Hi, >-----Original Message----- >From: ext Drews, Paul [mailto:[email protected]] >Sent: Tuesday, January 18, 2011 1:17 AM >To: Kodihalli Deepak (Nokia-MS/Bangalore); Ohly, Patrick; >Kavuri Sateesh (Nokia-MS/Bangalore); [email protected]; >Puranik Santosh (Nokia-MS/Bangalore); >[email protected]; [email protected]; >[email protected]; [email protected]; Duggirala >Karthik.2 (Nokia-MS/Helsinki) >Cc: [email protected] >Subject: RE: Buteo MTP + gadgetfs
> >Here's what I'm worried about: > >Case A: MTP responder sends data over bulk pipe where the >length is known ahead of time. Data takes multiple packets >and is an exact multiple of packet length. MTP responder asks >for "fsync" on the last sendData() call. Since the length is >known ahead of time, a ZLP should _not_ be sent according to >the USB spec. Even though the initiator knows the length in advance, it expects a ZLP if the packet size is a multiple of the USB max packet length. We have seen this with both windows and linux (libmtp) initiators. >Case B: MTP responder sends data over bulk pipe but the >length is _not_ known ahead of time. Data takes multiple >packets and is an exact multiple of packet length. MTP >responder asks for "fsync" on the last sendData() call. >Since the length is _not_ known ahead of time, a ZLP _should_ >be sent according to the USB spec. AFAIK, this is not possible with MTP1 at least, i.e the length _is_ always known. Regards, Deepak _______________________________________________ MeeGo-dev mailing list [email protected] http://lists.meego.com/listinfo/meego-dev
