bl2-el3: Add BL2_EL3 image

This patch enables BL2 to execute at the highest exception level
without any dependancy on TF BL1. This enables platforms which already
have a non-TF Boot ROM to directly load and execute BL2 and subsequent BL
stages without need for BL1.  This is not currently possible because
BL2 executes at S-EL1 and cannot jump straight to EL3.

Change-Id: Ief1efca4598560b1b8c8e61fbe26d1f44e929d69
Signed-off-by: Roberto Vargas <roberto.vargas@arm.com>
diff --git a/Makefile b/Makefile
index 1058b39..b97c048 100644
--- a/Makefile
+++ b/Makefile
@@ -500,6 +500,7 @@
 $(eval $(call assert_boolean,USE_COHERENT_MEM))
 $(eval $(call assert_boolean,USE_TBBR_DEFS))
 $(eval $(call assert_boolean,WARMBOOT_ENABLE_DCACHE_EARLY))
+$(eval $(call assert_boolean,BL2_AT_EL3))
 
 $(eval $(call assert_numeric,ARM_ARCH_MAJOR))
 $(eval $(call assert_numeric,ARM_ARCH_MINOR))
@@ -543,6 +544,7 @@
 $(eval $(call add_define,USE_COHERENT_MEM))
 $(eval $(call add_define,USE_TBBR_DEFS))
 $(eval $(call add_define,WARMBOOT_ENABLE_DCACHE_EARLY))
+$(eval $(call add_define,BL2_AT_EL3))
 
