Merge branch 'master' of ../master into next

38 files changed, 387 insertions, 287 deletions

diff --git a/arch/arm/cpu/arm1136/start.S b/arch/arm/cpu/arm1136/start.S
index eec8bee7eb..9a6f6cbee8 100644
--- a/arch/arm/cpu/arm1136/start.S
+++ b/arch/arm/cpu/arm1136/start.S
@@ -160,6 +160,7 @@ next:
 /* Set stackpointer in internal RAM to call board_init_f */
 call_board_init_f:
 	ldr	sp, =(CONFIG_SYS_INIT_SP_ADDR)
+	bic	sp, sp, #7 /* 8-byte alignment for ABI compliance */
 	ldr	r0,=0x00000000
 
 #ifdef CONFIG_NAND_SPL
@@ -186,22 +187,22 @@ relocate_code:
 	mov	r4, r0	/* save addr_sp */
 	mov	r5, r1	/* save addr of gd */
 	mov	r6, r2	/* save addr of destination */
-	mov	r7, r2	/* save addr of destination */
 
 	/* Set up the stack						    */
 stack_setup:
 	mov	sp, r4
 
 	adr	r0, _start
+	cmp	r0, r6
+	beq	clear_bss		/* skip relocation */
+	mov	r1, r6			/* r1 <- scratch for copy_loop */
 	ldr	r2, _TEXT_BASE
 	ldr	r3, _bss_start_ofs
 	add	r2, r0, r3		/* r2 <- source end address	    */
-	cmp	r0, r6
-	beq	clear_bss
 
 copy_loop:
 	ldmia	r0!, {r9-r10}		/* copy from source address [r0]    */
-	stmia	r6!, {r9-r10}		/* copy to   target address [r1]    */
+	stmia	r1!, {r9-r10}		/* copy to   target address [r1]    */
 	cmp	r0, r2			/* until source end address [r2]    */
 	blo	copy_loop
 
@@ -210,7 +211,7 @@ copy_loop:
 	 * fix .rel.dyn relocations
 	 */
 	ldr	r0, _TEXT_BASE		/* r0 <- Text base */
-	sub	r9, r7, r0		/* r9 <- relocation offset */
+	sub	r9, r6, r0		/* r9 <- relocation offset */
 	ldr	r10, _dynsym_start_ofs	/* r10 <- sym table ofs */
 	add	r10, r10, r0		/* r10 <- sym table in FLASH */
 	ldr	r2, _rel_dyn_start_ofs	/* r2 <- rel dyn start ofs */
@@ -221,10 +222,10 @@ fixloop:
 	ldr	r0, [r2]		/* r0 <- location to fix up, IN FLASH! */
 	add	r0, r0, r9		/* r0 <- location to fix up in RAM */
 	ldr	r1, [r2, #4]
-	and	r8, r1, #0xff
-	cmp	r8, #23			/* relative fixup? */
+	and	r7, r1, #0xff
+	cmp	r7, #23			/* relative fixup? */
 	beq	fixrel
-	cmp	r8, #2			/* absolute fixup? */
+	cmp	r7, #2			/* absolute fixup? */
 	beq	fixabs
 	/* ignore unknown type of fixup */
 	b	fixnext
@@ -233,7 +234,7 @@ fixabs:
 	mov	r1, r1, LSR #4		/* r1 <- symbol index in .dynsym */
 	add	r1, r10, r1		/* r1 <- address of symbol in table */
 	ldr	r1, [r1, #4]		/* r1 <- symbol value */
-	add	r1, r9			/* r1 <- relocated sym addr */
+	add	r1, r1, r9		/* r1 <- relocated sym addr */
 	b	fixnext
 fixrel:
 	/* relative fix: increase location by offset */
@@ -251,7 +252,7 @@ clear_bss:
 	ldr	r0, _bss_start_ofs
 	ldr	r1, _bss_end_ofs
 	ldr	r3, _TEXT_BASE		/* Text base */
-	mov	r4, r7			/* reloc addr */
+	mov	r4, r6			/* reloc addr */
 	add	r0, r0, r4
 	add	r1, r1, r4
 	mov	r2, #0x00000000		/* clear			    */
@@ -280,7 +281,7 @@ jump_2_ram:
 	add	lr, lr, r9
 	/* setup parameters for board_init_r */
 	mov	r0, r5		/* gd_t */
-	mov	r1, r7		/* dest_addr */
+	mov	r1, r6		/* dest_addr */
 	/* jump to it ... */
 	mov	pc, lr
 
diff --git a/arch/arm/cpu/arm1176/start.S b/arch/arm/cpu/arm1176/start.S
index cd8f01637a..237dcfef1b 100644
--- a/arch/arm/cpu/arm1176/start.S
+++ b/arch/arm/cpu/arm1176/start.S
@@ -223,6 +223,7 @@ skip_tcmdisable:
 /* Set stackpointer in internal RAM to call board_init_f */
 call_board_init_f:
 	ldr	sp, =(CONFIG_SYS_INIT_SP_ADDR)
+	bic	sp, sp, #7 /* 8-byte alignment for ABI compliance */
 	ldr	r0,=0x00000000
 	bl	board_init_f
 
@@ -240,22 +241,22 @@ relocate_code:
 	mov	r4, r0	/* save addr_sp */
 	mov	r5, r1	/* save addr of gd */
 	mov	r6, r2	/* save addr of destination */
-	mov	r7, r2	/* save addr of destination */
 
 	/* Set up the stack						    */
 stack_setup:
 	mov	sp, r4
 
 	adr	r0, _start
+	cmp	r0, r6
+	beq	clear_bss		/* skip relocation */
+	mov	r1, r6			/* r1 <- scratch for copy_loop */
 	ldr	r2, _TEXT_BASE
 	ldr	r3, _bss_start_ofs
 	add	r2, r0, r3		/* r2 <- source end address	    */
-	cmp	r0, r6
-	beq	clear_bss
 
 copy_loop:
 	ldmia	r0!, {r9-r10}		/* copy from source address [r0]    */
-	stmia	r6!, {r9-r10}		/* copy to   target address [r1]    */
+	stmia	r1!, {r9-r10}		/* copy to   target address [r1]    */
 	cmp	r0, r2			/* until source end address [r2]    */
 	blo	copy_loop
 
@@ -264,7 +265,7 @@ copy_loop:
 	 * fix .rel.dyn relocations
 	 */
 	ldr	r0, _TEXT_BASE		/* r0 <- Text base */
-	sub	r9, r7, r0		/* r9 <- relocation offset */
+	sub	r9, r6, r0		/* r9 <- relocation offset */
 	ldr	r10, _dynsym_start_ofs	/* r10 <- sym table ofs */
 	add	r10, r10, r0		/* r10 <- sym table in FLASH */
 	ldr	r2, _rel_dyn_start_ofs	/* r2 <- rel dyn start ofs */
@@ -275,10 +276,10 @@ fixloop:
 	ldr	r0, [r2]		/* r0 <- location to fix up, IN FLASH! */
 	add	r0, r0, r9		/* r0 <- location to fix up in RAM */
 	ldr	r1, [r2, #4]
-	and	r8, r1, #0xff
-	cmp	r8, #23			/* relative fixup? */
+	and	r7, r1, #0xff
+	cmp	r7, #23			/* relative fixup? */
 	beq	fixrel
-	cmp	r8, #2			/* absolute fixup? */
+	cmp	r7, #2			/* absolute fixup? */
 	beq	fixabs
 	/* ignore unknown type of fixup */
 	b	fixnext
@@ -287,7 +288,7 @@ fixabs:
 	mov	r1, r1, LSR #4		/* r1 <- symbol index in .dynsym */
 	add	r1, r10, r1		/* r1 <- address of symbol in table */
 	ldr	r1, [r1, #4]		/* r1 <- symbol value */
-	add	r1, r9			/* r1 <- relocated sym addr */
+	add	r1, r1, r9		/* r1 <- relocated sym addr */
 	b	fixnext
 fixrel:
 	/* relative fix: increase location by offset */
@@ -343,7 +344,7 @@ clear_bss:
 	ldr	r0, _bss_start_ofs
 	ldr	r1, _bss_end_ofs
 	ldr	r3, _TEXT_BASE		/* Text base */
-	mov	r4, r7			/* reloc addr */
+	mov	r4, r6			/* reloc addr */
 	add	r0, r0, r4
 	add	r1, r1, r4
 	mov	r2, #0x00000000		/* clear			    */
@@ -372,7 +373,7 @@ _nand_boot: .word nand_boot
 	add     lr, lr, r9
 	/* setup parameters for board_init_r */
 	mov	r0, r5		/* gd_t */
-	mov	r1, r7		/* dest_addr */
+	mov	r1, r6		/* dest_addr */
 	/* jump to it ... */
 	mov	pc, lr
 
diff --git a/arch/arm/cpu/arm720t/start.S b/arch/arm/cpu/arm720t/start.S
index 46050dab6e..abfa124638 100644
--- a/arch/arm/cpu/arm720t/start.S
+++ b/arch/arm/cpu/arm720t/start.S
@@ -138,6 +138,7 @@ reset:
 /* Set stackpointer in internal RAM to call board_init_f */
 call_board_init_f:
 	ldr	sp, =(CONFIG_SYS_INIT_SP_ADDR)
+	bic	sp, sp, #7 /* 8-byte alignment for ABI compliance */
 	ldr	r0,=0x00000000
 	bl	board_init_f
 
@@ -155,22 +156,22 @@ relocate_code:
 	mov	r4, r0	/* save addr_sp */
 	mov	r5, r1	/* save addr of gd */
 	mov	r6, r2	/* save addr of destination */
-	mov	r7, r2	/* save addr of destination */
 
 	/* Set up the stack						    */
 stack_setup:
 	mov	sp, r4
 
 	adr	r0, _start
+	cmp	r0, r6
+	beq	clear_bss		/* skip relocation */
+	mov	r1, r6			/* r1 <- scratch for copy_loop */
 	ldr	r2, _TEXT_BASE
 	ldr	r3, _bss_start_ofs
 	add	r2, r0, r3		/* r2 <- source end address	    */
-	cmp	r0, r6
-	beq	clear_bss
 
 copy_loop:
 	ldmia	r0!, {r9-r10}		/* copy from source address [r0]    */
-	stmia	r6!, {r9-r10}		/* copy to   target address [r1]    */
+	stmia	r1!, {r9-r10}		/* copy to   target address [r1]    */
 	cmp	r0, r2			/* until source end address [r2]    */
 	blo	copy_loop
 
@@ -179,7 +180,7 @@ copy_loop:
 	 * fix .rel.dyn relocations
 	 */
 	ldr	r0, _TEXT_BASE		/* r0 <- Text base */
-	sub	r9, r7, r0		/* r9 <- relocation offset */
+	sub	r9, r6, r0		/* r9 <- relocation offset */
 	ldr	r10, _dynsym_start_ofs	/* r10 <- sym table ofs */
 	add	r10, r10, r0		/* r10 <- sym table in FLASH */
 	ldr	r2, _rel_dyn_start_ofs	/* r2 <- rel dyn start ofs */
@@ -190,10 +191,10 @@ fixloop:
 	ldr	r0, [r2]		/* r0 <- location to fix up, IN FLASH! */
 	add	r0, r0, r9		/* r0 <- location to fix up in RAM */
 	ldr	r1, [r2, #4]
-	and	r8, r1, #0xff
-	cmp	r8, #23			/* relative fixup? */
+	and	r7, r1, #0xff
+	cmp	r7, #23			/* relative fixup? */
 	beq	fixrel
-	cmp	r8, #2			/* absolute fixup? */
+	cmp	r7, #2			/* absolute fixup? */
 	beq	fixabs
 	/* ignore unknown type of fixup */
 	b	fixnext
@@ -202,7 +203,7 @@ fixabs:
 	mov	r1, r1, LSR #4		/* r1 <- symbol index in .dynsym */
 	add	r1, r10, r1		/* r1 <- address of symbol in table */
 	ldr	r1, [r1, #4]		/* r1 <- symbol value */
-	add	r1, r9			/* r1 <- relocated sym addr */
+	add	r1, r1, r9		/* r1 <- relocated sym addr */
 	b	fixnext
 fixrel:
 	/* relative fix: increase location by offset */
@@ -220,7 +221,7 @@ clear_bss:
 	ldr	r0, _bss_start_ofs
 	ldr	r1, _bss_end_ofs
 	ldr	r3, _TEXT_BASE		/* Text base */
-	mov	r4, r7			/* reloc addr */
+	mov	r4, r6			/* reloc addr */
 	add	r0, r0, r4
 	add	r1, r1, r4
 	mov	r2, #0x00000000		/* clear			    */
@@ -244,7 +245,7 @@ clbss_l:str	r2, [r0]		/* clear loop...		    */
 	add	lr, lr, r9
 	/* setup parameters for board_init_r */
 	mov	r0, r5		/* gd_t */
-	mov	r1, r7		/* dest_addr */
+	mov	r1, r6		/* dest_addr */
 	/* jump to it ... */
 	mov	pc, lr
 
diff --git a/arch/arm/cpu/arm920t/at91/reset.c b/arch/arm/cpu/arm920t/at91/reset.c
index ce9c156154..51043ecddb 100644
--- a/arch/arm/cpu/arm920t/at91/reset.c
+++ b/arch/arm/cpu/arm920t/at91/reset.c
@@ -35,7 +35,10 @@
 #include <asm/arch/hardware.h>
 #include <asm/arch/at91_st.h>
 
