Same diff but with -U99999 to capture the full files for review.

This update of the patch addresses Adrian's concern that *_chan() are polluted by references to USB (*uc) data structures. Instead of setting uc->uc_write_delay directly the *chan() functions set a chan_flag in the sc to indicate that execution is in a sensitive part of channel processing, that a delay should be made before the write.

This is a much simplified patch. Plus a comment why we need to unset ntpd_user because rc.subr will use that to su(1) for us and we don't want this behavior. We want ntpd to setuid() itself.

This has been simplified. Will upload it.

In D48057#1095687, @manu wrote:

In D48057#1095665, @lwhsu wrote:

Will we (or how about directly) move them to wifi-firmware- ports?

That's not my plan no, since we're moving toward pkgbase users will have a way to install them or not, moving them to ports gives us nothing.

This would be welcome.

The other problem is the TL-WN823N, which did work on my old laptop, no longer does. Also tested on the sandbox server downstairs. Something has changed since it was last used. To rule out that this device has stopped working I connected it to my $JOB Windows laptop. It works. It would appear that this is a FreeBSD-specific problem I need to hunt down.

This version of the patch broke TL-WN823N on the ehci(4) laptop. Moving setting the USB delay flag out of the _chan functions caused this regression. The other two rtwn(4) devices I have still work.

In D47562#1089397, @adrian wrote:

I'm ok with this. Stuff in tree has to work. :-)

I'd like to dig into it some more to see if we can figure out the actual root cause (i swear we're just missing a register read somewhere just to ensure things are flushed) but I'd prefer to have everything in tree working whilst we go digging. And it's super non intrusive, so thank.

Thanks for digging into this!

In D47562#1089396, @adrian wrote:

oh interesting, so this now works on more?

This diff looks like it's putting a delay before /every/ write now?

Also, the previous patch worked with one of my RTL8192EU (the AC1200) but not the other (TL-WN823N). It now works with both. And with the RTL8188EU as well.

This update incorporates adrian@'s requested changes.

In D47522#1084776, @adrian wrote:

oh it's THIS bug, ha! ok, lemme go re-review D38854 and see if we can figure out how to get it in cleanly.

In D47562#1084777, @adrian wrote:

There's /gotta/ be a way to do this without having the channel code try get the USB pointer.

Lemme noodle on it, I think we should be able to move uc_delay_us into the rtwn softc, and make uc_write_delay a softc function? instead of how it's currently done?