On Sun, 5 Aug 2012, Tomas Sokorai wrote: > Hi guys! > > I think I might have hit a bug: > > * Description of the issue * > Continuous (variable time 1-10 mins) small bulk transfers (between 1-6 > bytes payload) either with libusb, or generic USB serial driver, leaves the > app in uninterruptible wait state (i.e., hung). > > The kernel hung task stack dump consistently show the USB stack getting stuck: ... > All the stacks are consistent of getting stuck in usb_kill_urb(). > > * Environment * > > This happens only in a OHCI-based system (AMD CPU, nVidia chipset), > since if I test the same program and USB device in an Intel-based, > EHCI laptop, I never hit this problem, and what's more interesting: > the operations are about 5x faster in the EHCI system, even when my > device is only 12Mbit-capable. > > I can reproduce this issue with a vanilla 3.5.0 kernel. ... > Any further ideas to shed more light into this issue? Build a kernel with CONFIG_USB_DEBUG enabled (and also CONFIG_DEBUG_FS). When the hang occurs, go into the subdirectory of /sys/kernel/debug/usb/ohci/ corresponding to the bus the device is plugged into (0000:00:04.0 if your setup hasn't changed), and post copies of the "registers" and "async" files. 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
