Fix compile error for at91rm9200 in latest git

Hi,

at91rm9200ek doesn't compile in latest git:

/here/workdir/linux-2.6/at91rm9200ek/src/arch/arm/mach-at91/board-ek.c: In function `ek_board_init':
/here/workdir/linux-2.6/at91rm9200ek/src/arch/arm/mach-at91/board-ek.c:148: error: `ek_i2c_devices' undeclared (first use in this function) /here/workdir/linux-2.6/at91rm9200ek/src/arch/arm/mach-at91/board-ek.c:148: error: (Each undeclared identifier is reported only once /here/workdir/linux-2.6/at91rm9200ek/src/arch/arm/mach-at91/board-ek.c:148: error: for each function it appears in.) /here/workdir/linux-2.6/at91rm9200ek/src/arch/arm/mach-at91/board-ek.c:148: warning: type defaults to int' in declaration of type name' /here/workdir/linux-2.6/at91rm9200ek/src/arch/arm/mach-at91/board-ek.c:148: warning: type defaults to int' in declaration of type name' /here/workdir/linux-2.6/at91rm9200ek/src/arch/arm/mach-at91/board-ek.c:148: error: size of array `type name' is negative

The offending commit is:

[jan@bender linux-2.6]$ git-bisect good f230d3f53d72d05bcb5666ab7e2eccd49c8b3a15 is first bad commit commit f230d3f53d72d05bcb5666ab7e2eccd49c8b3a15

Author: Andrew Victor <[email protected]>
Date:   Mon Nov 19 13:47:20 2007 +0100

[ARM] 4650/1: AT91: New-style init of I2C, support for i2c-gpio

The AT91 I2C driver is currently marked as "broken" due to hardware issues. This patch enables AT91-based platforms to also use the bitbanged GPIO for I2C.

This updates platform setup logic (setting up an i2c-gpio device using the same pins as the i2c-at91 device, unless only the BROKEN driver is enabled).

Also make use of the new-style initialization of I2C devices using i2c_register_board_info().

This patch adds the missing i2c_board_info struct (grabbed from Andrew Victor's tree).

---

index d05b1b2..53a5ef9 100644
@@ -109,6 +109,15 @@ static struct spi_board_info ek_spi_devices[] = {
 #endif
 };
 
+static struct i2c_board_info __initdata ek_i2c_devices[] = {
+	{
+		I2C_BOARD_INFO("ics1523", 0x26),
+	},
+	{
+		I2C_BOARD_INFO("dac3550", 0x4d),
+	}
+};
+
 #define EK_FLASH_BASE	AT91_CHIPSELECT_0
 #define EK_FLASH_SIZE	0x200000
 

Which is a combination of two patches, which seems to have dropped the chunk you show below. The reason this seems to have my signoff is that I originated one of those patches, and parts of the second, but in fact I had never seen f230d3f53d72d05bcb5666ab7e2eccd49c8b3a15 until after this email ...

Your patch looks correct, and seems to be the only obvious chunk that's missing. So, I'll ack it FWIW ... usual policy for these patches is to go through Russell.

I'll note that the whole "mark Atmel's TWI controller BROKEN" patchh series, making everything use the i2c-gpio instead, has turned up one more issue at the I2C level: the i2c_adapter.class for i2c-gpio needs to include at least the HWMON class bit.

- Dave

hi,

I guess in my rush to get those patches submitted before leaving SAN People I missed that chunk. Sorry.

David: The Signed-off line was kept as-is to keep the attribution correct.

You can add my Ack for what it's worth.

Regards,
Andrew Victor

Hi all,

OK, CC'ed Russell and added your Acked-by.

---

index d05b1b2..53a5ef9 100644
@@ -109,6 +109,15 @@ static struct spi_board_info ek_spi_devices[] = {
 #endif
 };
 
+static struct i2c_board_info __initdata ek_i2c_devices[] = {
+	{
+		I2C_BOARD_INFO("ics1523", 0x26),
+	},
+	{
+		I2C_BOARD_INFO("dac3550", 0x4d),
+	}
+};
+
 #define EK_FLASH_BASE	AT91_CHIPSELECT_0
 #define EK_FLASH_SIZE	0x200000
 

-> patch system please.