Replace mnt-rs with libmount

[tasleson]

The library mnt-rs utilzes /proc/mounts which contains inaccurate information and caused issue #695. After doing some research, it appears that /proc/self/mountinfo is the correct place to look for mount information, ref. https://bugzilla.redhat.com/show_bug.cgi?id=491924 . @agrover did a PR to utilize /proc/self/mountinfo to mnt-rs ref. stemjail/mnt-rs#4 , but that PR has stalled. Thus the decision was made to utilize libmount instead as it does utilize /proc/self/mountinfo.

[tasleson]

Rebased

[agrover]

Yeah, I think we just want to use maj/min for comparison here, is all. Something like:

let major = self.thin_dev.device().major as u64;
let minor = self.thin_dev.device().minor as u64;
if mount.major == major && mount.minor == minor {

[agrover]

Why 060326b? Aren't mount.major and .minor already u64?

[tasleson]

@agrover

Why 060326b? Aren't mount.major and .minor already u64?

pub struct Device {
    /// Device major number
    pub major: u32,
    /// Device minor number
    pub minor: u32,
}

Apparently devicemapper-0.21.0/src/device.rs has them as u32 and libmount has them as a c_ulong which can either be a u32 or a u64.

[tasleson]

Fixes #820

[agrover]

ok thanks. One of THOSE situations.

👍 once Clippy is made happy.

[mulkieran]

I'ld like an explanation in the PR message as to why this approach works and the previous didn't.

[agrover]

fwiw @tasleson you can find/copy the original problem description in #695.

[mulkieran]

I already read #695. It didn't give me any idea of why this fixes the problem.

[tasleson]

@mulkieran @agrover Please see updated PR message. Hopefully this clears things up.

[mulkieran]

A 21 comment bz from 2009 in which the last entry is a promise to fix this once and for all in util-linux couldn't really be said to clear anything up :( But I guess it's a starting point for further research if necessary.

[mulkieran]

I expect there will be some more minor clippy errors, but once those are fixed, I'm fine w/ it.

[mulkieran]

Except that you might as well close this and start a new PR that squashes the commits that should be squashed, like the clippy fixes.

Concerns have been addressed

[tasleson]

Passing all checks, I squashed, should be able to merge when CI is done