38 lines · plain
1What: /sys/devices/system/machinecheck/machinecheckX/tolerant2Contact: Borislav Petkov <bp@suse.de>3Date: Dec, 20214Description:5 Unused and obsolete after the advent of recoverable machine6 checks (see last sentence below) and those are present since7 2010 (Nehalem).8 9 Original description:10 11 The entries appear for each CPU, but they are truly shared12 between all CPUs.13 14 Tolerance level. When a machine check exception occurs for a15 non corrected machine check the kernel can take different16 actions.17 18 Since machine check exceptions can happen any time it is19 sometimes risky for the kernel to kill a process because it20 defies normal kernel locking rules. The tolerance level21 configures how hard the kernel tries to recover even at some22 risk of deadlock. Higher tolerant values trade potentially23 better uptime with the risk of a crash or even corruption24 (for tolerant >= 3).25 26 == ===========================================================27 0 always panic on uncorrected errors, log corrected errors28 1 panic or SIGBUS on uncorrected errors, log corrected errors29 2 SIGBUS or log uncorrected errors, log corrected errors30 3 never panic or SIGBUS, log all errors (for testing only)31 == ===========================================================32 33 Default: 134 35 Note this only makes a difference if the CPU allows recovery36 from a machine check exception. Current x86 CPUs generally37 do not.38