Allwinner A40i USB-OTG host GPIO pin ID interrupt doesn't run handler

Сергей Мосолов

unread,

Dec 21, 2023, 2:25:12 AM12/21/23

to linux...@googlegroups.com

Hello,

I kindly ask for an advice about interrupt working with USB-OTG usb0 on Allwinner A40i.
I need to make work usb0 as USB-OTG.

Looks like it's possible somehow:
https://lore.kernel.org/linux-arm-kernel/daf5c543-a1d1-04d2...@sholland.org/T/#md5e9ce01feb201041467c136f09e80628f414557

I've used usb-phy driver with support of muxing ohci0(host) and usb0(client) on usb0 pins, mentioned in the link above:
https://github.com/wirenboard/linux/blob/dev/v5.10.y/drivers/phy/allwinner/phy-sun4i-usb.c

In my case I have pin ID USB OTG = PI11.

Some excerpts from my dts file:
[quote]
/ {
...
/* ohci0 and ehci0 need to be included to work with USB OTG */
ohci0: usb@1c14400 {
compatible = "allwinner,sun8i-r40-ohci", "generic-ohci";
reg = <0x01c14400 0x100>;
interrupts = <GIC_SPI 40 IRQ_TYPE_LEVEL_HIGH>;
clocks = <&ccu CLK_BUS_EHCI0>, <&ccu CLK_BUS_OHCI0>,
<&ccu CLK_USB_OHCI0>;
resets = <&ccu RST_BUS_EHCI0>, <&ccu RST_BUS_OHCI0>;
status = "okay";
};

ehci0: usb@1c14000 {
compatible = "allwinner,sun8i-r40-ehci", "generic-ehci";
reg = <0x01c14000 0x100>;
interrupts = <GIC_SPI 39 IRQ_TYPE_LEVEL_HIGH>;
clocks = <&ccu CLK_BUS_EHCI0>, <&ccu CLK_BUS_OHCI0>;
resets = <&ccu RST_BUS_EHCI0>, <&ccu RST_BUS_OHCI0>;
status = "okay";
};
...
}
...
&usb_otg {
/* в dtsi defines of usb_otg as dr_mode = "peripheral"; */
status = "okay";
};

&usbphy {
dr_mode = "otg";
usb0_id_det-gpios = <PIN_PI 11 GPIO_ACTIVE_HIGH>;
usb0_vbus-supply = <&reg_vcc5v0>;
status = "okay";
};
[/quote]

USB-OTG works, but the problem is USB-OTG reads pin ID and choose client or host only on OS startup once.

After adding many printk and -DDEBUG for several subsystems it seems to me that the problem is PI11 GPIO irq just doesn't run it's handler requested here:
https://github.com/wirenboard/linux/blob/dev/v5.10.y/drivers/phy/allwinner/phy-sun4i-usb.c#L951

I observed EINT23 registration in pinctrl sunxi driver:
[quote]
[ 11.578919] sun4i-pinctrl 1c20800.pinctrl: 1c20800.pinctrl: request IRQ for GPIO 267, return 23
[ 11.587657] devm_request_irq data->id_det_irq: 74, ret: 0

74: 0 0 0 0 sunxi_pio_edge 23 Edge usb0-id-det

# cat /proc/interrupts
CPU0 CPU1 CPU2 CPU3
26: 0 0 0 0 GICv2 54 Level timer@1c20c00
27: 0 0 0 0 GICv2 29 Level arch_timer
28: 198416 45783 21540 10341 GICv2 30 Level arch_timer
31: 0 0 0 0 GICv2 152 Level arm-pmu
32: 0 0 0 0 GICv2 153 Level arm-pmu
33: 0 0 0 0 GICv2 154 Level arm-pmu
34: 0 0 0 0 GICv2 155 Level arm-pmu
35: 0 0 0 0 GICv2 59 Level 1c02000.dma-controller
37: 0 0 0 0 GICv2 102 Level gpmmu
38: 0 0 0 0 GICv2 104 Level ppmmu0
39: 0 0 0 0 GICv2 107 Level ppmmu1
40: 175 0 0 0 GICv2 101 Level gp
41: 128 0 0 0 GICv2 103 Level pp0
42: 128 0 0 0 GICv2 106 Level pp1
43: 0 0 0 0 GICv2 71 Level ehci_hcd:usb1
44: 0 0 0 0 GICv2 72 Level ohci_hcd:usb2
45: 1289 0 0 0 GICv2 61 Level sun4i-ts
46: 0 0 0 0 GICv2 56 Level 1c20400.rtc
47: 0 0 0 0 GICv2 125 Level 1400000.deinterlace
48: 0 0 0 0 GICv2 126 Level sun8i-ce-ns
49: 0 0 0 0 GICv2 85 Level 1c0e000.video-codec
74: 0 0 0 0 sunxi_pio_edge 23 Edge usb0-id-det
83: 28 0 0 0 GICv2 33 Level ttyS0
84: 0 0 0 0 GICv2 44 Level sun6i-spi
85: 80306 0 0 0 GICv2 39 Level mv64xxx_i2c
86: 0 0 0 0 sunxi-nmi 0 Level axp22x_irq_chip
109: 0 0 0 0 axp22x_irq_chip 22 Edge axp20x-pek-dbr
110: 0 0 0 0 axp22x_irq_chip 23 Edge axp20x-pek-dbf
113: 0 0 0 0 GICv2 40 Level mv64xxx_i2c
114: 0 0 0 0 GICv2 41 Level mv64xxx_i2c
115: 0 0 0 0 GICv2 120 Level mv64xxx_i2c
116: 0 0 0 0 GICv2 121 Level mv64xxx_i2c
117: 0 0 0 0 GICv2 129 Level tvd0
118: 0 0 0 0 GICv2 130 Level tvd1
119: 0 0 0 0 GICv2 131 Level tvd2
120: 0 0 0 0 GICv2 132 Level tvd3
121: 10281 0 0 0 GICv2 68 Level ths
122: 245 0 0 0 GICv2 66 Level sunxi-mmc
123: 5690 0 0 0 GICv2 64 Level sunxi-mmc
129: 0 0 0 0 GICv2 88 Level ahci-sunxi[1c18000.sata]
130: 168336 0 0 0 GICv2 87 Level eth0
131: 81204 0 0 0 GICv2 117 Level eth1
132: 401 0 0 0 GICv2 70 Level musb-hdrc.2.auto
133: 0 0 0 0 GICv2 96 Level ohci_hcd:usb3
134: 0 0 0 0 GICv2 97 Level ohci_hcd:usb4
135: 9114 0 0 0 GICv2 76 Level 1c71000.lcd-controller
136: 13939 0 0 0 GICv2 83 Level 1c73000.lcd-controller
137: 515 0 0 0 GICv2 90 Level 1ee0000.hdmi
138: 0 0 0 0 GICv2 110 Level ehci_hcd:usb5
139: 0 0 0 0 GICv2 108 Level ehci_hcd:usb6
IPI0: 0 0 0 0 CPU wakeup interrupts
IPI1: 0 0 0 0 Timer broadcast interrupts
IPI2: 71 81 134 106 Rescheduling interrupts
IPI3: 5850 17893 6323 6836 Function call interrupts
IPI4: 0 0 0 0 CPU stop interrupts
IPI5: 16353 2205 1054 417 IRQ work interrupts
IPI6: 0 0 0 0 completion interrupts
Err: 0
[/quote]

