Simple XP 32BIT 64Gb RAM (true Pae) Guide

[jaclaz]

12 hours ago, ruthan said:

Yeah, because otherwise someone could think that it working everywhere,

Rest assured that a lot of people will think anyway that, the concept that "sometimes it works, sometimes it doesn't" is written on almost any other post on this thread and on the "other" big one, some info that probably you missed, since you found needed to re-iterate it.

jaclaz
 

[Dibya]

2 hours ago, jaclaz said:

Rest assured that a lot of people will think anyway that, the concept that "sometimes it works, sometimes it doesn't" is written on almost any other post on this thread and on the "other" big one, some info that probably you missed, since you found needed to re-iterate it.

jaclaz
 

It happens to server 2003 itself . Some drivers are buggy . 

Those you have xeon or any server processor with server class mobo they can enjoy pae in full deal . I have fixed my issues with my drivers . In my new patch I have fixed 90% issues which ruthan is using now . Not publicly available.  I have fixed tons of usb driver , disk format issue etc .

Edited July 28, 2017 by Dibya

[jaclaz]

37 minutes ago, Dibya said:

It happens to server 2003 itself . Some drivers are buggy .

Well, since the PAE was disabled in XP SP2 EXACTLY BECAUSE there were tons of buggy drivers this is again not news.

The PAE feature was left in 2003 Server because (besides being actually *needed*) on that platform the amount of buggy drivers was much smaller and people actually buying/using Server 2003 had more possibilities/capabilities to use only certified (or however "better") hardware and drivers.

Let's see if citing directly the reference page by Mark Russinovich helps:

https://blogs.technet.microsoft.com/markrussinovich/2008/07/21/pushing-the-limits-of-windows-physical-memory/
 

Quote

However, by the time Windows XP SP2 was under development, client systems with more than 4GB were foreseeable, so the Windows team started broadly testing Windows XP on systems with more than 4GB of memory. Windows XP SP2 also enabled Physical Address Extensions (PAE) support by default on hardware that implements no-execute memory because its required for Data Execution Prevention (DEP), but that also enables support for more than 4GB of memory.

What they found was that many of the systems would crash, hang, or become unbootable because some device drivers, commonly those for video and audio devices that are found typically on clients but not servers, were not programmed to expect physical addresses larger than 4GB. As a result, the drivers truncated such addresses, resulting in memory corruptions and corruption side effects. Server systems commonly have more generic devices and with simpler and more stable drivers, and therefore hadn't generally surfaced these problems. The problematic client driver ecosystem led to the decision for client SKUs to ignore physical memory that resides above 4GB, even though they can theoretically address it.

Of course the decision (right or wrong) was made in or around late 2003/first half of 2004 and reflects the state of hardware (and drivers) at the time.

Later hardware (and drivers) may (or may not) be "better" or "worse", but given that next iteration of the Server line was Server 2008 (that came out early 2008, at a time where everyone was already on the 64 bit bandwagon) noone would have later extensively tested newer hardware (and drivers) with Server 2003 32 bit (of which BTW only Enterprise and Datacenter edition were licensed for more than 4 Gb via PAE), simply because *any* new Server install (be it 2003 or 2008) would have been on 64 bit hardware, and using the 64-bit version of the OS.

jaclaz
 

[ruthan]

What about some HW and SW boot drivers - white / black list? I dont mind to create one in Zoho or Google docs. In my case clean Windows working, but existing installation with lots of SW dont, so there are probably some evil imcompatible SW with own drivers needed

[jaclaz]

11 hours ago, ruthan said:

What about some HW and SW boot drivers - white / black list? I dont mind to create one in Zoho or Google docs. In my case clean Windows working, but existing installation with lots of SW dont, so there are probably some evil imcompatible SW with own drivers needed

Define "lots".  (and divide between hardware drivers and "software ones")

I mean, out of say 1,000 (one thousand) programs maybe 10 (ten) will have the need for a service/driver that may interfere with PAE and these would normally affect only the specific application working (i.e. only a small subset of the 10 should prevent booting or however affect other programs).

On the other hand hardware drivers will be the same no matter if it is a fresh install or if it is an older one (on the SAME hardware).

If you have (on the same hardware and with the same hardware drivers installed in both) 2 installs of XP, one PAE patched and one not, and if the issue is at boot time, making a boot log for each and comparing them should be enough to restrict the suspect to a handful of drivers, very likely one or two.

jaclaz
 

[ruthan]

Quote

I mean, out of say 1,000 (one thousand) programs maybe 10 (ten) will have the need for a service/driver that may interfere with PAE and these would normally affect only the specific application working (i.e. only a small subset of the 10 should prevent booting or however affect other programs).

   I use lots of such programs - when i check Autoruns log, in Drivers tab is lots things and less than i thing 1/2 is hardware related.

Quote

On the other hand hardware drivers will be the same no matter if it is a fresh install or if it is an older one (on the SAME hardware).

   Yeah clean i mean, installed XP with HW drivers but without other user programs.

Quote

If you have (on the same hardware and with the same hardware drivers installed in both) 2 installs of XP, one PAE patched and one not, and if the issue is at boot time, making a boot log for each and comparing them should be enough to restrict the suspect to a handful of drivers, very likely one or two.

  Logs, this is problem im using bootlog start options, but i have checked C:\Windows\ntbtlog.txt + minidump directory, but there wasnt any record for last boot on some computers, only some old ones.. im not sure in i can of PAGE_FAULT*IN_*NON_PAGED AREA is anything logged.. I or look into wrong log.

