Re: rtw88 import to main

On Fri, Apr 01, 2022 at 04:23:58PM +0000, Bjoern A. Zeeb wrote:
| On Fri, 1 Apr 2022, Doug Ambrisko wrote:
| 
| > On Fri, Apr 01, 2022 at 12:32:40AM +0000, Bjoern A. Zeeb wrote:
| > | I've just pushed Realtek's rtw88 driver into the tree along with
| > | firmware to support it and man pages.  Station mode only.
| >
| > 	[snip]
| >
| > | CAVEAT-1:  if you have more than 4GB of main memory, the driver will
| > | likely fail to load firmware or work.  Sounds silly these days but hey.
| >
| > 	[snip]
| >
| > Another potential hack with the 4GB limit is to use bhyve for the
| > wireless connectivity and then everything else outside.  It's
| > also handy to limit crashes, backtraces etc. to bhyve and not the
| > host.
| 
| Yes, I do that some for development and that got bhyve one or two
| fixes for passthru in the last months.  Cannot wait to have it on
| arm64 as well.
| 
| > I've thought of a suspend/resume laptop hack is stop bhyve on
| > on suspend and then restart it on resume.  Then we don't need
| > suspend/resume support in the driver.
| 
| There are projects like wifibox which try to do something like
| this running a Linux inside I believe (though I don't know if
| suspend/resume is a thing there)?
| 
| I also don't know how resume will "feel" as it'll take a few seconds
| extra if you have to completely restart a bhyve on resume.

A slow suspend & resume is better then a full power off/power on or
a crash recovery after forgetting to power off!

I've done bhyve for wireless WAN to make it easier to manage
for other people (pfSense etc.)
 
| It wasn't until a while ago when people were posting firmware crashes
| for iwm(4).  I think the goal remains to bring us back to more support
| and while I am sure drm-kmod is less troubled by a 4GB limit than
| realtek drivers, also remember that what we do here is potentially
| also helping and/or improving other parts of the system.

Yep.

Doug A.