Re: Unload problems with filter driver
- From: "Don Burn" <burn@xxxxxxxxxxxxxxxx>
- Date: Fri, 8 Jul 2005 06:50:20 -0400
Unless it is using the new mini-filter model, trying to unload a filter
driver is just asking for a crash. Your assumptions about when it is safe
are horribly wrong, there are multiple references to you device object still
around at that point.
--
Don Burn (MVP, Windows DDK)
Windows 2k/XP/2k3 Filesystem and Driver Consulting
Remove StopSpam from the email to reply
"helios" <helios@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote in message
news:DEB38E8A-FCE0-4B57-9FE8-B2E1B2DA50A3@xxxxxxxxxxxxxxxx
> is it possible to implement a safe way for filter driver unload?
>
> I have a filter driver on top of a FSD. If there are no pending IRPs and
> there is no other driver attached to mine, it should be safe to unload the
> driver.
>
> But how to tell the system, when its safe? Setting UnloadDriver will cause
> the object manager to call it even if the driver is still atached to a
> fsd.
> It seems to unload everything even devices stored in drivers structure
> field
> DeviceObject.
>
> The result is: there is still a reference to a non existing object in the
> AttachedDevice field of the FSD device object.
>
> Even worse: if there are pending IRPs the system bugchecks.
>
> Is it possible to deactivate unloading by temporarily remove the reference
> to UnloadDriver and set it explicitely by ioctl when everything is fine/
> detached etc?
>
> (tried to post this question yesterday but it did not work)
>
.
- References:
- Unload problems with filter driver
- From: helios
- Unload problems with filter driver
- Prev by Date: Questions about MSDN for some DDK functions
- Next by Date: Re: Help me in writing Pcmcia drivers
- Previous by thread: Unload problems with filter driver
- Next by thread: Re: Unload problems with filter driver
- Index(es):
Relevant Pages
|