-void board_reset(void) __attribute__((__weak__));
+void  __attribute__((weak)) board_reset(void)
+{
+	/* true empty function for defining weak symbol */
+}
 
 void reset_cpu(ulong ignored)
 {
@@ -45,8 +48,7 @@ void reset_cpu(ulong ignored)
 	serial_exit();
 #endif
 
-	if (board_reset)
-		board_reset();
+	board_reset();
 
 	/* Reset the cpu by setting up the watchdog timer */
 	writel(AT91_ST_WDMR_RSTEN | AT91_ST_WDMR_EXTEN | AT91_ST_WDMR_WDV(2),
diff --git a/arch/arm/cpu/arm920t/at91/timer.c b/arch/arm/cpu/arm920t/at91/timer.c
index 91377d47a6..d9a024fc0e 100644
--- a/arch/arm/cpu/arm920t/at91/timer.c
+++ b/arch/arm/cpu/arm920t/at91/timer.c
@@ -32,17 +32,16 @@
 
 #include <common.h>
 
-#include <asm/io.h>
-#include <asm/hardware.h>
+#include <asm/arch/io.h>
+#include <asm/arch/hardware.h>
 #include <asm/arch/at91_tc.h>
 #include <asm/arch/at91_pmc.h>
 
+DECLARE_GLOBAL_DATA_PTR;
+
 /* the number of clocks per CONFIG_SYS_HZ */
 #define TIMER_LOAD_VAL (CONFIG_SYS_HZ_CLOCK/CONFIG_SYS_HZ)
 
-static u32 timestamp;
-static u32 lastinc;
-
 int timer_init(void)
 {
 	at91_tc_t *tc = (at91_tc_t *) AT91_TC_BASE;
@@ -64,8 +63,8 @@ int timer_init(void)
 	writel(TIMER_LOAD_VAL, &tc->tc[0].rc);
 
 	writel(AT91_TC_CCR_SWTRG | AT91_TC_CCR_CLKEN, &tc->tc[0].ccr);
-	lastinc = 0;
-	timestamp = 0;
+	gd->lastinc = 0;
+	gd->tbl = 0;
 
 	return 0;
 }
@@ -86,7 +85,7 @@ ulong get_timer(ulong base)
 
 void set_timer(ulong t)
 {
-	timestamp = t;
+	gd->tbl = t;
 }
 
 void __udelay(unsigned long usec)
@@ -98,8 +97,8 @@ void reset_timer_masked(void)
 {
 	/* reset time */
 	at91_tc_t *tc = (at91_tc_t *) AT91_TC_BASE;
-	lastinc = readl(&tc->tc[0].cv) & 0x0000ffff;
-	timestamp = 0;
+	gd->lastinc = readl(&tc->tc[0].cv) & 0x0000ffff;
+	gd->tbl = 0;
 }
 
 ulong get_timer_raw(void)
@@ -109,16 +108,16 @@ ulong get_timer_raw(void)
 
 	now = readl(&tc->tc[0].cv) & 0x0000ffff;
 
-	if (now >= lastinc) {
+	if (now >= gd->lastinc) {
 		/* normal mode */
-		timestamp += now - lastinc;
+		gd->tbl += now - gd->lastinc;
 	} else {
 		/* we have an overflow ... */
-		timestamp += now + TIMER_LOAD_VAL - lastinc;
+		gd->tbl += now + TIMER_LOAD_VAL - gd->lastinc;
 	}
-	lastinc = now;
+	gd->lastinc = now;
 
-	return timestamp;
+	return gd->tbl;
 }
 
 ulong get_timer_masked(void)
diff --git a/arch/arm/cpu/arm920t/start.S b/arch/arm/cpu/arm920t/start.S
index 01edb9bdcb..08f178de10 100644
--- a/arch/arm/cpu/arm920t/start.S
+++ b/arch/arm/cpu/arm920t/start.S
@@ -184,6 +184,7 @@ copyex:
 /* Set stackpointer in internal RAM to call board_init_f */
 call_board_init_f:
 	ldr	sp, =(CONFIG_SYS_INIT_SP_ADDR)
+	bic	sp, sp, #7 /* 8-byte alignment for ABI compliance */
 	ldr	r0,=0x00000000
 	bl	board_init_f
 
@@ -201,22 +202,22 @@ relocate_code:
 	mov	r4, r0	/* save addr_sp */
 	mov	r5, r1	/* save addr of gd */
 	mov	r6, r2	/* save addr of destination */
-	mov	r7, r2	/* save addr of destination */
 
 	/* Set up the stack						    */
 stack_setup:
 	mov	sp, r4
 
 	adr	r0, _start
+	cmp	r0, r6
+	beq	clear_bss		/* skip relocation */
+	mov	r1, r6			/* r1 <- scratch for copy_loop */
 	ldr	r2, _TEXT_BASE
 	ldr	r3, _bss_start_ofs
 	add	r2, r0, r3		/* r2 <- source end address	    */
-	cmp	r0, r6
-	beq	clear_bss
 
 copy_loop:
 	ldmia	r0!, {r9-r10}		/* copy from source address [r0]    */
-	stmia	r6!, {r9-r10}		/* copy to   target address [r1]    */
+	stmia	r1!, {r9-r10}		/* copy to   target address [r1]    */
 	cmp	r0, r2			/* until source end address [r2]    */
 	blo	copy_loop
 
@@ -225,7 +226,7 @@ copy_loop:
 	 * fix .rel.dyn relocations
 	 */
 	ldr	r0, _TEXT_BASE		/* r0 <- Text base */
-	sub	r9, r7, r0		/* r9 <- relocation offset */
+	sub	r9, r6, r0		/* r9 <- relocation offset */
 	ldr	r10, _dynsym_start_ofs	/* r10 <- sym table ofs */
 	add	r10, r10, r0		/* r10 <- sym table in FLASH */
 	ldr	r2, _rel_dyn_start_ofs	/* r2 <- rel dyn start ofs */
@@ -236,10 +237,10 @@ fixloop:
 	ldr	r0, [r2]		/* r0 <- location to fix up, IN FLASH! */
 	add	r0, r0, r9		/* r0 <- location to fix up in RAM */
 	ldr	r1, [r2, #4]
-	and	r8, r1, #0xff
-	cmp	r8, #23			/* relative fixup? */
+	and	r7, r1, #0xff
+	cmp	r7, #23			/* relative fixup? */
 	beq	fixrel
-	cmp	r8, #2			/* absolute fixup? */
+	cmp	r7, #2			/* absolute fixup? */
 	beq	fixabs
 	/* ignore unknown type of fixup */
 	b	fixnext
@@ -248,7 +249,7 @@ fixabs:
 	mov	r1, r1, LSR #4		/* r1 <- symbol index in .dynsym */
 	add	r1, r10, r1		/* r1 <- address of symbol in table */
 	ldr	r1, [r1, #4]		/* r1 <- symbol value */
-	add	r1, r9			/* r1 <- relocated sym addr */
+	add	r1, r1, r9		/* r1 <- relocated sym addr */
 	b	fixnext
 fixrel:
 	/* relative fix: increase location by offset */
@@ -266,7 +267,7 @@ clear_bss:
 	ldr	r0, _bss_start_ofs
 	ldr	r1, _bss_end_ofs
 	ldr	r3, _TEXT_BASE		/* Text base */
-	mov	r4, r7			/* reloc addr */
+	mov	r4, r6			/* reloc addr */
 	add	r0, r0, r4
 	add	r1, r1, r4
 	mov	r2, #0x00000000		/* clear			    */
@@ -297,7 +298,7 @@ _nand_boot_ofs:
 	add	lr, lr, r9
 	/* setup parameters for board_init_r */
 	mov	r0, r5		/* gd_t */
-	mov	r1, r7		/* dest_addr */
+	mov	r1, r6		/* dest_addr */
 	/* jump to it ... */
 	mov	pc, lr
 
diff --git a/arch/arm/cpu/arm925t/start.S b/arch/arm/cpu/arm925t/start.S
index ce590a7038..2c0c869426 100644
--- a/arch/arm/cpu/arm925t/start.S
+++ b/arch/arm/cpu/arm925t/start.S
@@ -175,6 +175,7 @@ poll1:
 /* Set stackpointer in internal RAM to call board_init_f */
 call_board_init_f:
 	ldr	sp, =(CONFIG_SYS_INIT_SP_ADDR)
+	bic	sp, sp, #7 /* 8-byte alignment for ABI compliance */
 	ldr	r0,=0x00000000
 	bl	board_init_f
 
@@ -192,22 +193,22 @@ relocate_code:
 	mov	r4, r0	/* save addr_sp */
 	mov	r5, r1	/* save addr of gd */
 	mov	r6, r2	/* save addr of destination */
-	mov	r7, r2	/* save addr of destination */
 
 	/* Set up the stack						    */
 stack_setup:
 	mov	sp, r4
 
 	adr	r0, _start
+	cmp	r0, r6
+	beq	clear_bss		/* skip relocation */
+	mov	r1, r6			/* r1 <- scratch for copy_loop */
 	ldr	r2, _TEXT_BASE
 	ldr	r3, _bss_start_ofs
 	add	r2, r0, r3		/* r2 <- source end address	    */
-	cmp	r0, r6
-	beq	clear_bss
 
 copy_loop:
 	ldmia	r0!, {r9-r10}		/* copy from source address [r0]    */
-	stmia	r6!, {r9-r10}		/* copy to   target address [r1]    */
+	stmia	r1!, {r9-r10}		/* copy to   target address [r1]    */
 	cmp	r0, r2			/* until source end address [r2]    */
 	blo	copy_loop
 
