Skip to content

kboot: accept hyphenated USB4 PCIe-C aliases - #273

Closed
DjDeveloperr wants to merge 1 commit into
mainfrom
fix/j613-pcie-hyphen-alias
Closed

DjDeveloperr wants to merge 1 commit into
mainfrom
fix/j613-pcie-hyphen-alias

Conversation

@DjDeveloperr

@DjDeveloperr DjDeveloperr commented Nov 16, 2026 •

Copy link
Copy Markdown
Contributor

Superseded by Aura #302. Both PRs have the identical production fallback. Linux #1490 now tracks #302 for the paired Stage 2. The affected-hardware test and loader adoption are still required before DTB rollout. This branch retains the real-FDT unchanged-content check and compiled old-lookup control for a possible test-only follow-up.

CI checkpoint: all 15 distinct checks pass on 355cf2398015d99c086e88ab87a769de670086d6 (16 successful runs, including duplicate policy runs). The GCC Python suite reports 763 passed, 32 skipped and nine warnings; both USB4 alias tests ran successfully, including the compiled old-lookup control. Firmware builds, Stage 1/2 size checks and other platform suites pass. No hardware result or warning-free claim follows from this.

Linux #1490's new J613 PCIe-C aliases need valid hyphenated names to pass DT schema validation. dt_set_pcie_tunables() currently requests only usb4_N_pcie, so a renamed DTB would silently lose its controller and port tunables. Add the same legacy-first fallback used by ACIO/NHI/PCIe-adapter lookups: try usb4-N-pcie only when the legacy alias is absent.

Extend the existing real-FDT host replay for legacy-only, hyphen-only, both-name priority, missing aliases, mixed ports, and unchanged FDT contents. The new test reproduced a compiled runtime assertion before the production fix; with the fix it passes and a compiled old-lookup control fails its assertion. The existing adapter replay also passes. No private firmware or traces were used, and no hardware/loader/default changes were performed.

Keep draft: the repository requires affected-hardware testing before merge. This Aura must also be built, packaged and adopted before the renamed J613 DTBs roll out; source publication alone does not clear Linux #1490's loader-compatibility gate.

Signed-off-by: DjDeveloperr <hey@djdev.me>
DjDeveloperr added a commit to satchlj/linux that referenced this pull request Nov 16, 2026
Describe the T8122 PCIe-C compatible and its eight named resources,
and the ACIO post-init power-domain providers. Use valid hyphenated
names for the six new J613 USB4 aliases, retaining their node targets
and the inherited legacy aliases.

The renamed PCIe-C aliases require the legacy-first lookup fallback
in aurora-silicon/aura#273, built and adopted with the new DTBs.
Do not roll these DTBs out with an older incompatible Stage 2.

Signed-off-by: Diljit Singh <hey@djdev.me>
@RyanTheTide

Copy link
Copy Markdown
Contributor

This is the same production change as #302 (the same two lines in dt_set_pcie_tunables(), src/kboot.c:3523), and #302 is on current main and ready for review. Both PRs extend tests/python/test_usb4_alias_fallback.py at the same spot, so they will conflict. Suggest closing this one in favour of #302. The extra checks here (unchanged FDT contents and the compiled old-lookup control) are useful; they could follow as a small test-only PR on top of #302. Happy to help rebase them.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants