Factorize ARM startup code as mush as possible.

Signed-off-by Albert ARIBAUD <[email protected]>

---

Dear Albert ARIBAUD,

Quoted 2 lines · In message <[email protected]> you…

Signed-off-by Albert ARIBAUD <[email protected]>

---

Please move the patch description from the cover letter to this commit message.

Best regards,

Wolfgang Denk

the code sequence from isetting up the initial stack

Move all the C runtime setup code from every start.S in arch/arm into arch/arm/lib/crt0.S. This covers the code sequence from isetting up the initial stack to calling into board_init_r().

Also, rewrite the C runtime setup and make functions board_init_*() and relocate_code() behave according to normal C semantics (no jumping across the C stack any more, etc).

Some SPL targets had to be touched because they use start.S exolicitly or for some reason; the relevant maintainers and custodians are cc:ed.

Signed-off-by Albert ARIBAUD<[email protected]>

---
Changes in v2:
- moved description from cover letter to patch commit msg
- added note about tests in the cover letter
- fixed baords with CONFIG_SPL but not CONFIG_SPL_STACK

arch/arm/lib/crt0.S new file +53 -0

new file mode 100644
index 0000000..fd6bd92
@@ -0,0 +1,180 @@
+/*
+ *  crt0 - C-runtime startup Code for ARM U-Boot
+ *
+ *  Copyright (c) 2012  Albert ARIBAUD<[email protected]>
+ *
+ * See file CREDITS for list of people who contributed to this
+ * project.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation; either version 2 of
+ * the License, or (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston,
+ * MA 02111-1307 USA
+ */
+
+#include<config.h>
+#include<asm-offsets.h>
+
+/*
+ * This file handles the target-independent stages of the U-Boot
+ * start-up where a C runtime environment is needed. Its entry point
+ * is _main and is branched into from the target's start.S file.
+ *
+ * _main execution sequence is:
+ *
+ * 1. Set up initial environment for calling board_init_f().
+ *    This environment only provides a stack and a place to store
+ *    the GD ('global data') structure. In this context, VARIABLE
+ *    global data, initialized or not (BSS), are UNAVAILABLE; only
+ *    CONSTANT initialized data are available.
+ *
+ * 2. Call board_init_f(). This function prepares the hardware for
+ *    execution from DDR. As DDR may not be available, board_init_f()
+ *    must use GD to store any data which must be passed on to later,
+ *    stages, including the reloction destination and the new stack
+ *    pointer address, below which the stack resides and above it the
+ *    new GD resides.
+ *
+ * 3. Set up intermediate environment where the stack and GD are the
+ *    ones allocated by board_init_f() in DDR, but BSS and initialized
+ *    non-const data are still not available.
+ *
+ * 4. Call relocate_code(). This function relocates U-Boot from its
+ *    current location into the relocation destination computed by

arch/arm/lib/crt0.S new file +56 -0

+
+/*
+ * offset of the nand_boot() function for SPL crt
+ */
+
+#if defined(CONFIG_NAND_SPL)
+
+.globl nand_boot
+_nand_boot:
+	.word nand_boot
+
+#elif ! defined(CONFIG_SPL_BUILD)
+
+/*
+ * offset of the board_init_r() function for non-SPL crt
+ */
+
+.globl board_init_r
+_board_init_r:
+	.word board_init_r
+
+#endif
+
+/*
+ * start and end of BSS
+ */
+
+.globl __bss_start
+.globl __bss_end__
+
+/*
+ * entry point of crt0 sequence
+ */
+
+.global _main
+
+_main:
+
+/*
+ * Set up initial C runtime environment and call board_init_f(0).
+ */
+
+#if defined(CONFIG_NAND_SPL)
+	/* deprecated, use instead CONFIG_SPL_BUILD */
+	ldr	sp, =(CONFIG_SYS_INIT_SP_ADDR)
+#elif defined(CONFIG_SPL_BUILD)&&  defined(CONFIG_SPL_STACK)
+	ldr	sp, =(CONFIG_SPL_STACK)
+#else
+	ldr	sp, =(CONFIG_SYS_INIT_SP_ADDR)
+#endif
+	bic	sp, sp, #7 /* 8-byte alignment for ABI compliance */
+	mov	r8, sp		/* GD is above SP */
+	mov	r0, #0
+	bl	board_init_f
+
+/*

arch/arm/lib/crt0.S new file +8 -0

+
+	adr	lr, here
+	ldr	r0, [r8, #GD_RELOC_OFF]		/* lr = gd->start_addr_sp */
+	add	lr, lr, r0
+	ldr	r0, [r8, #GD_START_ADDR_SP]	/* r0 = gd->start_addr_sp */
+	mov	r1, r8				/* r1 = gd */
+	ldr	r2, [r8, #GD_RELOCADDR]		/* r2 = gd->relocaddr */
+	b	relocate_code

Hello Albert,

arch/arm/lib/crt0.S

Quoted 3 lines

arch/arm/lib/crt0.S +1 -0

 
+#ifndef CONFIG_SPL_BUILD

In some places the other style is used. "!(defined)". Any particular reasons for switching b/w these two styles?

arch/arm/lib/crt0.S line 147

Quoted 2 lines

arch/arm/lib/crt0.S +2 -0

+	ldr	r0, =__bss_start	/* this is auto-relocated! */
+	ldr	r1, =__bss_end__	/* this is aotu-relocated! */

s/aotu/auto

Regards,
Vikram

Move all the C runtime setup code from every start.S in arch/arm into arch/arm/lib/crt0.S. This covers the code sequence from isetting up the initial stack to calling into board_init_r().

Also, rewrite the C runtime setup and make functions board_init_*() and relocate_code() behave according to normal C semantics (no jumping across the C stack any more, etc).

Some SPL targets had to be touched because they use start.S exolicitly or for some reason; the relevant maintainers and custodians are cc:ed.

