* [PATCH] KVM: x86: Check KVM_REQ_IMMEDIATE_EXIT after enable_irq_window
@ 2013-04-28 7:23 Jan Kiszka
2013-04-28 8:29 ` Gleb Natapov
0 siblings, 1 reply; 5+ messages in thread
From: Jan Kiszka @ 2013-04-28 7:23 UTC (permalink / raw)
To: Gleb Natapov, Marcelo Tosatti; +Cc: kvm, Nakajima, Jun, Ren, Yongjie
From: Jan Kiszka <jan.kiszka@siemens.com>
The VMX implementation of enable_irq_window will raise
KVM_REQ_IMMEDIATE_EXIT if an interrupt window shall be opened during
nested vmrun. We have to transfer this request into the local variable
to avoid looping over a continuously re-asserted request on vmentry.
This issue only affects nested VMX scenarios.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
---
arch/x86/kvm/x86.c | 3 +++
1 files changed, 3 insertions(+), 0 deletions(-)
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 2a434bf..57ff14a 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -5759,6 +5759,9 @@ static int vcpu_enter_guest(struct kvm_vcpu *vcpu)
else if (kvm_cpu_has_injectable_intr(vcpu) || req_int_win)
kvm_x86_ops->enable_irq_window(vcpu);
+ req_immediate_exit =
+ kvm_check_request(KVM_REQ_IMMEDIATE_EXIT, vcpu);
+
if (kvm_lapic_enabled(vcpu)) {
/*
* Update architecture specific hints for APIC
--
1.7.3.4
^ permalink raw reply related [flat|nested] 5+ messages in thread
* Re: [PATCH] KVM: x86: Check KVM_REQ_IMMEDIATE_EXIT after enable_irq_window
2013-04-28 7:23 [PATCH] KVM: x86: Check KVM_REQ_IMMEDIATE_EXIT after enable_irq_window Jan Kiszka
@ 2013-04-28 8:29 ` Gleb Natapov
2013-04-28 8:50 ` [PATCH] KVM: x86: Rework request for immediate exit Jan Kiszka
0 siblings, 1 reply; 5+ messages in thread
From: Gleb Natapov @ 2013-04-28 8:29 UTC (permalink / raw)
To: Jan Kiszka; +Cc: Marcelo Tosatti, kvm, Nakajima, Jun, Ren, Yongjie
On Sun, Apr 28, 2013 at 09:23:58AM +0200, Jan Kiszka wrote:
> From: Jan Kiszka <jan.kiszka@siemens.com>
>
> The VMX implementation of enable_irq_window will raise
> KVM_REQ_IMMEDIATE_EXIT if an interrupt window shall be opened during
> nested vmrun. We have to transfer this request into the local variable
> to avoid looping over a continuously re-asserted request on vmentry.
>
> This issue only affects nested VMX scenarios.
>
> Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
So KVM_REQ_IMMEDIATE_EXIT does not behave like request bit should. What
about dropping it and letting enable_irq_window() return true if
immediate exit is needed.
> ---
> arch/x86/kvm/x86.c | 3 +++
> 1 files changed, 3 insertions(+), 0 deletions(-)
>
> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> index 2a434bf..57ff14a 100644
> --- a/arch/x86/kvm/x86.c
> +++ b/arch/x86/kvm/x86.c
> @@ -5759,6 +5759,9 @@ static int vcpu_enter_guest(struct kvm_vcpu *vcpu)
> else if (kvm_cpu_has_injectable_intr(vcpu) || req_int_win)
> kvm_x86_ops->enable_irq_window(vcpu);
>
> + req_immediate_exit =
> + kvm_check_request(KVM_REQ_IMMEDIATE_EXIT, vcpu);
> +
The code will become:
req_immediate_exit = kvm_x86_ops->enable_irq_window(vcpu);
> if (kvm_lapic_enabled(vcpu)) {
> /*
> * Update architecture specific hints for APIC
> --
> 1.7.3.4
--
Gleb.
^ permalink raw reply [flat|nested] 5+ messages in thread
* [PATCH] KVM: x86: Rework request for immediate exit
2013-04-28 8:29 ` Gleb Natapov
@ 2013-04-28 8:50 ` Jan Kiszka
2013-05-03 1:20 ` Marcelo Tosatti
0 siblings, 1 reply; 5+ messages in thread
From: Jan Kiszka @ 2013-04-28 8:50 UTC (permalink / raw)
To: Gleb Natapov, Marcelo Tosatti; +Cc: kvm, Nakajima, Jun, Ren, Yongjie
From: Jan Kiszka <jan.kiszka@siemens.com>
The VMX implementation of enable_irq_window raised
KVM_REQ_IMMEDIATE_EXIT after we checked it in vcpu_enter_guest. This
caused infinite loops on vmentry. Fix it by letting enable_irq_window
signal the need for an immediate exit via its return value and drop
KVM_REQ_IMMEDIATE_EXIT.
