[PATCH v3] Add sync to ensure flash_write_cmd is fully finished

Some CPUs like PPC, BLACKFIN need sync() to ensure cfi flash write command
is fully finished. The sync() is defined in each CPU's io.h file. For
those CPUs which do not need sync for now, a dummy sync() is defined in
their io.h as well.

Signed-off-by: Haiying Wang <Haiying.Wang@freescale.com>

drivers/cfi_flash.c

@@ -36,6 +36,7 @@
 
 #include <common.h>
 #include <asm/processor.h>
+#include <asm/io.h>
 #include <asm/byteorder.h>
 #include <environment.h>
 #ifdef	CFG_FLASH_CFI_DRIVER
@@ -931,27 +932,18 @@ static void flash_write_cmd (flash_info_t * info, flash_sect_t sect, uint offset
 		debug ("fwc addr %p cmd %x %x 8bit x %d bit\n", addr.cp, cmd,
 		       cword.c, info->chipwidth << CFI_FLASH_SHIFT_WIDTH);
 		*addr.cp = cword.c;
-#ifdef CONFIG_BLACKFIN
-		asm("ssync;");
-#endif
 		break;
 	case FLASH_CFI_16BIT:
 		debug ("fwc addr %p cmd %x %4.4x 16bit x %d bit\n", addr.wp,
 		       cmd, cword.w,
 		       info->chipwidth << CFI_FLASH_SHIFT_WIDTH);
 		*addr.wp = cword.w;
-#ifdef CONFIG_BLACKFIN
-		asm("ssync;");
-#endif
 		break;
 	case FLASH_CFI_32BIT:
 		debug ("fwc addr %p cmd %x %8.8lx 32bit x %d bit\n", addr.lp,
 		       cmd, cword.l,
 		       info->chipwidth << CFI_FLASH_SHIFT_WIDTH);
 		*addr.lp = cword.l;
-#ifdef CONFIG_BLACKFIN
-		asm("ssync;");
-#endif
 		break;
 	case FLASH_CFI_64BIT:
 #ifdef DEBUG
@@ -966,11 +958,11 @@ static void flash_write_cmd (flash_info_t * info, flash_sect_t sect, uint offset
 		}
 #endif
 		*addr.llp = cword.ll;
-#ifdef CONFIG_BLACKFIN
-		asm("ssync;");
-#endif
 		break;
 	}
+
+	/* Ensure all the instructions are fully finished */
+	sync();
 }
 
 static void flash_unlock_seq (flash_info_t * info, flash_sect_t sect)

include/asm-arm/io.h

@@ -29,6 +29,10 @@
 #include <asm/arch/hardware.h>
 #endif	/* XXX###XXX */
 
+static inline void sync(void)
+{
+}
+
 /*
  * Generic virtual read/write.  Note that we don't support half-word
  * read/writes.  We define __arch_*[bl] here, and leave __arch_*w

include/asm-avr32/io.h

@@ -89,4 +89,8 @@ static __inline__ void * phys_to_virt(unsigned long address)
 
 #endif /* __KERNEL__ */
 
+static inline void sync(void)
+{
+}
+
 #endif /* __ASM_AVR32_IO_H */

include/asm-blackfin/io.h

@@ -25,6 +25,11 @@
 #ifndef _BLACKFIN_IO_H
 #define _BLACKFIN_IO_H
 
+static inline void sync(void)
+{
+	__asm__ __volatile__ asm("ssync" : : : "memory");
+}
+
 #ifdef __KERNEL__
 
 #include <linux/config.h>

include/asm-i386/io.h

@@ -201,4 +201,8 @@ __OUTS(b)
 __OUTS(w)
 __OUTS(l)
 
+static inline void sync(void)
+{
+}
+
 #endif

include/asm-m68k/io.h

@@ -1 +1,8 @@
-/* */
+#ifndef __ASM_M68K_IO_H_
+#define __ASM_M68K_IO_H_
+
+static inline void sync(void)
+{
+}
+
+#endif /* __ASM_M68K_IO_H_ */

include/asm-microblaze/io.h

@@ -125,4 +125,8 @@ io_outsl (unsigned long port, const void *src, unsigned long count)
 #define ioremap_writethrough(physaddr, size)	(physaddr)
 #define ioremap_fullcache(physaddr, size)	(physaddr)
 
+static inline void sync(void)
+{
+}
+
 #endif /* __MICROBLAZE_IO_H__ */

include/asm-mips/io.h

@@ -447,4 +447,8 @@ extern void (*_dma_cache_inv)(unsigned long start, unsigned long size);
 #define dma_cache_wback(start,size)	_dma_cache_wback(start,size)
 #define dma_cache_inv(start,size)	_dma_cache_inv(start,size)
 
+static inline void sync(void)
+{
+}
+
 #endif /* _ASM_IO_H */

include/asm-nios/io.h

@@ -97,4 +97,8 @@ static inline void outsl (unsigned long port, const void *src, unsigned long cou
 	while (count--) outl (*p++, port);
 }
 
+static inline void sync(void)
+{
+}
+
 #endif /* __ASM_NIOS_IO_H_ */

include/asm-nios2/io.h

@@ -24,7 +24,10 @@
 #ifndef __ASM_NIOS2_IO_H_
 #define __ASM_NIOS2_IO_H_
 
-#define sync() asm volatile ("sync" : : : "memory");
+static inline void sync(void)
+{
+	__asm__ __volatile__ ("sync" : : : "memory");
+}
 
 extern unsigned char inb (unsigned char *port);
 extern unsigned short inw (unsigned short *port);

include/asm-ppc/io.h

@@ -95,8 +95,15 @@ extern void _outsl_ns(volatile u32 *port, const void *buf, int nl);
  * Acts as a barrier to ensure all previous I/O accesses have
  * completed before any further ones are issued.
  */
-#define eieio() __asm__ __volatile__ ("eieio" : : : "memory");
-#define sync()  __asm__ __volatile__ ("sync" : : : "memory");
+static inline void eieio(void)
+{
+	__asm__ __volatile__ ("eieio" : : : "memory");
+}
+
+static inline void sync(void)
+{
+	__asm__ __volatile__ ("sync" : : : "memory");
+}
 
 /* Enforce in-order execution of data I/O.
  * No distinction between read/write on PPC; use eieio for all three.