Signed-off-by Albert ARIBAUD <[email protected]>

---
Changes in v2:
- moved description from cover letter to patch commit msg
- added note about tests in the cover letter
- fixed baords with CONFIG_SPL but not CONFIG_SPL_STACK

 arch/arm/cpu/arm1136/start.S              |   62 ++--------
 arch/arm/cpu/arm1176/start.S              |   62 ++--------
 arch/arm/cpu/arm720t/start.S              |   53 ++-------
 arch/arm/cpu/arm920t/start.S              |   61 ++--------
 arch/arm/cpu/arm925t/start.S              |   61 ++--------
 arch/arm/cpu/arm926ejs/start.S            |   80 ++-----------
 arch/arm/cpu/arm946es/start.S             |   56 ++-------
 arch/arm/cpu/arm_intcm/start.S            |   63 ++--------
 arch/arm/cpu/armv7/start.S                |   58 +++-------
 arch/arm/cpu/ixp/start.S                  |   55 ++-------
 arch/arm/cpu/pxa/start.S                  |   63 ++--------
 arch/arm/cpu/s3c44b0/start.S              |   55 ++-------
 arch/arm/cpu/sa1100/start.S               |   50 ++------
 arch/arm/lib/Makefile                     |    2 +
 arch/arm/lib/board.c                      |   11 --
 arch/arm/lib/crt0.S                       |  180 +++++++++++++++++++++++++++++
 include/common.h                          |    2 +-
 include/configs/socfpga_cyclone5.h        |    2 +-
 lib/asm-offsets.c                         |   10 ++
 nand_spl/board/freescale/mx31pdk/Makefile |    6 +-
 nand_spl/board/karo/tx25/Makefile         |    6 +-
 21 files changed, 324 insertions(+), 674 deletions(-)
 create mode 100644 arch/arm/lib/crt0.S

arch/arm/lib/crt0.S new file +32 -0

new file mode 100644
index 0000000..fd6bd92
@@ -0,0 +1,180 @@
+/*
+ *  crt0 - C-runtime startup Code for ARM U-Boot
+ *
+ *  Copyright (c) 2012  Albert ARIBAUD <[email protected]>
+ *
+ * See file CREDITS for list of people who contributed to this
+ * project.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation; either version 2 of
+ * the License, or (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston,
+ * MA 02111-1307 USA
+ */
+
+#include <config.h>
+#include <asm-offsets.h>
+
+/*
+ * This file handles the target-independent stages of the U-Boot
+ * start-up where a C runtime environment is needed. Its entry point
+ * is _main and is branched into from the target's start.S file.
+ *

arch/arm/lib/crt0.S new file +52 -0

+ *    new GD resides.
+ *
+ * 3. Set up intermediate environment where the stack and GD are the
+ *    ones allocated by board_init_f() in DDR, but BSS and initialized
+ *    non-const data are still not available.
+ *
+ * 4. Call relocate_code(). This function relocates U-Boot from its
+ *    current location into the relocation destination computed by
+ *    board_init_f().
+ *
+ * 5. Set up final environment for calling board_init_r(). This
+ *    environment has BSS (initialized to 0), initialized non-const
+ *    data (initialized to their intended value), and stack in DDR.
+ *    GD has but retained values set by board_init_f(). Some CPUs
+ *    have some work to do at this point, so call c_runtime_cpu_setup.
+ *
+ * 6. Call noard_init_r(). If the function returns, reset the board.
+ */
+
+/*
+ * offset of the nand_boot() function for SPL crt
+ */
+
+#if defined(CONFIG_NAND_SPL)
+
+.globl nand_boot
+_nand_boot:
+	.word nand_boot
+
+#elif ! defined(CONFIG_SPL_BUILD)
+
+/*
+ * offset of the board_init_r() function for non-SPL crt
+ */
+
+.globl board_init_r
+_board_init_r:
+	.word board_init_r
+
+#endif
+
+/*
+ * start and end of BSS
+ */
+
+.globl __bss_start
+.globl __bss_end__
+
+/*
+ * entry point of crt0 sequence
+ */
+

arch/arm/lib/crt0.S new file +39 -0

+	ldr	sp, =(CONFIG_SYS_INIT_SP_ADDR)
+#elif defined(CONFIG_SPL_BUILD) && defined(CONFIG_SPL_STACK)
+	ldr	sp, =(CONFIG_SPL_STACK)
+#else
+	ldr	sp, =(CONFIG_SYS_INIT_SP_ADDR)
+#endif
+	bic	sp, sp, #7 /* 8-byte alignment for ABI compliance */
+	mov	r8, sp		/* GD is above SP */
+	mov	r0, #0
+	bl	board_init_f
+
+/*
+ * Set up intermediate environment (new sp and gd) and call
+ * relocate_code(addr_sp, gd, addr_moni). Trick here is that
+ * we'll return 'here' but relocated.
+ */
+
+	ldr	sp, [r8, #GD_START_ADDR_SP]	/* r8 = gd->start_addr_sp */
+	ldr	r8, [r8, #GD_BD]		/* r8 = gd->bd */
+	sub	r8, r8, #GD_SIZE		/* new GD is below bd */
+
+#ifndef CONFIG_SPL_BUILD
+
+	adr	lr, here
+	ldr	r0, [r8, #GD_RELOC_OFF]		/* lr = gd->start_addr_sp */
+	add	lr, lr, r0
+	ldr	r0, [r8, #GD_START_ADDR_SP]	/* r0 = gd->start_addr_sp */
+	mov	r1, r8				/* r1 = gd */
+	ldr	r2, [r8, #GD_RELOCADDR]		/* r2 = gd->relocaddr */
+	b	relocate_code
+here:
+
+#endif
+
+/* Set up final (full) environment */
+
+	bl	c_runtime_cpu_setup	/* we still call old routine here */
+
+	ldr	r0, =__bss_start	/* this is auto-relocated! */

arch/arm/lib/crt0.S new file +7 -0

+
+	mov	lr, pc
+
+#if defined(CONFIG_NAND_SPL)
+	ldr     pc, _nand_boot
+
+#elif ! defined(CONFIG_SPL_BUILD)

Dear Albert Aribaud,

DDR means DRAM in general?

Quoted 1 line

+ * must use GD to store any data which must be passed on to later,

-----------------------------------------------------------------------^

Quoted 2 lines

+ * stages, including the reloction destination and the new stack
+ * pointer address, below which the stack resides and above it the

------------------------^
Full stop? The rest of the sentence makes no sense to me.

Review

Quoted 2 lines

(quoted diff) +2 -0

+#if defined(CONFIG_NAND_SPL)
+	/* deprecated, use instead CONFIG_SPL_BUILD */

I wonder if we can emit a #warning here

Quoted 1 line

+#ifdef CONFIG_SPL_BUILD

I think this should be negated. The coloured LED stuff was enabled for non SPL builds before!

arch/arm/lib/crt0.S line 163

Quoted 7 lines

arch/arm/lib/crt0.S new file +7 -0

+
+	bl coloured_LED_init
+	bl red_led_on
+
+#endif
+
+	/* call board_init_r(gd_t *id, ulong dest_addr) */

Shouldn't this move down in the #elif path?

