x86: Put global data on the stack

Putting global data on the stack simplifies the init process (and makes it
slightly quicker). During the 'flash' stage of the init sequence, global
data is in the CAR stack. After SDRAM is initialised, global data is copied
from CAR to the SDRAM stack

Signed-off-by: Graeme Russ <graeme.russ@gmail.com>
Signed-off-by: Simon Glass <sjg@chromium.org>

7 files changed, 63 insertions, 52 deletions

diff --git a/arch/x86/cpu/cpu.c b/arch/x86/cpu/cpu.c
index 67de6bc..9c2db9f 100644
--- a/arch/x86/cpu/cpu.c
+++ b/arch/x86/cpu/cpu.c
@@ -90,12 +90,6 @@ static void load_gdt(const u64 *boot_gdt, u16 num_entries)
 	asm volatile("lgdtl %0\n" : : "m" (gdt));
 }
 
-void init_gd(gd_t *id, u64 *gdt_addr)
-{
-	id->gd_addr = id;
-	setup_gdt(id, gdt_addr);
-}
-
 void setup_gdt(gd_t *id, u64 *gdt_addr)
 {
 	/* CS: code, read/execute, 4 GB, base 0 */
diff --git a/arch/x86/cpu/start.S b/arch/x86/cpu/start.S
index ee0dabe..ec12e80 100644
--- a/arch/x86/cpu/start.S
+++ b/arch/x86/cpu/start.S
@@ -83,13 +83,33 @@ car_init_ret:
 	 * or fully initialised SDRAM - we really don't care which)
 	 * starting at CONFIG_SYS_CAR_ADDR to be used as a temporary stack
 	 */
-	movl	$CONFIG_SYS_INIT_SP_ADDR, %esp
 
-	/* Initialise the Global Data Pointer */
-	movl	$CONFIG_SYS_INIT_GD_ADDR, %eax
-	movl	%eax, %edx
-	addl	$GENERATED_GBL_DATA_SIZE, %edx
-	call	init_gd;
+	/* Stack grows down from top of CAR */
+	movl	$(CONFIG_SYS_CAR_ADDR + CONFIG_SYS_CAR_SIZE), %esp
+
+	/* Reserve space on stack for global data */
+	subl	$GENERATED_GBL_DATA_SIZE, %esp
+
+	/* Align global data to 16-byte boundary */
+	andl	$0xfffffff0, %esp
+
+	/* Setup first parameter to setup_gdt */
+	movl	%esp, %eax
+
+	/* Reserve space for global descriptor table */
+	subl	$X86_GDT_SIZE, %esp
+
+	/* Align temporary global descriptor table to 16-byte boundary */
+	andl	$0xfffffff0, %esp
+
+	/* Set second parameter to setup_gdt */
+	movl	%esp, %edx
+
+	/* gd->gd_addr = gd (Required to allow gd->xyz to work) */
+	movl	%eax, (%eax)
+
+	/* Setup global descriptor table so gd->xyz works */
+	call	setup_gdt
 
 	/* Set parameter to board_init_f() to boot flags */
 	xorl	%eax, %eax
@@ -113,9 +133,42 @@ board_init_f_r_trampoline:
 	 * %eax = Address of top of new stack
 	 */
 
-	/* Setup stack in RAM */
+	/* Stack grows down from top of SDRAM */
 	movl	%eax, %esp
 
+	/* Reserve space on stack for global data */
+	subl	$GENERATED_GBL_DATA_SIZE, %esp
+
+	/* Align global data to 16-byte boundary */
+	andl	$0xfffffff0, %esp
+
+	/* Setup first parameter to memcpy (and setup_gdt) */
+	movl	%esp, %eax
+
+	/* Setup second parameter to memcpy */
+	fs movl 0, %edx
+
+	/* Set third parameter to memcpy */
+	movl	$GENERATED_GBL_DATA_SIZE, %ecx
+
+	/* Copy global data from CAR to SDRAM stack */
+	call	memcpy
+
+	/* Reserve space for global descriptor table */
+	subl	$X86_GDT_SIZE, %esp
+
+	/* Align global descriptor table to 16-byte boundary */
+	andl	$0xfffffff0, %esp
+
+	/* Set second parameter to setup_gdt */
+	movl	%esp, %edx
+
+	/* gd->gd_addr = gd (Required to allow gd->xyz to work) */
+	movl	%eax, (%eax)
+
+	/* Setup global descriptor table so gd->xyz works */
+	call	setup_gdt
+
 	/* Re-enter U-Boot by calling board_init_f_r */
 	call	board_init_f_r
 