 # Define the EL3_PAYLOAD_BASE flag only if it is provided.
 ifdef EL3_PAYLOAD_BASE
diff --git a/bl2/aarch32/bl2_el3_entrypoint.S b/bl2/aarch32/bl2_el3_entrypoint.S
new file mode 100644
index 0000000..997b069
--- /dev/null
+++ b/bl2/aarch32/bl2_el3_entrypoint.S
@@ -0,0 +1,84 @@
+/*
+ * Copyright (c) 2017, ARM Limited and Contributors. All rights reserved.
+ *
+ * SPDX-License-Identifier: BSD-3-Clause
+ */
+
+#include <arch.h>
+#include <asm_macros.S>
+#include <bl_common.h>
+#include <el3_common_macros.S>
+
+
+	.globl	bl2_entrypoint
+	.globl	bl2_run_next_image
+
+
+func bl2_entrypoint
+	/* Save arguments x0-x3 from previous Boot loader */
+	mov	r9, r0
+	mov	r10, r1
+	mov	r11, r2
+	mov	r12, r3
+
+	el3_entrypoint_common                                   \
+                _init_sctlr=1                                   \
+                _warm_boot_mailbox=!PROGRAMMABLE_RESET_ADDRESS  \
+                _secondary_cold_boot=!COLD_BOOT_SINGLE_CPU      \
+                _init_memory=1                                  \
+                _init_c_runtime=1                               \
+                _exception_vectors=bl2_vector_table
+
+	/*
+	 * Restore parameters of boot rom
+	 */
+	mov	r0, r9
+	mov	r1, r10
+	mov	r2, r11
+	mov	r3, r12
+
+	bl	bl2_el3_early_platform_setup
+	bl	bl2_el3_plat_arch_setup
+
+	/* ---------------------------------------------
+	 * Jump to main function.
+	 * ---------------------------------------------
+	 */
+	bl	bl2_main
+
+	/* ---------------------------------------------
+	 * Should never reach this point.
+	 * ---------------------------------------------
+	 */
+	no_ret	plat_panic_handler
+
+endfunc bl2_entrypoint
+
+func bl2_run_next_image
+	mov	r8,r0
+
+	/*
+	 * MMU needs to be disabled because both BL2 and BL32 execute
+	 * in PL1, and therefore share the same address space.
+	 * BL32 will initialize the address space according to its
+	 * own requirement.
+	 */
+	bl	disable_mmu_icache_secure
+	stcopr	r0, TLBIALL
+	dsb	sy
+	isb
+	mov	r0, r8
+	bl	bl2_el3_plat_prepare_exit
+
+	/*
+	 * Extract PC and SPSR based on struct `entry_point_info_t`
+	 * and load it in LR and SPSR registers respectively.
+	 */
+	ldr	lr, [r8, #ENTRY_POINT_INFO_PC_OFFSET]
+	ldr	r1, [r8, #(ENTRY_POINT_INFO_PC_OFFSET + 4)]
+	msr	spsr, r1
+
+	add	r8, r8, #ENTRY_POINT_INFO_ARGS_OFFSET
+	ldm	r8, {r0, r1, r2, r3}
+	eret
+endfunc bl2_run_next_image
diff --git a/bl2/aarch32/bl2_el3_exceptions.S b/bl2/aarch32/bl2_el3_exceptions.S
new file mode 100644
index 0000000..11ddf37
--- /dev/null
+++ b/bl2/aarch32/bl2_el3_exceptions.S
@@ -0,0 +1,21 @@
+/*
+ * Copyright (c) 2017, ARM Limited and Contributors. All rights reserved.
+ *
+ * SPDX-License-Identifier: BSD-3-Clause
+ */
+
+#include <arch.h>
+#include <asm_macros.S>
+#include <bl_common.h>
+
+	.globl	bl2_vector_table
+
+vector_base bl2_vector_table
+	b	bl2_entrypoint
+	b	report_exception	/* Undef */
+	b	report_exception	/* SVC call */
+	b	report_exception	/* Prefetch abort */
+	b	report_exception	/* Data abort */
+	b	report_exception	/* Reserved */
