[PATCH 0/2] x86/kvm/vmx: Fix VMX interrupt injection vs hrtimer_rearm_deferred()

[PATCH 0/2] x86/kvm/vmx: Fix VMX interrupt injection vs hrtimer_rearm_deferred()

[Peter Zijlstra]

Hi,

Nicely split out in two patches, but otherwise unchanged (except for a small
comment).

[PATCH 1/2] x86/kvm/vmx: Move IRQ/NMI dispatch from KVM into x86 core

[Peter Zijlstra]

Move the VMX interrupt dispatch magic into the x86 core code. This
isolates KVM from the FRED/IDT decisions and reduces the amount of
EXPORT_SYMBOL_FOR_KVM().

Suggested-by: Sean Christopherson <seanjc@google.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: "Verma, Vishal L" <vishal.l.verma@intel.com>
---
 arch/x86/entry/Makefile             |    2 -
 arch/x86/entry/common.c             |   48 ++++++++++++++++++++++++++++++++++++
 arch/x86/entry/entry.S              |   46 ++++++++++++++++++++++++++++++++++
 arch/x86/entry/entry_64_fred.S      |    1 
 arch/x86/include/asm/desc.h         |    4 +++
 arch/x86/include/asm/desc_defs.h    |    2 -
 arch/x86/include/asm/entry-common.h |    2 +
 arch/x86/include/asm/fred.h         |    1 
 arch/x86/include/asm/idtentry.h     |   11 --------
 arch/x86/kernel/idt.c               |   13 +++++++++
 arch/x86/kernel/nmi.c               |    8 ------
 arch/x86/kvm/vmx/vmenter.S          |   46 ----------------------------------
 arch/x86/kvm/vmx/vmx.c              |   20 ++-------------
 13 files changed, 118 insertions(+), 86 deletions(-)

--- a/arch/x86/entry/Makefile
+++ b/arch/x86/entry/Makefile
@@ -13,7 +13,7 @@ CFLAGS_REMOVE_syscall_64.o	= $(CC_FLAGS_
 CFLAGS_syscall_32.o		+= -fno-stack-protector
 CFLAGS_syscall_64.o		+= -fno-stack-protector
 
-obj-y				:= entry.o entry_$(BITS).o syscall_$(BITS).o
+obj-y				:= entry.o entry_$(BITS).o syscall_$(BITS).o common.o
 
