portable/nrf5x: complete zero-length bulk OUT transfers

A zero-length OUT read (e.g. the MTP driver's read for the host's
terminating data ZLP) could never complete when the ZLP's EPDATA
arrived after the read was armed: actual_len < total_len is always
false for total_len == 0, so the handler only set data_received and the
armed transfer hung forever, wedging both bulk endpoints (host sees
pure NAK; EP0 stays alive). Run the 0-byte DMA in that case to drain
the endpoint and complete the transfer, matching what the arm path
already does when the ZLP arrives first.

Found via usbmon on the HIL rig: device/mtp on feather_nrf52840_express
wedged after SendObject's terminating ZLP whenever the class armed the
read before the host's (NAK-delayed) ZLP retry landed - deterministic
under CI load, which is why PR CI only passed this test on a re-flash
retry (21.4 s vs the normal 2.2 s).
This commit is contained in:
hathach
2026-08-14 11:29:37 +07:00
parent 7530f13031
commit f75b247304

View File

@ -805,7 +805,10 @@ void dcd_int_handler(uint8_t rhport) {
if (tu_bit_test(data_status, 16 + epnum) || (epnum == 0 && is_control_out)) { if (tu_bit_test(data_status, 16 + epnum) || (epnum == 0 && is_control_out)) {
xfer_td_t* xfer = get_td(epnum, TUSB_DIR_OUT); xfer_td_t* xfer = get_td(epnum, TUSB_DIR_OUT);
if (xfer->started && xfer->actual_len < xfer->total_len) { // total_len == 0: an armed zero-length read (e.g. MTP's terminating ZLP) still
// needs the 0-byte DMA to drain the endpoint and complete the transfer;
// actual_len < total_len can never be true for it.
if (xfer->started && (xfer->total_len == 0 || xfer->actual_len < xfer->total_len)) {
xact_out_dma(epnum); xact_out_dma(epnum);
} else { } else {
// Data overflow !!! Nah, nRF will auto accept next Bulk/Interrupt OUT packet // Data overflow !!! Nah, nRF will auto accept next Bulk/Interrupt OUT packet