On Tue, Sep 11, 2012 at 1:58 AM, Rusty Russell <rusty.russell@xxxxxxxxxx> wrote:
> We need to exhaust both tables we're walking: without this fix, we would
> not report all the CP15 registers for KVM_GET_VCPU_MSR_INDEX_LIST.
>
> Signed-off-by: Rusty Russell <rusty.russell@xxxxxxxxxx>
>
> diff --git a/arch/arm/kvm/coproc.c b/arch/arm/kvm/coproc.c
> index b454335..5fb1248 100644
> --- a/arch/arm/kvm/coproc.c
> +++ b/arch/arm/kvm/coproc.c
> @@ -850,7 +850,7 @@ static int walk_cp15(struct kvm_vcpu *vcpu, u64 __user *uind)
> BUG_ON(i1 == end1 || i2 == end2);
>
> /* Walk carefully, as both tables may refer to the same register. */
> - while (i1 && i2) {
> + while (i1 || i2) {
> int cmp = cmp_reg(i1, i2);
> /* target-specific overrides generic entry. */
> if (cmp <= 0) {
funny, thanks, applied.
-Christoffer
_______________________________________________
kvmarm mailing list
kvmarm@xxxxxxxxxxxxxxxxxxxxx
https://lists.cs.columbia.edu/cucslists/listinfo/kvmarm
