| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <aWJAT3n_KcND8bOz@monica.localdomain>
Date: Sat, 10 Jan 2026 20:04:31 +0800
From: Ze Huang <huang.ze@...ux.dev>
To: Samuel Holland <samuel.holland@...ive.com>,
Ze Huang <huang.ze@...ux.dev>, Yixun Lan <dlan@...too.org>
Cc: Rob Herring <robh@...nel.org>, Krzysztof Kozlowski <krzk+dt@...nel.org>,
Conor Dooley <conor+dt@...nel.org>, Paul Walmsley <pjw@...nel.org>,
Palmer Dabbelt <palmer@...belt.com>,
Albert Ou <aou@...s.berkeley.edu>, Alexandre Ghiti <alex@...ti.fr>,
devicetree@...r.kernel.org, linux-riscv@...ts.infradead.org,
spacemit@...ts.linux.dev, linux-kernel@...r.kernel.org,
Aurelien Jarno <aurelien@...el32.net>
Subject: Re: [PATCH v2 3/3] riscv: dts: spacemit: Enable USB3.0 on BananaPi-F3
On Thu, Jan 08, 2026 at 11:21:48PM -0600, Samuel Holland wrote:
> Hi Ze,
>
> On 2026-01-08 10:07 PM, Ze Huang wrote:
> > On Thu, Jan 08, 2026 at 07:36:05PM +0800, Yixun Lan wrote:
> >> Hi Ze,
> >>
> >> I've done a quick check of the "regulator-always-on" issue,
> >> it's ok to drop the one from hub, but not the vbus one..
> >>
> >
> > Thanks for your double check
> >
> >> On 09:21 Thu 08 Jan , Yixun Lan wrote:
> >>> Hi Ze,
> >>>
> >>> thanks for your patch, I have few comments, see below..
> >>>
> >>> On 20:05 Wed 07 Jan , Ze Huang wrote:
> >>>> Enable the DWC3 USB 3.0 controller and its associated usbphy2 on the
> >>>> Banana Pi F3 board.
> >>>>
> >>>> The board utilizes a VLI VL817 hub, which requires two separate power
> >>>> supplies: one VBUS and one for hub itself. Add two GPIO-controlled
> >>>> fixed-regulators to manage this.
> >>>>
> >>>> Tested-by: Aurelien Jarno <aurelien@...el32.net>
> >>>> Signed-off-by: Ze Huang <huang.ze@...ux.dev>
> >>>> ---
> >>>> arch/riscv/boot/dts/spacemit/k1-bananapi-f3.dts | 48 +++++++++++++++++++++++++
> >>>> 1 file changed, 48 insertions(+)
> >>>>
> >>>> diff --git a/arch/riscv/boot/dts/spacemit/k1-bananapi-f3.dts b/arch/riscv/boot/dts/spacemit/k1-bananapi-f3.dts
> >>>> index 3f10efd925dc..013df91c6a4c 100644
> >>>> --- a/arch/riscv/boot/dts/spacemit/k1-bananapi-f3.dts
> >>>> +++ b/arch/riscv/boot/dts/spacemit/k1-bananapi-f3.dts
> >>>> @@ -59,6 +59,26 @@ reg_vcc_4v: vcc-4v {
> >>>> regulator-always-on;
> >>>> vin-supply = <®_dc_in>;
> >>>> };
> >>>> +
> >>>> + usb3_vbus: regulator-vbus-5v {
> >>> I've checked the schematics, the name is 5V_VBUS there, so for the consistency
> >>> with previous naming convention, let's change to:
> >>> usb3_vbus_5v: usb3-vbus-5v
> >>>
> >>>> + compatible = "regulator-fixed";
> >>>> + regulator-name = "USB30_VBUS";
> >>>> + regulator-min-microvolt = <5000000>;
> >>>> + regulator-max-microvolt = <5000000>;
> >>> ..
> >>>> + regulator-always-on;
> >>> do you think the "always-on" property really necessary? it indicate the
> >>> power regulator is critical, and should never been disabled even during
> >>> suspend/resume state, for the case of USB, I think it should be totally
> >>> fine to poweroff once the device is not used(suspended) or even disabled
> >>>
> >>> besides, the regulator is designed with a gpio enabling/disabling control
> >>> which means it can be powered to on/off state?
> >>>
> >
> >
> >> not ok to drop, see comment below at &usb_dwc3
> >
> >>
> >>>> + gpio = <&gpio K1_GPIO(97) GPIO_ACTIVE_HIGH>;
> >>>> + enable-active-high;
> >>>> + };
> >>>> +
> >>>> + usb3_vhub: regulator-vhub-5v {
> >>> why use vhub, but not hub? where does this name come from?
> >>>
> >>> and for same reason, the name in schematics is VCC5V0_HUB, so how about
> >>> change it to:
> >>> usb3_hub_5v: usb3-hub-5v
> >>>
> >>>> + compatible = "regulator-fixed";
> >>>> + regulator-name = "USB30_VHUB";
> >>> ~~~need to fix too, if above is valid
> >>>> + regulator-min-microvolt = <5000000>;
> >>>> + regulator-max-microvolt = <5000000>;
> >>> ..
> >>>> + regulator-always-on;
> >>> ditto
> >
> >> ok to drop, as it's already handled at drivers/usb/misc/onboard_usb_dev.c
> >> which will explicitly request the regulator and enable it
> >
> >>
> >>>> + gpio = <&gpio K1_GPIO(123) GPIO_ACTIVE_HIGH>;
> >>>> + enable-active-high;
> >>>> + };
> >>>> };
> >>>>
> >>>> &combo_phy {
> >>>> @@ -67,6 +87,34 @@ &combo_phy {
> >>>> status = "okay";
> >>>> };
> >>>>
> >>> ..
> >>>> +&usbphy2 {
> >>>> + status = "okay";
> >>>> +};
> >>>> +
> >>>> +&usb_dwc3 {
> >>>> + dr_mode = "host";
> >>>> + vbus-supply = <&usb3_vbus>;
> >> due to drivers/usb/dwc3/dwc3-generic-plat.c has no handler to request regulator
> >> there will be problem if "regulator-always-on" property is removed..
> >
> > On the BananaPi-F3 and to Jupiter, the VL817 hub setup uses two distinct controls:
> >
> > GPIO97 controls the physical VBUS supply.
> > GPIO123 controls the VCC5V0_HUB supply for the hub logic.
> >
> > (Note: This differs from the Orange Pi RV2, where only GPIO123 is used to
> > control the VBUS supply).
> >
> > dwc3-generic-plat can not currently handle the regulator request explicitly,
> > keeping "regulator-always-on" forces the VBUS to remain active even when
> > suspended, which consumes unnecessary power.
> >
> > I prefer to drop regulator-always-on in the DTS to accurately describe the
> > hardware capability, and let the driver manage the VBUS state. This ensures
> > better power management.
> >
> > If anyone have other ideas, please let me know.
>
> If a USB device requires board-level resources to function, then the USB device
> itself (not the controller) needs a DT node that references these resources.
> This is the purpose of the usb-device.yaml binding. For the VL817 hub, there is
> a specific binding that links the highspeed and superspeed hubs together. See
> Documentation/devicetree/bindings/usb/vialab,vl817.yaml. These nodes are matched
> with the Linux usb-onboard-hub driver.
>
Hi Samuel,
Thank you for the guidance. I agree that VBUS control is a board-level
resource and handling it at host controller level is not appropriate.
I noticed Marco Felsch's patch series "[PATCH v4 0/5] Add onboard-dev
USB hub host managed vbus handling support", which adds support for
optional vbus-supply at the USB port level. This matches our hardware
situation perfectly.
To avoid dependencies on unmerged code, I plan to stick with the
"regulator-always-on" approach in the next version (v3). Once Marco's
series is merged into the kernel, I will follow up with a patch to switch
to the port-based VBUS supply method.
Link: https://lore.kernel.org/all/20250911-v6-16-topic-usb-onboard-dev-v4-0-1af288125d74@pengutronix.de/
Best regards, Ze
> Regards,
> Samuel
>
Powered by blists - more mailing lists