Regarding Win8 PrepareHardware and D0Exit

NTDEV
1

Hi All

I have a legacy driver (Win7) which I am porting to Win8. In this driver, a system thread is created using PsCreateSystemThread in the PrepareHardware callback.

I noticed that the corresponding code to remove this thread is made in the ReleaseHardware callback.

Now this leads to an issue. When I shutdown the system (by using powercfg, disable hibernate, and calling shutdown), or when I simply restart, ReleaseHardware is never called. This means that the System Thread is never killed.

Would it be “legal” to put the code for killing the thread in D0Exit ( if the final state is D3Final) ? This does make the design look “asymmetric” - PrepareHardware is called for creating the thread, but the removal is done in D0Exit conditionally - but would this be considered good or even acceptable design?

What is the general software design guidance for such cases ?

Thanks
Sharma

2

Your current design is fine. When the system is shut down, all devices are powered off and existing threads go away when the machine turns off.

d

-----Original Message-----
From: xxxxx@lists.osr.com [mailto:xxxxx@lists.osr.com] On Behalf Of xxxxx@gmail.com
Sent: Friday, September 28, 2012 11:38 AM
To: Windows System Software Devs Interest List
Subject: [ntdev] Regarding Win8 PrepareHardware and D0Exit

Hi All

I have a legacy driver (Win7) which I am porting to Win8. In this driver, a system thread is created using PsCreateSystemThread in the PrepareHardware callback.

I noticed that the corresponding code to remove this thread is made in the ReleaseHardware callback.

Now this leads to an issue. When I shutdown the system (by using powercfg, disable hibernate, and calling shutdown), or when I simply restart, ReleaseHardware is never called. This means that the System Thread is never killed.

Would it be “legal” to put the code for killing the thread in D0Exit ( if the final state is D3Final) ? This does make the design look “asymmetric” - PrepareHardware is called for creating the thread, but the removal is done in D0Exit conditionally - but would this be considered good or even acceptable design?

What is the general software design guidance for such cases ?

Thanks
Sharma

NTDEV is sponsored by OSR

For our schedule of WDF, WDM, debugging and other seminars visit:
http://www.osr.com/seminars

To unsubscribe, visit the List Server section of OSR Online at http://www.osronline.com/page.cfm?name=ListServer

3

That was the answer I was thinking of giving, but I realized that it
depends a lot on what the thread is doing. If the thread is actively
processing some request, and shutdown terminates that activity before it
has completed (say, the equivalent of lazy write-back), how would this be
handled?

[I used to get a lot of questions of the form, “My application doesn’t free the memory it allocates, so when the process stops, do I leak memory?”]
joe

Your current design is fine. When the system is shut down, all devices are
powered off and existing threads go away when the machine turns off.

d

-----Original Message-----
From: xxxxx@lists.osr.com
[mailto:xxxxx@lists.osr.com] On Behalf Of
xxxxx@gmail.com
Sent: Friday, September 28, 2012 11:38 AM
To: Windows System Software Devs Interest List
Subject: [ntdev] Regarding Win8 PrepareHardware and D0Exit

Hi All

I have a legacy driver (Win7) which I am porting to Win8. In this driver,
a system thread is created using PsCreateSystemThread in the
PrepareHardware callback.

I noticed that the corresponding code to remove this thread is made in the
ReleaseHardware callback.

Now this leads to an issue. When I shutdown the system (by using powercfg,
disable hibernate, and calling shutdown), or when I simply restart,
ReleaseHardware is never called. This means that the System Thread is
never killed.

Would it be “legal” to put the code for killing the thread in D0Exit ( if
the final state is D3Final) ? This does make the design look “asymmetric”

  • PrepareHardware is called for creating the thread, but the removal is
    done in D0Exit conditionally - but would this be considered good or even
    acceptable design?

What is the general software design guidance for such cases ?

Thanks
Sharma

NTDEV is sponsored by OSR

For our schedule of WDF, WDM, debugging and other seminars visit:
http://www.osr.com/seminars

To unsubscribe, visit the List Server section of OSR Online at
http://www.osronline.com/page.cfm?name=ListServer

NTDEV is sponsored by OSR

For our schedule of WDF, WDM, debugging and other seminars visit:
http://www.osr.com/seminars

To unsubscribe, visit the List Server section of OSR Online at
http://www.osronline.com/page.cfm?name=ListServer

4

Since releasehw() is called after there is no hardware access, he is already synchronizing power/device state with the thread (assuming the threads needs to know about these state changes), otherwise the design is broken at a more fundamental level outside of system shutdown

d

-----Original Message-----
From: xxxxx@lists.osr.com [mailto:xxxxx@lists.osr.com] On Behalf Of xxxxx@flounder.com
Sent: Friday, September 28, 2012 12:31 PM
To: Windows System Software Devs Interest List
Subject: RE: [ntdev] Regarding Win8 PrepareHardware and D0Exit

That was the answer I was thinking of giving, but I realized that it depends a lot on what the thread is doing. If the thread is actively processing some request, and shutdown terminates that activity before it has completed (say, the equivalent of lazy write-back), how would this be handled?

