On Fri, Nov 05, 2021 at 04:45:28PM -0400, Josef Bacik wrote:

Currently we do this awful thing where we get another ref on a trans
handle, async off that handle and commit the transaction from that work.
Because we do this we have to mess with current->journal_info and the
freeze counting stuff.

We already have an async thing to kick for the transaction commit, the
transaction kthread.  Replace this work struct with a flag on the
fs_info to tell the kthread to go ahead and commit even if it's before
our timeout.  Then we can drastically simplify the async transaction
commit path.

I wanted to get rid of the async commit completely and reusing the
pending operations that's basically the same what you add as the new fs
bit.

https://lore.kernel.org/linux-btrfs/808c557f1ae3034fdfa2d2361e864aac90ba5a94.1622733245.git.dsterba@suse.com/ (local)

There's also a followup removing transaction_blocked_wait, I got some
hangs in btrfs/011. I haven't tested 3/4 separately so I don't know the
exact cause.

This is duplicating functionality behind BTRFS_PENDING_COMMIT, otherwise
simplifying the async commit is good of course.