Sign and size the bias from omitting market beta

The true return model is $r = \beta_1 x + \beta_2 m + \varepsilon$, where $x$ is your signal and $m$ is the market return (variables centered, $\varepsilon$ exogenous). You regress $r$ on $x$ alone. Suppose the market loading is $\beta_2 = 0.5$, and the auxiliary regression of $m$ on $x$ has slope $\delta = \operatorname{Cov}(x, m)/\operatorname{Var}(x) = 0.8$.

What does your estimated slope on $x$ converge to, and is it biased up or down?