Excerpts from Ganesh Goudar's message of June 20, 2020 10:44 pm: > When an UE or memory error exception is encountered the MCE handler > tries to find the pfn using addr_to_pfn() which takes effective > address as an argument, later pfn is used to poison the page where > memory error occurred, recent rework in this area made addr_to_pfn > to run in realmode, which can be fatal as it may try to access > memory outside RMO region. > > To fix this move the use of addr_to_pfn to save_mce_event(), which > runs in virtual mode.
I'd rather this be done in the pseries mce handler rather than change bare metal, and leaving save_mce_event as purely recording the event, decoding remains in its caller. Radix guests also don't have an RMO, we could probably restrict that pseries hack to hash. I want to get rid of it but it will take longer to fix things properly. Thanks, Nick