KVM: VMX: Add Posted Interrupt supporting
v3 cover KVM: VMX: Add Posted Interrupt supporting
From: Yang Zhang <[email protected]>
The two patches are adding the Posted Interrupt supporting to KVM:
The first patch enables the feature 'acknowledge interrupt on vmexit'.Since it is required by Posted interrupt, we need to enable it firstly.
And the second patch is adding the posted interrupt supporting.
Please see the comments in the two patch to get more details.
Changes from v2 to v3:
* Add CONFIG_HAVE_KVM check when calling BUILD_INTERRUPT for posted interrupt.
* Rename send_notification_event() to deliver_posted_interrupt().
* Modify the algorithm of calculating interrupt coalescence: interrupt will be considered as delivered only when there is no previous interrupt pending in both irr and pir.
* Remove using new request to sync pir to irr, use KVM_REQ_EVENT for instead.
* Remove the optimization of checking outsanding notification bit after sending posted interrupt to remote cpu.
* Embed pi_desc inside struct vmx instead allocating it dynamically.
* Rebase on top of KVM upstream.
Yang Zhang (2):
KVM: VMX: enable acknowledge interupt on vmexit
KVM: VMX: Add Posted Interrupt supporting
arch/x86/include/asm/entry_arch.h | 4 + arch/x86/include/asm/hw_irq.h | 1 + arch/x86/include/asm/irq_vectors.h | 5 + arch/x86/include/asm/kvm_host.h | 4 + arch/x86/include/asm/vmx.h | 4 + arch/x86/kernel/entry_64.S | 5 + arch/x86/kernel/irq.c | 20 +++ arch/x86/kernel/irqinit.c | 4 + arch/x86/kvm/lapic.c | 19 +++- arch/x86/kvm/lapic.h | 1 + arch/x86/kvm/svm.c | 19 +++ arch/x86/kvm/vmx.c | 226 ++++++++++++++++++++++++++++++++---- arch/x86/kvm/x86.c | 5 +- 13 files changed, 289 insertions(+), 28 deletions(-)
PATCH v3 1/2 KVM: VMX: enable acknowledge interupt on vmexit
From: Yang Zhang <[email protected]>
The "acknowledge interrupt on exit" feature controls processor behavior for external interrupt acknowledgement. When this control is set, the processor acknowledges the interrupt controller to acquire the interrupt vector on VM exit.
After enabling this feature, an interrupt which arrived when target cpu is running in vmx non-root mode will be handled by vmx handler instead of handler in idt. Currently, vmx handler only fakes an interrupt stack and jump to idt table to let real handler to handle it. Further, we will recognize the interrupt and only delivery the interrupt which not belong to current vcpu through idt table. The interrupt which belonged to current vcpu will be handled inside vmx handler. This will reduce the interrupt handle cost of KVM.
Also, interrupt enable logic is changed if this feature is turnning on:
Before this patch, hypervior call local_irq_enable() to enable it directly. Now IF bit is set on interrupt stack frame, and will be enabled on a return from interrupt handler if exterrupt interrupt exists. If no external interrupt, still call local_irq_enable() to enable it.
Refer to Intel SDM volum 3, chapter 33.2.
--- arch/x86/include/asm/kvm_host.h | 1 + arch/x86/kvm/svm.c | 6 +++ arch/x86/kvm/vmx.c | 69 ++++++++++++++++++++++++++++++++++++-- arch/x86/kvm/x86.c | 4 ++- 4 files changed, 75 insertions(+), 5 deletions(-)
arch/x86/include/asm/kvm_host.h +1 -0
@@ -730,6 +730,7 @@ struct kvm_x86_ops {int (*check_intercept)(struct kvm_vcpu *vcpu,struct x86_instruction_info *info,enum x86_intercept_stage stage); +void (*handle_external_intr)(struct kvm_vcpu *vcpu); };struct kvm_arch_async_pf {
arch/x86/kvm/svm.c +6 -0
@@ -4247,6 +4247,11 @@ out:return ret; } +static void svm_handle_external_intr (struct kvm_vcpu *vcpu) +{ +local_irq_enable (); +} +static struct kvm_x86_ops svm_x86_ops = { .cpu_has_kvm_support = has_svm, .disabled_by_bios = is_disabled,
@@ -4342,6 +4347,7 @@ static struct kvm_x86_ops svm_x86_ops = { .set_tdp_cr3 = set_tdp_cr3, .check_intercept = svm_check_intercept, + .handle_external_intr = svm_handle_external_intr, };static int __init svm_init (void )
arch/x86/kvm/vmx.c +65 -4
117 lines not on this page · raw message
arch/x86/kvm/x86.c +3 -1
@@ -5784,7 +5784,9 @@ static int vcpu_enter_guest(struct kvm_vcpu *vcpu) vcpu->mode =OUTSIDE_GUEST_MODE ;smp_wmb (); -local_irq_enable (); + +/* Interrupt is enabled by handle_external_intr() */ + kvm_x86_ops->handle_external_intr (vcpu); ++vcpu->stat.exits;
Signature · 1 line
1.7.1
Re: [PATCH v3 1/2] KVM: VMX: enable acknowledge interupt on vmexit
Quoted 47 lines · Yang Zhang
From: Yang Zhang <[email protected]>
The "acknowledge interrupt on exit" feature controls processor behavior for external interrupt acknowledgement. When this control is set, the processor acknowledges the interrupt controller to acquire the interrupt vector on VM exit.
After enabling this feature, an interrupt which arrived when target cpu is running in vmx non-root mode will be handled by vmx handler instead of handler in idt. Currently, vmx handler only fakes an interrupt stack and jump to idt table to let real handler to handle it. Further, we will recognize the interrupt and only delivery the interrupt which not belong to current vcpu through idt table. The interrupt which belonged to current vcpu will be handled inside vmx handler. This will reduce the interrupt handle cost of KVM.
Also, interrupt enable logic is changed if this feature is turnning on:
Before this patch, hypervior calllocal_irq_enable()to enable it directly. Now IF bit is set on interrupt stack frame, and will be enabled on a return from interrupt handler if exterrupt interrupt exists. If no external interrupt, still calllocal_irq_enable()to enable it.Refer to Intel SDM volum 3, chapter 33.2.
arch/x86/kvm/vmx.c +23
-0+static void vmx_handle_external_intr (struct kvm_vcpu *vcpu) +{ +u32 exit_intr_info =vmcs_read32 (VM_EXIT_INTR_INFO ); + +/* +* If external interrupt exists, IF bit is set in rflags/eflags on the +* interrupt stack frame, and interrupt will be enabled on a return +* from interrupt handler. +*/ +if ((exit_intr_info & (INTR_INFO_VALID_MASK |INTR_INFO_INTR_TYPE_MASK )) + == (INTR_INFO_VALID_MASK |INTR_TYPE_EXT_INTR )) { +unsigned int vector; +unsigned long entry; +gate_desc *desc; +struct vcpu_vmx *vmx =to_vmx (vcpu); + + vector = exit_intr_info &INTR_INFO_VECTOR_MASK ; +#ifdef CONFIG_X86_64 + desc = (void *)vmx->host_idt_base + vector *16 ; +#else + desc = (void *)vmx->host_idt_base + vector *8 ; +#endif +
arch/x86/kvm/vmx.c static void vmx_complete_atomic_exit(struct vcpu_vmx *vmx) · line 6384
Quoted 10 lines
arch/x86/kvm/vmx.c +10
-0+ entry =gate_offset (*desc); + asm( + "mov %0, %%" _ASM_DX " \n\t" +#ifdef CONFIG_X86_64 + "mov %%" _ASM_SP ", %%" _ASM_BX " \n\t" + "and $0xfffffffffffffff0, %%" _ASM_SP " \n\t" + "mov %%ss, %%" _ASM_AX " \n\t" + "push %%" _ASM_AX " \n\t" + "push %%" _ASM_BX " \n\t" +#endif
Are we sure no interrupts are using the IST feature? I guess it's unlikely.
arch/x86/kvm/vmx.c static void vmx_complete_atomic_exit(struct vcpu_vmx *vmx) · line 6388
Quoted 4 lines
arch/x86/kvm/vmx.c +4
-0+"pushf \n\t" +"pop %%" _ASM_AX" \n\t" +"or $0x200, %%" _ASM_AX" \n\t" +"push %%" _ASM_AX" \n\t"
Can simplify to pushf; orl $0x200, %%rsp.
arch/x86/kvm/vmx.c static void vmx_complete_atomic_exit(struct vcpu_vmx *vmx) · line 6390
Quoted 2 lines
arch/x86/kvm/vmx.c +2
-0+"mov %%cs, %%" _ASM_AX" \n\t" +"push %%" _ASM_AX" \n\t"
push %%cs
Quoted 1 line
+ "push
intr_return\n\t"
push $1f. Or even combine with the next instruction, and call %rdx.
Quoted 17 lines
arch/x86/kvm/vmx.c +17
-0+ "jmp *%% " _ASM_DX " \n\t" + "1: \n\t" + ".pushsection .rodata \n\t" + ".global intr_return \n\t" + "intr_return: " _ASM_PTR " 1b \n\t" + ".popsection \n\t" + : : "m"(entry) : +#ifdef CONFIG_X86_64 +"rax" ,"rbx" ,"rdx" +#else +"eax" ,"edx" +#endif + ); + }else +local_irq_enable (); +} +
RE: [PATCH v3 1/2] KVM: VMX: enable acknowledge interupt on vmexit
Avi Kivity wrote on 2013-02-20:
Quoted 40 lines · Yang Zhang
Quoted 39 lines
From: Yang Zhang <[email protected]>
The "acknowledge interrupt on exit" feature controls processor behavior for external interrupt acknowledgement. When this control is set, the processor acknowledges the interrupt controller to acquire the interrupt vector on VM exit.
After enabling this feature, an interrupt which arrived when target cpu is running in vmx non-root mode will be handled by vmx handler instead of handler in idt. Currently, vmx handler only fakes an interrupt stack and jump to idt table to let real handler to handle it. Further, we will recognize the interrupt and only delivery the interrupt which not belong to current vcpu through idt table. The interrupt which belonged to current vcpu will be handled inside vmx handler. This will reduce the interrupt handle cost of KVM.
Also, interrupt enable logic is changed if this feature is turnning on:
Before this patch, hypervior calllocal_irq_enable()to enable it directly. Now IF bit is set on interrupt stack frame, and will be enabled on a return from interrupt handler if exterrupt interrupt exists. If no external interrupt, still calllocal_irq_enable()to enable it.Refer to Intel SDM volum 3, chapter 33.2.
+static void
vmx_handle_external_intr(struct kvm_vcpu *vcpu)+{ +
u32exit_intr_info=vmcs_read32(VM_EXIT_INTR_INFO); + + /* +
* If external interrupt exists, IF bit is set in rflags/eflags on
the + * interrupt stack frame, and interrupt will be enabled on
a return + * from interrupt handler. + */ + if
((exit_intr_info& (INTR_INFO_VALID_MASK|INTR_INFO_INTR_TYPE_MASK)) +== (INTR_INFO_VALID_MASK | INTR_TYPE_EXT_INTR)) {+ unsigned int vector; + unsigned long
entry; +gate_desc*desc; + structvcpu_vmx
*vmx =to_vmx(vcpu); + + vector =exit_intr_info&INTR_INFO_VECTOR_MASK; +#ifdefCONFIG_X86_64+ desc =
(void *)vmx->host_idt_base + vector * 16; +#else + desc =
(void *)vmx->host_idt_base + vector * 8; +#endif + +
entry =gate_offset(*desc); + asm( +
Quoted 8 lines
Quoted 6 lines
"mov %0, %%"
_ASM_DX" \n\t" +#ifdefCONFIG_X86_64+"mov %%" _ASM_SP ", %%" _ASM_BX " \n\t" +"and $0xfffffffffffffff0, %%"
_ASM_SP" \n\t" +
"mov %%ss, %%"_ASM_AX" \n\t" + "push %%"_ASM_AX" \n\t" + "push %%"_ASM_BX" \n\t"
+#endifAre we sure no interrupts are using the IST feature? I guess it's unlikely.
Linux uses IST for NMI, stack fault, machine-check, double fault and debug interrupt . No external interrupt will use it. This feature is only for external interrupt. So we don't need to check it here.
