Murdock results

✔️ PASSED

4d19f21 driver/usbdev_synopsys_dwc2: use correct number of EPs

Artifacts

• Documentation preview

bors merge

Build succeeded:

• Murdock
• static-tests
• tools-build-success

Thanks for reviewing.

This is really really strange. This very small change leads for the stm32f723e-disco (CID 2.000) to the problem that the enumeration with examples/usbus_minimal and tests/usbus_cdc_ecm only works about every 10th time, while it works for tests/usbus_hid every time. The difference is that tests/usbus_hid does not use auto_init and waits 3 seconds before initializing the USB OTG core. So it seems to be something time critical. By the way, for stm32f746g-disco (CID 3.100) it works perfectly. I really hate this buggy USB OTG core.

If I change it back here

Even more strange, if I leave the code as is, i.e. using _max_endpoints functions here, but just add a simple if (ep == NULL) { printf("NULL"); } after the iteration

diff --git a/drivers/usbdev_synopsys_dwc2/usbdev_synopsys_dwc2.c b/drivers/usbdev_synopsys_dwc2/usbdev_synopsys_dwc2.c
index 5ea29f682c..27804d220c 100644
--- a/drivers/usbdev_synopsys_dwc2/usbdev_synopsys_dwc2.c
+++ b/drivers/usbdev_synopsys_dwc2/usbdev_synopsys_dwc2.c
@@ -539,6 +539,7 @@ static usbdev_ep_t *_usbdev_new_ep(usbdev_t *dev, usb_ep_type_t type,
                 ep->num = idx;
             }
         }
+        if (ep == NULL) { printf ("NULL"); }
     }
 
     if (ep && ep->type == USB_EP_TYPE_NONE) {

it works although the NULL case doesn't happen 🤔

So there seems to be something that is soooo time critical that it works with very small code changes, but doesn't work without those small code changes.