On 04/04/2015 07:12 AM, Paul Kocialkowski wrote:
This may happen when using an USB1 device on a controller that only supports USB2 (e.g. EHCI). Reading the first descriptor will fail (read 0 byte), so we can abort the process at this point instead of failing later and wasting time.
FYI, this patch breaks USB keyboard (or perhaps any USB 1.x device) support on the RPi.
diff --git a/common/usb.c b/common/usb.c
@@ -956,7 +956,7 @@ int usb_new_device(struct usb_device *dev) */ #ifndef CONFIG_USB_XHCI err = usb_get_descriptor(dev, USB_DT_DEVICE, 0, desc, 64);
- if (err < 0) {
- if (err < sizeof(struct usb_device_descriptor)) { debug("usb_new_device: usb_get_descriptor() failed\n"); return -EIO; }
The value returned here is 8 not 18 (the sizeof the descriptor), so the code bails out with an error.
Given the description of what this patch is attempting to achieve, shouldn't the replacement check be:
if (err <= 0) {
My guess for why the value 8 is returned is because the device's maxpacket is 8, and the DWC2 driver only attempts to transfer 1 packet because the requested transfer size is 64, and the default packet size is 64, which means 1 packet. Note that DWC2 HW (perhaps unlike e.g. EHCI) requires the driver to specify the number of packets transferred, not a byte count. However, I haven't validated that yet. My device is a USB 1.x FS keyboard.