In terms of regioselectivity, at least for the first addition of the Grignard reagent, I suggest to have a look on the cumarin, that may be seen as an α,β-unsaturated carbonyl compound. Hard nucleophiles (in terms of HSAB principle) like $\ce{MeLi}$ would almost exclusively react with the carbonyl carbon, as you drew in your reaction equation. Soft nucleophiles, such as cuprates, would prefer the 1,4-addition.
Going from $\ce{MeMgBr}$, to $\ce{EtMgBr}$, $\ce{i-PrMgBr}$, and $\ce{t-BuMgBr}$ gradually lowers the polarity of the bond between the metal (magnesium) and the directly attached carbon atom. Hence, along with this line (in other words, as a trend), the preference to react in 1,4-position (more like Michael reaction) increases on expense of the reaction of this nucleophile on the carbonyl carbon. I speculate the bond in the Grignard reagent, between $\ce{Mg}$ and the phenyl group in $\ce{PhMgBr}$, is even less polarised, than in the instance of $\ce{t-BuMgBr}$, and hence conclude at least a significant portion of the product of the first addition formed will be the one of the 1,4-addition (second drawing).
As @Zhe, I however still have a difficulty to understand the subsequently suggested second addition of $\ce{PhMgBr}$ in the lower drawing, because the activation of the double bond once in conjugation with the carbonyl group is gone.