RT #72064. Added udelay(100) before checking the status of the write op.

drivers/mtd/cfi_flash.c

@@ -752,6 +752,10 @@ static int flash_write_cfiword (flash_info_t * info, ulong dest,
 	if (!sect_found)
 		sect = find_sector (info, dest);
 
+#ifdef CONFIG_SYS_STM32
+	/* FIXME: remove this workaround for ST-MEM */
+	udelay(100);
+#endif
 	return flash_full_status_check (info, sect, info->write_tout, "write");
 }
 
@@ -984,13 +988,16 @@ int flash_erase (flash_info_t * info, int s_first, int s_last)
 				flash_unlock_seq (info, sect);
 				flash_write_cmd (info, sect, 0,
 						 AMD_CMD_ERASE_SECTOR);
-				/* Wait before checking the status.
+#ifdef CONFIG_SYS_STM32
+				/* FIXME: remove this workaround for ST-MEM.
+				 * Wait before checking the status.
 				 * This is added after the similar delay in
 				 * linux/drivers/mtd/chips/cfi_cmdset_0002.c:do_erase_oneblock().
 				 * FIXME: why erase_blk_tout is so big, vs to
 				 * the value in the Linux mtd driver?
 				 */
 				udelay(500000);
+#endif
 				break;
 #ifdef CONFIG_FLASH_CFI_LEGACY
 			case CFI_CMDSET_AMD_LEGACY: