On Sat, 31 Jul 2004 15:08:20 +0200 Robert Wruck <[EMAIL PROTECTED]> wrote:
> I have an APC UPS connected to my 2.4.26 linux box (using a hiddev > device on an AMD756 OHCI controller). > Every night around 1am the kernel detects a BUG and the system freezes. > > This particular bug is triggered in sohci_free_dev when it's called from > an interrupt handler - comments state that it's likely caused by a > refcount bug in the interface driver (hiddev.c / hid.c?). > > sohci_free_dev shouldn't be called at all - the device is still > connected and working. Manually pulling the USB plug also doesn't > trigger the bug. >[...] I spent today looking at this, spurred on by the nice bug report and by the fact that this POS bombs in my code. The result, I'm afraid, is mostly negative: I've became convinced that someone decrements twice (unless we actually have a disconnect, but a) Robert clipped the dmesg OR b) disconnect was actually blocked in a method and invisible) It's kinda too long to list stones I turned over between hid, ohci, and devio. But if my count disbalance hypothesis is true, then pampering over this in usb-ohci is not feasible. The perpetrator has to be found and exterminated. The code is pretty nasty though. It might be easier to give up. Robert, can you try 2.6? -- Pete ------------------------------------------------------- This SF.Net email is sponsored by OSTG. Have you noticed the changes on Linux.com, ITManagersJournal and NewsForge in the past few weeks? Now, one more big change to announce. We are now OSTG- Open Source Technology Group. Come see the changes on the new OSTG site. www.ostg.com _______________________________________________ [EMAIL PROTECTED] To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