 obj-y				+= vdso/
 obj-y				+= vsyscall/
--- /dev/null
+++ b/arch/x86/entry/common.c
@@ -0,0 +1,48 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+
+#include <linux/entry-common.h>
+#include <linux/kvm_types.h>
+#include <asm/fred.h>
+#include <asm/desc.h>
+
+#if IS_ENABLED(CONFIG_KVM_INTEL)
+/*
+ * On VMX, NMIs and IRQs (as configured by KVM) are acknowledge by hardware as
+ * part of the VM-Exit, i.e. the event itself is consumed as part the VM-Exit.
+ * x86_entry_from_kvm() is invoked by KVM to effectively forward NMIs and IRQs
+ * to the kernel for servicing.  On SVM, a.k.a. AMD, the NMI/IRQ VM-Exit is
+ * purely a signal that an NMI/IRQ is pending, i.e. the event that triggered
+ * the VM-Exit is held pending until it's unblocked in the host.
+ */
+noinstr void x86_entry_from_kvm(unsigned int event_type, unsigned int vector)
+{
+	if (event_type == EVENT_TYPE_EXTINT) {
+#ifdef CONFIG_X86_64
+		/*
+		 * Use FRED dispatch, even when running IDT. The dispatch
+		 * tables are kept in sync between FRED and IDT, and the FRED
+		 * dispatch works well with CFI.
+		 */
+		fred_entry_from_kvm(event_type, vector);
+#else
+		idt_entry_from_kvm(vector);
+#endif
+		return;
+	}
+
+	WARN_ON_ONCE(event_type != EVENT_TYPE_NMI);
+
+#ifdef CONFIG_X86_64
+	if (cpu_feature_enabled(X86_FEATURE_FRED))
+		return fred_entry_from_kvm(event_type, vector);
+#endif
+
+	/*
+	 * Notably, we must use IDT dispatch for NMI when running in IDT mode.
+	 * The FRED NMI context is significantly different and will not work
+	 * right (speficially FRED fixed the NMI recursion issue).
+	 */
+	idt_entry_from_kvm(vector);
+}
+EXPORT_SYMBOL_FOR_KVM(x86_entry_from_kvm);
+#endif
--- a/arch/x86/entry/entry.S
+++ b/arch/x86/entry/entry.S
@@ -75,3 +75,49 @@ THUNK warn_thunk_thunk, __warn_thunk
 #if defined(CONFIG_STACKPROTECTOR) && defined(CONFIG_SMP)
 EXPORT_SYMBOL(__ref_stack_chk_guard);
 #endif
+
+#if IS_ENABLED(CONFIG_KVM_INTEL)
+.macro IDT_DO_EVENT_IRQOFF call_insn call_target
+	/*
+	 * Unconditionally create a stack frame, getting the correct RSP on the
+	 * stack (for x86-64) would take two instructions anyways, and RBP can
+	 * be used to restore RSP to make objtool happy (see below).
+	 */
+	push %_ASM_BP
+	mov %_ASM_SP, %_ASM_BP
+
+#ifdef CONFIG_X86_64
+	/*
+	 * Align RSP to a 16-byte boundary (to emulate CPU behavior) before
+	 * creating the synthetic interrupt stack frame for the IRQ/NMI.
+	 */
+	and  $-16, %rsp
+	push $__KERNEL_DS
+	push %rbp
+#endif
+	pushf
+	push $__KERNEL_CS
+	\call_insn \call_target
+
+	/*
+	 * "Restore" RSP from RBP, even though IRET has already unwound RSP to
+	 * the correct value.  objtool doesn't know the callee will IRET and,
+	 * without the explicit restore, thinks the stack is getting walloped.
+	 * Using an unwind hint is problematic due to x86-64's dynamic alignment.
+	 */
+	leave
+	RET
+.endm
+
+.pushsection .text, "ax"
+SYM_FUNC_START(idt_do_interrupt_irqoff)
+	IDT_DO_EVENT_IRQOFF CALL_NOSPEC _ASM_ARG1
+SYM_FUNC_END(idt_do_interrupt_irqoff)
+.popsection
+
+.pushsection .noinstr.text, "ax"
+SYM_FUNC_START(idt_do_nmi_irqoff)
+	IDT_DO_EVENT_IRQOFF call asm_exc_nmi
+SYM_FUNC_END(idt_do_nmi_irqoff)
+.popsection
+#endif
--- a/arch/x86/entry/entry_64_fred.S
+++ b/arch/x86/entry/entry_64_fred.S
@@ -147,5 +147,4 @@ SYM_FUNC_START(asm_fred_entry_from_kvm)
 	RET
 
 SYM_FUNC_END(asm_fred_entry_from_kvm)
-EXPORT_SYMBOL_FOR_KVM(asm_fred_entry_from_kvm);
 #endif
--- a/arch/x86/include/asm/desc.h
+++ b/arch/x86/include/asm/desc.h
@@ -438,6 +438,10 @@ extern void idt_setup_traps(void);
 extern void idt_setup_apic_and_irq_gates(void);
 extern bool idt_is_f00f_address(unsigned long address);
 
+extern void idt_do_interrupt_irqoff(unsigned int vector);
+extern void idt_do_nmi_irqoff(void);
+extern void idt_entry_from_kvm(unsigned int vector);
+
 #ifdef CONFIG_X86_64
 extern void idt_setup_early_pf(void);
 #else
--- a/arch/x86/include/asm/desc_defs.h
+++ b/arch/x86/include/asm/desc_defs.h
@@ -145,7 +145,7 @@ struct gate_struct {
 typedef struct gate_struct gate_desc;
 
 #ifndef _SETUP
-static inline unsigned long gate_offset(const gate_desc *g)
+static __always_inline unsigned long gate_offset(const gate_desc *g)
 {
 #ifdef CONFIG_X86_64
 	return g->offset_low | ((unsigned long)g->offset_middle << 16) |
--- a/arch/x86/include/asm/entry-common.h
+++ b/arch/x86/include/asm/entry-common.h
@@ -97,4 +97,6 @@ static __always_inline void arch_exit_to
 }
 #define arch_exit_to_user_mode arch_exit_to_user_mode
 
