KVM: x86: Fully defer to vendor code to decide how to force immediate exit

[ Upstream commit 0ec3d6d1f169baa7fc512ae4b78d17e7c94b7763 ]

Now that vmx->req_immediate_exit is used only in the scope of
vmx_vcpu_run(), use force_immediate_exit to detect that KVM should usurp
the VMX preemption to force a VM-Exit and let vendor code fully handle
forcing a VM-Exit.

Opportunsitically drop __kvm_request_immediate_exit() and just have
vendor code call smp_send_reschedule() directly.  SVM already does this
when injecting an event while also trying to single-step an IRET, i.e.
it's not exactly secret knowledge that KVM uses a reschedule IPI to force
an exit.

Link: https://lore.kernel.org/r/20240110012705.506918-7-seanjc@google.com
Signed-off-by: Sean Christopherson <seanjc@google.com>
[sean: resolve absurd conflict due to funky kvm_x86_ops.sched_in prototype]
Signed-off-by: Sean Christopherson <seanjc@google.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

arch/x86/include/asm/kvm-x86-ops.h

@@ -102,7 +102,6 @@ KVM_X86_OP(write_tsc_multiplier)
 KVM_X86_OP(get_exit_info)
 KVM_X86_OP(check_intercept)
 KVM_X86_OP(handle_exit_irqoff)
-KVM_X86_OP(request_immediate_exit)
 KVM_X86_OP(sched_in)
 KVM_X86_OP_OPTIONAL(update_cpu_dirty_logging)
 KVM_X86_OP_OPTIONAL(vcpu_blocking)

arch/x86/include/asm/kvm_host.h

@@ -1695,8 +1695,6 @@ struct kvm_x86_ops {
 			       struct x86_exception *exception);
 	void (*handle_exit_irqoff)(struct kvm_vcpu *vcpu);
 
-	void (*request_immediate_exit)(struct kvm_vcpu *vcpu);
-
 	void (*sched_in)(struct kvm_vcpu *kvm, int cpu);
 
 	/*
@@ -2182,7 +2180,6 @@ extern bool kvm_find_async_pf_gfn(struct kvm_vcpu *vcpu, gfn_t gfn);
 
 int kvm_skip_emulated_instruction(struct kvm_vcpu *vcpu);
 int kvm_complete_insn_gp(struct kvm_vcpu *vcpu, int err);
-void __kvm_request_immediate_exit(struct kvm_vcpu *vcpu);
 
 void __user *__x86_set_memory_region(struct kvm *kvm, int id, gpa_t gpa,
 				     u32 size);

arch/x86/kvm/svm/svm.c

@@ -4222,9 +4222,12 @@ static __no_kcsan fastpath_t svm_vcpu_run(struct kvm_vcpu *vcpu,
 		 * is enough to force an immediate vmexit.
 		 */
 		disable_nmi_singlestep(svm);
-		smp_send_reschedule(vcpu->cpu);
+		force_immediate_exit = true;
 	}
 
+	if (force_immediate_exit)
+		smp_send_reschedule(vcpu->cpu);
+
 	pre_svm_run(vcpu);
 
 	sync_lapic_to_cr8(vcpu);
@@ -5075,8 +5078,6 @@ static struct kvm_x86_ops svm_x86_ops __initdata = {
 	.check_intercept = svm_check_intercept,
 	.handle_exit_irqoff = svm_handle_exit_irqoff,
 
-	.request_immediate_exit = __kvm_request_immediate_exit,
-
 	.sched_in = svm_sched_in,
 
 	.nested_ops = &svm_nested_ops,

arch/x86/kvm/vmx/vmx.c

@@ -49,6 +49,8 @@
 #include <asm/spec-ctrl.h>
 #include <asm/vmx.h>
 
+#include <trace/events/ipi.h>
+
 #include "capabilities.h"
 #include "cpuid.h"
 #include "hyperv.h"
@@ -1304,8 +1306,6 @@ void vmx_prepare_switch_to_guest(struct kvm_vcpu *vcpu)
 	u16 fs_sel, gs_sel;
 	int i;
 
-	vmx->req_immediate_exit = false;
-
 	/*
 	 * Note that guest MSRs to be saved/restored can also be changed
 	 * when guest state is loaded. This happens when guest transitions
@@ -6015,7 +6015,8 @@ static int handle_pml_full(struct kvm_vcpu *vcpu)
 	return 1;
 }
 
-static fastpath_t handle_fastpath_preemption_timer(struct kvm_vcpu *vcpu)
+static fastpath_t handle_fastpath_preemption_timer(struct kvm_vcpu *vcpu,
+						   bool force_immediate_exit)
 {
 	struct vcpu_vmx *vmx = to_vmx(vcpu);
 
@@ -6031,7 +6032,7 @@ static fastpath_t handle_fastpath_preemption_timer(struct kvm_vcpu *vcpu)
 	 * If the timer expired because KVM used it to force an immediate exit,
 	 * then mission accomplished.
 	 */
-	if (vmx->req_immediate_exit)
+	if (force_immediate_exit)
 		return EXIT_FASTPATH_EXIT_HANDLED;
 
 	/*
@@ -7210,13 +7211,13 @@ static void atomic_switch_perf_msrs(struct vcpu_vmx *vmx)
 					msrs[i].host, false);
 }
 