diff --git a/arch/x86/include/asm/global_data.h b/arch/x86/include/asm/global_data.h
index 13a3ce8..59c2a90 100644
--- a/arch/x86/include/asm/global_data.h
+++ b/arch/x86/include/asm/global_data.h
@@ -54,7 +54,6 @@ struct global_data {
 	unsigned long	relocaddr;	/* Start address of U-Boot in RAM */
 	unsigned long	start_addr_sp;	/* start_addr_stackpointer */
 	unsigned long	gdt_addr;	/* Location of GDT */
-	unsigned long	new_gd_addr;	/* New location of Global Data */
 	phys_size_t	ram_size;	/* RAM size */
 	unsigned long	reset_status;	/* reset status register at boot */
 	void		**jt;		/* jump table */
diff --git a/arch/x86/include/asm/init_helpers.h b/arch/x86/include/asm/init_helpers.h
index 8afb443..ade694f 100644
--- a/arch/x86/include/asm/init_helpers.h
+++ b/arch/x86/include/asm/init_helpers.h
@@ -29,7 +29,6 @@ int display_dram_config(void);
 int init_baudrate_f(void);
 int calculate_relocation_address(void);
 
-int copy_gd_to_ram_f_r(void);
 int init_cache_f_r(void);
 
 int set_reloc_flag_r(void);
diff --git a/arch/x86/include/asm/processor.h b/arch/x86/include/asm/processor.h
index 6eb5180..17f27cb 100644
--- a/arch/x86/include/asm/processor.h
+++ b/arch/x86/include/asm/processor.h
@@ -41,6 +41,7 @@ enum {
 #else
 /* NOTE: If the above enum is modified, this define must be checked */
 #define X86_GDT_ENTRY_32BIT_DS	3
+#define X86_GDT_NUM_ENTRIES	7
 #endif
 
 #define X86_GDT_SIZE		(X86_GDT_NUM_ENTRIES * X86_GDT_ENTRY_SIZE)
diff --git a/arch/x86/lib/board.c b/arch/x86/lib/board.c
index e5caf13..a13f5c0 100644
--- a/arch/x86/lib/board.c
+++ b/arch/x86/lib/board.c
@@ -36,6 +36,7 @@
 #include <stdio_dev.h>
 #include <asm/u-boot-x86.h>
 #include <asm/relocate.h>
+#include <asm/processor.h>
 
 #include <asm/init_helpers.h>
 #include <asm/init_wrappers.h>
@@ -121,7 +122,6 @@ init_fnc_t *init_sequence_f[] = {
  * initialise the CPU caches (to speed up the relocation process)
  */
 init_fnc_t *init_sequence_f_r[] = {
-	copy_gd_to_ram_f_r,
 	init_cache_f_r,
 	copy_uboot_to_ram,
 	clear_bss,
diff --git a/arch/x86/lib/init_helpers.c b/arch/x86/lib/init_helpers.c
index 2f718d7..87c7263 100644
--- a/arch/x86/lib/init_helpers.c
+++ b/arch/x86/lib/init_helpers.c
@@ -83,18 +83,8 @@ int calculate_relocation_address(void)
 	 *       requirements
 	 */
 
-	/* Global Data is at top of available memory */
+	/* Stack is at top of available memory */
 	dest_addr = gd->ram_size;
-	dest_addr -= GENERATED_GBL_DATA_SIZE;
-	dest_addr &= ~15;
-	gd->new_gd_addr = dest_addr;
-
-	/* GDT is below Global Data */
-	dest_addr -= X86_GDT_SIZE;
-	dest_addr &= ~15;
-	gd->gdt_addr = dest_addr;
-
-	/* Stack is below GDT */
 	gd->start_addr_sp = dest_addr;
 
 	/* U-Boot is below the stack */
@@ -107,31 +97,6 @@ int calculate_relocation_address(void)
 	return 0;
 }
 
-int copy_gd_to_ram_f_r(void)
-{
-	gd_t *ram_gd;
-
-	/*
-	 * Global data is still in temporary memory (the CPU cache).
-	 * calculate_relocation_address() has set gd->new_gd_addr to
-	 * where the global data lives in RAM but getting it there
-	 * safely is a bit tricky due to the 'F-Segment Hack' that
-	 * we need to use for x86
-	 */
-	ram_gd = (gd_t *)gd->new_gd_addr;
-	memcpy((void *)ram_gd, gd, sizeof(gd_t));
-
-	/*
-	 * Reload the Global Descriptor Table so FS points to the
-	 * in-RAM copy of Global Data (calculate_relocation_address()
-	 * has already calculated the in-RAM location of the GDT)
-	 */
-	ram_gd->gd_addr = ram_gd;
-	init_gd(ram_gd, (u64 *)gd->gdt_addr);
-
-	return 0;
-}
-
 int init_cache_f_r(void)
 {
 	/* Initialise the CPU cache(s) */