@@ -216,7 +217,7 @@ copy_loop:
 	 * fix .rel.dyn relocations
 	 */
 	ldr	r0, _TEXT_BASE		/* r0 <- Text base */
-	sub	r9, r7, r0		/* r9 <- relocation offset */
+	sub	r9, r6, r0		/* r9 <- relocation offset */
 	ldr	r10, _dynsym_start_ofs	/* r10 <- sym table ofs */
 	add	r10, r10, r0		/* r10 <- sym table in FLASH */
 	ldr	r2, _rel_dyn_start_ofs	/* r2 <- rel dyn start ofs */
@@ -227,10 +228,10 @@ fixloop:
 	ldr	r0, [r2]		/* r0 <- location to fix up, IN FLASH! */
 	add	r0, r0, r9		/* r0 <- location to fix up in RAM */
 	ldr	r1, [r2, #4]
-	and	r8, r1, #0xff
-	cmp	r8, #23			/* relative fixup? */
+	and	r7, r1, #0xff
+	cmp	r7, #23			/* relative fixup? */
 	beq	fixrel
-	cmp	r8, #2			/* absolute fixup? */
+	cmp	r7, #2			/* absolute fixup? */
 	beq	fixabs
 	/* ignore unknown type of fixup */
 	b	fixnext
@@ -239,7 +240,7 @@ fixabs:
 	mov	r1, r1, LSR #4		/* r1 <- symbol index in .dynsym */
 	add	r1, r10, r1		/* r1 <- address of symbol in table */
 	ldr	r1, [r1, #4]		/* r1 <- symbol value */
-	add	r1, r9			/* r1 <- relocated sym addr */
+	add	r1, r1, r9		/* r1 <- relocated sym addr */
 	b	fixnext
 fixrel:
 	/* relative fix: increase location by offset */
@@ -257,7 +258,7 @@ clear_bss:
 	ldr	r0, _bss_start_ofs
 	ldr	r1, _bss_end_ofs
 	ldr	r3, _TEXT_BASE		/* Text base */
-	mov	r4, r7			/* reloc addr */
+	mov	r4, r6			/* reloc addr */
 	add	r0, r0, r4
 	add	r1, r1, r4
 	mov	r2, #0x00000000		/* clear			    */
@@ -288,7 +289,7 @@ _nand_boot_ofs:
 	add	lr, lr, r9
 	/* setup parameters for board_init_r */
 	mov	r0, r5		/* gd_t */
-	mov	r1, r7		/* dest_addr */
+	mov	r1, r6		/* dest_addr */
 	/* jump to it ... */
 	mov	pc, lr
 
diff --git a/arch/arm/cpu/arm926ejs/at91/at91sam9260_devices.c b/arch/arm/cpu/arm926ejs/at91/at91sam9260_devices.c
index f699f4d489..c1822b713d 100644
--- a/arch/arm/cpu/arm926ejs/at91/at91sam9260_devices.c
+++ b/arch/arm/cpu/arm926ejs/at91/at91sam9260_devices.c
@@ -204,6 +204,11 @@ void at91_macb_hw_init(void)
 #else
 	at91_set_b_periph(AT91_PIO_PORTA, 23, 0);	/* ETX2 */
 	at91_set_b_periph(AT91_PIO_PORTA, 24, 0);	/* ETX3 */
+#if defined(CONFIG_AT91SAM9G20)
+	/* 9G20 BOOT ROM initializes those pins to multi-drive, undo that */
+	at91_set_pio_multi_drive(AT91_PIO_PORTA, 23, 0);
+	at91_set_pio_multi_drive(AT91_PIO_PORTA, 24, 0);
+#endif
 #endif
 	at91_set_b_periph(AT91_PIO_PORTA, 22, 0);	/* ETXER */
 #endif
diff --git a/arch/arm/cpu/arm926ejs/davinci/timer.c b/arch/arm/cpu/arm926ejs/davinci/timer.c
index 9da7443f30..1c6fa4ae87 100644
--- a/arch/arm/cpu/arm926ejs/davinci/timer.c
+++ b/arch/arm/cpu/arm926ejs/davinci/timer.c
@@ -40,6 +40,8 @@
 #include <common.h>
 #include <asm/io.h>
 
+DECLARE_GLOBAL_DATA_PTR;
+
 struct davinci_timer {
 	u_int32_t	pid12;
 	u_int32_t	emumgt;
@@ -57,11 +59,9 @@ struct davinci_timer {
 static struct davinci_timer * const timer =
 	(struct davinci_timer *)CONFIG_SYS_TIMERBASE;
 
-#define TIMER_LOAD_VAL	(CONFIG_SYS_HZ_CLOCK / CONFIG_SYS_HZ)
-#define TIM_CLK_DIV	16
+#define TIMER_LOAD_VAL	0xffffffff
 
-static ulong timestamp;
-static ulong lastinc;
+#define TIM_CLK_DIV	16
 
 int timer_init(void)
 {
@@ -71,72 +71,51 @@ int timer_init(void)
 	writel(0x06 | ((TIM_CLK_DIV - 1) << 8), &timer->tgcr);
 	writel(0x0, &timer->tim34);
 	writel(TIMER_LOAD_VAL, &timer->prd34);
-	lastinc = 0;
-	timestamp = 0;
 	writel(2 << 22, &timer->tcr);
+	gd->timer_rate_hz = CONFIG_SYS_HZ_CLOCK / TIM_CLK_DIV;
+	gd->timer_reset_value = 0;
 
 	return(0);
 }
 
 void reset_timer(void)
 {
-	writel(0x0, &timer->tcr);
-	writel(0x0, &timer->tim34);
-	lastinc = 0;
-	timestamp = 0;
-	writel(2 << 22, &timer->tcr);
+	gd->timer_reset_value = get_ticks();
 }
 
-static ulong get_timer_raw(void)
+/*
+ * Get the current 64 bit timer tick count
+ */
+unsigned long long get_ticks(void)
 {
-	ulong now = readl(&timer->tim34);
-
-	if (now >= lastinc) {
-		/* normal mode */
-		timestamp += now - lastinc;
-	} else {
-		/* overflow ... */
-		timestamp += now + TIMER_LOAD_VAL - lastinc;
-	}
-	lastinc = now;
-	return timestamp;
+	unsigned long now = readl(&timer->tim34);
+
+	/* increment tbu if tbl has rolled over */
+	if (now < gd->tbl)
+		gd->tbu++;
+	gd->tbl = now;
+
+	return (((unsigned long long)gd->tbu) << 32) | gd->tbl;
 }
 
 ulong get_timer(ulong base)
 {
-	return((get_timer_raw() / (TIMER_LOAD_VAL / TIM_CLK_DIV)) - base);
-}
+	unsigned long long timer_diff;
 
-void set_timer(ulong t)
-{
-	timestamp = t;
+	timer_diff = get_ticks() - gd->timer_reset_value;
+
+	return (timer_diff / (gd->timer_rate_hz / CONFIG_SYS_HZ)) - base;
 }
 
 void __udelay(unsigned long usec)
 {
-	ulong tmo;
-	ulong endtime;
-	signed long diff;
-
-	tmo = CONFIG_SYS_HZ_CLOCK / 1000;
-	tmo *= usec;
-	tmo /= (1000 * TIM_CLK_DIV);
-
-	endtime = get_timer_raw() + tmo;
+	unsigned long long endtime;
 
-	do {
-		ulong now = get_timer_raw();
-		diff = endtime - now;
-	} while (diff >= 0);
-}
+	endtime = ((unsigned long long)usec * gd->timer_rate_hz) / 1000000UL;
+	endtime += get_ticks();
 
-/*
- * This function is derived from PowerPC code (read timebase as long long).
- * On ARM it just returns the timer value.
- */
-unsigned long long get_ticks(void)
-{
-	return(get_timer(0));
+	while (get_ticks() < endtime)
+		;
 }
 
 /*
diff --git a/arch/arm/cpu/arm926ejs/mx25/generic.c b/arch/arm/cpu/arm926ejs/mx25/generic.c
index b80a389ed9..c6e114634a 100644
--- a/arch/arm/cpu/arm926ejs/mx25/generic.c
+++ b/arch/arm/cpu/arm926ejs/mx25/generic.c
@@ -260,4 +260,16 @@ void mx25_fec_init_pins (void)
 	writel (outpadctl, &padctl->pad_fec_tdata1);
 
 }
+
+void imx_get_mac_from_fuse(unsigned char *mac)
+{
+	int i;
+	struct iim_regs *iim = (struct iim_regs *)IMX_IIM_BASE;
+	struct fuse_bank *bank = &iim->bank[0];
+	struct fuse_bank0_regs *fuse =
+			(struct fuse_bank0_regs *)bank->fuse_regs;
+
+	for (i = 0; i < 6; i++)
+		mac[i] = readl(&fuse->mac_addr[i]) & 0xff;
+}
 #endif /* CONFIG_FEC_MXC */
diff --git a/arch/arm/cpu/arm926ejs/mx27/generic.c b/arch/arm/cpu/arm926ejs/mx27/generic.c
index ae2ce5841c..27642bfe7d 100644
--- a/arch/arm/cpu/arm926ejs/mx27/generic.c
+++ b/arch/arm/cpu/arm926ejs/mx27/generic.c
@@ -313,6 +313,18 @@ void mx27_fec_init_pins(void)
 	for (i = 0; i < ARRAY_SIZE(mode); i++)
 		imx_gpio_mode(mode[i]);
 }
+
+void imx_get_mac_from_fuse(unsigned char *mac)
+{
+	int i;
+	struct iim_regs *iim = (struct iim_regs *)IMX_IIM_BASE;
+	struct fuse_bank *bank = &iim->bank[0];
+	struct fuse_bank0_regs *fuse =
+			(struct fuse_bank0_regs *)bank->fuse_regs;
+
+	for (i = 0; i < 6; i++)
+		mac[6 - 1 - i] = readl(&fuse->mac_addr[i]) & 0xff;
+}
 #endif /* CONFIG_FEC_MXC */
 
 #ifdef CONFIG_MXC_MMC
diff --git a/arch/arm/cpu/arm926ejs/start.S b/arch/arm/cpu/arm926ejs/start.S
index 6dcc9b4d67..55192521b7 100644
--- a/arch/arm/cpu/arm926ejs/start.S
+++ b/arch/arm/cpu/arm926ejs/start.S
@@ -174,6 +174,7 @@ reset:
 /* Set stackpointer in internal RAM to call board_init_f */
 call_board_init_f:
 	ldr	sp, =(CONFIG_SYS_INIT_SP_ADDR)
+	bic	sp, sp, #7 /* 8-byte alignment for ABI compliance */
 	ldr	r0,=0x00000000
 	bl	board_init_f
 
@@ -191,22 +192,22 @@ relocate_code:
 	mov	r4, r0	/* save addr_sp */
 	mov	r5, r1	/* save addr of gd */
 	mov	r6, r2	/* save addr of destination */
-	mov	r7, r2	/* save addr of destination */
 
 	/* Set up the stack						    */
 stack_setup:
 	mov	sp, r4
 
 	adr	r0, _start
+	cmp	r0, r6
+	beq	clear_bss		/* skip relocation */
+	mov	r1, r6			/* r1 <- scratch for copy loop */
 	ldr	r2, _TEXT_BASE
 	ldr	r3, _bss_start_ofs
 	add	r2, r0, r3		/* r2 <- source end address	    */
-	cmp	r0, r6
-	beq	clear_bss
 
 copy_loop:
 	ldmia	r0!, {r9-r10}		/* copy from source address [r0]    */
-	stmia	r6!, {r9-r10}		/* copy to   target address [r1]    */
+	stmia	r1!, {r9-r10}		/* copy to   target address [r1]    */
 	cmp	r0, r2			/* until source end address [r2]    */
 	blo	copy_loop
 
@@ -215,7 +216,7 @@ copy_loop:
 	 * fix .rel.dyn relocations
 	 */
 	ldr	r0, _TEXT_BASE		/* r0 <- Text base */
-	sub	r9, r7, r0		/* r9 <- relocation offset */
+	sub	r9, r6, r0		/* r9 <- relocation offset */
 	ldr	r10, _dynsym_start_ofs	/* r10 <- sym table ofs */
 	add	r10, r10, r0		/* r10 <- sym table in FLASH */
 	ldr	r2, _rel_dyn_start_ofs	/* r2 <- rel dyn start ofs */
@@ -226,10 +227,10 @@ fixloop:
 	ldr	r0, [r2]		/* r0 <- location to fix up, IN FLASH! */
 	add	r0, r0, r9		/* r0 <- location to fix up in RAM */
 	ldr	r1, [r2, #4]
-	and	r8, r1, #0xff
-	cmp	r8, #23			/* relative fixup? */
+	and	r7, r1, #0xff
+	cmp	r7, #23			/* relative fixup? */
 	beq	fixrel
-	cmp	r8, #2			/* absolute fixup? */
+	cmp	r7, #2			/* absolute fixup? */
 	beq	fixabs
 	/* ignore unknown type of fixup */
 	b	fixnext
@@ -238,7 +239,7 @@ fixabs:
 	mov	r1, r1, LSR #4		/* r1 <- symbol index in .dynsym */
 	add	r1, r10, r1		/* r1 <- address of symbol in table */
 	ldr	r1, [r1, #4]		/* r1 <- symbol value */
-	add	r1, r9			/* r1 <- relocated sym addr */
+	add	r1, r1, r9		/* r1 <- relocated sym addr */
 	b	fixnext
 fixrel:
 	/* relative fix: increase location by offset */
@@ -256,7 +257,7 @@ clear_bss:
 	ldr	r0, _bss_start_ofs
 	ldr	r1, _bss_end_ofs
 	ldr	r3, _TEXT_BASE		/* Text base */
-	mov	r4, r7			/* reloc addr */
+	mov	r4, r6			/* reloc addr */
 	add	r0, r0, r4
 	add	r1, r1, r4
 	mov	r2, #0x00000000		/* clear			    */
@@ -287,7 +288,7 @@ _nand_boot_ofs:
 	add	lr, lr, r9
 	/* setup parameters for board_init_r */
 	mov	r0, r5		/* gd_t */
-	mov	r1, r7		/* dest_addr */
+	mov	r1, r6		/* dest_addr */
 	/* jump to it ... */
 	mov	pc, lr
 
diff --git a/arch/arm/cpu/arm946es/start.S b/arch/arm/cpu/arm946es/start.S
index f4950f7072..f9c947024f 100644
--- a/arch/arm/cpu/arm946es/start.S
+++ b/arch/arm/cpu/arm946es/start.S
@@ -146,6 +146,7 @@ reset:
 /* Set stackpointer in internal RAM to call board_init_f */
 call_board_init_f:
 	ldr	sp, =(CONFIG_SYS_INIT_SP_ADDR)
+	bic	sp, sp, #7 /* 8-byte alignment for ABI compliance */
 	ldr	r0,=0x00000000
 	bl	board_init_f
 
