Could hurd use some mach port chicanery and `vm_map`?

My understanding is that currently the /proc/self/maps implementation on Hurd exposes memory objects via their relevant mach ports as the "path", like [mem_obj={port_num}]. This seems like it might provide us with enough information that we can "simply" use the ultra-verbose vm_map (...with some wrapper code, obviously)... to map in the relevant segments of the relevant files.

I think emulating specifics of the /proc/self/maps implementation Linux has on Hurd may have made sense at one point but is ultimately a losing battle if one actually wants to make a different OS. See e.g. https://github.com/rust-lang/backtrace-rs/issues/671 which the Linux implementation may start to use soon enough.

cc @sthibaul

No source file or test is named. Start by reviewing the Hurd /proc/self/maps behavior described here, the Mach port and vm_map interfaces, and rust-lang/backtrace-rs#671; done would mean a decided, tested approach for obtaining file-backed segments on Hurd without relying on Linux-specific maps details.