+	b	report_exception	/* IRQ */
+	b	report_exception	/* FIQ */
diff --git a/bl2/aarch64/bl2_el3_entrypoint.S b/bl2/aarch64/bl2_el3_entrypoint.S
new file mode 100644
index 0000000..2d3efd1
--- /dev/null
+++ b/bl2/aarch64/bl2_el3_entrypoint.S
@@ -0,0 +1,77 @@
+/*
+ * Copyright (c) 2017, ARM Limited and Contributors. All rights reserved.
+ *
+ * SPDX-License-Identifier: BSD-3-Clause
+ */
+
+#include <arch.h>
+#include <asm_macros.S>
+#include <bl_common.h>
+#include <el3_common_macros.S>
+
+	.globl	bl2_entrypoint
+	.globl	bl2_vector_table
+	.globl	bl2_el3_run_image
+	.globl	bl2_run_next_image
+
+func bl2_entrypoint
+	/* Save arguments x0-x3 from previous Boot loader */
+	mov	x20, x0
+	mov	x21, x1
+	mov	x22, x2
+	mov	x23, x3
+
+	el3_entrypoint_common                                   \
+		_init_sctlr=1                                   \
+		_warm_boot_mailbox=!PROGRAMMABLE_RESET_ADDRESS  \
+		_secondary_cold_boot=!COLD_BOOT_SINGLE_CPU      \
+		_init_memory=1                                  \
+		_init_c_runtime=1                               \
+		_exception_vectors=bl2_el3_exceptions
+
+	/*
+	 * Restore parameters of boot rom
+	 */
+	mov	x0, x20
+	mov	x1, x21
+	mov	x2, x22
+	mov	x3, x23
+
+	bl	bl2_el3_early_platform_setup
+	bl	bl2_el3_plat_arch_setup
+
+	/* ---------------------------------------------
+	 * Jump to main function.
+	 * ---------------------------------------------
+	 */
+	bl	bl2_main
+
+	/* ---------------------------------------------
+	 * Should never reach this point.
+	 * ---------------------------------------------
+	 */
+	no_ret	plat_panic_handler
+endfunc bl2_entrypoint
+
+func bl2_run_next_image
+	mov	x20,x0
+        /*
+         * MMU needs to be disabled because both BL2 and BL31 execute
+         * in EL3, and therefore share the same address space.
+         * BL31 will initialize the address space according to its
+         * own requirement.
+         */
+	bl	disable_mmu_icache_el3
+	tlbi	alle3
+	bl	bl2_el3_plat_prepare_exit
+
+	ldp	x0, x1, [x20, #ENTRY_POINT_INFO_PC_OFFSET]
+	msr	elr_el3, x0
+	msr	spsr_el3, x1
+
+	ldp	x6, x7, [x20, #(ENTRY_POINT_INFO_ARGS_OFFSET + 0x30)]
+	ldp	x4, x5, [x20, #(ENTRY_POINT_INFO_ARGS_OFFSET + 0x20)]
+	ldp	x2, x3, [x20, #(ENTRY_POINT_INFO_ARGS_OFFSET + 0x10)]
+	ldp	x0, x1, [x20, #(ENTRY_POINT_INFO_ARGS_OFFSET + 0x0)]
+	eret
+endfunc bl2_run_next_image
diff --git a/bl2/aarch64/bl2_el3_exceptions.S b/bl2/aarch64/bl2_el3_exceptions.S
new file mode 100644
index 0000000..987f6e3
--- /dev/null
+++ b/bl2/aarch64/bl2_el3_exceptions.S
@@ -0,0 +1,131 @@
+/*
+ * Copyright (c) 2017, ARM Limited and Contributors. All rights reserved.
+ *
+ * SPDX-License-Identifier: BSD-3-Clause
+ */
+
+#include <arch.h>
+#include <asm_macros.S>
+#include <bl1.h>
+#include <bl_common.h>
+#include <context.h>
+
+/* -----------------------------------------------------------------------------
+ * Very simple stackless exception handlers used by BL2.
+ * -----------------------------------------------------------------------------
+ */
+	.globl	bl2_el3_exceptions
+
+vector_base bl2_el3_exceptions
