OS/4 kernel and extended KEE

[OS4User]

Started from ver 4032 OS/4 kernel will load clock03b.sys instead of clock01.sys.

Clock03b uses Extended KEE, which includes KernSetIRQ and KernEOI (http://ru2.halfos.ru/core/docs/keeOS4.html)

Clock03b contains two drivers  CLOCK$ and TIMER0$ so timer0.sys  no more needed, it will not be loaded any way even if it will not be removed from config.sys

As you may guess Clock03b has  two _32_  bit only  interrupt service routines (for IRQ0 and IRQ8).

Clock03b  has many advantages comparative to pair clock01 + timer0  -  it is based on source code from clock01 by  Tsurgaev A.V. (Цургаев А.В.)

For those who is brave enough here is a link to download and test it. http://ru2.halfos.ru/core/downloads/kernel/os4/os2krnlSVN4032_unoff.zip

[OS4User]

A bug is fixed (freezing before switching to graphic mode)  http://ru2.halfos.ru/core/downloads/kernel/os4/os2krnlSVN4033_unoff.zip

[David McKenna]

  I gave this (4033) a try and when booting I get this error: 'File not found or corrupted: acpi4.sys'. I don't have this file (eCS 2.1). It seemed to work well for about 5 minutes, but then my desktop froze and I had to hard reset. Is there anything in particular you want tested?

[OS4User]

  I gave this (4033) a try and when booting I get this error: 'File not found or corrupted: acpi4.sys'. I don't have this file (eCS 2.1).

acpi4.sys  is optional  -  system just will not use it.

It seemed to work well for about 5 minutes, but then my desktop froze and I had to hard reset.

It would be great if you could show com port log, especially at the moment of such freeze.

Is there anything in particular you want tested?

Nothing special just stability and working in general.

[guzzi]

What is ACPI4.SYS?

[Doug Bissett]

I gave 4033 a try on my Lenovo ThinkPad L530. When I select any option from OS2LDR, it starts, the screen goes black for about half a second, and the machine reboots.

Using showmem=1 causes it to show a bit of interesting data, but it goes black, and reboots, after pressing a key (even when I select to boot the normal kernel). No sign of a log file, and I don't have a com port to play with.

It is possible that I didn't set it up properly. The instructions are about as clear as mud.

[walking_x]

This model has supplied with EFI BIOS and 1155 socket CPU?
Then this is the second side of famous 512Mb bug

IBM OS2LDR does not see memory above this border - and this is bad

OS/4 loader can see it, but something wrong with memory tables or  BIOS or ACPI implementation on EFI... wrong in BIOS - the same bug exists on most of new chipsets... but I'm also saw messages about successful working of all 4Gb on EFI...

I'm think - OS/4 Team may trace kernel's execution on this place if they can get new PC with the same bug.

Just add MEMLIMIT=512 to OS2LDR.INI for now...

[Doug Bissett]

This model has supplied with EFI BIOS and 1155 socket CPU?
Then this is the second side of famous 512Mb bug

It is EFI BIOS, I am not sure if it is 1155 socket CPU. Yes, it is the famous 512 MB bug.

IBM OS2LDR does not see memory above this border - and this is bad

Well, it is not real bad. eCS operates reasonably well in 512 MB.

OS/4 loader can see it, but something wrong with memory tables or  BIOS or ACPI implementation on EFI... wrong in BIOS - the same bug exists on most of new chipsets... but I'm also saw messages about successful working of all 4Gb on EFI...

I'm think - OS/4 Team may trace kernel's execution on this place if they can get new PC with the same bug.

Just add MEMLIMIT=512 to OS2LDR.INI for now...

From what I have found out, there is a 4KB memory hole, for some unknown reason, at the 512 MB location. The rest of the memory holes seem to be above the 3 GB mark.

Adding memlimit=512 to OS2LDR.INI changes nothing. I will try some other things, when I find the time (which may not be soon).

