PREfast (5744) fails some WDF headers, and thus my code too

[chris.a...@gmail.com]

I have some code in my driver that looks like this:

1) WdfSpinLockAcquire(......)
2) status = WdfCollectionAdd(....)
3) WdfSpinLockRelease(......)

To this, PREfast says, "so and so function modified the IRQL but did
not release it before exiting."

What?

Then, at the bottom of the report, it says (about wdfsync.h, among
others) such things as "no IRQL was saved into SpinLock", "leaking the
SpinLock", and "The IRQL in 'SpinLock' was never restored."

This is in the WDK RC2 that just came out last week.

I am going through this as I've been told that DTM requires a passing
PREfast report.

[Ray Trent]

chris.a...@gmail.com wrote:
> I am going through this as I've been told that DTM requires a passing
> PREfast report.

Check the logo requirements. Luckily, DTM only requires that you
successfully *run* Prefast, not that it report no warnings/errors.

Actually, last time I checked, it doesn't even seem to require that, but
that's probably a bug :-).
--
Ray

[chris.a...@gmail.com]

Ray Trent wrote:

> chris.a...@gmail.com wrote:
>
> > I am going through this as I've been told that DTM requires a passing
> > PREfast report.
>
> Check the logo requirements. Luckily, DTM only requires that you
> successfully *run* Prefast, not that it report no warnings/errors.

What?? Then what's the point?

I guess in any case, I'd like to understand why it's failing, if only
from a driver quality perspective.

[Ray Trent]

The point is to get people to at least run Prefast on their drivers.
What they do with that information is up to them.

At least you know they will be using the WDK build environment rather
than some hacked up thing of their own... :-).

As for the actual problem you're seeing, I'm guessing someone probably
hasn't finished annotating those WDF functions completely yet...

Prefast is rather notorious about generating noise on functions that try
to encapsulate memory de/allocations, spinlocks, etc. I just got
finished running it again and a bunch more of those cropped up for me.

If you really want to get rid of the warnings, I suspect your only
recourse will be to pragma them out until MS gets things cleaned up some
more. But if you want more in depth analysis, please post the entire
offending code fragment... it's impossible to tell what Prefast is doing
with a summary.

--
Ray

[chris.a...@gmail.com]

Ray Trent wrote:

> The point is to get people to at least run Prefast on their drivers.
> What they do with that information is up to them.

Well, Winqual failed our submission today. Of course, they don't say
why anywhere on the site (at least that I can see), so we have to email
the winqual alias and wait, wait, wait...

If they say it's because of PREfast failures I think I'm going to pop a
blood vessel.

[David J. Craig]

You have to submit the xml file and not the txt file to pass prefast.

<chris.a...@gmail.com> wrote in message
news:1162341108....@h48g2000cwc.googlegroups.com...

[Dan]

According to the DTM documentation you do not have to submitt the XML
file as this may contain IP that you may not want to show or have
Microsoft own.

>From the help file:
"Because the standard PFD log file, Defects.xml, might contain
intellectual properly from your driver source code, you have to submit
only the summary results file, DefectsSum.txt, which does not contains
intellectual property."

There well crafted WHQL legal agreements are pretty clear that if you
share IP

THEY OWN IT!!!!

So submitting the XML file should not be required.

And yes they did indicate that it failed because of Prefast. So I will
give you a towel to clean up after your blood vessel pops

Dan

[David J. Craig]

The docs are what SHOULD happen and not what really happens. Clean out your
xml file of any IP and submit it. It does not have a signature or hash of
the driver. When and if they get it fixed the txt has a hash that will link
it to the driver, but for now hand massage the xml file.

"Dan" <d.mu...@motorola.com> wrote in message
news:1162387817.2...@b28g2000cwb.googlegroups.com...

[David J. Craig]

Also your driver does not have to pass prefast with no errors or warnings.
They just want to see what errors are being flagged. I suspect they will
require a clean prefast sometime in the future, but not thus far.

"Dan" <d.mu...@motorola.com> wrote in message
news:1162387817.2...@b28g2000cwb.googlegroups.com...

[chris.a...@gmail.com]

David J. Craig wrote:

> The docs are what SHOULD happen and not what really happens. Clean out your
> xml file of any IP and submit it. It does not have a signature or hash of
> the driver. When and if they get it fixed the txt has a hash that will link
> it to the driver, but for now hand massage the xml file.

Are you speaking from experience from a successful DTM submission where
you were asked to do this?

Why is this little tidbit not included in the DTM documentation?

Why do we need to hand-edit tool output?

I'd like someone from DTM to answer if possible, at least for the
latter two questions.

[chris.a...@gmail.com]

chris.a...@gmail.com wrote:

> I have some code in my driver that looks like this:
>
> 1) WdfSpinLockAcquire(......)
> 2) status = WdfCollectionAdd(....)
> 3) WdfSpinLockRelease(......)
>
> To this, PREfast says, "so and so function modified the IRQL but did
> not release it before exiting."

Sadly, this problem still exists in the 6000 RTM version of the WDK.

[Doron Holan [MS]]

This a known issue and will be fixed for the next version of the WDK. Pls
ignore the warning for now.

thx
d

--
Please do not send e-mail directly to this alias. this alias is for
newsgroup purposes only.
This posting is provided "AS IS" with no warranties, and confers no rights.

<chris.a...@gmail.com> wrote in message
news:1163184579.0...@e3g2000cwe.googlegroups.com...