Re: [PATCH] x86/kvmclock: Stop kvmclocks for hibernate restore
From: Vitaly Kuznetsov <vkuznets@redhat.com>
Date: 2021-03-26 13:02:46
Also in:
lkml
Subsystem:
kvm paravirt (kvm/paravirt), the rest, x86 architecture (32-bit and 64-bit) · Maintainers:
Paolo Bonzini, Linus Torvalds, Thomas Gleixner, Ingo Molnar, Borislav Petkov, Dave Hansen
Vitaly Kuznetsov [off-list ref] writes: ..
(this is with your v2 included). There's nothing about CPU0 for e.g. async PF + timestamps are really interesting. Seems we have issues to fix) I'm playing with it right now.
What if we do the following (instead of your patch):
diff --git a/arch/x86/kernel/kvm.c b/arch/x86/kernel/kvm.c
index 78bb0fae3982..c32392d6329d 100644
--- a/arch/x86/kernel/kvm.c
+++ b/arch/x86/kernel/kvm.c@@ -26,6 +26,7 @@ #include <linux/kprobes.h> #include <linux/nmi.h> #include <linux/swait.h> +#include <linux/syscore_ops.h> #include <asm/timer.h> #include <asm/cpu.h> #include <asm/traps.h>
@@ -598,17 +599,21 @@ static void kvm_guest_cpu_offline(void) static int kvm_cpu_online(unsigned int cpu) { - local_irq_disable(); + unsigned long flags; + + local_irq_save(flags); kvm_guest_cpu_init(); - local_irq_enable(); + local_irq_restore(flags); return 0; } static int kvm_cpu_down_prepare(unsigned int cpu) { - local_irq_disable(); + unsigned long flags; + + local_irq_save(flags); kvm_guest_cpu_offline(); - local_irq_enable(); + local_irq_restore(flags); return 0; } #endif
@@ -639,6 +644,23 @@ static void kvm_flush_tlb_others(const struct cpumask *cpumask, native_flush_tlb_others(flushmask, info); } +static int kvm_suspend(void) +{ + kvm_guest_cpu_offline(); + + return 0; +} + +static void kvm_resume(void) +{ + kvm_cpu_online(raw_smp_processor_id()); +} + +static struct syscore_ops kvm_syscore_ops = { + .suspend = kvm_suspend, + .resume = kvm_resume, +}; + static void __init kvm_guest_init(void) { int i;
@@ -681,6 +703,8 @@ static void __init kvm_guest_init(void) kvm_guest_cpu_init(); #endif + register_syscore_ops(&kvm_syscore_ops); + /* * Hard lockup detection is enabled by default. Disable it, as guests * can get false positives too easily, for example if the host is
--
2.30.2
This seems to work fine (according to the log, I haven't checked yet
that PV features are actually working):
[ 20.678081] PM: hibernation: Creating image:
[ 20.689925] PM: hibernation: Need to copy 82048 pages
[ 2.302411] kvm-clock: cpu 0, msr 2c201001, primary cpu clock, resume
[ 2.302487] PM: Restoring platform NVS memory
[ 2.302498] kvm-guest: KVM setup async PF for cpu 0
[ 2.302502] kvm-guest: stealtime: cpu 0, msr 3ec2c080
[ 2.304754] Enabling non-boot CPUs ...
[ 2.304823] x86: Booting SMP configuration:
[ 2.304824] smpboot: Booting Node 0 Processor 1 APIC 0x1
[ 2.304952] kvm-clock: cpu 1, msr 2c201041, secondary cpu clock
[ 2.305400] kvm-guest: KVM setup async PF for cpu 1
[ 2.305405] kvm-guest: stealtime: cpu 1, msr 3ecac080
[ 2.305786] CPU1 is up
[ 2.305818] smpboot: Booting Node 0 Processor 2 APIC 0x2
[ 2.305920] kvm-clock: cpu 2, msr 2c201081, secondary cpu clock
[ 2.306325] kvm-guest: KVM setup async PF for cpu 2
[ 2.306330] kvm-guest: stealtime: cpu 2, msr 3ed2c080
[ 2.306599] CPU2 is up
[ 2.306627] smpboot: Booting Node 0 Processor 3 APIC 0x3
[ 2.306729] kvm-clock: cpu 3, msr 2c2010c1, secondary cpu clock
[ 2.307092] kvm-guest: KVM setup async PF for cpu 3
[ 2.307097] kvm-guest: stealtime: cpu 3, msr 3edac080
[ 2.307383] CPU3 is up
[ 2.307560] ACPI: Waking up from system sleep state S4
[ 2.318778] sd 0:0:0:0: [sda] Starting disk
[ 2.342335] OOM killer enabled.
[ 2.342817] Restarting tasks ... done.
[ 2.346209] PM: hibernation: hibernation exit
--
Vitaly