Author Topic: New OS/4 kernel :: SVN4227  (Read 149001 times)

OS4User

  • Sr. Member
  • ****
  • Posts: 406
  • Karma: +10/-0
    • View Profile
Re: New OS/4 kernel is available :: SVN4166 :: screen03.sys release
« Reply #15 on: February 05, 2014, 06:59:37 am »
I will try the USB to serial adapter, if I can find it. Meanwhile, I guess I will run the os4krnlr version, to see what else shows up.

USB to serial adapter will not help you  if your test machine does not have com port.

dbanet

  • Guest
Re: New OS/4 kernel is available :: SVN4166 :: screen03.sys release
« Reply #16 on: February 07, 2014, 07:43:27 pm »
Well, after a bit more testing, I found that using os4krnlr boots okay, but trying to use UNIMIX.EXE causes a strange problem. It works, as expected, with the base eCS 2.2 beta II install (IBM kernel). If I try it with os4krnlr, UNIMIX.EXE responds with "SYS0193: UNIAUD cannot be run in a OS/2 session." (yes, it does say "UNIAUD"). I am not sure what that is all about, but something is wrong.

I've (accidentally) reproduced this problem on my machine and I had to report it to OS/4 kernel developers. Fortunately, the problem entirely resides in the uniaud.dll: the bug in it was being compensated by the bug in the original kernel causing uniaud.dll to work without any problems. When the bug in the kernel was fixed by Phoenix people, uniaud.dll stopped to work properly on the OS/4 kernel.

You are welcome to download another version of uniaud.dll, which does not contain the bug in it and which works fine with the OS/4 kernel, as well as with the original one. Replace the existing one, which resides in the \MMOS2\DLL directory, with the downloaded one. To accomplish this, you may (or may not) need the UNLOCK.EXE utility from lxLite, the opensource utility package, to unlock and replace the DLL if it was locked by uniaud or someone else. After that, reboot your system.

Doug Bissett

  • Hero Member
  • *****
  • Posts: 1593
  • Karma: +4/-2
    • View Profile
Re: New OS/4 kernel is available :: SVN4166 :: screen03.sys release
« Reply #17 on: February 07, 2014, 09:06:26 pm »
Quote
USB to serial adapter will not help you  if your test machine does not have com port.

I expect that you are right, but I will try it anyway, if I can find the device.   :(

Quote
You are welcome to download another version of uniaud.dll,

Did that, and yes, it does fix the problem (will that fix be submitted to the official sources?). The sound still doesn't work properly, but it works the same as it does with the IBM kernels. I guess I need to rattle Paul Smedley's chain, to get an updated UNIAUD.

Paul Smedley

  • Hero Member
  • *****
  • Posts: 2092
  • Karma: +159/-0
    • View Profile
Re: New OS/4 kernel is available :: SVN4166 :: screen03.sys release
« Reply #18 on: February 07, 2014, 10:26:26 pm »
Quote
You are welcome to download another version of uniaud.dll,

Did that, and yes, it does fix the problem (will that fix be submitted to the official sources?). The sound still doesn't work properly, but it works the same as it does with the IBM kernels. I guess I need to rattle Paul Smedley's chain, to get an updated UNIAUD.

IIRC the source for uniaud.dll isn't at netlabs. Which uniaud build are you running? and which device? I have a build using alsa 1.0.24 source, but was never able to get 1.0.25 working.  Later builds are integrated into the linux kernel and not available as standalone packages which makes things more problematic in terms of updates.

dbanet

  • Guest
Re: New OS/4 kernel is available :: SVN4166 :: screen03.sys release
« Reply #19 on: February 07, 2014, 11:57:34 pm »
Paul, the DLL was not fixed by the Phoenix project, that's just a DLL from one of the binary uniaud distributions, and actually I don't know from which one is it. I have to do further testing and figure out what releases posted on netlabs' Trac are broken, but the DLL in the latest release is reported to work just okay with OS/4, although I didn't test it myself.

The old broken DLL may still remain on some users' systems, for example, due to not updating uniaud on time, or due to other reasons, so updating uniaud manually may be a good thought.

Doug Bissett

  • Hero Member
  • *****
  • Posts: 1593
  • Karma: +4/-2
    • View Profile
Re: New OS/4 kernel is available :: SVN4166 :: screen03.sys release
« Reply #20 on: February 08, 2014, 07:30:13 am »
Quote
Which uniaud build are you running?

It is from UniaudUpdate1.09.06-2.02.01.zip which is the latest that I know about. Earlier versions won't load at boot time (they pop up something about no hardware found, as I recall).

Quote
and which device?

