On 06/13/2018 05:58 PM, Ricardo Neri wrote:

On Tue, Jun 12, 2018 at 10:26:57PM -0700, Randy Dunlap wrote:

quoted

On 06/12/2018 05:57 PM, Ricardo Neri wrote:

quoted

diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
index f2040d4..a8833c7 100644
--- a/Documentation/admin-guide/kernel-parameters.txt
+++ b/Documentation/admin-guide/kernel-parameters.txt

@@ -2577,7 +2577,7 @@
 			Format: [state][,regs][,debounce][,die]
 
 	nmi_watchdog=	[KNL,BUGS=X86] Debugging features for SMP kernels
-			Format: [panic,][nopanic,][num]
+			Format: [panic,][nopanic,][num,][hpet]
 			Valid num: 0 or 1
 			0 - turn hardlockup detector in nmi_watchdog off
 			1 - turn hardlockup detector in nmi_watchdog on

This says that I can use "nmi_watchdog=hpet" without using 0 or 1.
Is that correct?

Yes, this what I meant. In my view, if you set nmi_watchdog=hpet it
implies that you want to activate the NMI watchdog. In this case, perf.

I can see how this will be ambiguous for the case of perf and arch NMI
watchdogs.

Alternative, a new parameter could be added; such as nmi_watchdog_type. I
didn't want to add it in this patchset as I think that a single parameter
can handle the enablement and type of the NMI watchdog.

What do you think?

I think it's OK like it is.

thanks,
-- 
~Randy