Author Topic: New OS/4 kernel :: SVN4227  (Read 148943 times)

dbanet

  • Guest
Re: New OS/4 kernel :: SVN4227
« Reply #120 on: August 01, 2015, 03:36:46 pm »
It's not open sourced at all is it?  I never can find a SVN server, should I even look?

No, unfortunately, not.

OS4User

  • Sr. Member
  • ****
  • Posts: 406
  • Karma: +10/-0
    • View Profile
Re: New OS/4 kernel :: SVN4227
« Reply #121 on: August 27, 2015, 07:39:23 pm »
 OS/4 tested HW list is updated http://www.os2world.com/wiki/index.php/Phoenix_OS/4#OS.2F4_Tested_Hardware_List

current   svn version  is 4410

OS4User

  • Sr. Member
  • ****
  • Posts: 406
  • Karma: +10/-0
    • View Profile
Re: New OS/4 kernel :: SVN4227
« Reply #122 on: December 20, 2015, 09:46:19 pm »
OS/4 kernel svn4500 is released. Starting from this ver OS/4 kernel loads clock03c.sys instead of  clock03b.sys
New clock driver is in a package with kernel.
« Last Edit: December 25, 2015, 08:56:43 pm by OS4User »

OS4User

  • Sr. Member
  • ****
  • Posts: 406
  • Karma: +10/-0
    • View Profile
Re: New OS/4 kernel :: SVN4502
« Reply #123 on: December 25, 2015, 08:56:14 pm »
OS/4 kernel svn4502 is released. The changes have occurred mostly in clock$.
If HPET is available in the computer, clock$ will use it for providing DosTmrQueryTime/Freq with 64bit time stamp.

Dariusz Piatkowski

  • Hero Member
  • *****
  • Posts: 1317
  • Karma: +26/-0
    • View Profile
Re: New OS/4 kernel :: SVN4502
« Reply #124 on: January 02, 2016, 02:04:02 am »
OS/4 kernel svn4502 is released. The changes have occurred mostly in clock$.
If HPET is available in the computer, clock$ will use it for providing DosTmrQueryTime/Freq with 64bit time stamp.

Re-tried this latest release...I am currently successfully running SMP in 5-core mode with ACPI 3.22.08, so that is my starting point.

Here are some brief results:

1) booting with OS4krnl produces a hard trap right after PMSHELL (I can not tell at all what is going on, so this looks like a candidate for a COM port debug, I need to get this organized though)

2) booting with OS4krnl but ACPI.PSD produces a trap with a complaint regarding missing ACPI4.BPD file...did some researching here but best I could do was to find an earlier forum post (http://www.os2world.com/forum/index.php?topic=785.0) and having re-read the on-line docs multiple times I still can not tell whether I should be attempting to patch the ACPI.PSD driver with the patch files supplied at the OS4 site (I am guessing the answer is a NO because those patches are meant for 3.22.06 release whereas I'm running 3.22.08 here)

I will a couple more combinations here, like RETAIL vs DEBUG kernels and /APIC vs none for OS4APIC.PSD driver.

OS4User

  • Sr. Member
  • ****
  • Posts: 406
  • Karma: +10/-0
    • View Profile
Re: New OS/4 kernel :: SVN4502
« Reply #125 on: January 02, 2016, 10:42:13 am »
1) booting with OS4krnl produces a hard trap right after PMSHELL (I can not tell at all what is going on, so this looks like a candidate for a COM port debug, I need to get this organized though)

Similar stuff sometimes happens on  my computer as well. In the comport log it looks like this:

Code: [Select]
DelayHardError SYS3171: 4 string(s):
 Pid 0020  Tid 0001  Slot 0045  HobMte 053A
 C:\OS2\PMSHELL.EXE
c0000005
1f93b94a
P1=00000002  P2=0003ff14  P3=XXXXXXXX  P4=XXXXXXXX 
EAX=00000000  EBX=1f93b93c  ECX=00040028  EDX=0004f564
ESI=00000000  EDI=0004f3f4 
DS=0053  DSACC=d0f3  DSLIM=ffffffff 
ES=0053  ESACC=d0f3  ESLIM=ffffffff 
FS=150b  FSACC=00f3  FSLIM=00000030
GS=0000  GSACC=****  GSLIM=********
CS:EIP=005b:1f93b94a  CSACC=d0df  CSLIM=ffffffff
SS:ESP=0053:0003ff18  SSACC=d0f3  SSLIM=ffffffff
EBP=00040054  FLG=00010206

PMWP.DLL 0001:0000b94a
This happens once every 100 boots approx. I don't know the reason and it happens much later after the init of the kernel has finished.

