os/, build/configs/: Segregate board specific changes from mkldscript.py by namanjain7 · Pull Request #7226 · Samsung/TizenRT

namanjain7 · 2026-03-31T11:05:22Z

Refactor mkldscript.py file. Add common code in mkldscript.py and board specific address calculation to build/configs//scripts

namanjain7 · 2026-03-31T11:07:16Z

Removed support for common_.ld and app1_.ld
Passing values as arguments to Makefile.unix and loadable.mk using --defsym

amandeep-samsung · 2026-03-31T11:46:59Z

@ewoodev : please review design so we can make changes accordingly in discussed design.

mkldscript does not have board specific code now.
related files are used for board specific address calculation
common_0.ld and app1_0.ld are not created now and variables are used to pass flash addresses in userspace_all.ld for XIP

ewoodev · 2026-04-01T05:31:04Z

 ifeq ($(CONFIG_XIP_ELF),y)
+MEM_VARS := $(shell python $(TOPDIR)/tools/mkldscript.py $(CONFIG_APP1_BIN_NAME))
+$(foreach v,$(MEM_VARS),$(eval $(v)))
+APP1_LD_DEFS := \


This file is common including from app1 app2.

please check app2 case

yes, we are adding both,

app2 and also

OTA index (just like before it was for both cases). We will take OTA index as argument and will fix it by defconfig as slot 0.

ewoodev · 2026-04-01T05:41:33Z

+get_flash_address(binary_type)  # Get flash address and size from board specific code

-    offset = offset + part_size
-    PART_IDX = PART_IDX + 1
+if binary_type == "common":
+    print("RAM_ADD={}".format(common_ram_start))
+    print("RAM_SIZE={}".format(hex(common_ram_size)))
+elif binary_type == "app1":
+    print("RAM_ADD={}".format(app1_ram_start))
+    print("RAM_SIZE={}".format(app1_ram_size))
+elif binary_type == "app2":
+    print("RAM_ADD={}".format(app2_ram_start))
+    print("RAM_SIZE={}".format(app2_ram_size))


why everytime calculate and return only one?

it's better calculating once of argument at that time.

Code modified.
Now, when script is called for common binary, then addresses for app1, app2 will also be calculated and saved in json file. When call for app1, app2 comes, it will fetch from json cache file and return it directly without calculating.

ewoodev · 2026-04-01T05:43:33Z

+def get_flash_address(binary_name):
+    global offset
+    """
+    Get flash address for specified binary (common/app1/app2) in dual OTA mode.
+    For rtl8730e, offset starts from base address and accumulates through partitions.
+    Returns FLASH_ADD and FLASH_SIZE for the specified binary.
+    """
+    index = 0
+    ota_index = 1
+
+    current_offset = offset
+
+    for name in NAME_LIST:
+        part_size = int(SIZE_LIST[index]) * 1024
+
+        if name == "kernel":
+            ota_index = (ota_index + 1) % 2
+        elif name == binary_name:
+            if binary_name == "common":
+                binary_start = hex(current_offset + 0x10 + signing_offset)
+                binary_size = hex(part_size - 0x10 - signing_offset)
+            else:
+                binary_start = hex(current_offset + 0x30 + signing_offset)
+                binary_size = hex(part_size - 0x30 - signing_offset)
+
+            print("FLASH_ADD={}".format(binary_start))
+            print("FLASH_SIZE={}".format(binary_size))
+            return
+
+        current_offset = current_offset + part_size
+        index += 1


This function is board specific?
i think, platform just needs only virture start address of kernel partition

sunghan-chang · 2026-04-07T23:40:08Z

@namanjain7 This is pending for long time. Please let me know status.

namanjain7 · 2026-04-08T05:23:39Z

@namanjain7 This is pending for long time. Please let me know status.

