flat assembler
Message board for the users of flat assembler.

Index > OS Construction > Undocumented MSRs?

Author
Thread Post new topic Reply to topic
bitRAKE



Joined: 21 Jul 2003
Posts: 3055
Location: vpcmipstrm
bitRAKE
Currently, I am curious about EXT_CONFIG ($EE) bit 2.

What is it?

Any information on the undocumented MSRs is helpful. Does anyone have a copy of the "BIOS Writers Guide" that was once public and seems to have disappeared? Both the P2 and P6 guide were on Intel's web site long ago...

AFAIK, the "BIOS Writer's Guides" are behind a NDA and the information is not public. From all the searching I've done the information is well guarded.


Last edited by bitRAKE on 17 Mar 2009, 15:13; edited 1 time in total
Post 17 Mar 2009, 01:43
View user's profile Send private message Visit poster's website Reply with quote
vid
Verbosity in development


Joined: 05 Sep 2003
Posts: 7105
Location: Slovakia
vid
I only googled this. Try asking about it on Intel forum (you might try to request more info on the footnote from datasheet).

bit 30 explained:
Quote:
Some steppings of the mobile Intel® Core™2 processor do indicate Tj to be approximately 85 or 100 via a single bit in the EXT_CONFIG register (msr 0EEh) but desktop, workstation and server processors do not. Nor is there a register implemented in those processors that software can read to get the Tj value for either the Pentium® 4 processor, Intel® Xeon® processors or Intel® Core™2 processors.


Many Core2Duo datasheets have this footnote:
page 83 wrote:

3. Disabling of any of the cores within the processor must be handled by configuring the EXT_CONFIG Model Specific Register (MSR). This MSR will allow for the disabling of a single core per die within the package.
Post 17 Mar 2009, 12:17
View user's profile Send private message Visit poster's website AIM Address MSN Messenger ICQ Number Reply with quote
tom tobias



Joined: 09 Sep 2003
Posts: 1320
Location: usa
tom tobias
here is a forum at Intel, from last month, which may prove instructive, though not satisfying.
Post 17 Mar 2009, 12:18
View user's profile Send private message Reply with quote
Alphonso



Joined: 16 Jan 2007
Posts: 294
Alphonso
Sometime ago I was looking at my BIOS and one of the things it does very very early on is set bit 2 of MSR EEh if a certain byte is set in CMOS. Never did find out what it does though Sad

I seem to have at least 176 MSRs on my P8400 of which about 40% don't seem to be documented, at least that is not in my Intel manuals from Feb 2008. Confused


Okay, I found this from Intel document 318080.pdf section 7.1
Quote:
Disabling of any of the cores within the Intel® Xeon® Processor 7200 Series and 7300
Series must be handled by configuring the EXT_CONFIG Model Specific Register (MSR).
This MSR will allow for the disabling of a single core per die within the Intel® Xeon®
Processor 7200 Series and 7300 Series package..
and sure enough setting bit 2 disables one of the cores on my P8400 Core2 Duo Smile

Some bad news, it appears the BIOS also sets bits 6 & 1 so not sure which bit(s) is(are) actually disabling one of the cores and the worse part is the MSR seems to be locked once the BIOS has finished with it. Evil or Very Mad

And why am I not getting any 'Last edited by...' each time I edit this message ? Rolling Eyes
Post 17 Mar 2009, 17:26
View user's profile Send private message Reply with quote
vid
Verbosity in development


Joined: 05 Sep 2003
Posts: 7105
Location: Slovakia
vid
tom tobias wrote:
here is a forum at Intel, from last month, which may prove instructive, though not satisfying.

Not instructive at all, in this matter, but surely satisfying to learn that "I can, however, promise that your position is well-represented in internal discussions."