Quoted 7 lines
arch/x86/kvm/vmx.c static void vmx_complete_atomic_exit(struct vcpu_vmx *vmx) · line 6388
Quoted 5 lines
arch/x86/kvm/vmx.c +4
-0+"pushf \n\t" +"pop %%" _ASM_AX" \n\t" +"or $0x200, %%" _ASM_AX" \n\t" +"push %%" _ASM_AX" \n\t" Can simplify to pushf; orl $0x200, %%rsp.
Sure.
Quoted 4 lines
arch/x86/kvm/vmx.c static void vmx_complete_atomic_exit(struct vcpu_vmx *vmx) · line 6390
Quoted 2 lines
arch/x86/kvm/vmx.c +2
-0+"mov %%cs, %%" _ASM_AX" \n\t" +"push %%" _ASM_AX" \n\t" push %%cs
"push %%cs" is invalid in x86_64.
Quoted 3 lines
Quoted 1 line
+ "push
intr_return\n\t"push $1f. Or even combine with the next instruction, and call %rdx.
Which is faster? jmp or call?
Quoted 9 lines
Quoted 9 lines
arch/x86/kvm/vmx.c +9
-0+ "jmp *%% " _ASM_DX " \n\t" + "1: \n\t" + ".pushsection .rodata \n\t" + ".global intr_return \n\t" + "intr_return: " _ASM_PTR " 1b \n\t" + ".popsection \n\t" + : : "m"(entry) : +#ifdef CONFIG_X86_64 +"rax" ,"rbx" ,"rdx"
Quoted 8 lines
arch/x86/kvm/vmx.c static void vmx_complete_atomic_exit(struct vcpu_vmx *vmx) · line 6408
Quoted 8 lines
arch/x86/kvm/vmx.c +8
-0+#else +"eax" ,"edx" +#endif + ); + }else +local_irq_enable (); +} +
Best regards,
Yang
Re: [PATCH v3 1/2] KVM: VMX: enable acknowledge interupt on vmexit
Quoted 64 lines · Yang Z
Avi Kivity wrote on 2013-02-20:
Quoted 40 lines · Yang Zhang
Quoted 39 lines
From: Yang Zhang <[email protected]>
The "acknowledge interrupt on exit" feature controls processor behavior for external interrupt acknowledgement. When this control is set, the processor acknowledges the interrupt controller to acquire the interrupt vector on VM exit.
After enabling this feature, an interrupt which arrived when target cpu is running in vmx non-root mode will be handled by vmx handler instead of handler in idt. Currently, vmx handler only fakes an interrupt stack and jump to idt table to let real handler to handle it. Further, we will recognize the interrupt and only delivery the interrupt which not belong to current vcpu through idt table. The interrupt which belonged to current vcpu will be handled inside vmx handler. This will reduce the interrupt handle cost of KVM.
Also, interrupt enable logic is changed if this feature is turnning on:
Before this patch, hypervior calllocal_irq_enable()to enable it directly. Now IF bit is set on interrupt stack frame, and will be enabled on a return from interrupt handler if exterrupt interrupt exists. If no external interrupt, still calllocal_irq_enable()to enable it.Refer to Intel SDM volum 3, chapter 33.2.
+static void
vmx_handle_external_intr(struct kvm_vcpu *vcpu)+{ +
u32exit_intr_info=vmcs_read32(VM_EXIT_INTR_INFO); + + /* +
* If external interrupt exists, IF bit is set in rflags/eflags on
the + * interrupt stack frame, and interrupt will be enabled on
a return + * from interrupt handler. + */ + if
((exit_intr_info& (INTR_INFO_VALID_MASK|INTR_INFO_INTR_TYPE_MASK)) +== (INTR_INFO_VALID_MASK | INTR_TYPE_EXT_INTR)) {+ unsigned int vector; + unsigned long
entry; +gate_desc*desc; + structvcpu_vmx
*vmx =to_vmx(vcpu); + + vector =exit_intr_info&INTR_INFO_VECTOR_MASK; +#ifdefCONFIG_X86_64+ desc =
(void *)vmx->host_idt_base + vector * 16; +#else + desc =
(void *)vmx->host_idt_base + vector * 8; +#endif + +
entry =gate_offset(*desc); + asm( +Quoted 8 lines
Quoted 6 lines
"mov %0, %%"
_ASM_DX" \n\t" +#ifdefCONFIG_X86_64+"mov %%" _ASM_SP ", %%" _ASM_BX " \n\t" +"and $0xfffffffffffffff0, %%"
_ASM_SP" \n\t" +
"mov %%ss, %%"_ASM_AX" \n\t" + "push %%"_ASM_AX" \n\t" + "push %%"_ASM_BX" \n\t"
+#endifAre we sure no interrupts are using the IST feature? I guess it's unlikely.
Linux uses IST for NMI, stack fault, machine-check, double fault and debug interrupt . No external interrupt will use it. This feature is only for external interrupt. So we don't need to check it here.
Quoted 7 lines
arch/x86/kvm/vmx.c static void vmx_complete_atomic_exit(struct vcpu_vmx *vmx) · line 6388
Quoted 5 lines
arch/x86/kvm/vmx.c +4
-0+"pushf \n\t" +"pop %%" _ASM_AX" \n\t" +"or $0x200, %%" _ASM_AX" \n\t" +"push %%" _ASM_AX" \n\t" Can simplify to pushf; orl $0x200, %%rsp.
Sure.
Quoted 3 lines
Quoted 2 lines
arch/x86/kvm/vmx.c +2
-0+"mov %%cs, %%" _ASM_AX" \n\t" +"push %%" _ASM_AX" \n\t"
Quoted 8 lines
Quoted 1 line
push %%cs
"push %%cs" is invalid in
x86_64.Quoted 3 lines
Quoted 1 line
+ "push
intr_return\n\t"push $1f. Or even combine with the next instruction, and call %rdx.
Which is faster? jmp or call?
Wrong question. You need to compare push+jmp with call. I do not see why later will be slower.
Quoted 13 lines
Quoted 13 lines
Quoted 13 lines
arch/x86/kvm/vmx.c +13
-0+ "jmp *%% " _ASM_DX " \n\t" + "1: \n\t" + ".pushsection .rodata \n\t" + ".global intr_return \n\t" + "intr_return: " _ASM_PTR " 1b \n\t" + ".popsection \n\t" + : : "m"(entry) : +#ifdef CONFIG_X86_64 +"rax" ,"rbx" ,"rdx" +#else +"eax" ,"edx" +#endif + );
Quoted 9 lines
Quoted 4 lines
arch/x86/kvm/vmx.c static void vmx_complete_atomic_exit(struct vcpu_vmx *vmx) · line 6408
Quoted 4 lines
arch/x86/kvm/vmx.c +4
-0+ }else +local_irq_enable (); +} +Best regards,
Yang
--
Gleb.
RE: [PATCH v3 1/2] KVM: VMX: enable acknowledge interupt on vmexit
Gleb Natapov wrote on 2013-02-20:
Quoted 71 lines · Yang Z
Quoted 2 lines
Avi Kivity wrote on 2013-02-20:
Quoted 1 line
On Tue, Feb 19, 2013 at 3:39 PM, Yang Zhang <[email protected]>
Quoted 26 lines · wrote:
Quoted 26 lines
Quoted 26 lines
From: Yang Zhang <[email protected]>
The "acknowledge interrupt on exit" feature controls processor behavior for external interrupt acknowledgement. When this control is set, the processor acknowledges the interrupt controller to acquire the interrupt vector on VM exit.
After enabling this feature, an interrupt which arrived when target cpu is running in vmx non-root mode will be handled by vmx handler instead of handler in idt. Currently, vmx handler only fakes an interrupt stack and jump to idt table to let real handler to handle it. Further, we will recognize the interrupt and only delivery the interrupt which not belong to current vcpu through idt table. The interrupt which belonged to current vcpu will be handled inside vmx handler. This will reduce the interrupt handle cost of KVM.
Also, interrupt enable logic is changed if this feature is turnning
on: Before this patch, hypervior calllocal_irq_enable()to enable it directly. Now IF bit is set on interrupt stack frame, and will be enabled on a return from interrupt handler if exterrupt interrupt exists. If no external interrupt, still calllocal_irq_enable()to enable it.Refer to Intel SDM volum 3, chapter 33.2.
Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
+static void
vmx_handle_external_intr(struct kvm_vcpu *vcpu)+{ +
u32exit_intr_info=vmcs_read32(VM_EXIT_INTR_INFO); + + /* +
* If external interrupt exists, IF bit is set in rflags/eflags on
the + * interrupt stack frame, and interrupt will be enabled on
a return + * from interrupt handler. + */ + if
((exit_intr_info& (INTR_INFO_VALID_MASK|
INTR_INFO_INTR_TYPE_MASK)) +Quoted 1 line
Quoted 1 line
Quoted 1 line
== (INTR_INFO_VALID_MASK |
INTR_TYPE_EXT_INTR)) {Quoted 32 lines
Quoted 14 lines
Quoted 8 lines
+ unsigned int vector; + unsigned long
entry; +gate_desc*desc; + structvcpu_vmx*vmx =to_vmx(vcpu); + + vector =exit_intr_info&INTR_INFO_VECTOR_MASK; +#ifdefCONFIG_X86_64+ desc = (void *)vmx->host_idt_base + vector * 16; +#else +
desc = (void *)vmx->host_idt_base + vector * 8; +#endif + + entry =gate_offset(*desc); + asm( +
"mov %0, %%"_ASM_DX" \n\t" +#ifdefCONFIG_X86_64+Quoted 5 lines
"mov %%" _ASM_SP ", %%" _ASM_BX " \n\t" +"and $0xfffffffffffffff0, %%"
_ASM_SP" \n\t" +
"mov %%ss, %%"_ASM_AX" \n\t" + "push %%"_ASM_AX" \n\t" + "push %%"_ASM_BX" \n\t"
+#endifQuoted 1 line
Are we sure no interrupts are using the IST feature? I guess it's unlikely.
Linux uses IST for NMI, stack fault, machine-check, double fault and debug interrupt . No external interrupt will use it. This feature is only for external interrupt. So we don't need to check it here.
Quoted 7 lines
arch/x86/kvm/vmx.c static void vmx_complete_atomic_exit(struct vcpu_vmx *vmx) · line 6388
Quoted 5 lines
arch/x86/kvm/vmx.c +4
-0+"pushf \n\t" +"pop %%" _ASM_AX" \n\t" +"or $0x200, %%" _ASM_AX" \n\t" +"push %%" _ASM_AX" \n\t" Can simplify to pushf; orl $0x200, %%rsp.
Sure.
Quoted 4 lines
arch/x86/kvm/vmx.c static void vmx_complete_atomic_exit(struct vcpu_vmx *vmx) · line 6390
Quoted 2 lines
arch/x86/kvm/vmx.c +2
-0+"mov %%cs, %%" _ASM_AX" \n\t" +"push %%" _ASM_AX" \n\t" push %%cs
Quoted 8 lines
Quoted 7 lines
"push %%cs" is invalid in
x86_64.Quoted 3 lines
Quoted 1 line
+ "push
intr_return\n\t"push $1f. Or even combine with the next instruction, and call %rdx.
Which is faster? jmp or call?
Wrong question. You need to compare push+jmp with call. I do not see why
Sorry, I didn't express it clearly. Yes, I want to compare push+jmp with call.
Quoted 1 line
later will be slower.
I think so. If push+jmp is not faster than call, I will use the latter.
Quoted 23 lines
Quoted 13 lines
Quoted 13 lines
Quoted 13 lines
arch/x86/kvm/vmx.c +13
-0+ "jmp *%% " _ASM_DX " \n\t" + "1: \n\t" + ".pushsection .rodata \n\t" + ".global intr_return \n\t" + "intr_return: " _ASM_PTR " 1b \n\t" + ".popsection \n\t" + : : "m"(entry) : +#ifdef CONFIG_X86_64 +"rax" ,"rbx" ,"rdx" +#else +"eax" ,"edx" +#endif + );Quoted 9 lines
Quoted 4 lines
arch/x86/kvm/vmx.c static void vmx_complete_atomic_exit(struct vcpu_vmx *vmx) · line 6408
Quoted 4 lines
arch/x86/kvm/vmx.c +4
-0+ }else +local_irq_enable (); +} +Best regards,
Yang
Quoted 2 lines
--
Gleb.
Best regards,
Yang
Re: [PATCH v3 1/2] KVM: VMX: enable acknowledge interupt on vmexit
In reply to Zhang, Yang ZQuoted 23 lines · Yang Z
Quoted 15 lines
Quoted 15 lines
+static void
vmx_handle_external_intr(struct kvm_vcpu *vcpu)+{ +
u32exit_intr_info=vmcs_read32(VM_EXIT_INTR_INFO); + + /* +
* If external interrupt exists, IF bit is set in rflags/eflags on
the + * interrupt stack frame, and interrupt will be enabled on
a return + * from interrupt handler. + */ + if
((exit_intr_info& (INTR_INFO_VALID_MASK|INTR_INFO_INTR_TYPE_MASK)) +== (INTR_INFO_VALID_MASK | INTR_TYPE_EXT_INTR)) {+ unsigned int vector; + unsigned long
entry; +gate_desc*desc; + structvcpu_vmx
*vmx =to_vmx(vcpu); + + vector =exit_intr_info&INTR_INFO_VECTOR_MASK; +#ifdefCONFIG_X86_64+ desc =
(void *)vmx->host_idt_base + vector * 16; +#else + desc =
(void *)vmx->host_idt_base + vector * 8; +#endif + +
entry =gate_offset(*desc); + asm( +Quoted 8 lines
Quoted 6 lines
"mov %0, %%"
_ASM_DX" \n\t" +#ifdefCONFIG_X86_64+"mov %%" _ASM_SP ", %%" _ASM_BX " \n\t" +"and $0xfffffffffffffff0, %%"
_ASM_SP" \n\t" +
"mov %%ss, %%"_ASM_AX" \n\t" + "push %%"_ASM_AX" \n\t" + "push %%"_ASM_BX" \n\t"
+#endifAre we sure no interrupts are using the IST feature? I guess it's unlikely.
Quoted 2 lines
Linux uses IST for NMI, stack fault, machine-check, double fault and debug interrupt . No external interrupt will use it. This feature is only for external interrupt. So we don't need to check it here.
Ok, thanks for checking.
Quoted 7 lines
Quoted 7 lines
Quoted 6 lines
arch/x86/kvm/vmx.c +4
-0+"pushf \n\t" +"pop %%" _ASM_AX" \n\t" +"or $0x200, %%" _ASM_AX" \n\t" +"push %%" _ASM_AX" \n\t"
Quoted 8 lines
Quoted 1 line
Can simplify to pushf; orl $0x200, %%rsp.
Sure.
Quoted 4 lines
arch/x86/kvm/vmx.c static void vmx_complete_atomic_exit(struct vcpu_vmx *vmx) · line 6390
Quoted 2 lines
arch/x86/kvm/vmx.c +2
-0+"mov %%cs, %%" _ASM_AX" \n\t" +"push %%" _ASM_AX" \n\t" push %%cs
"push %%cs" is invalid in
x86_64.
Oops. 'push[lq] $__KERNEL_CS' then.
Quoted 5 lines
Quoted 4 lines
Quoted 2 lines
+ "push
intr_return\n\t"push $1f. Or even combine with the next instruction, and call %rdx.
Which is faster? jmp or call?
Actually it doesn't matter, the processor is clever enough to minimize the difference. But the code is simpler and shorter with 'call'.
RE: [PATCH v3 1/2] KVM: VMX: enable acknowledge interupt on vmexit
Avi Kivity wrote on 2013-02-20:
Quoted 43 lines
Quoted 7 lines · Yang Z
Quoted 7 lines
Quoted 7 lines
+static void
vmx_handle_external_intr(struct kvm_vcpu *vcpu)+{ +
u32exit_intr_info=vmcs_read32(VM_EXIT_INTR_INFO); + + /* +
* If external interrupt exists, IF bit is set in rflags/eflags on
the + * interrupt stack frame, and interrupt will be enabled on
a return + * from interrupt handler. + */ + if
((exit_intr_info& (INTR_INFO_VALID_MASK|
INTR_INFO_INTR_TYPE_MASK)) +Quoted 1 line
Quoted 1 line
Quoted 1 line
== (INTR_INFO_VALID_MASK |
INTR_TYPE_EXT_INTR)) {Quoted 14 lines
Quoted 14 lines
Quoted 13 lines
+ unsigned int vector; + unsigned long
entry; +gate_desc*desc; + structvcpu_vmx*vmx =to_vmx(vcpu); + + vector =exit_intr_info&INTR_INFO_VECTOR_MASK; +#ifdefCONFIG_X86_64+ desc = (void *)vmx->host_idt_base + vector * 16; +#else +
desc = (void *)vmx->host_idt_base + vector * 8; +#endif + + entry =gate_offset(*desc); + asm( +
"mov %0, %%"_ASM_DX" \n\t" +#ifdefCONFIG_X86_64+"mov %%" _ASM_SP ", %%" _ASM_BX " \n\t" +"and $0xfffffffffffffff0, %%"
_ASM_SP" \n\t" +
"mov %%ss, %%"_ASM_AX" \n\t" + "push %%"_ASM_AX" \n\t" + "push %%"_ASM_BX" \n\t"
+#endifQuoted 4 lines
Quoted 1 line
Are we sure no interrupts are using the IST feature? I guess it's unlikely.
Linux uses IST for NMI, stack fault, machine-check, double fault and debug interrupt . No external interrupt will use it. This feature is only for external interrupt. So we don't need to check it here.
Ok, thanks for checking.
Quoted 10 lines
Quoted 8 lines
arch/x86/kvm/vmx.c static void vmx_complete_atomic_exit(struct vcpu_vmx *vmx) · line 6388
Quoted 6 lines
arch/x86/kvm/vmx.c +4
-0+"pushf \n\t" +"pop %%" _ASM_AX" \n\t" +"or $0x200, %%" _ASM_AX" \n\t" +"push %%" _ASM_AX" \n\t" Can simplify to pushf; orl $0x200, %%rsp.
Sure.
Quoted 7 lines
Quoted 5 lines
Quoted 4 lines
arch/x86/kvm/vmx.c static void vmx_complete_atomic_exit(struct vcpu_vmx *vmx) · line 6390
Quoted 2 lines
arch/x86/kvm/vmx.c +2
-0+"mov %%cs, %%" _ASM_AX" \n\t" +"push %%" _ASM_AX" \n\t" push %%cs
"push %%cs" is invalid in
x86_64.Oops. 'push[lq] $__KERNEL_CS' then.
Is this right? Just copy it from other file.
#define __STR(X) #X
#define STR(X) __STR(X)
#ifdef CONFIG_X86_64
"pushq $"STR(__KERNEL_CS)" \n\t"
#else
"pushl $"STR(__KERNEL_CS)" \n\t"#endif
#undef STR
#undef __STR
Quoted 8 lines
Quoted 5 lines
Quoted 4 lines
Quoted 2 lines
+ "push
intr_return\n\t"push $1f. Or even combine with the next instruction, and call %rdx.
Which is faster? jmp or call?
Actually it doesn't matter, the processor is clever enough to minimize the difference. But the code is simpler and shorter with 'call'. -- To
Yes, 'call' is better.
Best regards,
Yang
Re: [PATCH v3 1/2] KVM: VMX: enable acknowledge interupt on vmexit
Quoted 10 lines · Yang Z
Quoted 5 lines
Quoted 3 lines
Quoted 2 lines
push %%cs
"push %%cs" is invalid in
x86_64.Oops. 'push[lq] $__KERNEL_CS' then.
Is this right? Just copy it from other file.
#define
__STR(X)#X
#define STR(X)__STR(X)
Quoted 9 lines
#ifdef CONFIG_X86_64 "pushq $"STR(__KERNEL_CS)" \n\t" #else "pushl $"STR(__KERNEL_CS)" \n\t"#endif
#undef STR
#undef__STR
Use __ASM_SIZE and an immediate operand for __KERNEL_CS:
asm ( ... : : [cs]"i"(__KERNEL_CS) );
and the code will be cleaner.
RE: [PATCH v3 1/2] KVM: VMX: enable acknowledge interupt on vmexit
Avi Kivity wrote on 2013-02-20:
Quoted 17 lines
Quoted 15 lines · Yang Z
Quoted 5 lines
Quoted 3 lines
Quoted 2 lines
push %%cs
"push %%cs" is invalid in
x86_64.Oops. 'push[lq] $__KERNEL_CS' then.
Is this right? Just copy it from other file.
#define
__STR(X)#X
#define STR(X)__STR(X)#ifdef CONFIG_X86_64 "pushq $"STR(__KERNEL_CS)" \n\t" #else "pushl $"STR(__KERNEL_CS)" \n\t"#endif
Quoted 8 lines
Quoted 3 lines
#undef STR
#undef__STRUse
__ASM_SIZEand an immediate operand for__KERNEL_CS:asm ( ... : : [cs]"i"(__KERNEL_CS) );
and the code will be cleaner.
Thanks. Here is code after changing, please review it:
asm(
"mov %0, %%" _ASM_DX " \n\t"
#ifdef CONFIG_X86_64
"mov %%" _ASM_SP ", %%" _ASM_BX " \n\t"
"and $0xfffffffffffffff0, %%" _ASM_SP " \n\t"
"mov %%ss, %%" _ASM_AX " \n\t"
"push %%" _ASM_AX " \n\t"
"push %%" _ASM_BX " \n\t"
#endif
"pushf \n\t"
"orl $0x200, (%%" _ASM_SP ") \n\t"
__ASM_SIZE(push) " %c[cs] \n\t"
"call *%% " _ASM_DX " \n\t"
: : "m"(entry), [cs]"i"(__KERNEL_CS) :
#ifdef CONFIG_X86_64
"rax", "rbx", "rdx"
#else
"edx"
#endif
);Best regards,
Yang
Re: [PATCH v3 1/2] KVM: VMX: enable acknowledge interupt on vmexit
Quoted 14 lines · Yang Z
Thanks. Here is code after changing, please review it:
asm( "mov %0, %%" _ASM_DX " \n\t" #ifdef CONFIG_X86_64 "mov %%" _ASM_SP ", %%" _ASM_BX " \n\t" "and $0xfffffffffffffff0, %%" _ASM_SP " \n\t" "mov %%ss, %%" _ASM_AX " \n\t" "push %%" _ASM_AX " \n\t" "push %%" _ASM_BX " \n\t" #endif "pushf \n\t" "orl $0x200, (%%" _ASM_SP ") \n\t" __ASM_SIZE(push) " %c[cs] \n\t"
Quoted 8 lines
"call *%% " _ASM_DX " \n\t"
: : "m"(entry), [cs]"i"(__KERNEL_CS) :
#ifdef CONFIG_X86_64
"rax", "rbx", "rdx"
#else
"edx"
#endif
);For cleanliness, you can provide more parameters via the constraints:
"m"(entry) -> [entry]"r"(entry) (and then use "call *[entry]") use "push %c[ss]" for ss
and [sp]"=&r"(tmp) instead of rdx/edx as the temporary for rsp
RE: [PATCH v3 1/2] KVM: VMX: enable acknowledge interupt on vmexit
Avi Kivity wrote on 2013-02-21:
Quoted 16 lines
Quoted 13 lines · Yang Z
Thanks. Here is code after changing, please review it:
asm( "mov %0, %%" _ASM_DX " \n\t" #ifdef CONFIG_X86_64 "mov %%" _ASM_SP ", %%" _ASM_BX " \n\t" "and $0xfffffffffffffff0, %%" _ASM_SP " \n\t" "mov %%ss, %%" _ASM_AX " \n\t" "push %%" _ASM_AX " \n\t" "push %%" _ASM_BX " \n\t" #endif "pushf \n\t" "orl $0x200, (%%" _ASM_SP ") \n\t" __ASM_SIZE(push) " %c[cs] \n\t" "call *%% " _ASM_DX " \n\t" : : "m"(entry), [cs]"i"(__KERNEL_CS) : #ifdef CONFIG_X86_64 "rax", "rbx", "rdx" #else "edx" #endif );
Quoted 5 lines
For cleanliness, you can provide more parameters via the constraints:
"m"(entry) -> [entry]"r"(entry) (and then use "call *[entry]") use "push %c[ss]" for ss
and [sp]"=&r"(tmp) instead of rdx/edx as the temporary for rsp
I didn't found the __KERNEL_SS definition. But I noticed that KVM use __KERNEL_DS as HOST_SS_SELECTOR. Please review the following changes.
asm volatile(
#ifdef CONFIG_X86_64
"mov %%" _ASM_SP ", %[sp] \n\t"
"and $0xfffffffffffffff0, %%" _ASM_SP " \n\t"
"push $%c[ss] \n\t"
"push %[sp] \n\t"
#endif
"pushf \n\t"
"orl $0x200, (%%" _ASM_SP ") \n\t"
__ASM_SIZE(push) " $%c[cs] \n\t"
"call *%[entry] \n\t"
:
#ifdef CONFIG_X86_64
[sp]"=&r"(tmp)
#endif
: [entry]"r"(entry),
[ss]"i"(__KERNEL_DS),
[cs]"i"(__KERNEL_CS)
);Best regards,
Yang
A message that is not in the archive (sent off-list or not yet received): CAG7+5M1c7mtENHao+1yFCQkQus78HXK+QQBi3vwE6chAr_ZxVA@mail.gmail.com
RE: [PATCH v3 1/2] KVM: VMX: enable acknowledge interupt on vmexit
Eric Northup wrote on 2013-02-21:
Quoted 56 lines · Yang Z
Quoted 38 lines
Avi Kivity wrote on 2013-02-20:
Quoted 36 lines
Quoted 34 lines · Yang Zhang
From: Yang Zhang <[email protected]>
The "acknowledge interrupt on exit" feature controls processor behavior for external interrupt acknowledgement. When this control is set, the processor acknowledges the interrupt controller to acquire the interrupt vector on VM exit.
