esp32 s2 tiny usb driver lost complete interrupt during concurrent IN & OUT test (IDFGH-6285)

[chuanjinpang]

Environment
verion:esp-idf-4.3-beta3
EPS32 S2 as a usb device
host : centos7
Describe your problem
esp-idf-4.3-beta3\esp-idf-4.3-beta3\components\tinyusb\tinyusb\src\portable\espressif\esp32s2\dcd_esp32s2.c
when concurrent issue USB IN & OUT transcation in stress test. it seem lost complete interrupt that cause USB can't work.
(USB0.out_ep_reg[n].doepint & USB_XFERCOMPL0_M) or(USB0.in_ep_reg[n].diepint & USB_D_XFERCOMPL0_M) can't ocurr in bad case.

I suspect the esp-idf tiny usb have known limitation beacuse the IOT project use a Semphore for switch IN or OUT transfer.
refer:
esp-iot-solution\components\usb\tinyusb\additions\tusb\src\class\net\net_device.c
3. my queston:
is that it's a limitation for usb transfer? or it's a bug ?

[chuanjinpang]

let me update status .
my diy project is a usb 802.11 wireless network card with esp32s2,
the root cause maybe the wifi EMI affect the USB transfer that cause esp32s2 usb transfer failed. and the esp32s2 usb tinyusb seem doesn't have any error recovery flow, then the USB endpoint work abnormal.
the USB recieve the data from host, then issue the data to wifi interface, then wifi start transmit, and the wifi also input data to host, then the EMI affect the USB transfer.
I doesn't have the USB sniffer now, maybe 202201 got one . i will check the bus status .

[chuanjinpang]

20211231 update status:

1. it's no t EMI related issue. I check the USB signal with Oscilloscope. the D+ D- signal seem ok.
2. it's maybe esp32s2 usb chip hardware issue.

[tore-espressif]

Hello @chuanjinpang sorry for such a long delay.
this may be the same case as #7404 and #8241

Could you please try your code on latest master branch of esp-idf?

[Alvin1Zhang]

Thanks for reporting, will close due to short of feedback, feel free to reopen with more updates. Thanks.