On 19.03.2024 15:48, Andrew Cooper wrote: > Right now, if the user requests the watchdog on the command line, > setup_apic_nmi_watchdog() will blindly assume that setting up the watchdog > worked. Reuse nmi_perfctr_msr to identify when the watchdog has been set up. > > Rearrange setup_p6_watchdog() to not set nmi_perfctr_msr until the sanity > checks are performed. Turn setup_p4_watchdog() into a void function, matching > the others. > > If the watchdog isn't set up, inform the user and override to NMI_NONE, which > will prevent check_nmi_watchdog() from claiming that all CPUs are stuck. > > e.g.: > > (XEN) alt table ffff82d040697c38 -> ffff82d0406a97f0 > (XEN) Failed to configure NMI watchdog > (XEN) Brought up 512 CPUs > (XEN) Scheduling granularity: cpu, 1 CPU per sched-resource > > Signed-off-by: Andrew Cooper <andrew.coop...@citrix.com>
Reviewed-by: Jan Beulich <jbeul...@suse.com>
