An emerging paradigm posits that the abyssal overturning circulation is driven by bottom-enhanced mixing, which results in vigorous upwelling in the bottom boundary layer (BBL) along the sloping seafloor and downwelling in the stratified mixing layer (SML) above; their residual is the overturning circulation. This boundary-controlled circulation fundamentally alters abyssal tracer distributions, with implications for global climate. Chapter 1 describes how a basin-scale overturning circulation arises from the coupling between the ocean interior and mixing-driven boundary layers over rough topography, such as the sloping flanks of mid-ocean ridges. BBL upwelling is well predicted by boundary layer theory, whereas the compensation by SML downwelling is weakened by the upward increase of the basin-wide stratification, which supports a finite net overturning. These simulated watermass transformations are comparable to best-estimate diagnostics but are sustained by a crude parameterization of boundary layer restratification processes. In Chapter 2, I run a realistic simulation of a fracture zone canyon in the Brazil Basin to decipher the non-linear dynamics of abyssal mixing layers and their interactions with rough topography. Using a hierarchy of progressively idealized simulations, I identify three physical processes that set the stratification of abyssal mixing layers (in addition to the weak buoyancy-driven cross-slope circulation): submesoscale baroclinic eddies on the ridge flanks, enhanced up-canyon flow due to inhibition of the cross-canyon thermal wind, and homogenization of canyon troughs below the level of blocking sills. Combined, these processes maintain a sufficiently large near-boundary stratification for mixing to drive globally significant BBL upwelling. In Chapter 3, simulated Tracer Release Experiments illustrate how passive tracers are mixed, stirred, and advected in abyssal mixing layers. Exact diagnostics reveal that while a tracer’s diapycnal motion is directly proportional to the mean divergence of mixing rates, its diapycnal spreading depends on both the mean mixing rate and an additional non-linear stretching term. These simulations suggest that the theorized boundary-layer control on the abyssal circulation is falsifiable: downwelling in the SML has already been confirmed by the Brazil Basin Tracer Release Experiment, while an upcoming experiment in the Rockall Trough will confirm or deny the existence of upwelling in the BBL.Ph.D.