It's possible that it's some old bug that has no relation to OS/4. I remember on the original kernel there sometimes happened situations when there appeared PMshell background and nothing more. Somebody offered a solution with organizing a kind of delay  before PMshell start, but I cannot find that post.

Of course, this can be something else, in this situation comport log should help.

2) booting with OS4krnl but ACPI.PSD produces a trap with a complaint regarding missing ACPI4.BPD file...did some researching here but best I could do was to find an earlier forum post (http://www.os2world.com/forum/index.php?topic=785.0) and having re-read the on-line docs multiple times I still can not tell whether I should be attempting to patch the ACPI.PSD driver with the patch files supplied at the OS4 site (I am guessing the answer is a NO because those patches are meant for 3.22.06 release whereas I'm running 3.22.08 here)

Yes, ACPI.PSD has to be patched and used with /ST=0. Indeed,  the patch does very simple thing:
in ACPI.PSD there is something like this:
Code: [Select]
if (ST==0)
{
    ST=4096
}

patch makes this code out of it:
Code: [Select]
if (ST==0)
{
    ST=0
}

thus, if one asks for ST=0 he does get ST=0. I would like to stress that this patch does not "spoil"  ACPI.PSD and it still can be used with original kernel.

ACPI4.BPD - it is an optional acpi driver. It is not obligatory for booting the system.

It is true that there are patches only for 3.22.06 and 3.22.03, that's because nobody had asked for something else. If you insist,  patch for 08 will also be produced. At the same time you may try OS/4 with 06 (ACPI.PSD only  the rest may remain 08)

I will a couple more combinations here, like RETAIL vs DEBUG kernels and /APIC vs none for OS4APIC.PSD driver.

waiting for result  (better with comport log ;) )
« Last Edit: January 02, 2016, 12:31:39 pm by OS4User »

Dariusz Piatkowski

  • Hero Member
  • *****
  • Posts: 1317
  • Karma: +26/-0
    • View Profile
Re: New OS/4 kernel :: SVN4502
« Reply #126 on: January 02, 2016, 03:14:00 pm »
1) booting with OS4krnl produces a hard trap right after PMSHELL (I can not tell at all what is going on, so this looks like a candidate for a COM port debug, I need to get this organized though)

Similar stuff sometimes happens on  my computer as well. In the comport log it looks like this:

Code: [Select]
DelayHardError SYS3171: 4 string(s):
 Pid 0020  Tid 0001  Slot 0045  HobMte 053A
 C:\OS2\PMSHELL.EXE
c0000005
1f93b94a
P1=00000002  P2=0003ff14  P3=XXXXXXXX  P4=XXXXXXXX 
EAX=00000000  EBX=1f93b93c  ECX=00040028  EDX=0004f564
ESI=00000000  EDI=0004f3f4 
DS=0053  DSACC=d0f3  DSLIM=ffffffff 
ES=0053  ESACC=d0f3  ESLIM=ffffffff 
FS=150b  FSACC=00f3  FSLIM=00000030
GS=0000  GSACC=****  GSLIM=********
CS:EIP=005b:1f93b94a  CSACC=d0df  CSLIM=ffffffff
SS:ESP=0053:0003ff18  SSACC=d0f3  SSLIM=ffffffff
EBP=00040054  FLG=00010206

PMWP.DLL 0001:0000b94a
This happens once every 100 boots approx. I don't know the reason and it happens much later after the init of the kernel has finished.

It's possible that it's some old bug that has no relation to OS/4. I remember on the original kernel there sometimes happened situations when there appeared PMshell background and nothing more. Somebody offered a solution with organizing a kind of delay  before PMshell start, but I cannot find that post.

