Elliott Mitchell

This was simply rebased to aid keeping track of local branches, everything is identical to the original. Notably the CI pass is still valid.

> That makes the whole basis for this change in question. And the change itself isn't just this, but has other bits in it as well. I'm inclined to just...

An alternative patch set to fix the issue: [altfix](https://github.com/ehem/freebsd-src/tree/altfix) As GitHub functions on branch names, I could readily push that on top of "intrfix" for this pull. It is nominally...

Okay, try this. Note the later commits are cleaning out some bits which I would tend to, but they are optional to the goal of this pull request. I suspect...

It actually only takes the two at the front to achieve my primary objective. With those `intr_isrc_deregister()` would genuinely function and fully release the portions allocated by INTRNG. The as...

Okay, figured out what was going on with `->isrc_handlers`, so that can now be nuked. The goal for this pull is achieved at 3068c822dcb784d5f66f73343d86510fbd359a6b, though the cleanup in 56aaa294392145dd64476da6ab3335204d52f978 is...

I finally realized `INTR_SOLO` isn't actually incompatible with always allocating events. It is doing using `INTR_SOLO` and event-style handlers at the same time which is incompatible. As a result removing...

I'm planning to update to FreeBSD's HEAD soon. There will be a rebase in a few hours.

In fact, the two at the front really do fix the problem sufficiently. If the event is always allocated, then `intr_event_destroy()` won't ever be called with a `NULL` event. Then...

Given how things look now, I plan to rebase in a bit. Some of what is in here now looks worthy of being split into a separate branch.