[PATCH] riscv: kvm: move preempt_disable() call in kvm_arch_vcpu_ioctl_run

[Nikolay Borisov]

local_irq_disable provides stronger guarantees than preempt_disable so
calling the latter is redundant when interrupts are disabled. Instead,
explicitly disable preemption right before interrupts are
enabled/disabled to ensure that the time accounted in guest_timing_exit_irqoff
includes time taken by the guest or interrupts.

Signed-off-by: Nikolay Borisov <nborisov at suse.com>
---
 arch/riscv/kvm/vcpu.c | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)

diff --git a/arch/riscv/kvm/vcpu.c b/arch/riscv/kvm/vcpu.c
index 7f4ad5e4373a..0c527b2c3c63 100644
--- a/arch/riscv/kvm/vcpu.c
+++ b/arch/riscv/kvm/vcpu.c
@@ -888,8 +888,6 @@ int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu)
 
 		kvm_riscv_check_vcpu_requests(vcpu);
 
-		preempt_disable();
-
 		local_irq_disable();
 
 		/*
@@ -926,7 +924,6 @@ int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu)
 		    kvm_request_pending(vcpu)) {
 			vcpu->mode = OUTSIDE_GUEST_MODE;
 			local_irq_enable();
-			preempt_enable();
 			kvm_vcpu_srcu_read_lock(vcpu);
 			continue;
 		}
@@ -960,6 +957,7 @@ int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu)
 		/* Syncup interrupts state with HW */
 		kvm_riscv_vcpu_sync_interrupts(vcpu);
 
+		preempt_disable();
 		/*
 		 * We must ensure that any pending interrupts are taken before
 		 * we exit guest timing so that timer ticks are accounted as
-- 
2.25.1