card oops with "Data Cache Parity Error"

[zwg132883]

Kernel oops happened on our cards a couple of times with "Data Cache Parity Error".  We met such issue on different cards with different SW version.

CPU:T1042
OS: linux 4.14

<4>[7798506.531955] Machine check in kernel mode.
<4>[7798506.531957] Machine check in kernel mode.
<4>[7798506.531961] Caused by (from MCSR=20080000):
<4>[7798506.531964] Caused by (from MCSR=8000):
<4>[7798506.531966] Data Cache Parity Error
<4>[7798506.531967] Load Error Report
<4>[7798506.531970] Machine Check Physical Address: 0x2dbaca48
<4>[7798506.531972] Oops: Machine check, sig: 7 [#1]
<4>[7798506.531973] BE SMP NR_CPUS=8 CoreNet

What can cause CPU "Data Cache Parity Error"? How to debug/resolve such issue?

[ufedor]

DCPE should not occur during normal processor operation and could be caused by several factors:

1) processor connection is not implemented in accordance with the Data Sheet and Design Checklist requirements

2) processor is overclocked

3) processor is overheated

4) there is external strong source of EMI

Considering above it is required to perform thorough design check referring the Data Sheet and Design Checklist, ensure that configured processor clocking mode is valid, measure the processor die temperature, ensure that proper shielding is implemented.