What you describe is precisely what I see with my standard 3.22.08 setup upon nearly every re-boot. Meaning, the PMSHELL window displays (the blue background) and halts...CTL-ALT-DEL still reboots the PC though. I find that if I hit the CTL-ESC key combination repeatedly as the PMSHELL window is coming up I can sucessfully boot though. This procedure works pretty well when I run my 6-core CPU in 5-core mode, as soon as I enable all the cores the PMSHELL nearly always hangs regardless of what I do during bootup. So this certainly appers to be some type of timing issue...

...Yes, ACPI.PSD has to be patched and used with /ST=0. Indeed,  the patch does very simple thing:
in ACPI.PSD there is something like this:
Code: [Select]
if (ST==0)
{
    ST=4096
}

patch makes this code out of it:
Code: [Select]
if (ST==0)
{
    ST=0
}

thus, if one asks for ST=0 he does get ST=0. I would like to stress that this patch does not "spoil"  ACPI.PSD and it still can be used with original kernel.

ACPI4.BPD - it is an optional acpi driver. It is not obligatory for booting the system.

It is true that there are patches only for 3.22.06 and 3.22.03, that's because nobody had asked for something else. If you insist,  patch for 08 will also be produced. At the same time you may try OS/4 with 06 (ACPI.PSD only  the rest may remain 08)

Well, I will try 3.22.06, but if at all possible (not sure how much work this is) I'd prefer to run the matching 3.22.08 instead.

I will a couple more combinations here, like RETAIL vs DEBUG kernels and /APIC vs none for OS4APIC.PSD driver.

waiting for result  (better with comport log ;) )

...nah, no go here...all other combinations produced the same result...a hang at PMSHELL, so I will focus on teh comport log and will attempt to re-test with the patched 3.22.06 first.

Co-incidentally, what is the reason for the different version of DOSCALL1.DLL? I repeated all my tests with both versions (14.106 SMP and the one included in the 4502 drop) but the results were the same.

Thanks for the quick response!!!

Sergey Posokhov

  • Full Member
  • ***
  • Posts: 169
  • Karma: +8/-6
    • View Profile
    • OS/2 API Research
Re: New OS/4 kernel :: SVN4227
« Reply #127 on: January 02, 2016, 08:34:19 pm »
It looks like a bug in VBE2GRAD.DLL, also known as Panorama.
Version "3 Dec 2015" works with OS/4 SVN4506 (both debug and retail), previous doesn't... even with SVN4200.
« Last Edit: January 02, 2016, 08:37:39 pm by Sergey Posokhov »

Lars

  • Hero Member
  • *****
  • Posts: 1271
  • Karma: +65/-0
    • View Profile
Re: New OS/4 kernel :: SVN4227
« Reply #128 on: January 04, 2016, 05:56:57 pm »
It remains a secret of the OS4Krnl developer team why they removed the trace and strace facilities.
Without that none of the USB drivers can be traced as they are making good use of OS/2 built in tracing mechanism. Tracing is about the only chance for a developer to do offline analysis of problems surfacing at the user's site (unless you want to recode your own tracing facility).

OS4User

  • Sr. Member
  • ****
  • Posts: 406
  • Karma: +10/-0
    • View Profile
Re: New OS/4 kernel :: SVN4227
« Reply #129 on: January 04, 2016, 09:37:15 pm »
It remains a secret of the OS4Krnl developer team why they removed the trace and strace facilities.
Without that none of the USB drivers can be traced as they are making good use of OS/2 built in tracing mechanism. Tracing is about the only chance for a developer to do offline analysis of problems surfacing at the user's site (unless you want to recode your own tracing facility).

I have never use trace facilities, same time KernPrintf was sufficient  to debug clock03 and some other drivers.

Lars

  • Hero Member
  • *****
  • Posts: 1271
  • Karma: +65/-0
    • View Profile
Re: New OS/4 kernel :: SVN4227
« Reply #130 on: January 05, 2016, 12:49:42 am »
That's because you obviously love to reinvent the wheel.
And enabling trace for the kernel components of OS/2 can help, for example, in order to investigate when OS2DASD is called with what parameters. You cannot replace that with your home brewn way of outputting text to whereever.

