On Tuesday, September 17, 2002, at 03:58 PM, David Brownell wrote:
> Nathanael Noblet wrote: >> Hello, >> I'm having a problem with some hardware and I think it is a bug. > > Maybe. But mostly your issues are warn() messages from the driver, > which won't always indicate bugs. I suspect maybe this driver should > use dbg() not warn(). Should I try simply replacing the warns with dbg()? where is dbg declared? is it the same argument list/types? True, I do have the odd kernel panic and wierd reboot though, but those could easily be symptoms caused by the driver... >> 1) "Rx status -84" - scrolls incessantly down my screen. This is >> being called from the bulk_read_callback function in the rlt8150.c >> file. >> interspersed within those messages is sometimes the message > > "-EILSEQ" ... CRC error or data toggle problem. That often suggests > some kind of hardware problem, like a bad cable. But drivers can goof > up the data toggle and cause -EILSEQ. Well I'd have to say that it is not a hardware problem because this usb ethernet thing works perfectly on the UHCI based machine. The ethernet cables and everything is identical to both machines. How would I try to figure out why this is being sent? where is it sent from? What determines the CRC or toggle problem and how can I figure out why it is incorrect >> 2) "reset needed may be?.." which is always spewed on the screen when >> I disconnect the device from the usb bus. > > Live Yoda long! :) > > That message is returned on -ETIMEDOUT, which is what OHCI returns > when the device stops responding (like maybe because it's gone). > UHCI returns -EILSEQ, which produces your other message. > > Spewing is an undesirable response, but is hard to avoid since if > those errors happen at other times, they're trouble. But khubd > will disconnect() the driver when it notices, so the messages will > stop at some point. Well these messages will come in between the "Rx status" messages. I'm not sure on what frequency or what not but they are there. I have yet to determine whether there is a pattern. > >> 3) System hard reboots - this is wierd, and has happened only 3 >> times, but I'll start something like ping apple.com or using >> links/lynx to go somewhere, screen blanks, and I get the BIOS screen >> and it's starting back up. >> 3) Kernel Panic - This is even more rare, I think mainly with kernel >> 2.5.33 with the preemptive stuff included. > > Are you running with "debug memory allocations" enabled? I tend to run > with that at all times; it'll notice many bugs before they get a chance > to corrupt things very much. Both of these #3 modes are pretty far > gone. I'm not sure. Probably not. Is that a kernel compile option? I haven't seen it. Where do I find it, how do I enable it? > My initial suspicion is that if there's a bug it's not in the ohci > code, > since if it were then a lot of other people would see the same > problems. I agree. I read in the archives though that the rlt8150 driver went through some "conversion" to use the urb something or other, I thought perhaps that when doing so, something in the interaction between the different hubs was incorrect for ohci..? But I am too lost to know left from right in this situation. -- Nathanael Noblet Gnat Solutions 4604 Monterey Ave NW Calgary, AB T3B 5K4 P: (403) 288-4613 C: (403) 809-5368 ------------------------------------------------------- This SF.NET email is sponsored by: AMD - Your access to the experts on Hammer Technology! Open Source & Linux Developers, register now for the AMD Developer Symposium. Code: EX8664 http://www.developwithamd.com/developerlab _______________________________________________ [EMAIL PROTECTED] To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
