On Thu, 23 Sep 2010, Felipe Balbi wrote: > On Thu, Sep 23, 2010 at 03:36:40AM -0500, Ming Lei wrote: > >2010/9/23 Felipe Balbi <balbi@xxxxxx>: > >> Hi, > >> > >> On Wed, Sep 22, 2010 at 10:15:09AM -0500, tom.leiming@xxxxxxxxx wrote: > >>> > >>> + if ((request->zero && request->length > >> > >> I think you can actually remove the request->length check, I doubt any > >> gadget driver will queue a request without length sent to something more > >> than 0. could you test that please ? Gadget drivers queue requests with length set to 0 all the time. How else could they send a ZLP? Maybe they don't also set the request->zero flag, but I wouldn't depend on it. Checking is safer. > >I think so too. But I am not sure, since I don't know why the > >request->length check is introduced, so I don't remove the check in > >the patch and just fix the ZLP issue only. The check is there in order to prevent the controller driver from sending _two_ ZLPs when the gadget driver queues a request with request.length == 0 and request.zero set. > >Also seems req->zero is used in some corner case only, I don't think > >there are good test cases to verify the change of removing the check, > >right? > > > >So suggest to keep the check until we are sure it is safe to remove it. > > makes sense :-) I'll try to play around with removing request->length > check after merge window then :-) Even if it never gets used, better safe than sorry. It's not like this test takes up multiple megabytes of code space. 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
