x86: Use fs for global data

Use the base address of the 'F' segment as a pointer to the global data
structure. By adding the linear address (i.e. the 'D' segment address) as
the first word of the global data structure, the address of the global data
relative to the 'D' segment can be found simply, for example, by:

	fs movl 0, %eax

This makes the gd 'pointer' writable prior to relocation (by reloading the
Global Desctriptor Table) which brings x86 into line with all other arches

NOTE: Writing to the gd 'pointer' is expensive (but we only do it
twice) but using it to access global data members (read and write) is
still fairly cheap

--
Changes for v2:
 - Rebased against changes made to patch #3
 - Removed extra indent
 - Tweaked commit message

6 files changed, 94 insertions, 47 deletions

diff --git a/arch/x86/cpu/cpu.c b/arch/x86/cpu/cpu.c
index 8102fb951..8c3b92c78 100644
--- a/arch/x86/cpu/cpu.c
+++ b/arch/x86/cpu/cpu.c
@@ -90,6 +90,37 @@ 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 = (ulong)id;
+	setup_gdt(id, gdt_addr);
+}
+
+void setup_gdt(gd_t *id, u64 *gdt_addr)
+{
+	/* CS: code, read/execute, 4 GB, base 0 */
+	gdt_addr[X86_GDT_ENTRY_32BIT_CS] = GDT_ENTRY(0xc09b, 0, 0xfffff);
+
+	/* DS: data, read/write, 4 GB, base 0 */
+	gdt_addr[X86_GDT_ENTRY_32BIT_DS] = GDT_ENTRY(0xc093, 0, 0xfffff);
+
+	/* FS: data, read/write, 4 GB, base (Global Data Pointer) */
+	gdt_addr[X86_GDT_ENTRY_32BIT_FS] = GDT_ENTRY(0xc093, (ulong)id, 0xfffff);
+
+	/* 16-bit CS: code, read/execute, 64 kB, base 0 */
+	gdt_addr[X86_GDT_ENTRY_16BIT_CS] = GDT_ENTRY(0x109b, 0, 0x0ffff);
+
+	/* 16-bit DS: data, read/write, 64 kB, base 0 */
+	gdt_addr[X86_GDT_ENTRY_16BIT_DS] = GDT_ENTRY(0x1093, 0, 0x0ffff);
+
+	load_gdt(gdt_addr, X86_GDT_NUM_ENTRIES);
+	load_ds(X86_GDT_ENTRY_32BIT_DS);
+	load_es(X86_GDT_ENTRY_32BIT_DS);
+	load_gs(X86_GDT_ENTRY_32BIT_DS);
+	load_ss(X86_GDT_ENTRY_32BIT_DS);
+	load_fs(X86_GDT_ENTRY_32BIT_FS);
+}
+
 int x86_cpu_init_f(void)
 {
 	const u32 em_rst = ~X86_CR0_EM;
@@ -117,28 +148,6 @@ int x86_cpu_init_r(void)
 	    "movl	%%eax, %%cr0\n"
 	    "wbinvd\n" : : "i" (nw_cd_rst) : "eax");
 
