From f2f14588b38eb16ec6cdee7eb0cc82c8114f1594 Mon Sep 17 00:00:00 2001 From: Lai Jiangshan Date: Wed, 19 Oct 2011 12:31:14 +0200 Subject: [PATCH] qemu-kvm: fix improper nmi emulation Bugzilla: 738565 Upstream: http://article.gmane.org/gmane.comp.emulators.kvm.devel/80339, reviewed by Jan and myself Currently, NMI interrupt is blindly sent to all the vCPUs when NMI button event happens. This doesn't properly emulate real hardware on which NMI button event triggers LINT1. Because of this, NMI is sent to the processor even when LINT1 is maskied in LVT. For example, this causes the problem that kdump initiated by NMI sometimes doesn't work on KVM, because kdump assumes NMI is masked on CPUs other than CPU0. With this patch, inject-nmi request is handled as follows. - When in-kernel irqchip is disabled, deliver LINT1 instead of NMI interrupt. - When in-kernel irqchip is enabled, get the in-kernel LAPIC states and test the APIC_LVT_MASKED, if LINT1 is unmasked, and then delivering the NMI directly. (Suggested by Jan Kiszka) Changes from upstream: - use APICState instead of DeviceState - move apic/noapic decision to qemu-kvm.c Testing: The nmi command now hits just cpu 0, as shown in /proc/interrupts. Signed-off-by: Lai Jiangshan Reported-by: Kenji Kaneshige Signed-off-by: Avi Kivity Signed-off-by: Michal Novotny --- hw/apic.c | 31 +++++++++++++++++++++++++++++++ hw/pc.h | 1 + monitor.c | 6 +++++- qemu-kvm.c | 9 ++++++++- 4 files changed, 45 insertions(+), 2 deletions(-) diff --git a/hw/apic.c b/hw/apic.c index 383eeae..7546ea6 100644 --- a/hw/apic.c +++ b/hw/apic.c @@ -208,6 +208,37 @@ void apic_deliver_pic_intr(CPUState *env, int level) } } +static inline uint32_t kapic_reg(struct kvm_lapic_state *kapic, int reg_id); + +static void kvm_irqchip_deliver_nmi(void *p) +{ + APICState *s = p; + struct kvm_lapic_state klapic; + uint32_t lvt; + + kvm_get_lapic(s->cpu_env, &klapic); + lvt = kapic_reg(&klapic, 0x32 + APIC_LVT_LINT1); + + if (lvt & APIC_LVT_MASKED) { + return; + } + + if (((lvt >> 8) & 7) != APIC_DM_NMI) { + return; + } + + kvm_vcpu_ioctl(s->cpu_env, KVM_NMI); +} + +void apic_deliver_nmi(APICState *s) +{ + if (kvm_irqchip_in_kernel()) { + kvm_irqchip_deliver_nmi(s); + } else { + apic_local_deliver(s->cpu_env, APIC_LVT_LINT1); + } +} + #define foreach_apic(apic, deliver_bitmask, code) \ {\ int __i, __j, __mask;\ diff --git a/hw/pc.h b/hw/pc.h index 8547516..916a595 100644 --- a/hw/pc.h +++ b/hw/pc.h @@ -44,6 +44,7 @@ void apic_deliver_irq(uint8_t dest, uint8_t dest_mode, int apic_init(CPUState *env); int apic_accept_pic_intr(CPUState *env); void apic_deliver_pic_intr(CPUState *env, int level); +void apic_deliver_nmi(struct APICState *d); int apic_get_interrupt(CPUState *env); qemu_irq *ioapic_init(void); void ioapic_set_irq(void *opaque, int vector, int level); diff --git a/monitor.c b/monitor.c index f5ee250..3273914 100644 --- a/monitor.c +++ b/monitor.c @@ -2380,7 +2380,11 @@ static int do_inject_nmi(Monitor *mon, const QDict *qdict, QObject **ret_data) if (kvm_enabled()) { kvm_inject_interrupt(env, CPU_INTERRUPT_NMI); } else { - cpu_interrupt(env, CPU_INTERRUPT_NMI); + if (!env->apic_state) { + cpu_interrupt(env, CPU_INTERRUPT_NMI); + } else { + apic_deliver_nmi(env->apic_state); + } } } diff --git a/qemu-kvm.c b/qemu-kvm.c index 34e9916..a7ab2c0 100644 --- a/qemu-kvm.c +++ b/qemu-kvm.c @@ -66,6 +66,9 @@ pthread_cond_t qemu_pause_cond = PTHREAD_COND_INITIALIZER; pthread_cond_t qemu_work_cond = PTHREAD_COND_INITIALIZER; __thread CPUState *current_env; +/* minovotn: Copied from hw/pc.h since file excluded because of conflicts */ +void apic_deliver_nmi(struct APICState *d); + static int qemu_system_ready; #define SIG_IPI (SIGRTMIN+4) @@ -1630,7 +1633,11 @@ void kvm_cpu_synchronize_state(CPUState *env) static void inject_interrupt(void *data) { - cpu_interrupt(current_env, (long) data); + if (!current_env->apic_state || (long)data != CPU_INTERRUPT_NMI) { + cpu_interrupt(current_env, (long) data); + } else { + apic_deliver_nmi(current_env->apic_state); + } } void kvm_inject_interrupt(CPUState *env, int mask) -- 1.7.4.4