[Doug Bissett]

Just as a quick test, to be sure that I have it installed properly, I put this on my Lenovo ThinkPad T510. It seems to work fine, so I am now confident that I have installed it properly on the L530.

[walking_x]

Adding memlimit=512 to OS2LDR.INI changes nothing. I will try some other things, when I find the time (which may not be soon).

This is actually strange, 512mb memory limit always helps.
May be 511 or 510? I'm don't know how it calculates - with or without 1st mb.

[OS4User]

What is ACPI4.SYS?

It will be an acpi driver for os/4 kernel. Current kernel is trying to load it, but in case it is missing nothing fatal happens.

[Doug Bissett]

Adding memlimit=512 to OS2LDR.INI changes nothing. I will try some other things, when I find the time (which may not be soon).

This is actually strange, 512mb memory limit always helps.
May be 511 or 510? I'm don't know how it calculates - with or without 1st mb.

I tried 500 too. Still the same problem.

[David McKenna]

  Since I last posted I have been using the OS4 kernel (retail) with no trouble. Maybe the freeze I got then was from something else (Flash??). I did have a POPUPLOG from DMISL.EXE which is something I've never had before...

  I have one question: Is os2krnl.ini necessary? I ask because I don't see anything in ___hlp$ that would make me think I need to add anything to os2krnl.ini, but maybe I don't understand the instructions....

[OS4User]

  Since I last posted I have been using the OS4 kernel (retail) with no trouble. Maybe the freeze I got then was from something else (Flash??). I did have a POPUPLOG from DMISL.EXE which is something I've never had before...

would be interesting to see

  I have one question: Is os2krnl.ini necessary? I ask because I don't see anything in ___hlp$ that would make me think I need to add anything to os2krnl.ini, but maybe I don't understand the instructions....

OS/4 kernel with switched ON preload option will load all files which will be necessary on stage 1 of boot process into a kind of memory disk and then the hard drive will not be touched until the stage 2.

For this purpose kernel  builds a list of such files. This list includes  all BASEDEVs + PSD from config.sys,   BASEDEVs which loads without config.sys,  SNOOPERS from snoop.lst  and files mentioned in os4krnl.ini

Actually os4krnl.ini serves for a user to enable him to add files which are needed to be preloaded and about which the kernel didn't guess.

If during boot kernel will try to load some file which was not preloaded - user will receive a message something like  "Please add driver.sys to os4kernel.ini".

In majority of cases os4krnl.ini will be absent or empty. I personally can't imagine a situation when something has to be added there, but who knows.

[David McKenna]

 Here's the POPUPLOG... always exactly the same, and I never get this with the eCS 2.1 kernel:

06-19-2013  10:34:26  SYS3175  PID 0021  TID 0002  Slot 0030
C:\OS2\DMISL.EXE
c0000005
00018b1a
P1=00000001  P2=0000001e  P3=XXXXXXXX  P4=XXXXXXXX 
EAX=00000000  EBX=17b20000  ECX=17b20000  EDX=00000000
ESI=17b20000  EDI=00000000 
DS=0053  DSACC=d0f3  DSLIM=5fffffff 
ES=0053  ESACC=d0f3  ESLIM=5fffffff 
FS=150b  FSACC=00f3  FSLIM=00000030
GS=0000  GSACC=****  GSLIM=********
CS:EIP=005b:00018b1a  CSACC=d0df  CSLIM=5fffffff
SS:ESP=0053:000d3e4c  SSACC=d0f3  SSLIM=5fffffff
EBP=000d3ef8  FLG=00010246

DMISL.EXE 0001:00008b1a

  I don't know what I do to cause it... I don't intentionally run DMISL and the POPUPLOG just appears every once in awhile.

  Thanks for the explanation of os2krnl.ini. I was not using preload=1 so that is why I didn't see any 'Please add..' messages. When I use preload=1, then I see a couple of 'Please add...' messages, mostly *.sym files which the retail kernel won't use anyway. Also, my boot drive is an SSD drive, which boots very fast with or without preload=1 (Intel CPU too).