Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.

Messages - Roderick Klein

Pages: [1] 2 3 ... 35
Programming / Re: Building Qt5
« on: November 23, 2022, 12:00:07 am »
Dear builders Qt5 please turn off the AVX (and AVX2) code. Due to its presence, the browser does not display the pages. And it is impossible to use libraries. I don’t understand why this code is included if it does not work on OS/2 correctly?

Please open a ticket for this at Github for this issue on the BWW page.

And on which websites does happen so it can be reproduced ?


Roderick Klein

Programming / Re: UEFI and GPT not developed by ArcaNoae - by whom else?
« on: November 12, 2022, 04:25:29 pm »
Hi Tom,

thank you very much for the explanation, my comment before was a bit misleading, sorry.

I mean I have no clue who is doing the work.

I do know what UEFI, Secure Boot and GPT means.

Thanks again!!!

My problem with doing a presentation on this topic was that I felt I could just NOT start with secure boot. Without explaining what UEFI and GPT do.
Part of this has todo with the fact that you have an ESP and what a UEFI loader is and what it does in the OS. Maybe that I made to complicated.

What we did at OS/2 VOICE was port 2 Linux tools to OS/2 that enable you to sign a UEFI binary with your own certificate. This is what Arca Noae has done. On a system that has key management in the UEFI BIOS menu you can import this cert and the system will then accept that UEFI binaries from Arca Noae.

Another tool that was created was to import the cert under ArcaOS if the UEFI BIOS menu does not provide this feature.

I primarly had this all of this created as I read that different things in Windows 11 MIGHT fail. Such as certain Windows updates. This was not clear at the time what that could be (I lost the link to that article). I myself also use Bitlocker on all my Windows 10 + 11 installations. And if you switch of secure boot Windows sometimes asks for your Bitlocker recovery key. I do not find that convient.

Its still being tested right now. All tools where ported from Linux and there for once finished will be put into a public repo on


Programming / Re: UEFI and GPT not developed by ArcaNoae - by whom else?
« on: November 12, 2022, 04:12:48 pm »
Good sum up Sigurd. Only -
UEFI BIOS needs GPT HDD Layout (only Exception: The UEFI BIOS uses the CSM Mode if present!)
is wrong I think. I'm not really sure but I think UEFI can work with MBR disks without problems. Only Windoze does not support UEFI without booting from a GPT disk. But....

MBR disc can work but are not common in a UEFI envrioment. It also comes with all the legacy headaches MBR disc support bring.
From my perspective its easier to setup GPT dual boot then MBR dual boot.


Hardware / Re: Multi-stream Audio - A Theory
« on: November 12, 2022, 04:01:02 pm »
I'd like to preface this by saying that I'm not very familiar with OS/2's driver structure/methodology.
This idea might be completely moot from the get-go but I figured there's no harm in seeing where discussion regarding it's feasibility might lead; especially considering how essential multi-stream audio is for any desktop OS and in particular, nearing the release of a browser which is very liable to support multimedia playback in a way we haven't seen for quite a few years.

I was setting up a software audio mixer on Windows earlier today. Like a lot of audio mixers - it interfaces with programs using a driver which relays all audio streams to the mixer program, the mixer program then sends the audio through to the hardware driver as a single combined stream.

It's worth noting a lot of Linux distros which use ALSA, from which Uniaud is based, run PulseAudio as a middleware audio mixer much in the way I have described in the above paragraph. I'm unsure if this concept has ever been toyed with in the OS/2 ecosystem, let alone if anything like it has existed in the past decade (accounting for the fact that some software designed for ECS no longer works in ArcaOS, i.e QEMU) .

Since ALSA(Uniaud) has already been ported to OS/2 with Paul Smedley to thank for the updated 32 bit builds, could there be any possibility for a PulseAudio port? Considering that the project has been active since 2004, perhaps an older build could be a candidate for this undertaking. Keeping in mind that ALSA and PulseAudio are often used in tandem as the two halves of the Linux audio package and it's prominence in being a go-to solution, I am hoping they would be able to cooperate as intended to a similar extent in OS/2.

Lastly, for what it's worth, I put together a rough diagram for how such a solution may possibly work - it does match quite closely to Pulseaudio's flow however.

- Ibrahim

That drawing is only the simple part.

MMOS/2 supports multiple audio sttreams. According to some documentation in MCP 2 (which eCS and ArcaOS are based on), this got "broken".
However I have heard that people can install the SBLIVE driver from Sander van Leeuwen and play MMOS/2 audio system sounds AND MP3 music at the same time. That is because both the audio hardware, the driver and MMOS/2 support this.

Having a maxier DLL in one application that can mix audio is great. But if you have another OS/2 application, then what ?
It means in the browser you can mix audio from lets say two youtube video's ?. But anything outisde the browser could not access the card.

