On Fri, Mar 27, 2009 at 5:02 AM, Kung James <kong1191@xxxxxxxxx> wrote: > Hi, Dan > > I have a question about the "tt_available()" function in > drivers/usb/host/ehci-sched.c ... It's been a while since I was into the EHCI code, but I think that you are right, that looks like a bug there. You can see that my first draft of the patch didn't include the - 1 http://marc.info/?l=linux-usb-devel&m=112319318001393&w=2 but then my second (and final) draft had it added. I don't remember why. http://marc.info/?l=linux-usb-devel&m=113510731024887&w=2 David, does removing the - 1 look right to you also? The -1 appears to be a bug that prevents checking the last fully used uframe for a split isoc transfer. Signed-off-by: Dan Streetman <ddstreet@xxxxxxxx> --- linux-2.6.29/drivers/usb/host/ehci-sched.c 2009-03-23 19:12:14.000000000 -0400 +++ b/drivers/usb/host/ehci-sched.c 2009-03-27 17:11:07.000000000 -0400 @@ -323,7 +323,7 @@ static int tt_available ( * already scheduled transactions */ if (125 < usecs) { - int ufs = (usecs / 125) - 1; + int ufs = (usecs / 125); int i; for (i = uframe; i < (uframe + ufs) && i < 8; i++) if (0 < tt_usecs[i]) { > The following source code is part of "tt_available()" function. > > [kernel 2.6.29] > ============================================ > 320 /* special case for isoc transfers larger than 125us: > 321 * the first and each subsequent fully used uframe > 322 * must be empty, so as to not illegally delay > 323 * already scheduled transactions > 324 */ > 325 if (125 < usecs) { > 326 int ufs = (usecs / 125) - 1; > 327 int i; > 328 for (i = uframe; i < (uframe + ufs) && i < 8; i++) > 329 if (0 < tt_usecs[i]) { > 330 ehci_vdbg(ehci, > 331 "multi-uframe xfer can't fit " > 332 "in frame %d uframe %d\n", > 333 frame, i); > 334 return 0; > 335 } > 336 } > ============================================ > > My question is described as follow: > > I have an full-speed USB audio device connected on a USB HUB. > When I play music via the USB audio device, the SITD of the > isoc. out endpoint will be scheduled via EHCI HCD driver. The > max. packet size of the isoc. out endpoint is 200 bytes. So > the bus transaction time (tt_usecs) is equal to 165 us. > > The 165 us contains a fully used uframe, so the for loop need to be > executed one time theoretically. But the for loop does not to be > executed, because the value of 'ufs' is zero. Why the value of 'ufs' > need to "minus one" ? > > > Best Regard, > James Kung > -- 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
