The Electric Dipole Moment [itex]\vec{d}_f[/itex] of some particle [itex]f[/itex], is defined by its interaction with an electric field [itex]\vec{E}[/itex]:
[itex]H_{EDM} = - \vec{d}_f \cdot \vec{E}[/itex]
For a spin-1/2 particle, this corresponds to the effective lagrangian density:
[itex]\mathcal{L}_{EDM} = \frac{-i}{2} d_f F_{\mu \nu} \bar{\psi}_f \sigma^{\mu \nu} \gamma^5 \psi_f[/itex]
you can have a similar term if you look at quarks [itex]\mathcal{q}_r[/itex], by replacing the electromagnetism with strong ints:
[itex]\mathcal{L}_{CEDM} = \frac{-i}{2} d_r^{CEDM} G_{\mu \nu}^a \bar{\mathcal{q}}_r \sigma^{\mu \nu} \gamma^5 \frac{\lambda^a}{2} \mathcal{q}_r[/itex]
and that's how it arises... In general both EDMs and CEDMs generate P and T (so CP) violations.