This issue only affects nested VMX scenarios.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
---
arch/x86/include/asm/kvm_host.h | 2 +-
arch/x86/kvm/svm.c | 3 ++-
arch/x86/kvm/vmx.c | 15 ++++++++-------
arch/x86/kvm/x86.c | 7 +++----
include/linux/kvm_host.h | 15 +++++++--------
5 files changed, 21 insertions(+), 21 deletions(-)
diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
index 18635ae..111b4a0 100644
--- a/arch/x86/include/asm/kvm_host.h
+++ b/arch/x86/include/asm/kvm_host.h
@@ -694,7 +694,7 @@ struct kvm_x86_ops {
bool (*get_nmi_mask)(struct kvm_vcpu *vcpu);
void (*set_nmi_mask)(struct kvm_vcpu *vcpu, bool masked);
void (*enable_nmi_window)(struct kvm_vcpu *vcpu);
- void (*enable_irq_window)(struct kvm_vcpu *vcpu);
+ int (*enable_irq_window)(struct kvm_vcpu *vcpu);
void (*update_cr8_intercept)(struct kvm_vcpu *vcpu, int tpr, int irr);
int (*vm_has_apicv)(struct kvm *kvm);
void (*hwapic_irr_update)(struct kvm_vcpu *vcpu, int max_irr);
diff --git a/arch/x86/kvm/svm.c b/arch/x86/kvm/svm.c
index d6713e1..91a7431 100644
--- a/arch/x86/kvm/svm.c
+++ b/arch/x86/kvm/svm.c
@@ -3632,7 +3632,7 @@ static int svm_interrupt_allowed(struct kvm_vcpu *vcpu)
return ret;
}
-static void enable_irq_window(struct kvm_vcpu *vcpu)
+static int enable_irq_window(struct kvm_vcpu *vcpu)
{
struct vcpu_svm *svm = to_svm(vcpu);
@@ -3646,6 +3646,7 @@ static void enable_irq_window(struct kvm_vcpu *vcpu)
svm_set_vintr(svm);
svm_inject_irq(svm, 0x0);
}
+ return 0;
}
static void enable_nmi_window(struct kvm_vcpu *vcpu)
diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
index d663a59..1795512 100644
--- a/arch/x86/kvm/vmx.c
+++ b/arch/x86/kvm/vmx.c
@@ -4398,22 +4398,23 @@ static bool nested_exit_on_nmi(struct kvm_vcpu *vcpu)
PIN_BASED_NMI_EXITING;
}
-static void enable_irq_window(struct kvm_vcpu *vcpu)
+static int enable_irq_window(struct kvm_vcpu *vcpu)
{
u32 cpu_based_vm_exec_control;
- if (is_guest_mode(vcpu) && nested_exit_on_intr(vcpu)) {
+
+ if (is_guest_mode(vcpu) && nested_exit_on_intr(vcpu))
/*
* We get here if vmx_interrupt_allowed() said we can't
- * inject to L1 now because L2 must run. Ask L2 to exit
- * right after entry, so we can inject to L1 more promptly.
+ * inject to L1 now because L2 must run. The caller will have
+ * to make L2 exit right after entry, so we can inject to L1
+ * more promptly.
*/
- kvm_make_request(KVM_REQ_IMMEDIATE_EXIT, vcpu);
- return;
- }
+ return -EBUSY;
cpu_based_vm_exec_control = vmcs_read32(CPU_BASED_VM_EXEC_CONTROL);
cpu_based_vm_exec_control |= CPU_BASED_VIRTUAL_INTR_PENDING;
vmcs_write32(CPU_BASED_VM_EXEC_CONTROL, cpu_based_vm_exec_control);
+ return 0;
}
static void enable_nmi_window(struct kvm_vcpu *vcpu)
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 2a434bf..c522260 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -5692,7 +5692,7 @@ static int vcpu_enter_guest(struct kvm_vcpu *vcpu)
int r;
bool req_int_win = !irqchip_in_kernel(vcpu->kvm) &&
vcpu->run->request_interrupt_window;
- bool req_immediate_exit = 0;
+ bool req_immediate_exit = false;
if (vcpu->requests) {
if (kvm_check_request(KVM_REQ_MMU_RELOAD, vcpu))
@@ -5734,8 +5734,6 @@ static int vcpu_enter_guest(struct kvm_vcpu *vcpu)
record_steal_time(vcpu);
if (kvm_check_request(KVM_REQ_NMI, vcpu))
process_nmi(vcpu);
- req_immediate_exit =
- kvm_check_request(KVM_REQ_IMMEDIATE_EXIT, vcpu);
if (kvm_check_request(KVM_REQ_PMU, vcpu))
kvm_handle_pmu_event(vcpu);
if (kvm_check_request(KVM_REQ_PMI, vcpu))
@@ -5757,7 +5755,8 @@ static int vcpu_enter_guest(struct kvm_vcpu *vcpu)
if (vcpu->arch.nmi_pending)
kvm_x86_ops->enable_nmi_window(vcpu);
else if (kvm_cpu_has_injectable_intr(vcpu) || req_int_win)
- kvm_x86_ops->enable_irq_window(vcpu);
