We should probably add a test for various valid and/or problematic entries.

Do you have any test cases in mind ?

By reading the source code of glib:

• Names appear to be encoded like \040 for space https://code.woboq.org/userspace/glibc/misc/mntent_r.c.html#192
• Line can be empty or commented https://code.woboq.org/userspace/glibc/misc/mntent_r.c.html#155

I tested the encoded part

$ sudo mount -t tmpfs a-b /mnt/a\ b
$ cat /proc/self/mountinfo
...
94 27 0:38 / /mnt/a\040b rw,relatime shared:263 - tmpfs a-b rw,inode64

There was no need to go so far ... https://man7.org/linux/man-pages/man5/fstab.5.html

The second field (fs_file) => If the name of the mount point contains spaces or tabs these can be escaped as `\040' and '\011' respectively.

EDIT:
Because of the apparent "complexity", I think a dedicated function is needed. Don't you?
If it is good for you, I will add a mounts_linux_internal_test.go file

@muesli ?

Do you have any test cases in mind ?

• regular unproblematic entries
• certainly the problematic entry from the report in mount -t tmpfs - /tmp -> "found invalid line" #153
• comments
• empty lines
• ...

The second field (fs_file) => If the name of the mount point contains spaces or tabs these can be escaped as `\040' and '\011' respectively.

That should we test for as well.

EDIT: Because of the apparent "complexity", I think a dedicated function is needed. Don't you? If it is good for you, I will add a mounts_linux_internal_test.go file

I agree.

Nice work so far, will give this a proper review asap.

@muesli , can you review ? 🙂

Ok, I've simplified the mountinfo parsing a bit. Everything seems to be working and testing fine, but maybe you wanna give it one more review as well. unescapeFstab seems to decode the strings just fine, so I've removed the newly added decodeName function again.

Cool, I hadn't seen it.
I'm not sure if it's a good idea to not respect the official mountinfo index, like defined by proc man pages.

Otherwise, it's good to me