+extern void x86_entry_from_kvm(unsigned int entry_type, unsigned int vector);
+
 #endif
--- a/arch/x86/include/asm/fred.h
+++ b/arch/x86/include/asm/fred.h
@@ -110,7 +110,6 @@ static __always_inline unsigned long fre
 static inline void cpu_init_fred_exceptions(void) { }
 static inline void cpu_init_fred_rsps(void) { }
 static inline void fred_complete_exception_setup(void) { }
-static inline void fred_entry_from_kvm(unsigned int type, unsigned int vector) { }
 static inline void fred_sync_rsp0(unsigned long rsp0) { }
 static inline void fred_update_rsp0(void) { }
 #endif /* CONFIG_X86_FRED */
--- a/arch/x86/include/asm/idtentry.h
+++ b/arch/x86/include/asm/idtentry.h
@@ -633,17 +633,6 @@ DECLARE_IDTENTRY_RAW(X86_TRAP_MC,	xenpv_
 #endif
 
 /* NMI */
-
-#if IS_ENABLED(CONFIG_KVM_INTEL)
-/*
- * Special entry point for VMX which invokes this on the kernel stack, even for
- * 64-bit, i.e. without using an IST.  asm_exc_nmi() requires an IST to work
- * correctly vs. the NMI 'executing' marker.  Used for 32-bit kernels as well
- * to avoid more ifdeffery.
- */
-DECLARE_IDTENTRY(X86_TRAP_NMI,		exc_nmi_kvm_vmx);
-#endif
-
 DECLARE_IDTENTRY_NMI(X86_TRAP_NMI,	exc_nmi);
 #ifdef CONFIG_XEN_PV
 DECLARE_IDTENTRY_RAW(X86_TRAP_NMI,	xenpv_exc_nmi);
--- a/arch/x86/kernel/idt.c
+++ b/arch/x86/kernel/idt.c
@@ -268,6 +268,19 @@ void __init idt_setup_early_pf(void)
 }
 #endif
 
+noinstr void idt_entry_from_kvm(unsigned int vector)
+{
+	if (vector == NMI_VECTOR)
+		return idt_do_nmi_irqoff();
+
+	/*
+	 * Only the NMI path requires noinstr.
+	 */
+	instrumentation_begin();
+	idt_do_interrupt_irqoff(gate_offset(idt_table + vector));
+	instrumentation_end();
+}
+
 static void __init idt_map_in_cea(void)
 {
 	/*
--- a/arch/x86/kernel/nmi.c
+++ b/arch/x86/kernel/nmi.c
@@ -609,14 +609,6 @@ DEFINE_IDTENTRY_RAW(exc_nmi)
 		goto nmi_restart;
 }
 
-#if IS_ENABLED(CONFIG_KVM_INTEL)
-DEFINE_IDTENTRY_RAW(exc_nmi_kvm_vmx)
-{
-	exc_nmi(regs);
-}
-EXPORT_SYMBOL_FOR_KVM(asm_exc_nmi_kvm_vmx);
-#endif
-
 #ifdef CONFIG_NMI_CHECK_CPU
 
 static char *nmi_check_stall_msg[] = {
--- a/arch/x86/kvm/vmx/vmenter.S
+++ b/arch/x86/kvm/vmx/vmenter.S
@@ -31,38 +31,6 @@
 #define VCPU_R15	__VCPU_REGS_R15 * WORD_SIZE
 #endif
 
-.macro VMX_DO_EVENT_IRQOFF call_insn call_target
-	/*
-	 * Unconditionally create a stack frame, getting the correct RSP on the
-	 * stack (for x86-64) would take two instructions anyways, and RBP can
-	 * be used to restore RSP to make objtool happy (see below).
-	 */
-	push %_ASM_BP
-	mov %_ASM_SP, %_ASM_BP
-
-#ifdef CONFIG_X86_64
-	/*
-	 * Align RSP to a 16-byte boundary (to emulate CPU behavior) before
-	 * creating the synthetic interrupt stack frame for the IRQ/NMI.
-	 */
-	and  $-16, %rsp
-	push $__KERNEL_DS
-	push %rbp
-#endif
-	pushf
-	push $__KERNEL_CS
-	\call_insn \call_target
-
-	/*
-	 * "Restore" RSP from RBP, even though IRET has already unwound RSP to
-	 * the correct value.  objtool doesn't know the callee will IRET and,
-	 * without the explicit restore, thinks the stack is getting walloped.
-	 * Using an unwind hint is problematic due to x86-64's dynamic alignment.
-	 */
-	leave
-	RET
-.endm
-
 .section .noinstr.text, "ax"
 
