From: Mehmet Fide <mehmet.fide@screeningeagle.com>

This adds a device tree for the Toradex Iris carrier board carrying the
Vybrid Colibri modules, plus two small preparatory patches for the
module-level dtsi that the i.MX Colibri variants already had.

Everything was tested on a Colibri VF50 on an Iris V1.1 running 6.18.

Changes in v4:
- gpio-line-names packed four per line (Frank Li)
- keep the EXT_IO pin group applied next to the carrier hog group
  instead of overwriting pinctrl-0 in the Iris dtsi, verified in
  pinmux-pins on hardware (Sashiko review)
- enable pwm0/pwm1 like the evaluation board does, so the dts matches
  the commit message; both chips register on hardware (Sashiko review)
- the compatibles keep their hyphens; Krzysztof asked for hyphens in
  every newly introduced compatible in the v2 review

Changes in v3, following Krzysztof's review:
- fold the Iris compatibles into the existing -on-eval enum entries
- use hyphens in the newly introduced compatibles
- drop the fixed 5V regulator that only fed the VBUS regulator
- sent as a fresh thread instead of in reply to v2

Changes in v2:
- add the dt-bindings patch for the new compatibles
- describe the USB host VBUS regulator instead of hogging USBH_PEN

Mehmet Fide (4):
  ARM: dts: vf-colibri: apply the EXT_IO pin group
  ARM: dts: vf-colibri: name the SODIMM gpio lines
  dt-bindings: arm: fsl: add the Colibri VF50 and VF61 on Iris
  ARM: dts: vf: add Iris carrier board support for Colibri VF50 and VF61

 .../devicetree/bindings/arm/fsl.yaml          |  12 +-
 arch/arm/boot/dts/nxp/vf/Makefile             |   2 +
 arch/arm/boot/dts/nxp/vf/vf-colibri-iris.dtsi | 126 ++++++++++++++++++
 arch/arm/boot/dts/nxp/vf/vf-colibri.dtsi      |  56 ++++++++
 .../boot/dts/nxp/vf/vf500-colibri-iris.dts    |  14 ++
 .../boot/dts/nxp/vf/vf610-colibri-iris.dts    |  14 ++
 6 files changed, 220 insertions(+), 4 deletions(-)
 create mode 100644 arch/arm/boot/dts/nxp/vf/vf-colibri-iris.dtsi
 create mode 100644 arch/arm/boot/dts/nxp/vf/vf500-colibri-iris.dts
 create mode 100644 arch/arm/boot/dts/nxp/vf/vf610-colibri-iris.dts


base-commit: 3d6d817622b0a9721e3cc404df3469171582be13
-- 
2.54.0

Hi Mehmet,

On Thu, Aug 13, 2026 at 1:11 PM Mehmet Fide <mehmet.fide@gmail.com> wrote:
>
> From: Mehmet Fide <mehmet.fide@screeningeagle.com>
>
> This adds a device tree for the Toradex Iris carrier board carrying the
> Vybrid Colibri modules, plus two small preparatory patches for the
> module-level dtsi that the i.MX Colibri variants already had.
>
> Everything was tested on a Colibri VF50 on an Iris V1.1 running 6.18.

What about the latest mainline kernel, 7.2-rc7?

From: Mehmet Fide <mehmet.fide@screeningeagle.com>

A correction on the gpiolib warning I mentioned: it is not a gpio-vf610
matter after all, so no driver patch from me.

Half of it is already fixed: d761c7e38a00 ("gpiolib: Check
gc->get_direction() before calling gpiod_get_direction()") in the GPIO
tree covers the debugfs dump path. The remaining backtrace comes
through the i2c generic GPIO bus recovery, which calls
gpiod_get_direction() on any controller without .get_direction(), and
for that I have sent a gpiolib patch:

  https://lore.kernel.org/all/20260813193715.2346477-1-mehmet.fide@gmail.com/

With both in place, 7.2-rc7 plus this series boots the Colibri VF50 on
the Iris with a clean log.

Thanks,
Mehmet

From: Mehmet Fide <mehmet.fide@screeningeagle.com>

Hi Fabio,

> What about the latest mainline kernel, 7.2-rc7?

The series is based on it, the base-commit in the cover letter
(3d6d817622b0) is 7.2-rc7; the 6.18 mentioned there is the LTS our
reference image runs.

To make sure, I have now also boot tested exactly this combination on
the hardware: 7.2-rc7 (multi_v7_defconfig) plus this series on the
Colibri VF50 on an Iris boots to login, with Ethernet up, both PWM
chips registering, the microSD slot, all three UARTs, the carrier RTC
probing, the SODIMM gpio names in place, the EXT_IO pin group applied
and both transceiver hogs holding their pins.

One thing unrelated to this series showed up on the way: gpio-vf610
has no get_direction callback, and the gpiolib in 7.2-rc7 now WARNs on
every gpiod_get_direction() call because of that, so the boot log
carries a number of identical gpiolib.c:431 backtraces on any Vybrid
board. I will send a small gpio-vf610 patch for it separately.

Thanks,
Mehmet