@@ -163,22 +164,22 @@ relocate_code:
 	mov	r4, r0	/* save addr_sp */
 	mov	r5, r1	/* save addr of gd */
 	mov	r6, r2	/* save addr of destination */
-	mov	r7, r2	/* save addr of destination */
 
 	/* Set up the stack						    */
 stack_setup:
 	mov	sp, r4
 
 	adr	r0, _start
+	cmp	r0, r6
+	beq	clear_bss		/* skip relocation */
+	mov	r1, r6			/* r1 <- scratch for copy_loop */
 	ldr	r2, _TEXT_BASE
 	ldr	r3, _bss_start_ofs
 	add	r2, r0, r3		/* r2 <- source end address	    */
-	cmp	r0, r6
-	beq	clear_bss
 
 copy_loop:
 	ldmia	r0!, {r9-r10}		/* copy from source address [r0]    */
-	stmia	r6!, {r9-r10}		/* copy to   target address [r1]    */
+	stmia	r1!, {r9-r10}		/* copy to   target address [r1]    */
 	cmp	r0, r2			/* until source end address [r2]    */
 	blo	copy_loop
 
@@ -187,7 +188,7 @@ copy_loop:
 	 * fix .rel.dyn relocations
 	 */
 	ldr	r0, _TEXT_BASE		/* r0 <- Text base */
-	sub	r9, r7, r0		/* r9 <- relocation offset */
+	sub	r9, r6, r0		/* r9 <- relocation offset */
 	ldr	r10, _dynsym_start_ofs	/* r10 <- sym table ofs */
 	add	r10, r10, r0		/* r10 <- sym table in FLASH */
 	ldr	r2, _rel_dyn_start_ofs	/* r2 <- rel dyn start ofs */
@@ -198,10 +199,10 @@ fixloop:
 	ldr	r0, [r2]		/* r0 <- location to fix up, IN FLASH! */
 	add	r0, r0, r9		/* r0 <- location to fix up in RAM */
 	ldr	r1, [r2, #4]
-	and	r8, r1, #0xff
-	cmp	r8, #23			/* relative fixup? */
+	and	r7, r1, #0xff
+	cmp	r7, #23			/* relative fixup? */
 	beq	fixrel
-	cmp	r8, #2			/* absolute fixup? */
+	cmp	r7, #2			/* absolute fixup? */
 	beq	fixabs
 	/* ignore unknown type of fixup */
 	b	fixnext
@@ -210,7 +211,7 @@ fixabs:
 	mov	r1, r1, LSR #4		/* r1 <- symbol index in .dynsym */
 	add	r1, r10, r1		/* r1 <- address of symbol in table */
 	ldr	r1, [r1, #4]		/* r1 <- symbol value */
-	add	r1, r9			/* r1 <- relocated sym addr */
+	add	r1, r1, r9		/* r1 <- relocated sym addr */
 	b	fixnext
 fixrel:
 	/* relative fix: increase location by offset */
@@ -228,7 +229,7 @@ clear_bss:
 	ldr	r0, _bss_start_ofs
 	ldr	r1, _bss_end_ofs
 	ldr	r3, _TEXT_BASE		/* Text base */
-	mov	r4, r7			/* reloc addr */
+	mov	r4, r6			/* reloc addr */
 	add	r0, r0, r4
 	add	r1, r1, r4
 	mov	r2, #0x00000000		/* clear			    */
@@ -254,7 +255,7 @@ _nand_boot: .word nand_boot
 	add	lr, lr, r9
 	/* setup parameters for board_init_r */
 	mov	r0, r5		/* gd_t */
-	mov	r1, r7		/* dest_addr */
+	mov	r1, r6		/* dest_addr */
 	/* jump to it ... */
 	mov	pc, lr
 
diff --git a/arch/arm/cpu/arm_intcm/start.S b/arch/arm/cpu/arm_intcm/start.S
index fcaba30f58..d28e745b2d 100644
--- a/arch/arm/cpu/arm_intcm/start.S
+++ b/arch/arm/cpu/arm_intcm/start.S
@@ -142,6 +142,7 @@ reset:
 /* Set stackpointer in internal RAM to call board_init_f */
 call_board_init_f:
 	ldr	sp, =(CONFIG_SYS_INIT_SP_ADDR)
+	bic	sp, sp, #7 /* 8-byte alignment for ABI compliance */
 	ldr	r0,=0x00000000
 	bl	board_init_f
 
@@ -159,22 +160,22 @@ relocate_code:
 	mov	r4, r0	/* save addr_sp */
 	mov	r5, r1	/* save addr of gd */
 	mov	r6, r2	/* save addr of destination */
-	mov	r7, r2	/* save addr of destination */
 
 	/* Set up the stack						    */
 stack_setup:
 	mov	sp, r4
 
 	adr	r0, _start
+	cmp	r0, r6
+	beq	clear_bss		/* skip relocation */
+	mov	r1, r6			/* r1 <- scratch for copy_loop */
 	ldr	r2, _TEXT_BASE
 	ldr	r3, _bss_start_ofs
 	add	r2, r0, r3		/* r2 <- source end address	    */
-	cmp	r0, r6
-	beq	clear_bss
 
 copy_loop:
 	ldmia	r0!, {r9-r10}		/* copy from source address [r0]    */
-	stmia	r6!, {r9-r10}		/* copy to   target address [r1]    */
+	stmia	r1!, {r9-r10}		/* copy to   target address [r1]    */
 	cmp	r0, r2			/* until source end address [r2]    */
 	blo	copy_loop
 
@@ -183,7 +184,7 @@ copy_loop:
 	 * fix .rel.dyn relocations
 	 */
 	ldr	r0, _TEXT_BASE		/* r0 <- Text base */
-	sub	r9, r7, r0		/* r9 <- relocation offset */
+	sub	r9, r6, r0		/* r9 <- relocation offset */
 	ldr	r10, _dynsym_start_ofs	/* r10 <- sym table ofs */
 	add	r10, r10, r0		/* r10 <- sym table in FLASH */
 	ldr	r2, _rel_dyn_start_ofs	/* r2 <- rel dyn start ofs */
@@ -194,10 +195,10 @@ fixloop:
 	ldr	r0, [r2]		/* r0 <- location to fix up, IN FLASH! */
 	add	r0, r0, r9		/* r0 <- location to fix up in RAM */
 	ldr	r1, [r2, #4]
-	and	r8, r1, #0xff
-	cmp	r8, #23			/* relative fixup? */
+	and	r7, r1, #0xff
+	cmp	r7, #23			/* relative fixup? */
 	beq	fixrel
-	cmp	r8, #2			/* absolute fixup? */
+	cmp	r7, #2			/* absolute fixup? */
 	beq	fixabs
 	/* ignore unknown type of fixup */
 	b	fixnext
@@ -206,7 +207,7 @@ fixabs:
 	mov	r1, r1, LSR #4		/* r1 <- symbol index in .dynsym */
 	add	r1, r10, r1		/* r1 <- address of symbol in table */
 	ldr	r1, [r1, #4]		/* r1 <- symbol value */
-	add	r1, r9			/* r1 <- relocated sym addr */
+	add	r1, r1, r9		/* r1 <- relocated sym addr */
 	b	fixnext
 fixrel:
 	/* relative fix: increase location by offset */
@@ -224,7 +225,7 @@ clear_bss:
 	ldr	r0, _bss_start_ofs
 	ldr	r1, _bss_end_ofs
 	ldr	r3, _TEXT_BASE		/* Text base */
-	mov	r4, r7			/* reloc addr */
+	mov	r4, r6			/* reloc addr */
 	add	r0, r0, r4
 	add	r1, r1, r4
 	mov	r2, #0x00000000		/* clear			    */
@@ -255,7 +256,7 @@ _nand_boot_ofs:
 	add	lr, lr, r9
 	/* setup parameters for board_init_r */
 	mov	r0, r5		/* gd_t */
-	mov	r1, r7		/* dest_addr */
+	mov	r1, r6		/* dest_addr */
 	/* jump to it ... */
 	mov	pc, lr
 
diff --git a/arch/arm/cpu/armv7/mx5/soc.c b/arch/arm/cpu/armv7/mx5/soc.c
index 7c7a565665..290011923e 100644
--- a/arch/arm/cpu/armv7/mx5/soc.c
+++ b/arch/arm/cpu/armv7/mx5/soc.c
@@ -100,6 +100,20 @@ int cpu_eth_init(bd_t *bis)
 	return rc;
 }
 
+#if defined(CONFIG_FEC_MXC)
+void imx_get_mac_from_fuse(unsigned char *mac)
+{
+	int i;
+	struct iim_regs *iim = (struct iim_regs *)IMX_IIM_BASE;
+	struct fuse_bank *bank = &iim->bank[1];
+	struct fuse_bank1_regs *fuse =
+			(struct fuse_bank1_regs *)bank->fuse_regs;
+
+	for (i = 0; i < 6; i++)
+		mac[i] = readl(&fuse->mac_addr[i]) & 0xff;
+}
+#endif
+
 /*
  * Initializes on-chip MMC controllers.
  * to override, implement board_mmc_init()
diff --git a/arch/arm/cpu/armv7/omap-common/timer.c b/arch/arm/cpu/armv7/omap-common/timer.c
index 6b8cf7ba2d..9beebb1e74 100644
--- a/arch/arm/cpu/armv7/omap-common/timer.c
+++ b/arch/arm/cpu/armv7/omap-common/timer.c
@@ -35,8 +35,8 @@
 #include <common.h>
 #include <asm/io.h>
 
-static ulong timestamp;
-static ulong lastinc;
+DECLARE_GLOBAL_DATA_PTR;
+
 static struct gptimer *timer_base = (struct gptimer *)CONFIG_SYS_TIMERBASE;
 
 /*
@@ -74,7 +74,7 @@ ulong get_timer(ulong base)
 
 void set_timer(ulong t)
 {
-	timestamp = t;
+	gd->tbl = t;
 }
 
 /* delay x useconds */
@@ -96,8 +96,8 @@ void __udelay(unsigned long usec)
 void reset_timer_masked(void)
 {
 	/* reset time, capture current incrementer value time */
-	lastinc = readl(&timer_base->tcrr) / (TIMER_CLOCK / CONFIG_SYS_HZ);
-	timestamp = 0;		/* start "advancing" time stamp from 0 */
+	gd->lastinc = readl(&timer_base->tcrr) / (TIMER_CLOCK / CONFIG_SYS_HZ);
+	gd->tbl = 0;		/* start "advancing" time stamp from 0 */
 }
 
 ulong get_timer_masked(void)
@@ -105,14 +105,14 @@ ulong get_timer_masked(void)
 	/* current tick value */
 	ulong now = readl(&timer_base->tcrr) / (TIMER_CLOCK / CONFIG_SYS_HZ);
 
