cvs commit: src/sys/boot/i386/cdboot cdboot.s

[Harti Brandt]

On Wed, 12 Apr 2006, Alexander Leidinger wrote:

AL>John Baldwin <jhb at FreeBSD.org> wrote:
AL>
AL>> On Tuesday 11 April 2006 23:40, Maxim Sobolev wrote:
AL>
AL>> > BTW, can you please take a look at the problem with SMP bootstrap on
AL>> > Aplintel notebooks? For some reason our SMP kernel can't start the
AL>> > second processor. You can find more details here:
AL>> > 
AL>> > http://groups.google.ca/group/mailing.freebsd.current/browse_thread/thread/2b554e7a6cf3d3cd/b4f74b7c7907cb41?lnk=st&q=%22Intel+Macs+that+boot+FreeBSD%3F%22&rnum=1&hl=en#b4f74b7c7907cb41
AL>> 
AL>> I looked but unfortunately there isn't much to go on.  We follow the
AL>> sequence
AL>> Intel specifies.  If you want to debug this you'll have to probably do
AL>> something like bring back the postcode stuff in mp startup and dump the
AL>> various postcode values to see how far the AP got, etc.
AL>
AL>If it helps: I've read somewhere that the dual-core CPUs from Intel changed
AL>the semantic of the HT <something> (don't remember if it's the feature bit or
AL>the register or whatever... I don't have that much low-level knowledge of
AL>current CPUs). They made this so that Windows XP is able to run with the
AL>dual-core CPU instead of requiring the Professional version to use both CPU
AL>cores.

There is very good article in the last C't (www.heise.de/ct, but the 
article is only in the printed version) which explains all the different 
versions of the HT bits and processor enumeration variants for both intel 
and Amd CPUs. It's in german, though.

The article also points to several inconsistencies in the Intel 
documentation with regard to this.

AL>So maybe we need to change the part in FreeBSD which handles the HT part of
AL>the CPU...

harti