Mark the vCPU CSRs as dirty after successfully setting an FWFT feature
value. FWFT features may modify CSRs (e.g., pointer masking modifies
henvcfg.PMM), and failing to mark them dirty can lead to the guest
observing stale CSR state after vCPU scheduling or migration.

Fixes: 1323a5cfe52c ("KVM: riscv: Skip CSR restore if VCPU is reloaded on the 
same core")

Signed-off-by: Yong-Xuan Wang <[email protected]>
---
 arch/riscv/kvm/vcpu_sbi_fwft.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/arch/riscv/kvm/vcpu_sbi_fwft.c b/arch/riscv/kvm/vcpu_sbi_fwft.c
index 2eab15339694..5e4aafb0cbf1 100644
--- a/arch/riscv/kvm/vcpu_sbi_fwft.c
+++ b/arch/riscv/kvm/vcpu_sbi_fwft.c
@@ -521,6 +521,7 @@ static int kvm_sbi_ext_fwft_set_reg(struct kvm_vcpu *vcpu, 
unsigned long reg_num
                break;
        case 2:
                ret = conf->feature->set(vcpu, conf, true, value);
+               vcpu->arch.csr_dirty = true;
                break;
        default:
                return -ENOENT;

-- 
2.43.7


Reply via email to