All,
I have a problem that I could really use some help with regarding the
headerpayloadsplit NDIS test offered by ndistest. I first experienced
this issue when running the DTM LWF test against a driver I’m trying to
get certified. All the DTM tests pass except for headerpayloadsplit. I
wish it was a problem with my driver, but this test fails both with or
without my LWF driver installed
I have fallen back to running this test directly using ndistest as a
simplified test case. Per the Driver Test Manager docs, I install two
“NDISTest 6.0 - CL - IM/Filter Test” adapters using drivers distributed
along with ndistest. The two miniport adapters show up under device
manager as working properly. When setting up DTM for LWF, it asks me to
select a Test Device and a Support device. I select the first NDISTest
IM/Filter adapter as the Test Device and the second IM/Filter adapter as
the Support Device, exactly the way it is described in the Driver Test
Manager docs. When running ndistest directly, I follow a similar
procedure ( first NDISTest IM/Filter adapter selected as Test Device,
None selected as Message Device and second NDISTest IM/Filter adapter
selected as Support Device ). As for a job list, I just make sure the
headerpayloadsplit test is selected along with the mandatory preconfig
and postconfig jobs. Then I click start.
The resulting headerpayloadsplit.htm file always shows the following
output …
Unexpected Failure Outside of a Test Case or Action
88888 A breakpoint was hit in the protocol driver while this test was
executing
… To investigate this, I setup a remote debug session and watched for
the debugger to display some kind of error. Unfortunately, it just spit
out ndistest log messages and didn’t hit any breakpoints. There was log
output that looked suspect, a slew of these …
NOTE: This is without EVER installing my LWF drivers on the system.
CNDTHeaderDataSplitVerifyModule::OnPacketListsIn: Packet has been
indicated with NDIS_FLAGS_PHYSICAL_ADDRESS_PRESENT flag set. But the
physical address (e93) in the Packet OOB information does not match the
physical address (e8e) for the start of packet data in the ‘Data’ MDL
which is at the virtual address XXXXXXXX.
To rule out a wonky OS install, I re-installed to a pristine state and
tried again only to get the same result. To rule out hardware problems,
I tried running the test using a clean VMWare installation and again got
the same result.
The only way I found to get this particular test to pass is to use a
real Ethernet adapter as the support device. This is contrary to the
Driver Test Manager documentation, but when a real support device is
used, both with and without my LWF installed, the suspicious debugger
output disappears and all the tests pass without any errors.
Does anyone have any insight into this? After 8 hours of pulling my hair
out, I’m all out of ideas
Thanks in advance,
-Matthew