It is in a Lenovo ThinkPad L530. From PCI.EXE:
Code: [Select]
Bus 0 (PCI Express), Device Number 27, Device Function 0
 Vendor 8086h Intel Corporation
 Device 1E20h 7 Series/C210 Series Chipset Family High Definition Audio Controller
 Command 0006h (Memory Access, BusMaster)
 Status 0010h (Has Capabilities List, Fast Timing)
 Revision 04h, Header Type 00h, Bus Latency Timer 00h
 Self test 00h (Self test not supported)
 Cache line size 64 Bytes (16 DWords)
 PCI Class Multimedia, type Unknown!
 Subsystem ID 21F717AAh Unknown
 Subsystem Vendor 17AAh Lenovo
 Address 0 is a Memory Address (64-bit) : F2610000h..F2613FFFh
 System IRQ 22, INT# A
 New Capabilities List Present:
   Power Management Capability, Version 1.1
     Does not support low power State D1 or D2
     Supports PME# signalling from mode(s) D0, D3hot, D3cold
     PME# signalling is currently disabled
     3.3v AUX Current required : 55 mA
     Current Power State : D0 (Device operational, no power saving)
   Message Signalled Interrupt Capability
     MSI is disabled
     MSI function can generate 64-bit addresses
   PCI Express Capability, Version 1
     Device/Port Type :
       Root Complex Integrated Endpoint Device
       Unsupported Request Severity is Non-Fatal
     Device Status :
       AUX Power Detected
     Link Capabilities :
       Maximum Link speed : Unknown (00h)!!
       Maximum Link Width : xReserved
       Link Port Number   : 0
     Link Control :
       Asynchronous Clocking in Use
     Link Status :
       Current Link speed : Unknown (00)!!
       Current Link Width : x0
but Win7 Pro 64 claims there is also a Realtek High Definition Audio device, that I don't see with PCI.EXE. I think one of them might be for the HDMI interface.

The other "complication", and what I think is causing my problem, is that windows uses a DOLBY driver for it. I think that may be the problem because I do hear sounds, but it is like being in an echo chamber. There is no sign of hangs, so I don't think it is IRQ related. I suspect that there may be a new control, that will turn the DOLBY off, and there are probably other controls that make the DOLBY feature do different things, that UNIMIX doesn't know about. There is no sound, at all, until I use UNIMIX to turn on the speakers, and turn the volume up.

Quote
The old broken DLL may still remain on some users' systems, for example, due to not updating uniaud on time, or due to other reasons, so updating uniaud manually may be a good thought.

The UNIAUD.DLL, that I had, is from eCS 2.2 beta 2. Bldlevel says:
Code: [Select]
Signature:       @#UNIAUD:0.3.0#@ Vladest Software
Vendor:          UNIAUD
Revision:        0.03.0
File Version:    0.3
Description:     Vladest Software
The one that works says there is no version information. The date on the file is today, so that is meaningless. The same "broken" file (bldlevel) is what came with eCS 2.1 as well (dated June 26, 2008), so anybody with a reasonably current eCS will have the broken file. I am pretty sure that reporting the problem to the eCS bug tracker would simply get discarded as being a problem with an unsupported kernel, unless we can find something else that fails with it.

FWIW, I do have Linux Mint 15, and the sound works properly. I will need to figure out what driver it uses.

Quote
Meanwhile, I guess I will run the os4krnlr version, to see what else shows up.

So far, everything else seems to work well, but I really haven't done a lot with it.

Paul Smedley

  • Hero Member
  • *****
  • Posts: 2092
  • Karma: +159/-0
    • View Profile
Re: New OS/4 kernel is available :: SVN4166 :: screen03.sys release
« Reply #21 on: February 08, 2014, 09:21:03 am »
Quote
Which uniaud build are you running?

It is from UniaudUpdate1.09.06-2.02.01.zip which is the latest that I know about. Earlier versions won't load at boot time (they pop up something about no hardware found, as I recall).

Quote
and which device?

