PCIE test card causes BSOD

[Mark_Huang]

Hi,

We are using Burn In v8.1 and PCIE test cards to stress the PCIE interface.
The interposer card is the EVM of DIODES PI7C9X2G606PR (https://www.diodes.com/products/conn.../PI7C9X2G606PR)

After running for 5 hours, Windows 10 64bits happened BSOD (refer to attached pictures)

Does anyone have ideas about this behavior?

Thanks,
Mark.

[David (PassMark)]

The version of BurnInTest you are using is old. Can you try with V9.

When were these cards purchased? What firmware release are you using?
What device driver release are you using?
Is the problem reproducible? If yes, does it always happen after 5 hours?
Did you get a crash dump from the BSOD? Your photo indicated that it was a driver fault. But didn't indicate which driver.

We have never done any testing with the EVM of DIODES PI7C9X2G606PR extender card.

[Mark_Huang]

Okay, I would like to confirm the setup first.

Attached photo is the card I used.
I consider my card should use PCIe Test Card device driver (Old release for Firmware V1.8 ), correct?
Before I was using PCIe Test Card device driver (For Firmware V2.0) on it.

Today I change driver to the old one.
But I can't run PCIE test item.
The behavior is weird.
Before I run Burn In test, it can detect those cards.
After I run Burn In test, it CAN'T detect those cards.

Is there anything wrong on my side?
Thanks.

[David (PassMark)]

The version of BurnInTest you are using is old. Can you try with V9.

Are you sure you have the V1.8 firmware? There were not many units sold with that older firmware release. Can you post the log from the PCIeTest software so that we can see the firmware version.

[Mark_Huang]

I tried Burn In v9.
PCIE and ethernet advance test does not work. (CPU/RAM test does work)
It keeps pending. no progress....

I run PCIe test alone by installing new and old driver.
I think it should be v2.0.....do you agree?

[HamidK (PassMark)]

From the log file, it seems you have the V2.0 firmware. So please make sure you are using the latest driver and test application.
https://www.passmark.com.au/download...t_download.htm

The short run in the PCIeTestLog_new driver.txt seems to be fine, without error.

Is the 5h crash problem reproducible? If yes, does it always happen after 5 hours?
Can you send the crash dump file for further investigation?

[Mark_Huang]

I'd like to confirm Burn In v9.0 test problem first.
I can't run the test successfully.

PCIE and ethernet advance test does not work. (CPU/RAM test does work)
It keeps pending. no progress.... (please refer to attached photo)

I use below conditions...
OS: Win 10 64 bits
Test SW: Burn In v9.0
Test fixture: Pass Mark PCIE test cards (FW 2.0, according to your check)
Driver: PCIe Test Card device driver (For Firmware V2.0)

[David (PassMark)]

In the screen shot the test was only running for 30 seconds. And there appears to be some activity on the PCIe test. So I am not sure why you are saying there was no progress.

If it is just the networking test that has the problem, what settings are you using for the network test.

[Mark_Huang]

Now we can run Burn In v9.0.
We only run PCIE test cards items. (using 4 cards)
And the BSOD issue is still existed.
It happened after 30 mins running.

Would you have other ideas?
Thanks a lot.

[David (PassMark)]

Does it happen if you don't use the extender card?
Does it happen if you just use 1 card at a time?

Is there any additional information about the BSOD. This tool might help
https://www.nirsoft.net/utils/blue_screen_view.html

[Mark_Huang]

- Currently we didn't use the extender card. We plug 4 PCIE test cards on motherboard directly.
- We can't see BSOD by using 1 card for 12 hours. We didn't test it longer time.
- Attached is the mini dump file for your reference.

I refer to "https://docs.microsoft.com/en-us/windows-hardware/drivers/debugger/bug-check-0x10d---wdf-violation"
It showed "Parameter 1 indicates 0x8. The handle of the DMA transaction object. An operation occurred on a DMA transaction object while it was not in the correct state."

[HamidK (PassMark)]

Can you please run 4 instances of the PCITest application to see if you can reproduce the BSOD?
https://www.passmark.com.au/ftp/PCIeTestV2.0.1001.zip

[Mark_Huang]

Yes, we have tried to use standalone PCIE loopback test tool by using 4 cards and open 4 x application.
It ran normally for 12 hours.
Then I closed all 4 x application.

When I click restart in windows 10, the same BSOD happened.
However it can not happen every time while I restart Win 10.

[David (PassMark)]

From the crash it looks like the error was in the device driver for the PCIe bus controller (i.e. a motherboard issue). Not in our code.

But we are in the process of setting up a couple of test machines to make sure the problem can't be reproduced on other hardware (as we don't have access to the same hardware you are using).

[Mark_Huang]

Thanks for your feedback.
It seems PCIeTestV2.0.1001.zip is not easy to see BSOD.

For your reference,
- we can see BurnIn BSOD on both AMD V1000 and Skylake-H platform in Win10.
- we analysis minidump file through http://www.osronline.com/page.cfm?name=Analyze
It shows it comes from IGLx.sys