DTM and Device Path Excerciser

Michal_Vodicka-2 · March 6, 2007, 2:33pm

We’re trying to sign new build of our USB driver with DTM. For Vista there was no problem for both x86 and x64. But Device Path Exerciser Test fails at XP. After few attempts it passed at XP64 but we can’t manage to make it working at 32-bit version.

The test always fails after about hour of running “Miscellaneous Tests (part 1)”. To find what can cause problem I’d need to know what exactly failed. Unfortunately, we can’t find this info. DTM copies DC2.LOG files when test finishes but they don’t contain anything useful; the last line is always the info about misc test start and that’s all.

I tried to start manually the appropriate command line (DevPathExer.exe /dr /HCT /lwa) with an option which creates diagnostic log but this way the test doesn’t fail. Only when started as DTM job. I tried to turn on my driver traces but there was nothing interesting. I’m affraid it isn’t my driver what fails but some other driver lower in the stack.

Any idea how to diagnose such a problem? Is is possible to add more parameters to this DTM test command line (to create diag info) and disable DTM from deleting it when test finishes? (really ingenious feature :-/) Or how to tell DevPathExer to dump info about the operation which failed?

(With DTM MS reached the next level with creating obnoxius and unmanageable monsters…)

Best regards,

Michal Vodicka
UPEK, Inc.
[xxxxx@upek.com, http://www.upek.com]

Rob_Linegar · March 7, 2007, 5:05am

Hi Michal,

Been through similar myself in the last few weeks on XP, but with the
last HCT kit.

Is there still a decent help file with the DTM like there was with with
HCT 12.1.01? (which is still valid for XP 32 submissions till june
1st…)…

The problems I had with Misc Tests 1 were with a virtual serial port
driver.

reads + writes from invalid offsets
reads + writes for 0 bytes
‘sub opens’ (basically open is called again even though the device
object is exclusive…)

All from a driver that sailed through a couple years ago…

The command line looks different too… I was using “DevPathExer.exe /m
/dr ” to run misc tests 1.

The documentation with HCT was good as far as covering what tests were
run and the command line params. If DTM doesn’t include that info yet
I’d suspect that most of it hasn’t really changed. Check against the HCT
doc’s and maybe even try reproducing with that?

I did have an option to make a “diags.log” which held lots of info…
HOWEVER… Creating these big logs can actually make the test fail so
only enable that if you have to.

>> (With DTM MS reached the next level with creating obnoxius and
unmanageable monsters…)
( Great! Guess what I’m doing after this last submission with HCT? )

BR,

Rob Linegar
Software Engineer
Data Encryption Systems Limited
www.des.co.uk | www.deslock.com

-----Original Message-----
From: xxxxx@lists.osr.com
[mailto:xxxxx@lists.osr.com] On Behalf Of Michal Vodicka
Sent: 06 March 2007 19:33
To: Windows System Software Devs Interest List
Subject: [ntdev] DTM and Device Path Excerciser

We’re trying to sign new build of our USB driver with DTM. For Vista
there was no problem for both x86 and x64. But Device Path Exerciser
Test fails at XP. After few attempts it passed at XP64 but we can’t
manage to make it working at 32-bit version.

The test always fails after about hour of running “Miscellaneous Tests
(part 1)”. To find what can cause problem I’d need to know what exactly
failed. Unfortunately, we can’t find this info. DTM copies DC2.LOG files
when test finishes but they don’t contain anything useful; the last line
is always the info about misc test start and that’s all.

I tried to start manually the appropriate command line (DevPathExer.exe
/dr /HCT /lwa) with an option which creates diagnostic log
but this way the test doesn’t fail. Only when started as DTM job. I
tried to turn on my driver traces but there was nothing interesting. I’m
affraid it isn’t my driver what fails but some other driver lower in the
stack.

Any idea how to diagnose such a problem? Is is possible to add more
parameters to this DTM test command line (to create diag info) and
disable DTM from deleting it when test finishes? (really ingenious
feature :-/) Or how to tell DevPathExer to dump info about the operation
which failed?

