-ixp4xx/include/mach/ixp4xx-regs.h
+++ b/arch/arm/mach-ixp4xx/include/mach/ixp4xx-regs.h
@@ -595,6 +595,13 @@ 
 
 #define DCMD_LENGTH	0x01fff		/* length mask (max = 8K - 1) */
 
+/*
+ * EHCI USB host registers
+ */
+#define IXP4XX_EHCI1_BASE_PHYS		0xCD000000
+#define IXP4XX_EHCI2_BASE_PHYS		0xCE000000
+#define IXP4XX_EHCI_REGION_SIZE		0x00000200
+
 /* "fuse" bits of IXP_EXP_CFG2 */
 /* All IXP4xx CPUs */
 #define IXP4XX_FEATURE_RCOMP		(1 << 0)
diff --git a/arch/arm/tools/mach-types b/arch/arm/tools/mach-types
index a10297d..9ada1a7 100644
--- a/arch/arm/tools/mach-types
+++ b/arch/arm/tools/mach-types
@@ -261,6 +261,7 @@  halibut			MACH_HALIBUT		HALIBUT			1439
 trout			MACH_TROUT		TROUT			1440
 tct_hammer		MACH_TCT_HAMMER		TCT_HAMMER		1460
 herald			MACH_HERALD		HERALD			1461
+cambria			MACH_CAMBRIA		CAMBRIA			1468
 sim_one			MACH_SIM_ONE		SIM_ONE			1476
 jive			MACH_JIVE		JIVE			1490
 sam9_l9260		MACH_SAM9_L9260		SAM9_L9260		1501