[WIP] linux 4.11 compat: avoid refcount_t name conflict

[ofaaland]

Description

Rename the ZFS type refcount_t to zfs_refcount_t.

Rather than touching all the ZFS code that uses refcount_t, use a preprocessor
macro to rename the type at compile time to zfs_refcount_t. The ZFS code
never refers to the linux type with the conflicting name so this does not break
anything.

Similarly rename refcount_add() which also conflicts.

Motivation and Context

Linux 4.11 introduces a new type, refcount_t, which conflicts with the
type of the same name defined within ZFS. The build fails when building
against that kernel or later.

Issue #5823

How Has This Been Tested?

Built and ran ztest on my desktop.

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)

Checklist:

• My code follows the ZFS on Linux code style requirements.
• I have updated the documentation accordingly.
• I have read the CONTRIBUTING document.
• I have added tests to cover my changes.
• [] All new and existing tests passed.
• Change has been approved by a ZFS on Linux member.

[behlendorf]

LGTM. I agree a minimal fix is best here to avoid potential future conflicts.

[tuxoko]

To be honest, I don't like this kind of macro game very much. If you're not careful about the include order, it's very easy to accidentally modify the kernel header and cause some strange thing.

We did this kind of thing with kmem_cache thing, but kmem_cache object and function are rarely used inside kernel header files so it's arguably not a big deal. However, you can't say the same thing for refcount_t, it's definitely going to be used in a lot of header files.

[behlendorf]

The long term plan discussed at the OpenZFS dev summit would be to add a prefix, maybe zk_ to all these interfaces to prevent a name conflict across platforms. So this should be a medium term build fix. I'd rather not have to play these macro games either but for the moment it's probably the least disruptive build fix.