|Summary:||Newer Ultra/AXi systems w/NCR scsi severely unstable.|
|Product:||[Retired] Red Hat Linux||Reporter:||David Miller <davem>|
|Component:||kernel||Assignee:||David Miller <davem>|
|Status:||CLOSED CURRENTRELEASE||QA Contact:|
|Fixed In Version:||Doc Type:||Bug Fix|
|Doc Text:||Story Points:||---|
|Last Closed:||1999-08-13 23:17:21 UTC||Type:||---|
|oVirt Team:||---||RHEL 7.3 requirements from Atomic Host:|
|Cloudforms Team:||---||Target Upstream Version:|
Description David Miller 1999-08-03 07:39:51 UTC
UltraSparc AXi systems produced with APB PCI bridge revision 13 or greater cannot withstain any amount of heavy disk activity on NCR scsi controller. The problem has to do with PCI write posting semantics of the APB bridge chipset, which were dampened by some retry semantic differences in the older chips so we never saw it on older AXi machines. Gerard Roudier has some attempts to work around this problem, and after testing and verifications this fix if working will go into the upcoming kernel updates. Worst part about this bug is that a machine will successfully fully install, then be unusable on first boot, therefore it will be triky to distribute a "fix" since people afflicted by the problem cannot even get the machine up far enough to install an RPM update. Suggestions are welcome on this latter issue.
Comment 1 David Miller 1999-08-13 23:17:59 UTC
I've fixed this kernel bug, it will be in our kernel updates coming up.