On 4/21/22 06:18, Farhan Khan wrote:
On Wed, 2022-04-20 at 20:37 +0200, Hans Petter Selasky wrote:
Hi,
It appears you've programmed the interrupt endpoint to only receive
one
USB packet, because there is no /* FALLTHROUGH */, but a "break;"
statement instead!
break;
case USB_ST_SETUP:
Maybe if you change that, it will work.
--HPS
Hi!
I removed that break and the issue remains.
I would not expect this to make a difference because the USB_ST_SETUP
condition is below the USB_ST_TRANSFERRED condition and the
USB_ST_TRANSFERRED condition is never called. Since I am only getting
USB_ST_SETUP calls, USB_ST_TRANSFERRED's fallthrough never happens
anyways.
The timeline of events is this:
----
1. athn_data_rx_callback called via usbd_transfer_start() when pipes
are opened, with USB_ST_SETUP condition - no USB_ST_TRANSFERRED occurs
2. athn_usb_intr callback called via usbd_transfer_start(), also when
pipes are opened. This creates a USB_ST_SETUP condition, but no
USB_ST_TRANSFFERRED condition.
3. Loading firmware content via usbd_do_request_flags()
4. Sending final usbd_do_request() with length of 0.
5. msleep() called, times out with EWOUDLBLOCK condition. This should
be woke up by wakeup(9), which would happen if athn_usb_intr with the
USB_ST_TRANSFERRED condition happened, but it does not.
6. Firmware loading failure due to timeout.
7. Driver unloaded
8. athn_usb_intr called with USB_ST_ERROR/USB_ST_CANCELLED (Why?)
Hi,
8) usbdump will answer this question.
I see some more bugs:
[ATHN_RX_INTR] = {
.type = UE_INTERRUPT,
.endpoint = 0x83, // AR_PIPE_RX_INTR,
.direction = UE_DIR_ANY,
^^^^ try to use UE_DIR_RX here, else you might
bind for TX!
And use UE_DIR_TX for TX. Ditto for the other transfers. Any change?
Second note, is that you should set some buffer size for the
non-interrupt endpoints, else the buffer size will only be
wMaxPacketSize bytes big!
--HPS
