You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Publish cheetah's PCIe, GNSS, AoC and modem drivers from omamobile-int to omamobile #117
The Pixel 6 Pro (raven) runs Wi-Fi, GNSS, the AoC (speakers, microphones and sensors) and the modem on the same drivers the Pixel 7 Pro uses. Those drivers are only in omamobile-int, so raven's support for them can't be sent as pull requests against omamobile: each branch would have to carry a copy of cheetah's code.
In omamobile-int but not in omamobile:
Wi-Fi: the gs201 PCIe controller and PHY (drivers/pci/controller/dwc/pcie-gs201.c, drivers/phy/samsung/phy-gs201-pcie.c)
GNSS: the BCM4775x driver (drivers/misc/bcm4775x)
AoC: Trusty (drivers/trusty), GSA and the AoC stack (drivers/soc/google/aoc)
Modem: CPIF (drivers/soc/google/cpif) and the CP PCIe bring-up
Once these are on omamobile, in whatever form suits you, raven's side of each is small: 5 commits for Wi-Fi, 1 for GNSS, 14 for the AoC, and about 26 for the modem. Is publishing them planned?
Thanks for laying this out, and for counting what raven's side would be. It helps to see how much is waiting on these.
Yes, it's planned, but those drivers aren't all equally close, so here's the honest picture (KERNEL-PUBLISH §3 records why each one is still held):
Wi-Fi is next. It's the BCM4389 brcmfmac series and the gs201 PCIe host and PHY it runs on. I prepared it in a curated form once already, and I promised the brcmfmac half on wifi: brcmfmac: shared-memory version 9 firmware, and the BCM4383 (Pixel 9a) #56: the Pixel 9a's work there is waiting on the same series. It needs re-picking onto today's omamobile along with the D3 series, then a run on my phone, and then it gets published.
GNSS, the AoC with Trusty and GSA, and the modem need porting before they can go on omamobile, which is meant to be upstream-ready. The BCM4775x driver is a misc driver with Broadcom's private interface and belongs in drivers/gnss/. The AoC and Trusty are imported Google drivers, tens of thousands of lines not yet in upstream style, and they're the largest porting job left. CPIF and the modem's PCIe are still in their stepped bring-up form. Copying them over as they are would make omamobile a second integration branch, and I'd rather not.
I won't give dates. Wi-Fi comes first, and I'll post here as each piece lands on omamobile.
In the meantime, omamobile-int is published on GitHub as a snapshot. Keeping raven's commits on branches based on it would let people build and test them now. When a piece reaches omamobile, rebase that part and open the PR then. If a raven change turns out not to depend on these drivers, it can come as a PR against omamobile straight away.
Activity
derekross commentedon Oct 28, 2026
Thanks for laying this out, and for counting what raven's side would be. It helps to see how much is waiting on these.
Yes, it's planned, but those drivers aren't all equally close, so here's the honest picture (KERNEL-PUBLISH §3 records why each one is still held):
omamobilealong with the D3 series, then a run on my phone, and then it gets published.omamobile, which is meant to be upstream-ready. The BCM4775x driver is a misc driver with Broadcom's private interface and belongs indrivers/gnss/. The AoC and Trusty are imported Google drivers, tens of thousands of lines not yet in upstream style, and they're the largest porting job left. CPIF and the modem's PCIe are still in their stepped bring-up form. Copying them over as they are would makeomamobilea second integration branch, and I'd rather not.I won't give dates. Wi-Fi comes first, and I'll post here as each piece lands on
omamobile.In the meantime,
omamobile-intis published on GitHub as a snapshot. Keeping raven's commits on branches based on it would let people build and test them now. When a piece reachesomamobile, rebase that part and open the PR then. If a raven change turns out not to depend on these drivers, it can come as a PR againstomamobilestraight away.