Best regards

Andreas Bie?mann

The goal of this series is to scrub the start.S files which have proliferated across arch/arm and eliminate code redundancy.

Hi Tom,

On Sun, 4 Nov 2012 12:32:03 +0100, Albert ARIBAUD <[email protected]> wrote:

Quoted 3 lines

The goal of this series is to scrub the start.S files which have proliferated across arch/arm and eliminate code redundancy.

I know this came a bit late in early nov 4th, but I really would like it to be considered for 2013.01. Would you agree to make an exception for it? Thanks in advance.

Quoted 9 lines · Albert ARIBAUD

Hi Tom,

Quoted 3 lines · Albert ARIBAUD

The goal of this series is to scrub the start.S files which have proliferated across arch/arm and eliminate code redundancy.

Quoted 3 lines

I know this came a bit late in early nov 4th, but I really would like it to be considered for 2013.01. Would you agree to make an exception for it? Thanks in advance.

If you can collect a diverse set of Tested-by's, yes.

Quoted 15 lines · Tom Rini

Quoted 12 lines · Albert ARIBAUD

Hi Tom,

Quoted 3 lines · Albert ARIBAUD

The goal of this series is to scrub the start.S files which have proliferated across arch/arm and eliminate code redundancy.

I know this came a bit late in early nov 4th, but I really would like it to be considered for 2013.01. Would you agree to make an exception for it? Thanks in advance.

Quoted 1 line

If you can collect a diverse set of Tested-by's, yes.

I have not gone through the patch, but will test it in a day or two on the hawkboard, and report the findings.

The hawkboard comes with the arm926ejs core, so that part of the code would be tested.

hi Albert,

Quoted 16 lines · Sughosh Ganu

Quoted 15 lines · Tom Rini

Quoted 12 lines · Albert ARIBAUD

Hi Tom,

Quoted 3 lines · Albert ARIBAUD

The goal of this series is to scrub the start.S files which have proliferated across arch/arm and eliminate code redundancy.

I know this came a bit late in early nov 4th, but I really would like it to be considered for 2013.01. Would you agree to make an exception for it? Thanks in advance.

Quoted 7 lines

Quoted 1 line

If you can collect a diverse set of Tested-by's, yes.

I have not gone through the patch, but will test it in a day or two on the hawkboard, and report the findings.

The hawkboard comes with the arm926ejs core, so that part of the code would be tested.

I tried the 1st patch of the series, and with that u-boot does not come up on the board. It is also printing out some random values for the dram and nand sizes.

The patch was applied on top of commit 1cc619be8b7. Also, with the mentioned commit, u-boot boots up fine on the board. Also to be noted is that the spl image compiled with these changes is booting up fine, loading the main u-boot image, and jumping to it -- the issue is with booting the main u-boot image.

Hi Sughosh,

Quoted 16 lines · Sughosh Ganu

hi Albert,

Quoted 13 lines · Sughosh Ganu

Quoted 12 lines · Tom Rini

Quoted 10 lines · Albert ARIBAUD

Hi Tom,

Quoted 3 lines · Albert ARIBAUD

The goal of this series is to scrub the start.S files which have proliferated across arch/arm and eliminate code redundancy.

I know this came a bit late in early nov 4th, but I really would

like

Quoted 1 line

Quoted 1 line

Quoted 1 line

Quoted 1 line

it to be considered for 2013.01. Would you agree to make an

exception

Quoted 14 lines

Quoted 9 lines

Quoted 3 lines

Quoted 1 line

for it? Thanks in advance.

If you can collect a diverse set of Tested-by's, yes.

I have not gone through the patch, but will test it in a day or two on the hawkboard, and report the findings.

The hawkboard comes with the arm926ejs core, so that part of the code would be tested.

I tried the 1st patch of the series, and with that u-boot does not come up on the board. It is also printing out some random values for the dram and nand sizes.

Quoted 5 lines

The patch was applied on top of commit 1cc619be8b7. Also, with the mentioned commit, u-boot boots up fine on the board. Also to be noted is that the spl image compiled with these changes is booting up fine, loading the main u-boot image, and jumping to it -- the issue is with booting the main u-boot image.

Thanks Sughosh. Can you build an U-Boot with the following defined in the hawkboard.h config file?

#define DEBUG
#if defined(CONFIG_SPL_BUILD) && ! defined (__ASSEMBLY__) static inline int printf(const char *fmt, ...)

{
        return 0;
}

#endif

Note: only the #define DEBUG matter to me, but with it alone, SPL build fails due to some code now requiring printf(). This is why I add a dummy printf definition for C code during SPL build -- ASM code does not need printf() and actually chokes on the definition, hence the condition on __ASSEMBLY__.

This debug U-Boot should print a lot more info. Can you please try it and copy/paste its output here? Thanks in advance.

hi Albert,

Quoted 55 lines · Albert ARIBAUD

Hi Sughosh,

Quoted 16 lines · Sughosh Ganu

hi Albert,

Quoted 13 lines · Sughosh Ganu

Quoted 12 lines · Tom Rini

Quoted 10 lines · Albert ARIBAUD

Hi Tom,

Quoted 3 lines · Albert ARIBAUD

The goal of this series is to scrub the start.S files which have proliferated across arch/arm and eliminate code redundancy.

I know this came a bit late in early nov 4th, but I really would

like

Quoted 1 line

Quoted 1 line

Quoted 1 line

Quoted 1 line

it to be considered for 2013.01. Would you agree to make an

exception

Quoted 14 lines

Quoted 9 lines

Quoted 3 lines

Quoted 1 line

for it? Thanks in advance.

If you can collect a diverse set of Tested-by's, yes.

I have not gone through the patch, but will test it in a day or two on the hawkboard, and report the findings.

The hawkboard comes with the arm926ejs core, so that part of the code would be tested.

I tried the 1st patch of the series, and with that u-boot does not come up on the board. It is also printing out some random values for the dram and nand sizes.

Quoted 5 lines

The patch was applied on top of commit 1cc619be8b7. Also, with the mentioned commit, u-boot boots up fine on the board. Also to be noted is that the spl image compiled with these changes is booting up fine, loading the main u-boot image, and jumping to it -- the issue is with booting the main u-boot image.

Thanks Sughosh. Can you build an U-Boot with the following defined in the hawkboard.h config file?

#define DEBUG
#if defined(CONFIG_SPL_BUILD) && ! defined (__ASSEMBLY__) static inline int printf(const char *fmt, ...)