Hello Mr. chang,
we have done the changes that need to be done.
there are some recent makefile changes, and we are merging it with our commit.
We will update ASAP when it's done.

namanjain7 · 2026-04-09T14:53:56Z

we have added dynamic ram start calculation for each partition.
the ram allocation starts from the end.
the script will be called everytime we build for common, app1, and app2.
currently, we are facing issues related to removing duplicate calculation. calculating app1 offset will require calculating common binary offset also which creates duplicate calculation. we are optimizing this.
this task is on priority and we will push the final version ASAP when done.

amandeep-samsung · 2026-04-09T15:08:30Z

+    print(sys.argv[5])
+
+if __name__ == "__main__":
+    get_mem_config()


please remove it

namanjain7 · 2026-04-09T15:44:37Z

we will update flash addressing related changes after more verify

ewoodev · 2026-04-13T07:03:47Z

+COMMON_BINARY_NAME=common
+MEM_VARS := $(shell python $(TOPDIR)/tools/mkldscript.py $(COMMON_BINARY_NAME))
+$(foreach v,$(MEM_VARS),$(eval $(v)))
+COMMON_LD_DEFS := \
+	--defsym __FLASH_START_ADDRESS__=$(FLASH_ADD) \
+	--defsym __FLASH_SIZE__=$(FLASH_SIZE) \
+	--defsym __RAM_START_ADDRESS__=$(RAM_ADD) \
+	--defsym __RAM_SIZE__=$(RAM_SIZE)


this code is added severel code(common app1 app2)
how about make function??

The common binary is build in Makefile.unix. So, this code will execute once.
For app1, app2, --> The makefile is loadable.mk
Both have different makefiles.

ewoodev · 2026-04-13T09:05:47Z

+cfg_file = os.path.join(os_folder, '.config')
+build_folder = os.path.join(os_folder, '..', 'build')
+output_folder = os.path.join(build_folder, 'output', 'bin')
+saved_config_file_path = os.path.join(output_folder, 'mem_layout_cache.json')


i'd like to know why this is needed.

which file is using this file?

cfg_file ->used to get CONFIG_RAM_START, common app1 app2... binary size to get address.
build_folder, output_folder, saved_config_file_path -> For saving json file in output folder. This file will have config values extracted from .config file and memory values.

namanjain7 · 2026-04-16T07:37:17Z

All issues resolved. We have tested in tp1x, tp1x plus, and ailite boards.

mkldscript.py was used to make .ld script containing flash and ram start address and size. The file contained code changes for both rtl8730e and bk7239n. This commit removes this file and make a common file and board specific file. We added new file bin_mem_layout.py which is used to print the data. In makefile, we are reading the data and passing it as an argument in makefile, instead of directly importing .ld scripts.

namanjain7 marked this pull request as draft March 31, 2026 11:05

namanjain7 force-pushed the mkldscript_draft branch from 74890f1 to e5a232b Compare March 31, 2026 11:19

ewoodev reviewed Apr 1, 2026

View reviewed changes

amandeep-samsung reviewed Apr 9, 2026

View reviewed changes

namanjain7 force-pushed the mkldscript_draft branch 2 times, most recently from 00dffd1 to cc05d68 Compare April 9, 2026 15:43

ewoodev reviewed Apr 13, 2026

View reviewed changes

namanjain7 force-pushed the mkldscript_draft branch 2 times, most recently from d22117d to bb87918 Compare April 15, 2026 11:28

namanjain7 marked this pull request as ready for review April 15, 2026 11:28

namanjain7 force-pushed the mkldscript_draft branch from bb87918 to 547e5d4 Compare April 16, 2026 08:30

Conversation

namanjain7 commented Mar 31, 2026

Uh oh!

namanjain7 commented Mar 31, 2026

Uh oh!

amandeep-samsung commented Mar 31, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

sunghan-chang commented Apr 7, 2026

Uh oh!

namanjain7 commented Apr 8, 2026

Uh oh!

namanjain7 commented Apr 9, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

namanjain7 commented Apr 9, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

namanjain7 commented Apr 16, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants