Rtl8139 and 'One-Card Network Card Miniport Driver Test'

Hi,

I have a problem running CETK's 'One-Card Network Card Miniport Driver Test',
on the PB 4.2 and XScale-based board with an on-board ethernet interface
and 2 additional RTL8139 PCI cards.

There are two configuration I use for testing:
1.
- the 1st RTL NIC used for ethernet debug (and VMINI),
- the 2nd RTL NIC used for test,
- the internal NIC not active

2.
- the 1st RTL NIC used for test,
- the 2nd RTL NIC used for test,
- the internal NIC used fro ethernet debug (and VMINI)

The problem is that I cannot pass the test for the 1st configuration.
However the second one works fine. Also if the first one is compiled
in the debug mode it also passes.

By failing I mean that the system hangs during the test. There is no
fixed moment when it hangs. Instead it happens sometimes during the
first subtest and sometimes it goes as far as to 'sending to a group
of addresses' which is the last subtest. But it has never passed.

I examined the RTL8139 ETHDBG driver and it looks OK for me. Other
things that come into my mind are: a kitl problem, a pci problem or
maybe the test itself has some defect(s) (I've notied for example that
there are memory leaks within the test - but not sure if they're the
reason of my problem).

Can anyone help ?
Thanks,
Zbyszek

Harish
Fri Oct 17 15:03:34 CDT 2003

I believe the test is fine & does not have any memory leaks. We keep track
of memory allocation throughout the test & ensure that it does get freed. If
you have something that proves the leak then please let me know I'll be
happy to receive such information.

In case of you first configuration, where you use 1st RTL NIC for KITL, are
you running the test on VMINI1 or on the 2nd RTL NIC, which in this case
will get shown up as PCI\RTL81391.
That is in your first configuration you'll have VMINI1 (1st RTL NIC) &
PCI\RTL81391 (2nd RTL NIC).
In your 2nd configuration you'll have VMINI1 (internal NIC), PCI\RTL81391
(1st NIC) & PCI\RTL81392 (2nd NIC).

Try running the test on PCI\RTL81391. FYI, we have run the test on RTL8139 &
it did pass for OS 4.2. Besides that the test on VMINI1 did pass for the
platforms those we shipped in PB 4.2. If you could give test log or PB debug
output log then that would be great.

Thanks,
Harish.

--
This posting is provided "AS IS" with no warranties, and confers no rights.
"Zbyszek" <zskoczek@go2.pl> wrote in message
news:2910d5e6.0310170645.216fe3a9@posting.google.com...
> Hi,
>
> I have a problem running CETK's 'One-Card Network Card Miniport Driver
Test',
> on the PB 4.2 and XScale-based board with an on-board ethernet interface
> and 2 additional RTL8139 PCI cards.
>
> There are two configuration I use for testing:
> 1.
> - the 1st RTL NIC used for ethernet debug (and VMINI),
> - the 2nd RTL NIC used for test,
> - the internal NIC not active
>
> 2.
> - the 1st RTL NIC used for test,
> - the 2nd RTL NIC used for test,
> - the internal NIC used fro ethernet debug (and VMINI)
>
> The problem is that I cannot pass the test for the 1st configuration.
> However the second one works fine. Also if the first one is compiled
> in the debug mode it also passes.
>
> By failing I mean that the system hangs during the test. There is no
> fixed moment when it hangs. Instead it happens sometimes during the
> first subtest and sometimes it goes as far as to 'sending to a group
> of addresses' which is the last subtest. But it has never passed.
>
> I examined the RTL8139 ETHDBG driver and it looks OK for me. Other
> things that come into my mind are: a kitl problem, a pci problem or
> maybe the test itself has some defect(s) (I've notied for example that
> there are memory leaks within the test - but not sure if they're the
> reason of my problem).
>
> Can anyone help ?
> Thanks,
> Zbyszek

zskoczek
Thu Oct 30 02:27:02 CST 2003

Hi,

first of all, thanks for your replay.

As the situation has changed a bit (yet hasn't been completly
clarified) I'll give you a short backgorund:

As I stated previously the problem appeared when a RTL card was
tested and another RTL card was used for EDBG/VMINI at the same
time. It DIDN'T appear if RTL card was tested and the internal
adapter was used for EDBG/VMINI.

There was however one more factor I forgot to mention previously.
Because initially I had had problems starting the test I was adviced
to connect a remote tool (like registry editor) before running the
test. It did help to run the test but as I said one configuration
passed another failed.

So having the test keep failing I tried to deactivate VMINI and
separate ETHDBG and CETK server connections. At that stage my
configurations were:

Config 1.
- internal adpater for EDBG connection
- 1st RTL card for CETK server connection
- 2nd RTL card for testing
This is the configration, for which I spotted some memory leaks
(see the log below). However, the test passed.

*** ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
472786 PID:43e3d50a TID:43e3d47a *** TEST COMPLETED
472786 PID:43e3d50a TID:43e3d47a ***
472787 PID:43e3d50a TID:43e3d47a *** Test Name: 1c_Send
472788 PID:43e3d50a TID:43e3d47a *** Test ID: 2
472788 PID:43e3d50a TID:43e3d47a *** Library Path: \ndt_1c.dll
472789 PID:43e3d50a TID:43e3d47a *** Command Line: -t PCI\RTL81392
472790 PID:43e3d50a TID:43e3d47a *** Result: Passed
472791 PID:43e3d50a TID:43e3d47a *** Random Seed: 31999
472791 PID:43e3d50a TID:43e3d47a *** Thread Count: 1
472792 PID:43e3d50a TID:43e3d47a *** Execution Time: 0:00:51.540
472793 PID:43e3d50a TID:43e3d47a ***
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
472793 PID:43e3d50a TID:43e3d47a
472794 PID:43e3d50a TID:43e3d47a </TESTCASE RESULT="PASSED">
472795 PID:43e3d50a TID:43e3d47a END GROUP: ndt_1c.dll
472802 PID:a3fc9bc2 TID:43e3d47a Memory: Memory leaks detected
472804 PID:a3fc9bc2 TID:43e3d47a Memory: The memory block at
0x002e28e0 with total length = 36 bytes
472804 PID:a3fc9bc2 TID:43e3d47a Memory: 0000 = 78 69 5a 4b 24 00 00
00 00 00 00 00 b0 28 2e 00
472805 PID:a3fc9bc2 TID:43e3d47a Memory: 0010 = 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00
472805 PID:a3fc9bc2 TID:43e3d47a Memory: 0020 = 3c 2d 1e 0f
472806 PID:a3fc9bc2 TID:43e3d47a Memory: The memory block at
0x002e28b0 with total length = 36 bytes
472806 PID:a3fc9bc2 TID:43e3d47a Memory: 0000 = 78 69 5a 4b 24 00 00
00 e0 28 2e 00 70 28 2e 00
472807 PID:a3fc9bc2 TID:43e3d47a Memory: 0010 = 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00
472807 PID:a3fc9bc2 TID:43e3d47a Memory: 0020 = 3c 2d 1e 0f
472807 PID:a3fc9bc2 TID:43e3d47a Memory: The memory block at
0x002e2870 with total length = 44 bytes
472808 PID:a3fc9bc2 TID:43e3d47a Memory: 0000 = 78 69 5a 4b 2c 00 00
00 b0 28 2e 00 f0 9b 32 00
472809 PID:a3fc9bc2 TID:43e3d47a Memory: 0010 = 00 02 04 06 08 0a 00
02 04 06 08 0a ff ff ff ff
472809 PID:a3fc9bc2 TID:43e3d47a Memory: 0020 = ff ff ff ff ff ff ff
ff 3c 2d 1e 0f
472810 PID:a3fc9bc2 TID:43e3d47a Memory: The memory block at
0x00329bf0 with total length = 260 bytes
472810 PID:a3fc9bc2 TID:43e3d47a Memory: 0000 = 78 69 5a 4b 04 01 00
00 70 28 2e 00 00 00 00 00
472811 PID:a3fc9bc2 TID:43e3d47a Memory: 0010 = 10 13 f4 01 00 00 00
00 00 00 00 00 00 00 00 00
472811 PID:a3fc9bc2 TID:43e3d47a Memory: 0020 = 56 64 01 17 00 00 00
00 00 00 00 00 00 00 00 00
472812 PID:a3fc9bc2 TID:43e3d47a Memory: 0030 = 9a 00 50 46 dc 07 00
00 34 00 50 46 06 00 00 00
472817 PID:a3fc9bc2 TID:43e3d47a Memory: 0040 = e0 9d 04 00 00 00 00
00 c0 a5 03 00 90 a5 fd 01
472818 PID:a3fc9bc2 TID:43e3d47a Memory: 0050 = 58 9c 32 00 00 00 00
00 00 00 00 00 02 00 00 00
472819 PID:a3fc9bc2 TID:43e3d47a Memory: 0060 = 00 00 00 00 00 00 00
00 00 00 00 00 98 a4 fd 01
472819 PID:a3fc9bc2 TID:43e3d47a Memory: 0070 = 98 a4 fd 01 f8 72 f4
01 00 9c 32 00 00 00 00 00
472820 PID:a3fc9bc2 TID:43e3d47a Memory: 0080 = 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00
472821 PID:a3fc9bc2 TID:43e3d47a Memory: 0090 = 98 d4 d2 83 00 00 00
00 00 00 00 00 c0 28 2e 00
472821 PID:a3fc9bc2 TID:43e3d47a Memory: 00a0 = f0 28 2e 00 64 00 00
00 00 00 00 00 00 00 00 00
472822 PID:a3fc9bc2 TID:43e3d47a Memory: 00b0 = 00 00 00 00 00 00 00
00 06 00 00 00 04 00 00 00
472823 PID:a3fc9bc2 TID:43e3d47a Memory: 00c0 = 80 28 2e 00 00 00 00
00 dc 05 00 00 64 00 00 00
472823 PID:a3fc9bc2 TID:43e3d47a Memory: 00d0 = 28 00 00 00 01 00 00
00 80 3e 00 00 90 01 00 00
472823 PID:a3fc9bc2 TID:43e3d47a Memory: 00e0 = 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00
472824 PID:a3fc9bc2 TID:43e3d47a Memory: 00f0 = 7a 26 07 00 be 36 07
00 00 00 00 00 00 00 00 00
472825 PID:a3fc9bc2 TID:43e3d47a Memory: 0100 = 3c 2d 1e 0f
472826 PID:a3fc9bc2 TID:43fa57ee Prefetch Abort: Thread=83fa5c20
Proc=80526da8 'device.exe'
472826 PID:a3fc9bc2 TID:43fa57ee AKY=00000009 PC=01f47d30 RA=01f47d30
BVA=01f47d30 FSR=000004f0
472828 PID:43e3d50a TID:43e3d47a ***
==================================================================
472828 PID:43e3d50a TID:43e3d47a *** MEMORY INFO
472829 PID:43e3d50a TID:43e3d47a ***
472829 PID:43e3d50a TID:43e3d47a *** Memory Total: 30,818,304
bytes
472830 PID:43e3d50a TID:43e3d47a *** Memory Used: 3,055,616
bytes
472831 PID:43e3d50a TID:43e3d47a *** Memory Free: 27,762,688
bytes
472831 PID:43e3d50a TID:43e3d47a ***
472832 PID:43e3d50a TID:43e3d47a *** Kernel Used: 229,376
bytes
472833 PID:43e3d50a TID:43e3d47a *** Water Mark: 6,759
pages
472833 PID:43e3d50a TID:43e3d47a ***
472834 PID:43e3d50a TID:43e3d47a *** Store Total: 30,740,480
bytes
472835 PID:43e3d50a TID:43e3d47a *** Store Used: 200,228
bytes
472836 PID:43e3d50a TID:43e3d47a *** Store Free: 30,540,252
bytes
472836 PID:43e3d50a TID:43e3d47a ***
==================================================================
472837 PID:43e3d50a TID:43e3d47a
472838 PID:43e3d50a TID:43e3d47a ***
==================================================================
472838 PID:43e3d50a TID:43e3d47a *** SUITE SUMMARY
472839 PID:43e3d50a TID:43e3d47a ***
472839 PID:43e3d50a TID:43e3d47a *** Passed: 1
472840 PID:43e3d50a TID:43e3d47a *** Failed: 0
472840 PID:43e3d50a TID:43e3d47a *** Skipped: 0
472841 PID:43e3d50a TID:43e3d47a *** Aborted: 0
472842 PID:43e3d50a TID:43e3d47a *** -------- ---------
472842 PID:43e3d50a TID:43e3d47a *** Total: 1
472843 PID:43e3d50a TID:43e3d47a ***
472843 PID:43e3d50a TID:43e3d47a *** Cumulative Test Execution Time:
0:00:51.540
472845 PID:43e3d50a TID:43e3d47a *** Total Tux Suite Execution Time:
0:00:51.603
472846 PID:43e3d50a TID:43e3d47a *** CPU Idle Time:
0:00:00.000
472846 PID:43e3d50a TID:43e3d47a ***
==================================================================
472847 PID:43e3d50a TID:43e3d47a </TESTGROUP>
472848 PID:43e3d50a TID:43e3d47a @@@@@@0
472848 PID:43e3d50a TID:43e3d47a
472958 PID:a3e5679a TID:a3e9e62e Done Running Test