  Logs (ntblog + autoruns) from main computer attached, its old installation  - very customized: / I know, i know, some ghost drivers should be removed.. and there are strange things like paragon GPT driver + etc.. I but i tried to remove some usuall suspects, without success.
https://dl.dropboxusercontent.com/u/2741044/Win32PAEDebug/PaeDebug.7z

HW: Asrock Z97 Extreme 4 / 3.1  board  (with Asmedia Sata + USB and Intel LAN crapp) + Core i7 + 32 GB an Geforce 970. Non pae boot is fine, Safe mode with PAE is fine, PAE no safe boot mode is failing.

Edited August 5, 2017 by ruthan

[jaclaz]

It's queer.

At least the kernel and hal should be logged.

In any case with a single ntbtlog.txt you are not likely to go very far.

The idea is to have TWO  of them, on the same machine, with same hardware drivers, coming from two installations, one "old" (and not working with the PAE patch) and a new, clean one.

Or having two of them from the same install, one when it is working (without PAE patch) and one when it is not working (with the PAE patch installed).

It is by comparing the two logs that you can likely find the issue.

jaclaz

 

 

[ruthan]

Hmm i would extected something simple like - im starting this, done, im starting that.. crash or missing entry, Linux boot logs is quite simple in this logic. Why it is more complex and complicated, its boot multithreaded, or some already started drive could crash after a while?  If logging such, could be coded some logging wrapper to make it better, because if debug sucks, development usually sucks too.

  Here is good boot log, same machine, same installation:
https://dl.dropboxusercontent.com/u/2741044/Win32PAEDebug/ntbtlog-MasterOnd-GoodBootNoPae.txt

Edited August 6, 2017 by ruthan

[Destro]

Ya I tried a couple different patches.  Still waiting for a 95% working on all machines stable one for XP.

[Dave-H]

So, I have a server motherboard, with dual Xeon processors.
Do you think I should give Dibya's patch a shot?
I have a full backup and can easily repair Windows XP using the other operating systems I have installed as regards substituting system files.
I've downloaded the package, but it's a bit vague about what I actually have to do to try it.
Is it just a matter of substituting all the system files with those in the package, or do I have to edit boot.ini as well?
I have 8GB of RAM installed (for 64bit Windows 10) and it would be nice if XP could also use it all.

[LoneCrusader]

21 minutes ago, Dave-H said:

So, I have a server motherboard, with dual Xeon processors.
Do you think I should give Dibya's patch a shot?
I have a full backup and can easily repair Windows XP using the other operating systems I have installed as regards substituting system files.
I've downloaded the package, but it's a bit vague about what I actually have to do to try it.
Is it just a matter of substituting all the system files with those in the package, or do I have to edit boot.ini as well?
I have 8GB of RAM installed (for 64bit Windows 10) and it would be nice if XP could also use it all.

I'm not sure what all is in Dibya's patch, but I have used the XP PAE patch in an experiment and it works... mind you it wasn't tested extensively though. You need two patched files (NTOSKRNL.EXE and HAL.DLL [the correct ones for your setup; these are chosen during installation]) and a Server 2003 version of USBPORT.SYS. Other drivers out there may suffer from the dreaded "bugs" that everyone seems to be afraid of, but USBPORT.SYS is the only one known for sure to be a problem at this time.

[Dave-H]

Thanks, I hope Dibya will come back and say exactly what has to be done with his patch.
I use USB a lot on my system, so I don't want to risk anything that might compromise the stability of that function, and Dibya reckons he's cracked that issue.

[dencorso]

My friends, IMO it's not mature yet. In any case, please do read this thread. With 8 GiB one can create a RAMDISK and then put the swapfile in it, the rest can be used to hold the Temporary Internet Files, and the caches for FF, Opera and Chrome: and it'll get auto cleaned up at every reboot.

[LoneCrusader]

1 hour ago, dencorso said:

My friends, IMO it's not mature yet. In any case, please do read this thread. With 8 GiB one can create a RAMDISK and then put the swapfile in it, the rest can be used to hold the Temporary Internet Files, and the caches for FF, Opera and Chrome: and it'll get auto cleaned up at every reboot.

I agree that it's not "mature" in the sense that it has not been widely tested. However I believe there exists a slight bias of opinion around here against the PAE solution.

I guess it all comes down to which reverse engineer's knowledge, information, and experiments you put the most faith in. Most of the discussion revolves around attempting to use the SP1 version of HAL.DLL because something was supposedly "removed" from it. I discussed this issue with rloew last year and he ran some tests and examined the different HAL's. He said there was nothing of interest in the SP1 HAL. This, and seeing the patch work is good enough for me. If I weren't already running XP x64 when I need it instead of XP x86 I would probably be using it.

[ruthan]

There machines with lazy BIOS memory areas implementation, like my X99 MSI board, where you dont have after XP 32 installation typical - 2.8, 3.5 GB of avalaible RAM, only in 9xx MB - and it that case is such patch necesary, because otherwise after you install typical set of resident apps for poweruser you end in constant swapping and imposibility to execute apps.

   Its same with Vista and Windows 7 32bit.. but there even doesnt exist such patches.. Jokes is that those OS are official on paper supported.. I have not energy to contact official support and force them to make proper bios..

Update: I had bright moment of decency and wrote MSI to fix it, in first round of support, i got question with MB is detecting all memory, these guys, they need MB serial number, but usually you can add to requst picture or file or even link it, because they even cant download something from dropbox or other link.. I directed them directly into this supermicro manual table:

It would be nice, if in all manual would be such table, it saw such thing first time, even i tried to select proper X99 MB.

Edited August 9, 2017 by ruthan