After enabling this feature, an interrupt which arrived when target cpu is running in vmx non-root mode will be handled by vmx handler instead of handler in idt. Currently, vmx handler only fakes an interrupt stack and jump to idt table to let real handler to handle it. Further, we will recognize the interrupt and only delivery the interrupt which not belong to current vcpu through idt table. The interrupt which belonged to current vcpu will be handled inside vmx handler. This will reduce the interrupt handle cost of KVM.
Also, interrupt enable logic is changed if this feature is turnning on:
Before this patch, hypervior calllocal_irq_enable()to enable it directly.
Now IF bit is set on interrupt stack frame, and will be enabled on a return from
interrupt handler if exterrupt interrupt exists. If no external interrupt, still
calllocal_irq_enable()to enable it.Refer to Intel SDM volum 3, chapter 33.2.
+static void
vmx_handle_external_intr(struct kvm_vcpu *vcpu)+{ +
u32exit_intr_info=vmcs_read32(VM_EXIT_INTR_INFO); + + /* +
* If external interrupt exists, IF bit is set in rflags/eflags on
the + * interrupt stack frame, and interrupt will be enabled on
a return + * from interrupt handler. + */ + if
((exit_intr_info& (INTR_INFO_VALID_MASK|
INTR_INFO_INTR_TYPE_MASK)) +Quoted 1 line
Quoted 1 line
Quoted 1 line
== (INTR_INFO_VALID_MASK |
INTR_TYPE_EXT_INTR)) {Quoted 14 lines
Quoted 14 lines
Quoted 13 lines
+ unsigned int vector; + unsigned long
entry; +gate_desc*desc; + structvcpu_vmx*vmx =to_vmx(vcpu); + + vector =exit_intr_info&INTR_INFO_VECTOR_MASK; +#ifdefCONFIG_X86_64+ desc = (void *)vmx->host_idt_base + vector * 16; +#else +
desc = (void *)vmx->host_idt_base + vector * 8; +#endif + + entry =gate_offset(*desc); + asm( +
"mov %0, %%"_ASM_DX" \n\t" +#ifdefCONFIG_X86_64+"mov %%" _ASM_SP ", %%" _ASM_BX " \n\t" +"and $0xfffffffffffffff0, %%"
_ASM_SP" \n\t" +
"mov %%ss, %%"_ASM_AX" \n\t" + "push %%"_ASM_AX" \n\t" + "push %%"_ASM_BX" \n\t"
+#endif
Quoted 8 lines
Quoted 6 lines
Quoted 2 lines
Are we sure no interrupts are using the IST feature? I guess it's unlikely.
Linux uses IST for NMI, stack fault, machine-check, double fault and debug interrupt . No external interrupt will use it.
This feature is only for external interrupt. So we don't need to check it here.Would it be silly paranoia to
BUG_ON(desc->ist)?
I think the BUG_ON is reasonable. I will add it in next version.
Best regards,
Yang
PATCH v3 2/2 KVM: VMX: Add Posted Interrupt supporting
In reply to Yang ZhangFrom: Yang Zhang <[email protected]>
Posted Interrupt allows APIC interrupts to inject into guest directly without any vmexit.
- When delivering a interrupt to guest, if target vcpu is running, update Posted-interrupt requests bitmap and send a notification event to the vcpu. Then the vcpu will handle this interrupt automatically, without any software involvemnt.
- If target vcpu is not running or there already a notification event pending in the vcpu, do nothing. The interrupt will be handled by next vm entry
--- arch/x86/include/asm/entry_arch.h | 4 + arch/x86/include/asm/hw_irq.h | 1 + arch/x86/include/asm/irq_vectors.h | 5 + arch/x86/include/asm/kvm_host.h | 3 + arch/x86/include/asm/vmx.h | 4 + arch/x86/kernel/entry_64.S | 5 + arch/x86/kernel/irq.c | 20 +++++ arch/x86/kernel/irqinit.c | 4 + arch/x86/kvm/lapic.c | 19 ++++- arch/x86/kvm/lapic.h | 1 + arch/x86/kvm/svm.c | 13 +++ arch/x86/kvm/vmx.c | 157 +++++++++++++++++++++++++++++++----- arch/x86/kvm/x86.c | 1 + 13 files changed, 214 insertions(+), 23 deletions(-)
arch/x86/include/asm/entry_arch.h +4 -0
@@ -19,6 +19,10 @@ BUILD_INTERRUPT(reboot_interrupt,REBOOT_VECTOR) BUILD_INTERRUPT(x86_platform_ipi, X86_PLATFORM_IPI_VECTOR) +#ifdef CONFIG_HAVE_KVM +BUILD_INTERRUPT(kvm_posted_intr_ipi, POSTED_INTR_VECTOR) +#endif +/* * every pentium local APIC has two 'local interrupts', with a * soft-definable vector attached to both interrupts, one of
arch/x86/include/asm/hw_irq.h +1 -0
@@ -28,6 +28,7 @@/* Interrupt handlers registered during init_IRQ */ extern void apic_timer_interrupt (void );extern void x86_platform_ipi (void ); +extern void kvm_posted_intr_ipi (void );extern void error_interrupt (void );extern void irq_work_interrupt (void );
arch/x86/include/asm/irq_vectors.h +5 -0
@@ -102,6 +102,11 @@*/ #define X86_PLATFORM_IPI_VECTOR 0xf7 +/* Vector for KVM to deliver posted interrupt IPI */ +#ifdef CONFIG_HAVE_KVM +#define POSTED_INTR_VECTOR 0xf2 +#endif +/* * IRQ work vector: */
arch/x86/include/asm/kvm_host.h +3 -0
@@ -704,6 +704,9 @@ struct kvm_x86_ops {void (*hwapic_isr_update)(struct kvm *kvm,int isr);void (*load_eoi_exitmap)(struct kvm_vcpu *vcpu,u64 *eoi_exit_bitmap);void (*set_virtual_x2apic_mode)(struct kvm_vcpu *vcpu,bool set); +bool (*deliver_posted_interrupt)(struct kvm_vcpu *vcpu,int vector, +int *result,bool *delivered); +void (*sync_pir_to_irr)(struct kvm_vcpu *vcpu);int (*set_tss_addr)(struct kvm *kvm,unsigned int addr);int (*get_tdp_level)(void );u64 (*get_mt_mask)(struct kvm_vcpu *vcpu,gfn_t gfn,bool is_mmio);
arch/x86/include/asm/vmx.h +4 -0
@@ -158,6 +158,7 @@#define PIN_BASED_EXT_INTR_MASK 0x00000001#define PIN_BASED_NMI_EXITING 0x00000008#define PIN_BASED_VIRTUAL_NMIS 0x00000020 +#define PIN_BASED_POSTED_INTR 0x00000080#define VM_EXIT_SAVE_DEBUG_CONTROLS 0x00000002#define VM_EXIT_HOST_ADDR_SPACE_SIZE 0x00000200
@@ -180,6 +181,7 @@/* VMCS Encodings */ enum vmcs_field {VIRTUAL_PROCESSOR_ID =0x00000000 , +POSTED_INTR_NV =0x00000002 ,GUEST_ES_SELECTOR =0x00000800 ,GUEST_CS_SELECTOR =0x00000802 ,GUEST_SS_SELECTOR =0x00000804 ,
@@ -214,6 +216,8 @@ enum vmcs_field {VIRTUAL_APIC_PAGE_ADDR_HIGH =0x00002013 ,APIC_ACCESS_ADDR =0x00002014 ,APIC_ACCESS_ADDR_HIGH =0x00002015 , +POSTED_INTR_DESC_ADDR =0x00002016 , +POSTED_INTR_DESC_ADDR_HIGH =0x00002017 ,EPT_POINTER =0x0000201a ,EPT_POINTER_HIGH =0x0000201b , EOI_EXIT_BITMAP0 = 0x0000201c,
arch/x86/kernel/entry_64.S +5 -0
@@ -1177,6 +1177,11 @@ apicinterrupt LOCAL_TIMER_VECTOR \ apicinterrupt X86_PLATFORM_IPI_VECTOR \ x86_platform_ipi smp_x86_platform_ipi +#ifdef CONFIG_HAVE_KVM +apicinterrupt POSTED_INTR_VECTOR \ + kvm_posted_intr_ipi smp_posted_intr_ipi +#endif + apicinterrupt THRESHOLD_APIC_VECTOR \ threshold_interrupt smp_threshold_interrupt apicinterrupt THERMAL_APIC_VECTOR \
arch/x86/kernel/irq.c +20 -0
@@ -228,6 +228,26 @@ void smp_x86_platform_ipi(struct pt_regs *regs)set_irq_regs (old_regs); } +#ifdef CONFIG_HAVE_KVM +/* +* Handler for POSTED_INTERRUPT_VECTOR. +*/ +void smp_posted_intr_ipi (struct pt_regs *regs) +{ +struct pt_regs *old_regs =set_irq_regs (regs); + +ack_APIC_irq (); + +irq_enter (); + +exit_idle (); + +irq_exit (); + +set_irq_regs (old_regs); +} +#endif +EXPORT_SYMBOL_GPL (vector_used_by_percpu_irq);#ifdef CONFIG_HOTPLUG_CPU
arch/x86/kernel/irqinit.c +4 -0
@@ -205,6 +205,10 @@ static void __init apic_intr_init(void)/* IPI for X86 platform specific use */ alloc_intr_gate (X86_PLATFORM_IPI_VECTOR , x86_platform_ipi); +#ifdef CONFIG_HAVE_KVM +/* IPI for KVM to deliver posted interrupt */ +alloc_intr_gate (POSTED_INTR_VECTOR , kvm_posted_intr_ipi); +#endif /* IPI vectors for APIC spurious and error interrupts */ alloc_intr_gate (SPURIOUS_APIC_VECTOR , spurious_interrupt);
arch/x86/kvm/lapic.c +16 -3
@@ -357,6 +357,12 @@ static u8 count_vectors(void *bitmap)return count; } +int kvm_apic_test_irr (int vec,struct kvm_lapic *apic) +{ +return apic_test_vector (vec, apic->regs +APIC_IRR ); +} +EXPORT_SYMBOL_GPL (kvm_apic_test_irr); +static inline int apic_test_and_set_irr (int vec,struct kvm_lapic *apic) { apic->irr_pending = true;
@@ -379,6 +385,7 @@ static inline int apic_find_highest_irr(struct kvm_lapic *apic)if (!apic->irr_pending)return -1 ; + kvm_x86_ops->sync_pir_to_irr (apic->vcpu); result =apic_search_irr (apic);ASSERT (result ==-1 || result >=16 );
@@ -685,6 +692,7 @@ static int __apic_accept_irq(struct kvm_lapic *apic, int delivery_mode, {int result =0 ;struct kvm_vcpu *vcpu = apic->vcpu; +bool delivered = false;switch (delivery_mode) {case APIC_DM_LOWEST :
@@ -700,7 +708,10 @@ static int __apic_accept_irq(struct kvm_lapic *apic, int delivery_mode, }else apic_clear_vector (vector, apic->regs +APIC_TMR ); - result = !apic_test_and_set_irr (vector, apic); +if (!kvm_x86_ops->deliver_posted_interrupt (vcpu, vector, + &result, &delivered)) + result = !apic_test_and_set_irr (vector, apic); +trace_kvm_apic_accept_irq (vcpu->vcpu_id, delivery_mode, trig_mode, vector, !result);if (!result) {
@@ -710,8 +721,10 @@ static int __apic_accept_irq(struct kvm_lapic *apic, int delivery_mode,break ; } -kvm_make_request (KVM_REQ_EVENT , vcpu); -kvm_vcpu_kick (vcpu); +if (!delivered) { +kvm_make_request (KVM_REQ_EVENT , vcpu); +kvm_vcpu_kick (vcpu); + }break ;case APIC_DM_REMRD :
arch/x86/kvm/lapic.h +1 -0
@@ -157,5 +157,6 @@ static inline u16 apic_logical_id(struct kvm_apic_map *map, u32 ldr)void kvm_calculate_eoi_exitmap (struct kvm_vcpu *vcpu,struct kvm_lapic_irq *irq,u64 *eoi_bitmap); +int kvm_apic_test_irr (int vec,struct kvm_lapic *apic);#endif
arch/x86/kvm/svm.c +13 -0
@@ -3591,6 +3591,17 @@ static void svm_hwapic_isr_update(struct kvm *kvm, int isr)return ; } +static void svm_sync_pir_to_irr (struct kvm_vcpu *vcpu) +{ +return ; +} + +static bool svm_deliver_posted_interrupt (struct kvm_vcpu *vcpu, +int vector,int *result,bool *delivered) +{ +return false; +} +static int svm_nmi_allowed (struct kvm_vcpu *vcpu) {struct vcpu_svm *svm =to_svm (vcpu);
@@ -4319,6 +4330,8 @@ static struct kvm_x86_ops svm_x86_ops = { .vm_has_apicv = svm_vm_has_apicv, .load_eoi_exitmap = svm_load_eoi_exitmap, .hwapic_isr_update = svm_hwapic_isr_update, + .sync_pir_to_irr = svm_sync_pir_to_irr, + .deliver_posted_interrupt = svm_deliver_posted_interrupt, .set_tss_addr = svm_set_tss_addr, .get_tdp_level = get_npt_level,
arch/x86/kvm/vmx.c +137 -20
254 lines not on this page · raw message
arch/x86/kvm/x86.c +1 -0
@@ -2679,6 +2679,7 @@ void kvm_arch_vcpu_put(struct kvm_vcpu *vcpu)static int kvm_vcpu_ioctl_get_lapic (struct kvm_vcpu *vcpu,struct kvm_lapic_state *s) { + kvm_x86_ops->sync_pir_to_irr (vcpu);memcpy (s->regs, vcpu->arch.apic->regs,sizeof *s);return 0 ;
Signature · 1 line
1.7.1
Hi Marcelo,
Can you help to review this patch? Many thanks if you can review it quickly.
