"usb explore" thread sometimes crashes shortly after desktop is loaded

[stippi]

PANIC: vm_page_fault: unhandled page fault in kernael space at 0x7, ip 0x9639f7b0

Welcome to KDL...
kdebug> sc
stack trace for thread 0x24c "usb explore"
...
frame                     caller                    <image>:function + offset
...
iframe at 0x96552ef4  (end = 0x96552f48)
 eax 0x3              ebx 0x3             ecx 0x938c3000      edx 0x200246
 esi 0x90cbd500       edi 0x90cbd594      ebp 0x96552f58      esp 0x96552f24
 vector: 0xe, error code 0x0
96552eec (+ 108) 9639f7b0  </boot/.../bus_managers/usb>:RescanDrivers__5StackP11rescan_item + 0x0014
96552f58 (+  80) 963a12a3  <...>:Explore__3Hub + 0x028b
96552fa8 (+  48) 9639f5a9  <...>:ExploreThread__5StackPv + 0x0045
96552fd8 (+  32) 8003ae43  <...>:_create_kernel_thread_kentry + 0x001b
96552ff8 (+1772802056) 8003add8  <kernel>:thread_kthread_exit + 0x0000
kdebug> 

Puh... I typed that all. Hope I didn't sneak in any errors.

[axeld]

Actual screenshots are welcome, too, though :-)

[stippi]

Any news? :-)

[mmlr]

Not really. It's difficult without being able to reproduce. I have looked over the explore code many times now and thought it was finally approaching stability. Do you have anything additional installed to the usb_hid driver when this occurs?

[mmlr]

I have redesigned the explore process in hrev22929 and this might just have an effect on this bug. Can you please retest with a current revision and tell me whether it still happens?

[stippi]

I was planning to test this as soon as I first read the commit, but have not gotten around it yet. Unfortunately, the notebook (my beloved T40p which ran Haiku so well) was stolen last week. This was the machine that had this error reproducably. I am not even sure if I ever saw that KDL on my other machines. But I will check.

[stippi]

Since I cannot reproduce this bug anymore, I am closing it for now, if I do happen to see it again, I can reopen it anyways...