+
+	/* -----------------------------------------------------
+	 * Current EL with SP0 : 0x0 - 0x200
+	 * -----------------------------------------------------
+	 */
+vector_entry SynchronousExceptionSP0
+	mov	x0, #SYNC_EXCEPTION_SP_EL0
+	bl	plat_report_exception
+	no_ret	plat_panic_handler
+	check_vector_size SynchronousExceptionSP0
+
+vector_entry IrqSP0
+	mov	x0, #IRQ_SP_EL0
+	bl	plat_report_exception
+	no_ret	plat_panic_handler
+	check_vector_size IrqSP0
+
+vector_entry FiqSP0
+	mov	x0, #FIQ_SP_EL0
+	bl	plat_report_exception
+	no_ret	plat_panic_handler
+	check_vector_size FiqSP0
+
+vector_entry SErrorSP0
+	mov	x0, #SERROR_SP_EL0
+	bl	plat_report_exception
+	no_ret	plat_panic_handler
+	check_vector_size SErrorSP0
+
+	/* -----------------------------------------------------
+	 * Current EL with SPx: 0x200 - 0x400
+	 * -----------------------------------------------------
+	 */
+vector_entry SynchronousExceptionSPx
+	mov	x0, #SYNC_EXCEPTION_SP_ELX
+	bl	plat_report_exception
+	no_ret	plat_panic_handler
+	check_vector_size SynchronousExceptionSPx
+
+vector_entry IrqSPx
+	mov	x0, #IRQ_SP_ELX
+	bl	plat_report_exception
+	no_ret	plat_panic_handler
+	check_vector_size IrqSPx
+
+vector_entry FiqSPx
+	mov	x0, #FIQ_SP_ELX
+	bl	plat_report_exception
+	no_ret	plat_panic_handler
+	check_vector_size FiqSPx
+
+vector_entry SErrorSPx
+	mov	x0, #SERROR_SP_ELX
+	bl	plat_report_exception
+	no_ret	plat_panic_handler
+	check_vector_size SErrorSPx
+
+	/* -----------------------------------------------------
+	 * Lower EL using AArch64 : 0x400 - 0x600
+	 * -----------------------------------------------------
+	 */
+vector_entry SynchronousExceptionA64
+	mov	x0, #SYNC_EXCEPTION_AARCH64
+	bl	plat_report_exception
+	no_ret	plat_panic_handler
+	check_vector_size SynchronousExceptionA64
+
+vector_entry IrqA64
+	mov	x0, #IRQ_AARCH64
+	bl	plat_report_exception
+	no_ret	plat_panic_handler
+	check_vector_size IrqA64
+
+vector_entry FiqA64
+	mov	x0, #FIQ_AARCH64
+	bl	plat_report_exception
+	no_ret	plat_panic_handler
+	check_vector_size FiqA64
+
+vector_entry SErrorA64
+	mov	x0, #SERROR_AARCH64
+	bl	plat_report_exception
+	no_ret	plat_panic_handler
+	check_vector_size SErrorA64
+
+	/* -----------------------------------------------------
+	 * Lower EL using AArch32 : 0x600 - 0x800
+	 * -----------------------------------------------------
+	 */
+vector_entry SynchronousExceptionA32
+	mov	x0, #SYNC_EXCEPTION_AARCH32
+	bl	plat_report_exception
+	no_ret	plat_panic_handler
+	check_vector_size SynchronousExceptionA32
+
+vector_entry IrqA32
+	mov	x0, #IRQ_AARCH32
+	bl	plat_report_exception
+	no_ret	plat_panic_handler
+	check_vector_size IrqA32
+
+vector_entry FiqA32
+	mov	x0, #FIQ_AARCH32
+	bl	plat_report_exception
+	no_ret	plat_panic_handler
+	check_vector_size FiqA32
+
+vector_entry SErrorA32
+	mov	x0, #SERROR_AARCH32
+	bl	plat_report_exception
+	no_ret	plat_panic_handler
+	check_vector_size SErrorA32
diff --git a/bl2/bl2.mk b/bl2/bl2.mk
index 32e3284..f6d69eb 100644
--- a/bl2/bl2.mk
+++ b/bl2/bl2.mk
@@ -5,7 +5,6 @@
 #
 
