As Jean-Marie Prival comments, Ceres doesn't have a rocky crust. It has an icy crust in which rocky matter is embedded from asteroid impacts. I'd rather see this comment turned into an answer than reinvent the wheel on that dwarf planet. In case they "die", the relevant comments are quoted below the main body of the answer..
Instead, here I consider a celestial body closer at hand. Earth does not have a subterranean ocean but it can produce water deep in the mantle, extracted from rock. As described in this answer from Space Exploration Stack Exchange, a high-pressure solid phase of water, Ice VII, has been identified as inclusions in diamonds; the presence of these inclusions implies that water exists in the molecular state ($\text{H}_2\text{O}$) as distinct from hydroxyl groups in minerals. The water is first trapped as a supercritical fluid as the diamonds form deep in the mantle, then condenses to Ice VII as the diamonds cool at the surface but retain mantle-level pressures internally.
Water in Earth's mantle likely does not come directly from surface oceans, but from hydroxide-bearing minerals in the rock. When the rock is cycled downwards by plate tectonics, it becomes hot enough to decompose the hydroxides (hydroxides of silicon, aluminum, iron, calcium, and magnesium, the elements to which hydroxyl groups are most likely to be bound in crustal rocks, all decompose well below 1000°C), thus generating the subterranean water.
Previous comments
There seems to be a whole bunch of publications on the matter: nature.com/articles/s41550-020-1146-8, nature.com/articles/s41550-020-1019-1, nature.com/articles/s41561-020-0581-6. I don't have access to all of them (too expensive ...). Maybe there's some useful information on the matter. – user20217 Aug 12 at 12:59
Interesting! The first one states that "the interior of Ceres consists of (1) a thin lag deposit (regolith) layer, (2) an icy crust that contains the bulk of an origin ocean [...]". That would explain a lot, the rocky layer being created on top of the icy surface by impacts. What got me was this passage in doi.org/10.1038/s41467-020-15973-8 : "Ceres [...] is partially differentiated into a rocky interior and a comparatively more volatile-rich crust, which is composed of rock, salts, clathrates, and ≤ 40% water ice." Looks like the crust might not be so "rocky" after all... – Jean-Marie Prival Aug 12 at 14:10