Based on all the testing with different kernels I suspect this might be a HW/FW issue, and we might want to seek guidance from Cavium on how to effectively address this issue on Seuss. Please note that the issue is not reproduced on a similar CRB system, and occurs *only on* Seuss. Where there any older kernel versions on which these tests passed on Seuss? If so we can do a reverse bisect to find the offending patch.
Based on all the testing with different kernels I suspect this might be a HW/FW issue, and we might want to seek guidance from Cavium on how to effectively address this issue on Seuss. Please note that the issue is not reproduced on a similar CRB system, and occurs *only on* Seuss. Where there any older kernel versions on which these tests passed on Seuss? If so we can do a reverse bisect to find the offending patch.