Question for Roberto about your basic files...

[Roderick Klein]

Hello Roberto,

Since I do not have your email address, I wanted to ask here. You posted in the QT thread the following:
https://www.os2world.com/forum/index.php/topic,3252.msg39838.html#msg39838
You write:
"The UA.Bas file eliminates a block of one or more processors that tends to go 100%, and when executing it"

So why do you "poke" certain memory addresses inside the DOS VDM itself ?
Or put differently the DOS VDM can not read or write to the memory where the Dooble code is loaded.

Or is purely based on the idea to increase the CPU load on the CPU and slow down the Dooble code execution ?

I am just curious but I am having a hard time understanding how this is supposed even have an effect ?

Roderick

[roberto]

Thanks for opening, a new thread. I have my job and I do this for hoby and time does not give me
for all.
It would be good for someone else to learn to use this test tool, before making modifications
firm in the system.
I will try to answer your questions and try to teach you to do it.

1- Remember that I am modifying the dll s of the / mmos2 / not of the dooble, when the dooble uses those DLL.
2- The dooble, use it as a reference to see whether or not (for this you need a program that almost
It works, but that something lacks)
3- DVM sessions are executed before the dooble, so the dooble, when executing uses
The DLLs of the system, and these DLLs already have some strategically placed poke.
4-To execute poke & hxxxxxxxxxx, 1, that one tells the system runs as fast as you can
Always, he always says runs, another value, different from 1, could tell the system, quiet
  Slowly, you are with an 8088 processor, take time we are not in a hurry, the processor is slow, or it is
  A 80286 or ....You can choose worse values

5- It is never about loading the processor with more work, you can see major speeds with
one more load of the VDM , sessions are despicable charges.

6- Everything I say is not based on anything more than my experience, and my tests. I can be in error.

7- To be able to get the pokes that can interbose during the use of the dooble, you have to run the dooble
and Thesseus, because
analyze. You always have to use physical memory, (thanks Lars, this helped me a lot)
8- In the first five or 10 poke of the DLL or exe or sys that you want to modify you will have some type of result,
good or bad. And that will tell you if that DLL is likely to be modified.

9- And now we have practiced, from yesterday to today the UA.Bas to change, I send you a little advance, ua3.bas.
With these modifications, the dooble with 512mb Cache, and the unraveled single procesor, runs as if it would be
marked, with the advantages of multiprocess. But it fails and gives an error in Socks in console.
I look at the UA3.Bas and I look at the socks, what pokes I can modify, if they are few for one better.
Then I arrive at UA4.Bas, also attached. And Sock errors are removed, but slow, very slow,
but it works. What usually happens here is that there is a lack of continuing to correct things, and when you correct them
Flow, we will see it in the future.

10- You comment, that you are supposed to do, because you must be careful that it does not negatively
affect other programs. I want to say, for example, running UA.Bas, the Thunderbird when closing produces
 an error. I had never done it before. I hope the Thunderbird problem is solved, if the problem is solved
  by closing the dooble tabs, some of them especially the second Tab, tends to generate an error when
  closing. How is it possible if you do nothing, that the Thunderbid fails?

Saludos

[roberto]

Hi Roderick
I attach au5.bas and au6.bas
UA5.Bas was a mistake and you disagree, I didn't like the results
UA6.Bas detects a failure in the USB WiFi mouse, and search and incorporate a new DLL amou034.dll, it seems to me that the
Mouse problem is solved, some more tests and can be confirmed. I have left a video and state more than 3
Hours function, with 6 tabs open, about 750MB free. This is improving.
Current errors:
www.ebay.es fail in single procesor unmark:
js: A preload for 'https://ir.ebaystatic.com/pictures/aw/pics/s_1x2.gif' is foun
d, but is not used because the request credentials mode does not match. Consider
 taking a look at crossorigin attribute.
js: A preload for 'https://ir.ebaystatic.com/pictures/aw/pics/s_1x2.gif' is foun
d, but is not used because the request credentials mode does not match. Consider
 taking a look at crossorigin attribute.
js: A preload for 'https://ir.ebaystatic.com/pictures/aw/pics/s_1x2.gif' is foun
d, but is not used because the request credentials mode does not match. Consider
 taking a look at crossorigin attribute.
[497:13:0520/205051.926000:FATAL:scoped_file.cc(43)] Check failed: . : Bad file
number (9)

