arm: cortex_a_r linker: move data section between rodata and bss

By moving data section between rodata and bss, and disabling
Kconfig LINKER_LAST_SECTION_ID by default if non-XIP (it is
seems only needed for XIP), the size of zephyr.bin can be
reduced significantly on non-XIP system.

Signed-off-by: Dat Nguyen Duy <[email protected]>

Author: Dat-NguyenDuy

include/zephyr/arch/arm/cortex_a_r/scripts/linker.ld

@@ -282,34 +282,6 @@ SECTIONS
     _app_smem_rom_start = LOADADDR(_APP_SMEM_SECTION_NAME);
 #endif  /* CONFIG_USERSPACE */
 
-    SECTION_DATA_PROLOGUE(_BSS_SECTION_NAME,(NOLOAD), BSS_ALIGN)
-    {
-        /*
-         * For performance, BSS section is assumed to be 4 byte aligned and
-         * a multiple of 4 bytes
-         */
-        . = ALIGN(4);
-        __bss_start = .;
-        __kernel_ram_start = .;
-
-        *(.bss)
-        *(".bss.*")
-        *(COMMON)
-        *(".kernel_bss.*")
-
-#ifdef CONFIG_CODE_DATA_RELOCATION
-#include <linker_sram_bss_relocate.ld>
-#endif
-
-        /*
-         * As memory is cleared in words only, it is simpler to ensure the BSS
-         * section ends on a 4 byte boundary. This wastes a maximum of 3 bytes.
-         */
-        __bss_end = ALIGN(4);
-    } GROUP_DATA_LINK_IN(RAMABLE_REGION, RAMABLE_REGION)
-
-#include <zephyr/linker/common-noinit.ld>
-
     SECTION_DATA_PROLOGUE(_DATA_SECTION_NAME,,)
     {
         __data_region_start = .;
@@ -345,12 +317,6 @@ SECTIONS
 
     __data_region_end = .;
 
-
-    /* Define linker symbols */
-
-    __kernel_ram_end = RAM_ADDR + RAM_SIZE;
-    __kernel_ram_size = __kernel_ram_end - __kernel_ram_start;
-
 #if DT_NODE_HAS_STATUS(DT_CHOSEN(zephyr_ocm), okay)
 GROUP_START(OCM)
 
@@ -382,12 +348,6 @@ GROUP_END(OCM)
  */
 #include <snippets-sections.ld>
 
-#define LAST_RAM_ALIGN . = ALIGN(_region_min_align);
-
-#include <zephyr/linker/ram-end.ld>
-
-    GROUP_END(RAMABLE_REGION)
-
 #include <zephyr/linker/debug-sections.ld>
 
     SECTION_PROLOGUE(.ARM.attributes, 0,)
@@ -413,4 +373,43 @@ GROUP_END(OCM)
      * calculate this value here. */
     _flash_used = LOADADDR(.last_section) + SIZEOF(.last_section) - __rom_region_start;
 
+    SECTION_DATA_PROLOGUE(_BSS_SECTION_NAME,(NOLOAD), BSS_ALIGN)
+    {
+        /*
+         * For performance, BSS section is assumed to be 4 byte aligned and
+         * a multiple of 4 bytes
+         */
+        . = ALIGN(4);
+        __bss_start = .;
+        __kernel_ram_start = .;
+
+        *(.bss)
+        *(".bss.*")
+        *(COMMON)
+        *(".kernel_bss.*")
+
+#ifdef CONFIG_CODE_DATA_RELOCATION
+#include <linker_sram_bss_relocate.ld>
+#endif
+
+        /*
+         * As memory is cleared in words only, it is simpler to ensure the BSS
+         * section ends on a 4 byte boundary. This wastes a maximum of 3 bytes.
+         */
+        __bss_end = ALIGN(4);
+    } GROUP_DATA_LINK_IN(RAMABLE_REGION, RAMABLE_REGION)
+
+#include <zephyr/linker/common-noinit.ld>
+
+    /* Define linker symbols */
+
+    __kernel_ram_end = RAM_ADDR + RAM_SIZE;
+    __kernel_ram_size = __kernel_ram_end - __kernel_ram_start;
+
+#define LAST_RAM_ALIGN . = ALIGN(_region_min_align);
+
+#include <zephyr/linker/ram-end.ld>
+
+    GROUP_END(RAMABLE_REGION)
+
 }