Config 2.
- 1st RTL card for EDBG connection
- 2nd RTL card for CETK server connection
- 3rd RTL card for testing
The failing configuration (see the log below).

vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv
67857 PID:c3e3b1a2 TID:c3e2079e *** TEST STARTING
67858 PID:c3e3b1a2 TID:c3e2079e ***
67859 PID:c3e3b1a2 TID:c3e2079e *** Test Name: 1c_Send
67862 PID:c3e3b1a2 TID:c3e2079e *** Test ID: 2
67863 PID:c3e3b1a2 TID:c3e2079e *** Library Path: \ndt_1c.dll
67864 PID:c3e3b1a2 TID:c3e2079e *** Command Line: -t PCI\RTL81392
67865 PID:c3e3b1a2 TID:c3e2079e *** Random Seed: 24866
67866 PID:c3e3b1a2 TID:c3e2079e *** Thread Count: 0
67867 PID:c3e3b1a2 TID:c3e2079e ***
vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv
67868 PID:c3e3b1a2 TID:c3e2079e BEGIN TEST: "1c_Send", Threads=0,
Seed=24866
67869 PID:c3e3b1a2 TID:c3e2079e c3e2079e: ndt_1c Msg: Start
1c_Send test on the adapter PCI\RTL81392
67870 PID:c3e3b1a2 TID:c3e2079e c3e2079e: ndt_1c Msg: Getting
basic info about the adapter
67873 PID:c3e3b1a2 TID:c3e2079e c3e2079e: ndt_1c Msg: Get
OID_GEN_MAXIMUM_TOTAL_SIZE = 1514
67874 PID:c3e3b1a2 TID:c3e2079e c3e2079e: ndt_1c Msg: Get
OID_GEN_MAXIMUM_FRAME_SIZE = 1500
67879 PID:c3e3b1a2 TID:c3e2079e c3e2079e: ndt_1c Msg: Testing
sending to a (random) directed address
67880 PID:c3e3b1a2 TID:c3e2079e c3e2079e: ndt_1c Msg: Sending:
Send
67881 PID:c3e3b1a2 TID:c3e2079e c3e2079e: ndt_1c Msg: Sending
to a random address
74414 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74414 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74415 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74416 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74416 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74417 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74418 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74419 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74419 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74420 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74421 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74421 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74422 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74422 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74423 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74424 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74424 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74425 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74426 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74426 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74427 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74428 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74428 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74429 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74430 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74431 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74431 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74432 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74433 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74433 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74434 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74435 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74436 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74436 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74437 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74438 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74438 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74439 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74440 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74441 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74441 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001
74942 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c001000c
74942 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c001000c
74943 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c001000c
74944 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c001000c
74944 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c001000c
74945 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c001000c
74946 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c001000c
74946 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c001000c
74947 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c001000c
74948 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c001000c
74948 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c001000c
74949 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c001000c
74950 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c001000c
74950 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c001000c
74951 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c001000c
74952 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c001000c
80426 PID:63fc9bc2 TID:a3facf72 CRequestSend::SendComplete:Status =
c0000001


