Hi Junjie, marcandre.lureau > Subject: Re: [PATCH v6 7/8] hw/usb/redirect-server: Stream interrupt IN > endpoints > > From: Junjie Cao <[email protected]> > > Hi Jamin, > > On v6, the local test device delivers data after an IOERROR without another > start request. STALL stops the stream. > > With a bulk IN endpoint at 0x83, v6 accepts the start request and sends the > data as an interrupt_packet after the device wakes. > > Could we also reject non-interrupt endpoints in start_interrupt_receiving()? > > Thanks, > Junjie
Thanks for the review and for testing. I am reworking patches 5, 6 and 7 to fix the error cases you both raised. Junjie, on your first point: the stream staying up after an IOERROR is on purpose. redirect.c clears interrupt_started only on a stall, so ending the stream on any other error would leave the host thinking it still runs. The non-interrupt endpoint you found is a real bug, and v7 will refuse it. Thanks, Jamin
