Linux 6.9: Call add_disk() from workqueue to fix zfs_allow_010_pos

[tonyhutter]

Motivation and Context

Fixes: #16089

Description

The 6.9 kernel behaves differently in how it releases block devices. In the common case it will async release the device only after the return to userspace. This is different from the 6.8 and older kernels which release the block devices synchronously. To get around this, call add_disk() from a workqueue so that the kernel uses a different codepath to release our zvols in the way we expect. This stops zfs_allow_010_pos from hanging.

How Has This Been Tested?

Ran zfs_allow_010_pos with this PR and it passed.

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.

[robn]

Oh wow, great work hunting that down. I don't feel so bad now! 😆

I tried compiling on older kernels. Builds ok on 6.1, failed on 5.10: __flush_workqueue doesn't exist there. But, I wonder if this is better gated to the 6.9 APIs anyway, since the old code works great there and deciphering workqueue behaviour back into medieval times is not very much fun.

Regardless, I will be delighted to stick a +1 on this. Cheers :)

[tonyhutter]

Builds ok on 6.1, failed on 5.10: __flush_workqueue doesn't exist there. But, I wonder if this is better gated to the 6.9 APIs anyway,

Yea, I was hoping to use a single codepath, but I had a hunch that might be the case. I'll break it out so that HAVE_BDEV_FILE_OPEN_BY_PATH (6.9 kernel) codepaths call the workqueue stuff, and the older kernels call __zvol_os_add_disk() directly.

[robn]

Compile checked against:

• 6.9.1
• 6.6.31
• 6.1.94
• 5.10.217
• 4.9.337

Seems good to me, nice work!

[robn]

@tonyhutter again, great work on this one!