I think we can see that by fact that Intel has recently been the leader in releasing "internal" documentation for their technologies (even though it is still by far less than we would like to).
Post 17 Mar 2009, 23:45
View user's profile Send private message Visit poster's website AIM Address MSN Messenger ICQ Number Reply with quote
bitRAKE



Joined: 21 Jul 2003
Posts: 3055
Location: vpcmipstrm
bitRAKE
Alphonso wrote:
And why am I not getting any 'Last edited by...' each time I edit this message ? Rolling Eyes
This happens when a message is that last one in a thread.
(I use it all the time, lol.)

Without re-writing the BIOS there doesn't appear to be a way around the locked settings, or SMM for that matter (controls fan speeds and throttles processors, as well as legacy emulation). In general this is a "good thing" for safety and security. Yet, if the features are locked, why not release information about them? Obviously, another case of security by obscurity.

My EE value is: 877D4B01 ( an edit Very Happy )
Post 18 Mar 2009, 01:06
View user's profile Send private message Visit poster's website Reply with quote
Alphonso



Joined: 16 Jan 2007
Posts: 294
Alphonso
bitRAKE wrote:
This happens when a message is that last one in a thread.
Ahh, I see, makes sense. Thanks.

FWIW for me normally MSR EEh would be 9EB91000h but if I trigger bit 2 via CMOS I get 9EB91046h and just the one core. EDX remains unchanged (0).

I had thought it might be a nice idea to have a socketable flash chip and write my own code from the jump vector FFFFFFF0h, not a fully fledged BIOS as such but just a few routines to see how things work, however due to a severe lack of skill and resources it will probably just remain a thought Sad

I didn't even realize MSR EEh was called EXT_CONFIG until your post Wink


Last edited by Alphonso on 18 Mar 2009, 10:28; edited 2 times in total
Post 18 Mar 2009, 08:30
View user's profile Send private message Reply with quote
revolution
When all else fails, read the source


Joined: 24 Aug 2004
Posts: 17716
Location: In your JS exploiting you and your system
revolution
Just the other day I found an undocumented temperature sensor inside the TI OMAP3503 chip we are upgrading to. No mention of it in the manuals, but the few gaps in the config memory space was screaming out to me "we are undocumented features". I was in the process of designing a board for it with an external temperature sensor for monitoring but now I don't have to, saves a few components and makes the routing a bit easier. Very Happy

Discovering undocumented features can be very useful sometimes. The only problem is, what happens if in the next silicon update they remove it? Or change it? Or put something else there to make the code crash and burn? Basically, lots of pitfalls can be had also.
Post 18 Mar 2009, 08:47
View user's profile Send private message Visit poster's website Reply with quote
Alphonso



Joined: 16 Jan 2007
Posts: 294
Alphonso
That's the nice thing about just programming for oneself, most of the what ifs disappear and it makes the programming so much easier that even someone like myself has a good chance of being successful. Smile
Post 18 Mar 2009, 10:30
View user's profile Send private message Reply with quote
bitRAKE



Joined: 21 Jul 2003
Posts: 3055
Location: vpcmipstrm
bitRAKE
Some BIOS boot blocks allow preempting BIOS hardware configuration. Mine has support for floppy/CD/USB, but I haven't figured out how to activate it - manufacture says they don't support that feature. The BIOS chip is soldered on the board, and I'm not ready to tear it apart just yet.

If the BIOS boot block loading isn't supported then why leave the code there to slow down the boot process? It has to be a major portion of the boot block.

Yeah, I'm not trying to take over the world - just my PC. MSR EE has EDX at 04000000 on my PC - add another undocumented bit, lol.
Post 18 Mar 2009, 15:39
View user's profile Send private message Visit poster's website Reply with quote
vid
Verbosity in development


Joined: 05 Sep 2003
Posts: 7105
Location: Slovakia
vid
Quote:
Some BIOS boot blocks allow preempting BIOS hardware configuration. Mine has support for floppy/CD/USB, but I haven't figured out how to activate it - manufacture says they don't support that feature.