+ req_immediate_exit =
+ kvm_x86_ops->enable_irq_window(vcpu) != 0;
if (kvm_lapic_enabled(vcpu)) {
/*
diff --git a/include/linux/kvm_host.h b/include/linux/kvm_host.h
index 93a5005..7bde424 100644
--- a/include/linux/kvm_host.h
+++ b/include/linux/kvm_host.h
@@ -119,14 +119,13 @@ static inline bool is_error_page(struct page *page)
#define KVM_REQ_APF_HALT 12
#define KVM_REQ_STEAL_UPDATE 13
#define KVM_REQ_NMI 14
-#define KVM_REQ_IMMEDIATE_EXIT 15
-#define KVM_REQ_PMU 16
-#define KVM_REQ_PMI 17
-#define KVM_REQ_WATCHDOG 18
-#define KVM_REQ_MASTERCLOCK_UPDATE 19
-#define KVM_REQ_MCLOCK_INPROGRESS 20
-#define KVM_REQ_EPR_EXIT 21
-#define KVM_REQ_SCAN_IOAPIC 22
+#define KVM_REQ_PMU 15
+#define KVM_REQ_PMI 16
+#define KVM_REQ_WATCHDOG 17
+#define KVM_REQ_MASTERCLOCK_UPDATE 18
+#define KVM_REQ_MCLOCK_INPROGRESS 19
+#define KVM_REQ_EPR_EXIT 20
+#define KVM_REQ_SCAN_IOAPIC 21
#define KVM_USERSPACE_IRQ_SOURCE_ID 0
#define KVM_IRQFD_RESAMPLE_IRQ_SOURCE_ID 1
--
1.7.3.4
^ permalink raw reply related [flat|nested] 5+ messages in thread
* Re: [PATCH] KVM: x86: Rework request for immediate exit
2013-04-28 8:50 ` [PATCH] KVM: x86: Rework request for immediate exit Jan Kiszka
@ 2013-05-03 1:20 ` Marcelo Tosatti
2013-05-03 5:33 ` Gleb Natapov
0 siblings, 1 reply; 5+ messages in thread
From: Marcelo Tosatti @ 2013-05-03 1:20 UTC (permalink / raw)
To: Jan Kiszka; +Cc: Gleb Natapov, kvm, Nakajima, Jun, Ren, Yongjie
On Sun, Apr 28, 2013 at 10:50:52AM +0200, Jan Kiszka wrote:
> From: Jan Kiszka <jan.kiszka@siemens.com>
>
> The VMX implementation of enable_irq_window raised
> KVM_REQ_IMMEDIATE_EXIT after we checked it in vcpu_enter_guest. This
> caused infinite loops on vmentry. Fix it by letting enable_irq_window
> signal the need for an immediate exit via its return value and drop
> KVM_REQ_IMMEDIATE_EXIT.
>
> This issue only affects nested VMX scenarios.
>
> Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Please rebase against "KVM: x86: Account for failing enable_irq_window
for NMI window request".
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [PATCH] KVM: x86: Rework request for immediate exit
2013-05-03 1:20 ` Marcelo Tosatti
@ 2013-05-03 5:33 ` Gleb Natapov
0 siblings, 0 replies; 5+ messages in thread
From: Gleb Natapov @ 2013-05-03 5:33 UTC (permalink / raw)
To: Marcelo Tosatti; +Cc: Jan Kiszka, kvm, Nakajima, Jun, Ren, Yongjie
On Thu, May 02, 2013 at 10:20:39PM -0300, Marcelo Tosatti wrote:
> On Sun, Apr 28, 2013 at 10:50:52AM +0200, Jan Kiszka wrote:
> > From: Jan Kiszka <jan.kiszka@siemens.com>
> >
> > The VMX implementation of enable_irq_window raised
> > KVM_REQ_IMMEDIATE_EXIT after we checked it in vcpu_enter_guest. This
> > caused infinite loops on vmentry. Fix it by letting enable_irq_window
> > signal the need for an immediate exit via its return value and drop
> > KVM_REQ_IMMEDIATE_EXIT.
> >
> > This issue only affects nested VMX scenarios.
> >
> > Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
>
> Please rebase against "KVM: x86: Account for failing enable_irq_window
> for NMI window request".
This one is applied already. Looks like I forgot to send an email about
it.
--
Gleb.
^ permalink raw reply [flat|nested] 5+ messages in thread
end of thread, other threads:[~2013-05-03 5:33 UTC | newest]
Thread overview: 5+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2013-04-28 7:23 [PATCH] KVM: x86: Check KVM_REQ_IMMEDIATE_EXIT after enable_irq_window Jan Kiszka
2013-04-28 8:29 ` Gleb Natapov
2013-04-28 8:50 ` [PATCH] KVM: x86: Rework request for immediate exit Jan Kiszka
2013-05-03 1:20 ` Marcelo Tosatti
2013-05-03 5:33 ` Gleb Natapov
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox