On Fri, Jun 19, 2009 at 8:20 PM, David Brownell<david-b@xxxxxxxxxxx> wrote: > On Tuesday 16 June 2009, Laurent Pinchart wrote: >> > We don't need to let gadget device user space application do the >> > conversion. >> >> I'm afraid that won't work. The host and device clocks are not synchronized, >> so you'll experience buffer underruns or overruns. > > Maybe someone is going to finally get into how to do synchronization > with ISO streams in Linux? Read the USB spec, there are several > mechanisms. I believe there are likely to be some missing lowlevel > primitives in the kernel-to-kernel interfaces, needed to make those > things work right. > > Though I think you're talking about a different issue, a bit higher > up in the stack. There might be missing features there too. ;) Are you thinking of Async type of synchronisation by feedback endpoint? Very few USB audio device supports this. And it seems not many people in the whole word understand this (for one I do not). It is said this it the best to counter clock jitter. http://www.audioasylum.com/forums/pcaudio/messages/7719.html According to sc6po (who has done one device), it works better in Linux than Windows XP. He says that his device works great under Linux but can crash XP from time to time. http://www.microchip.com/forums/tm.aspx?m=344751 http://www.microchip.com/forums/tm.aspx?m=270049&mpage=7 (post 136) But Vista seems to catch up. http://www.microchip.com/forums/tm.aspx?m=424170 -- Xiaofan http://mcuee.blogspot.com -- 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
