I've noticed that the Gimp Layer Mask, which I believe should be at worst an 8-bit (256 values) grayscale layer does not have full precision. In RGB 8-bit precision mode, I can create an opaque image with a smooth transition between white and black, using all the expected 0-255 color values R, G, B... i.e (0,0,0), (1,1,1), (2,2,2) ... (255,255,255)
When I create a layer mask from it and view the layer mask, the result is an immediate jump from 0,0,0 to 13,13,13.
The layer mask should have 0-255 values from black to white, no? It is grayscale (you can't colorize it). It's not an issue with the conversion, because attempting to edit the layer mask with a smooth gradient results in the same blotchy transition.
Working in a higher color precision should not be necessary, and it only helps when viewing the image. Exporting a PNG in 8 bit RBGA results in the same lack of precision in the A channel.
The layermask is in linear light mode and this is a known issue: https://gitlab.gnome.org/GNOME/gimp/-/issues/3136