What makes it even more complicated is the adding VDM support to this mix. (WINOS2 and DOS support).

The bottom line is that you need todo most of this stuff at driver level.  Then the audio mixing is done by the hardware.


Programming / Re: Building Qt5
« on: October 30, 2022, 11:58:29 pm »
I'll try with PROTECTONLY=YES, which IIRC, will by itself stop all the mdos/winos2 stuff loading. The build died this time with,
Code: [Select]
cc1plus.exe: out of memory allocating 65536 bytes after a total of 65536 bytesThis is getting quite frustrating, must have spent over 24 hours of actual compiling, though with the hangs etc it is hard to judge, obj directory is now 2GB, and this is a rel build, a debug probably would use more memory, but maybe not if less optimized.
It is hard to imagine doing much developing on the code if it takes a day to recompile and as this is my usual use machine, it also means not using it while compiling due to the tight memory. Of course this machine is a dozen years old so a newer fast machine would help. Still would be the memory problem and the way things go, the next update to the code is likely to use more memory.
The future does not look bright

While your conclusion about the future not looking because of the issue you encountered  compiling the webkit DLL,  based on the results, is logical. I would hwoever like to suggest to be more reserved drawing that conclusion the "future not looking bright". Its again a lack of time we are having as community. The issue you encrountered is known but we have not looked into how to resolve this further.

So idea's exists how the we *migh*t be able to work around this issue. But this requires research. Nobody best I can tell ever did research why the system is running out of memory. I discussed the issue with Steve Levine months ago. I forgot what he had mentioned as options to look into.  Its month ago.
But I would not say we have hit the end of the track. No research has been done why the lockup occurs/what is the memory consumed by.

It might also be possible to drastically reduce the compile time on OS/2. But will require some changes in the compiler. Again I discussed this months ago. So I might have not worded this all properly.
On a MacOS box I think compiling the webkit DLL takes about 4 hours. So we can still improve this all.
But it needs to hit the top of the list.

I BTW wonder how much RAM protectonly=yes setting will give you back. Since you do not load any DOS or WINOS2 sessions I think the implact will be minimal....


Programming / Re: Building Qt5
« on: October 30, 2022, 11:27:49 pm »
Im afraid that's because OS2 sees only 1MB of mem on this machine.
Also a exceptq log in var about CC1PLUS

The crash happens when configure  tests for precompiled header support, expected.
I take it you mean 1GB of ram. You still have the swap file. When I was building earlier versions of Firefox, around 10 IIRC, needed VIRTUALADDRESSLIMIT=3072 or linking xul.dll would gracefully die with an out of spill memory error. Then I had a simm die leaving me with 1.5 GBs of memory, I could still compile as long as not much else was running but if something like SM was running, the system would die with a swap file full error. There was lots of room for the swap file, but it is limited to 2GB.
Basically without enough ram, it should gracefully die rather then issue weird errors. A long time back I had a 486, OS/2 was trapping all the time so I moved to Linux, which at the time meant recompiling the kernel for your hardware, as no module support. Took a week as it kept dying with bus errors when trying to compile the kernel. Eventually discovered the computer had a broken 16 bit DMA thingy and things were stable if I didn't use 16 bit DMA. Just an example of broken hardware screwing up compilation.
Currently here, I'm still having problems with detecting libvpx and even after hacking qt5\qt5\webengine\src\buildtools\configure.json to detect libvpx, it died in webrtc trying to compile the intree libvpx.
I also get hard lockups, perhaps caused by my Snap system having to high of a VIRTUALADDRESSLIMIT. I think I'll wait for the next beta, do a new install and retry then.

I have seen this occur with SNAP loaded on some systems where setting the virtualaddresslimit to high can cause the system to lockup.
I have however NEVER seen this happen with Panorama loaded.


Programming / Re: Building Qt5
« on: October 23, 2022, 11:56:41 pm »
A bug was opened by Gregg Young about how to bulld QT.

Do you know the issue number for Gregg's bug?

No sorry.


Programming / Re: Building Qt5
« on: October 23, 2022, 10:50:48 pm »
Hey All,

Given it's been a LONG time since there have been any Qt5 updates, I figured I'd start a thread on trying to build Qt5 from the sourcecode at the bww github, and attempting to follow the build instructions at

I've previously tried building and failed, but have never documented my experience or compared it to what others have experienced.

At the moment, I'm doing a fresh checkout of the code, on a new bare metal install of AOS 5.07

More to come...



Dmitry just finished a import QT fix related to menu;s not working. He is working on the genereal stability of the Dooble browser. As been stated his quick depeature from Russia and getting settled somewhere else in the E.U. has taken a lot of time. This has resulted in development being more or less at a standstill for about 6 months. But a fix on LIBCX has been worked on and genereal stabilisation of Dooble is being worked on.

A bug was opened by Gregg Young about how to bulld QT.