{
        return 0;
}

#endif

Quoted 9 lines

Note: only the #define DEBUG matter to me, but with it alone, SPL build fails due to some code now requiring printf(). This is why I add a dummy printf definition for C code during SPL build -- ASM code does not need printf() and actually chokes on the definition, hence the condition on __ASSEMBLY__.

This debug U-Boot should print a lot more info. Can you please try it and copy/paste its output here? Thanks in advance.

Unfortunately i am currently on vacation, with no access to the board, and
would be able to try out your suggested changes only after i am back, by
the end of next week. In the meantime, i am also trying to get myself a jtag debugger -- it is quite frustrating to provide half baked information.

I had tried enabling DEBUG in the board.c file, and with that, i saw that board_init_f completes, and the board hangs after returning from the function. Now i really need a debugger to figure out where exactly is the problem, as it could be in the relocation part, or while trying to jump to the board_init_r after having relocated the code. Hopefully, with a jtag debugger, i will be able to provide you a lot more info.

-sughosh

Quoted 27 lines

Quoted 27 lines

Quoted 9 lines

I tried the 1st patch of the series, and with that u-boot does not come up on the board. It is also printing out some random values for the dram and nand sizes.

The patch was applied on top of commit 1cc619be8b7. Also, with the mentioned commit, u-boot boots up fine on the board. Also to be noted is that the spl image compiled with these changes is booting up fine, loading the main u-boot image, and jumping to it -- the issue is with booting the main u-boot image.

Thanks Sughosh. Can you build an U-Boot with the following defined in the hawkboard.h config file?

#define DEBUG
#if defined(CONFIG_SPL_BUILD) && ! defined (__ASSEMBLY__) static inline int printf(const char *fmt, ...)

{
        return 0;
}

#endif

Note: only the #define DEBUG matter to me, but with it alone, SPL build fails due to some code now requiring printf(). This is why I add a dummy printf definition for C code during SPL build -- ASM code does not need printf() and actually chokes on the definition, hence the condition on __ASSEMBLY__.

Quoted 9 lines

Quoted 3 lines

This debug U-Boot should print a lot more info. Can you please try it and copy/paste its output here? Thanks in advance.

Unfortunately i am currently on vacation, with no access to the board, and
would be able to try out your suggested changes only after i am back, by
the end of next week. In the meantime, i am also trying to get myself a jtag debugger -- it is quite frustrating to provide half baked information.

Thanks for the feedback. Meanwhile, I have found two ugly bugs which would certainly affect execution past board_init_f(), so maybe your hawkboard was bitten by theses; next time please try v3, which fixes these bugs, or the latest patch version if v3 is obsolete by the time you come back.

Move all the C runtime setup code from every start.S in arch/arm into arch/arm/lib/crt0.S. This covers the code sequence from isetting up the initial stack to calling into board_init_r().

Also, rewrite the C runtime setup and make functions board_init_*() and relocate_code() behave according to normal C semantics (no jumping across the C stack any more, etc).

Some SPL targets had to be touched because they use start.S exolicitly or for some reason; the relevant maintainers and custodians are cc:ed.

Signed-off-by Albert ARIBAUD <[email protected]>

---
Changes in v3:
- various clarifications and typo fixes
- fixed wrong LED calls conditional
- fixed open comment eating some code
- fixed code overrun in SPL case
Series-changes: 2
- moved description from cover letter to patch commit msg
- added note about tests in the cover letter
- fixed baords with CONFIG_SPL but not CONFIG_SPL_STACK

 arch/arm/cpu/arm1136/start.S              |   69 +++--------
 arch/arm/cpu/arm1176/start.S              |   62 ++--------
 arch/arm/cpu/arm720t/start.S              |   53 ++-------
 arch/arm/cpu/arm920t/start.S              |   61 ++--------
 arch/arm/cpu/arm925t/start.S              |   61 ++--------
 arch/arm/cpu/arm926ejs/start.S            |   80 ++-----------
 arch/arm/cpu/arm946es/start.S             |   56 ++-------
 arch/arm/cpu/arm_intcm/start.S            |   63 ++--------
 arch/arm/cpu/armv7/start.S                |   58 +++-------
 arch/arm/cpu/ixp/start.S                  |   55 ++-------
 arch/arm/cpu/pxa/start.S                  |   63 ++--------
 arch/arm/cpu/s3c44b0/start.S              |   55 ++-------
 arch/arm/cpu/sa1100/start.S               |   50 ++------
 arch/arm/lib/Makefile                     |    2 +
 arch/arm/lib/board.c                      |   11 --
 arch/arm/lib/crt0.S                       |  180 +++++++++++++++++++++++++++++
 include/common.h                          |    2 +-
 include/configs/socfpga_cyclone5.h        |    2 +-
 lib/asm-offsets.c                         |   10 ++
 nand_spl/board/freescale/mx31pdk/Makefile |    6 +-
 nand_spl/board/karo/tx25/Makefile         |    6 +-
 21 files changed, 329 insertions(+), 676 deletions(-)
 create mode 100644 arch/arm/lib/crt0.S