-	if (now >= lastinc)	/* normal mode (non roll) */
+	if (now >= gd->lastinc)	/* normal mode (non roll) */
 		/* move stamp fordward with absoulte diff ticks */
-		timestamp += (now - lastinc);
+		gd->tbl += (now - gd->lastinc);
 	else	/* we have rollover of incrementer */
-		timestamp += ((TIMER_LOAD_VAL / (TIMER_CLOCK / CONFIG_SYS_HZ))
-				- lastinc) + now;
-	lastinc = now;
-	return timestamp;
+		gd->tbl += ((TIMER_LOAD_VAL / (TIMER_CLOCK / CONFIG_SYS_HZ))
+			     - gd->lastinc) + now;
+	gd->lastinc = now;
+	return gd->tbl;
 }
 
 /*
diff --git a/arch/arm/cpu/armv7/omap3/emif4.c b/arch/arm/cpu/armv7/omap3/emif4.c
index 0870857ede..30856372b3 100644
--- a/arch/arm/cpu/armv7/omap3/emif4.c
+++ b/arch/arm/cpu/armv7/omap3/emif4.c
@@ -29,6 +29,7 @@
 #include <asm/arch/sys_proto.h>
 #include <asm/arch/emif4.h>
 
+DECLARE_GLOBAL_DATA_PTR;
 extern omap3_sysinfo sysinfo;
 
 static emif4_t *emif4_base = (emif4_t *)OMAP34XX_SDRC_BASE;
@@ -48,10 +49,11 @@ u32 is_mem_sdr(void)
  */
 u32 get_sdr_cs_size(u32 cs)
 {
-	u32 size;
+	u32 size = 0;
 
 	/* TODO: Calculate the size based on EMIF4 configuration */
-	size = CONFIG_SYS_CS0_SIZE;
+	if (cs == CS0)
+		size = CONFIG_SYS_CS0_SIZE;
 
 	return size;
 }
@@ -138,7 +140,6 @@ void do_emif4_init(void)
  */
 int dram_init(void)
 {
-	DECLARE_GLOBAL_DATA_PTR;
 	unsigned int size0 = 0, size1 = 0;
 
 	size0 = get_sdr_cs_size(CS0);
@@ -156,7 +157,6 @@ int dram_init(void)
 
 void dram_init_banksize (void)
 {
-	DECLARE_GLOBAL_DATA_PTR;
 	unsigned int size0 = 0, size1 = 0;
 
 	size0 = get_sdr_cs_size(CS0);
diff --git a/arch/arm/cpu/armv7/omap3/sdrc.c b/arch/arm/cpu/armv7/omap3/sdrc.c
index a4979ce61d..2a7970b4d0 100644
--- a/arch/arm/cpu/armv7/omap3/sdrc.c
+++ b/arch/arm/cpu/armv7/omap3/sdrc.c
@@ -37,6 +37,7 @@
 #include <asm/arch/mem.h>
 #include <asm/arch/sys_proto.h>
 
+DECLARE_GLOBAL_DATA_PTR;
 extern omap3_sysinfo sysinfo;
 
 static struct sdrc *sdrc_base = (struct sdrc *)OMAP34XX_SDRC_BASE;
@@ -172,7 +173,6 @@ void do_sdrc_init(u32 cs, u32 early)
  */
 int dram_init(void)
 {
-	DECLARE_GLOBAL_DATA_PTR;
 	unsigned int size0 = 0, size1 = 0;
 
 	size0 = get_sdr_cs_size(CS0);
@@ -194,7 +194,6 @@ int dram_init(void)
 
 void dram_init_banksize (void)
 {
-	DECLARE_GLOBAL_DATA_PTR;
 	unsigned int size0 = 0, size1 = 0;
 
 	size0 = get_sdr_cs_size(CS0);
diff --git a/arch/arm/cpu/armv7/s5p-common/timer.c b/arch/arm/cpu/armv7/s5p-common/timer.c
index 04906503e6..651fd5ddff 100644
--- a/arch/arm/cpu/armv7/s5p-common/timer.c
+++ b/arch/arm/cpu/armv7/s5p-common/timer.c
@@ -65,15 +65,12 @@ int timer_init(void)
 	writel((PRESCALER_1 & 0xff) << 8, &timer->tcfg0);
 	writel((MUX_DIV_2 & 0xf) << MUX4_DIV_SHIFT, &timer->tcfg1);
 
-	if (count_value == 0) {
-		/* reset initial value */
-		/* count_value = 2085937.5(HZ) (per 1 sec)*/
-		count_value = get_pwm_clk() / ((PRESCALER_1 + 1) *
-				(MUX_DIV_2 + 1));
-
-		/* count_value / 100 = 20859.375(HZ) (per 10 msec) */
-		count_value = count_value / 100;
-	}
+	/* count_value = 2085937.5(HZ) (per 1 sec)*/
+	count_value = get_pwm_clk() / ((PRESCALER_1 + 1) *
+			(MUX_DIV_2 + 1));
+
+	/* count_value / 100 = 20859.375(HZ) (per 10 msec) */
+	count_value = count_value / 100;
 
 	/* set count value */
 	writel(count_value, &timer->tcntb4);
@@ -114,8 +111,11 @@ void set_timer(unsigned long t)
 /* delay x useconds */
 void __udelay(unsigned long usec)
 {
+	struct s5p_timer *const timer = s5p_get_base_timer();
 	unsigned long tmo, tmp;
 
+	count_value = readl(&timer->tcntb4);
+
 	if (usec >= 1000) {
 		/*
 		 * if "big" number, spread normalization
diff --git a/arch/arm/cpu/armv7/start.S b/arch/arm/cpu/armv7/start.S
index 2dfdafe982..684f2d2538 100644
--- a/arch/arm/cpu/armv7/start.S
+++ b/arch/arm/cpu/armv7/start.S
@@ -142,6 +142,7 @@ next:
 /* Set stackpointer in internal RAM to call board_init_f */
 call_board_init_f:
 	ldr	sp, =(CONFIG_SYS_INIT_SP_ADDR)
+	bic	sp, sp, #7 /* 8-byte alignment for ABI compliance */
 	ldr	r0,=0x00000000
 	bl	board_init_f
 
@@ -159,24 +160,24 @@ relocate_code:
 	mov	r4, r0	/* save addr_sp */
 	mov	r5, r1	/* save addr of gd */
 	mov	r6, r2	/* save addr of destination */
-	mov	r7, r2	/* save addr of destination */
 
 	/* Set up the stack						    */
 stack_setup:
 	mov	sp, r4
 
 	adr	r0, _start
+#ifndef CONFIG_PRELOADER
+	cmp	r0, r6
+	beq	clear_bss		/* skip relocation */
+#endif
+	mov	r1, r6			/* r1 <- scratch for copy_loop */
 	ldr	r2, _TEXT_BASE
 	ldr	r3, _bss_start_ofs
 	add	r2, r0, r3		/* r2 <- source end address	    */
-	cmp	r0, r6
-#ifndef CONFIG_PRELOADER
-	beq	jump_2_ram
-#endif
 
 copy_loop:
 	ldmia	r0!, {r9-r10}		/* copy from source address [r0]    */
-	stmia	r6!, {r9-r10}		/* copy to   target address [r1]    */
+	stmia	r1!, {r9-r10}		/* copy to   target address [r1]    */
 	cmp	r0, r2			/* until source end address [r2]    */
 	blo	copy_loop
 
@@ -185,7 +186,7 @@ copy_loop:
 	 * fix .rel.dyn relocations
 	 */
 	ldr	r0, _TEXT_BASE		/* r0 <- Text base */
-	sub	r9, r7, r0		/* r9 <- relocation offset */
+	sub	r9, r6, r0		/* r9 <- relocation offset */
 	ldr	r10, _dynsym_start_ofs	/* r10 <- sym table ofs */
 	add	r10, r10, r0		/* r10 <- sym table in FLASH */
 	ldr	r2, _rel_dyn_start_ofs	/* r2 <- rel dyn start ofs */
@@ -196,10 +197,10 @@ fixloop:
 	ldr	r0, [r2]		/* r0 <- location to fix up, IN FLASH! */
 	add	r0, r0, r9		/* r0 <- location to fix up in RAM */
 	ldr	r1, [r2, #4]
-	and	r8, r1, #0xff
-	cmp	r8, #23			/* relative fixup? */
+	and	r7, r1, #0xff
+	cmp	r7, #23			/* relative fixup? */
 	beq	fixrel
-	cmp	r8, #2			/* absolute fixup? */
+	cmp	r7, #2			/* absolute fixup? */
 	beq	fixabs
 	/* ignore unknown type of fixup */
 	b	fixnext
@@ -208,7 +209,7 @@ fixabs:
 	mov	r1, r1, LSR #4		/* r1 <- symbol index in .dynsym */
 	add	r1, r10, r1		/* r1 <- address of symbol in table */
 	ldr	r1, [r1, #4]		/* r1 <- symbol value */
-	add	r1, r9			/* r1 <- relocated sym addr */
+	add	r1, r1, r9		/* r1 <- relocated sym addr */
 	b	fixnext
 fixrel:
 	/* relative fix: increase location by offset */
@@ -224,7 +225,7 @@ clear_bss:
 	ldr	r0, _bss_start_ofs
 	ldr	r1, _bss_end_ofs
 	ldr	r3, _TEXT_BASE		/* Text base */
-	mov	r4, r7			/* reloc addr */
+	mov	r4, r6			/* reloc addr */
 	add	r0, r0, r4
 	add	r1, r1, r4
 	mov	r2, #0x00000000		/* clear			    */
@@ -246,7 +247,7 @@ jump_2_ram:
 	add	lr, lr, r9
 	/* setup parameters for board_init_r */
 	mov	r0, r5		/* gd_t */
-	mov	r1, r7		/* dest_addr */
+	mov	r1, r6		/* dest_addr */
 	/* jump to it ... */
 	mov	pc, lr
 
diff --git a/arch/arm/cpu/armv7/syslib.c b/arch/arm/cpu/armv7/syslib.c
index f9ed9a307b..84d17f0f21 100644
--- a/arch/arm/cpu/armv7/syslib.c
+++ b/arch/arm/cpu/armv7/syslib.c
@@ -23,7 +23,6 @@
 
 #include <common.h>
 #include <asm/io.h>
-#include <asm/arch/sys_proto.h>
 
 /************************************************************
  * sdelay() - simple spin loop.  Will be constant time as
diff --git a/arch/arm/cpu/ixp/start.S b/arch/arm/cpu/ixp/start.S
index d1ab19bf5d..9f8c15b877 100644
--- a/arch/arm/cpu/ixp/start.S
+++ b/arch/arm/cpu/ixp/start.S
@@ -268,6 +268,7 @@ reset:
 /* Set stackpointer in internal RAM to call board_init_f */
 call_board_init_f:
 	ldr	sp, =(CONFIG_SYS_INIT_SP_ADDR)
+	bic	sp, sp, #7 /* 8-byte alignment for ABI compliance */
 	ldr	r0,=0x00000000
 	bl	board_init_f
 
@@ -285,22 +286,22 @@ relocate_code:
 	mov	r4, r0	/* save addr_sp */
 	mov	r5, r1	/* save addr of gd */
 	mov	r6, r2	/* save addr of destination */
-	mov	r7, r2	/* save addr of destination */
 
 	/* Set up the stack						    */
 stack_setup:
 	mov	sp, r4
 
 	adr	r0, _start
+	cmp	r0, r6
+	beq	clear_bss		/* skip relocation */
+	mov	r1, r6			/* r1 <- scratch for copy_loop */
 	ldr	r2, _TEXT_BASE
 	ldr	r3, _bss_start_ofs
 	add	r2, r0, r3		/* r2 <- source end address	    */
-	cmp	r0, r6
-	beq	clear_bss
 
 copy_loop:
 	ldmia	r0!, {r9-r10}		/* copy from source address [r0]    */
-	stmia	r6!, {r9-r10}		/* copy to   target address [r1]    */
+	stmia	r1!, {r9-r10}		/* copy to   target address [r1]    */
 	cmp	r0, r2			/* until source end address [r2]    */
 	blo	copy_loop
 