(With DTM MS reached the next level with creating obnoxius and
unmanageable monsters…)

Best regards,

Michal Vodicka
UPEK, Inc.
[xxxxx@upek.com, http://www.upek.com]

—
Questions? First check the Kernel Driver FAQ at
http://www.osronline.com/article.cfm?id=256

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

William_Michael_Jones-2 · March 7, 2007, 9:19am

Hi Michal,

I have two problems with the DTM.
(1) One is with the Common Scenario Stress with IO failing with BSOD. This
fails with our driver installed running the tests with our driver but also
BSODs without our driver installed running the test against the LPT port.
This happens on two different machines which both support S2. I have tried
to report his problem to Microsoft but have had no success so far.

(2) The second Plug Play Driver Test fails with a timeout. Now this gets
fixed for the x32 by applying the DTM fixes found at
http://www.microsoft.com/whdc/DevTools/WDK/DTM/DTMQFE.mspx

But the fix breaks the x64 by copying the ia64 binaries for the x64. duh! We
need a DTM contact point you guys. OSR HELP. Don Burns has recently wrote
to me that a DTM newgroups will be up in a month. We need it now.

(3) Now I see Michal you are having a problem. When you get the contact
point at MS let me know too, please.

-William Michael Jones “Mike”
“Michal Vodicka” wrote in message
news:xxxxx@ntdev…
We’re trying to sign new build of our USB driver with DTM. For Vista there
was no problem for both x86 and x64. But Device Path Exerciser Test fails at
XP. After few attempts it passed at XP64 but we can’t manage to make it
working at 32-bit version.

The test always fails after about hour of running “Miscellaneous Tests (part
1)”. To find what can cause problem I’d need to know what exactly failed.
Unfortunately, we can’t find this info. DTM copies DC2.LOG files when test
finishes but they don’t contain anything useful; the last line is always the
info about misc test start and that’s all.

I tried to start manually the appropriate command line (DevPathExer.exe /dr
/HCT /lwa) with an option which creates diagnostic log but this
way the test doesn’t fail. Only when started as DTM job. I tried to turn on
my driver traces but there was nothing interesting. I’m affraid it isn’t my
driver what fails but some other driver lower in the stack.

Any idea how to diagnose such a problem? Is is possible to add more
parameters to this DTM test command line (to create diag info) and disable
DTM from deleting it when test finishes? (really ingenious feature :-/) Or
how to tell DevPathExer to dump info about the operation which failed?

(With DTM MS reached the next level with creating obnoxius and unmanageable
monsters…)

Best regards,

Michal Vodicka
UPEK, Inc.
[xxxxx@upek.com, http://www.upek.com]

Michal_Vodicka-2 · March 7, 2007, 12:54pm

Rob,

The command line looks different too… I was using “DevPathExer.exe /m
/dr ” to run misc tests 1.
>
Yes, the command line I used was for whole test which includes misc tests. /m works there, too, but we weren’t able to reproduce problem when the test wasn’t started by DTM itself.

> The documentation with HCT was good as far as covering what tests were
> run and the command line params. If DTM doesn’t include that info yet
> I’d suspect that most of it hasn’t really changed. Check against the HCT
> doc’s and maybe even try reproducing with that?
>
I’m affraid the problem started with the last DTM update. Previously the same driver was successfully signed several times; there are changes but nothing apparently related to this kind of test.

WDK contains dc2.exe which reports version 3.2 and DevPathExer.exe is the same tool reporting version 3.3.

I don’t understand why the tool doesn’t report details about the failure… Or maybe I’m missing something. Any sane developer writing a tool which invokes tens or hundreds thousands tests would implement detailed failure reports as one of the first things.

> I did have an option to make a “diags.log” which held lots of info…
> HOWEVER… Creating these big logs can actually make the test fail so
> only enable that if you have to.
>
I have this option. But DTM doesn’t use it and as I said, the problem can’t be even reproduced when I start the same test using the same command line on the same hardware… So I’d need to persuade DTM to add one more option to its test.

> >> (With DTM MS reached the next level with creating obnoxius and
> unmanageable monsters…)
> ( Great! Guess what I’m doing after this last submission with HCT? )
>
Switch to Linux? Actually, if I think about both Vista and DTM at once, Linux doesn’t look like a bad possibility…

Best regards,

Michal Vodicka
UPEK, Inc.
[xxxxx@upek.com, http://www.upek.com]