-	/*
-	 * There are machines which are known to not boot with the GDT
-	 * being 8-byte unaligned. Intel recommends 16 byte alignment
-	 */
-	static const u64 boot_gdt[] __aligned(16) = {
-		/* CS: code, read/execute, 4 GB, base 0 */
-		[X86_GDT_ENTRY_32BIT_CS] = GDT_ENTRY(0xc09b, 0, 0xfffff),
-		/* DS: data, read/write, 4 GB, base 0 */
-		[X86_GDT_ENTRY_32BIT_DS] = GDT_ENTRY(0xc093, 0, 0xfffff),
-		/* 16-bit CS: code, read/execute, 64 kB, base 0 */
-		[X86_GDT_ENTRY_16BIT_CS] = GDT_ENTRY(0x109b, 0, 0x0ffff),
-		/* 16-bit DS: data, read/write, 64 kB, base 0 */
-		[X86_GDT_ENTRY_16BIT_DS] = GDT_ENTRY(0x1093, 0, 0x0ffff),
-	};
-
-	load_gdt(boot_gdt, X86_GDT_NUM_ENTRIES);
-	load_ds(X86_GDT_ENTRY_32BIT_DS);
-	load_es(X86_GDT_ENTRY_32BIT_DS);
-	load_fs(X86_GDT_ENTRY_32BIT_DS);
-	load_gs(X86_GDT_ENTRY_32BIT_DS);
-	load_ss(X86_GDT_ENTRY_32BIT_DS);
-
 	/* Initialize core interrupt and exception functionality of CPU */
 	cpu_init_interrupts();
 	return 0;
diff --git a/arch/x86/cpu/start.S b/arch/x86/cpu/start.S
index 69a9b2cdd..ee0dabe4b 100644
--- a/arch/x86/cpu/start.S
+++ b/arch/x86/cpu/start.S
@@ -31,7 +31,7 @@
 #include <asm/global_data.h>
 #include <asm/processor.h>
 #include <asm/processor-flags.h>
-#include <generated/asm-offsets.h>
+#include <generated/generic-asm-offsets.h>
 
 .section .text
 .code32
@@ -85,6 +85,12 @@ car_init_ret:
 	 */
 	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;
+
 	/* Set parameter to board_init_f() to boot flags */
 	xorl	%eax, %eax
 	movw	%bx, %ax
diff --git a/arch/x86/include/asm/global_data.h b/arch/x86/include/asm/global_data.h
index 05a2139d0..908a02c90 100644
--- a/arch/x86/include/asm/global_data.h
+++ b/arch/x86/include/asm/global_data.h
@@ -36,6 +36,8 @@
 #ifndef __ASSEMBLY__
 
 typedef	struct global_data {
+	/* NOTE: gd_addr MUST be first member of struct global_data! */
+	unsigned long	gd_addr;	/* Location of Global Data */
 	bd_t		*bd;
 	unsigned long	flags;
 	unsigned long	baudrate;
@@ -51,13 +53,24 @@ typedef	struct global_data {
 	unsigned long	bus_clk;
 	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 */
 	char		env_buf[32];	/* buffer for getenv() before reloc. */
 } gd_t;
 
-extern gd_t *gd;
+static inline gd_t *get_fs_gd_ptr(void)
+{
+	gd_t *gd_ptr;
+
+	asm volatile("fs movl 0, %0\n" : "=r" (gd_ptr));
+
+	return gd_ptr;
+}
+
+#define gd	get_fs_gd_ptr()
 
 #endif
 
@@ -73,12 +86,6 @@ extern gd_t *gd;
 #define GD_FLG_DISABLE_CONSOLE	0x00040	/* Disable console (in & out)		*/
 #define GD_FLG_ENV_READY	0x00080	/* Environment imported into hash table	*/
 
-#if 0
 #define DECLARE_GLOBAL_DATA_PTR
-#else
-#define XTRN_DECLARE_GLOBAL_DATA_PTR    extern
-#define DECLARE_GLOBAL_DATA_PTR     XTRN_DECLARE_GLOBAL_DATA_PTR \
-gd_t *gd
-#endif
 
 #endif /* __ASM_GBL_DATA_H */
diff --git a/arch/x86/include/asm/processor.h b/arch/x86/include/asm/processor.h
index aa8188e51..6eb518063 100644
--- a/arch/x86/include/asm/processor.h
+++ b/arch/x86/include/asm/processor.h
@@ -33,6 +33,7 @@ enum {
 	X86_GDT_ENTRY_UNUSED,
 	X86_GDT_ENTRY_32BIT_CS,
 	X86_GDT_ENTRY_32BIT_DS,
+	X86_GDT_ENTRY_32BIT_FS,
 	X86_GDT_ENTRY_16BIT_CS,
 	X86_GDT_ENTRY_16BIT_DS,
 	X86_GDT_NUM_ENTRIES
diff --git a/arch/x86/include/asm/u-boot-x86.h b/arch/x86/include/asm/u-boot-x86.h
index c3d227783..5540d5176 100644
--- a/arch/x86/include/asm/u-boot-x86.h
+++ b/arch/x86/include/asm/u-boot-x86.h
@@ -37,6 +37,8 @@ int x86_cpu_init_r(void);
 int cpu_init_r(void);
 int x86_cpu_init_f(void);
 int cpu_init_f(void);
+void init_gd(gd_t *id, u64 *gdt_addr);
+void setup_gdt(gd_t *id, u64 *gdt_addr);
 
 /* cpu/.../timer.c */
 void timer_isr(void *);
diff --git a/arch/x86/lib/board.c b/arch/x86/lib/board.c
index 6f075b7aa..b64c2d3ca 100644
--- a/arch/x86/lib/board.c
+++ b/arch/x86/lib/board.c
@@ -42,20 +42,12 @@
 #include <serial.h>
 #include <asm/u-boot-x86.h>
 #include <elf.h>
+#include <asm/processor.h>
 
 #ifdef CONFIG_BITBANGMII
 #include <miiphy.h>
 #endif
 
-/*
- * Pointer to initial global data area
- *
- * Here we initialize it.
- */
-#undef	XTRN_DECLARE_GLOBAL_DATA_PTR
-#define XTRN_DECLARE_GLOBAL_DATA_PTR	/* empty = allocate here */
-DECLARE_GLOBAL_DATA_PTR = (gd_t *) (CONFIG_SYS_INIT_GD_ADDR);
-
 /************************************************************************
  * Init Utilities							*
  ************************************************************************
@@ -128,6 +120,7 @@ static int calculate_relocation_address(void);
 static int copy_uboot_to_ram(void);
 static int clear_bss(void);
 static int do_elf_reloc_fixups(void);
+static int copy_gd_to_ram(void);
 
 init_fnc_t *init_sequence_f[] = {
 	cpu_init_f,
@@ -146,6 +139,7 @@ init_fnc_t *init_sequence_f[] = {
 };
 
 init_fnc_t *init_sequence_r[] = {
+	copy_gd_to_ram,
 	cpu_init_r,		/* basic cpu dependent setup */
 	board_early_init_r,	/* basic board dependent setup */
 	dram_init,		/* configure available RAM banks */
@@ -157,8 +151,6 @@ init_fnc_t *init_sequence_r[] = {
 	NULL,
 };
 
-gd_t *gd;
-
 static int calculate_relocation_address(void)
 {
 	ulong text_start = (ulong)&__text_start;
@@ -171,8 +163,18 @@ static int calculate_relocation_address(void)
 	 *       requirements
 	 */
 
-	/* Stack is at top of available memory */
+	/* Global Data 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 */
@@ -279,6 +281,31 @@ void board_init_f_r(void)
 		;
 }
 
+static int copy_gd_to_ram(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 = (ulong)ram_gd;
+	init_gd(ram_gd, (u64 *)gd->gdt_addr);
+
+	return 0;
+}
+
 void board_init_r(gd_t *id, ulong dest_addr)
 {
 #if defined(CONFIG_CMD_NET)
@@ -288,15 +315,10 @@ void board_init_r(gd_t *id, ulong dest_addr)
 	ulong size;
 #endif
 	static bd_t bd_data;
-	static gd_t gd_data;
 	init_fnc_t **init_fnc_ptr;
 
 	show_boot_progress(0x21);
 
-	/* Global data pointer is now writable */
-	gd = &gd_data;
-	memcpy(gd, id, sizeof(gd_t));
-
 	/* compiler optimization barrier needed for GCC >= 3.4 */
 	__asm__ __volatile__("" : : : "memory");