 BL2_SOURCES		+=	bl2/bl2_main.c				\
-				bl2/${ARCH}/bl2_entrypoint.S		\
 				bl2/${ARCH}/bl2_arch_setup.c		\
 				lib/locks/exclusive/${ARCH}/spinlock.S	\
 				plat/common/${ARCH}/platform_up_stack.S
@@ -20,4 +19,15 @@
 BL2_SOURCES		+=	bl2/bl2_image_load.c
 endif
 
+ifeq (${BL2_AT_EL3},0)
+BL2_SOURCES		+=	bl2/${ARCH}/bl2_entrypoint.S
 BL2_LINKERFILE		:=	bl2/bl2.ld.S
+
+else
+BL2_SOURCES		+=	bl2/${ARCH}/bl2_el3_entrypoint.S	\
+				bl2/${ARCH}/bl2_el3_exceptions.S	\
+				plat/common/plat_bl2_el3_common.c	\
+				lib/cpus/${ARCH}/cpu_helpers.S		\
+				lib/cpus/errata_report.c
+BL2_LINKERFILE		:=	bl2/bl2_el3.ld.S
+endif
diff --git a/bl2/bl2_el3.ld.S b/bl2/bl2_el3.ld.S
new file mode 100644
index 0000000..7ec4646
--- /dev/null
+++ b/bl2/bl2_el3.ld.S
@@ -0,0 +1,175 @@
+/*
+ * Copyright (c) 2017, ARM Limited and Contributors. All rights reserved.
+ *
+ * SPDX-License-Identifier: BSD-3-Clause
+ */
+
+#include <platform_def.h>
+#include <xlat_tables_defs.h>
+
+OUTPUT_FORMAT(PLATFORM_LINKER_FORMAT)
+OUTPUT_ARCH(PLATFORM_LINKER_ARCH)
+ENTRY(bl2_entrypoint)
+
+MEMORY {
+    RAM (rwx): ORIGIN = BL2_BASE, LENGTH = BL2_LIMIT - BL2_BASE
+}
+
+
+SECTIONS
+{
+    . = BL2_BASE;
+    ASSERT(. == ALIGN(PAGE_SIZE),
+           "BL2_BASE address is not aligned on a page boundary.")
+
+#if SEPARATE_CODE_AND_RODATA
+    .text . : {
+        __TEXT_START__ = .;
+        *bl2_el3_entrypoint.o(.text*)
+        *(.text*)
+        *(.vectors)
+        . = NEXT(PAGE_SIZE);
+        __TEXT_END__ = .;
+     } >RAM
+
+    .rodata . : {
+        __RODATA_START__ = .;
+        *(.rodata*)
+
+        /* Ensure 8-byte alignment for descriptors and ensure inclusion */
+        . = ALIGN(8);
+        __PARSER_LIB_DESCS_START__ = .;
+        KEEP(*(.img_parser_lib_descs))
+        __PARSER_LIB_DESCS_END__ = .;
+
+        /*
+         * Ensure 8-byte alignment for cpu_ops so that its fields are also
+         * aligned. Also ensure cpu_ops inclusion.
+         */
+        . = ALIGN(8);
+        __CPU_OPS_START__ = .;
+        KEEP(*(cpu_ops))
+        __CPU_OPS_END__ = .;
+
+        . = NEXT(PAGE_SIZE);
+        __RODATA_END__ = .;
+    } >RAM
+#else
+    ro . : {
+        __RO_START__ = .;
+        *bl2_el3_entrypoint.o(.text*)
+        *(.text*)
+        *(.rodata*)
+
+        /*
+         * Ensure 8-byte alignment for cpu_ops so that its fields are also
+         * aligned. Also ensure cpu_ops inclusion.
+         */
+        . = ALIGN(8);
+        __CPU_OPS_START__ = .;
+        KEEP(*(cpu_ops))
+        __CPU_OPS_END__ = .;
+
+        /* Ensure 8-byte alignment for descriptors and ensure inclusion */
+        . = ALIGN(8);
+        __PARSER_LIB_DESCS_START__ = .;
+        KEEP(*(.img_parser_lib_descs))
+        __PARSER_LIB_DESCS_END__ = .;
+
+        *(.vectors)
+        __RO_END_UNALIGNED__ = .;
+        /*
+         * Memory page(s) mapped to this section will be marked as
+         * read-only, executable.  No RW data from the next section must
+         * creep in.  Ensure the rest of the current memory page is unused.
+         */
+        . = NEXT(PAGE_SIZE);
+
+        __RO_END__ = .;
+    } >RAM
+#endif
+
+    ASSERT(__CPU_OPS_END__ > __CPU_OPS_START__,
+          "cpu_ops not defined for this platform.")
+
+    /*
+     * Define a linker symbol to mark start of the RW memory area for this
+     * image.
+     */
+    __RW_START__ = . ;
+
+    /*
+     * .data must be placed at a lower address than the stacks if the stack
+     * protector is enabled. Alternatively, the .data.stack_protector_canary
+     * section can be placed independently of the main .data section.
+     */
+    .data . : {
+        __DATA_START__ = .;
+        *(.data*)
+        __DATA_END__ = .;
+    } >RAM
+
+    stacks (NOLOAD) : {
+        __STACKS_START__ = .;
+        *(tzfw_normal_stacks)
+        __STACKS_END__ = .;
+    } >RAM
+
+    /*
+     * The .bss section gets initialised to 0 at runtime.
+     * Its base address should be 16-byte aligned for better performance of the
+     * zero-initialization code.
+     */
+    .bss : ALIGN(16) {
+        __BSS_START__ = .;
+        *(SORT_BY_ALIGNMENT(.bss*))
+        *(COMMON)
+        __BSS_END__ = .;
+    } >RAM
+
+    /*
+     * The xlat_table section is for full, aligned page tables (4K).
+     * Removing them from .bss avoids forcing 4K alignment on
+     * the .bss section and eliminates the unnecessary zero init
+     */
+    xlat_table (NOLOAD) : {
+        *(xlat_table)
+    } >RAM
+
+#if USE_COHERENT_MEM
+    /*
+     * The base address of the coherent memory section must be page-aligned (4K)
+     * to guarantee that the coherent data are stored on their own pages and
+     * are not mixed with normal data.  This is required to set up the correct
+     * memory attributes for the coherent data page tables.
+     */
+    coherent_ram (NOLOAD) : ALIGN(PAGE_SIZE) {
+        __COHERENT_RAM_START__ = .;
+        *(tzfw_coherent_mem)
+        __COHERENT_RAM_END_UNALIGNED__ = .;
+        /*
+         * Memory page(s) mapped to this section will be marked
+         * as device memory.  No other unexpected data must creep in.
+         * Ensure the rest of the current memory page is unused.
+         */
+        . = NEXT(PAGE_SIZE);
+        __COHERENT_RAM_END__ = .;
+    } >RAM
+#endif
+
+    /*
+     * Define a linker symbol to mark end of the RW memory area for this
+     * image.
+     */
+    __RW_END__ = .;
+    __BL2_END__ = .;
+
+    __BSS_SIZE__ = SIZEOF(.bss);
+
+#if USE_COHERENT_MEM
+    __COHERENT_RAM_UNALIGNED_SIZE__ =
+        __COHERENT_RAM_END_UNALIGNED__ - __COHERENT_RAM_START__;
+#endif
+
+    ASSERT(. <= BL2_LIMIT, "BL2 image has exceeded its limit.")
+}
diff --git a/bl2/bl2_main.c b/bl2/bl2_main.c
index 018deb3..c85db2d 100644
--- a/bl2/bl2_main.c
+++ b/bl2/bl2_main.c
@@ -13,6 +13,11 @@
 #include <platform.h>
 #include "bl2_private.h"
 
