16.05.24 12:46, Jan Beulich:
--- a/xen/arch/x86/cpu/mcheck/non-fatal.c
+++ b/xen/arch/x86/cpu/mcheck/non-fatal.c
@@ -29,9 +29,13 @@ static int __init cf_check init_nonfatal_mce_checker(void)
/* Assume we are on K8 or newer AMD or Hygon CPU here */
amd_nonfatal_mcheck_init(c);
break;
+
case X86_VENDOR_INTEL:
intel_nonfatal_mcheck_init(c);
break;
+
+ default:
+ return -ENODEV;
}
printk(KERN_INFO "mcheck_poll: Machine check polling timer started.\n");
return 0;
... converting this to an error return still isn't justified. On one hand it's
benign because we still don't check return values from initcalls. Yet otoh it
isn't really an error, is it?
I realize earlier in the function there are error returns, too. I question at
least the former of them as well. And the latter shouldn't be an error at least
when the vendor isn't going to be handled in the switch().
I'll just return 0 then, and maybe leave a comment, as this code will
start to look a bit unnatural to me.
-Sergiy
