svn commit: r250339 - head/sys/x86/acpica

[John Baldwin]

On Wednesday, May 08, 2013 2:21:12 pm Attilio Rao wrote:
> On Wed, May 8, 2013 at 6:01 PM, John Baldwin <jhb at freebsd.org> wrote:
> > On Tuesday, May 07, 2013 6:49:57 pm Attilio Rao wrote:
> >> Author: attilio
> >> Date: Tue May  7 22:49:56 2013
> >> New Revision: 250339
> >> URL: http://svnweb.freebsd.org/changeset/base/250339
> >>
> >> Log:
> >>   Add functions to do ACPI System Locality Information Table parsing
> >>   and printing at boot.
> >>   For reference on table informations and purposes please review ACPI specs.
> >>
> >>   Sponsored by:       EMC / Isilon storage division
> >>   Obtained from:      jeff
> >>   Reviewed by:        jhb (earlier version)
> >
> > Eh, I don't recall reviewing this.  However, acpidump already dumps this
> > table, and we should dump tables via acpidump, not in dmesg.
> 
> Now that I read better this patch is based upon a patch you submitted
> originally (see the private thread NUMA WIP).
> We already print some useful ACPI table information at boot really
> (APIC ACPI table case) and this is only printed out when having
> MAXMEMDOM > 1. Finally I have also sent you this patch and you didn't
> object.

Humm, that was in an old thread, and I never responded to say "ok" to
this.

However, the patch I posted in that thread is the patch to dump the SLIT
in acpidump and it was comitted back in r241198.  That is the place to dump
these tables, not in dmesg.  We do not dump the raw contents of any other
ACPI tables in dmesg, and have actively moved other tables out to userland
(e.g. $PIR is dumped via a userland tool only).
 
> However if you really dislike it I can print under bootverbose maybe?

No, it does not belong in dmesg at all.  Code to use its contents and honor
it when setting the domain lists would be fine :), but printing it when you
can more easily get it via acpidump -t is just clutter.

-- 
John Baldwin