Hide Forgot
Created attachment 511598 [details] failed test log Description of problem: The cpuscaling test abort during the test on RH 5.6 on x86 using intel Pentium(R) CPU G630 with v7-1-3-46 but pass with v7-1-2-25 Version-Release number of selected component (if applicable): RH5.6 v7-1-3-46 How reproducible: Every time Steps to Reproduce: 1. Install proc listed above in a DL120 G7. Latest rom. 2. Execute cpuscaling Actual results: Test abort Expected results: Pass Additional info:Same system/rom/proc passes x64 successfully
Created attachment 511600 [details] Pass log with v7-1-2
Created attachment 511601 [details] Pass log with v7-1-3-46 on RH5.6 x64
Garry is this repeatable? Unfortunately I don't see anything that helps determine the issue in the logs except what appears may be a duplicate of bz#722384. What information is in /sys/devices/system/cpu/* ?
Created attachment 513722 [details] /sys/devices/system/cpu/* info /sys/devices/system/cpu/* info
Hi Rob, See attached cpu.zip. Yes, it's repeatable. Thanks for your support. Regards, Garry
Hi Garry, I suspect the cpu topology is being reported wrong in the kernel as it current defines every CPU as no relation to any other core. This data I thought was provided by the BIOS to the kernel, but also in RHEL5 I believe the -xen kernel can mask the topology in a DOM. Does this same problem occur on the BM kernel?
(In reply to comment #6) > Hi Garry, I suspect the cpu topology is being reported wrong in the kernel as > it current defines every CPU as no relation to any other core. This data I > thought was provided by the BIOS to the kernel, but also in RHEL5 I believe the > -xen kernel can mask the topology in a DOM. Does this same problem occur on > the BM kernel? Hi Rob, Yes, Hi Rob, Yes, this same problem occurs on the BM kernel. Thanks,
Hi Garry, is this issue still occurring? The topology is reported by the kernel so I don't think there's much we can alter in v7 to resolve; however the cpuscaling test did receive a fairly large update in v7-1.4.