p-value cap at 0.5 REML

Hello,

I'm running GCTA REML on genetic data that I simulated with msprime. For every locus I am creating a GRM and then testing it for association with random phenotypes, in order to see the distribution of the p-values under a Null simulation. I noticed that the p-values < 0.5 seem to be distributed uniformly, as expected, but that there are no p-values > 0.5. Is there a cap on the p-values in GCTA?

Thank you for your help!

Vivian