On Fri 090515, Ethan wrote: > On Fri, May 15, 2009 at 8:46 AM, Alan Stern <stern@xxxxxxxxxxxxxxxxxxx> wrote: [.......] > > This patch may fix the problem. ?I haven't tested it, so you guys will > > have to be the guinea pigs. ?:-) > > Wow! This discussion got very technical very fast. Let me see if > I---a relatively USB-naive user---have this right. My DUB-7 hub may > have an ISP1521 chip that shares a single TT among all 7 ports. This > fact may trigger a bug in the Linux USB driver when low or full speed > devices are plugged into the hub. However, since all of my devices > are high-speed USB 2.0 devices (sorry I forgot to mention that in my > original post), this bug and your patch are unfortunately irrelevant > to my problem. I also think that you are seeing a diffent issue. Just to make it more clear: in my situation I have one full speed device (Prolific USB<>RS232) permanently connected to one port of the ISP1521 and memory sticks are not being recognized on arrival or the removal is not seen by the OS (XP). As soon as I remove the Prolific these problems cannot be reproduced any more. The probability goes from roughly 1 problem per 3 plug in to 1 per 50 or even more. In rare cases it fails also on the first attempt. (problem: device not being recognized) The "removal not seen" effect happens considerably less frequently. If I remember correctly all other ports of the ISP1521 did not work any more. XP still showed the device as fully functional but when I connected the very same usb stick to another still working port I got a blue screens three times (its even a separate stop code, something like: device with certain serialnumber connected twice or similar) It is especially interesting to me that the probability is much higher while our application is running, it does not do very much with the Prolific, only on startup and when the user presses some button. However the application has registered a callback for device arrival and removal and tries to collect info (drive letters etc.) on each such event and this additional activity seems to raise the error probability. Perhaps these activities only modify some timing but who knows ... I have not yet managed to trigger this bug under Linux, so if there are ways to increase the probability I would like to know for testing. (e.g.: data transfers to the USB<>RS232) Knowing mitigating factors would also be desired of course. BTW, the main PC board is Intel ICH8 based which fortunately has two ehci root hubs. There is no lack on "interesting" USB effects here: I have also one particular 4 port high speed hub (05e3:0605 Genesys Logic, Inc. USB 2.0 Hub) which - given a certain cable length - will cause all kinds of havoc when connected to a free root hub port: other devices on the same root hub disappear or are being re-enumerated with full speed instead of high speed. After inserting a USB extension cable of 1 meter length all seems normal again. Alan, special thanks for your helpful comments. The other readers please forgive me talking about a problem we have in XP but it seems to affect Linux as well which is my favorite OS and this thread may explain why some feelings about general flakey-ness of USB have built up over time here. (not even trying to discuss cable and connector quality) Greetings, Karl > > Alan: I'll try the troubleshooting steps you recommended---testing > the hub with a different cable, computer, and OS---and let you know > how I fare. > > Thanks to everyone for your help so far. > > --Ethan > -- > 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 -- D.I. Karl Kiniger mailto:karl.kiniger@xxxxxxxxxx Tel: (++43) 7682-3800-710 Voip(GE): 1662 710 Fax: (++43) 7682-3800-47 C: +43 6991 3800 710 - GE Medical Systems Kretztechnik GmbH & Co OHG - Tiefenbach 15 – A-4871 Zipf Landesgericht Wels FN 222387s – UID (VAT-ID): ATU54086108 - DVR: 2111103 – ARA: 12816 Persönlich haftende Gesellschafterin: GE Medical Systems Kretztechnik GmbH Tiefenbach 15 – A-4871 Zipf - Landesgericht Wels FN 218689b -- 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
