Re: IBM T60 dmsg 6.1-RC1 (problems will test patches)

From: Jung-uk Kim <jkim_at_FreeBSD.org>
Date: Wed, 19 Apr 2006 15:28:20 -0400
On Wednesday 19 April 2006 02:35 pm, Alexander Leidinger wrote:
> Am Wed, 19 Apr 2006 07:28:51 -0400
>
> schrieb John Baldwin <jhb_at_freebsd.org>:
> > On Wednesday 19 April 2006 01:22 pm, Benjamin Adams wrote:
> > > I don't think the kernel is using dual core corrently. 6.0 and
> > > 7.0 did not reconize my acpi_bus for my hard drive.
> > > Also built in Wireless is not detected, Maybe other problems?
> >
> > Aside from the wireless it all looks (mostly) ok to me.
> >
> > > FreeBSD/SMP: Multiprocessor System Detected: 2 CPUs
> > >  cpu0 (BSP): APIC ID:  0
> > >  cpu1 (AP): APIC ID:  1
> >
> > These are your two cores, though FreeBSD thinks they are HTT
> > instead of dual cores for some reason.  You can toggle the
> > 'machdep.hyperthreading_allowed' sysctl to make the kernel
> > schedule threads onto the second core though.
>
> This is because Intel wants the Dual-Core to work with Windows XP.
> They (and AMD) decided to change the semantic of the HTT bits. A
> summary of a german article about this is at
>   
> http://lists.freebsd.org/pipermail/cvs-all/2006-April/166131.html
>
> Feel free to ask for some specific details, I will translate those
> parts then.

RELENG_6 does not *display* number of cores but -CURRENT should:

http://www.freebsd.org/cgi/cvsweb.cgi/src/sys/amd64/amd64/identcpu.c.diff?r1=1.136&r2=1.140
http://www.freebsd.org/cgi/cvsweb.cgi/src/sys/i386/i386/identcpu.c.diff?r1=1.150&r2=1.145

One thing that I neglected to MFC... :-(

Actual HTT core detection is done with different algorithm by nectar:

http://www.freebsd.org/cgi/cvsweb.cgi/src/sys/amd64/amd64/mp_machdep.c.diff?r1=1.253&r2=1.254

Jung-uk Kim
Received on Wed Apr 19 2006 - 17:28:41 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:38:55 UTC