Check to see if the kernel has cd9660 before attempting to mount created images

[brooks]

Summary

This extends the set in r316028 to allow all tests to pass or be skipped
on a system without cd9660 support.

Obtained from: CheriBSD
Sponsored by: DARPA, AFRL
MFC After: 1 week

Test Plan

tested on a mips64 qemu system without cd9660 support

[ngie]

Is this solving an issue seen on CheriBSD, or a hypothetical issue?

[brooks]

In D10516#217955, @ngie wrote:

Is this solving an issue seen on CheriBSD, or a hypothetical issue?

Real in that the MALTA64 kernel doesn't contain a cd9660. r316028 seems to have only covered half the cases that require mounting the created filesystem.

[ngie]

I'm ok with all of the changes but to :o_flag_isolevel_3 (it would eliminate potential coverage today)

usr.sbin/makefs/tests/makefs_cd9660_tests.sh

269 ↗

(On Diff #27769)

This isn't necessary, yet, but it will be in the future. Maybe this should be moved down?

[brooks]

FWIW, I'm not convinced mounting the file systems is the best option. We should probably mount one of them, but extracting the files with tar would be more portable.

usr.sbin/makefs/tests/makefs_cd9660_tests.sh

269 ↗

(On Diff #27769)

I don't see any reason not to add it since the mount_image call requires it. Otherwise we add bogus failures to systems where the test will fail as soon as makefs is updated.

[brooks]

After realizing I misunderstood how atf-sh works (atf-check returns immediate failure if it fails, not at script exit), I wonder if check_cd9660_support should be rolled into mount_image. Then we'll do unvalidated image creation testing even when kernel support is missing.

[brooks]

In D10516#218274, @brooks wrote:

After realizing I misunderstood how atf-sh works (atf-check returns immediate failure if it fails, not at script exit), I wonder if check_cd9660_support should be rolled into mount_image. Then we'll do unvalidated image creation testing even when kernel support is missing.

Hmm, it can't actually go in mount_image. I'll revise this diff and move all the check_cd9660_support calls down.

[brooks]

• Go ahead and create the images even if we won't be able to mount them.

[ngie]

I'll open another review to use tar instead of mount_cd9660. I think your input is correct there.