@@ -309,7 +310,7 @@ copy_loop:
 	 * fix .rel.dyn relocations
 	 */
 	ldr	r0, _TEXT_BASE		/* r0 <- Text base */
-	sub	r9, r7, r0		/* r9 <- relocation offset */
+	sub	r9, r6, r0		/* r9 <- relocation offset */
 	ldr	r10, _dynsym_start_ofs	/* r10 <- sym table ofs */
 	add	r10, r10, r0		/* r10 <- sym table in FLASH */
 	ldr	r2, _rel_dyn_start_ofs	/* r2 <- rel dyn start ofs */
@@ -320,10 +321,10 @@ fixloop:
 	ldr	r0, [r2]		/* r0 <- location to fix up, IN FLASH! */
 	add	r0, r0, r9		/* r0 <- location to fix up in RAM */
 	ldr	r1, [r2, #4]
-	and	r8, r1, #0xff
-	cmp	r8, #23			/* relative fixup? */
+	and	r7, r1, #0xff
+	cmp	r7, #23			/* relative fixup? */
 	beq	fixrel
-	cmp	r8, #2			/* absolute fixup? */
+	cmp	r7, #2			/* absolute fixup? */
 	beq	fixabs
 	/* ignore unknown type of fixup */
 	b	fixnext
@@ -332,7 +333,7 @@ fixabs:
 	mov	r1, r1, LSR #4		/* r1 <- symbol index in .dynsym */
 	add	r1, r10, r1		/* r1 <- address of symbol in table */
 	ldr	r1, [r1, #4]		/* r1 <- symbol value */
-	add	r1, r9			/* r1 <- relocated sym addr */
+	add	r1, r1, r9		/* r1 <- relocated sym addr */
 	b	fixnext
 fixrel:
 	/* relative fix: increase location by offset */
@@ -350,7 +351,7 @@ clear_bss:
 	ldr	r0, _bss_start_ofs
 	ldr	r1, _bss_end_ofs
 	ldr	r3, _TEXT_BASE		/* Text base */
-	mov	r4, r7			/* reloc addr */
+	mov	r4, r6			/* reloc addr */
 	add	r0, r0, r4
 	add	r1, r1, r4
 	mov	r2, #0x00000000		/* clear			    */
@@ -374,7 +375,7 @@ clbss_l:str	r2, [r0]		/* clear loop...		    */
 	add	lr, lr, r9
 	/* setup parameters for board_init_r */
 	mov	r0, r5		/* gd_t */
-	mov	r1, r7		/* dest_addr */
+	mov	r1, r6		/* dest_addr */
 	/* jump to it ... */
 	mov	pc, lr
 
diff --git a/arch/arm/cpu/lh7a40x/start.S b/arch/arm/cpu/lh7a40x/start.S
index 5cf477b1a3..32dfe8b5f7 100644
--- a/arch/arm/cpu/lh7a40x/start.S
+++ b/arch/arm/cpu/lh7a40x/start.S
@@ -155,6 +155,7 @@ reset:
 /* Set stackpointer in internal RAM to call board_init_f */
 call_board_init_f:
 	ldr	sp, =(CONFIG_SYS_INIT_SP_ADDR)
+	bic	sp, sp, #7 /* 8-byte alignment for ABI compliance */
 	ldr	r0,=0x00000000
 	bl	board_init_f
 
@@ -172,22 +173,22 @@ relocate_code:
 	mov	r4, r0	/* save addr_sp */
 	mov	r5, r1	/* save addr of gd */
 	mov	r6, r2	/* save addr of destination */
-	mov	r7, r2	/* save addr of destination */
 
 	/* Set up the stack						    */
 stack_setup:
 	mov	sp, r4
 
 	adr	r0, _start
+	cmp	r0, r6
+	beq	clear_bss		/* skip relocation */
+	mov	r1, r6			/* r1 <- scratch for copy_loop */
 	ldr	r2, _TEXT_BASE
 	ldr	r3, _bss_start_ofs
 	add	r2, r0, r3		/* r2 <- source end address	    */
-	cmp	r0, r6
-	beq	clear_bss
 
 copy_loop:
 	ldmia	r0!, {r9-r10}		/* copy from source address [r0]    */
-	stmia	r6!, {r9-r10}		/* copy to   target address [r1]    */
+	stmia	r1!, {r9-r10}		/* copy to   target address [r1]    */
 	cmp	r0, r2			/* until source end address [r2]    */
 	blo	copy_loop
 
@@ -196,7 +197,7 @@ copy_loop:
 	 * fix .rel.dyn relocations
 	 */
 	ldr	r0, _TEXT_BASE		/* r0 <- Text base */
-	sub	r9, r7, r0		/* r9 <- relocation offset */
+	sub	r9, r6, r0		/* r9 <- relocation offset */
 	ldr	r10, _dynsym_start_ofs	/* r10 <- sym table ofs */
 	add	r10, r10, r0		/* r10 <- sym table in FLASH */
 	ldr	r2, _rel_dyn_start_ofs	/* r2 <- rel dyn start ofs */
@@ -207,10 +208,10 @@ fixloop:
 	ldr	r0, [r2]		/* r0 <- location to fix up, IN FLASH! */
 	add	r0, r0, r9		/* r0 <- location to fix up in RAM */
 	ldr	r1, [r2, #4]
-	and	r8, r1, #0xff
-	cmp	r8, #23			/* relative fixup? */
+	and	r7, r1, #0xff
+	cmp	r7, #23			/* relative fixup? */
 	beq	fixrel
-	cmp	r8, #2			/* absolute fixup? */
+	cmp	r7, #2			/* absolute fixup? */
 	beq	fixabs
 	/* ignore unknown type of fixup */
 	b	fixnext
@@ -219,7 +220,7 @@ fixabs:
 	mov	r1, r1, LSR #4		/* r1 <- symbol index in .dynsym */
 	add	r1, r10, r1		/* r1 <- address of symbol in table */
 	ldr	r1, [r1, #4]		/* r1 <- symbol value */
-	add	r1, r9			/* r1 <- relocated sym addr */
+	add	r1, r1, r9		/* r1 <- relocated sym addr */
 	b	fixnext
 fixrel:
 	/* relative fix: increase location by offset */
@@ -237,7 +238,7 @@ clear_bss:
 	ldr	r0, _bss_start_ofs
 	ldr	r1, _bss_end_ofs
 	ldr	r3, _TEXT_BASE		/* Text base */
-	mov	r4, r7			/* reloc addr */
+	mov	r4, r6			/* reloc addr */
 	add	r0, r0, r4
 	add	r1, r1, r4
 	mov	r2, #0x00000000		/* clear			    */
@@ -258,7 +259,7 @@ clbss_l:str	r2, [r0]		/* clear loop...		    */
 	add	lr, lr, r9
 	/* setup parameters for board_init_r */
 	mov	r0, r5		/* gd_t */
-	mov	r1, r7		/* dest_addr */
+	mov	r1, r6		/* dest_addr */
 	/* jump to it ... */
 	mov	pc, lr
 
diff --git a/arch/arm/cpu/pxa/start.S b/arch/arm/cpu/pxa/start.S
index ae358a56e6..fbd0def465 100644
--- a/arch/arm/cpu/pxa/start.S
+++ b/arch/arm/cpu/pxa/start.S
@@ -220,6 +220,7 @@ zerojmp:
 /* Set stackpointer in internal RAM to call board_init_f */
 call_board_init_f:
 	ldr	sp, =(CONFIG_SYS_INIT_SP_ADDR)
+	bic	sp, sp, #7 /* 8-byte alignment for ABI compliance */
 	ldr	r0,=0x00000000
 	bl	board_init_f
 
@@ -237,23 +238,23 @@ relocate_code:
 	mov	r4, r0	/* save addr_sp */
 	mov	r5, r1	/* save addr of gd */
 	mov	r6, r2	/* save addr of destination */
-	mov	r7, r2	/* save addr of destination */
 
 	/* Set up the stack						    */
 stack_setup:
 	mov	sp, r4
 
 	adr	r0, _start
+	cmp	r0, r6
+	beq	clear_bss		/* skip relocation */
+	mov	r1, r6			/* r1 <- scratch for copy_loop */
 	ldr	r2, _TEXT_BASE
 	ldr	r3, _bss_start_ofs
 	add	r2, r0, r3		/* r2 <- source end address	    */
-	cmp	r0, r6
-	beq	clear_bss
 
 	stmfd sp!, {r0-r12}
 copy_loop:
 	ldmia	r0!, {r3-r5, r7-r11}	/* copy from source address [r0]    */
-	stmia	r6!, {r3-r5, r7-r11}	/* copy to   target address [r1]    */
+	stmia	r1!, {r3-r5, r7-r11}	/* copy to   target address [r1]    */
 	cmp	r0, r2			/* until source end address [r2]    */
 	blo	copy_loop
 	ldmfd sp!, {r0-r12}
@@ -263,7 +264,7 @@ copy_loop:
 	 * fix .rel.dyn relocations
 	 */
 	ldr	r0, _TEXT_BASE		/* r0 <- Text base */
-	sub	r9, r7, r0		/* r9 <- relocation offset */
+	sub	r9, r6, r0		/* r9 <- relocation offset */
 	ldr	r10, _dynsym_start_ofs	/* r10 <- sym table ofs */
 	add	r10, r10, r0		/* r10 <- sym table in FLASH */
 	ldr	r2, _rel_dyn_start_ofs	/* r2 <- rel dyn start ofs */
@@ -274,10 +275,10 @@ fixloop:
 	ldr	r0, [r2]	/* r0 <- location to fix up, IN FLASH! */
 	add	r0, r9		/* r0 <- location to fix up in RAM */
 	ldr	r1, [r2, #4]
-	and	r8, r1, #0xff
-	cmp	r8, #23		/* relative fixup? */
+	and	r7, r1, #0xff
+	cmp	r7, #23		/* relative fixup? */
 	beq	fixrel
-	cmp	r8, #2		/* absolute fixup? */
+	cmp	r7, #2		/* absolute fixup? */
 	beq	fixabs
 	/* ignore unknown type of fixup */
 	b	fixnext
@@ -286,7 +287,7 @@ fixabs:
 	mov	r1, r1, LSR #4		/* r1 <- symbol index in .dynsym */
 	add	r1, r10, r1		/* r1 <- address of symbol in table */
 	ldr	r1, [r1, #4]		/* r1 <- symbol value */
-	add	r1, r9			/* r1 <- relocated sym addr */
+	add	r1, r1, r9		/* r1 <- relocated sym addr */
 	b	fixnext
 fixrel:
 	/* relative fix: increase location by offset */
@@ -304,7 +305,7 @@ clear_bss:
 	ldr	r0, _bss_start_ofs
 	ldr	r1, _bss_end_ofs
 	ldr	r3, _TEXT_BASE		/* Text base */
-	mov	r4, r7			/* reloc addr */
+	mov	r4, r6			/* reloc addr */
 	add	r0, r0, r4
 	add	r1, r1, r4
 	mov	r2, #0x00000000		/* clear			    */
@@ -332,7 +333,7 @@ _start_oneboot_ofs
 	add	lr, lr, r9
 	/* setup parameters for board_init_r */
 	mov	r0, r5		/* gd_t */
-	mov	r1, r7		/* dest_addr */
+	mov	r1, r6		/* dest_addr */
 	/* jump to it ... */
 	mov	pc, lr
 
diff --git a/arch/arm/cpu/s3c44b0/start.S b/arch/arm/cpu/s3c44b0/start.S
index 8c7100c2bf..9379af67b2 100644
--- a/arch/arm/cpu/s3c44b0/start.S
+++ b/arch/arm/cpu/s3c44b0/start.S
@@ -127,6 +127,7 @@ reset:
 /* Set stackpointer in internal RAM to call board_init_f */
 call_board_init_f:
 	ldr	sp, =(CONFIG_SYS_INIT_SP_ADDR)
+	bic	sp, sp, #7 /* 8-byte alignment for ABI compliance */
 	ldr	r0,=0x00000000
 	bl	board_init_f
 
@@ -144,22 +145,22 @@ relocate_code:
 	mov	r4, r0	/* save addr_sp */
 	mov	r5, r1	/* save addr of gd */
 	mov	r6, r2	/* save addr of destination */
-	mov	r7, r2	/* save addr of destination */
 
