On 2013-03-14 20:39, Jan Kiszka wrote:
> vmx_vcpu_reset may now be called while already holding the srcu lock, so
> we may overwrite what was already saved there. Save and restore it.
>
> Signed-off-by: Jan Kiszka <jan.kiszka@xxxxxxxxxxx>
> ---
>
> Even if this should be unneeded, it looks more consistent. In any case,
> all versions on the table, pick what you prefer.
>
> arch/x86/kvm/vmx.c | 3 +++
> 1 files changed, 3 insertions(+), 0 deletions(-)
>
> diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
> index 958ac3a..7bc49ca 100644
> --- a/arch/x86/kvm/vmx.c
> +++ b/arch/x86/kvm/vmx.c
> @@ -4117,6 +4117,7 @@ static void vmx_vcpu_reset(struct kvm_vcpu *vcpu)
> {
> struct vcpu_vmx *vmx = to_vmx(vcpu);
> u64 msr;
> + int idx;
>
> vmx->rmode.vm86_active = 0;
>
> @@ -4190,9 +4191,11 @@ static void vmx_vcpu_reset(struct kvm_vcpu *vcpu)
> vmcs_write16(VIRTUAL_PROCESSOR_ID, vmx->vpid);
>
> vmx->vcpu.arch.cr0 = X86_CR0_NW | X86_CR0_CD | X86_CR0_ET;
> + idx = vcpu->srcu_idx;
> vcpu->srcu_idx = srcu_read_lock(&vcpu->kvm->srcu);
> vmx_set_cr0(&vmx->vcpu, kvm_read_cr0(vcpu)); /* enter rmode */
> srcu_read_unlock(&vcpu->kvm->srcu, vcpu->srcu_idx);
> + vcpu->srcu_idx = idx;
> vmx_set_cr4(&vmx->vcpu, 0);
> vmx_set_efer(&vmx->vcpu, 0);
> vmx_fpu_activate(&vmx->vcpu);
>
This cannot work either: I think we really need to drop the srcu lock
before calling vmx_set_tss_addr. But if we nest the lock, we may only
drop it once now in enter_rmode.
OK, I'll propose a patch to remove that TSS bug workaround from
enter_rmode. Will dig a bit in the archives as well to check which
version of qemu-kvm was actually exposing this.
Jan
Attachment:
signature.asc
Description: OpenPGP digital signature
