Skip to content

J873: enable the native N1 PCIe radio path - #1013

Merged
Acelogic merged 3 commits into
aurora-wipfrom
j873-n1-public
Nov 6, 2026
Merged

Acelogic merged 3 commits into
aurora-wipfrom
j873-n1-public

Conversation

@Acelogic

@Acelogic Acelogic commented Nov 6, 2026 •

Copy link
Copy Markdown
Member

J873 boots from aurora-wip without exposing its N1 radio. Add the T8152 PCIe reference-clock/PIPE-reset sequence and describe APCIE port 0, GPIO reset, MSI allocation and per-function DART streams in the existing J873 RAM-boot tree. The existing CPU park and RAM handoff remain in place.

The PCIe sequence adapts Thomas Glanzmann’s public linux-aurora commit d85dcbe23a28aa6db3629140510a64b2b046c2df, preserving authorship. Device-tree resources were checked against the physical J873g ADT. This depends on #998 for the native N1 Wi-Fi/Bluetooth drivers, including its AP-switching fix. It does not import the APIF driver from #828.

The previous merge hold is resolved by the tested #998 disconnect-budget fix. The native firmware starts, all three PCI functions appear, and Control/Alpha each receive 16 MSI vectors. Prior targeted validation of the three binding schemas and board DT passed; this revision adds no further device-tree changes.

Validation of the combined public-base + #998 + #1013 source tree on physical J813 and J873, using fresh guarded RAM boots:

  • Kernel, 190 modules and both board DTBs built; 185/185 existing host-suite jobs passed locally. No hosted check run was available for this revision, so this is not a hosted-CI result.
  • 16 AP changes on a TP-Link Deco mesh, including four cross-AP cases per board and shared-AP controls: 240/240 ping replies, immediate first replies, and 32 SHA-256-verified 8 MiB transfers. Wi-Fi power saving remained enabled.
  • Ten-minute simultaneous bidirectional TCP: 30.00/30.00 Mbit/s received at a configured 30 Mbit/s cap, with concurrent Bluetooth LE discovery for 15 seconds per board. These are capped smoke tests, not peak-throughput measurements.
  • J813 passes 16/16 initial GPU/health checks; J873 passes 23/23. Both repeat the GPU smoke after the radio soak with 12,288 exact pixels each, no kernel taint and no recorded kernel crash.

Earlier failed trials and packet captures are retained privately. Isolated packet loss also occurred without an AP change during control testing; this result qualifies the reproduced handoff stall, not loss-free RF operation. Bluetooth pairing/ACL/audio, suspend/resume and installed-rootfs stability remain outside this qualification. No firmware or internal-disk installation was performed.

Private firmware remains locally extracted and is not included. The full combined tree exactly matches the tree used for the hardware build; PMU guards, payload readback and CPU-park checks remain intact.

Attribution

  • Original T8152 PCIe initialization sequence: Thomas Glanzmann (@sithglan), source d85dcbe23a28, retained with his authorship in commit 8bbc3ed23fb4.
  • Current-tree PCIe/DART/GPIO bindings, J873 radio device-tree integration, ADT verification, and both-board qualification: @Acelogic, with Codex assistance.
  • The Linux #998 dependency combines several authors' work: native Wi-Fi/Bluetooth foundations by @Acelogic (efa93c24e38d, a03e01fbb531); J873 board, per-function DMA, firmware scan/join, and Bluetooth adaptations by Joshua Warren (@joshuaswarren); and the real-MSI data path and Alpha MSI allocation by @sithglan (68d0acaba211, eb01c70f9380). The AP-switch disconnect-budget fix is @Acelogic's commit 298fee3294c6, developed and hardware-tested with Codex assistance.

Acelogic and others added 3 commits November 6, 2026 01:21
Add the T8152 compatibles used by the J873 N1 path. Describe the interleaved port and PHY register resources of its PCIe host.

Signed-off-by: Acelogic <miguelc4600@gmail.com>
Power the reference clock buffer and release the lane PIPE reset after PERST#. Adapt the public J873 host-controller change to the current aurora-wip hardware descriptors without changing other SoC sequences.

Adapted from aurora-silicon/linux-aurora commit d85dcbe23a28aa6db3629140510a64b2b046c2df.

Signed-off-by: Thomas Glanzmann <thomas@glanzmann.de>
Signed-off-by: Acelogic <miguelc4600@gmail.com>
Describe APCIE port 0, its GPIO reset, MSI window and separate DART streams for the N1 control, Wi-Fi and Bluetooth functions. Retain the Aura RAM handoff and secondary CPU park contract.

Resources follow the captured J873g ADT and Thomas Glanzmann’s public t8152-j873-pcie.dts. Ethernet remains outside this change.

Signed-off-by: Acelogic <miguelc4600@gmail.com>
@Acelogic
Acelogic marked this pull request as ready for review November 6, 2026 09:01
@Acelogic
Acelogic merged commit 728febe into aurora-wip Nov 6, 2026
RyanTheTide added a commit that referenced this pull request Nov 6, 2026
Carries the work merged on aurora-wip since #1018 into the release
branch: M5 T8142/J813 bring-up (#995 #996 #997 #998 #1001 #1002 #1003),
T8152/J873 bring-up (#984 #1009 #1013 #1022), J813 AVD/ISP (#1026)
and the selftest harness updates (#1014). The merge is clean; the 13
integration/aurora commits since #1018 are kept.

Signed-off-by: Ryan Murray <81541646+RyanTheTide@users.noreply.github.com>
Assisted-by: LLM
@RyanTheTide
RyanTheTide deleted the j873-n1-public branch November 7, 2026 00:54
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