Author Topic: New OS/4 kernel :: SVN4227  (Read 149000 times)

Lars

  • Hero Member
  • *****
  • Posts: 1271
  • Karma: +65/-0
    • View Profile
Re: New OS/4 kernel :: SVN4227
« Reply #135 on: January 06, 2016, 10:03:14 am »
1) I am not paid by ArcaNoae, at least up to this point. I am free to say and do what I want.
2) I neither have the time nor the interest to thrash any project. I am stating an opinion.
3) you need not expect any statement from ArcaNoae. But I am in discussion with David every now and then. And he is certainly not monitoring this group. And he tests all his stuff with a limited set of kernels. If they find a problem with the kernel they will concentrate on patching the official kernels.
4) the next ACPI.PSD is in the queue. Do you think it is sensible for the OS4 team to offer patches to ACPI.PSD to make it work with their kernel ad infinitum ?

Yes, I'll move on.

Martin Iturbide

  • OS2World NewsMaster
  • Global Moderator
  • Hero Member
  • *****
  • Posts: 4713
  • Karma: +41/-1
  • Your Friend Wil Declares...
    • View Profile
    • Martin's Personal Blog
Re: New OS/4 kernel :: SVN4227
« Reply #136 on: January 06, 2016, 02:13:54 pm »
Hi Lars

I also have accumulated frustration of all the things that I want for the platform, but telling other projects and people that "your project is a waste of time, you better do this" it is not polite and you are getting that people blocked to any suggestions you may have.

You are free to create your own forum thread and specify what should be the agenda for the platform under your own opinion and try to motivate people to help you.
If you ask me to organize any documentation on the EDM/2 wiki, to request IBM republishing permission for some OS/2 documentation, to help raising money for an OSS project, to post anything on OS2World to help your projects or anything that is on my hands, I will help you for sure. But it has to be a positive collaborative thing.

The final word Lars, is that I really want to help you and do something for the benefits for platform users, but it needs to be positive.

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

Dariusz Piatkowski

  • Hero Member
  • *****
  • Posts: 1317
  • Karma: +26/-0
    • View Profile
Re: New OS/4 kernel :: SVN4227
« Reply #137 on: January 07, 2016, 02:23:34 am »
Lars,

...
4) the next ACPI.PSD is in the queue. Do you think it is sensible for the OS4 team to offer patches to ACPI.PSD to make it work with their kernel ad infinitum ?
...

So here is a real-world example why alternative approaches may be worth pursuing.

I have directly benefitted by subscribing to the Arca Noae's DRIVER package, thanks to that I now have a pretty stable SMP environment, although (and this is the big enough of a justification in my mind) of the total of 6 CPU cores I can only boot the machine up with 5, otherwise I end up with the dreadful "PMSHELL blue screen". Having logged a ticket with support organization it was prompty written of as a "known PMSHELL issue"...hmm...so really?...no further debug, no additional follow-up, just a plain "...can't be fixed..."? And this is for software that I actually paid money for...

Now, from my point of view, I am literally locked-out of using nearly 20% of my computing power simply b/c the current de-facto SMP solution won't let me do it. I'm not saying that this is the SOLE reason...heck, I am no os2ldr/krnl/PMSHELL guru, but as long as I have apparently reached the end of the road in one approach I will pursue others if they arise so that maybe in the course of testing/researching I will learn something and maybe someone else in our tiny community will benefit from.

Beyond just that very selfish reason above, here is a more precise one: if the /ST=0 parameter in the ACPI.PSD drive does in fact get ignored, and the authors have set some arbitrary value for the size of the stack, then I would consider this to be a defect. So if the OS4 team patches the binary to actually correct this behaviour then I'm actually questioning why the "core" OS2 maintainer team refuses to fix the code itself?

OS4User

  • Sr. Member
  • ****
  • Posts: 406
  • Karma: +10/-0
    • View Profile
Re: New OS/4 kernel :: SVN4227
« Reply #138 on: January 07, 2016, 11:09:38 am »
Beyond just that very selfish reason above, here is a more precise one: if the /ST=0 parameter in the ACPI.PSD drive does in fact get ignored, and the authors have set some arbitrary value for the size of the stack, then I would consider this to be a defect. So if the OS4 team patches the binary to actually correct this behaviour then I'm actually questioning why the "core" OS2 maintainer team refuses to fix the code itself?

The case is a bit more complicated. OS/2 kernel offers to PSD very small stack (2672 bytes). It is ok for os2apic.psd and  it looks like not enouph for ACPI.PSD. That is why ACPI.PSD does some actions to increase PSD stack offered by the kernel. It is not an easy thing since PSD stack is organized on per processor base. At the same time ACPI.PSD recognizes /ST key to set desirable PSD stack size on user demand.  ACPI.PSD does not allow the user to set stack size = 0 (as it is really senseless), but if one finds a way (the patch helps) to set "ST=0", ACPI.PSD will not go through the procedure of increasing PSD stack offered by the kernel.

OS/4 starting from svn ver 4250 has unified PSD/interrupt stack with 12k size thus ACPI.PSD has no need to increase it. At the same time, "not easy" ASPI.PSD procedure for increasing PSD stack, simply spoils internal data structure of OS/4 kernel. That is way ACPI.PSD has to be patched and /ST=0 has to be used in order to work with OS/4.
« Last Edit: January 07, 2016, 02:54:50 pm by OS4User »

Sergey Posokhov

  • Full Member
  • ***
  • Posts: 169
  • Karma: +8/-6
    • View Profile
    • OS/2 API Research
Re: New OS/4 kernel :: SVN4227
« Reply #139 on: January 14, 2016, 10:49:41 pm »
« Last Edit: January 14, 2016, 10:53:40 pm by Sergey Posokhov »

Sergey Posokhov

  • Full Member
  • ***
  • Posts: 169
  • Karma: +8/-6
    • View Profile
    • OS/2 API Research
Re: New OS/4 kernel :: SVN4227
« Reply #140 on: January 14, 2016, 10:54:05 pm »

muffetta

  • Guest
Re: New OS/4 kernel :: SVN4227
« Reply #141 on: January 14, 2016, 11:16:02 pm »
..SVN4512, 2Core, HD(1080p). Both in range 50% - 80%, max 97% (CPU 1)

Sergey Posokhov

  • Full Member
  • ***
  • Posts: 169
  • Karma: +8/-6
    • View Profile
    • OS/2 API Research
Re: New OS/4 kernel :: SVN4227
« Reply #142 on: January 15, 2016, 10:58:05 pm »
SVN 4512.
http://www.youtube.com/watch?v=LiyS-DnkMeY
100% sadness (endogenous depression).

muffetta

  • Guest
Re: New OS/4 kernel :: SVN4227
« Reply #143 on: January 25, 2016, 11:34:18 am »
..SVN4515. The same test running in flash and html5, using both OS4 and original eCS kernels: in this case the high CPU load seems not connected to OS/4. For the html5, obviously is light, since 1028HD is not available (max av. 360p).