On Thu, 29 Mar 2012, Paul Zimmerman wrote: > Hi Pratyush, > > I tried my "isoc support for gadget zero and usbtest" patches with > the dwc3 driver with the above patch, and the isoc IN tests from > usbtest still didn't work. I discovered that I needed to queue up > 8 requests in source_sink_start_ep(), or else the transfer would > stop after the first 4 packets were sent. ... > This makes sense, because the dwc3 driver is only setting the IOC > bit in every 8th TRB, so if less than 8 TRBs are queued up > initially, an interrupt event will never be generated, so the > gadget driver will never be notified so it can queue up the next > requests. This is probably obvious to you guys, but I'll say it anyway: The behavior you describe is a bug in the DWC3 driver. Transfer completions have to be reported promptly unless the client specifically says otherwise by setting the "no_interrupt" flag in the usb_request structure. Alan Stern -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
