[danielinux]

Target-specific assignments in arch.mk/options.mk/board.mk used hard '=', silently clobbering values the user set via command line, environment, or a sourced .config (e.g. LSCRIPT_IN from PR #958 was defeated on every target with a dedicated script).

Convert the target-default assignments to '?=' (first-set wins) and, for LSCRIPT_IN, capture the user value before the target files are parsed so the last-resort default (hal/$(TARGET).ld) and the ARMClang .sct default only apply when the user provided nothing.

CMake: WOLFBOOT_LSCRIPT_TEMPLATE had the same flaw (plain set() over a user -D value); guard it with NOT DEFINED and let test-app use an absolute user template as-is.

Document LSCRIPT_IN as a user option in docs/compile.md.

[Copilot]

🟡 Changes recommended

Critical make assignment and linker-script selection errors can break builds, use incorrect memory layouts, overwrite custom templates, and leave ChaCha code in flash when encrypted self-updates require it in RAM.

7 open issues

• UBL target selects nonexistent linker script · New
• Invalid make assignment operator breaks affected builds · New
• Stage-1 fallback prevents Tiger Lake script selection · New
• HSM flattening overwrites custom linker template · New
• Full-system CSRM selection is shadowed by earlier default · New
• Architecture default suppresses encrypted-update RAM requirements · New
• Generated bootloader script collides with application output · New

What changed in this PR

This PR adjusts wolfBoot’s Make and CMake configuration so user-provided options take precedence over target defaults, extending the custom linker-script support introduced in #958.

Changes:

• Converts target defaults to conditional assignments and defers the Make linker-script fallback.
• Preserves CMake linker-template overrides and supports absolute template paths.
• Documents the LSCRIPT_IN option and its usage.

File	Description
tools/​config.mk	Makes the multi-sector erase default conditional.
test-app/​CMakeLists.txt	Handles absolute bootloader template paths.
options.mk	Makes the ChaCha RAM linker-script selection conditional.
Makefile	Captures user script selection and defers fallback defaults.
hal/​boards/​stm32wba55cg_nucleo/​board.mk	Makes the board linker-script default conditional.
hal/​boards/​stm32wb_nucleo/​board.mk	Makes the board linker-script default conditional.
docs/​compile.md	Documents custom linker-script configuration.
CMakeLists.txt	Guards linker-template defaults against user overrides.
arch.mk	Converts architecture-specific linker-script defaults to conditional assignments.

_{Review effort: Balanced}

[Copilot]

🟡 Changes recommended

Active moderate findings remain in configuration persistence, stage1 override handling, and CMake path quoting, so the change is not ready for approval.

1 open issue

• Preserve user linker script in stage1 builds · New

3 previously missed issues

In code that hasn't changed since last review

Persist LSCRIPT_IN in generated .config

Makefile:44

LSCRIPT_IN is documented as a .config option, but config/Makefile regenerates .config exclusively from CONFIG_VARS, and that list does not contain LSCRIPT_IN. Therefore make config silently deletes a custom linker-script path before this capture runs, and the next build falls back to the target default. Add LSCRIPT_IN to the persisted configuration variables.

Quote linker-template paths in CMake arguments

test-app/​CMakeLists.txt:114

This new path is intended to accept an arbitrary absolute user template, but the unquoted expansion is split into multiple CMake arguments when the path contains whitespace. That makes IS_ABSOLUTE (and then configure_file) fail for valid user paths such as /work trees/custom.ld; quote the template, resolved source, and destination arguments.

Persist FLASH_MULTI_SECTOR_ERASE in generated config

tools/​config.mk:105

Changing this to ?= only preserves a value while it is present in .config; config/Makefile rebuilds that file from CONFIG_VARS, which omits FLASH_MULTI_SECTOR_ERASE. A user setting from a generated configuration is consequently discarded and this default becomes 0 again, defeating the precedence change. Add FLASH_MULTI_SECTOR_ERASE to CONFIG_VARS as well.

7 resolved since last review

• Generated bootloader script collides with application output
• Architecture default suppresses encrypted-update RAM requirements
• Full-system CSRM selection is shadowed by earlier default
• HSM flattening overwrites custom linker template
• Stage-1 fallback prevents Tiger Lake script selection
• Invalid make assignment operator breaks affected builds
• UBL target selects nonexistent linker script

_{Review effort: Lite}

[Copilot]

🟡 Changes recommended

Two moderate active issues remain: generated configuration drops LSCRIPT_IN, and root-invoked stage1 builds resolve user-supplied relative linker-script paths from the wrong directory.

2 open issues

• Persist LSCRIPT_IN in generated .config · New
• Resolve relative LSCRIPT_IN from the repository root · New

1 resolved since last review

• Preserve user linker script in stage1 builds

_{Review effort: Lite}

[Copilot]

🟡 Changes recommended

Critical linker-script configuration bugs leave empty values from make config and mishandle absolute custom paths, causing normal and x86 FSP builds to fail until defaults and user-provided scripts are handled correctly.

3 open issues

• Empty saved LSCRIPT_IN prevents the fallback script from being selected · New
• Normalize empty LSCRIPT_IN before target and stage1 defaults · New
• Custom x86 FSP linker scripts are mishandled by the stage1 rule · New

2 resolved since last review

• Resolve relative LSCRIPT_IN from the repository root
• Persist LSCRIPT_IN in generated .config

_{Review effort: Lite}

[Copilot]

🟡 Changes recommended

An active moderate Makefile issue causes empty LSCRIPT_IN values to bypass the intended target-specific linker-script selection, so changes are required.

1 open issue

• Empty LSCRIPT_IN skips target-specific linker-script defaults · New

3 resolved since last review

• Normalize empty LSCRIPT_IN before target and stage1 defaults
• Empty saved LSCRIPT_IN prevents the fallback script from being selected
• Custom x86 FSP linker scripts are mishandled by the stage1 rule

_{Review effort: Lite}

[Copilot]

🔵 Needs a closer look

Human assessment is needed because moderate hidden findings reveal a shared command-line precedence issue in fallback linker-script assignments, making the documented empty-value behavior unreliable across targets.

No open issues 🎉

1 resolved since last review

• Empty LSCRIPT_IN skips target-specific linker-script defaults

_{Review effort: Lite}