Killed by SIGTRAP
...
Moved 01f1_0d.TRP to C:\var\log\app\646932ad-01f1_0d-QTWEBENGINEPROCESS-exceptq.
txt


www.ebay.es -- work ok with single procesor mark

www.ibm.com/es-es --- show this:
js: Uncaught (in promise) TypeError: Array.from(...).filter(...).at is not a fun
ction
js: Uncaught (in promise) TypeError: Array.from(...).filter(...).at is not a fun
ction
js: jQuery.Deferred exception: Cannot read property 'attributes' of undefined Ty
peError: Cannot read property 'attributes' of undefined
    at h (https://1.www.s81c.com/common/v18/drupal/js/www.js:9:208475)
    at HTMLDocument.<anonymous> (https://1.www.s81c.com/common/v18/drupal/js/www
.js:9:216402)
    at e (https://1.www.s81c.com/common/v18/drupal/js/www.js:9:43513)
    at t (https://1.www.s81c.com/common/v18/drupal/js/www.js:9:43815) undefined
js: Uncaught TypeError: Cannot read property 'attributes' of undefined
js: Access to XMLHttpRequest at 'https://login.ibm.com/v1/mgmt/idaas/user/status
/' from origin 'https://www.ibm.com' has been blocked by CORS policy: No 'Access
-Control-Allow-Origin' header is present on the requested resource.
js: +++DBDM-ERROR > ajaxCall: IDaaS-User-Status : -1 : error
js: Survey: isIBMer value is:  0
js: Survey: isIBMer value in SSA is :  true
js: Survey : Events counts in session : 0
js: Survey : Events counts in session : from SSA
js: Survey : Events counts in page : 0
js: Survey : Events counts in page : from SSA undefined
js: Survey : Events counts in total views : 2
js: Survey : Events counts in total views : from SSA 2

I have edited the link with reference to the FFMPEG-chromium-6.3.2.ZIP, there are two files with the same name, one with
3 Files, and another with hours of difference with 4 files, use the 4 files.
saludos

[Roderick Klein]

I read it all and I am gladly proven wrong.

But best I can tell you are stating that your basic files modify the mmos2 files
Dll files in memory ? Is that correct ,?

You see best I can tell an os/2 dos vdm  can in no way
Modify os2 kernel memory.  Its protect mode operating system.
The memory is locked between os/2 vdm and the os/2 program.section.

I am really curious how Lars explained thuis to you..

[fusion]

You see best I can tell an os/2 dos vdm  can in no way
Modify os2 kernel memory.  Its protect mode operating system.
The memory is locked between os/2 vdm and the os/2 program.section.

You're totally right.. I'm not sure about any of this.

roberto, are you running these dll files through a disassembler so you know what you're changing? Where are the addresses coming from? What basic interpreter are you using, and do you perhaps have a helper driver to allow you to access memory like that? I have a strong suspicion that those POKEs aren't actually doing anything at all.

If we make the leap and assume they are doing something, every POKE in these files changes the memory value to 1. It simply makes no sense:

I could see it working if you were to POKE in a RET so that a bit of code that crashes doesn't get run. "1" is not RET, but would be the first part of an ADD instruction.

If you're changing 16 or 32 bit numbers, none of the examples change a complete number, but only a portion (POKE only POKEs one byte). If the 1 is meant to change a bit mask, again, where does this information come from?

We're hoping to be convinced, but also don't want you to buy in to snake oil, so to speak.

[Lars]

I read it all and I am gladly proven wrong.

But best I can tell you are stating that your basic files modify the mmos2 files
Dll files in memory ? Is that correct ,?

You see best I can tell an os/2 dos vdm  can in no way
Modify os2 kernel memory.  Its protect mode operating system.
The memory is locked between os/2 vdm and the os/2 program.section.

I am really curious how Lars explained thuis to you..

Hi Roderick,

I cannot remember having explained anything to Roberto.
At least I hope, I didn't...

Lars

[roberto]

All your questions are answered in other messages.
Do you have something to contribute?
saludos

[Roderick Klein]

All your questions are answered in other messages.
Do you have something to contribute?
saludos

Well sorry about this. I was just wanting to know how your basic scripts would work. And if they work for you then no problemo.
I just wanted to know how that is supposed to work. And we just do not understand how you can modify OS/2 OS section for a DOS VDM.

Basic point is you cannot modify the kernel memory of the OS/2 sub-system from a DOS VDM. So no my questions where not answered.
I am just left with more questions then answers. But no problem.

Thanks,

Roderick

[fusion]

Do you have something to contribute?

Yes: I'm sorry to have to be frank, but the posts about QBASIC are jibberish. I tried to ask some leading questions to hopefully make you reconsider what you think, but it doesn't seem to have worked.

There is no "1 to make it go fast" or "some other number for quiet" or any of this nonsense. People have specifically used OS/2 to run DOS programs because OS/2 DOESN'T ALLOW the type of thing you claim to be doing.

To use your C64/Vic20 references, you're basically doing the equivalent of POKEing the C64 machine and then going across the room to look for whether the Vic20 is running better or not.

Since you now know this, anyone who reads your QBASIC posts and tries what you describe will henceforth have been purposefully misled by you. You're wasting their time both for having read it, and possibly having tried it themselves.

I would focus my efforts on posting just the TRAP logs.

[Dave Yeo]

A simple test, start 2 vdm sessions, poke some memory in one and peek at the same memory in the other and see if the pokes affect the other VDM.