ARM: Add relocatable binary support

For making the same binary executable on different SoCs which have
different DRAM addresses we have to be independent of the compile
time link address.

This patch adds relocatable binary support for the ARM architecture.
With this two new functions are available. relocate_to_current_adr
will fixup the binary to continue executing from the current position.
relocate_to_adr will copy the binary to a given address, fixup the
binary and continue executing from there.

For the PBL and the real image relocatable support can be enabled
independently. This is done to (hopefully) better cope with setups
where the PBL runs from SRAM or ROM and the real binary does not.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

5 files changed, 147 insertions, 4 deletions

diff --git a/arch/arm/cpu/Makefile b/arch/arm/cpu/Makefile
index 44410eec95..5935e1c5e4 100644
--- a/arch/arm/cpu/Makefile
+++ b/arch/arm/cpu/Makefile
@@ -21,3 +21,6 @@ pbl-$(CONFIG_CPU_32v7) += cache-armv7.o
 obj-$(CONFIG_CACHE_L2X0) += cache-l2x0.o
 
 pbl-y += start-pbl.o setupc.o
+
+obj-y += common.o
+pbl-y += common.o
diff --git a/arch/arm/cpu/common.c b/arch/arm/cpu/common.c
new file mode 100644
index 0000000000..40f784f02a
--- /dev/null
+++ b/arch/arm/cpu/common.c
@@ -0,0 +1,66 @@
+/*
+ * Copyright (c) 2010 Sascha Hauer <s.hauer@pengutronix.de>, Pengutronix
+ *
+ * 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 version 2
+ * as published by the Free Software Foundation.
+ *
+ * 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.
+ *
+ */
+
+#include <common.h>
+#include <init.h>
+#include <sizes.h>
+#include <asm/barebox-arm.h>
+#include <asm/barebox-arm-head.h>
+#include <asm-generic/memory_layout.h>
+#include <asm/sections.h>
+#include <asm/pgtable.h>
+#include <asm/cache.h>
+
+/*
+ * relocate binary to the currently running address
+ */
+void relocate_to_current_adr(void)
+{
+	uint32_t offset;
+	uint32_t *dstart, *dend, *dynsym, *dynend;
+
+	/* Get offset between linked address and runtime address */
+	offset = get_runtime_offset();
+
+	dstart = (void *)(ld_var(__rel_dyn_start) - offset);
+	dend = (void *)(ld_var(__rel_dyn_end) - offset);
+
+	dynsym = (void *)(ld_var(__dynsym_start) - offset);
+	dynend = (void *)(ld_var(__dynsym_end) - offset);
+
+	while (dstart < dend) {
+		uint32_t *fixup = (uint32_t *)(*dstart - offset);
+		uint32_t type = *(dstart + 1);
+
+		if ((type & 0xff) == 0x17) {
+			*fixup = *fixup - offset;
+		} else {
+			int index = type >> 8;
+			uint32_t r = dynsym[index * 4 + 1];
+
+			*fixup = *fixup + r - offset;
+		}
+
+		*dstart -= offset;
+		dstart += 2;
+	}
+
+	memset(dynsym, 0, (unsigned long)dynend - (unsigned long)dynsym);
+
+	arm_early_mmu_cache_flush();
+	flush_icache();
+}
diff --git a/arch/arm/cpu/setupc.S b/arch/arm/cpu/setupc.S
index d0de87dd9b..7fd5d012f0 100644
--- a/arch/arm/cpu/setupc.S
+++ b/arch/arm/cpu/setupc.S
@@ -1,4 +1,5 @@
 #include <linux/linkage.h>
+#include <asm/sections.h>
 