+#ifdef AARCH32
+#define NEXT_IMAGE	"BL32"
+#else
+#define NEXT_IMAGE	"BL31"
+#endif
 
 /*******************************************************************************
  * The only thing to do in BL2 is to load further images and pass control to
@@ -49,6 +54,8 @@
 	disable_mmu_icache_secure();
 #endif /* AARCH32 */
 
+
+#if !BL2_AT_EL3
 	console_flush();
 
 	/*
@@ -57,4 +64,11 @@
 	 * be passed to next BL image as an argument.
 	 */
 	smc(BL1_SMC_RUN_IMAGE, (unsigned long)next_bl_ep_info, 0, 0, 0, 0, 0, 0);
+#else
+	NOTICE("BL2: Booting " NEXT_IMAGE "\n");
+	print_entry_point_info(next_bl_ep_info);
+	console_flush();
+
+	bl2_run_next_image(next_bl_ep_info);
+#endif
 }
diff --git a/bl2/bl2_private.h b/bl2/bl2_private.h
index 83b8047..ea2f33a 100644
--- a/bl2/bl2_private.h
+++ b/bl2/bl2_private.h
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2013-2016, ARM Limited and Contributors. All rights reserved.
+ * Copyright (c) 2013-2017, ARM Limited and Contributors. All rights reserved.
  *
  * SPDX-License-Identifier: BSD-3-Clause
  */
@@ -17,5 +17,6 @@
  *****************************************/
 void bl2_arch_setup(void);
 struct entry_point_info *bl2_load_images(void);
+void bl2_run_next_image(const entry_point_info_t *bl_ep_info);
 
 #endif /* __BL2_PRIVATE_H__ */
diff --git a/include/common/aarch32/el3_common_macros.S b/include/common/aarch32/el3_common_macros.S
index 59e99f8..d654b65 100644
--- a/include/common/aarch32/el3_common_macros.S
+++ b/include/common/aarch32/el3_common_macros.S
@@ -260,9 +260,9 @@
 	 * ---------------------------------------------------------------------
 	 */
 	.if \_init_c_runtime
-#ifdef IMAGE_BL32
+#if defined(IMAGE_BL32) || (defined(IMAGE_BL2) && BL2_AT_EL3)
 		/* -----------------------------------------------------------------
-		 * Invalidate the RW memory used by the BL32 (SP_MIN) image. This
+		 * Invalidate the RW memory used by the image. This
 		 * includes the data and NOBITS sections. This is done to
 		 * safeguard against possible corruption of this memory by
 		 * dirty cache lines in a system cache as a result of use by
@@ -273,7 +273,7 @@
 		ldr	r1, =__RW_END__
 		sub	r1, r1, r0
 		bl	inv_dcache_range
-#endif /* IMAGE_BL32 */
+#endif
 
 		ldr	r0, =__BSS_START__
 		ldr	r1, =__BSS_SIZE__
diff --git a/include/common/aarch64/el3_common_macros.S b/include/common/aarch64/el3_common_macros.S
index 63a0fa7..ebaf7aa 100644
--- a/include/common/aarch64/el3_common_macros.S
+++ b/include/common/aarch64/el3_common_macros.S
@@ -269,7 +269,7 @@
 	 * ---------------------------------------------------------------------
 	 */
 	.if \_init_c_runtime
-#ifdef IMAGE_BL31
+#if defined(IMAGE_BL31) || (defined(IMAGE_BL2) && BL2_AT_EL3)
 		/* -------------------------------------------------------------
 		 * Invalidate the RW memory used by the BL31 image. This
 		 * includes the data and NOBITS sections. This is done to
@@ -282,7 +282,7 @@
 		adr	x1, __RW_END__
 		sub	x1, x1, x0
 		bl	inv_dcache_range
-#endif /* IMAGE_BL31 */
+#endif
 
 		ldr	x0, =__BSS_START__
 		ldr	x1, =__BSS_SIZE__
diff --git a/include/lib/cpus/aarch32/cpu_macros.S b/include/lib/cpus/aarch32/cpu_macros.S
index e2e4316..0f3a572 100644
--- a/include/lib/cpus/aarch32/cpu_macros.S
+++ b/include/lib/cpus/aarch32/cpu_macros.S
@@ -9,6 +9,10 @@
 #include <arch.h>
 #include <errata_report.h>
 
+#if defined(IMAGE_BL1) || defined(IMAGE_BL32)  || (defined(IMAGE_BL2) && BL2_AT_EL3)
+#define IMAGE_AT_EL3
+#endif
+
 #define CPU_IMPL_PN_MASK	(MIDR_IMPL_MASK << MIDR_IMPL_SHIFT) | \
 				(MIDR_PN_MASK << MIDR_PN_SHIFT)
 
