Am 30.09.24 um 08:27 schrieb Stefan Wahren:
Hi,
recently i submitted the commit d483f034f032 ("usb: dwc2: Skip clock
gating on Broadcom SoCs") in order to fix an issue with suspend
handling. But this change reveals another issue at least on Raspberry
Pi 3 B (arm64/defconfig) for the following scenario:
1. Power off and Disconnect all external USB from Raspberry Pi 3 B Plus
2. Power on Raspberry Pi
3. Wait until successfully booted
4. Connect USB devices like keyboard
The expected behavior would be that all devices are enumerated, but
this doesn't happend. Instead i observe that the DWC2 stays in
lx_state DWC2_L2 forever (bad case):
[ 2.334366] dwc2 3f980000.usb: supply vusb_d not found, using dummy
regulator
[ 2.341892] dwc2 3f980000.usb: supply vusb_a not found, using dummy
regulator
[ 2.400027] dwc2 3f980000.usb: DWC OTG Controller
[ 2.404868] dwc2 3f980000.usb: new USB bus registered, assigned bus
number 1
[ 2.412087] dwc2 3f980000.usb: irq 51, io mem 0x3f980000
[ 2.711826] usb 1-1: new high-speed USB device number 2 using dwc2
[ 3.195838] usb 1-1.1: new high-speed USB device number 3 using dwc2
[ 3.435829] dwc2 3f980000.usb: dwc2_port_suspend
[ 3.459914] dwc2 3f980000.usb: _dwc2_hcd_suspend
[ 9.009743] dwc2 3f980000.usb: _dwc2_hcd_resume
[ 9.030667] dwc2 3f980000.usb: dwc2_port_suspend
[ 9.044137] dwc2 3f980000.usb: _dwc2_hcd_suspend
[ 9.044222] dwc2 3f980000.usb: _dwc2_hcd_resume
[ 9.354370] usb 1-1.1: new high-speed USB device number 4 using dwc2
[ 9.584095] dwc2 3f980000.usb: dwc2_port_suspend
[ 9.599997] dwc2 3f980000.usb: _dwc2_hcd_suspend
Now i spend several hours into investigating this issue and gained at
least some insights. The reason why the DWC2 stuck in the suspend state
is because the relevant irq 51 doesn't fire anymore in the bad case. I
didn't figure out what's causing this, but i suspect this is related to
the call of dwc2_port_suspend() by dwc2_hcd_hub_control(). According to
the implementation it looks like that dwc2_hcd_hub_control assumes that
if dwc2_port_suspend() returns 0 the bus is suspended, but for the
corner case (hsotg->params.power_down == DWC2_POWER_DOWN_PARAM_NONE &&
hsotg->params.no_clock_gating) the function returns 0 but bus_suspended
stays false. So it seems to me that usb/core and dwc2 becomes async
about their states.
As a hack i made dwc2_port_suspend() to return an error for this corner
case, which prevents this issue but also pm_runtime.
Maybe this is a relevant side node: in the bad case the onboard Ethernet
chip LAN7800 is also not probed after startup (just root and the hubs).
It looks like a race between LAN7800 enumeration and pm_runtime.
Another hint why this seems only happen on the Raspberry Pi is because
there is no PHY or clock control available to Linux.
