x86: Simplify Flash-to-RAM code execution transition

Move the relocation offset calculation out of assembler and into C. This
also paves the way for the upcoming init sequence simplification by adding
the board_init_f_r flash to RAM transitional function

--
Changes for v2:
 - Added commit message
 - Minor adjustment to new stack address comment

arch/x86/cpu/start.S

@@ -96,32 +96,22 @@ car_init_ret:
 	movw	$0x85, %ax
 	jmp	die
 
-.globl relocate_code
-.type relocate_code, @function
-relocate_code:
+.globl board_init_f_r_trampoline
+.type board_init_f_r_trampoline, @function
+board_init_f_r_trampoline:
 	/*
 	 * SDRAM has been initialised, U-Boot code has been copied into
 	 * RAM, BSS has been cleared and relocation adjustments have been
 	 * made. It is now time to jump into the in-RAM copy of U-Boot
 	 *
-	 * %eax = Address of top of stack
-	 * %edx = Address of Global Data
-	 * %ecx = Base address of in-RAM copy of U-Boot
+	 * %eax = Address of top of new stack
 	 */
 
 	/* Setup stack in RAM */
 	movl	%eax, %esp
 
-	/* Setup call address of in-RAM copy of board_init_r() */
-	movl	$board_init_r, %ebp
-	addl	(GENERATED_GD_RELOC_OFF)(%edx), %ebp
-
-	/* Setup parameters to board_init_r() */
-	movl	%edx, %eax
-	movl	%ecx, %edx
-
-	/* Jump to in-RAM copy of board_init_r() */
-	call	*%ebp
+	/* Re-enter U-Boot by calling board_init_f_r */
+	call	board_init_f_r
 
 die:
 	hlt

arch/x86/include/asm/u-boot-x86.h

@@ -61,5 +61,7 @@ u32 isa_map_rom(u32 bus_addr, int size);
 int video_bios_init(void);
 int video_init(void);
 
+void	board_init_f_r_trampoline(ulong) __attribute__ ((noreturn));
+void	board_init_f_r(void) __attribute__ ((noreturn));
 
 #endif	/* _U_BOOT_I386_H_ */

arch/x86/lib/board.c

@@ -253,10 +253,29 @@ void board_init_f(ulong boot_flags)
 
 	gd->flags |= GD_FLG_RELOC;
 
-	/* Enter the relocated U-Boot! */
-	relocate_code(gd->start_addr_sp, gd, gd->relocaddr);
+	/*
+	 * SDRAM is now initialised, U-Boot has been copied into SDRAM,
+	 * the BSS has been cleared etc. The final stack can now be setup
+	 * in SDRAM. Code execution will continue (momentarily) in Flash,
+	 * but with the stack in SDRAM and Global Data in temporary memory
+	 * (CPU cache)
+	 */
+	board_init_f_r_trampoline(gd->start_addr_sp);
+
+	/* NOTREACHED - board_init_f_r_trampoline() does not return */
+	while (1)
+		;
+}
+
+void board_init_f_r(void)
+{
+	/*
+	 * Transfer execution from Flash to RAM by calculating the address
+	 * of the in-RAM copy of board_init_r() and calling it
+	 */
+	(board_init_r + gd->reloc_off)(gd, gd->relocaddr);
 
-	/* NOTREACHED - relocate_code() does not return */
+	/* NOTREACHED - board_init_r() does not return */
 	while (1)
 		;
 }