 /**
@@ -320,10 +288,6 @@ SYM_INNER_LABEL_ALIGN(vmx_vmexit, SYM_L_
 
 SYM_FUNC_END(__vmx_vcpu_run)
 
-SYM_FUNC_START(vmx_do_nmi_irqoff)
-	VMX_DO_EVENT_IRQOFF call asm_exc_nmi_kvm_vmx
-SYM_FUNC_END(vmx_do_nmi_irqoff)
-
 #ifndef CONFIG_CC_HAS_ASM_GOTO_OUTPUT
 
 /**
@@ -375,13 +339,3 @@ SYM_FUNC_START(vmread_error_trampoline)
 	RET
 SYM_FUNC_END(vmread_error_trampoline)
 #endif
-
-.section .text, "ax"
-
-#ifndef CONFIG_X86_FRED
-
-SYM_FUNC_START(vmx_do_interrupt_irqoff)
-	VMX_DO_EVENT_IRQOFF CALL_NOSPEC _ASM_ARG1
-SYM_FUNC_END(vmx_do_interrupt_irqoff)
-
-#endif
--- a/arch/x86/kvm/vmx/vmx.c
+++ b/arch/x86/kvm/vmx/vmx.c
@@ -7083,9 +7083,6 @@ void vmx_load_eoi_exitmap(struct kvm_vcp
 	vmcs_write64(EOI_EXIT_BITMAP3, eoi_exit_bitmap[3]);
 }
 
-void vmx_do_interrupt_irqoff(unsigned long entry);
-void vmx_do_nmi_irqoff(void);
-
 static void handle_nm_fault_irqoff(struct kvm_vcpu *vcpu)
 {
 	/*
@@ -7127,17 +7124,9 @@ static void handle_external_interrupt_ir
 	    "unexpected VM-Exit interrupt info: 0x%x", intr_info))
 		return;
 
-	/*
-	 * Invoke the kernel's IRQ handler for the vector.  Use the FRED path
-	 * when it's available even if FRED isn't fully enabled, e.g. even if
-	 * FRED isn't supported in hardware, in order to avoid the indirect
-	 * CALL in the non-FRED path.
-	 */
+	/* For the IRQ to the core kernel for processing. */
 	kvm_before_interrupt(vcpu, KVM_HANDLING_IRQ);
-	if (IS_ENABLED(CONFIG_X86_FRED))
-		fred_entry_from_kvm(EVENT_TYPE_EXTINT, vector);
-	else
-		vmx_do_interrupt_irqoff(gate_offset((gate_desc *)host_idt_base + vector));
+	x86_entry_from_kvm(EVENT_TYPE_EXTINT, vector);
 	kvm_after_interrupt(vcpu);
 
 	vcpu->arch.at_instruction_boundary = true;
@@ -7447,10 +7436,7 @@ noinstr void vmx_handle_nmi(struct kvm_v
 		return;
 
 	kvm_before_interrupt(vcpu, KVM_HANDLING_NMI);
-	if (cpu_feature_enabled(X86_FEATURE_FRED))
-		fred_entry_from_kvm(EVENT_TYPE_NMI, NMI_VECTOR);
-	else
-		vmx_do_nmi_irqoff();
+	x86_entry_from_kvm(EVENT_TYPE_NMI, NMI_VECTOR);
 	kvm_after_interrupt(vcpu);
 }
 

[PATCH 2/2] x86/kvm/vmx: Fix VMX vs hrtimer_rearm_deferred()

[Peter Zijlstra]

Vishal reported that KVM unit test 'x2apic' started failing after commit
0e98eb14814e ("entry: Prepare for deferred hrtimer rearming").

The reason is that KVM/VMX is injecting interrupts while it has IRQs disabled,
for a context that will enable IRQs, this means that regs->flags.X86_EFLAGS_IF
== 0 and the irqentry_exit() will not DTRT.

