Bug#887250: systemd should depend on e2fsprogs explicitly

[Helmut Grohne]

Control: severity -1 minor

I think thus far, nobody has done the analysis part for systemd.

On Sun, Jan 14, 2018 at 08:11:47PM +0100, Helmut Grohne wrote:
> /bin/journalctl contains chattr. According to file it is a ELF 64-bit LSB shared object, x86-64, version 1 (SYSV)
> /bin/systemd-tmpfiles contains chattr. According to file it is a ELF 64-bit LSB shared object, x86-64, version 1 (SYSV)

False positives: chattr_fd

> /lib/systemd/libsystemd-shared-236.so contains chattr and debugfs. According to file it is a ELF 64-bit LSB shared object, x86-64, version 1 (SYSV)

False positives:
 * chattr_fd
 * chattr_path
 * chattr-util.c
 * ..."consider turning off the copy-on-write file attribute on the
   journal directory, using chattr +C."
 * debugfs is used in a filesystem type list.

> /lib/systemd/system-generators/systemd-cryptsetup-generator contains mke2fs. According to file it is a ELF 64-bit LSB shared object, x86-64, version 1 (SYSV)

Actually generates a .service file with:

    ExecStartPost=/sbin/mke2fs '/dev/mapper/%s'

It's used when a crypttab entry uses the tmp=<fstype> option. This is
not part of any default install as we tend to use a tmpfs for /tmp
there. Despite the crypttab option supporting fstype, the generator
really only supports mke2fs. Yes, it will create an ext2 filesystem.
Does anybody really use that feature?

> /lib/systemd/system/sys-kernel-debug.mount contains debugfs. According to file it is a ASCII text

False positive. Yeah, it's about the kernel filesystem type.

> /lib/udev/rules.d/99-systemd.rules contains mke2fs. According to file it is a ASCII text

False positive. It does contain mke2fs in a comment.

Given these, the only actual use is the systemd-cryptsetup-generator.
The relevant option is not used by default. The option parsing is
inconsistent with cryptsetup's description of the option. (Should I file
a bug about that?) Poking random developers at the option resulted in
"weird option".

I think it warrants a "Suggests" at most.

I'm in favour of just closing the bug. Does anyone disagree?

Helmut