Roderick Klein

Setup & Installation / Re: I think I've found the magic number
« on: October 22, 2022, 11:33:38 pm »
The only thing the SET ORD_DOS32PERFSYSCALL=931322574.615478067200 is set the environment variable.
But OS/2 will never pick this up, best I can tell. Is there any code in OS/2 that uses this SET statement ?


Setup & Installation / Re: I think I've found the magic number
« on: October 22, 2022, 11:23:54 pm »
I mean how does this set statement influence the complete system stability ?
You previously made some sort of basic file that runs in a DOS VDM (where you not that person ).
How can the OS2KRNL and other ring 0 code be influenced by a VDM basic program ?
I am not saying your work does not work, I am just puzzled how it exactly works.



Setup & Installation / Re: MARKEXE vs EXECMODE - differences?
« on: October 21, 2022, 10:20:38 pm »
Hi everyone,

I've had a persistent problem with some apps putting a complete soft-lock of my machine. This normally happens during application start-up, but occasionally it would also happen throughout execution. I know, it sounds weird, but it's absolutely consistent and in appears to be related to an underlying component DLL, although I'm still working through pegging that detail.

If you could make a list of these apps and if they are tha letst version see if you make memory dump by pressing CTRL ALT F10 key combination twice.  With a memory dump it might be possible to determine why the hang occurs.


Setup & Installation / Re: MARKEXE vs EXECMODE - differences?
« on: October 21, 2022, 08:40:48 pm »
One other thing, that seems to cause the problem, is to use VirtualAddressLimit greater than 2560 (this may be machine dependent).

It is machine dependent, and dependent on things like the graphic card. The PCI address space is at the top of memory, including graphics memory and I believe if the VirtualAddressLimit is too high, they can overlap causing instability and crashes.
I currently have it set at 3072 and just went over 500 hours of uptime with SeaMonkey running most of that time steadily, Thunderbird much of the time and Dooble at times. It finally locked up hard when opening a QT4 card game.

I think the PCI address space does not interfere with the virtual address limit. I think the issue with the graphics memory and virtual address limit only occurs with SNAP and not Panroama, but I could be wrong.


Setup & Installation / Re: MARKEXE vs EXECMODE - differences?
« on: October 21, 2022, 08:26:29 pm »
I can't go above VIRTUALADDRESSLIMIT=2816 with this system. Otherwise AOO can't open some dlls anymore. But it also depends on other settings.

I've solid hangs a while after I start Thunderbird (beside SM) with my different calendars located at different sites. On my older system I had solid hangs often when playing videos. No clue if audio or video related.

This happens with SNAP or Panroama I seem to recall this virtualaddresslimit issue's with SNAP ?


Hardware / Re: Turn off Laptop Display
« on: October 21, 2022, 08:16:47 pm »
When using Panorama in UEFI mode, AOS 5.1 is incapable of changing any video setting itself because it doesn't contain a bona-fide video driver - i.e. code that knows how to manipulate a video card's registers.. This "mini-driver" is a really thin BIOS emulation that returns data cached by os2ldr at boot time and (I believe) allows you to issue video commands that will be ignored. Instead, AOS uses a workaround that capitalizes on the fact that OS/2 has never been able to change resolutions on-the-fly.

When 'aosldr' starts up, it uses the UEFI firmware's boot-time services to read 'os2.ini' and determine the desired resolution / refresh rate / color depth. It then uses the firmware to set the video accordingly. Later, when it loads the OS/2 kernel, it loses access to those boot-time video services, but the video settings remain in place, unchanged and unchangable for the rest of the session.

Consequently, if turning off the display requires any sort of direct manipulation of the video card, you are S-O-L.

So how about any INT10h calls that do not change the resolution but actually want to write something to the screen like int 10h / AH = 0Ah or int 10h / AH = 13h ? Will these work ? Or is AN int10h support limited to only the calls that control/query the resolution ?

You mean during boot when device drivers are loading or when the OS is started ?
Or are you talking ab0ut DOS session that use that INT 10 interface. in DOS sessions that should work via VEFI.SYS (replacement for VSVGA.SYS).



Hardware / Re: Turn off Laptop Display
« on: October 20, 2022, 10:17:29 pm »
Rich his summary is correct. If you want to control the video output port, or the brightness a screen on a UEFI booted system you need a native video driver. The UEFI interface does not provide standards for this. The only thing that MIGHT work is look if the ACPI backlight can be implemented. But I know not enough about the code and what more code is needed. (See the link my previous post in this thread).  But this will not work on all equipment...

The bottom line you should see UEFI is that compared to BIOS it provides as little services to an OS as possible. The OS should do as much stuff as possible.  UEFI does provide services to itself so an UEFI OS loader can load the needed files from a boot device for example. But when exitfrombootservices is called, the disc I/O services go away.


Pages: [1] 2 3 ... 35