Started a ThreadX module port for the Cortex-R52, headers first - #639

Draft

fdesbiens wants to merge 33 commits into

eclipse-threadx:devfrom

fdesbiens:feature/r52-module-port-foundation

Commits on Aug 19, 2026

Started a ThreadX module port for the Cortex-R52, headers first
fdesbiens
committed
7811ef1
Added the PMSAv8-R region setup and fault handler for the module port
fdesbiens
committed
315fb72
Completed the portable half of the R52 module manager sources
fdesbiens
committed
21210e5
Programmed MPU regions directly instead of through PRSELR, and moved the module block
fdesbiens
committed
efe85af
Added the module scheduler's region switch, with the offsets checked at build time
fdesbiens
committed
c95e594
Added the abort-vector fault capture, and found three bugs writing it
fdesbiens
committed
c102b7c
Added the module user-mode entry, the only way a module reaches the kernel
fdesbiens
committed
3e8570d
Added the supervisor call handler: the privilege boundary itself
fdesbiens
committed
140140e
Added the module thread stack build, which decides the mode a module starts in
fdesbiens
committed
fdedbed
Moved module-port threads to SYS mode, which the context paths depend on
fdesbiens
committed
451cb11
Disabled the port dispatch, and added the module-side library
fdesbiens
committed
d5194a4
Added the module preamble, which checks its own declared size
fdesbiens
committed
d02d9f4
Added a sample module that tests the boundary rather than the API
fdesbiens
committed
523ccef
Added the module manager sample, whose expected result is a fault
fdesbiens
committed
dabb02f
Routed the board's SVC and abort vectors to the module port, when built for it
fdesbiens
committed
23c653a
Built the module manager example, and found five ways it can look right and be wrong
fdesbiens
committed
970542e
Built the module as its own image, because one link cannot hold both
fdesbiens
committed
87536af
Gave the kernel a window onto the module area, owned by the scheduler
fdesbiens
committed
a07d21f
Moved the scheduler into System mode, and borrowed SVC for the return
fdesbiens
committed
d8f898e
Seeded r0 and r1 so the module shell entry has its arguments
fdesbiens
committed
6f4b211
Moved the module's kernel gateway out of the kernel's code region
fdesbiens
committed
c1c9396
Drove the manager from a thread, and gave it the object pool it needs
fdesbiens
committed
4d201ae
Stored the preamble's entry points relative to their own words
fdesbiens
committed
61b8111
Added the silicon run and diagnostic scripts for the module demonstration
fdesbiens
committed
f8ed88b
Made modules relocatable, so one blob runs at any address
fdesbiens
committed
09a5c76
Proved the relocation by running the same blob from two addresses
fdesbiens
committed
08776d4
Fixed the module fault-notify path and covered both abort types
fdesbiens
committed
89bc0fe

Navigation Menu

Uh oh!

Started a ThreadX module port for the Cortex-R52, headers first - #639

Started a ThreadX module port for the Cortex-R52, headers first#639
fdesbiens wants to merge 33 commits into
eclipse-threadx:devfrom
fdesbiens:feature/r52-module-port-foundation

Commits on Aug 19, 2026

Started a ThreadX module port for the Cortex-R52, headers first

Added the PMSAv8-R region setup and fault handler for the module port

Completed the portable half of the R52 module manager sources

Programmed MPU regions directly instead of through PRSELR, and moved the module block

Added the module scheduler's region switch, with the offsets checked at build time

Added the abort-vector fault capture, and found three bugs writing it

Added the module user-mode entry, the only way a module reaches the kernel

Added the supervisor call handler: the privilege boundary itself

Added the module thread stack build, which decides the mode a module starts in

Moved module-port threads to SYS mode, which the context paths depend on

Disabled the port dispatch, and added the module-side library

Added the module preamble, which checks its own declared size

Added a sample module that tests the boundary rather than the API

Added the module manager sample, whose expected result is a fault

Routed the board's SVC and abort vectors to the module port, when built for it

Built the module manager example, and found five ways it can look right and be wrong

Built the module as its own image, because one link cannot hold both

Gave the kernel a window onto the module area, owned by the scheduler

Moved the scheduler into System mode, and borrowed SVC for the return

Seeded r0 and r1 so the module shell entry has its arguments

Moved the module's kernel gateway out of the kernel's code region

Drove the manager from a thread, and gave it the object pool it needs

Stored the preamble's entry points relative to their own words

Added the silicon run and diagnostic scripts for the module demonstration

Made modules relocatable, so one blob runs at any address

Proved the relocation by running the same blob from two addresses

Fixed the module fault-notify path and covered both abort types

Commits on Aug 20, 2026

Masked the shared region's limit to the protection granule

Added the module isolation regression on the Armv8-R AEM FVP

Added the shared-region regression to both module examples

Applied the kernel warning set to the module port, and unified the mode names

Woke the CLZ priority search, which had never once run under GCC

Documented the CLZ search, and corrected the readme's AP claim