chore(radio): collapse carried patches into OSSF fork integration pins

[Mikefly123]

Collapse carried module patches into OSSF fork integration pins

All radio-stack module fixes formerly carried as patches/*.patch now live on Open-Source-Space-Foundation fork integration branches (feat/proves-usp-radio). This PR re-pins the modules and deletes the migrated carried-patch machinery. fprime-yamcs-noapp-path.patch (pip-package patch) stays, with its make fprime-venv apply step.

Pin changes

Module	Old	New (feat/proves-usp-radio head)
west.yml usp_zephyr	Lora-net @ bfacd43	OSSF @ 0bf3e208124d43c55fcb7f91bae6a9ac9812ee71
west.yml usp	Lora-net @ 351b201	OSSF @ 8a60e4a16092f1bfa030b18494da267d8ca6a3ba
west.yml + .gitmodules zephyr	zephyrproject-rtos @ v4.4.1	OSSF @ 3838a2802c916accdfa671de77633ee45b69441c
.gitmodules lib/fprime	nasa/fprime @ 8a62e45	OSSF @ baf163f3ba52ecfabaa39b4aa5847a3cecfb2ae6
lib/fprime-zephyr (already OSSF)	c997272	4b8652e15849d21a3655943d30eda65c2dc886c0

Removed patches → integration PRs

Removed patch	Fork PR	Integration PR
0001 RF-switch GPIO	usp_zephyr#1	Open-Source-Space-Foundation/usp_zephyr#7
0002 Zephyr-4.3 Kconfig	usp_zephyr#2	usp_zephyr#7
0003 LR_FHSS_SRC_PATH	usp_zephyr#3	usp_zephyr#7
0006 wakeup t_woff settle	usp_zephyr#4	usp_zephyr#7
0008 RAC API mutex	usp_zephyr#5	usp_zephyr#7
0010 board.yml full_name	usp_zephyr#6	usp_zephyr#7
0009 RP failsafe UNLOCK exemption	usp#1	Open-Source-Space-Foundation/usp#3
0005 CDC-ACM stuck TX FIFO	zephyr#1	Open-Source-Space-Foundation/zephyr#3
0007 CDC-ACM bounded poll-out	zephyr#2	zephyr#3
fprime-com-aggregator-bounded-timeout	fprime#1	Open-Source-Space-Foundation/fprime#5
fprime-sched-tick-drop	fprime#2	fprime#5
(fprime-zephyr pin bump, no carried patch)	fprime-zephyr#19/#20/#21	Open-Source-Space-Foundation/fprime-zephyr#22

Each integration PR links its constituent PRs. Note: fprime-zephyr#22 is draft because it carries the draft P5 PR #20 (flight-adoption conditions pending).

Removed machinery: Makefile targets usp-patches, usp-core-patches, zephyr-patches, the two fprime apply steps inside submodules, and CI's "Apply carried patches" step. patches/README.md rewritten.

Tree equivalence (verified in a scratch workspace built from this branch)

west update + submodule init with NO patch targets; every module checked out at the integration head. Byte-compare vs (old pin + carried patches):

• usp_zephyr: byte-identical.
• zephyr: byte-identical (base v4.4.1-base = v4.4.1 tag).
• usp: only delta = radio-planner lazy-sleep hysteresis (+136 lines, usp PR Simple led-blinker in Proves Reference #2 — perf change in the integration branch, never a carried patch).
• fprime: only deltas = ComAggregator UT extras (+52 lines, part of fprime#1's upstream-shaped commit) and TlmPacketizer I16 offset fix (fprime PR Add Watchdog Petter Component  #3 / issue Propose upstream fprime fix: TlmPacketizer packetOffset is 64-bit, wastes ~27 KiB on 32-bit targets #469 — in the integration branch, carried on main by fix(ram): carry TlmPacketizer I16 packetOffset patch (−27.5 KiB RAM Reduction) #468, never on this branch).
• fprime-zephyr: only deltas = re-arm semaphore (PR Queue.hpp Build Error: Expected identifier before "'" token #19), P5_GMSK_83K profile (PR Board Fails to Compile Due to sx1276, SPI not connected #20), GFSK RX→TX wedge fix (PR works with radio in the v5d board #21) — all intentional integration-branch content.

Regression evidence

Fork-branch stack HIL-regressed on the two-board v5e/GRC bench: Phase A rungs 1–7 (boot, CDC, radio RX, 100× wakeup hammer, TM load, RF profile matrix, end-to-end authenticated commanding over RF) + Phase B rungs 8–10 including the 24 h unattended soak, green 2026-07-26 (28 profile cycles, zero wedge/reboot/CDC events). Reports: logs/HIL-REGRESSION-REPORT-phaseA.md, logs/HIL-REGRESSION-REPORT-phaseB.md.

Bench smoke on THIS branch's build (direct-boot v5e image, banner 3838a2802c91 verified in flash): clean boot, authenticated commanding round-trip (SEND_PKT post seq-align), P0→P4 switch + RF NO_OP on both boards (wedge-fix regression) and P4→P0 restore, //dl200k.bin (178,704 B) downlink completed cleanly. Timing was ~85 s rather than the stacked-image ~35 s because draft perf PRs #475 (ComDelay divider-0 passthrough) and #476 (v5e SPI/TCXO DTS) are PCR-side changes not on this branch — consistent with ComDelay's 10 Hz cap at 721 frames; not a module regression.

🤖 Generated with Claude Code

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on base/target branches other than the default branch.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Organization UI

Review profile: ASSERTIVE

Plan: Pro Plus

Run ID: 8c157ddf-4530-4d05-9c94-2896c26bc573

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}