Then, as the next step, for the two configuration I tried NOT
to connect the remote tool and what suprised me a lot both
configuration started working and it was a stable behaviour.

So knowing that NOT connecting the registry helps I went back
to the original configuration (with EDBG/VMINI/CETK server
connected over the same interface) and tried that and it also
worked fine. However, the test needed about 20 minutes to finish
whereas for the previous configuration (with separate EDBG and
CETK), or with the registry connected, it only requires about
1.5 minute.

So I guess at the moment the questions would be:
- why remote registry speeds up the tests (is it becasue of the
KITL connection)?
- is it possible (it looks like) that the remote registry
connection causes the test to break ?
- why the test is breaking for one configuration only, whereas
it works fine for another one (non-PCI based one) ?

Any help would be more than welcome.
Kind regards,
Zbyszek

Harishr
Mon Dec 29 20:11:15 CST 2003

-why remote registry speeds up the tests (is it becasue of the
KITL connection)?

No. Running remote registry should not cause any things. Note that these
tools use KITL (over EDBG) connection & hence should have nothing to do
with production drivers.

- is it possible (it looks like) that the remote registry
connection causes the test to break ?

NO.

- why the test is breaking for one configuration only, whereas
it works fine for another one (non-PCI based one) ?

I still believe that it is not the test, as we could run the test
successfully on all sort of miniport drivers across all supported platforms.
I would suggest to ensure that it is not the problem posed by sharing of
IRQs.

Now the configs. & logs you attached.

Config 1.
- internal adpater for EDBG connection
- 1st RTL card for CETK server connection
- 2nd RTL card for testing
This is the configration, for which YOU spotted some memory leaks.

Please note that here the test is not complete. 1card test has 10 sub tests
& to have all the subtests to be finsihed it takes around ~20/30 minutes.
It seems that since there was memory leak reported by the first sub test
the whole test suite halted (& hence it took less time). As I explained
earlier we keep track of all the memory allocated, so looking at the log I
can suspect that, some of the NDIS packets were never returned
back to our test protocol driver & that may get indicated as leaks.

Config 2.
- 1st RTL card for EDBG connection
- 2nd RTL card for CETK server connection
- 3rd RTL card for testing
The failing configuration (see the log below).

Here the error indicates STATUS_UNSUCCESSFUL = c0000001 & this is indicated
in Sendcomplete, which indicates that during this config. not even the
single packet could be sent.

I want to know following things:
<1> Which BSP you are using? Are you using any std. BSP that is shipped in
PB?
<2> Are you using RTL8139 miniport driver shipped in PB?

Also please do following things.

<3> Please try to ensure that EDBG does not share any IRQ with any other
device on that BSP. Also you can run the test without EDBG connection. Just
bootup the device with iamge with KITL & Debugger disabled & use one card
for CETK connection & run the CETK test on any other network card.

<4> If you doubt the test then for that particular config. run continuous
ping test with different buffer sizes & see the response. Obviously you'll
have to put the test card on the network where you can be able to ping to
any other machine.

Please note that we have run tests on GEODE platform which has 2 RTL8139 on
board & our tests ran fine even when one card was used for EDBG & other
card was being used as production driver.

I hope this would help.