Skip to content

arm64: dts: exynos: google: zuma-husky: the BCM4775x GNSS node on spi21, disabled - #239

Merged
derekross merged 2 commits into
derekross:omamobilefrom
Somnius:husky-gnss-node-1109
Nov 11, 2026
Merged

derekross merged 2 commits into
derekross:omamobilefrom
Somnius:husky-gnss-node-1109

Conversation

@Somnius

@Somnius Somnius commented Nov 10, 2026

Copy link
Copy Markdown
Contributor

@derekross, this is the GNSS node I said would follow #211. It's two commits on today's omamobile (253208c).

The first only says "Tensor G3 and G4" in the S2MPG15 PMIC and regulator binding descriptions, as you suggested on #211. No property changes.

The second adds gnss@0 under spi21 for the brcm,bcm4775-gnss binding, like the 7 Pro's node, disabled. The four lines are stock's bcm4775@0: host request gpa5-3 (also the interrupt, level high), mcu request gpp4-3, nstandby gph3-3, mcu response gph3-2. The supplies are the S2MPG15 LDOs 9, 10 and 11 from #211. The pin groups are in zuma-husky.dts next to the Wi-Fi ones, with the function, pull and power-down values from stock, and no drive strength because stock's groups don't set one. The GPIOs are raw levels with nstandby active high (stock writes it active low), because the driver sets nstandby with raw values. spi-max-frequency 26 MHz and feedback delay 1 are stock's values for this node.

I compared the pins, banks and rails with stock's device tree. checkpatch --strict is clean on both, zuma-husky.dtb builds with W=1 without warnings, and dt-validate on the dtb is quiet for these nodes both as they are and with status switched to okay (the only messages left are the eusb2 phy supplies, which this doesn't touch).

I didn't boot it, so none of it is tested on the phone. It stays disabled until a bring-up boot.

Two things I'm unsure about. Chip select is the controller's native one, since stock has no cs-gpios and no pin group for spi_21. Stock's node also has samsung,spi-chip-select-mode = 0, which this tree's spi driver doesn't read, so I left it out; I don't know if that mode matters for this chip, and I haven't looked at how it fits with #204. And I haven't confirmed the chip itself from the phone, only from stock's node name.

…and G4

The Pixel 8 Pro (Tensor G3) has an S2MPG15 sub-PMIC too, with the same
three GNSS LDOs, as its zuma-husky node shows. Say G3 and G4 in the
descriptions of the PMIC and of its regulators. No property changes.

Signed-off-by: Lef <me@lefteros.com>
… spi21, disabled

The Pixel 8 Pro's GNSS receiver is a BCM4775x on HSI0 USI3 (Google's
spi_21). Add a gnss@0 node for the brcm,bcm4775-gnss binding, as the
Pixel 7 Pro does, with the lines from stock's bcm4775@0: host request
gpa5-3, mcu request gpp4-3, nstandby gph3-3 and mcu response gph3-2, and
the three supplies from the S2MPG15 LDOs (ldo9s, ldo10s, ldo11s). The
four pin groups sit in this file beside the Wi-Fi ones, with the pull and
power-down states stock has.

The GPIOs are raw levels with nstandby active high, not active low as
stock writes it: the driver sets nstandby with raw values. The host
request line is also the interrupt (level high), as on the 7 Pro.

spi-max-frequency (26 MHz) and samsung,spi-feedback-delay (1) are stock's
values for this node. Chip select is the controller's native one, as
stock has no cs-gpios and no pin group for spi_21; the spi driver's
manual_native_cs path handles that. Stock's node also has
samsung,spi-chip-select-mode = <0>, which this tree's spi driver does
not read, so it is left out.

The node is status = "disabled" and sits on the pin controllers and the
gpa5, gph3 and gpp4 banks that omamobile already has, and on spi21 and
the S2MPG15 sub-PMIC (both already in, the latter from the S2MPG15
commit that merged earlier).

Built, not booted.

Signed-off-by: Lef <me@lefteros.com>
@derekross

Copy link
Copy Markdown
Owner

@Somnius thank you, this is a clean follow-up to #211 and it checks out. I compared the node with Google's zuma-board-gps.dtsi for shusky: the bus, 26 MHz and mode 3, the four lines (gpa5-3, gpp4-3, gph3-2, gph3-3) with their functions, pulls and power-down states, feedback delay 1 and the three S2MPG15 rails all match, and the groups sit under the right pin controllers. nstandby as active high is right for our driver, which sets the line with raw values everywhere, and it matches the 7 Pro's node and the binding's wording. Your note on chip select is correct too: google,zuma-spi uses the port configuration with manual_native_cs, and nothing in this tree reads samsung,spi-chip-select-mode. I also looked for other users of the four pins in Google's husky files; gpp4-3 only shows up in the SoC's generic USI2 UART and SPI groups, and the 8 Pro runs USI2 as I2C for the cameras, so it's free.

On today's omamobile it merges cleanly, checkpatch and the sign-offs are fine, all ten device trees build with W=1 with no new warnings, and in the built husky tree the node is disabled inside a disabled bus. The Pixel 7 Pro's, 6 Pro's and 9a's trees are byte-identical before and after. The "G3 and G4" wording is what I asked for on #211, thank you.

I'm merging this (rebase, both commits). Three optional things for the bring-up overlay PR, none for this one:

  • spi21 has no num-cs, so the SPI driver will log "number of chip select lines not specified, assuming 1" at probe; num-cs = <1> in the overlay (or here) quiets it.
  • The overlay needs usi3_hsi0 enabled next to spi21 and pmic-2 (its mode is already SPI in zuma.dtsi).
  • Stock sets swap-mode and dma-mode on this bus, which our tree doesn't model, and our google,zuma-spi uses zumapro's port configuration. If the first frames come back short or shifted, I'd look there before the pins. And since stock keeps the three rails always on while our driver switches them, load the driver before reading anything from the chip.

The usual rules for the first boot apply: the driver blacklisted before the DTB that enables its node, a log stream and the watchdog first, one change per boot, and a runbook in device/husky before it runs.

@derekross
derekross merged commit fc7086a into derekross:omamobile Nov 11, 2026
1 check passed
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