Mailinglist Archive: opensuse (4570 mails)

["Grahame M. Kelly" <grahame@xxxxxxxxxxxxxx>]

On Thu, 24 Nov 2005 09:16 am, John Craig wrote:

> I have a dual Opteron File server with 4GB ram, and just "upgraded" to
> SLES9 x86-64 (was previously running SUSE 9.3 x86-64 without problem).  The
> system keeps crashing, and I am starting to suspect mtrr.
>
> Some other posts on the subject indicate that mtrr is used to remap the
> memory that is hidden by the PCI bus to a region above the existing
> physical memory, so it can be accessible.
>
> The system on boot always reports something like this:
>
> mtrr: 0xfd000000,0x800000 overlaps existing 0xfd000000,0x400000
> mtrr: 0xfd000000,0x800000 overlaps existing 0xfd000000,0x400000
>
> And after running a while it suddently locks up, requiring a cold start.
>
> Something else that I thought funny is that there are 2 PCI -X cards in it
> that work fine but attempting to put in another (an Intel dual GigE card)
> prevents the system from booting or even passing POST (even with the other
> cards removed). The same card works in another dual Opteron machine
> (different mother board) running SLES9 without problem.
>
> So, if erroneous memory mapping by mtrr is causing the lockups, how can I
> remap or disable memory mapping?
>
> Thanks for any clues
> John Craig

Hi John.

As MTRR is only applicable to Intel P6 family processors - I don't think it is 
implemented at all with AMD CPU's and Opteron in particular.

You can simply remove the MTRR by recompile the Kernel at configuration time. 
The CONFIG_MTRR option creates a /proc/mtrr file which may be used to 
manipulate your MTRRs.  MTTR is used only to speed up the X server /Video 
Card memory response times (seem to remember this from a lecture Richard 
Gooch gave about the MTTR using Intel MTRR he developed several years ago). 
The /usr/src/linux/Documentation/mtrr.txt information is more forthcoming - 
Its was a INTEL CPU family thingy anyhow originally - not AMD Opteron.

I suspect either memory implementation (motherboard design) or mother board 
advance BIOS settings  - memory configuration wise at the h/w level.   MTRR / 
MTTR if used at all probably aggravates the memory design/timing issues if 
anything - This has been my experience to date.

---

Reference some cards not working (or the system hanging) at all in using PCI-X 
slots - and / or rather only in some motherboards - this has been my 
frustrated experience too. It is a h/w design implementation issue either in 
the card's firmware (if it has either a CPLD/FPGA or uP Controller on-board) 
most of the time otherwise the card itself may never had tests under strict 
high speed demands of the PCI-X conditions in the first place. Some cards are 
compatible, others not unfortunately. Its motherboard implementation of PCI-X 
and the cards timing restraints typically.

I base my experiences on:
My system is a Tyan S2892 Thunder K8SE motherboard with Dual 270 Opteron CPU's 
(with Dual Core = Quad CPU's) + 4GB Memory DDR400 Reg ECC (128Bit wide 
interleaving) + 2 x 3Ware controllers (133MHz) PCI-X slots with WD SATA2 - 
800GB disks. Was originally SuSE 9.3 Professional x64, now SuSE 10.0 x64 and 
I have no problems related to the MTRR or MTTR although I get the same mtrr 
message on boot up.

Hope this assists you.
-- 
Cheers. Grahame