 .section .text.setupc
 
@@ -35,3 +36,61 @@ ENTRY(setup_c)
 	pop	{r4, r5}
 	mov	pc, lr
 ENDPROC(setup_c)
+
+#ifdef CONFIG_RELOCATABLE
+/*
+ * void relocate_to_adr(unsigned long targetadr)
+ *
+ * Copy binary to targetadr, relocate code, clear bss and continue
+ * executing at new address.
+ */
+.section .text.relocate_to_adr
+ENTRY(relocate_to_adr)
+					/* r0: target address */
+	push	{r3, r4, r5, r6, r7, r8}
+	mov	r7, lr
+
+	mov	r6, r0
+
+	bl	get_runtime_offset
+
+	mov	r5, r0
+
+	ld_var	_text, r0, r4
+	mov	r8, r0
+
+	sub	r1, r0, r5		/* r1: from address */
+
+	cmp	r1, r6			/* already at correct address? */
+	beq	1f			/* yes, skip copy to new address */
+
+	ld_var	__bss_start, r2, r4
+
+	sub	r2, r2, r0		/* r2: size */
+	mov	r0, r6			/* r0: target */
+
+	add	r7, r7, r0		/* adjust return address */
+	sub	r7, r7, r1		/* lr += offset */
+
+	bl	memcpy			/* copy binary */
+
+#ifdef CONFIG_MMU
+	bl	arm_early_mmu_cache_flush
+#endif
+	mov	r0,#0
+	mcr	p15, 0, r0, c7, c5, 0	/* flush icache */
+
+	ldr	r0,=1f
+	sub	r0, r0, r8
+	add	r0, r0, r6
+	mov	pc, r0			/* jump to relocated address */
+1:
+	bl	relocate_to_current_adr	/* relocate binary */
+
+	mov	lr, r7
+
+	pop	{r3, r4, r5, r6, r7, r8}
+	mov	pc, lr
+
+ENDPROC(relocate_to_adr)
+#endif
diff --git a/arch/arm/cpu/start-pbl.c b/arch/arm/cpu/start-pbl.c
index 91bc8fe9ff..6f03c4a7df 100644
--- a/arch/arm/cpu/start-pbl.c
+++ b/arch/arm/cpu/start-pbl.c
@@ -55,9 +55,13 @@ static noinline __noreturn void __barebox_arm_entry(uint32_t membase,
 	uint32_t pg_start, pg_end, pg_len;
 	void __noreturn (*barebox)(uint32_t, uint32_t, uint32_t);
 	uint32_t endmem = membase + memsize;
+	unsigned long barebox_base;
 
 	endmem -= STACK_SIZE; /* stack */
 
+	if (IS_ENABLED(CONFIG_PBL_RELOCATABLE))
+		relocate_to_current_adr();
+
 	/* Get offset between linked address and runtime address */
 	offset = get_runtime_offset();
 
@@ -65,8 +69,13 @@ static noinline __noreturn void __barebox_arm_entry(uint32_t membase,
 	pg_end = (uint32_t)&input_data_end - offset;
 	pg_len = pg_end - pg_start;
 
+	if (IS_ENABLED(CONFIG_RELOCATABLE))
+		barebox_base = arm_barebox_image_place(membase + memsize);
+	else
+		barebox_base = TEXT_BASE;
+
 	if (offset && (IS_ENABLED(CONFIG_PBL_FORCE_PIGGYDATA_COPY) ||
-				region_overlap(pg_start, pg_len, TEXT_BASE, pg_len * 4))) {
+				region_overlap(pg_start, pg_len, barebox_base, pg_len * 4))) {
 		/*
 		 * copy piggydata binary to its link address
 		 */
@@ -86,14 +95,15 @@ static noinline __noreturn void __barebox_arm_entry(uint32_t membase,
 	free_mem_ptr = endmem;
 	free_mem_end_ptr = free_mem_ptr + SZ_128K;
 
-	pbl_barebox_uncompress((void*)TEXT_BASE, (void *)pg_start, pg_len);
+	pbl_barebox_uncompress((void*)barebox_base, (void *)pg_start, pg_len);
 
+	arm_early_mmu_cache_flush();
 	flush_icache();
 
 	if (IS_ENABLED(CONFIG_THUMB2_BAREBOX))
-		barebox = (void *)(TEXT_BASE + 1);
+		barebox = (void *)(barebox_base + 1);
 	else
-		barebox = (void *)TEXT_BASE;
+		barebox = (void *)barebox_base;
 
 	barebox(membase, memsize, boarddata);
 }
diff --git a/arch/arm/cpu/start.c b/arch/arm/cpu/start.c
index cd34d9c66d..7c2bcd0631 100644
--- a/arch/arm/cpu/start.c
+++ b/arch/arm/cpu/start.c
@@ -37,6 +37,11 @@ static noinline __noreturn void __start(uint32_t membase, uint32_t memsize,
 	unsigned long endmem = membase + memsize;
 	unsigned long malloc_start, malloc_end;
 
+	if (IS_ENABLED(CONFIG_RELOCATABLE)) {
+		unsigned long barebox_base = arm_barebox_image_place(endmem);
+		relocate_to_adr(barebox_base);
+	}
+
 	setup_c();
 
 	arm_stack_top = endmem;