It is in a Lenovo ThinkPad L530. From PCI.EXE:
Code: [Select]
Bus 0 (PCI Express), Device Number 27, Device Function 0
 Vendor 8086h Intel Corporation
 Device 1E20h 7 Series/C210 Series Chipset Family High Definition Audio Controller
 Command 0006h (Memory Access, BusMaster)
 Status 0010h (Has Capabilities List, Fast Timing)
 Revision 04h, Header Type 00h, Bus Latency Timer 00h
 Self test 00h (Self test not supported)
 Cache line size 64 Bytes (16 DWords)
 PCI Class Multimedia, type Unknown!
 Subsystem ID 21F717AAh Unknown
 Subsystem Vendor 17AAh Lenovo
 Address 0 is a Memory Address (64-bit) : F2610000h..F2613FFFh
 System IRQ 22, INT# A
 New Capabilities List Present:
   Power Management Capability, Version 1.1
     Does not support low power State D1 or D2
     Supports PME# signalling from mode(s) D0, D3hot, D3cold
     PME# signalling is currently disabled
     3.3v AUX Current required : 55 mA
     Current Power State : D0 (Device operational, no power saving)
   Message Signalled Interrupt Capability
     MSI is disabled
     MSI function can generate 64-bit addresses
   PCI Express Capability, Version 1
     Device/Port Type :
       Root Complex Integrated Endpoint Device
       Unsupported Request Severity is Non-Fatal
     Device Status :
       AUX Power Detected
     Link Capabilities :
       Maximum Link speed : Unknown (00h)!!
       Maximum Link Width : xReserved
       Link Port Number   : 0
     Link Control :
       Asynchronous Clocking in Use
     Link Status :
       Current Link speed : Unknown (00)!!
       Current Link Width : x0
but Win7 Pro 64 claims there is also a Realtek High Definition Audio device, that I don't see with PCI.EXE. I think one of them might be for the HDMI interface.

There are some fixes in ALSA 1.0.25 for a "Lenovo Y530" - I wonder if these would help?

dbanet

  • Guest
Re: New OS/4 kernel is available :: SVN4166 :: screen03.sys release
« Reply #22 on: February 08, 2014, 04:15:19 pm »
Doug, your Lenovo Thinkpad seems to have one of the following graphic adapters: Intel GMA HD / Intel HD Graphics 3000 / Intel HD Graphics 4000. Each of them seem to be unsupported by SNAP, you are likely to use Panorama driver with it.

Please, report what Panorama driver version are you using, how does it work with the original kernel, and with OS/4 kernel.
Turning ACPI.PSD off and booting with OS/4 may be a good test for the new screen03.sys driver -- how then Panorama will work?

Sigurd Fastenrath

  • Hero Member
  • *****
  • Posts: 576
  • Karma: +27/-0
  • OS/2 Versus Hardware - Maximum Warp!
    • View Profile
Re: New OS/4 kernel is available :: SVN4166 :: screen03.sys release
« Reply #23 on: February 08, 2014, 04:40:26 pm »
There are some fixes in ALSA 1.0.25 for a "Lenovo Y530" - I wonder if these would help?

I would be happy as well with the X230T I have.   ::)

Doug Bissett

  • Hero Member
  • *****
  • Posts: 1593
  • Karma: +4/-2
    • View Profile
Re: New OS/4 kernel is available :: SVN4166 :: screen03.sys release
« Reply #24 on: February 08, 2014, 06:31:11 pm »
Quote
There are some fixes in ALSA 1.0.25 for a "Lenovo Y530" - I wonder if these would help?

I wouldn't be too surprised if they did. It seems that the 530 (perhaps the 520, 540, and 230) series started using some new hardware (keyboard, touchpad, stickmouse, sound). Some of them work okay, others only partly.

Quote
Please, report what Panorama driver version are you using, how does it work with the original kernel, and with OS/4 kernel.

I am currently using Panorama-1.06, and it works great with either kernel. It is AUDIO that doesn't work very well (apparently, no fault of OS/4). I tried SNAP, and it also works, in VESA mode, with the IBM kernel (I haven't tried it with OS/4). The video is described, by PCI.EXE, as:
Code: [Select]
Vendor 8086h Intel Corporation
 Device 0166h 3rd Gen Core processor Graphics Controller

Quote
Turning ACPI.PSD off and booting with OS/4 may be a good test for the new screen03.sys driver

I am using acpi-3.22.03 (no parameters), with no apparent problems. As far as I can tell, screen03.sys loads, and I have seen no video problems, so far. Timer0.sys, on the other hand, does fail, but that doesn't seem to hurt anything. I did turn off ACPI, and the only things that changed were that it uses 1 CPU, and it won't power off at shutdown (both expected).

I just tried os4apic. It works, as expected, using /APIC (two CPUs, and high IRQs), and without /APIC (two CPUs, and low IRQs).

There are also a couple of DOSCALL1 files in the package. I haven't tried to use them. What are they for?

Doug Bissett

  • Hero Member
  • *****
  • Posts: 1593
  • Karma: +4/-2
    • View Profile
Re: New OS/4 kernel is available :: SVN4166 :: screen03.sys release
« Reply #25 on: February 08, 2014, 06:39:11 pm »
Quote
and which device?

I should add, that it says "Mixer Intel ID 2806" while booting.

dbanet

  • Guest
Re: New OS/4 kernel is available :: SVN4166 :: screen03.sys release
« Reply #26 on: February 08, 2014, 07:57:03 pm »
Doug, as I recall, 16-bit timer0.sys fail is okay because its function is now fully substituted by the clock03b.sys system driver written by Phoenix.

I'm sorry I can't help you with the sound, maybe you may try messing with unimix controls more ;)

Thank you for your positive feedback, it's really appreciated.

Doug Bissett

  • Hero Member
  • *****
  • Posts: 1593
  • Karma: +4/-2
    • View Profile
Re: New OS/4 kernel is available :: SVN4166 :: screen03.sys release
« Reply #27 on: February 08, 2014, 08:58:42 pm »
I just tried OS/4 on my Lenovo ThinkPad T510. This thing is a little different than the L530:

It does not have a real COM port (it does have a software modem), but it hangs after the first "Loading" line, if I don't enable the debug port.

Hmmm. eCS 2.2 beta 2, on the T510, does NOT have the new UNIAUD.DLL file. I guess I need to discover where that came from.  I remember that the UNIAUD driver was automatically detected on the T510 (in eCS 2.1, and eCS 2.2 beta 2), but I had to select it on the L530 (no sound chip was detected, in both versions of eCS). In all cases, I manually updated to UniaudUpdate1.09.06-2.02.01, after install, but that does not change UNIAUD.DLL.

Then, I tried loading OS/4 with the QSINIT loader. That works, and I can use the QSINIT RAMDISK. It works with the L530 too.

Quote
as I recall, 16-bit timer0.sys fail is okay because its function is now fully substituted by the clock03b.sys system driver

I was thinking that that was the case. I didn't see any indication that it should be REMed. I REMed it, and everything seems to be okay, so far (a few minutes).

Quote
I'm sorry I can't help you with the sound, maybe you may try messing with unimix controls more

I think that UNIMIX is missing a few controls that I need. Anyway, that all works the same as with the IBM kernel, so it is not an OS/4 problem.

Quote
Thank you for your positive feedback, it's really appreciated.

It would be more positive, if the documentation was a little better. Somebody who hasn't played with this stuff would be in real trouble, I think.

The OS/4 kernel, using the QSINIT loader, or the OS/4 loader,  seems to be working well, and it does enable all of my memory, which is a great breakthrough. I can use it with ACPI, with no parameters, which enables both processors, and uses high IRQs. I need to try hyperthreading.

Now, I need to get back to real life, for a while.   ;D

dbanet

  • Guest
Re: New OS/4 kernel is available :: SVN4166 :: screen03.sys release
« Reply #28 on: February 08, 2014, 10:32:24 pm »
I just tried OS/4 on my Lenovo ThinkPad T510. This thing is a little different than the L530:

It does not have a real COM port (it does have a software modem), but it hangs after the first "Loading" line, if I don't enable the debug port.
Hm? Can you, please, describe more precisely? I think I didn't understand that much.
So, it hangs just after you select the kernel in the loader, and it does work if you comment out the dbport statement in os2ldr.ini, doesn't it?
What's with the software modem? Is it, urgh, connected to the COM port?


Quote
as I recall, 16-bit timer0.sys fail is okay because its function is now fully substituted by the clock03b.sys system driver

I was thinking that that was the case. I didn't see any indication that it should be REMed. I REMed it, and everything seems to be okay, so far (a few minutes).
There is no indication it should be REMed just because you don't actually need it to be REMed -- that's done for compatibility and prevents you from the necessity to have two different CONFIG.SYS files.


Quote
Thank you for your positive feedback, it's really appreciated.

It would be more positive, if the documentation was a little better. Somebody who hasn't played with this stuff would be in real trouble, I think.
That's the thing, unlike kernel developing, I think, I can do myself. At least documentation for the end-user, not a developer. I'll consider making somewhat a FAQ.

dbanet

  • Guest
Re: New OS/4 kernel is available :: SVN4166 :: screen03.sys release
« Reply #29 on: February 08, 2014, 10:48:22 pm »
All, there are actually two options of getting a kernel log, excluding using real COM port, which are very useful if you don't have one. The second option is experimental and allows obtaining the log even if it is impossible to load the system properly. Unfortunately, you are not able to use this now, because it is highly experimental and is not yet released.

The first one is old enough to be forgotten by everyone, but seems to work :). Just do
copy oemhlp$ log.txt
and read it
e log.txt
VoilĂ  :)

UPD:
I've just been corrected.
You should read oemhlp$ if you do not have ACPI and
you should read ___hlp$ if you do have ACPI.PSD, and,
you should read oemhl_$ if you have ACPI4.SYS (well, you don't).
« Last Edit: February 08, 2014, 11:04:20 pm by Boris »