[I used to get a lot of questions of the form, “My application doesn’t free the memory it allocates, so when the process stops, do I leak memory?”]
joe

Your current design is fine. When the system is shut down, all devices
are powered off and existing threads go away when the machine turns off.

d

-----Original Message-----
From: xxxxx@lists.osr.com
[mailto:xxxxx@lists.osr.com] On Behalf Of
xxxxx@gmail.com
Sent: Friday, September 28, 2012 11:38 AM
To: Windows System Software Devs Interest List
Subject: [ntdev] Regarding Win8 PrepareHardware and D0Exit

Hi All

I have a legacy driver (Win7) which I am porting to Win8. In this
driver, a system thread is created using PsCreateSystemThread in the
PrepareHardware callback.

I noticed that the corresponding code to remove this thread is made in
the ReleaseHardware callback.

Now this leads to an issue. When I shutdown the system (by using
powercfg, disable hibernate, and calling shutdown), or when I simply
restart, ReleaseHardware is never called. This means that the System
Thread is never killed.

Would it be “legal” to put the code for killing the thread in D0Exit (
if the final state is D3Final) ? This does make the design look “asymmetric”

  • PrepareHardware is called for creating the thread, but the removal
    is done in D0Exit conditionally - but would this be considered good
    or even acceptable design?

What is the general software design guidance for such cases ?

Thanks
Sharma

NTDEV is sponsored by OSR

For our schedule of WDF, WDM, debugging and other seminars visit:
http://www.osr.com/seminars

To unsubscribe, visit the List Server section of OSR Online at
http://www.osronline.com/page.cfm?name=ListServer

NTDEV is sponsored by OSR

For our schedule of WDF, WDM, debugging and other seminars visit:
http://www.osr.com/seminars

To unsubscribe, visit the List Server section of OSR Online at
http://www.osronline.com/page.cfm?name=ListServer

NTDEV is sponsored by OSR

For our schedule of WDF, WDM, debugging and other seminars visit:
http://www.osr.com/seminars

To unsubscribe, visit the List Server section of OSR Online at http://www.osronline.com/page.cfm?name=ListServer

5

That’s good to know. I didn’t know how the pieces fit together, but yes,
I agree; given the conditions under which it is called, as you outline
below, it should be safe to just let the system kill the thread.
joe

Since releasehw() is called after there is no hardware access, he is
already synchronizing power/device state with the thread (assuming the
threads needs to know about these state changes), otherwise the design is
broken at a more fundamental level outside of system shutdown

d

-----Original Message-----
From: xxxxx@lists.osr.com
[mailto:xxxxx@lists.osr.com] On Behalf Of
xxxxx@flounder.com
Sent: Friday, September 28, 2012 12:31 PM
To: Windows System Software Devs Interest List
Subject: RE: [ntdev] Regarding Win8 PrepareHardware and D0Exit

That was the answer I was thinking of giving, but I realized that it
depends a lot on what the thread is doing. If the thread is actively
processing some request, and shutdown terminates that activity before it
has completed (say, the equivalent of lazy write-back), how would this be
handled?

[I used to get a lot of questions of the form, “My application doesn’t free the memory it allocates, so when the process stops, do I leak memory?”]
joe

> Your current design is fine. When the system is shut down, all devices
> are powered off and existing threads go away when the machine turns off.
>
> d
>
> -----Original Message-----
> From: xxxxx@lists.osr.com
> [mailto:xxxxx@lists.osr.com] On Behalf Of
> xxxxx@gmail.com
> Sent: Friday, September 28, 2012 11:38 AM
> To: Windows System Software Devs Interest List
> Subject: [ntdev] Regarding Win8 PrepareHardware and D0Exit
>
> Hi All
>
> I have a legacy driver (Win7) which I am porting to Win8. In this
> driver, a system thread is created using PsCreateSystemThread in the
> PrepareHardware callback.
>
> I noticed that the corresponding code to remove this thread is made in
> the ReleaseHardware callback.
>
> Now this leads to an issue. When I shutdown the system (by using
> powercfg, disable hibernate, and calling shutdown), or when I simply
> restart, ReleaseHardware is never called. This means that the System
> Thread is never killed.
>
> Would it be “legal” to put the code for killing the thread in D0Exit (
> if the final state is D3Final) ? This does make the design look
> “asymmetric”
> - PrepareHardware is called for creating the thread, but the removal
> is done in D0Exit conditionally - but would this be considered good
> or even acceptable design?
>
> What is the general software design guidance for such cases ?
>
> Thanks
> Sharma
>
> —
> NTDEV is sponsored by OSR
>
> For our schedule of WDF, WDM, debugging and other seminars visit:
> http://www.osr.com/seminars
>
> To unsubscribe, visit the List Server section of OSR Online at
> http://www.osronline.com/page.cfm?name=ListServer
>
>
>
> —
> NTDEV is sponsored by OSR
>
> For our schedule of WDF, WDM, debugging and other seminars visit:
> http://www.osr.com/seminars
>
> To unsubscribe, visit the List Server section of OSR Online at
> http://www.osronline.com/page.cfm?name=ListServer
>

NTDEV is sponsored by OSR

For our schedule of WDF, WDM, debugging and other seminars visit:
http://www.osr.com/seminars

To unsubscribe, visit the List Server section of OSR Online at
http://www.osronline.com/page.cfm?name=ListServer

NTDEV is sponsored by OSR

For our schedule of WDF, WDM, debugging and other seminars visit:
http://www.osr.com/seminars

To unsubscribe, visit the List Server section of OSR Online at
http://www.osronline.com/page.cfm?name=ListServer