Re: ECC Memory question
8-§ > Aug 6 22:57:25 stahlw06 vmunix: Detected Low Priority Machine Check.
8-§ > Aug 6 22:57:25 stahlw06 vmunix: LPMC type : SEDC (ECC-corrected single-bit error)
8-§ > Aug 6 22:57:25 stahlw06 vmunix: Error Address : 0x3123c58
8-§ >
8-§ > appears in the log whenever a correctable one-bit error is found. But
8-§ > what happens on parisc-linux in such a case?
8-§ >
LPMC is a kind of error which has a less criticity than HPMC. In your case is due to a problem either on the CPU cache or on the RAM error. If you use the tool provided in the PDC (like the BIOS of x86 but with much more capabilities and troubleshooting functionalities) you can indeed recognize the bank where the error occurred! Well after a certain amount of "single-bit error" the system provides a reboot; the same happens with HPMCs: the system is able to provide a "correction action" when single-bit errors occurs. I mean it maps the corrupted areas and "jumps them over". When, instead, one double-bit error occurs (or a certain number of them, exactly I don't remember!), the system automatily reboot the OS, signaling the presence of the HPMC/LPMC.
If, in your case, it didn't happen, that's fine, but in a certain sense you are advised! I don't mean it could happen tomorrow, or the day after tomorrow, it could happen in a year or more, depending how huch the system is either stressed or dirty or ....
When HPMC occurs on a memory bank the PDC provides the disabling of the couple of bank (the memory usually consists in a couple of bank, ok?): so if it is the only one couple you must provide a new one to make the system work, otherwise it can work as well but with less mem!
linuxabbestia
<<Unix IS user friendly... It's just selective about who its friends are!>>
Reply to: