Re: GSoC progress report

On Thu, Jul 06, 2023 at 11:08:20AM +1200, Jeandre Kruger wrote:
> Hi Martin,
> 
> I removed sc_ec/sc_if from ath(4) and made a few other minor changes.
> Some criticism of my work on ath(4) would be very helpful (probably in
> particular code I disabled, some of which I was unsure about -- though
> it seems like the driver functions fine).
> 
> One other question -- very early on sc_mtx (as it is in FreeBSD) was
> removed from ath(4) in NetBSD.  (TX buffers and queues still have
> mutexes on them.  A few other mutexes were added to the softc in
> FreeBSD in the meantime.)  Should that be added back, or should spl*
> simply stay for now?  How does it relate to MPSAFE?

I don't remember anything about sc_mtx but the spl-based
"synchronization" that I used in the early 2000s should go away.  If it
works at all, I believe it relies either on the kernel "big lock" or
some other lock (softnet mutex?) to keep threads from stepping on each
other, especially on an MP machine.

David

-- 
David Young
[email protected]    Urbana, IL    (217) 721-9981