-static void vmx_update_hv_timer(struct kvm_vcpu *vcpu)
+static void vmx_update_hv_timer(struct kvm_vcpu *vcpu, bool force_immediate_exit)
 {
 	struct vcpu_vmx *vmx = to_vmx(vcpu);
 	u64 tscl;
 	u32 delta_tsc;
 
-	if (vmx->req_immediate_exit) {
+	if (force_immediate_exit) {
 		vmcs_write32(VMX_PREEMPTION_TIMER_VALUE, 0);
 		vmx->loaded_vmcs->hv_timer_soft_disabled = false;
 	} else if (vmx->hv_deadline_tsc != -1) {
@@ -7269,7 +7270,8 @@ void noinstr vmx_spec_ctrl_restore_host(struct vcpu_vmx *vmx,
 	barrier_nospec();
 }
 
-static fastpath_t vmx_exit_handlers_fastpath(struct kvm_vcpu *vcpu)
+static fastpath_t vmx_exit_handlers_fastpath(struct kvm_vcpu *vcpu,
+					     bool force_immediate_exit)
 {
 	/*
 	 * If L2 is active, some VMX preemption timer exits can be handled in
@@ -7283,7 +7285,7 @@ static fastpath_t vmx_exit_handlers_fastpath(struct kvm_vcpu *vcpu)
 	case EXIT_REASON_MSR_WRITE:
 		return handle_fastpath_set_msr_irqoff(vcpu);
 	case EXIT_REASON_PREEMPTION_TIMER:
-		return handle_fastpath_preemption_timer(vcpu);
+		return handle_fastpath_preemption_timer(vcpu, force_immediate_exit);
 	default:
 		return EXIT_FASTPATH_NONE;
 	}
@@ -7425,7 +7427,9 @@ static fastpath_t vmx_vcpu_run(struct kvm_vcpu *vcpu, bool force_immediate_exit)
 		vmx_passthrough_lbr_msrs(vcpu);
 
 	if (enable_preemption_timer)
-		vmx_update_hv_timer(vcpu);
+		vmx_update_hv_timer(vcpu, force_immediate_exit);
+	else if (force_immediate_exit)
+		smp_send_reschedule(vcpu->cpu);
 
 	kvm_wait_lapic_expire(vcpu);
 
@@ -7489,7 +7493,7 @@ static fastpath_t vmx_vcpu_run(struct kvm_vcpu *vcpu, bool force_immediate_exit)
 	vmx_recover_nmi_blocking(vmx);
 	vmx_complete_interrupts(vmx);
 
-	return vmx_exit_handlers_fastpath(vcpu);
+	return vmx_exit_handlers_fastpath(vcpu, force_immediate_exit);
 }
 
 static void vmx_vcpu_free(struct kvm_vcpu *vcpu)
@@ -7988,11 +7992,6 @@ static __init void vmx_set_cpu_caps(void)
 		kvm_cpu_cap_check_and_set(X86_FEATURE_WAITPKG);
 }
 
-static void vmx_request_immediate_exit(struct kvm_vcpu *vcpu)
-{
-	to_vmx(vcpu)->req_immediate_exit = true;
-}
-
 static int vmx_check_intercept_io(struct kvm_vcpu *vcpu,
 				  struct x86_instruction_info *info)
 {
@@ -8404,8 +8403,6 @@ static struct kvm_x86_ops vmx_x86_ops __initdata = {
 	.check_intercept = vmx_check_intercept,
 	.handle_exit_irqoff = vmx_handle_exit_irqoff,
 
-	.request_immediate_exit = vmx_request_immediate_exit,
-
 	.sched_in = vmx_sched_in,
 
 	.cpu_dirty_log_size = PML_ENTITY_NUM,
@@ -8663,7 +8660,6 @@ static __init int hardware_setup(void)
 	if (!enable_preemption_timer) {
 		vmx_x86_ops.set_hv_timer = NULL;
 		vmx_x86_ops.cancel_hv_timer = NULL;
-		vmx_x86_ops.request_immediate_exit = __kvm_request_immediate_exit;
 	}
 
 	kvm_caps.supported_mce_cap |= MCG_LMCE_P;

arch/x86/kvm/vmx/vmx.h

@@ -331,8 +331,6 @@ struct vcpu_vmx {
 	unsigned int ple_window;
 	bool ple_window_dirty;
 
-	bool req_immediate_exit;
-
 	/* Support for PML */
 #define PML_ENTITY_NUM		512
 	struct page *pml_pg;

arch/x86/kvm/x86.c

@@ -10505,12 +10505,6 @@ static void kvm_vcpu_reload_apic_access_page(struct kvm_vcpu *vcpu)
 	static_call_cond(kvm_x86_set_apic_access_page_addr)(vcpu);
 }
 
-void __kvm_request_immediate_exit(struct kvm_vcpu *vcpu)
-{
-	smp_send_reschedule(vcpu->cpu);
-}
-EXPORT_SYMBOL_GPL(__kvm_request_immediate_exit);
-
 /*
  * Called within kvm->srcu read side.
  * Returns 1 to let vcpu_run() continue the guest execution loop without
@@ -10756,10 +10750,8 @@ static int vcpu_enter_guest(struct kvm_vcpu *vcpu)
 		goto cancel_injection;
 	}
 
-	if (req_immediate_exit) {
+	if (req_immediate_exit)
 		kvm_make_request(KVM_REQ_EVENT, vcpu);
-		static_call(kvm_x86_request_immediate_exit)(vcpu);
-	}
 
 	fpregs_assert_state_consistent();
 	if (test_thread_flag(TIF_NEED_FPU_LOAD))