Subject: Re: IPL_LOCK oddity
To: Manuel Bouyer <firstname.lastname@example.org>
From: Bill Studenmund <email@example.com>
Date: 08/24/2006 10:55:42
Content-Type: text/plain; charset=us-ascii
On Thu, Aug 24, 2006 at 09:38:11AM +0200, Manuel Bouyer wrote:
> On Thu, Aug 24, 2006 at 12:18:12PM +0900, YAMAMOTO Takashi wrote:
> > > In chatting with dbj, I was reminded that on a number of ports, we ha=
> > > IPL_SERIAL higher than IPL_LOCK. IPL_IPI also is higher than IPL_LOCK=
> > > x86 (I didn't check extensively).
> > > Also, is the IPI code structured so that it doesn't have to take lock=
> > x86 ipi code uses raw __cpu_simple_lock for that reason.
> And maybe the serial code should use that too ... The locks here are not
> supposed to be hold for long times, from what I remember.
That would probably be the best solution. I'd rather not have spllock()=20
block serial interrupts.
The one problem is that the lock that the com driver takes for interrupts=
is also the one it takes for normal uses. The latter uses may want to stay=
with normal spin locks. If so, that'd be an arguement against changing=20
serial drivers to use __cpu_simple_lock.
I just talked myself into and out of something. :-|
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.3 (NetBSD)
-----END PGP SIGNATURE-----