Zhang, Yang Z wrote on 2013-02-19:
Quoted 454 lines
From: Yang Zhang <[email protected]>
Posted Interrupt allows APIC interrupts to inject into guest directly without any vmexit.
- When delivering a interrupt to guest, if target vcpu is running, update Posted-interrupt requests bitmap and send a notification event to the vcpu. Then the vcpu will handle this interrupt automatically, without any software involvemnt.
- If target vcpu is not running or there already a notification event pending in the vcpu, do nothing. The interrupt will be handled by next vm entrySigned-off-by Yang Zhang <[email protected]>--- arch/x86/include/asm/entry_arch.h | 4 + arch/x86/include/asm/hw_irq.h | 1 + arch/x86/include/asm/irq_vectors.h | 5 + arch/x86/include/asm/kvm_host.h | 3 + arch/x86/include/asm/vmx.h | 4 + arch/x86/kernel/entry_64.S | 5 + arch/x86/kernel/irq.c | 20 +++++ arch/x86/kernel/irqinit.c | 4 + arch/x86/kvm/lapic.c | 19 ++++- arch/x86/kvm/lapic.h | 1 + arch/x86/kvm/svm.c | 13 +++ arch/x86/kvm/vmx.c | 157 +++++++++++++++++++++++++++++++----- arch/x86/kvm/x86.c | 1 + 13 files changed, 214 insertions(+), 23deletions(-)
(quoted diff) +0
-0
b/arch/x86/include/asm/entry_arch.hindex 40afa00..9bd4eca 100644 ---a/arch/x86/include/asm/entry_arch.h+++b/arch/x86/include/asm/entry_arch.h@@ -19,6 +19,10 @@BUILD_INTERRUPT(reboot_interrupt,REBOOT_VECTOR)
BUILD_INTERRUPT(x86_platform_ipi, X86_PLATFORM_IPI_VECTOR)
+#ifdefCONFIG_HAVE_KVM+BUILD_INTERRUPT(kvm_posted_intr_ipi,POSTED_INTR_VECTOR) +#endif +
/*
* every pentium local APIC has two 'local interrupts', with a
* soft-definable vector attached to both interrupts, one ofarch/x86/include/asm/hw_irq.h +0
-0@@ -28,6 +28,7 @@/* Interrupt handlers registered during init_IRQ */ extern void apic_timer_interrupt(void); extern void x86_platform_ipi(void); +externvoid kvm_posted_intr_ipi (void );extern void error_interrupt (void );extern void irq_work_interrupt (void );(quoted diff) +0
-0
b/arch/x86/include/asm/irq_vectors.hindex 1508e51..774dc9f 100644 ---a/arch/x86/include/asm/irq_vectors.h+++b/arch/x86/include/asm/irq_vectors.h@@ -102,6 +102,11 @@
*/
#defineX86_PLATFORM_IPI_VECTOR0xf7arch/x86/include/asm/irq_vectors.h +5
-0+/* Vector for KVM to deliver posted interrupt IPI */ +#ifdef CONFIG_HAVE_KVM +#define POSTED_INTR_VECTOR 0xf2 +#endif +/* * IRQ work vector: */ (quoted diff) +0
-0
b/arch/x86/include/asm/kvm_host.hindex b8388e9..79da55e 100644 ---a/arch/x86/include/asm/kvm_host.h+++b/arch/x86/include/asm/kvm_host.harch/x86/include/asm/kvm_host.h +3
-0@@ -704,6 +704,9 @@ struct kvm_x86_ops {void (*hwapic_isr_update)(struct kvm *kvm,int isr);void (*load_eoi_exitmap)(struct kvm_vcpu *vcpu,u64 *eoi_exit_bitmap);void (*set_virtual_x2apic_mode)(struct kvm_vcpu *vcpu,bool set); +bool (*deliver_posted_interrupt)(struct kvm_vcpu *vcpu,int vector, +int *result,bool *delivered); +void (*sync_pir_to_irr)(struct kvm_vcpu *vcpu);int (*set_tss_addr)(struct kvm *kvm,unsigned int addr);int (*get_tdp_level)(void );u64 (*get_mt_mask)(struct kvm_vcpu *vcpu,gfn_t gfn,bool is_mmio);arch/x86/include/asm/vmx.h +1
-0@@ -158,6 +158,7 @@#define PIN_BASED_EXT_INTR_MASK 0x00000001#define PIN_BASED_NMI_EXITING 0x00000008#define PIN_BASED_VIRTUAL_NMIS 0x00000020 +#define PIN_BASED_POSTED_INTR 0x00000080#define VM_EXIT_SAVE_DEBUG_CONTROLS 0x00000002 #define VM_EXIT_HOST_ADDR_SPACE_SIZE 0x00000200 @@ -180,6 +181,7 @@/* VMCS Encodings */ enum
vmcs_field{VIRTUAL_PROCESSOR_ID
=0x00000000, +POSTED_INTR_NV=0x00000002,arch/x86/include/asm/vmx.h +2
-0GUEST_ES_SELECTOR = 0x00000800, GUEST_CS_SELECTOR = 0x00000802, GUEST_SS_SELECTOR = 0x00000804, @@ -214,6 +216,8 @@ enum vmcs_field { VIRTUAL_APIC_PAGE_ADDR_HIGH =0x00002013 ,APIC_ACCESS_ADDR =0x00002014 ,APIC_ACCESS_ADDR_HIGH =0x00002015 , +POSTED_INTR_DESC_ADDR =0x00002016 , +POSTED_INTR_DESC_ADDR_HIGH =0x00002017 ,EPT_POINTER =0x0000201a ,EPT_POINTER_HIGH =0x0000201b ,EOI_EXIT_BITMAP0 =0x0000201c ,arch/x86/kernel/entry_64.S +5
-0@@ -1177,6 +1177,11 @@ apicinterrupt LOCAL_TIMER_VECTOR \ apicinterrupt X86_PLATFORM_IPI_VECTOR \ x86_platform_ipi smp_x86_platform_ipi +#ifdef CONFIG_HAVE_KVM +apicinterrupt POSTED_INTR_VECTOR \ + kvm_posted_intr_ipi smp_posted_intr_ipi +#endif + apicinterrupt THRESHOLD_APIC_VECTOR \ threshold_interrupt smp_threshold_interrupt apicinterrupt THERMAL_APIC_VECTOR \arch/x86/kernel/irq.c +20
-0@@ -228,6 +228,26 @@ void smp_x86_platform_ipi(struct pt_regs *regs)set_irq_regs (old_regs); } +#ifdef CONFIG_HAVE_KVM +/* +* Handler for POSTED_INTERRUPT_VECTOR. +*/ +void smp_posted_intr_ipi (struct pt_regs *regs) +{ +struct pt_regs *old_regs =set_irq_regs (regs); + +ack_APIC_irq (); + +irq_enter (); + +exit_idle (); + +irq_exit (); + +set_irq_regs (old_regs); +} +#endif +EXPORT_SYMBOL_GPL (vector_used_by_percpu_irq);#ifdef CONFIG_HOTPLUG_CPU arch/x86/kernel/irqinit.c +2
-0@@ -205,6 +205,10 @@ static void __init apic_intr_init(void)/* IPI for X86 platform specific use */ alloc_intr_gate (X86_PLATFORM_IPI_VECTOR , x86_platform_ipi); +#ifdef CONFIG_HAVE_KVM +/* IPI for KVM to deliver posted interrupt */ + alloc_intr_gate(POSTED_INTR_VECTOR, kvm_posted_intr_ipi); +#endif/* IPI vectors for APIC spurious and error interrupts */ alloc_intr_gate (SPURIOUS_APIC_VECTOR , spurious_interrupt);arch/x86/kvm/lapic.c +8
-0@@ -357,6 +357,12 @@ static u8 count_vectors(void *bitmap)return count; } +int kvm_apic_test_irr (int vec,struct kvm_lapic *apic) +{ +return apic_test_vector (vec, apic->regs +APIC_IRR ); +} +EXPORT_SYMBOL_GPL (kvm_apic_test_irr); +static inline int apic_test_and_set_irr (int vec,struct kvm_lapic *apic) { apic->irr_pending = true;@@ -379,6 +385,7 @@ static inline int apic_find_highest_irr(struct kvm_lapic *apic)if (!apic->irr_pending)return -1 ; + kvm_x86_ops->sync_pir_to_irr (apic->vcpu); result =apic_search_irr (apic);ASSERT (result ==-1 || result >=16 );@@ -685,6 +692,7 @@ static int __apic_accept_irq(struct kvm_lapic *apic, int delivery_mode, {int result =0 ;struct kvm_vcpu *vcpu = apic->vcpu; +bool delivered = false;switch (delivery_mode) {arch/x86/kvm/lapic.c +5
-2case APIC_DM_LOWEST :@@ -700,7 +708,10 @@ static int __apic_accept_irq(struct kvm_lapic *apic, int delivery_mode, }else apic_clear_vector (vector, apic->regs +APIC_TMR ); - result = !apic_test_and_set_irr (vector, apic); +if (!kvm_x86_ops->deliver_posted_interrupt (vcpu, vector, + &result, &delivered)) + result = !apic_test_and_set_irr (vector, apic); +trace_kvm_apic_accept_irq (vcpu->vcpu_id, delivery_mode, trig_mode, vector, !result);if (!result) {@@ -710,8 +721,10 @@ static int __apic_accept_irq(struct kvm_lapic *apic, int delivery_mode, break; } - kvm_make_request(KVM_REQ_EVENT, vcpu); - kvm_vcpu_kick(vcpu); + if (!delivered) { + kvm_make_request(KVM_REQ_EVENT, vcpu); + kvm_vcpu_kick(vcpu); + }break ;case APIC_DM_REMRD :arch/x86/kvm/lapic.h +1
-0@@ -157,5 +157,6 @@ static inline u16 apic_logical_id(struct kvm_apic_map *map, u32 ldr)void kvm_calculate_eoi_exitmap (struct kvm_vcpu *vcpu,struct kvm_lapic_irq *irq,u64 *eoi_bitmap); +int kvm_apic_test_irr (int vec,struct kvm_lapic *apic);#endif
arch/x86/kvm/svm.c +11
-0@@ -3591,6 +3591,17 @@ static void svm_hwapic_isr_update(struct kvm *kvm, int isr) return; } +static void svm_sync_pir_to_irr (struct kvm_vcpu *vcpu) +{ +return ; +} + +static bool svm_deliver_posted_interrupt (struct kvm_vcpu *vcpu, +int vector,int *result,bool *delivered) +{ +return false; +} + static int svm_nmi_allowed(struct kvm_vcpu *vcpu) { struct vcpu_svm *svm = to_svm(vcpu); @@ -4319,6 +4330,8 @@ static struct kvm_x86_ops svm_x86_ops = { .vm_has_apicv = svm_vm_has_apicv, .load_eoi_exitmap =
svm_load_eoi_exitmap, .hwapic_isr_update =svm_hwapic_isr_update,arch/x86/kvm/svm.c +2
-0+ .sync_pir_to_irr = svm_sync_pir_to_irr, + .deliver_posted_interrupt = svm_deliver_posted_interrupt, .set_tss_addr = svm_set_tss_addr, .get_tdp_level = get_npt_level,arch/x86/kvm/vmx.c +6
