https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=288607
John Baldwin <[email protected]> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |[email protected] Resolution|--- |Not A Bug Status|New |Closed --- Comment #3 from John Baldwin <[email protected]> --- The MCA bank doesn't mean "RAM" or "DIMM" bank. It indicates which register is reporting the error (MCA uses different banks of registers to report different types of errors, and the mapping of errors to banks is vendor-specific and also uarch-specific). So dmidecode isn't useful here, mcelog from ports (which should support AMD) is the only tool that can provide more details than is in dmesg, but often it doesn't provide any more detail. The bits decoded in dmesg do indicate an uncorrected checksum error in your CPU's instruction-cache. That would generally be a hardware failure. Software can't really trigger ECC errors inside the CPU die. -- You are receiving this mail because: You are the assignee for the bug.
