mce_schedule_work();
}
-static void mce_report_event(struct pt_regs *regs)
-{
- if (regs->flags & (X86_VM_MASK|X86_EFLAGS_IF)) {
- mce_notify_irq();
- /*
- * Triggering the work queue here is just an insurance
- * policy in case the syscall exit notify handler
- * doesn't run soon enough or ends up running on the
- * wrong CPU (can happen when audit sleeps)
- */
- mce_schedule_work();
- return;
- }
-
- irq_work_queue(&mce_irq_work);
-}
-
/*
* Check if the address reported by the CPU is in a format we can parse.
* It would be possible to add code for most other cases, but all would
mce_panic("Fatal machine check on current CPU", &m, msg);
if (worst > 0)
- mce_report_event(regs);
+ irq_work_queue(&mce_irq_work);
+
mce_wrmsrl(MSR_IA32_MCG_STATUS, 0);
sync_core();