nios2: fix r15 issue for gcc4

The "-ffixed-r15" option doesn't work well for gcc4. Since we
don't use gp for small data with option "-G0", we can use gp
as global data pointer. This allows compiler to use r15. It
is necessary for gcc4 to work properly.

Signed-off-by: Thomas Chou <thomas@wytron.com.tw>
Signed-off-by: Scott McNutt <smcnutt@psyent.com>
diff --git a/README b/README
index 81692c0..acb3308 100644
--- a/README
+++ b/README
@@ -4023,6 +4023,14 @@
 
     ==> U-Boot will use R8 to hold a pointer to the global data
 
+On Nios II, the ABI is documented here:
+	http://www.altera.com/literature/hb/nios2/n2cpu_nii51016.pdf
+
+    ==> U-Boot will use gp to hold a pointer to the global data
+
+    Note: on Nios II, we give "-G0" option to gcc and don't use gp
+    to access small data sections, so gp is free.
+
 NOTE: DECLARE_GLOBAL_DATA_PTR must be used with file-global scope,
 or current versions of GCC may "optimize" the code too much.
 
diff --git a/arch/nios2/config.mk b/arch/nios2/config.mk
index 8e5d6ef..6789038 100644
--- a/arch/nios2/config.mk
+++ b/arch/nios2/config.mk
@@ -27,6 +27,6 @@
 STANDALONE_LOAD_ADDR = 0x02000000 -L $(gcclibdir)
 
 PLATFORM_CPPFLAGS += -DCONFIG_NIOS2 -D__NIOS2__
-PLATFORM_CPPFLAGS += -ffixed-r15 -G0
+PLATFORM_CPPFLAGS += -G0
 
 LDSCRIPT ?= $(SRCTREE)/$(CPUDIR)/u-boot.lds
diff --git a/arch/nios2/cpu/start.S b/arch/nios2/cpu/start.S
index d1016ea..76d3b52 100644
--- a/arch/nios2/cpu/start.S
+++ b/arch/nios2/cpu/start.S
@@ -113,13 +113,6 @@
 	 bne	r5, r6, 4b
 5:
 
-	/* GLOBAL POINTER -- the global pointer is used to reference
-	 * "small data" (see -G switch). The linker script must
-	 * provide the gp address.
-	 */
-	 movhi	gp, %hi(_gp)
-	 ori	gp, gp, %lo(_gp)
-
 	/* JUMP TO RELOC ADDR */
 	movhi	r4, %hi(_reloc)
 	ori	r4, r4, %lo(_reloc)
diff --git a/arch/nios2/include/asm/global_data.h b/arch/nios2/include/asm/global_data.h
index 34aa962..f1b3482 100644
--- a/arch/nios2/include/asm/global_data.h
+++ b/arch/nios2/include/asm/global_data.h
@@ -48,6 +48,6 @@
 #define	GD_FLG_LOGINIT	0x00020		/* Log Buffer has been initialized	*/
 #define GD_FLG_DISABLE_CONSOLE	0x00040		/* Disable console (in & out)	 */
 
-#define DECLARE_GLOBAL_DATA_PTR     register gd_t *gd asm ("r15")
+#define DECLARE_GLOBAL_DATA_PTR     register gd_t *gd asm ("gp")
 
 #endif /* __ASM_NIOS2_GLOBALDATA_H_ */
diff --git a/doc/README.standalone b/doc/README.standalone
index 885c92f..6381087 100644
--- a/doc/README.standalone
+++ b/doc/README.standalone
@@ -19,12 +19,12 @@
    thus the compiler cannot perform type checks on these assignments.
 
 2. The pointer to the jump table is passed to the application in a
-   machine-dependent way. PowerPC, ARM, MIPS and Blackfin architectures
-   use a dedicated register to hold the pointer to the 'global_data'
-   structure: r2 on PowerPC, r8 on ARM, k0 on MIPS, and P3 on Blackfin.
-   The x86 architecture does not use such a register; instead, the
-   pointer to the 'global_data' structure is passed as 'argv[-1]'
-   pointer.
+   machine-dependent way. PowerPC, ARM, MIPS, Blackfin and Nios II
+   architectures use a dedicated register to hold the pointer to the
+   'global_data' structure: r2 on PowerPC, r8 on ARM, k0 on MIPS,
+   P3 on Blackfin and gp on Nios II. The x86 architecture does not
+   use such a register; instead, the pointer to the 'global_data'
+   structure is passed as 'argv[-1]' pointer.
 
    The application can access the 'global_data' structure in the same
    way as U-Boot does:
@@ -56,6 +56,7 @@
 	ARM		0x0c100000	0x0c100000
 	MIPS		0x80200000	0x80200000
 	Blackfin	0x00001000	0x00001000
+	Nios II		0x02000000	0x02000000
 
    For example, the "hello world" application may be loaded and
    executed on a PowerPC board with the following commands:
diff --git a/examples/standalone/stubs.c b/examples/standalone/stubs.c
index ce3371d..0187708 100644
--- a/examples/standalone/stubs.c
+++ b/examples/standalone/stubs.c
@@ -84,7 +84,7 @@
 	: : "i"(offsetof(gd_t, jt)), "i"(XF_ ## x) : "r0");
 #elif defined(CONFIG_NIOS2)
 /*
- * r15 holds the pointer to the global_data, r8 is call-clobbered
+ * gp holds the pointer to the global_data, r8 is call-clobbered
  */
 #define EXPORT_FUNC(x) \
 	asm volatile (			\
@@ -92,11 +92,11 @@
 #x ":\n"				\
 "	movhi	r8, %%hi(%0)\n"		\
 "	ori	r8, r0, %%lo(%0)\n"	\
-"	add	r8, r8, r15\n"		\
+"	add	r8, r8, gp\n"		\
 "	ldw	r8, 0(r8)\n"		\
 "	ldw	r8, %1(r8)\n"		\
 "	jmp	r8\n"			\
-	: : "i"(offsetof(gd_t, jt)), "i"(XF_ ## x * sizeof(void *)) : "r15");
+	: : "i"(offsetof(gd_t, jt)), "i"(XF_ ## x * sizeof(void *)) : "gp");
 #elif defined(CONFIG_M68K)
 /*
  * d7 holds the pointer to the global_data, a0 is a call-clobbered