-1@@ -84,7 +84,8 @@ module_param(vmm_exclusive, bool, S_IRUGO);static bool __read_mostly fasteoi =1 ;module_param (fasteoi, bool,S_IRUGO ); -static bool __read_mostly enable_apicv_reg_vid; +static bool __read_mostly enable_apicv =1 ; +module_param (enable_apicv, bool,S_IRUGO );/* * If nested=1, nested virtualization is supported, i.e., guests may use @@ -365,6 +366,36 @@ struct nested_vmx {struct page *apic_access_page; }; +#define POSTED_INTR_ON 0 +/* Posted-Interrupt Descriptor */ +struct pi_desc { + u32 pir[8]; /* Posted interrupt requested */ + union { + struct { + u8 on:1, + rsvd:7; + } control; + u32 rsvd[8]; + } u; +} __aligned(64); + +static bool pi_test_and_set_on(struct pi_desc *pi_desc) +{ + return test_and_set_bit(POSTED_INTR_ON, + (unsigned long *)&pi_desc->u.control); +} + +static bool
pi_test_and_clear_on(struct pi_desc *pi_desc)+{ + return
test_and_clear_bit(POSTED_INTR_ON, + (unsigned long
*)&pi_desc->u.control); +} + +static int pi_test_and_set_pir(int vector,
structpi_desc*pi_desc) +{ + return test_and_set_bit(vector, (unsigned
long *)pi_desc->pir); +} +
structvcpu_vmx{arch/x86/kvm/vmx.c +4
-0struct kvm_vcpu vcpu;unsigned long host_rsp;@@ -429,6 +460,9 @@ struct vcpu_vmx {bool rdtscp_enabled; +/* Posted interrupt descriptor */ +struct pi_desc pi_desc; +/* Support for a guest hypervisor (nested VMX) */ struct nested_vmx nested; };@@ -783,6 +817,18 @@ static inline boolcpu_has_vmx_virtual_intr_delivery (void )SECONDARY_EXEC_VIRTUAL_INTR_DELIVERY ; } +static inline bool cpu_has_vmx_posted_intr(void) +{ + returnvmcs_config.pin_based_exec_ctrl &
PIN_BASED_POSTED_INTR; +} + +static
inline boolcpu_has_vmx_apicv(void)+{ + returncpu_has_vmx_apic_register_virt()&&arch/x86/kvm/vmx.c +16
-9+cpu_has_vmx_virtual_intr_delivery () && +cpu_has_vmx_posted_intr (); +} + static inline bool cpu_has_vmx_flexpriority(void) { return cpu_has_vmx_tpr_shadow() &&@@ -2530,12 +2576,6 @@ static __init int setup_vmcs_config(struct vmcs_config *vmcs_conf)u32 _vmexit_control =0 ;u32 _vmentry_control =0 ; - min =PIN_BASED_EXT_INTR_MASK |PIN_BASED_NMI_EXITING ; - opt =PIN_BASED_VIRTUAL_NMIS ; -if (adjust_vmx_controls (min, opt,MSR_IA32_VMX_PINBASED_CTLS , - &_pin_based_exec_control) <0 ) -return -EIO ; - min = CPU_BASED_HLT_EXITING |#ifdef CONFIG_X86_64 CPU_BASED_CR8_LOAD_EXITING |@@ -2612,6 +2652,17 @@ static __init int setup_vmcs_config(struct vmcs_config *vmcs_conf) &_vmexit_control) < 0)return -EIO ; + min =PIN_BASED_EXT_INTR_MASK |PIN_BASED_NMI_EXITING ; + opt =PIN_BASED_VIRTUAL_NMIS |PIN_BASED_POSTED_INTR ; +if (adjust_vmx_controls (min, opt,MSR_IA32_VMX_PINBASED_CTLS , + &_pin_based_exec_control) <0 ) +return -EIO ; + +if (!(_cpu_based_2nd_exec_control & +SECONDARY_EXEC_VIRTUAL_INTR_DELIVERY ) || + !(_vmexit_control &VM_EXIT_ACK_INTR_ON_EXIT )) + _pin_based_exec_control &= ~PIN_BASED_POSTED_INTR ; + min = 0; opt = VM_ENTRY_LOAD_IA32_PAT; if (adjust_vmx_controls(min, opt, MSR_IA32_VMX_ENTRY_CTLS, @@ -2790,11 +2841,10 @@ static __init int hardware_setup(void) if (!cpu_has_vmx_ple()) ple_gap = 0; - if (!cpu_has_vmx_apic_register_virt() || - !cpu_has_vmx_virtual_intr_delivery()) - enable_apicv_reg_vid = 0; + if (!cpu_has_vmx_apicv()) + enable_apicv = 0; - if (enable_apicv_reg_vid) + if (enable_apicv) kvm_x86_ops->update_cr8_intercept = NULL; else kvm_x86_ops->hwapic_irr_update = NULL;@@ -3871,6 +3921,62 @@ static void vmx_disable_intercept_msr_write_x2apic(u32 msr) msr, MSR_TYPE_W); } +static int vmx_vm_has_apicv(struct kvm *kvm) +{ + return enable_apicv&&
irqchip_in_kernel(kvm); +} + +static bool
vmx_deliver_posted_interrupt(structkvm_vcpu*vcpu, + int vector, int
*result, bool *delivered) +{ + structvcpu_vmx*vmx =to_vmx(vcpu); +arch/x86/kvm/vmx.c +1
-0+ if (!vmx_vm_has_apicv(vcpu->kvm)) + return false; + + if(
kvm_apic_test_irr(vector, vcpu->arch.apic)) + goto out; + else {arch/x86/kvm/vmx.c +1
-0+ *result = !pi_test_and_set_pir (vector, &vmx->pi_desc); + if(!*result) + goto out; + } + + if (!pi_test_and_set_on(&vmx->pi_desc)
&& + (vcpu->mode==IN_GUEST_MODE)) {arch/x86/kvm/vmx.c +2
-0+kvm_make_request (KVM_REQ_EVENT , vcpu); + apic->send_IPI_mask(get_cpu_mask(vcpu->cpu), +
POSTED_INTR_VECTOR); + *delivered = true; + } +out: + return true; +} +
+static voidvmx_sync_pir_to_irr(struct kvm_vcpu *vcpu)+{ + structvcpu_vmx*vmx =to_vmx(vcpu); + structkvm_lapic*apic =vcpu->arch.apic; + unsigned int i, old, new,ret_val,irr_offset,pir_val; + + if (!vmx_vm_has_apicv(vcpu->kvm) ||arch/x86/kvm/vmx.c +1
-0+ !pi_test_and_clear_on(&vmx->pi_desc)) + return; + + for (i = 0; i<= 7; i++) { +
pir_val=xchg(&vmx->pi_desc.pir[i], 0); + if (pir_val)
{ +irr_offset=APIC_IRR+ i *0x10; + do { + old =kvm_apic_get_reg(apic, irr_offset); + new = old |pir_val;arch/x86/kvm/vmx.c +1
-0+ ret_val = cmpxchg((u32 *)(apic->regs + + irr_offset), old,new); + } while (
unlikely(ret_val != old)); + } + } +} +
/*
* Set up the vmcs's constant host-state fields, i.e., host-state fields that
* will not change in the lifetime of the guest.arch/x86/kvm/vmx.c +10
-7@@ -3931,6 +4037,15 @@ static void set_cr4_guest_host_mask(struct vcpu_vmx *vmx)vmcs_writel (CR4_GUEST_HOST_MASK , ~vmx->vcpu.arch.cr4_guest_owned_bits); } +static u32 vmx_pin_based_exec_ctrl (struct vcpu_vmx *vmx) +{ +u32 pin_based_exec_ctrl = vmcs_config.pin_based_exec_ctrl; + +if (!vmx_vm_has_apicv (vmx->vcpu.kvm)) + pin_based_exec_ctrl &= ~PIN_BASED_POSTED_INTR ; +return pin_based_exec_ctrl; +} +static u32 vmx_exec_control (struct vcpu_vmx *vmx) {u32 exec_control = vmcs_config.cpu_based_exec_ctrl; @@ -3948,11 +4063,6 @@ static u32 vmx_exec_control(struct vcpu_vmx *vmx) return exec_control; } -static int vmx_vm_has_apicv (struct kvm *kvm) -{ -return enable_apicv_reg_vid &&irqchip_in_kernel (kvm); -} -static u32 vmx_secondary_exec_control (struct vcpu_vmx *vmx) {u32 exec_control = vmcs_config.cpu_based_2nd_exec_ctrl; @@ -4008,8 +4118,7 @@ static int vmx_vcpu_setup(struct vcpu_vmx *vmx) vmcs_write64(VMCS_LINK_POINTER, -1ull); /* 22.3.1.5 *//* Control */ -vmcs_write32 (PIN_BASED_VM_EXEC_CONTROL , - vmcs_config.pin_based_exec_ctrl); +vmcs_write32 (PIN_BASED_VM_EXEC_CONTROL ,vmx_pin_based_exec_ctrl (vmx)); vmcs_write32(CPU_BASED_VM_EXEC_CONTROL,
vmx_exec_control(vmx));arch/x86/kvm/vmx.c +4
-1@@ -4018,13 +4127,16 @@ static int vmx_vcpu_setup(struct vcpu_vmx *vmx) vmx_secondary_exec_control(vmx)); } -if (enable_apicv_reg_vid) { +if (vmx_vm_has_apicv (vmx->vcpu.kvm)) {vmcs_write64 (EOI_EXIT_BITMAP0 ,0 );vmcs_write64 (EOI_EXIT_BITMAP1 ,0 );vmcs_write64 (EOI_EXIT_BITMAP2 ,0 );vmcs_write64 (EOI_EXIT_BITMAP3 ,0 );vmcs_write16 (GUEST_INTR_STATUS ,0 ); + +vmcs_write64 (POSTED_INTR_NV ,POSTED_INTR_VECTOR ); +vmcs_write64 (POSTED_INTR_DESC_ADDR ,__pa ((&vmx->pi_desc))); } if (ple_gap) { @@ -4174,6 +4286,9 @@ static int vmx_vcpu_reset(struct kvm_vcpu *vcpu) vmcs_write64(APIC_ACCESS_ADDR,
page_to_phys(vmx->vcpu.kvm->arch.apic_access_page));arch/x86/kvm/vmx.c +5
-0+if (vmx_vm_has_apicv (vcpu->kvm)) +memset (&vmx->pi_desc,0 ,sizeof (struct pi_desc )); +if (vmx->vpid !=0 )vmcs_write16 (VIRTUAL_PROCESSOR_ID , vmx->vpid);@@ -7650,6 +7765,8 @@ static struct kvm_x86_ops vmx_x86_ops = { .load_eoi_exitmap = vmx_load_eoi_exitmap, .hwapic_irr_update = vmx_hwapic_irr_update, .hwapic_isr_update = vmx_hwapic_isr_update, + .sync_pir_to_irr = vmx_sync_pir_to_irr, + .deliver_posted_interrupt = vmx_deliver_posted_interrupt, .set_tss_addr = vmx_set_tss_addr, .get_tdp_level = get_ept_level, @@ -7753,7 +7870,7 @@ static int __init vmx_init(void)arch/x86/kvm/vmx.c +1
-1memcpy (vmx_msr_bitmap_longmode_x2apic, vmx_msr_bitmap_longmode,PAGE_SIZE ); -if (enable_apicv_reg_vid) { +if (enable_apicv) {for (msr =0x800 ; msr <=0x8ff ; msr++)vmx_disable_intercept_msr_read_x2apic (msr);arch/x86/kvm/x86.c +0
-1@@ -2679,6 +2679,7 @@ void kvm_arch_vcpu_put(struct kvm_vcpu *vcpu)static int kvm_vcpu_ioctl_get_lapic (struct kvm_vcpu *vcpu,struct kvm_lapic_state *s) { + kvm_x86_ops->sync_pir_to_irr (vcpu);memcpy (s->regs, vcpu->arch.apic->regs,sizeof *s);return 0 ; --
Quoted 1 line
1.7.1
Best regards,
Yang
Quoted 4 lines · Yang Z
Hi Marcelo,
Can you help to review this patch? Many thanks if you can review it quickly.
