I've never dealt with a Code 38, but this blog post might help.
"Why doesn't my driver unload?"
http://blogs.msdn.com/b/usbcoreblog/...er-unload.aspx
Specifically you could try the debugger commands near the end.
On 10/5/2010 2:32 AM, Pavel A. wrote:
> From your description, the eu3eusbser.sys seems to be the culprit.
> Possibly it holds some IRP pending, or something like that.
> --pa
>
> "Carl-Fredrik Neikter" <>
> wrote in message news:16C6BD99-2FAC-4FCE-99E6-...
>> Hello
>>
>> We are facing a problem with a device driver to an USB-device. The USB
>> device is a Cinterion EU3, which is connected to our computer. During
>> certain
>> circumstances, we need to restart the USB device by the AT^SMSO
>> (Switch Off
>> Mobile Station) command. We can also control the shutdown of the USB
>> device
>> by certain control signals (emergency reset).
>>
>> However, after a shutdown and restart, the USB device is not recognized
>> correctly in Windows any longer. Windows reports Code 38 ("Windows cannot
>> load the device driver for this hardware because a previous instance
>> of the
>> device driver is still in memory"). In other words, the removal (plug
>> off) of
>> the device seems to be the culprit. The driver(s) cannot unload itself
>> completely during removal. The driver in question is USB Composite Device
>> (usbccgp.sys).
>>
>> The USB Composite Device is a standard Microsoft driver. We have tried to
>> plug-in, remove and plug-in another composite device (keyboard/mouse)
>> with no
>> problem at all. So it might also be the case that the dependent Cinterion
>> driver eu3eusbser.sys have an inherent problem that spreads to the
>> composite
>> device driver.
>>
>> Please note that before shut off and removal of the device, we try to
>> close
>> all possible handles to the device. So this should not be a problem. Also
>> note that just trying do disable/enable the composite device in Device
>> Manager also gives the same driver loading problem, i.e. Code 38.
>>
>> How should we continue the error tracing to pinpoint the problem?
>>
>> --Carl-Fredrik
>>