Hi, > On Sat, 28 Aug 2010, Artem Makhutov wrote: > >> Hi, >> >>> On Sat, 28 Aug 2010, Artem Makhutov wrote: >>> >>> >>>> It is much harder to reproduce the problem when no hubs are in use. >>>> What is the big deal in not using hubs? >>>> >>>> >>> There are several reasons. First, not having the extra hardware >>> removes a whole class of possible causes of errors. >>> >> The problem is much easier to reproduce by using hubs. >> > Which would seem to indicate a hardware component is involved somehow. > There's no software difference between a device plugged into a hub and > a directly-attached device. > Maybe this are the RF-interferences of the modems. I can try to rework my "RF-shielding-case" for the usb-hubs by soldering all remaining holes. >>> Second, not having >>> the hubs makes the logs easier to follow since they won't contain a lot >>> of uninteresting extra entries. >>> >> Ok. >> >>> Third, without the hubs you won't be able to connect as many modems, >>> which again makes the logs easier to follow. >>> >> Thats not true :) I can connect the same amount of modems without hubs :) >> > Yes, that's what I saw when I looked at your error log! A bunch of > modems were connected to your motherboard's EHCI controller and several > others were running at full speed on your add-on UHCI controllers. The > problem occurred only with one of the modems running at high speed. I > would expect that unplugging those attached to the add-on card wouldn't > make any difference at all. > No, it makes no difference. I was also able to reproduce this problem with only 3 modems directly connected to the motherboard - without add-on cards. Also on different computers. >> Unfortunately the problem can only be triggered by connecting a lot of >> modems. >> > That's another indication of a hardware problem. Furthermore, this > particular error code (xacterr in the log) comes directly from the > controller itself. It indicates that the computer failed to receive a > packet the modem should have sent. > Can it be possible that the USB traffic from one modem is somehow influencing other modems? > Note that in the error log, all the errors occurred for the device > plugged into port 1 -- the modem providing ttyUSB0, ttyUSB1, and > ttyUSB2. Maybe that particular port or that particular modem has a bad > cable connection. (And don't overlook the possibility of a bad cable > inside the computer case, connecting the motherboard to the USB port.) > This kind of problems were reported by a lot of different users. As I said before I have also this kind of problems on different computers. So I don't think that this is a problem with my cables. I have also replaced the modems with other model types. (Huawei K3520 with K3765). The problems are the same. >>> Fourth, without the high-speed hubs the modems will have to >>> connect to a full-speed USB-1.1 controller instead of a high-speed >>> USB-2.0 controller, removing yet another possible source of errors and >>> complexity. >>> >>> >> How do you know that they will connect to USB-1.1 instead of USB-2.0? >> I was believing that the modems are USB-2.0 devices? >> How can I find out if they are USB-1.1 or 2.0? >> > I was wrong. They are indeed all USB-2.0 devices and they all run at > high speed if connected to a high-speed controller. > > As for FreeBSD... How can you be sure that these errors don't occur > under that OS? Maybe they are simply being ignored instead of reported > by the driver and in the log. I am not absolutely sure here. Usually I am loosing the connection to the modem after such a problem. I can not send any commands to the modem any more and I also can't receive any "updates" from the modem after such a "crash". The only thing that I can do is replugging the modem or restarting the PC. This did not happend on FreeBSD. All modems were fully operational the whole time. Tomorrow I will do some more tests. During the last test all modems were idle. Only the "control connection" was opened. I want to prepare a 2nd PC running FreeBSD and connect 8 modems via hubs to it. The first PC will run Linux and all modems will be directly attached to it. Then I want to make voice calls from the Linux box and terminate them on the FreeBSD box and vice versa. This should put real load on the modems and I can see if the FreeBSD box will start getting buggy (while having the worse "modem constilation"). Thanks for your help, Artem -- 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