What exactly do you mean by this? Do you mean BIOS recovery process (prioritized loading of specially formatted floppy to reflash BIOS in case it is fucked up somehow)? Or something else?
Post 18 Mar 2009, 23:35
View user's profile Send private message Visit poster's website AIM Address MSN Messenger ICQ Number Reply with quote
bitRAKE



Joined: 21 Jul 2003
Posts: 3055
Location: vpcmipstrm
bitRAKE
vid, yeah that is exactly what I mean. Phoenix calls it a "crisis recovery disk". Basically, it loads the boot sector at $700 and jumps to offset $80 in recovery mode; but what (if anything) triggers recovery mode is left for my discovery. There is no jumper - maybe it's a serial port key type like my Toughbook uses (Phoenix BIOS as well).

Very little setup is done at this stage, and most of it is guided by CMOS settings. So, in theory I could create a custom "crisis recovery disk" and setup the machine as I desire (rewrite of CMOS and reboot might be required for some changes). All this while keeping the BIOS I presently have (i.e. non-destructive).
Post 19 Mar 2009, 01:45
View user's profile Send private message Visit poster's website Reply with quote
Alphonso



Joined: 16 Jan 2007
Posts: 294
Alphonso
bitRAKE wrote:
vid, yeah that is exactly what I mean. Phoenix calls it a "crisis recovery disk". Basically, it loads the boot sector at $700 and jumps to offset $80 in recovery mode; but what (if anything) triggers recovery mode is left for my discovery. There is no jumper - maybe it's a serial port key type like my Toughbook uses (Phoenix BIOS as well).

Very little setup is done at this stage, and most of it is guided by CMOS settings. So, in theory I could create a custom "crisis recovery disk" and setup the machine as I desire (rewrite of CMOS and reboot might be required for some changes). All this while keeping the BIOS I presently have (i.e. non-destructive).
bitRAKE what are you trying to use this on?

For my notebook (HP InsydeH2O BIOS) to initiate the crisis recovery I have to disconnect the battery and AC then while holding down the 'win' key plus 'B' key connect the AC power and press the power button to start the notebook then release the two keys. AFAIK the ACER uses the 'fn' key plus the 'esc' key. That's the easy part, the hard part is knowing what the BIOS image file should be named. I'm not sure why the manufacturers do not supply this info' since IMHO a lot of the BIOS flashes that go bad could be fixed without having to send the thing in for repair and I would think a lot of those are warranty repairs.

Any way, it sounds like your trying to use a different recovery method which actually boots the media and executes the flash software except you want to run your own program in place of the flash program. Interesting idea.
Post 19 Mar 2009, 05:59
View user's profile Send private message Reply with quote
bitRAKE



Joined: 21 Jul 2003
Posts: 3055
Location: vpcmipstrm
bitRAKE
X7DCA-L - tech support obviously was not forth coming with any information - they prefer the motherboard be return in instances of a bad BIOS.
Post 19 Mar 2009, 15:48
View user's profile Send private message Visit poster's website Reply with quote
Alphonso



Joined: 16 Jan 2007
Posts: 294
Alphonso
Is the phoenix crisis recovery disk you mentioned the one that works with a floppy disk only? If so I wouldn't think it would work for your machine since the BIOS image file appears to be 2MB. Perhaps the recovery method is similar to the one I mentioned above, ie looks for the image file on the media and flashes using the existing BB code. I'm only guessing though.

As an alternative method what about the Boot BIOS Straps, particularly PCI mode.
ICH9 Chapter 10.1.75 wrote:
When PCI is selected, the top 16MB of memory below 4GB (FF00_0000h to
FFFF_FFFFh) is accepted by the primary side of the PCI P2P bridge and forwarded to
the PCI bus. This allows systems with corrupted or unprogrammed flash to boot from
a PCI device. The PCI-to-PCI bridge Memory Space Enable bit does not need to be set
(nor any other bits) in order for these cycles to go to PCI. Note that BIOS decode
range bits and the other BIOS protection bits have no effect when PCI is selected.
This functionality is intended for debug/testing only.
May be there's some hardware available for this.