arch/arm/lib/crt0.S new file +139 -0

new file mode 100644
index 0000000..ad6c71d
@@ -0,0 +1,180 @@
+/*
+ *  crt0 - C-runtime startup Code for ARM U-Boot
+ *
+ *  Copyright (c) 2012  Albert ARIBAUD <[email protected]>
+ *
+ * See file CREDITS for list of people who contributed to this
+ * project.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation; either version 2 of
+ * the License, or (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston,
+ * MA 02111-1307 USA
+ */
+
+#include <config.h>
+#include <asm-offsets.h>
+
+/*
+ * This file handles the target-independent stages of the U-Boot
+ * start-up where a C runtime environment is needed. Its entry point
+ * is _main and is branched into from the target's start.S file.
+ *
+ * _main execution sequence is:
+ *
+ * 1. Set up initial environment for calling board_init_f().
+ *    This environment only provides a stack and a place to store
+ *    the GD ('global data') structure, both located in some readily
+ *    available RAM (SRAM, locked cache...). In this context, VARIABLE
+ *    global data, initialized or not (BSS), are UNAVAILABLE; only
+ *    CONSTANT initialized data are available.
+ *
+ * 2. Call board_init_f(). This function prepares the hardware for
+ *    execution from system RAM (DRAM, DDR...) As system RAM may not
+ *    be available yet, , board_init_f() must use the current GD to
+ *    store any data which must be passed on to later stages. These
+ *    data include the relocation destination, the future stack, and
+ *    the future GD location.
+ *
+ * 3. Set up intermediate environment where the stack and GD are the
+ *    ones allocated by board_init_f() in system RAM, but BSS and
+ *    initialized non-const data are still not available.
+ *
+ * 4. Call relocate_code(). This function relocates U-Boot from its
+ *    current location into the relocation destination computed by
+ *    board_init_f().
+ *
+ * 5. Set up final environment for calling board_init_r(). This
+ *    environment has BSS (initialized to 0), initialized non-const
+ *    data (initialized to their intended value), and stack in system
+ *    RAM. GD has retained values set by board_init_f(). Some CPUs
+ *    have some work left to do at this point regarding memory, so
+ *    call c_runtime_cpu_setup.
+ *
+ * 6. Branch to either _nand_boot() or board_init_r().
+ */
+
+/*
+ * offset of the nand_boot() function for SPL crt
+ */
+
+#if defined(CONFIG_NAND_SPL)
+
+.globl nand_boot
+_nand_boot:
+       .word nand_boot
+
+#elif ! defined(CONFIG_SPL_BUILD)
+
+/*
+ * offset of the board_init_r() function for non-SPL crt
+ */
+
+.globl board_init_r
+_board_init_r:
+       .word board_init_r
+
+#endif
+
+/*
+ * start and end of BSS
+ */
+
+.globl __bss_start
+.globl __bss_end__
+
+/*
+ * entry point of crt0 sequence
+ */
+
+.global _main
+
+_main:
+
+/*
+ * Set up initial C runtime environment and call board_init_f(0).
+ */
+
+#if defined(CONFIG_NAND_SPL)
+       /* deprecated, use instead CONFIG_SPL_BUILD */
+       ldr     sp, =(CONFIG_SYS_INIT_SP_ADDR)
+#elif defined(CONFIG_SPL_BUILD) && defined(CONFIG_SPL_STACK)
+       ldr     sp, =(CONFIG_SPL_STACK)
+#else
+       ldr     sp, =(CONFIG_SYS_INIT_SP_ADDR)
+#endif
+       bic     sp, sp, #7 /* 8-byte alignment for ABI compliance */
+       mov     r8, sp          /* GD is above SP */
+       mov     r0, #0
+       bl      board_init_f
+
+/*
+ * Set up intermediate environment (new sp and gd) and call
+ * relocate_code(addr_sp, gd, addr_moni). Trick here is that
+ * we'll return 'here' but relocated.
+ */
+
+       ldr     sp, [r8, #GD_START_ADDR_SP]     /* r8 = gd->start_addr_sp */
+       ldr     r8, [r8, #GD_BD]                /* r8 = gd->bd */
+       sub     r8, r8, #GD_SIZE                /* new GD is below bd */
+
+#if ! defined(CONFIG_SPL_BUILD)
+
+       adr     lr, here
+       ldr     r0, [r8, #GD_RELOC_OFF]         /* lr = gd->start_addr_sp */
+       add     lr, lr, r0
+       ldr     r0, [r8, #GD_START_ADDR_SP]     /* r0 = gd->start_addr_sp */
+       mov     r1, r8                          /* r1 = gd */
+       ldr     r2, [r8, #GD_RELOCADDR]         /* r2 = gd->relocaddr */
+       b       relocate_code

arch/arm/lib/crt0.S new file +10 -0

+
+#if ! defined(CONFIG_SPL_BUILD)
+
+       bl coloured_LED_init
+       bl red_led_on
+
+#endif
+
+#if defined(CONFIG_NAND_SPL)
+

Hi Albert,

Quoted 14 lines · Albert ARIBAUD

Move all the C runtime setup code from every start.S in arch/arm into arch/arm/lib/crt0.S. This covers the code sequence from isetting up the initial stack to calling into board_init_r().

Also, rewrite the C runtime setup and make functions board_init_*() and relocate_code() behave according to normal C semantics (no jumping across the C stack any more, etc).

Some SPL targets had to be touched because they use start.S exolicitly or for some reason; the relevant maintainers and custodians are cc:ed.

Quoted 1 line

Signed-off-by Albert ARIBAUD <[email protected]>

I am very pleased to see this. My generic relocation series from some time ago is still hanging around. Your series tidies up some things which would make that series easier to do. I'm happy to take another look at it.

I have tested your series on a Tegra seaboard, and got it running with a few changes. Please see comments below.

arch/arm/lib/crt0.S line 149

Quoted 2 lines

arch/arm/lib/crt0.S new file +2 -0

+       ldr     r0, =__bss_start        /* this is auto-relocated! */
+       ldr     r1, =__bss_end__        /* this is auto-relocated! */

I noticed this last time around - really we should have __bss_end__, not __bss_end. Something to look at in future work, perhaps.

Quoted 1 line

+ bhs clbss_l

I think this should be blo, not bhs. Otherwise the BSS does not get cleared and my seaboard doesn't boot properly.

arch/arm/lib/crt0.S line 176

Quoted 2 lines

arch/arm/lib/crt0.S new file +2 -0

+       /* call board_init_r */
+       ldr     pc, _board_init_r       /* this is auto-relocated! */

For my toolchain I need:

ldr pc, =board_init_r

otherwise I get:

crt0.S: Assembler messages:
crt0.S:176: Error: internal_relocation (type: OFFSET_IMM) not fixed up

Also _board_init_r seems to be the wrong symbol name for me:

arch/arm/lib/libarm.o: In function `clbss_l':
/home/sjg/trunk/src/third_party/u-boot/files/arch/arm/lib/crt0.S:176:
undefined reference to `_board_init_r'

Regards,
Simon

The goal of this series is to scrub the start.S files which have proliferated across arch/arm and eliminate code redundancy.

Currently this series only factorizes the C runtime framework, that is, the assembly code responsible for setting up the C environments needed by board_init_*() and relocate_code().

Further factorization is possible: relocate_code() can be moved out and possibly rewritten in C now that it has plain C function semantics; exception handlers should also be pretty much cpu-independent.

Eventually, start.S files should disappear or contain CPU-specific code only.

The new C runtime setup sequence has been validated step by step on target versatileqemu using toolchain ELDK4.2.

The whole changes have been build-tested across all ARM targets using MAKEALL -a arm.

Changes in v3:
- various clarifications and typo fixes
- fixed wrong LED calls conditional
- fixed open comment eating some code
- fixed code overrun in SPL case

Remove code such as unneeded includes or labels.
These removals have no functional effect.