Notably, irqentry_exit() must not call hrtimer_rearm_deferred() when the return
context does not have IF set, because this will cause problems vs NMIs.

Therefore, fix up the state after the injection.

Fixes: 0e98eb14814e ("entry: Prepare for deferred hrtimer rearming")
Reported-by: "Verma, Vishal L" <vishal.l.verma@intel.com>
Closes: https://lore.kernel.org/r/70cd3e97fbb796e2eb2ff8cd4b7614ada05a5f24.camel%40intel.com
Suggested-by: Thomas Gleixner <tglx@kernel.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: "Verma, Vishal L" <vishal.l.verma@intel.com>
---
 arch/x86/entry/common.c |   13 +++++++++++++
 1 file changed, 13 insertions(+)

--- a/arch/x86/entry/common.c
+++ b/arch/x86/entry/common.c
@@ -2,6 +2,7 @@
 
 #include <linux/entry-common.h>
 #include <linux/kvm_types.h>
+#include <linux/hrtimer_rearm.h>
 #include <asm/fred.h>
 #include <asm/desc.h>
 
@@ -27,6 +28,18 @@ noinstr void x86_entry_from_kvm(unsigned
 #else
 		idt_entry_from_kvm(vector);
 #endif
+		/*
+		 * Strictly speaking, only the NMI path requires noinstr.
+		 */
+		instrumentation_begin();
+		/*
+		 * KVM/VMX will dispatch from IRQ-disabled but for a context
+		 * that will have IRQs-enabled. This confuses the entry code
+		 * and it will not have reprogrammed the timer. Do so now.
+		 */
+		hrtimer_rearm_deferred();
+		instrumentation_end();
+
 		return;
 	}
 

Re: [PATCH 1/2] x86/kvm/vmx: Move IRQ/NMI dispatch from KVM into x86 core

[Xin Li]

Few nits below, but

Reviewed-by: Xin Li <xin@zytor.com <mailto:xin@zytor.com>>


> On Apr 23, 2026, at 8:56 AM, Peter Zijlstra <peterz@infradead.org> wrote:
> 
> Move the VMX interrupt dispatch magic into the x86 core code. This
> isolates KVM from the FRED/IDT decisions and reduces the amount of
> EXPORT_SYMBOL_FOR_KVM().
> 
> Suggested-by: Sean Christopherson <seanjc@google.com>
> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
> Tested-by: "Verma, Vishal L" <vishal.l.verma@intel.com>
> --- /dev/null
> +++ b/arch/x86/entry/common.c
> @@ -0,0 +1,48 @@
> +/* SPDX-License-Identifier: GPL-2.0 */
> +
> +#include <linux/entry-common.h>
> +#include <linux/kvm_types.h>
> +#include <asm/fred.h>
> +#include <asm/desc.h>

Swap the two header inclusion order?

> +
> +#if IS_ENABLED(CONFIG_KVM_INTEL)
> +/*
> + * On VMX, NMIs and IRQs (as configured by KVM) are acknowledge by hardware as
> + * part of the VM-Exit, i.e. the event itself is consumed as part the VM-Exit.
> + * x86_entry_from_kvm() is invoked by KVM to effectively forward NMIs and IRQs
> + * to the kernel for servicing.  On SVM, a.k.a. AMD, the NMI/IRQ VM-Exit is
> + * purely a signal that an NMI/IRQ is pending, i.e. the event that triggered
> + * the VM-Exit is held pending until it's unblocked in the host.
> + */
> +noinstr void x86_entry_from_kvm(unsigned int event_type, unsigned int vector)

Is u32 better than unsigned int as is close to hardware definition?

Currently, the event type and event vector are defined as 4-bit and 8-bit
fields, respectively; and they can be readily expanded to 8 and 16 bits by
utilizing adjacent reserved bits.

> +{
> + if (event_type == EVENT_TYPE_EXTINT) {
> +#ifdef CONFIG_X86_64
> + /*
> + * Use FRED dispatch, even when running IDT. The dispatch
> + * tables are kept in sync between FRED and IDT, and the FRED
> + * dispatch works well with CFI.
> + */
> + fred_entry_from_kvm(event_type, vector);

Then should we rename the “fred_” prefix, say “x86_64_"?

Sorry naming is hard to me ;)