ARM: OMAP: Add minimal OMAP2430 support

Tony/Richard,

* NFS is still not working. NETDEV Watchdog timeouts.
* Intel Starta flash gets detected - JFFS2 is not yet tested.
* platform_add_devices() is not working - Let me know, where I am making mistake :)

Please review.

Well one thing we should change is to make it possible to compile in all
24xx boards. This involves changing things like PRCM register access to use a base address instead of ifdefs.

I know we this is quite a bit of work now, but it pays off in the long
run. Something like this has proven to work well for portable code across various omaps/platforms:

#define OMAP2420_PRCM_SOME_BASE 0xaaaaaaaa
#define OMAP2430_PRCM_SOME_BASE 0xbbbbbbbb

#define PRCM_SOME_REG1_OFFSET 0x10
#define REG1_AAAA (1 << 31)
#define REG1_BBBB (1 << 30)
#define REG1_CCCC (1 << 29)

#define PRCM_SOME_REG2_OFFSET 0x20
#define REG2_AAAA (1 << 2)
#define REG2_BBBB (1 << 1)
#define REG2_CCCC (1 << 0)

#define PRCM_SOME_REG3_OFFSET 0x30

static unsigned long prcm_vbase;

#define prcm_readl(r) __raw_readl(prcm_vbase + (r))
#define prcm_writel(v, r) __raw_writel((v), prcm_vbase + (r))

static int prcm_do_something(void)

{
	u32 reg;

	reg = prcm_readl(PRCM_SOME_REG1_OFFSET);
	reg |= REG1_BBBB;
	prcm_writel(reg, PRCM_SOME_REG1_OFFSET);
}

static int __init prcm_init(void)

{
	if (cpu_is_omap2420())
		prcm_vbase = OMAP2420_PRCM_SOME_BASE;
	else if (cpu_is_omap2430())
		prcm_vbase = OMAP2430_PRCM_SOME_BASE;

	return 0;
}

Quoted 41 lines · --- Tony Lindgren

Well one thing we should change is to make it possible to compile in all
24xx boards. This involves changing things like PRCM register access to use a base address instead of ifdefs.

I know we this is quite a bit of work now, but it pays off in the long
run. Something like this has proven to work well for portable code across various omaps/platforms:

#define OMAP2420_PRCM_SOME_BASE 0xaaaaaaaa
#define OMAP2430_PRCM_SOME_BASE 0xbbbbbbbb

#define PRCM_SOME_REG1_OFFSET 0x10
#define REG1_AAAA (1 << 31)
#define REG1_BBBB (1 << 30)
#define REG1_CCCC (1 << 29)

#define PRCM_SOME_REG2_OFFSET 0x20
#define REG2_AAAA (1 << 2)
#define REG2_BBBB (1 << 1)
#define REG2_CCCC (1 << 0)

#define PRCM_SOME_REG3_OFFSET 0x30

static unsigned long prcm_vbase;

#define prcm_readl(r) __raw_readl(prcm_vbase + (r))
#define prcm_writel(v, r) __raw_writel((v), prcm_vbase + (r))

static int prcm_do_something(void)

{
	u32 reg;

	reg = prcm_readl(PRCM_SOME_REG1_OFFSET);
	reg |= REG1_BBBB;
	prcm_writel(reg, PRCM_SOME_REG1_OFFSET);
}

static int __init prcm_init(void)

Quoted 9 lines

{
	if (cpu_is_omap2420())
		prcm_vbase = OMAP2420_PRCM_SOME_BASE;
	else if (cpu_is_omap2430())
		prcm_vbase = OMAP2430_PRCM_SOME_BASE;

	return 0;
}

Thanx for the review. I will work on this, once I come back from Japan :). At home, both the 2430 boards are not-booting and I had tough time with CSST and switch settings, so was not able to flash bootloader on 2430SDP-EVM ES1.0 chips. But it will not be hard to make compile-only patch for review.

Quoted 4 lines · Komal Shah

Quoted 2 lines · --- Tony Lindgren

Well one thing we should change is to make it possible to compile

in

Quoted 2 lines

Quoted 2 lines

all
24xx boards. This involves changing things like PRCM register

access

Quoted 5 lines

Quoted 5 lines

to use a base address instead of ifdefs.

I know we this is quite a bit of work now, but it pays off in the long
run. Something like this has proven to work well for portable

code

Quoted 42 lines

Quoted 41 lines

across various omaps/platforms:

#define OMAP2420_PRCM_SOME_BASE 0xaaaaaaaa
#define OMAP2430_PRCM_SOME_BASE 0xbbbbbbbb

#define PRCM_SOME_REG1_OFFSET 0x10
#define REG1_AAAA (1 << 31)
#define REG1_BBBB (1 << 30)
#define REG1_CCCC (1 << 29)

#define PRCM_SOME_REG2_OFFSET 0x20
#define REG2_AAAA (1 << 2)
#define REG2_BBBB (1 << 1)
#define REG2_CCCC (1 << 0)

#define PRCM_SOME_REG3_OFFSET 0x30

static unsigned long prcm_vbase;

#define prcm_readl(r) __raw_readl(prcm_vbase + (r))
#define prcm_writel(v, r) __raw_writel((v), prcm_vbase + (r))

static int prcm_do_something(void)

{
	u32 reg;

	reg = prcm_readl(PRCM_SOME_REG1_OFFSET);
	reg |= REG1_BBBB;
	prcm_writel(reg, PRCM_SOME_REG1_OFFSET);
}

static int __init prcm_init(void)

{
	if (cpu_is_omap2420())
		prcm_vbase = OMAP2420_PRCM_SOME_BASE;
	else if (cpu_is_omap2430())
		prcm_vbase = OMAP2430_PRCM_SOME_BASE;

	return 0;
}

Quoted 1 line

Thanx for the review. I will work on this, once I come back from

Japan

Quoted 1 line

:). At home, both the 2430 boards are not-booting and I had tough

time

Quoted 1 line

with CSST and switch settings, so was not able to flash bootloader

on

Quoted 1 line

2430SDP-EVM ES1.0 chips. But it will not be hard to make

compile-only

Quoted 1 line

patch for review.

After looking at the files which are needed to change for prcm_read/write[l], clock.c needs the major update. I hope Richard is OK with going from __REG32(...) to __raw_read/write[l] based prcm_rw wrappers?