Signed-off-by Albert ARIBAUD <[email protected]>

---
Changes in v2:
- removed useless includes in arm926ejs and arm925t

Move all the C runtime setup code from every start.S in arch/arm into arch/arm/lib/crt0.S. This covers the code sequence from setting up the initial stack to calling into board_init_r().

Also, rewrite the C runtime setup and make functions board_init_*() and relocate_code() behave according to normal C semantics (no jumping across the C stack any more, etc).

Some SPL targets had to be touched because they use start.S explicitly or for some reason; the relevant maintainers and custodians are cc:ed.

Signed-off-by Albert ARIBAUD <[email protected]>

Hi Albert,

Quoted 10 lines · Albert ARIBAUD

Move all the C runtime setup code from every start.S in arch/arm into arch/arm/lib/crt0.S. This covers the code sequence from setting up the initial stack to calling into board_init_r().

Also, rewrite the C runtime setup and make functions board_init_*() and relocate_code() behave according to normal C semantics (no jumping across the C stack any more, etc).

Quoted 5 lines

Some SPL targets had to be touched because they use start.S explicitly or for some reason; the relevant maintainers and custodians are cc:ed.

Signed-off-by Albert ARIBAUD <[email protected]>

I tested this on Seaboard (Tegra 20).

Tested-by Simon Glass <[email protected]>

I tried to test it on a snow (exynos5250) but couldn't really sync up with our tree, so gave up. I am not completely sure about how the CONFIG_SPL_BUILD stuff fits together in start.S and I got a build error I wasn't sure how to correct (arch/arm/cpu/armv7/start.S:179:
undefined reference to `relocate_done')

Regards,
Simon

Hi Simon,

Quoted 22 lines · Simon Glass

Hi Albert,

Quoted 10 lines · Albert ARIBAUD

Move all the C runtime setup code from every start.S in arch/arm into arch/arm/lib/crt0.S. This covers the code sequence from setting up the initial stack to calling into board_init_r().

Also, rewrite the C runtime setup and make functions board_init_*() and relocate_code() behave according to normal C semantics (no jumping across the C stack any more, etc).

Quoted 5 lines

Some SPL targets had to be touched because they use start.S explicitly or for some reason; the relevant maintainers and custodians are cc:ed.

Signed-off-by Albert ARIBAUD <[email protected]>

I tested this on Seaboard (Tegra 20).

Quoted 7 lines

Tested-by Simon Glass <[email protected]>

I tried to test it on a snow (exynos5250) but couldn't really sync up with our tree, so gave up. I am not completely sure about how the CONFIG_SPL_BUILD stuff fits together in start.S and I got a build error I wasn't sure how to correct (arch/arm/cpu/armv7/start.S:179:
undefined reference to `relocate_done')

Thansk Simon.

In file arch/arm/cpu/armv7/start.S, line 179 uses relocate_done and line 228 provides it, and there is no preprocessor conditional in-between, so I fail to see how you could get this error.

Is your tree (and branch) available so that I can try and see the issue for myself? Either it'll uncover a problem in my patch series, or it'll help you merge it (or both).

Quoted 3 lines

Regards,
Simon

Amicalement,
--
Albert.

Quoted 12 lines

Quoted 5 lines

I tried to test it on a snow (exynos5250) but couldn't really sync up with our tree, so gave up. I am not completely sure about how the CONFIG_SPL_BUILD stuff fits together in start.S and I got a build error I wasn't sure how to correct (arch/arm/cpu/armv7/start.S:179:
undefined reference to `relocate_done')

Thansk Simon.

In file arch/arm/cpu/armv7/start.S, line 179 uses relocate_done and line 228 provides it, and there is no preprocessor conditional in-between, so I fail to see how you could get this error.

Quoted 3 lines

Is your tree (and branch) available so that I can try and see the issue for myself? Either it'll uncover a problem in my patch series, or it'll help you merge it (or both).

OK I will try a bit harder and come back to you. Sorry I can't do that immediately but will be soon...

Regards,
Simon

Hi Simon,

Quoted 12 lines · Simon Glass

Quoted 12 lines

Quoted 5 lines

I tried to test it on a snow (exynos5250) but couldn't really sync up with our tree, so gave up. I am not completely sure about how the CONFIG_SPL_BUILD stuff fits together in start.S and I got a build error I wasn't sure how to correct (arch/arm/cpu/armv7/start.S:179:
undefined reference to `relocate_done')

Thansk Simon.

In file arch/arm/cpu/armv7/start.S, line 179 uses relocate_done and line 228 provides it, and there is no preprocessor conditional in-between, so I fail to see how you could get this error.

Quoted 6 lines

Quoted 3 lines

Is your tree (and branch) available so that I can try and see the issue for myself? Either it'll uncover a problem in my patch series, or it'll help you merge it (or both).

OK I will try a bit harder and come back to you. Sorry I can't do that immediately but will be soon...

Did you manage to reproduce or sort out this issue?

Quoted 3 lines

Regards,
Simon

Amicalement,
--
Albert.

Hi Albert,

Quoted 24 lines · Albert ARIBAUD

Hi Simon,

Quoted 18 lines · Simon Glass

Quoted 8 lines

Quoted 5 lines

I tried to test it on a snow (exynos5250) but couldn't really sync up with our tree, so gave up. I am not completely sure about how the CONFIG_SPL_BUILD stuff fits together in start.S and I got a build error I wasn't sure how to correct (arch/arm/cpu/armv7/start.S:179:
undefined reference to `relocate_done')

Thansk Simon.

Quoted 7 lines

In file arch/arm/cpu/armv7/start.S, line 179 uses relocate_done and line 228 provides it, and there is no preprocessor conditional in-between, so I fail to see how you could get this error.

Is your tree (and branch) available so that I can try and see the issue for myself? Either it'll uncover a problem in my patch series, or it'll help you merge it (or both).

OK I will try a bit harder and come back to you. Sorry I can't do that immediately but will be soon...

Quoted 1 line

Did you manage to reproduce or sort out this issue?

No, I was hoping that I would be able to test upstream U-Boot on snow (waiting on some memory patches to appear), but that hasn't happened. I will have another go at working out the problem - either this week or next depending on when I can set aside the time for it. However, please don't hold up the series for it.

Quoted 8 lines

Quoted 4 lines

Regards,
Simon

Amicalement,
--
Albert.

Regards,
Simon

Move all the C runtime setup code from every start.S in arch/arm into arch/arm/lib/crt0.S. This covers the code sequence from setting up the initial stack to calling into board_init_r().

Also, rewrite the C runtime setup and make functions board_init_*() and relocate_code() behave according to normal C semantics (no jumping across the C stack any more, etc).

Some SPL targets had to be touched because they use start.S explicitly or for some reason; the relevant maintainers and custodians are cc:ed.

Signed-off-by Albert ARIBAUD <[email protected]>

---

Move all the C runtime setup code from every start.S in arch/arm into arch/arm/lib/crt0.S. This covers the code sequence from setting up the initial stack to calling into board_init_r().

Also, rewrite the C runtime setup and make functions board_init_*() and relocate_code() behave according to normal C semantics (no jumping across the C stack any more, etc).

Some SPL targets had to be touched because they use start.S explicitly or for some reason; the relevant maintainers and custodians are cc:ed.

Signed-off-by Albert ARIBAUD <[email protected]>

Move all the C runtime setup code from every start.S in arch/arm into arch/arm/lib/crt0.S. This covers the code sequence from setting up the initial stack to calling into board_init_r().

Also, rewrite the C runtime setup and make functions board_init_*() and relocate_code() behave according to normal C semantics (no jumping across the C stack any more, etc).

Some SPL targets had to be touched because they use start.S explicitly or for some reason; the relevant maintainers and custodians are cc:ed.

Signed-off-by Albert ARIBAUD <[email protected]>

---
Changes in v4:
- fixed bhs into blo in BSS init loop
- switched from explicit literals to '=target' notation
- fixed location of GD in initial environment
- stopped SPL boot sequence at calling board_init_f

Changes in v3:
- various clarifications and typo fixes
- fixed wrong LED calls conditional
- fixed open comment eating some code
- fixed code overrun in SPL case

Changes in v2:
- moved description from cover letter to patch commit msg
- added note about tests in the cover letter
- fixed baords with CONFIG_SPL but not CONFIG_SPL_STACK

Quoted 28 lines · Albert ARIBAUD

Move all the C runtime setup code from every start.S in arch/arm into arch/arm/lib/crt0.S. This covers the code sequence from setting up the initial stack to calling into board_init_r().

Also, rewrite the C runtime setup and make functions board_init_*() and relocate_code() behave according to normal C semantics (no jumping across the C stack any more, etc).

Some SPL targets had to be touched because they use start.S explicitly or for some reason; the relevant maintainers and custodians are cc:ed.

Signed-off-by Albert ARIBAUD <[email protected]>

---
Changes in v4:
- fixed bhs into blo in BSS init loop
- switched from explicit literals to '=target' notation
- fixed location of GD in initial environment
- stopped SPL boot sequence at calling board_init_f

Changes in v3:
- various clarifications and typo fixes
- fixed wrong LED calls conditional
- fixed open comment eating some code
- fixed code overrun in SPL case

Quoted 4 lines

Changes in v2:
- moved description from cover letter to patch commit msg
- added note about tests in the cover letter
- fixed baords with CONFIG_SPL but not CONFIG_SPL_STACK

This change breaks sandbox, powerpc, mips and m68k. I suspect fixing sandbox will be the easiest test-case and catch all of the others.

Signature · 1 line

Tom

The goal of this series is to scrub the start.S files which have proliferated across arch/arm and eliminate code redundancy.

This series only factorizes the C runtime framework, that is, the assembly code responsible for setting up the C environments needed by board_init_*() and relocate_code().

Further factorization is possible: relocate_code() can be moved out and possibly rewritten in C now that it has plain C function semantics; exception handlers should also be pretty much cpu-independent.

Also, SPL currently makes board_init_f() perform all of the setting up, including BSS clearing and chain- loading. This could be streamlined by only putting basic inits in board_init_f(), letting crt0 do the BSS initialization and chain-loading U-Boot from inside board_init_r().

Eventually, start.S files should disappear or contain CPU-specific code only.

The new C runtime setup sequence has been validated step by step on targets versatileqemu using gcc version 4.2.2 from the ELDK4.2 toolchain, and wireless_space using gcc version 4.7.2 (Ubuntu/Linaro 4.7.2-1ubuntu1).

The whole changes have been build-tested across all ARM targets using MAKEALL -a arm.

Changes in v5:
- Fixed build on non-ARM architectures

Changes in v4:
- fixed bhs into blo in BSS init loop
- switched from explicit literals to '=target' notation
- fixed location of GD in initial environment
- stopped SPL boot sequence at calling board_init_f

Changes in v3:
- various clarifications and typo fixes
- fixed wrong LED calls conditional
- fixed open comment eating some code
- fixed code overrun in SPL case

Changes in v2:
- moved description from cover letter to patch commit msg
- added note about tests in the cover letter
- fixed baords with CONFIG_SPL but not CONFIG_SPL_STACK
- removed useless includes in arm926ejs and arm925t

Albert ARIBAUD (2):
arm: move C runtime setup code in crt0.S
arm: remove useless code in start.S files

 arch/arm/cpu/arm1136/start.S              |   71 +++---------
 arch/arm/cpu/arm1176/start.S              |   62 ++---------
 arch/arm/cpu/arm720t/start.S              |   53 ++-------
 arch/arm/cpu/arm920t/start.S              |   61 ++--------
 arch/arm/cpu/arm925t/start.S              |   65 ++---------
 arch/arm/cpu/arm926ejs/start.S            |   86 ++------------
 arch/arm/cpu/arm946es/start.S             |   56 ++--------
 arch/arm/cpu/arm_intcm/start.S            |   63 ++---------
 arch/arm/cpu/armv7/start.S                |   58 +++-------
 arch/arm/cpu/ixp/start.S                  |   55 ++-------
 arch/arm/cpu/pxa/start.S                  |   63 ++---------
 arch/arm/cpu/s3c44b0/start.S              |   55 ++-------
 arch/arm/cpu/sa1100/start.S               |   50 ++-------
 arch/arm/lib/Makefile                     |    2 +
 arch/arm/lib/board.c                      |   11 --

 arch/arm/lib/crt0.S                       |  173 +++++++++++++++++++++++++++++
 include/common.h                          |    2 +-
 include/configs/socfpga_cyclone5.h        |    2 +-
 lib/asm-offsets.c                         |   14 +++
 nand_spl/board/freescale/mx31pdk/Makefile |    6 +-
 nand_spl/board/karo/tx25/Makefile         |    6 +-
 21 files changed, 326 insertions(+), 688 deletions(-)
 create mode 100644 arch/arm/lib/crt0.S

The goal of this series is to scrub the start.S files which have proliferated across arch/arm and eliminate code redundancy.

