On Thursday 08 January 2009, Qiuping Chen wrote: > The tests passed include: > - hotplug, with and without hubs. > - g_file_storage: file copying for 12 hours at full speed and high > speed. > - g_file_storage: USBCV chap9 tests. > - g_file_storage: USBCV MSC tests. > - g_ether: scp files for 12 hours at full speed and high speed. > - g_ether: USBCV chap9 tests. That's good basic coverage; it means the main code paths are behaving well. How does it do on http://www.linux-usb.org/usbtest/#gadgets test scripts? That is, run "test.sh" with a Linux host. The usual problem is that while test #9 works OK (hard for it to fail if USBCV ch9 passes), test #10 turns up races in fault handling paths for control transfers. You should be able to run "test.sh control" overnight, too. (USBCV was clearly not intended to include stress tests modes.) It's also worth trying the other network tests outlined on that web page. "scp" by itself generates a pretty regular load ... "ping -f -l 50 -c 1000" (both sides) generates queues and IRQ patterns that TCP won't, for example. > +config USB_GADGET_IUSBC > + boolean "Intel USB Client Controller" Any issue with changing this to "Intel Poulsbo USB ..."? I'm reviewing this now, and that's the first externally-visible change I wanted. There may be a label other than "Poulsbo" that makes Intel happier ... but Intel is a big company, and has more than one USB client controller, so *some* label is needed. - Dave > + depends on PCI > + select USB_GADGET_DUALSPEED > + help > + Intel USB client controller is a PCI peripheral > + which supports both full speed and high speed USB 2.0 > + data transfers. > + It has three IN and three OUT configurable endpoints, > + as well as endpoint zero (for control transfers). -- 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