I've reconfigured PI11 as GPIO and I've observed change of input state on USB-OTG cable.
External interrupt is possible on PI11 - PI11/SPI0_CLK/UART5_RX/EINT23 - everything else is turned of like spi0 or routed to different pins as uart5.

Link to syslog excerpts and dts with dtsi:
https://disk.yandex.ru/d/0H5tnCl1qQIE2g

I really appreciate an advice with links to how debug that.

Please, notify my if I can provide some additional info.

Andre Przywara

unread,

Dec 21, 2023, 12:14:29 PM12/21/23

to Сергей Мосолов, linux...@googlegroups.com, linux...@lists.linux.dev

On Thu, 21 Dec 2023 11:24:57 +0400
Сергей Мосолов <mosolo...@gmail.com> wrote:

(adding the proper sunxi mailing list for a wider audience)

Hi Sergej,

> I kindly ask for an advice about interrupt working with USB-OTG usb0 on
> Allwinner A40i.
> I need to make work usb0 as USB-OTG.
>
> Looks like it's possible somehow:
> https://lore.kernel.org/linux-arm-kernel/daf5c543-a1d1-04d2...@sholland.org/T/#md5e9ce01feb201041467c136f09e80628f414557
>
> I've used usb-phy driver with support of muxing ohci0(host) and
> usb0(client) on usb0 pins, mentioned in the link above:
> https://github.com/wirenboard/linux/blob/dev/v5.10.y/drivers/phy/allwinner/phy-sun4i-usb.c

some good checking and detective work!

I am afraid the problem is somewhere deeper, though, in the way our sunxi
PHY interacts with the OTG framework. I actually believe it's broken
for every Allwinner system at the moment. Some months ago I spent some
time in the code, and it looks like there is some race between the
OHCI/EHCI controllers and the MUSB controller for the PHY. Both the MUSB
and the HCI drivers claim the PHY and set their preferred mode (OTG and
host, respectively), but if I remember correctly, the HCI driver always
wins, regardless of the order. I can't remember exactly, but I think the
problem was in sun4i_usb_phy_set_mode. You can try to insert debug
messages there to see what's going on.
The short summary at the moment seems to be: forcing the mode (peripheral
or host) works, but the dynamic switching does not.

You might want to try disabling the ehci0/ohci0 nodes, as a quick check, to
see if that improves thing. I haven't tried that, though, as doing so is
not correct and breaks U-Boot.
Also keep an eye on the interrupt counter in /proc/interrupts, you should
see some non-zero number if you connect and disconnect the cable a few
times. But if the PHY is in host mode, then it won't activate the IRQ, I
believe.

Пока,
Andre

Сергей Мосолов

unread,

Jan 12, 2024, 2:04:02 AMJan 12

to linux-sunxi

Hi Andre!

Thanks for sharing that info.

Actually the problem was that the GPIO interrupts not working, there just no irq 60 (PIO) generated for some reason at all, at any EXTI PIN.

After conforming to recommendation in A40i datasheet:

"(24).A 10MΩ resister is connected in parallel between X32KOUT and X32KIN, the resistor can create negative feedback in an inverter to ensure amplifier in linear amplifier region."

GPIO interrupts start working and, in particular, USB OTG PIN correctly detects Host or Client connected. I've used mentioned above:

https://github.com/wirenboard/linux/blob/dev/v5.10.y/drivers/phy/allwinner/phy-sun4i-usb.c