The SSH configuration is at least a distribution-level thing

(unlike the PAM situation)

am I correct in assmuing the nvme driver currently does not post fault events?

(i.e. I didn't miss it somehow reading the code)

I see the comment "FMA is then called to fault the device"

in "Error Handling", in the top

the only thing I see is pci_ereport_post, but a bunch of TODO: post ereport comments

so I'm thinking like a bad read is probably not going to generate an ereport unless it triggers something lower level..

probably needs hans or robert to know more