The patch is only 2 days on the list. Be patient.
Quoted 453 lines
Zhang, Yang Z wrote on 2013-02-19:
Quoted 452 lines
From: Yang Zhang <[email protected]>
Posted Interrupt allows APIC interrupts to inject into guest directly without any vmexit.
- When delivering a interrupt to guest, if target vcpu is running, update Posted-interrupt requests bitmap and send a notification event to the vcpu. Then the vcpu will handle this interrupt automatically, without any software involvemnt.
- If target vcpu is not running or there already a notification event pending in the vcpu, do nothing. The interrupt will be handled by next vm entrySigned-off-by Yang Zhang <[email protected]>--- arch/x86/include/asm/entry_arch.h | 4 + arch/x86/include/asm/hw_irq.h | 1 + arch/x86/include/asm/irq_vectors.h | 5 + arch/x86/include/asm/kvm_host.h | 3 + arch/x86/include/asm/vmx.h | 4 + arch/x86/kernel/entry_64.S | 5 + arch/x86/kernel/irq.c | 20 +++++ arch/x86/kernel/irqinit.c | 4 + arch/x86/kvm/lapic.c | 19 ++++- arch/x86/kvm/lapic.h | 1 + arch/x86/kvm/svm.c | 13 +++ arch/x86/kvm/vmx.c | 157 +++++++++++++++++++++++++++++++----- arch/x86/kvm/x86.c | 1 + 13 files changed, 214 insertions(+), 23deletions(-)
(quoted diff) +0
-0
b/arch/x86/include/asm/entry_arch.hindex 40afa00..9bd4eca 100644 ---a/arch/x86/include/asm/entry_arch.h+++b/arch/x86/include/asm/entry_arch.h@@ -19,6 +19,10 @@BUILD_INTERRUPT(reboot_interrupt,REBOOT_VECTOR)
BUILD_INTERRUPT(x86_platform_ipi, X86_PLATFORM_IPI_VECTOR)
+#ifdefCONFIG_HAVE_KVM+BUILD_INTERRUPT(kvm_posted_intr_ipi,POSTED_INTR_VECTOR) +#endif +
/*
* every pentium local APIC has two 'local interrupts', with a
* soft-definable vector attached to both interrupts, one ofarch/x86/include/asm/hw_irq.h +0
-0@@ -28,6 +28,7 @@/* Interrupt handlers registered during init_IRQ */ extern void apic_timer_interrupt(void); extern void x86_platform_ipi(void); +externvoid kvm_posted_intr_ipi (void );extern void error_interrupt (void );extern void irq_work_interrupt (void );(quoted diff) +0
-0
b/arch/x86/include/asm/irq_vectors.hindex 1508e51..774dc9f 100644 ---a/arch/x86/include/asm/irq_vectors.h+++b/arch/x86/include/asm/irq_vectors.h@@ -102,6 +102,11 @@
*/
#defineX86_PLATFORM_IPI_VECTOR0xf7arch/x86/include/asm/irq_vectors.h +5
-0+/* Vector for KVM to deliver posted interrupt IPI */ +#ifdef CONFIG_HAVE_KVM +#define POSTED_INTR_VECTOR 0xf2 +#endif +/* * IRQ work vector: */ (quoted diff) +0
-0
b/arch/x86/include/asm/kvm_host.hindex b8388e9..79da55e 100644 ---a/arch/x86/include/asm/kvm_host.h+++b/arch/x86/include/asm/kvm_host.harch/x86/include/asm/kvm_host.h +3
-0@@ -704,6 +704,9 @@ struct kvm_x86_ops {void (*hwapic_isr_update)(struct kvm *kvm,int isr);void (*load_eoi_exitmap)(struct kvm_vcpu *vcpu,u64 *eoi_exit_bitmap);void (*set_virtual_x2apic_mode)(struct kvm_vcpu *vcpu,bool set); +bool (*deliver_posted_interrupt)(struct kvm_vcpu *vcpu,int vector, +int *result,bool *delivered); +void (*sync_pir_to_irr)(struct kvm_vcpu *vcpu);int (*set_tss_addr)(struct kvm *kvm,unsigned int addr);int (*get_tdp_level)(void );u64 (*get_mt_mask)(struct kvm_vcpu *vcpu,gfn_t gfn,bool is_mmio);arch/x86/include/asm/vmx.h +1
-0@@ -158,6 +158,7 @@#define PIN_BASED_EXT_INTR_MASK 0x00000001#define PIN_BASED_NMI_EXITING 0x00000008#define PIN_BASED_VIRTUAL_NMIS 0x00000020 +#define PIN_BASED_POSTED_INTR 0x00000080#define VM_EXIT_SAVE_DEBUG_CONTROLS 0x00000002 #define VM_EXIT_HOST_ADDR_SPACE_SIZE 0x00000200 @@ -180,6 +181,7 @@/* VMCS Encodings */ enum
vmcs_field{VIRTUAL_PROCESSOR_ID
=0x00000000, +POSTED_INTR_NV=0x00000002,arch/x86/include/asm/vmx.h +2
-0GUEST_ES_SELECTOR = 0x00000800, GUEST_CS_SELECTOR = 0x00000802, GUEST_SS_SELECTOR = 0x00000804, @@ -214,6 +216,8 @@ enum vmcs_field { VIRTUAL_APIC_PAGE_ADDR_HIGH =0x00002013 ,APIC_ACCESS_ADDR =0x00002014 ,APIC_ACCESS_ADDR_HIGH =0x00002015 , +POSTED_INTR_DESC_ADDR =0x00002016 , +POSTED_INTR_DESC_ADDR_HIGH =0x00002017 ,EPT_POINTER =0x0000201a ,EPT_POINTER_HIGH =0x0000201b ,EOI_EXIT_BITMAP0 =0x0000201c ,arch/x86/kernel/entry_64.S +5
-0@@ -1177,6 +1177,11 @@ apicinterrupt LOCAL_TIMER_VECTOR \ apicinterrupt X86_PLATFORM_IPI_VECTOR \ x86_platform_ipi smp_x86_platform_ipi +#ifdef CONFIG_HAVE_KVM +apicinterrupt POSTED_INTR_VECTOR \ + kvm_posted_intr_ipi smp_posted_intr_ipi +#endif + apicinterrupt THRESHOLD_APIC_VECTOR \ threshold_interrupt smp_threshold_interrupt apicinterrupt THERMAL_APIC_VECTOR \arch/x86/kernel/irq.c +20
-0@@ -228,6 +228,26 @@ void smp_x86_platform_ipi(struct pt_regs *regs)set_irq_regs (old_regs); } +#ifdef CONFIG_HAVE_KVM +/* +* Handler for POSTED_INTERRUPT_VECTOR. +*/ +void smp_posted_intr_ipi (struct pt_regs *regs) +{ +struct pt_regs *old_regs =set_irq_regs (regs); + +ack_APIC_irq (); + +irq_enter (); + +exit_idle (); + +irq_exit (); + +set_irq_regs (old_regs); +} +#endif +EXPORT_SYMBOL_GPL (vector_used_by_percpu_irq);#ifdef CONFIG_HOTPLUG_CPU arch/x86/kernel/irqinit.c +2
-0@@ -205,6 +205,10 @@ static void __init apic_intr_init(void)/* IPI for X86 platform specific use */ alloc_intr_gate (X86_PLATFORM_IPI_VECTOR , x86_platform_ipi); +#ifdef CONFIG_HAVE_KVM +/* IPI for KVM to deliver posted interrupt */ + alloc_intr_gate(POSTED_INTR_VECTOR, kvm_posted_intr_ipi); +#endif/* IPI vectors for APIC spurious and error interrupts */ alloc_intr_gate (SPURIOUS_APIC_VECTOR , spurious_interrupt);arch/x86/kvm/lapic.c +8
-0@@ -357,6 +357,12 @@ static u8 count_vectors(void *bitmap)return count; } +int kvm_apic_test_irr (int vec,struct kvm_lapic *apic) +{ +return apic_test_vector (vec, apic->regs +APIC_IRR ); +} +EXPORT_SYMBOL_GPL (kvm_apic_test_irr); +static inline int apic_test_and_set_irr (int vec,struct kvm_lapic *apic) { apic->irr_pending = true;@@ -379,6 +385,7 @@ static inline int apic_find_highest_irr(struct kvm_lapic *apic)if (!apic->irr_pending)return -1 ; + kvm_x86_ops->sync_pir_to_irr (apic->vcpu); result =apic_search_irr (apic);ASSERT (result ==-1 || result >=16 );@@ -685,6 +692,7 @@ static int __apic_accept_irq(struct kvm_lapic *apic, int delivery_mode, {int result =0 ;struct kvm_vcpu *vcpu = apic->vcpu; +bool delivered = false;switch (delivery_mode) {arch/x86/kvm/lapic.c +5
-2case APIC_DM_LOWEST :@@ -700,7 +708,10 @@ static int __apic_accept_irq(struct kvm_lapic *apic, int delivery_mode, }else apic_clear_vector (vector, apic->regs +APIC_TMR ); - result = !apic_test_and_set_irr (vector, apic); +if (!kvm_x86_ops->deliver_posted_interrupt (vcpu, vector, + &result, &delivered)) + result = !apic_test_and_set_irr (vector, apic); +trace_kvm_apic_accept_irq (vcpu->vcpu_id, delivery_mode, trig_mode, vector, !result);if (!result) {@@ -710,8 +721,10 @@ static int __apic_accept_irq(struct kvm_lapic *apic, int delivery_mode, break; } - kvm_make_request(KVM_REQ_EVENT, vcpu); - kvm_vcpu_kick(vcpu); + if (!delivered) { + kvm_make_request(KVM_REQ_EVENT, vcpu); + kvm_vcpu_kick(vcpu); + }break ;case APIC_DM_REMRD :arch/x86/kvm/lapic.h +1
-0@@ -157,5 +157,6 @@ static inline u16 apic_logical_id(struct kvm_apic_map *map, u32 ldr)void kvm_calculate_eoi_exitmap (struct kvm_vcpu *vcpu,struct kvm_lapic_irq *irq,u64 *eoi_bitmap); +int kvm_apic_test_irr (int vec,struct kvm_lapic *apic);#endif
arch/x86/kvm/svm.c +11
-0@@ -3591,6 +3591,17 @@ static void svm_hwapic_isr_update(struct kvm *kvm, int isr) return; } +static void svm_sync_pir_to_irr (struct kvm_vcpu *vcpu) +{ +return ; +} + +static bool svm_deliver_posted_interrupt (struct kvm_vcpu *vcpu, +int vector,int *result,bool *delivered) +{ +return false; +} + static int svm_nmi_allowed(struct kvm_vcpu *vcpu) { struct vcpu_svm *svm = to_svm(vcpu); @@ -4319,6 +4330,8 @@ static struct kvm_x86_ops svm_x86_ops = { .vm_has_apicv = svm_vm_has_apicv, .load_eoi_exitmap =
svm_load_eoi_exitmap, .hwapic_isr_update =svm_hwapic_isr_update,arch/x86/kvm/svm.c +2
-0+ .sync_pir_to_irr = svm_sync_pir_to_irr, + .deliver_posted_interrupt = svm_deliver_posted_interrupt, .set_tss_addr = svm_set_tss_addr, .get_tdp_level = get_npt_level,arch/x86/kvm/vmx.c +6
-1@@ -84,7 +84,8 @@ module_param(vmm_exclusive, bool, S_IRUGO);static bool __read_mostly fasteoi =1 ;module_param (fasteoi, bool,S_IRUGO ); -static bool __read_mostly enable_apicv_reg_vid; +static bool __read_mostly enable_apicv =1 ; +module_param (enable_apicv, bool,S_IRUGO );/* * If nested=1, nested virtualization is supported, i.e., guests may use @@ -365,6 +366,36 @@ struct nested_vmx {struct page *apic_access_page; }; +#define POSTED_INTR_ON 0 +/* Posted-Interrupt Descriptor */ +struct pi_desc { + u32 pir[8]; /* Posted interrupt requested */ + union { + struct { + u8 on:1, + rsvd:7; + } control; + u32 rsvd[8]; + } u; +} __aligned(64); + +static bool pi_test_and_set_on(struct pi_desc *pi_desc) +{ + return test_and_set_bit(POSTED_INTR_ON, + (unsigned long *)&pi_desc->u.control); +} + +static bool
pi_test_and_clear_on(struct pi_desc *pi_desc)+{ + return
test_and_clear_bit(POSTED_INTR_ON, + (unsigned long
*)&pi_desc->u.control); +} + +static int pi_test_and_set_pir(int vector,
structpi_desc*pi_desc) +{ + return test_and_set_bit(vector, (unsigned
long *)pi_desc->pir); +} +
structvcpu_vmx{arch/x86/kvm/vmx.c +4
-0struct kvm_vcpu vcpu;unsigned long host_rsp;@@ -429,6 +460,9 @@ struct vcpu_vmx {bool rdtscp_enabled; +/* Posted interrupt descriptor */ +struct pi_desc pi_desc; +/* Support for a guest hypervisor (nested VMX) */ struct nested_vmx nested; };@@ -783,6 +817,18 @@ static inline boolcpu_has_vmx_virtual_intr_delivery (void )SECONDARY_EXEC_VIRTUAL_INTR_DELIVERY ; } +static inline bool cpu_has_vmx_posted_intr(void) +{ + returnvmcs_config.pin_based_exec_ctrl &
PIN_BASED_POSTED_INTR; +} + +static
inline boolcpu_has_vmx_apicv(void)+{ + returncpu_has_vmx_apic_register_virt()&&arch/x86/kvm/vmx.c +16
-9+cpu_has_vmx_virtual_intr_delivery () && +cpu_has_vmx_posted_intr (); +} + static inline bool cpu_has_vmx_flexpriority(void) { return cpu_has_vmx_tpr_shadow() &&@@ -2530,12 +2576,6 @@ static __init int setup_vmcs_config(struct vmcs_config *vmcs_conf)u32 _vmexit_control =0 ;u32 _vmentry_control =0 ; - min =PIN_BASED_EXT_INTR_MASK |PIN_BASED_NMI_EXITING ; - opt =PIN_BASED_VIRTUAL_NMIS ; -if (adjust_vmx_controls (min, opt,MSR_IA32_VMX_PINBASED_CTLS , - &_pin_based_exec_control) <0 ) -return -EIO ; - min = CPU_BASED_HLT_EXITING |#ifdef CONFIG_X86_64 CPU_BASED_CR8_LOAD_EXITING |@@ -2612,6 +2652,17 @@ static __init int setup_vmcs_config(struct vmcs_config *vmcs_conf) &_vmexit_control) < 0)return -EIO ; + min =PIN_BASED_EXT_INTR_MASK |PIN_BASED_NMI_EXITING ; + opt =PIN_BASED_VIRTUAL_NMIS |PIN_BASED_POSTED_INTR ; +if (adjust_vmx_controls (min, opt,MSR_IA32_VMX_PINBASED_CTLS , + &_pin_based_exec_control) <0 ) +return -EIO ; + +if (!(_cpu_based_2nd_exec_control & +SECONDARY_EXEC_VIRTUAL_INTR_DELIVERY ) || + !(_vmexit_control &VM_EXIT_ACK_INTR_ON_EXIT )) + _pin_based_exec_control &= ~PIN_BASED_POSTED_INTR ; + min = 0; opt = VM_ENTRY_LOAD_IA32_PAT; if (adjust_vmx_controls(min, opt, MSR_IA32_VMX_ENTRY_CTLS, @@ -2790,11 +2841,10 @@ static __init int hardware_setup(void) if (!cpu_has_vmx_ple()) ple_gap = 0; - if (!cpu_has_vmx_apic_register_virt() || - !cpu_has_vmx_virtual_intr_delivery()) - enable_apicv_reg_vid = 0; + if (!cpu_has_vmx_apicv()) + enable_apicv = 0; - if (enable_apicv_reg_vid) + if (enable_apicv) kvm_x86_ops->update_cr8_intercept = NULL; else kvm_x86_ops->hwapic_irr_update = NULL;@@ -3871,6 +3921,62 @@ static void vmx_disable_intercept_msr_write_x2apic(u32 msr) msr, MSR_TYPE_W); } +static int vmx_vm_has_apicv(struct kvm *kvm) +{ + return enable_apicv&&
irqchip_in_kernel(kvm); +} + +static bool
vmx_deliver_posted_interrupt(structkvm_vcpu*vcpu, + int vector, int
*result, bool *delivered) +{ + structvcpu_vmx*vmx =to_vmx(vcpu); +arch/x86/kvm/vmx.c +1
-0+ if (!vmx_vm_has_apicv(vcpu->kvm)) + return false; + + if(
kvm_apic_test_irr(vector, vcpu->arch.apic)) + goto out; + else {arch/x86/kvm/vmx.c +1
-0+ *result = !pi_test_and_set_pir (vector, &vmx->pi_desc); + if(!*result) + goto out; + } + + if (!pi_test_and_set_on(&vmx->pi_desc)
&& + (vcpu->mode==IN_GUEST_MODE)) {arch/x86/kvm/vmx.c +2
-0+kvm_make_request (KVM_REQ_EVENT , vcpu); + apic->send_IPI_mask(get_cpu_mask(vcpu->cpu), +
POSTED_INTR_VECTOR); + *delivered = true; + } +out: + return true; +} +
+static voidvmx_sync_pir_to_irr(struct kvm_vcpu *vcpu)+{ + structvcpu_vmx*vmx =to_vmx(vcpu); + structkvm_lapic*apic =vcpu->arch.apic; + unsigned int i, old, new,ret_val,irr_offset,pir_val; + + if (!vmx_vm_has_apicv(vcpu->kvm) ||arch/x86/kvm/vmx.c +1
-0+ !pi_test_and_clear_on(&vmx->pi_desc)) + return; + + for (i = 0; i<= 7; i++) { +
pir_val=xchg(&vmx->pi_desc.pir[i], 0); + if (pir_val)
{ +irr_offset=APIC_IRR+ i *0x10; + do { + old =kvm_apic_get_reg(apic, irr_offset); + new = old |pir_val;arch/x86/kvm/vmx.c +1
-0+ ret_val = cmpxchg((u32 *)(apic->regs + + irr_offset), old,new); + } while (
unlikely(ret_val != old)); + } + } +} +
/*
* Set up the vmcs's constant host-state fields, i.e., host-state fields that
* will not change in the lifetime of the guest.arch/x86/kvm/vmx.c +10
-7@@ -3931,6 +4037,15 @@ static void set_cr4_guest_host_mask(struct vcpu_vmx *vmx)vmcs_writel (CR4_GUEST_HOST_MASK , ~vmx->vcpu.arch.cr4_guest_owned_bits); } +static u32 vmx_pin_based_exec_ctrl (struct vcpu_vmx *vmx) +{ +u32 pin_based_exec_ctrl = vmcs_config.pin_based_exec_ctrl; + +if (!vmx_vm_has_apicv (vmx->vcpu.kvm)) + pin_based_exec_ctrl &= ~PIN_BASED_POSTED_INTR ; +return pin_based_exec_ctrl; +} +static u32 vmx_exec_control (struct vcpu_vmx *vmx) {u32 exec_control = vmcs_config.cpu_based_exec_ctrl; @@ -3948,11 +4063,6 @@ static u32 vmx_exec_control(struct vcpu_vmx *vmx) return exec_control; } -static int vmx_vm_has_apicv (struct kvm *kvm) -{ -return enable_apicv_reg_vid &&irqchip_in_kernel (kvm); -} -static u32 vmx_secondary_exec_control (struct vcpu_vmx *vmx) {u32 exec_control = vmcs_config.cpu_based_2nd_exec_ctrl; @@ -4008,8 +4118,7 @@ static int vmx_vcpu_setup(struct vcpu_vmx *vmx) vmcs_write64(VMCS_LINK_POINTER, -1ull); /* 22.3.1.5 *//* Control */ -vmcs_write32 (PIN_BASED_VM_EXEC_CONTROL , - vmcs_config.pin_based_exec_ctrl); +vmcs_write32 (PIN_BASED_VM_EXEC_CONTROL ,vmx_pin_based_exec_ctrl (vmx)); vmcs_write32(CPU_BASED_VM_EXEC_CONTROL,
vmx_exec_control(vmx));arch/x86/kvm/vmx.c +4
-1@@ -4018,13 +4127,16 @@ static int vmx_vcpu_setup(struct vcpu_vmx *vmx) vmx_secondary_exec_control(vmx)); } -if (enable_apicv_reg_vid) { +if (vmx_vm_has_apicv (vmx->vcpu.kvm)) {vmcs_write64 (EOI_EXIT_BITMAP0 ,0 );vmcs_write64 (EOI_EXIT_BITMAP1 ,0 );vmcs_write64 (EOI_EXIT_BITMAP2 ,0 );vmcs_write64 (EOI_EXIT_BITMAP3 ,0 );vmcs_write16 (GUEST_INTR_STATUS ,0 ); + +vmcs_write64 (POSTED_INTR_NV ,POSTED_INTR_VECTOR ); +vmcs_write64 (POSTED_INTR_DESC_ADDR ,__pa ((&vmx->pi_desc))); } if (ple_gap) { @@ -4174,6 +4286,9 @@ static int vmx_vcpu_reset(struct kvm_vcpu *vcpu) vmcs_write64(APIC_ACCESS_ADDR,
page_to_phys(vmx->vcpu.kvm->arch.apic_access_page));arch/x86/kvm/vmx.c +5
-0+if (vmx_vm_has_apicv (vcpu->kvm)) +memset (&vmx->pi_desc,0 ,sizeof (struct pi_desc )); +if (vmx->vpid !=0 )vmcs_write16 (VIRTUAL_PROCESSOR_ID , vmx->vpid);@@ -7650,6 +7765,8 @@ static struct kvm_x86_ops vmx_x86_ops = { .load_eoi_exitmap = vmx_load_eoi_exitmap, .hwapic_irr_update = vmx_hwapic_irr_update, .hwapic_isr_update = vmx_hwapic_isr_update, + .sync_pir_to_irr = vmx_sync_pir_to_irr, + .deliver_posted_interrupt = vmx_deliver_posted_interrupt, .set_tss_addr = vmx_set_tss_addr, .get_tdp_level = get_ept_level, @@ -7753,7 +7870,7 @@ static int __init vmx_init(void)arch/x86/kvm/vmx.c +1
-1memcpy (vmx_msr_bitmap_longmode_x2apic, vmx_msr_bitmap_longmode,PAGE_SIZE ); -if (enable_apicv_reg_vid) { +if (enable_apicv) {for (msr =0x800 ; msr <=0x8ff ; msr++)vmx_disable_intercept_msr_read_x2apic (msr);arch/x86/kvm/x86.c +0
-0@@ -2679,6 +2679,7 @@ void kvm_arch_vcpu_put(struct kvm_vcpu *vcpu)static int kvm_vcpu_ioctl_get_lapic (struct kvm_vcpu *vcpu,struct kvm_lapic_state *s) { + kvm_x86_ops->sync_pir_to_irr (vcpu);memcpy (s->regs, vcpu->arch.apic->regs,sizeof *s);
Quoted 8 lines
Quoted 3 lines
arch/x86/kvm/x86.c +0
-1return 0 ; --1.7.1
Best regards,
Yang
--
Gleb.