Sorry, I seemed to have gone way OT and being a bit of a noob probably shouldn't even be talking about these things.
Post 20 Mar 2009, 07:50
View user's profile Send private message Reply with quote
bitRAKE



Joined: 21 Jul 2003
Posts: 3055
Location: vpcmipstrm
bitRAKE
I've spent several hours RE'ing the boot block code - I'd say over 50% is to support loading in the recovery mode (floppy/CD/USB). Very, very little is actually configuring the machine. There is one subroutine that's chipset specific and one that handles processor specific stuff - the rest is all general. Doesn't appear to be any self-flashing support in this BIOS (nor any other Phoenix BIOS I've encountered).

No need to apologize Alphonso - I should change the topic to match the discussion (any suggestions?) - the MSR bit wasn't going anywhere really. Maybe, the PCI Memory bridge is how they flashed the BIOS originally. Even if it's not the answer I'm looking for, I feel the code will provide an answer before too long - which I'll gladly share.
Post 20 Mar 2009, 15:23
View user's profile Send private message Visit poster's website Reply with quote
tom tobias



Joined: 09 Sep 2003
Posts: 1320
Location: usa
tom tobias
bitRAKE wrote:

I should change the topic to match the discussion (any suggestions?)
No, but, since the topic has evolved a bit away from the original question regarding undocumented MSR bits, may I ask, imprudently, why you use this particular motherboard, with two cpu's and dual ethernet connection capabilities? To be a tad more precise, is it worth almost 260$ to have two cpu's, each with multiple cores, on a motherboard with 48 gbytes dram, RATHER THAN two motherboards, each with a single cpu, and perhaps half as much DRAM?
What do you gain from such a configuration????Is the throughput more than double the throughput from a motherboard with only a single cpu? I would have expected it to be less than that, perhaps 1.2 x throughput of a single cpu motherboard with same amount of memory. To my way of thinking, double the price for 1.2x performance gain is not justified.
Post 20 Mar 2009, 16:51
View user's profile Send private message Reply with quote
bitRAKE



Joined: 21 Jul 2003
Posts: 3055
Location: vpcmipstrm
bitRAKE
Performance wasn't the goal, tom. When I sought out this build the primary goal was something that would get me through 5+ years of use with minimal wattage. Knowing the full details of the system might make this more obvious:

NVS290 Video card, 21 watts
(the on board video really sucks for watching HD movies Very Happy)
12GB DDR2 memory
(most DDR3 is more power hungry)
L5410 processors, 50 watts max each
(usually, much less power)
5100 chipset is power optimized

My last desktop system was pulling ~400 watts at idle! I aimed for half of that at maximum load. Although performance wasn't my primary goal there was certainly no reason I couldn't get some GFLOPS and extended functionality within my power budget. Yes, I paid up front (I don't use credit either), but if you factor the cost of electricity over five years you will find this system is inexpensive.

From my experience with previous dual processor motherboards, I enjoyed the responsiveness and was willing to trade away some preformance for this flexibility. It is a mistake to think in terms of linear performance - that isn't how I use this PC. When I do upgrade this system will serve media throughout the house (desktop and two/three laptops).

Maybe, it is still not justified, but it works for me. I did go to the extreme first - trying to adapt to a laptop as my primary system, but it didn't work. The mobility actually proved more costly than the power savings! Not just in terms of time - I'm more comfortable in my little office here. Then there is the reduced functionality: small screen/keyboard/storage/etc...
Post 21 Mar 2009, 02:29
View user's profile Send private message Visit poster's website Reply with quote
tom tobias



Joined: 09 Sep 2003
Posts: 1320
Location: usa
tom tobias
Agree, bitRAKE, with your excellent point about portable/laptops. I know many FASM forumers disagree with this perspective.

But, I am not buying your main point, i.e. the high performance, low power system you have described serving as principal computer of the house/office environment.

