I'm CC'ing this reply to the USB development mailing list, in the hope that other people will be able to help.
On Fri, 23 Apr 2004 [EMAIL PROTECTED] wrote: > Hi, > I have a problem with getting USB "harddrive" devices to work properly > since I got my hands on a camera (Konica Revio C2). I have previously been > using (still am) an MP3 player with built in memory accessable as an > harddrive via USB this has always worked fine. > > Anyways, the problem occurs as follows, my MP3 mplayer can be > plugged/unplugged as many times as one wishes, no problem always gets > detected. However once I connect the camera, or rather once I unplug the > camera, neither the camera or the MP3 mplayer can be reconnected > (they are not detected by the kernel anymore at any level from what I can > see when looking at dmesg output and there is no OOPS when > disconnecting, however it seems the kernel goes ballistic and kills most > processes). You're quite sure this happens only with the camera, not with the MP3 player? Does this happen every time you unplug the camera, or only some times? Does it happen even if you don't transfer any data between the computer and the camera? Can you test what happens if you don't load the usb-storage driver (and prevent it from being loaded automatically)? > Since I didnt have a clue what to do or how to debug this someone I asked > suggested that I should modulize my kernel in order to be able to unload > and reload the modules. This way I can _provoke_ an OOPS, however I am not > so sure about what is cause and effect. > > The tests have been conducted using a 2.6.5-mmX kernel, a 2.6.0-mm2 kernel Should that be 2.6.6-mm2? > and a 2.6.6-rc2-bk2 kernel, all have the same problem. What happens if you go back to a 2.6.4 kernel? > The logs are from the 2.6.6-rc2-bk2 kernel. During the tests I had the > nvidia gfx driver loaded, since I figured that may annoy you after I read > the oops tracing document I retried the test after a boot without loading > the module at boot time, it yielded the same results, however I didnt feel > like collecting all data again so I did not. I hope this is acceptable. That's fine. > Steps: > > 1. boot > 2. load uhci_hcd.ko module > 3. plugin camera (it gets detected) > 4. plugout camera (more or less every process gets shutdown, attached is > ps before and after output if that helps). > > << after this point nothing just plain works if it is usb related, the > kernel will not display anything in regard to it if doing dmesg after > pluggin/unplugging >> > > 5. unload all usb related modules (usbcore has to be force unloaded). > 6. load usbcore > 7. load uhci_hci (3x oops) There's no point carrying out steps 5-7. Once things start going wrong (processes dying) you're already screwed. Anything you do later will be so messed up that it probably won't help pinpoint the original source of the problem. You log looked perfectly normal, up to the time when you unloaded all the drivers. So that's not going to be any help. (By the way, you don't need to use ksymoops under Linux 2.6 -- its function has been moved into the kernel.) I saw in your logs that you're using ACPI. What happens if you turn that off? This sounds like some form of memory corruption. I would say that it seems very unlikely except that another report describing the same sort of problem (lots of processes suddenly dumping core) was also posted today. It's possible that this isn't a USB problem at all. Maybe the disconnection is just triggering something bad in a different part of the kernel. Without knowing more we can't really say. Alan Stern ------------------------------------------------------- This SF.net email is sponsored by: The Robotic Monkeys at ThinkGeek For a limited time only, get FREE Ground shipping on all orders of $35 or more. Hurry up and shop folks, this offer expires April 30th! http://www.thinkgeek.com/freeshipping/?cpg=12297 _______________________________________________ [EMAIL PROTECTED] To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
