PCIe under mainline on mica (2026-11-02)
Attempt 2026-11-02-6: kernel/mica-pcie.config on top of the USB and ADSP fragments.
Source: the AP console over SuzyQ (raw log dumps/firstboot/2026-11-02-ap.log).
Result: both PCIe controllers up, both devices enumerate
| Address | Device | Class | Link | Driver |
|---|---|---|---|---|
0006:00:00.0 | root port 17cb:0111 (pcie@1bf8000, gen4x4 PHY 1bfc000) | 0x060400 | 16 GT/s x4 | pcieport |
0006:01:00.0 | NVMe SSD 1344:5427 (vendor 0x1344: Micron) | 0x010802 | 16 GT/s x4 | none (on purpose) |
0004:00:00.0 | root port 17cb:0111 (pcie@1c08000, gen3x2 PHY 1c0e000) | 0x060400 | 8 GT/s x2 | pcieport |
0004:01:00.0 | WCN7850 Wi-Fi 17cb:1107 | 0x028000 | 8 GT/s x2 | none (ath12k probe failed) |
qcom-pcie 1bf8000.pcie: PCIe Gen.4 x4 link upat 1.6 s;1c08000.pcie: PCIe Gen.3 x2 link upat 1.75 s.- Wi-Fi: the WCN7850 PMU and PCI power control powered the chip;
ath12k_wifi7_pcibound, thenDirect firmware load for ath12k/WCN7850/hw2.0/amss.bin failed with error -2→failed to start mhi: -110. No Wi-Fi firmware was shipped. - SSD safety:
BLK_DEV_NVMEis=mand guarded by# forbid-y:in the fragment, so enumeration read config space only; no NVMe command was sent. The SSD is the stock disk and is never written (ADR 0003, ticket 07). - Deferred now: only
arm-scmi(“failed to setup channels”). PCIe, USB, the GPU SMMU, the DP aux bridges andpmic-glinkall probed.
What made it work
The QMP PCIe PHY was =m. The Wi-Fi port also needs PCI_PWRCTRL_PWRSEQ, which has no prompt and is selected only
by ATH11K/ATH12K; ATH12K in turn needs MAC80211, CFG80211 and RFKILL built in
(scripts/dt-kconfig-map.py, scripts/build-kernel.sh demotion check).