This series only factorizes the C runtime framework, that is, the assembly code responsible for setting up the C environments needed by board_init_*() and relocate_code().

Further factorization is possible: relocate_code() can be moved out and possibly rewritten in C now that it has plain C function semantics; exception handlers should also be pretty much cpu-independent.

Also, SPL currently makes board_init_f() perform all of the setting up, including BSS clearing and chain- loading. This could be streamlined by only putting basic inits in board_init_f(), letting crt0 do the BSS initialization and chain-loading U-Boot from inside board_init_r().

Eventually, start.S files should disappear or contain CPU-specific code only.

The new C runtime setup sequence has been validated step by step on targets versatileqemu using gcc version 4.2.2 from the ELDK4.2 toolchain, and wireless_space using gcc version 4.7.2 (Ubuntu/Linaro 4.7.2-1ubuntu1).

The whole changes have been build-tested across all ARM targets using MAKEALL -a arm.

Changes in v4:
- fixed bhs into blo in BSS init loop
- switched from explicit literals to '=target' notation
- fixed location of GD in initial environment
- stopped SPL boot sequence at calling board_init_f

Changes in v3:
- various clarifications and typo fixes
- fixed wrong LED calls conditional
- fixed open comment eating some code
- fixed code overrun in SPL case

Changes in v2:
- moved description from cover letter to patch commit msg
- added note about tests in the cover letter
- fixed baords with CONFIG_SPL but not CONFIG_SPL_STACK
- removed useless includes in arm926ejs and arm925t

Albert ARIBAUD (2):
arm: move C runtime setup code in crt0.S
arm: remove useless code in start.S files

 arch/arm/cpu/arm1136/start.S              |   71 +++---------
 arch/arm/cpu/arm1176/start.S              |   62 ++---------
 arch/arm/cpu/arm720t/start.S              |   53 ++-------
 arch/arm/cpu/arm920t/start.S              |   61 ++--------
 arch/arm/cpu/arm925t/start.S              |   65 ++---------
 arch/arm/cpu/arm926ejs/start.S            |   86 ++------------
 arch/arm/cpu/arm946es/start.S             |   56 ++--------
 arch/arm/cpu/arm_intcm/start.S            |   63 ++---------
 arch/arm/cpu/armv7/start.S                |   58 +++-------
 arch/arm/cpu/ixp/start.S                  |   55 ++-------
 arch/arm/cpu/pxa/start.S                  |   63 ++---------
 arch/arm/cpu/s3c44b0/start.S              |   55 ++-------
 arch/arm/cpu/sa1100/start.S               |   50 ++-------
 arch/arm/lib/Makefile                     |    2 +
 arch/arm/lib/board.c                      |   11 --

 arch/arm/lib/crt0.S                       |  173 +++++++++++++++++++++++++++++
 include/common.h                          |    2 +-
 include/configs/socfpga_cyclone5.h        |    2 +-
 lib/asm-offsets.c                         |   10 ++
 nand_spl/board/freescale/mx31pdk/Makefile |    6 +-
 nand_spl/board/karo/tx25/Makefile         |    6 +-
 21 files changed, 322 insertions(+), 688 deletions(-)
 create mode 100644 arch/arm/lib/crt0.S

Signature · 1 line

1.7.10.4

The goal of this series is to scrub the start.S files which have proliferated across arch/arm and eliminate code redundancy.

This series only factorizes the C runtime framework, that is, the assembly code responsible for setting up the C environments needed by board_init_*() and relocate_code().

Further factorization is possible: relocate_code() can be moved out and possibly rewritten in C now that it has plain C function semantics; exception handlers should also be pretty much cpu-independent.

Also, SPL currently makes board_init_f() perform all of the setting up, including BSS clearing and chain- loading. This could be streamlined by only putting basic inits in board_init_f(), letting crt0 do the BSS initialization and chain-loading U-Boot from inside board_init_r().

Eventually, start.S files should disappear or contain CPU-specific code only.

The new C runtime setup sequence has been validated step by step on targets versatileqemu using gcc version 4.2.2 from the ELDK4.2 toolchain, and wireless_space using gcc version 4.7.2 (Ubuntu/Linaro 4.7.2-1ubuntu1).

The whole changes have been build-tested across all ARM targets using MAKEALL -a arm.

Changes in v4:
- fixed bhs into blo in BSS init loop
- switched from explicit literals to '=target' notation
- fixed location of GD in initial environment
- stopped SPL boot sequence at calling board_init_f

Changes in v3:
- various clarifications and typo fixes
- fixed wrong LED calls conditional
- fixed open comment eating some code
- fixed code overrun in SPL case

Changes in v2:
- moved description from cover letter to patch commit msg
- added note about tests in the cover letter
- fixed baords with CONFIG_SPL but not CONFIG_SPL_STACK
- removed useless includes in arm926ejs and arm925t

Albert ARIBAUD (2):
arm: move C runtime setup code in crt0.S
arm: remove useless code in start.S files

 arch/arm/cpu/arm1136/start.S              |   71 +++---------
 arch/arm/cpu/arm1176/start.S              |   62 ++---------
 arch/arm/cpu/arm720t/start.S              |   53 ++-------
 arch/arm/cpu/arm920t/start.S              |   61 ++--------
 arch/arm/cpu/arm925t/start.S              |   65 ++---------
 arch/arm/cpu/arm926ejs/start.S            |   86 ++------------
 arch/arm/cpu/arm946es/start.S             |   56 ++--------
 arch/arm/cpu/arm_intcm/start.S            |   63 ++---------
 arch/arm/cpu/armv7/start.S                |   58 +++-------
 arch/arm/cpu/ixp/start.S                  |   55 ++-------
 arch/arm/cpu/pxa/start.S                  |   63 ++---------
 arch/arm/cpu/s3c44b0/start.S              |   55 ++-------
 arch/arm/cpu/sa1100/start.S               |   50 ++-------
 arch/arm/lib/Makefile                     |    2 +
 arch/arm/lib/board.c                      |   11 --

 arch/arm/lib/crt0.S                       |  173 +++++++++++++++++++++++++++++
 include/common.h                          |    2 +-
 include/configs/socfpga_cyclone5.h        |    2 +-
 lib/asm-offsets.c                         |   10 ++
 nand_spl/board/freescale/mx31pdk/Makefile |    6 +-
 nand_spl/board/karo/tx25/Makefile         |    6 +-
 21 files changed, 322 insertions(+), 688 deletions(-)
 create mode 100644 arch/arm/lib/crt0.S