On Thu, Jan 25, 2024 at 09:05:13AM +0100, Jan Beulich wrote: > On 24.01.2024 16:55, Andrew Cooper wrote: > > On 24/01/2024 3:21 pm, Jan Beulich wrote: > > but if you felt > > turning this into using a scratch cpumask and %*pb then I think that > > would be nicer still. > > I expected you might say that, and indeed I considered it. But then > I wanted to keep code churn limited. The way it's done it's imo an > almost purely mechanical change. Much like with Roger's consideration > of further refining the message text in the error case.
One suggestion I had in mind (which is not explicitly related to the fix here) would be to taint the hypervisor when we hit such selftests errors, as that would make it more obvious to go search the dmesg for reported failures. I think it's still fairly easy to miss noticing the watchdog selftest failed on some CPUs. Regards, Roger.