Here I must disagree. I think a far wiser solution, for the money, is to have two or three different units, each considerably less expensive than your single unit. These multiple devices would be perhaps not identical, with one, for example designed for high def video, another perhaps controlling printer/scanner, a third used for accessing the internet, rather than one single box running all operations.

I add to these three, another three computers, one for business operations, another just for internet based music, and a third for various assembly language programming projects. Two other computers are available for testing software. I don't find, in my experience, any difficulty using off the shelf, slightly outdated motherboards/cpus, i.e. NOT the most recent, best, top of the line, 64 bit, quad core cpus. I do find that even my most expensive motherboards/cpus/hard drives fail from time to time. What does one do, in the situation you describe, where, 38 months out, a critical component fails?

In my case, I simply salvage whichever components I can, and the rest of the unit goes into the recycle bin. My computers are "throw away", not top of the line. As one fails, I replace it with another, non state of the art device. How fast does the computer have to be? To listen to internet music, for example, I use a P3 866 MHz cpu on a ten year old motherboard. This works well to pick up my favorite stations in Czech republic, Norway, Denmark, Holland, France, Germany, and Switzerland. Low power? Its power consumption is probably not as low as your newest board, but my P3 866MHz cpu requires 26 watts, much less power than my more recent boards with the P4 3.0 GHz cpus, which typically operate using about 80 watts of power (still a bit less than your dual 50 watt cpus). What about the Yamaha receiver converting my digital signal into actual music??? OOPS...
Embarassed

How much power does your 32 inch flat screen monitor require, compared with the old 17 inch gargantuan behometh from a decade ago? Have you performed the test, between on chip graphics controllers, versus dedicated video card, vis a vis display of high resolution cinema, for example from DVD or Blu ray?

I find the "need to save electricity" portion of your argument, bitRAKE, very persuasive, i.e. the need to use modern, electricity sparing cpus. However, even there, saving electricity, I am dissuaded by history. Five years is a VERY long period of time, in this business. Today's "low power chipset", i.e. Intel 5100, is not particularly unique in that respect. Here is a forthcoming
two watt chipset from Via, not yet released....Do you suppose it will be released within the next calendar year?

Your dual, quad core cpu motherboard is low power compared with what? Do you really require that many horses pulling the hay? In order to estimate cost/benefit amortized over several years, I believe one needs to adopt a very conservative perspective on the life span of the components, and the near certainty of technological improvement, rendering obsolete one or another beloved feature of the current products. Take for example, "dual channel memory". I seem to recall many tests, several years ago, demonstrating the superiority of this architecture. Does the 5100 support this novelty? I don't know the answer, but I have the impression that it does not. If so, would that feature be not supported by the 5100 chip set because dual channel memory operation cost/benefit ratio proved unsatisfactory? Would that assessment, in hindsight, have prevented you from purchasing, several years ago, a more expensive motherboard which supported, at higher cost, dual channel memory operation? Mind you, I still admire my vesa local bus video cards, so I confess to being entirely outdated....
Smile
Post 21 Mar 2009, 09:41
View user's profile Send private message Reply with quote
bitRAKE



Joined: 21 Jul 2003
Posts: 3055
Location: vpcmipstrm
bitRAKE
When the time comes to choose it is always a leap.

My grandfather said to at least look at 10 used cars before buying one. Using reflection to deconstruct ones choices is not a happy way to live. I reflect to constructively make better future choices - knowing that I did the best I could at the time.

I've had my whole apartment filled with computers, but I'm not there right now. I've purchased on the rising edge of the curve, but I'm not there right now. We'll see where I am in five years.
Post 21 Mar 2009, 16:46
View user's profile Send private message Visit poster's website Reply with quote
Display posts from previous:
Post new topic Reply to topic

Jump to:  


< Last Thread | Next Thread >
Forum Rules:
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
You cannot attach files in this forum
You can download files in this forum


Copyright © 1999-2020, Tomasz Grysztar. Also on GitHub, YouTube, Twitter.

Website powered by rwasa.