 	/* Set up the stack						    */
 stack_setup:
 	mov	sp, r4
 
 	adr	r0, _start
+	cmp	r0, r6
+	beq	clear_bss		/* skip relocation */
+	mov	r1, r6			/* r1 <- scratch for copy_loop */
 	ldr	r2, _TEXT_BASE
 	ldr	r3, _bss_start_ofs
 	add	r2, r0, r3		/* r2 <- source end address	    */
-	cmp	r0, r6
-	beq	clear_bss
 
 copy_loop:
 	ldmia	r0!, {r9-r10}		/* copy from source address [r0]    */
-	stmia	r6!, {r9-r10}		/* copy to   target address [r1]    */
+	stmia	r1!, {r9-r10}		/* copy to   target address [r1]    */
 	cmp	r0, r2			/* until source end address [r2]    */
 	blo	copy_loop
 
@@ -168,7 +169,7 @@ copy_loop:
 	 * fix .rel.dyn relocations
 	 */
 	ldr	r0, _TEXT_BASE		/* r0 <- Text base */
-	sub	r9, r7, r0		/* r9 <- relocation offset */
+	sub	r9, r6, r0		/* r9 <- relocation offset */
 	ldr	r10, _dynsym_start_ofs	/* r10 <- sym table ofs */
 	add	r10, r10, r0		/* r10 <- sym table in FLASH */
 	ldr	r2, _rel_dyn_start_ofs	/* r2 <- rel dyn start ofs */
@@ -179,10 +180,10 @@ fixloop:
 	ldr	r0, [r2]		/* r0 <- location to fix up, IN FLASH! */
 	add	r0, r0, r9		/* r0 <- location to fix up in RAM */
 	ldr	r1, [r2, #4]
-	and	r8, r1, #0xff
-	cmp	r8, #23			/* relative fixup? */
+	and	r7, r1, #0xff
+	cmp	r7, #23			/* relative fixup? */
 	beq	fixrel
-	cmp	r8, #2			/* absolute fixup? */
+	cmp	r7, #2			/* absolute fixup? */
 	beq	fixabs
 	/* ignore unknown type of fixup */
 	b	fixnext
@@ -191,7 +192,7 @@ fixabs:
 	mov	r1, r1, LSR #4		/* r1 <- symbol index in .dynsym */
 	add	r1, r10, r1		/* r1 <- address of symbol in table */
 	ldr	r1, [r1, #4]		/* r1 <- symbol value */
-	add	r1, r9			/* r1 <- relocated sym addr */
+	add	r1, r1, r9		/* r1 <- relocated sym addr */
 	b	fixnext
 fixrel:
 	/* relative fix: increase location by offset */
@@ -209,7 +210,7 @@ clear_bss:
 	ldr	r0, _bss_start_ofs
 	ldr	r1, _bss_end_ofs
 	ldr	r3, _TEXT_BASE		/* Text base */
-	mov	r4, r7			/* reloc addr */
+	mov	r4, r6			/* reloc addr */
 	add	r0, r0, r4
 	add	r1, r1, r4
 	mov	r2, #0x00000000		/* clear			    */
@@ -233,7 +234,7 @@ clbss_l:str	r2, [r0]		/* clear loop...		    */
 	add	lr, lr, r9
 	/* setup parameters for board_init_r */
 	mov	r0, r5		/* gd_t */
-	mov	r1, r7		/* dest_addr */
+	mov	r1, r6		/* dest_addr */
 	/* jump to it ... */
 	mov	pc, lr
 
diff --git a/arch/arm/cpu/sa1100/start.S b/arch/arm/cpu/sa1100/start.S
index 36d44da1f8..7c2db4f146 100644
--- a/arch/arm/cpu/sa1100/start.S
+++ b/arch/arm/cpu/sa1100/start.S
@@ -131,6 +131,7 @@ reset:
 /* Set stackpointer in internal RAM to call board_init_f */
 call_board_init_f:
 	ldr	sp, =(CONFIG_SYS_INIT_SP_ADDR)
+	bic	sp, sp, #7 /* 8-byte alignment for ABI compliance */
 	ldr	r0,=0x00000000
 	bl	board_init_f
 
@@ -148,22 +149,22 @@ relocate_code:
 	mov	r4, r0	/* save addr_sp */
 	mov	r5, r1	/* save addr of gd */
 	mov	r6, r2	/* save addr of destination */
-	mov	r7, r2	/* save addr of destination */
 
 	/* Set up the stack						    */
 stack_setup:
 	mov	sp, r4
 
 	adr	r0, _start
+	cmp	r0, r6
+	beq	clear_bss		/* skip relocation */
+	mov	r1, r6			/* r1 <- scratch for copy_loop */
 	ldr	r2, _TEXT_BASE
 	ldr	r3, _bss_start_ofs
 	add	r2, r0, r3		/* r2 <- source end address	    */
-	cmp	r0, r6
-	beq	clear_bss
 
 copy_loop:
 	ldmia	r0!, {r9-r10}		/* copy from source address [r0]    */
-	stmia	r6!, {r9-r10}		/* copy to   target address [r1]    */
+	stmia	r1!, {r9-r10}		/* copy to   target address [r1]    */
 	cmp	r0, r2			/* until source end address [r2]    */
 	blo	copy_loop
 
@@ -172,7 +173,7 @@ copy_loop:
 	 * fix .rel.dyn relocations
 	 */
 	ldr	r0, _TEXT_BASE		/* r0 <- Text base */
-	sub	r9, r7, r0		/* r9 <- relocation offset */
+	sub	r9, r6, r0		/* r9 <- relocation offset */
 	ldr	r10, _dynsym_start_ofs	/* r10 <- sym table ofs */
 	add	r10, r10, r0		/* r10 <- sym table in FLASH */
 	ldr	r2, _rel_dyn_start_ofs	/* r2 <- rel dyn start ofs */
@@ -183,10 +184,10 @@ fixloop:
 	ldr	r0, [r2]		/* r0 <- location to fix up, IN FLASH! */
 	add	r0, r0, r9		/* r0 <- location to fix up in RAM */
 	ldr	r1, [r2, #4]
-	and	r8, r1, #0xff
-	cmp	r8, #23			/* relative fixup? */
+	and	r7, r1, #0xff
+	cmp	r7, #23			/* relative fixup? */
 	beq	fixrel
-	cmp	r8, #2			/* absolute fixup? */
+	cmp	r7, #2			/* absolute fixup? */
 	beq	fixabs
 	/* ignore unknown type of fixup */
 	b	fixnext
@@ -195,7 +196,7 @@ fixabs:
 	mov	r1, r1, LSR #4		/* r1 <- symbol index in .dynsym */
 	add	r1, r10, r1		/* r1 <- address of symbol in table */
 	ldr	r1, [r1, #4]		/* r1 <- symbol value */
-	add	r1, r9			/* r1 <- relocated sym addr */
+	add	r1, r1, r9		/* r1 <- relocated sym addr */
 	b	fixnext
 fixrel:
 	/* relative fix: increase location by offset */
@@ -213,7 +214,7 @@ clear_bss:
 	ldr	r0, _bss_start_ofs
 	ldr	r1, _bss_end_ofs
 	ldr	r3, _TEXT_BASE		/* Text base */
-	mov	r4, r7			/* reloc addr */
+	mov	r4, r6			/* reloc addr */
 	add	r0, r0, r4
 	add	r1, r1, r4
 	mov	r2, #0x00000000		/* clear			    */
@@ -234,7 +235,7 @@ clbss_l:str	r2, [r0]		/* clear loop...		    */
 	add	lr, lr, r9
 	/* setup parameters for board_init_r */
 	mov	r0, r5		/* gd_t */
-	mov	r1, r7		/* dest_addr */
+	mov	r1, r6		/* dest_addr */
 	/* jump to it ... */
 	mov	pc, lr
 
diff --git a/arch/arm/include/asm/arch-mx25/imx-regs.h b/arch/arm/include/asm/arch-mx25/imx-regs.h
index f5a2929e22..55ad115a76 100644
--- a/arch/arm/include/asm/arch-mx25/imx-regs.h
+++ b/arch/arm/include/asm/arch-mx25/imx-regs.h
@@ -36,6 +36,7 @@
 #ifndef __ASSEMBLY__
 #ifdef CONFIG_FEC_MXC
 extern void mx25_fec_init_pins(void);
+extern void imx_get_mac_from_fuse(unsigned char *mac);
 #endif
 
 /* Clock Control Module (CCM) registers */
@@ -129,12 +130,17 @@ struct iim_regs {
 	u32 iim_srev;
 	u32 iim_prog_p;
 	u32 res1[0x1f5];
-	u32 iim_bank_area0[0x20];
-	u32 res2[0xe0];
-	u32 iim_bank_area1[0x20];
-	u32 res3[0xe0];
-	u32 iim_bank_area2[0x20];
+	struct fuse_bank {
+		u32 fuse_regs[0x20];
+		u32 fuse_rsvd[0xe0];
+	} bank[3];
 };
+
+struct fuse_bank0_regs {
+	u32 fuse0_25[0x1a];
+	u32 mac_addr[6];
+};
+
 #endif
 
 /* AIPS 1 */
@@ -312,7 +318,4 @@ struct iim_regs {
 #define WSR_UNLOCK1		0x5555
 #define WSR_UNLOCK2		0xAAAA
 
-/* FUSE bank offsets */
-#define IIM0_MAC		0x1a
-
 #endif				/* _IMX_REGS_H */
diff --git a/arch/arm/include/asm/arch-mx27/imx-regs.h b/arch/arm/include/asm/arch-mx27/imx-regs.h
index 6ecddaae85..8f40aa728c 100644
--- a/arch/arm/include/asm/arch-mx27/imx-regs.h
+++ b/arch/arm/include/asm/arch-mx27/imx-regs.h
@@ -34,6 +34,7 @@ extern void mx27_uart_init_pins(void);
 
 #ifdef CONFIG_FEC_MXC
 extern void mx27_fec_init_pins(void);
+extern void imx_get_mac_from_fuse(unsigned char *mac);
 #endif /* CONFIG_FEC_MXC */
 
 #ifdef CONFIG_MXC_MMC
@@ -202,9 +203,19 @@ struct iim_regs {
 	u32 iim_scs1;
 	u32 iim_scs2;
 	u32 iim_scs3;
-	u32 res[0x1F0];
-	u32 iim_bank_area0[0x100];
+	u32 res[0x1f1];
+	struct fuse_bank {
+		u32 fuse_regs[0x20];
+		u32 fuse_rsvd[0xe0];
+	} bank[1];
 };
+
+struct fuse_bank0_regs {
+	u32 fuse0_3[5];
+	u32 mac_addr[6];
+	u32 fuse10_31[0x16];
+};
+
 #endif
 
 #define IMX_IO_BASE		0x10000000
@@ -512,9 +523,4 @@ struct iim_regs {
 #define IIM_ERR_SNSE	(1 << 2)
 #define IIM_ERR_PARITYE	(1 << 1)
 
-/* Definitions for i.MX27 TO2 */
-#define IIM0_MAC		5
-#define IIM0_SCC_KEY		11
-#define IIM1_SUID		1
-
 #endif				/* _IMX_REGS_H */
diff --git a/arch/arm/include/asm/arch-mx5/imx-regs.h b/arch/arm/include/asm/arch-mx5/imx-regs.h
index 0b6249a9b8..b45026d175 100644
--- a/arch/arm/include/asm/arch-mx5/imx-regs.h
+++ b/arch/arm/include/asm/arch-mx5/imx-regs.h
@@ -205,9 +205,13 @@
 #define BOARD_REV_1_0           0x0
 #define BOARD_REV_2_0           0x1
 
+#define IMX_IIM_BASE            (IIM_BASE_ADDR)
+
 #if !(defined(__KERNEL_STRICT_NAMES) || defined(__ASSEMBLY__))
 #include <asm/types.h>
 
