majbthrd

@hathach, if you add an “#error” to ./src/portable/raspberrypi/rp2040/dcd_rp2040.c at about line 328 inside the #if TUD_OPT_RP2040_USB_DEVICE_ENUMERATION_FIX block, and then attempt a compile, it doesn’t report the error. The “rp2040_usb_device_enumeration_fix” appears...

No, not exactly 128 bytes. The intent was to transfer request->wLength, but no more than 128. tu_control_xfer() calls tu_min() to cap all transfers to request->wLength.

I've submitted a PR that should address the STM32 RNDIS issue, and I've regression tested it with SAMD21 and NUC126. It may be worth re-trying on the other targets mentioned...

Just adding to what @pigrew already said... The problem transfer happens with EP0. A host-to-device class request is handled, and this results in a tud_control_xfer() (./src/class/net/net_device.c:334). A valid transaction should...

OK, @duempel, whilst waiting for someone to volunteer to decipher and fix the dcd_synopsys driver, changing the notify EP from EP1 to EP3 causes the driver to work as expected,...

FYI: another standalone STM32F4(07)DISCO workaround that enables net_lwip_webserver is to use the dcd_int_handler as a poll routine instead of an interrupt handler: In ./src/portable/st/synopsys/dcd_synopsis.c, comment out the interrupt enable: ````...

That this modification causes things to work with the STM32F4(07)DISCO is bizarre. By adding this to transmit_packet() in ./src/portable/st/synopsys/dcd_synopsis.c: ```` if (1==fifo_num) { (* tx_fifo) = 1; (* tx_fifo) =...

@hathach, well done! I've done a quick check with net_lwip_webserver operation using the STM32F4(07)DISCO with Linux, Windows, and macOS, and everything seems functional.

@hathach, I just noticed that NUC505 is listed at the top of this thread as having not been tested yet. I've been using it locally for a month or two,...

With #531, the iMX RT1010 now works, and I suspect the same minor tweak to the stack size will remedy others.