Re: [Libusb-devel] Improving kernel -> userspace (usbfs) usb device hand off

[Xiaofan C. <xia...@gm...>]

On Sat, Jun 11, 2011 at 1:04 AM, Tim Roberts <ti...@pr...> wrote:
> Hans de Goede wrote:
>>
>> Not sure what you mean here, but we need for a way for drivers to say
>> no to a software caused disconnection. See my usb mass storage device
>> which is still mounted getting redirected to a vm example. This cannot
>> be reliably done from userspace. Where as it is trivial to do this
>> from kernel space.
>
> FWIW, this is somewhat similar to the Windows shutdown mechanism.  When
> the system wants to tear down a driver stack for any reason, it sends
> down an IRP_MN_QUERY_STOP_DEVICE request.  That asks each driver in the
> stack "are you ready to stop?"  Any driver can veto the request.  If a
> driver objects, then the system sends IRP_MN_CANCEL_STOP_DEVICE, and
> things go back to normal.  If all the drivers agree, then
> IRP_MN_STOP_DEVICE gets sent, and that's when things really shut down.

Are there any timeout mechanism? Is this the reason why
sometimes it takes a long time for Windows to shut down? :-)
And sometimes Windows hangs during shutting down. It is
quite often in the Windows 98SE days but still happens
occasionally under XP and then Vista and Win7.

> The problem with this mechanism is that there are a vast number of
> corner cases and edge conditions.  Most existing drivers handle these
> requests crudely, in a way that works most of the time, but cracks under
> pressure.  In KMDF, the state machine that handles plug-and-play state
> changes has something like 100 states.  (The power management state
> machine is much, much worse, but that's a tale for another day.)
>


-- 
Xiaofan