git status --porcelain <path>, but includes untracced files unless ignored. It's for general script checks for local modifications. And is the Fastest 🚀. It also handles staged files.

Correct copy-pasteo

style(9) from br@

Reported by: @naddy

Note that 16384 is not an intentionally set limit, it is just a consequence of an expression to ensure we don't have an integer overflow. It is somewhat conservative and could be made higher if there's a desire to do so.

In that case what about just "idle_on_suspend" (or "enable_idle_on_suspend") with a default of 1

Negative-sense sysctls can be confusing -- i.e., enable the disable sysctl to avoid idling CPUs upon suspend. That said, this sysctl is configuring the system to do something undesired/unusual, so I do like having 0 be the normal state. What about calling it something like "suspend_idle_workaround"?

Yeah it didn't work before, even from the initial commit

/*
 * XXX: This will be wrong in the multipacket case
 * I suspect FS should be for the first packet and
 * LS should be for the last packet
 */

@imp yeah for me it's less about "people are probably using this" than a general "FreeBSD might silently drop support for something that I care about"

IMO this one warrants some specific attention to it (e.g. referencing it explicitly in the HEADS-UP mails rather than just included in "i386-only drivers are being removed" as it theoretically could be used on amd64. But yes, I suspect this has not been used on other than i386 at all, or at least not in the last decade+

Documentation needs to match what is actually going on, not what could go on with a trivial rewrite.

also undo mlink changes

I like it

release/Makefile and sys/conf/newvers.sh look ok, will have to find some time to review the other bits.
IMO it would be fine to commit this in a few parts - e.g. newvers.sh is useful independent of how release builds are produced

There's no reason for mgb to be x86-only

I assume you used Linux source as a reference, but this code is not derived from Linux and so "Obtained from" isn't right. I would just write as prose in the commit message something like "Configuration based on Linux commit 1234abcd" if you found a specific Linux commit that added support, or something simple like "Linux (at commit <hash>) used as a reference."

Overall looks good to me, a couple of inline questions and if feasible I'd like to see this as a few separate commits.

Do you want to include the usual dance I do for BEs?