Re: git: ce881170088c - main - atkbd: Disable periodic polling by default.
Date: Thu, 06 Jan 2022 00:06:35 UTC
On Wed, Jan 5, 2022 at 5:04 PM Gleb Smirnoff <glebius@freebsd.org> wrote: > On Wed, Jan 05, 2022 at 04:18:25PM -0700, Warner Losh wrote: > W> On Wed, Jan 5, 2022, 3:47 PM Gleb Smirnoff <glebius@freebsd.org> wrote: > W> > W> > On Wed, Jan 05, 2022 at 08:13:03PM +0000, Alexander Motin wrote: > W> > A> commit ce881170088c4c98c036fe561f8ee8413c2e2585 > W> > A> Author: Alexander Motin <mav@FreeBSD.org> > W> > A> AuthorDate: 2022-01-05 20:00:36 +0000 > W> > A> Commit: Alexander Motin <mav@FreeBSD.org> > W> > A> CommitDate: 2022-01-05 20:12:58 +0000 > W> > A> > W> > A> atkbd: Disable periodic polling by default. > W> > A> > W> > A> It is one of the few remaining Giant-locked callouts. It > would be > W> > A> good to remove it, not mentioning that polling itself is not > good. > W> > A> > W> > A> If this cause keyboard/mouse freezes on some hardware, please > set > W> > A> loader tunable hw.atkbd.hz=1 as workaround and report the > issue. > W> > A> > W> > A> Submitted by: imp, jhb > W> > > W> > Lame question: this is related only to a keyboard plugged into PS/2, > isn't > W> > W> Yes. I think this part of the code is only for "legacy" keyboards. I > don't > W> think mice are affected, but since both attach to the keyboard > controller > W> there is a tiny chance mice might be affected in some weird hardware... > > I'd even say to legacy mainboards. Cause I have PS/2 keybooard plugged into > USB port and AFAIU it is not affected by this change in any way. > correct. Going through USB is unaffected. > I think that even if this change creates some regression, we never know > that. > Nobody would today run so old mainboard as a desktop. Might have it as a > headless appliance. > You'd be surprised what people run today :(. Doing the graphics stuff has shown me that. Nobody would build a new one today, but there's lots of systems that have been around a few years that have keyboards connected via PS/2. I doubt, though, any will see a regression because of the extreme age of this workaround. Warner