@@ -38,7 +42,7 @@
 CPU_MIDR: /* cpu_ops midr */
 	.space  4
 /* Reset fn is needed during reset */
-#if defined(IMAGE_BL1) || defined(IMAGE_BL32)
+#if defined(IMAGE_AT_EL3)
 CPU_RESET_FUNC: /* cpu_ops reset_func */
 	.space  4
 #endif
@@ -54,7 +58,7 @@
 #if REPORT_ERRATA
 CPU_ERRATA_FUNC: /* CPU errata status printing function */
 	.space  4
-#ifdef IMAGE_BL32
+#if defined(IMAGE_BL32)
 CPU_ERRATA_LOCK:
 	.space	4
 CPU_ERRATA_PRINTED:
@@ -120,7 +124,7 @@
 	.align 2
 	.type cpu_ops_\_name, %object
 	.word \_midr
-#if defined(IMAGE_BL1) || defined(IMAGE_BL32)
+#if defined(IMAGE_AT_EL3)
 	.word \_resetfunc
 #endif
 #ifdef IMAGE_BL32
diff --git a/include/lib/cpus/aarch64/cpu_macros.S b/include/lib/cpus/aarch64/cpu_macros.S
index a8c23e5..ccf5306 100644
--- a/include/lib/cpus/aarch64/cpu_macros.S
+++ b/include/lib/cpus/aarch64/cpu_macros.S
@@ -21,6 +21,10 @@
 /* Word size for 64-bit CPUs */
 #define CPU_WORD_SIZE			8
 
+#if defined(IMAGE_BL1) || defined(IMAGE_BL31) ||(defined(IMAGE_BL2) && BL2_AT_EL3)
+#define IMAGE_AT_EL3
+#endif
+
 /*
  * Whether errata status needs reporting. Errata status is printed in debug
  * builds for both BL1 and BL31 images.
@@ -38,7 +42,7 @@
 CPU_MIDR: /* cpu_ops midr */
 	.space  8
 /* Reset fn is needed in BL at reset vector */
-#if defined(IMAGE_BL1) || defined(IMAGE_BL31)
+#if defined(IMAGE_AT_EL3)
 CPU_RESET_FUNC: /* cpu_ops reset_func */
 	.space  8
 #endif
@@ -54,7 +58,7 @@
 #if REPORT_ERRATA
 CPU_ERRATA_FUNC:
 	.space	8
-#ifdef IMAGE_BL31
+#if defined(IMAGE_BL31)
 CPU_ERRATA_LOCK:
 	.space	8
 CPU_ERRATA_PRINTED:
@@ -124,7 +128,7 @@
 	.align 3
 	.type cpu_ops_\_name, %object
 	.quad \_midr
-#if defined(IMAGE_BL1) || defined(IMAGE_BL31)
+#if defined(IMAGE_AT_EL3)
 	.quad \_resetfunc
 #endif
 #ifdef IMAGE_BL31
diff --git a/include/lib/xlat_tables/xlat_tables_v2_helpers.h b/include/lib/xlat_tables/xlat_tables_v2_helpers.h
index 1be99b7..de1c2d4 100644
--- a/include/lib/xlat_tables/xlat_tables_v2_helpers.h
+++ b/include/lib/xlat_tables/xlat_tables_v2_helpers.h
@@ -168,7 +168,7 @@
  * This IMAGE_EL macro must not to be used outside the library, and it is only
  * used in AArch64.
  */
-#if defined(IMAGE_BL1) || defined(IMAGE_BL31)
+#if defined(IMAGE_BL1) || defined(IMAGE_BL31) || (defined(IMAGE_BL2) && BL2_AT_EL3)
 # define IMAGE_EL	3
 # define IMAGE_XLAT_DEFAULT_REGIME EL3_REGIME
 #else
diff --git a/include/plat/common/platform.h b/include/plat/common/platform.h
index f11bee9..0960105 100644
--- a/include/plat/common/platform.h
+++ b/include/plat/common/platform.h
@@ -235,6 +235,21 @@
  * Optional BL2 functions (may be overridden)
  ******************************************************************************/
 
+
+/*******************************************************************************
+ * Mandatory BL2 at EL3 functions: Must be implemented if BL2_AT_EL3 image is
+ * supported
+ ******************************************************************************/
+void bl2_el3_early_platform_setup(u_register_t arg0, u_register_t arg1,
+				  u_register_t arg2, u_register_t arg3);
+void bl2_el3_plat_arch_setup(void);
+
+
+/*******************************************************************************
+ * Optional BL2 at EL3 functions (may be overridden)
+ ******************************************************************************/
+void bl2_el3_plat_prepare_exit(void);
+
 /*******************************************************************************
  * Mandatory BL2U functions.
  ******************************************************************************/
diff --git a/lib/cpus/aarch32/cpu_helpers.S b/lib/cpus/aarch32/cpu_helpers.S
index bfdc1e4..72e42c6 100644
--- a/lib/cpus/aarch32/cpu_helpers.S
+++ b/lib/cpus/aarch32/cpu_helpers.S
@@ -10,7 +10,7 @@
 #include <cpu_data.h>
 #include <cpu_macros.S>
 
