ZTS: add mount_loopback to test zfs behind loop dev

[tonyhutter]

Motivation and Context

Add test for #17277

Description

Add a test case to reproduce issue #17277. This has actually been fixed by #17298, but add a test case for good measure. The test case verifies that we can make an xfs filesystem on a ZFS-backed loopback device.

How Has This Been Tested?

Test case added

Types of changes

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Performance enhancement (non-breaking change which improves efficiency)
• Code cleanup (non-breaking change which makes code smaller or more readable)
• Breaking change (fix or feature that would cause existing functionality to change)
• Library ABI change (libzfs, libzfs_core, libnvpair, libuutil and libzfsbootenv)
• Documentation (a change to man pages or other documentation)

Checklist:

• My code follows the OpenZFS code style requirements.
• I have updated the documentation accordingly.
• I have read the contributing document.
• I have added tests to cover my changes.
• I have run the ZFS Test Suite with this change applied.
• All commit messages are properly formatted and contain Signed-off-by.

[tonyhutter]

@behlendorf I included your fixes in my latest push.

[robn]

Two thoughts:

• maybe a skip if you don't have mkfs.xfs and/or XFS support in your kernel (I routinely don't have those in my tiny VMs)
• it's more work, but would we be better to identify the IO pattern that tickled the bug, and write a program to test that? If XFS or the kernel ever changes in the future, the test could end up quietly not actually testing anything useful, and not alerting us to a future break. On the other hand, there's no harm in having a straight up functional "can we even mount a loopback filesystem" test, so as long as we know what this test is, probably nbd.

[tonyhutter]

@robn -

maybe a skip if you don't have mkfs.xfs and/or XFS support in your kernel (I routinely don't have those in my tiny VMs)

Thanks - I just added a check in my latest push

On the other hand, there's no harm in having a straight up functional "can we even mount a loopback filesystem" test, so as long as we know what this test is, probably nbd.

Yes, its just a basic sanity test to verify "filesystem over loopback over ZFS".

[robn]

Nice, thanks!

[amotin]

It should probably be specified somewhere that skipped test is OK:

Tests with results other than PASS that are unexpected:
    SKIP mount/mount_loopback (expected PASS)

[behlendorf]

Since xfs may not be built as a kmod it's probably more reliable to check that either CONFIG_XFS_FS=m or CONFIG_XFS_FS=y is set in /boot/config-$(uname -r).

[robn]

That's the /proc/filesystems check though, right? If it's not a module, then it must be builtin and so in the filesystem list?

[tonyhutter]

@robn that's right - the two checks are to look for modules & built-in filesystems. I actually think the /boot/config-* check is a little more elegant, but I wasn't 100% sure all distros put their configs in /boot.

[behlendorf]

This seems to work for at least the distributions covered by the CI. We already do a similar check for uring support in functional/io/io_uring.ksh. If you wanted to be fancy you could first check for /proc/config.gz. Providing the kernel config in /proc/ has been supported for years now, but I'm not sure how commonly it's enabled in distribution kernels.

At least on RHEL8 it's disabled:

/boot/config-4.18.0-553.50.1.el8_10.x86_64:# CONFIG_IKCONFIG is not set
/boot/config-4.18.0-553.51.1.el8_10.x86_64:# CONFIG_IKCONFIG is not set

[behlendorf]

Not to be greedy, but it'd be nice to have an identical test which uses ext4.