UMDF debugging problems

NTDEV
1

I’m building a UMDF driver based on the Echo sample provided (WDK 6000).

I don’t seem to be having any problems attaching a debugger and breaking as
required in the source code, but as I step through the code, it appears the
line information in the program debug database is all wrong (as the debugger
steps in all the wrong places).

I initially thought (as always) I had an old .pdb lying around somewhere on
my path that the debugger was picking up, but I’ve since triple checked
there aren’t any old files lying around and verified proof-positive via the
modules window which symbol file was being loaded.

I’m now wondering if there’s a compiler/debugger related bug that I’m
unaware of and need an update of some sort. As well, I’m using the Windows
Vista build environment (as instructed to do so by 6000’s release
documentation) for development under XP targeting XP , so I’m not too sure
if that’s possibly causing me some grief.

I get the same debugging behaviour in either the Visual Studio (2005) user
mode debugger or windbg.

Does anyone have any ideas?

Thanks,

2

Are you debugging a release build? If so, line numbers and locals can
be wrong due
to optimization.

-Jeff

-----Original Message-----
From: xxxxx@lists.osr.com
[mailto:xxxxx@lists.osr.com] On Behalf Of Greg Thomson
Sent: Wednesday, January 10, 2007 12:00 PM
To: Windows System Software Devs Interest List
Subject: [ntdev] UMDF debugging problems

I’m building a UMDF driver based on the Echo sample provided (WDK 6000).

I don’t seem to be having any problems attaching a debugger and breaking
as
required in the source code, but as I step through the code, it appears
the
line information in the program debug database is all wrong (as the
debugger
steps in all the wrong places).

I initially thought (as always) I had an old .pdb lying around somewhere
on
my path that the debugger was picking up, but I’ve since triple checked
there aren’t any old files lying around and verified proof-positive via
the
modules window which symbol file was being loaded.

I’m now wondering if there’s a compiler/debugger related bug that I’m
unaware of and need an update of some sort. As well, I’m using the
Windows
Vista build environment (as instructed to do so by 6000’s release
documentation) for development under XP targeting XP , so I’m not too
sure
if that’s possibly causing me some grief.

I get the same debugging behaviour in either the Visual Studio (2005)
user
mode debugger or windbg.

Does anyone have any ideas?

Thanks,

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

3

Thanks, that was the problem. When I rebuilt everything using the Checked
Build Environment, it all debugged fine.

I assumed the checked/free build environments were for the OS installed and
not for the release/debug version of the driver. Also, wouldn’t expect a
release build to output a .pdb file.

Guess this newbie better read-up a little more on the build environment
before I get myself into more trouble.

Thanks a million !

wrote in message news:xxxxx@ntdev…
Are you debugging a release build? If so, line numbers and locals can
be wrong due
to optimization.

-Jeff

-----Original Message-----
From: xxxxx@lists.osr.com
[mailto:xxxxx@lists.osr.com] On Behalf Of Greg Thomson
Sent: Wednesday, January 10, 2007 12:00 PM
To: Windows System Software Devs Interest List
Subject: [ntdev] UMDF debugging problems

I’m building a UMDF driver based on the Echo sample provided (WDK 6000).

I don’t seem to be having any problems attaching a debugger and breaking
as
required in the source code, but as I step through the code, it appears
the
line information in the program debug database is all wrong (as the
debugger
steps in all the wrong places).

I initially thought (as always) I had an old .pdb lying around somewhere
on
my path that the debugger was picking up, but I’ve since triple checked
there aren’t any old files lying around and verified proof-positive via
the
modules window which symbol file was being loaded.

I’m now wondering if there’s a compiler/debugger related bug that I’m
unaware of and need an update of some sort. As well, I’m using the
Windows
Vista build environment (as instructed to do so by 6000’s release
documentation) for development under XP targeting XP , so I’m not too
sure
if that’s possibly causing me some grief.

I get the same debugging behaviour in either the Visual Studio (2005)
user
mode debugger or windbg.

Does anyone have any ideas?

Thanks,

—
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