ASRock ALiveDUAL-eSATA2 || Matrox G550 AGP

M

manfred

More detailed information is provided
- in Bug "kern/128331: [agp] nForce3-250 AGP not detected correctly" :
- -> http://www.freebsd.org/cgi/query-pr.cgi?pr=kern/128331,
- at http://www.asrock.com/mb/overview.asp?Model=Alivedual-esata2,
- at http://en.wikipedia.org/wiki/NForce3 and http://www.nvidia.com/page/nf3.html,
- at http://www.matrox.com/graphics/de/products/graphics_cards/g_series/g550/.

My understanding of this motherboard is that
a) the "Northbridge": NVIDIA ( former ULi / ALi ) M1695 directly connects to the CPU via pcib0
as well as
b) the "Southbridge": NVIDIA nForce3 250 (including the AGP bridge) directly via pcib1
via their respective HyperTransport interfaces, each using 16 / 8 of the (up to 32) lane links in parallel.

IFF this understanding is correct, the finding in the bug report _could perhaps_ indicate a severe flaw in the logic of FreeBSD's kernel handling multiple host bridges and the corresponding assignment of (in principle, correctly detected) devices to their host bridges.

Otherwise I would be really glad for every hint to any misunderstanding on my side.
(I am exploiting Gentoo for years now, avoiding all the distro- & release-mania, but I am quite new to the 7.x FreeBSD kernel.)

Any kernel developer on this newly created forum (++ :) !) yet already ?
 
ADDENDUM: Digging further on, I came across
http://www.ocworkbench.com/2005/uli/m1695/g1.htm
which states:
<cite>
M1695 ... Tunnel Chip ... is ... chipset that uses a tunneling technique to interconnect the two "North Bridges" as the South Bridge ...
</cite> .

Thus, the hardware is actually set up as follows:

[ CPU ]
<-- HyperTransport Link -->
[ M1695 ]
<-- HyperTransport Link -->
[ nForce3 250 ]
<-- AGP Bus -->
[ G550 ]

which implies that indeed we have two host bridges,
but _sequentially_ - not in parallel.
 
7.1-rc1

With 7.1-RC1, the problem still persists the same.
I checked it with a clean fresh install from the amd-64 DVD1.
 
7.1 final RELEASE

With 7.1, the problem still persists the same.

I checked it with a clean fresh install from the amd-64 DVD1.

Afterwards, just to prove sure,
I upgraded to the most recent mga driver port -
still the same, as expected.
 
You must log in or register to reply here.
Back
Top