-#if defined(IMAGE_BL1) || defined(IMAGE_BL32)
+#if defined(IMAGE_BL1) || defined(IMAGE_BL32) || (defined(IMAGE_BL2) && BL2_AT_EL3)
 	/*
 	 * The reset handler common to all platforms.  After a matching
 	 * cpu_ops structure entry is found, the correponding reset_handler
@@ -42,7 +42,7 @@
 	bx	lr
 endfunc reset_handler
 
-#endif /* IMAGE_BL1 || IMAGE_BL32 */
+#endif
 
 #ifdef IMAGE_BL32 /* The power down core and cluster is needed only in  BL32 */
 	/*
diff --git a/lib/cpus/aarch64/cpu_helpers.S b/lib/cpus/aarch64/cpu_helpers.S
index 2384553..ae1c3c2 100644
--- a/lib/cpus/aarch64/cpu_helpers.S
+++ b/lib/cpus/aarch64/cpu_helpers.S
@@ -7,7 +7,7 @@
 #include <arch.h>
 #include <asm_macros.S>
 #include <assert_macros.S>
-#ifdef IMAGE_BL31
+#if defined(IMAGE_BL31) || (defined(IMAGE_BL2) && BL2_AT_EL3)
 #include <cpu_data.h>
 #endif
 #include <cpu_macros.S>
@@ -15,7 +15,7 @@
 #include <errata_report.h>
 
  /* Reset fn is needed in BL at reset vector */
-#if defined(IMAGE_BL1) || defined(IMAGE_BL31)
+#if defined(IMAGE_BL1) || defined(IMAGE_BL31) || (defined(IMAGE_BL2) && BL2_AT_EL3)
 	/*
 	 * The reset handler common to all platforms.  After a matching
 	 * cpu_ops structure entry is found, the correponding reset_handler
@@ -47,7 +47,7 @@
 	ret
 endfunc reset_handler
 
-#endif /* IMAGE_BL1 || IMAGE_BL31 */
+#endif
 
 #ifdef IMAGE_BL31 /* The power down core and cluster is needed only in  BL31 */
 	/*
diff --git a/lib/cpus/errata_report.c b/lib/cpus/errata_report.c
index 8d9f704..182679d 100644
--- a/lib/cpus/errata_report.c
+++ b/lib/cpus/errata_report.c
@@ -20,6 +20,8 @@
 # define BL_STRING	"BL31"
 #elif defined(AARCH32) && defined(IMAGE_BL32)
 # define BL_STRING	"BL32"
+#elif defined(IMAGE_BL2) && BL2_AT_EL3
+# define BL_STRING "BL2"
 #else
 # error This image should not be printing errata status
 #endif
diff --git a/make_helpers/defaults.mk b/make_helpers/defaults.mk
index fa0d17d..643890f 100644
--- a/make_helpers/defaults.mk
+++ b/make_helpers/defaults.mk
@@ -27,6 +27,9 @@
 # Base commit to perform code check on
 BASE_COMMIT			:= origin/master
 
+# Execute BL2 at EL3
+BL2_AT_EL3			:= 0
+
 # By default, consider that the platform may release several CPUs out of reset.
 # The platform Makefile is free to override this value.
 COLD_BOOT_SINGLE_CPU		:= 0
diff --git a/plat/common/plat_bl2_el3_common.c b/plat/common/plat_bl2_el3_common.c
new file mode 100644
index 0000000..358a02d
--- /dev/null
+++ b/plat/common/plat_bl2_el3_common.c
@@ -0,0 +1,24 @@
+/*
+ * Copyright (c) 2017, ARM Limited and Contributors. All rights reserved.
+ *
+ * SPDX-License-Identifier: BSD-3-Clause
+ */
+
+#include <arch_helpers.h>
+#include <assert.h>
+#include <bl_common.h>
+#include <debug.h>
+#include <errno.h>
+#include <platform_def.h>
+
+/*
+ * The following platform functions are weakly defined. They
+ * are default implementations that allow BL2 to compile in
+ * absence of real definitions. The Platforms may override
+ * with more complex definitions.
+ */
+#pragma weak bl2_el3_plat_prepare_exit
+
+void bl2_el3_plat_prepare_exit(void)
+{
+}