+extern void imx_get_mac_from_fuse(unsigned char *mac);
+
 #define __REG(x)	(*((volatile u32 *)(x)))
 #define __REG16(x)	(*((volatile u16 *)(x)))
 #define __REG8(x)	(*((volatile u8 *)(x)))
@@ -275,6 +279,36 @@ struct src {
 	u32	sisr;
 	u32	simr;
 };
+
+struct iim_regs {
+	u32	stat;
+	u32	statm;
+	u32     err;
+	u32	emask;
+	u32	fctl;
+	u32	ua;
+	u32	la;
+	u32	sdat;
+	u32	prev;
+	u32	srev;
+	u32	preg_p;
+	u32	scs0;
+	u32	scs1;
+	u32	scs2;
+	u32	scs3;
+	u32	res0[0x1f1];
+	struct fuse_bank {
+		u32	fuse_regs[0x20];
+		u32	fuse_rsvd[0xe0];
+	} bank[4];
+};
+
+struct fuse_bank1_regs {
+	u32	fuse0_8[9];
+	u32	mac_addr[6];
+	u32	fuse15_31[0x11];
+};
+
 #endif /* __ASSEMBLER__*/
 
 #endif				/*  __ASM_ARCH_MXC_MX51_H__ */
diff --git a/arch/arm/include/asm/global_data.h b/arch/arm/include/asm/global_data.h
index ada3fbb641..2a84d27a4e 100644
--- a/arch/arm/include/asm/global_data.h
+++ b/arch/arm/include/asm/global_data.h
@@ -55,11 +55,14 @@ typedef	struct	global_data {
 	unsigned long	plla_rate_hz;
 	unsigned long	pllb_rate_hz;
 	unsigned long	at91_pllb_usb_init;
-	/* "static data" needed by at91's timer.c */
+#endif
+#ifdef CONFIG_ARM
+	/* "static data" needed by most of timer.c on ARM platforms */
 	unsigned long	timer_rate_hz;
 	unsigned long	tbl;
 	unsigned long	tbu;
 	unsigned long long	timer_reset_value;
+	unsigned long	lastinc;
 #endif
 	unsigned long	relocaddr;	/* Start address of U-Boot in RAM */
 	phys_size_t	ram_size;	/* RAM size */
diff --git a/arch/arm/lib/board.c b/arch/arm/lib/board.c
index 1fd5f8362f..96c0e30161 100644
--- a/arch/arm/lib/board.c
+++ b/arch/arm/lib/board.c
@@ -276,7 +276,7 @@ void board_init_f (ulong bootflag)
 	ulong addr, addr_sp;
 
 	/* Pointer is writable since we allocated a register for it */
-	gd = (gd_t *) (CONFIG_SYS_INIT_SP_ADDR);
+	gd = (gd_t *) ((CONFIG_SYS_INIT_SP_ADDR) & ~0x07);
 	/* compiler optimization barrier needed for GCC >= 3.4 */
 	__asm__ __volatile__("": : :"memory");
 
diff --git a/arch/powerpc/cpu/mpc83xx/pcie.c b/arch/powerpc/cpu/mpc83xx/pcie.c
index 1771c4823c..46a706d349 100644
--- a/arch/powerpc/cpu/mpc83xx/pcie.c
+++ b/arch/powerpc/cpu/mpc83xx/pcie.c
@@ -201,18 +201,18 @@ static void mpc83xx_pcie_init_bus(int bus, struct pci_region *reg)
 	out_le32(&out_win->tarl, 0);
 	out_le32(&out_win->tarh, 0);
 
-	for (i = 0; i < 2; i++, reg++) {
+	for (i = 0; i < 2; i++) {
 		u32 ar;
 
-		if (reg->size == 0)
+		if (reg[i].size == 0)
 			break;
 
 		out_win = &pex->bridge.pex_outbound_win[i + 1];
-		out_le32(&out_win->bar, reg->phys_start);
-		out_le32(&out_win->tarl, reg->bus_start);
+		out_le32(&out_win->bar, reg[i].phys_start);
+		out_le32(&out_win->tarl, reg[i].bus_start);
 		out_le32(&out_win->tarh, 0);
-		ar = PEX_OWAR_EN | (reg->size & PEX_OWAR_SIZE);
-		if (reg->flags & PCI_REGION_IO)
+		ar = PEX_OWAR_EN | (reg[i].size & PEX_OWAR_SIZE);
+		if (reg[i].flags & PCI_REGION_IO)
 			ar |= PEX_OWAR_TYPE_IO;
 		else
 			ar |= PEX_OWAR_TYPE_MEM;
diff --git a/arch/powerpc/cpu/mpc83xx/start.S b/arch/powerpc/cpu/mpc83xx/start.S
index 515be4c871..460ac9aabb 100644
--- a/arch/powerpc/cpu/mpc83xx/start.S
+++ b/arch/powerpc/cpu/mpc83xx/start.S
@@ -1158,6 +1158,10 @@ map_flash_by_law1:
 	bne 1b
 
 	stw r4, LBLAWAR1(r3) /* LBLAWAR1 <= 8MB Flash Size */
+	/* Wait for HW to catch up */
+	lwz r4, LBLAWAR1(r3)
+	twi 0,r4,0
+	isync
 	blr
 
 	/* Though all the LBIU Local Access Windows and LBC Banks will be
@@ -1196,5 +1200,9 @@ remap_flash_by_law0:
 	xor r4, r4, r4
 	stw r4, LBLAWBAR1(r3)
 	stw r4, LBLAWAR1(r3) /* Off LBIU LAW1 */
+	/* Wait for HW to catch up */
+	lwz r4, LBLAWAR1(r3)
+	twi 0,r4,0
+	isync
 	blr
 #endif /* CONFIG_SYS_FLASHBOOT */
diff --git a/arch/powerpc/cpu/mpc85xx/cpu_init.c b/arch/powerpc/cpu/mpc85xx/cpu_init.c
index 27236a0bad..4b8faa5daf 100644
--- a/arch/powerpc/cpu/mpc85xx/cpu_init.c
+++ b/arch/powerpc/cpu/mpc85xx/cpu_init.c
@@ -327,7 +327,7 @@ int cpu_init_r(void)
 	if (l2cache->l2ctl & MPC85xx_L2CTL_L2E) {
 		puts("already enabled");
 		l2srbar = l2cache->l2srbar0;
-#ifdef CONFIG_SYS_INIT_L2_ADDR
+#if defined(CONFIG_SYS_INIT_L2_ADDR) && defined(CONFIG_SYS_FLASH_BASE)
 		if (l2cache->l2ctl & MPC85xx_L2CTL_L2SRAM_ENTIRE
 				&& l2srbar >= CONFIG_SYS_FLASH_BASE) {
 			l2srbar = CONFIG_SYS_INIT_L2_ADDR;
diff --git a/arch/powerpc/lib/board.c b/arch/powerpc/lib/board.c
index b21c1d6ff0..9759e23a53 100644
--- a/arch/powerpc/lib/board.c
+++ b/arch/powerpc/lib/board.c
@@ -645,6 +645,17 @@ void board_init_r (gd_t *id, ulong dest_addr)
 	gd->cpu += dest_addr - CONFIG_SYS_MONITOR_BASE;
 #endif
 
+#ifdef CONFIG_SYS_EXTRA_ENV_RELOC
+	/*
+	 * Some systems need to relocate the env_addr pointer early because the
+	 * location it points to will get invalidated before env_relocate is
+	 * called.  One example is on systems that might use a L2 or L3 cache
+	 * in SRAM mode and initialize that cache from SRAM mode back to being
+	 * a cache in cpu_init_r.
+	 */
+	gd->env_addr += dest_addr - CONFIG_SYS_MONITOR_BASE;
+#endif
+
 #ifdef CONFIG_SERIAL_MULTI
 	serial_initialize();
 #endif
diff --git a/arch/sh/cpu/sh2/start.S b/arch/sh/cpu/sh2/start.S
index 77043f686a..8e0e64067b 100644
--- a/arch/sh/cpu/sh2/start.S
+++ b/arch/sh/cpu/sh2/start.S
@@ -26,7 +26,7 @@
 	.align	2
 
 	.global	_start
-_start:
+_sh_start:
 	.long 0x00000010	/* Ppower ON reset PC*/
 	.long 0x00000000
 	.long 0x00000010	/* Manual reset PC */
@@ -39,7 +39,7 @@ _init:
 	nop
 1:	sts	pr, r5
 	mov.l	._reloc_dst, r4
-	add	#(_start-1b), r5
+	add	#(_sh_start-1b), r5
 	mov.l	._reloc_dst_end, r6
 
 2:	mov.l	@r5+, r1
@@ -74,6 +74,6 @@ loop:
 ._reloc_dst_end:	.long	reloc_dst_end
 ._bss_start:		.long	bss_start
 ._bss_end:		.long	bss_end
-._gd_init:		.long	(_start - GENERATED_GBL_DATA_SIZE)
-._stack_init:	.long	(_start - GENERATED_GBL_DATA_SIZE - CONFIG_SYS_MALLOC_LEN - 16)
+._gd_init:		.long	(_sh_start - GENERATED_GBL_DATA_SIZE)
+._stack_init:	.long	(_sh_start - GENERATED_GBL_DATA_SIZE - CONFIG_SYS_MALLOC_LEN - 16)
 ._sh_generic_init:	.long	sh_generic_init
diff --git a/arch/sh/cpu/sh3/start.S b/arch/sh/cpu/sh3/start.S
index 9dd2303265..d96ca910c6 100644
--- a/arch/sh/cpu/sh3/start.S
+++ b/arch/sh/cpu/sh3/start.S
@@ -29,7 +29,7 @@
 	.align	2
 
 	.global	_start
-_start:
+_sh_start:
 	mov.l	._lowlevel_init, r0
 100:	bsrf	r0
 	nop
@@ -38,7 +38,7 @@ _start:
 	nop
 1:	sts	pr, r5
 	mov.l	._reloc_dst, r4
-	add	#(_start-1b), r5
+	add	#(_sh_start-1b), r5
 	mov.l	._reloc_dst_end, r6
 
 2:	mov.l	@r5+, r1
@@ -73,6 +73,6 @@ loop:
 ._reloc_dst_end:	.long	reloc_dst_end
 ._bss_start:		.long	bss_start
 ._bss_end:		.long	bss_end
-._gd_init:		.long	(_start - GENERATED_GBL_DATA_SIZE)
-._stack_init:	.long	(_start - GENERATED_GBL_DATA_SIZE - CONFIG_SYS_MALLOC_LEN - 16)
+._gd_init:		.long	(_sh_start - GENERATED_GBL_DATA_SIZE)
+._stack_init:	.long	(_sh_start - GENERATED_GBL_DATA_SIZE - CONFIG_SYS_MALLOC_LEN - 16)
 ._sh_generic_init:	.long	sh_generic_init
diff --git a/arch/sh/cpu/sh4/start.S b/arch/sh/cpu/sh4/start.S
index 4b5f606fff..a1d5ee42ef 100644
--- a/arch/sh/cpu/sh4/start.S
+++ b/arch/sh/cpu/sh4/start.S
@@ -1,5 +1,5 @@
 /*
- * (C) Copyright 2007
+ * (C) Copyright 2007, 2010
  * Nobuhiro Iwamatsu <iwamatsu@nigauri.org>
  *
  * This program is free software; you can redistribute it and/or
@@ -26,7 +26,7 @@
 	.align	2
 
 	.global	_start
-_start:
+_sh_start:
 	mov.l	._lowlevel_init, r0
 100:	bsrf	r0
 	nop
@@ -35,7 +35,7 @@ _start:
 	nop
 1:	sts	pr, r5
 	mov.l	._reloc_dst, r4
-	add	#(_start-1b), r5
+	add	#(_sh_start-1b), r5
 	mov.l	._reloc_dst_end, r6
 
 2:	mov.l	@r5+, r1
@@ -70,6 +70,6 @@ loop:
 ._reloc_dst_end:	.long	reloc_dst_end
 ._bss_start:		.long	bss_start
 ._bss_end:		.long	bss_end
-._gd_init:		.long	(_start - GENERATED_GBL_DATA_SIZE)
-._stack_init:		.long	(_start - GENERATED_GBL_DATA_SIZE - CONFIG_SYS_MALLOC_LEN - 16)
+._gd_init:		.long	(_sh_start - GENERATED_GBL_DATA_SIZE)
+._stack_init:		.long	(_sh_start - GENERATED_GBL_DATA_SIZE - CONFIG_SYS_MALLOC_LEN - 16)
 ._sh_generic_init:	.long	sh_generic_init