Martin Iturbide

  • OS2World NewsMaster
  • Global Moderator
  • Hero Member
  • *****
  • Posts: 4711
  • Karma: +41/-1
  • Your Friend Wil Declares...
    • View Profile
    • Martin's Personal Blog
Re: New OS/4 kernel :: SVN4227
« Reply #131 on: January 05, 2016, 01:05:05 pm »
Hi Lars

It remains a secret of the OS4Krnl developer team why they removed the trace and strace facilities.

You can always create your own kernel and then you can add the trace functionality.

Quote
That's because you obviously love to reinvent the wheel.
Or you can suggest it to the OS4 team in a more constructive way.... or offer yourself to help on that project.

Regards
Martin Iturbide
OS2World NewsMaster
... just share the dream.

Lars

  • Hero Member
  • *****
  • Posts: 1271
  • Karma: +65/-0
    • View Profile
Re: New OS/4 kernel :: SVN4227
« Reply #132 on: January 05, 2016, 10:41:09 pm »
Hi Lars

It remains a secret of the OS4Krnl developer team why they removed the trace and strace facilities.

You can always create your own kernel and then you can add the trace functionality.

Quote
That's because you obviously love to reinvent the wheel.
Or you can suggest it to the OS4 team in a more constructive way.... or offer yourself to help on that project.

Regards

Create my own kernel ? You must be kidding. What sense does it make if every OS/2 user (guess there are about 250 left on planet earth) creates his own kernel ?
Suggesting anything to the OS4 team is lost ground. They are proud of their own proprietary solution.
They have given NO reason whatsoever why they removed something that worked just fine and that is WELL DOCUMENTED to exist.
Two things are for sure:
1) I am certainly not wasting any time testing their kernel
2) ArcaNoae is certainly not wasting any time to get things working also with the OS4 kernel

The sad thing is that these guys could really help in developing sensible things for OS/2: USB3 Support, help fix UNIAUD, whatever.
What they do is wasting time but of course they can waste all the time they want, after all, it's their time.

Lars

OS4User

  • Sr. Member
  • ****
  • Posts: 406
  • Karma: +10/-0
    • View Profile
Re: New OS/4 kernel :: SVN4227
« Reply #133 on: January 06, 2016, 12:21:22 am »
With due respect to you, I can't agree with all your statements.

Suggesting anything to the OS4 team is lost ground.

On the contrary. OS/4 team will gladly hear reasonable advice, suggestions, bug reports, etc.

They have given NO reason whatsoever why they removed something that worked just fine and that is WELL DOCUMENTED to exist.

May be one day trace facility will be re-enabled.

What they do is wasting time but of course they can waste all the time they want, after all, it's their time.

There are people who use OS/4 and find it useful. Moreover, there are several HW where the original kernel wasn't able to run at all(in any configurations with any loaders), only OS/4 could run there. Thus OS/4 isn't that much of a waste of time as it may seem.

Martin Iturbide

  • OS2World NewsMaster
  • Global Moderator
  • Hero Member
  • *****
  • Posts: 4711
  • Karma: +41/-1
  • Your Friend Wil Declares...
    • View Profile
    • Martin's Personal Blog
Re: New OS/4 kernel :: SVN4227
« Reply #134 on: January 06, 2016, 04:19:57 am »
Hi Lars

I regret your attitude, you are not being constructive.  If you don't like OS4 kernel and you can not give constructive feedback, just move on.

Quote
2) ArcaNoae is certainly not wasting any time to get things working also with the OS4 kernel
Now you speak for Arca Noae? Are they paying your to thrash other projects?  Or is it that Arca Noea has an interest on seeing OS4Kernel fail? Should we expect to see on Arca Noae's next announcement something like "Do not use OS4 Kernel"?

Quote
The sad thing is that these guys could really help in developing sensible things for OS/2: USB3 Support, help fix UNIAUD, whatever.
You can suggest but people does with their time whatever they like. I had seen many developers choking on their own arrogance telling others what to do (for free and in their own time) in very "ugly" ways. It is better to ask for things in a constructive way.

Quote
What they do is wasting time...
Sorry Lars, but it even sounds like you are talking to bad influences over the phone, you must chill out and understand that everybody does with their time whatever they want. This is a community and people does what they want.

Regards.

« Last Edit: January 06, 2016, 01:44:39 pm by Martin Iturbide »
Martin Iturbide
OS2World NewsMaster
... just share the dream.