Author Topic: Final Request for Help for the Final Warp 4 Image Clone  (Read 10736 times)

Sigurd Fastenrath

  • Sr. Member
  • ****
  • Thank You
  • -Given: 13
  • -Receive: 35
  • Posts: 337
    • View Profile
Final Request for Help for the Final Warp 4 Image Clone
« on: April 05, 2013, 11:55:07 am »
I would like to ask for help to solve the last problems I do have before I would like to freeze my DFSEE Installation Image of the tailored OS/2 Warp 4. Wim Brul is so kind to help me once again to recreate the Finger/Touch Support for the Samsung Serie 7 Slate as it was broken because of a BIOS Update some time ago. I am able to use this Image and install it with DFSEE in about 5 minutes on each and every PC i have, but I am customizing as well one special edition for my Slate.
I already have most things done (Warp 4.52 with FP5, all the free zip things like: Cups, Java, ODIN, QT 473, Samba and so on, a very few parts of eCS like ACPI, AHCI, and Panorama; some more free drivers USB (183),Stylus,WINOS2,; some more programs Injoy+InjoyFirewall,HPFS386, Air Boot Manager 1.08,Freetype, Warpsans, SMPlayer, PM123 and and and) and the Image is now at 1.5GB. This image can be copied to a Partition that could range from 4 to 64 GB as DFSEE is able to "Sync the file to the partition size", and 64GB is the limit for HPFS.

And - very important to me - I do not have one file from ecomstation.ru running, no ecoruntime or usbwidget or or.. - and that makes the installations to become much more stable, especially compared with eCS 2.2 beta. I did not include YUM/RPM as well es it was at least not necessary till now, I was able to install everything with the ZIP files.

In addition I added all the files IBM or Mensys cut off - like Warpguides, OS/2 Warp Tutorial and so on. I extracted the OS2 Warp Logo from the Warp Center and integrated it in xcenter of xworkplace.

At least I would say "my" Warp Clone is up to date with the software and hardwaresupport. And this Image/Clone file is working great in combination with DFSEE. I reinstalled it serveral times.

In some case I would like to say I followed the Path Doug Bisset describes for Beta testing of ecomstation 2.2. I added something new to the clone, if it works I build a new image, if not I just replaced it with the image before and started again. As it takes only a few minutes to image/restore I was able to build the clone in my free time starting from an OS/2 Warp 4 scratch to what I have now in almost three weeks.

And here are the things I would like to fix at least and ask for your help - Thank you very much in advance!

- PEN for OS/2 - does someone know how to change the keyboard layout from US to German? The keyboard is really great, no need for me to use the "Virtualkeyboard" of ecs.ru

- ACPI - I have to use the /VW switch on the Samsung Slate, otherwise the boot process stops with "Can not load OS2AHCI.ADD because IRQ 19 allready claimed" -> so obvious for me tha the handling of hig IRQ is not working as it should as eCS 2.1 with ACPI 3.21.07 does not have this problem (but some more I do not have with Warp 4). I only have the ACPI32107.wpi file installed, no more ACPI related files. Do I need something else to handle the high IRQ in the right way?

- Atheros WLAN - I still try to figure out to run the Atheros WLAN 9285 chipset. I already opened the Slate and could change the MiniPCIE Option GM67 card, and put an Intel inside, but I would loose Bluetooth or 3G,a s there is no Intel chipset Mini PCIE card around with Bluetooth, Wlan and 3G - I do use and need all three for windows. And I am not sure if the WLAN Multimac for Intel will see the light of the day (hopefully it will). Someone posted to me a mail with some hints for this card (I already got the driver loading in UNSUPPORTED mode but the boot process stops - I guess because of netbind.exe or an IRQ conflict) - but unfortunately I lost this mail. May be thsi person is reading here and can send it to me once again?

- Sound - is working with latest UNIAUD with Warp 4 much better then with eCS 2.2 (do not know why exactly, but I have less boots without sound than before) - and strange - I can adjust loudness with the usual thing, the uniaudmixer and with SMPlayer -b ut not with PM123  :o . I am using a startscript in startup.cmd to enable sound at all.

- Memory management - the Installation "only" sees 512MB of the internal 4GB of RAM - according to the xworkplace program "OS/2 kernel" - as it is an Intel Core i5 second generation cpu the same problems occur with eCS 2.2 beta. Is there a program that can show me the amount of space OS/2 "really" uses or sees?

Thanks in Advance and Bye!

Sigurd

« Last Edit: April 05, 2013, 12:07:33 pm by Sigurd Fastenrath »



Eugene Gorbunoff

  • Full Member
  • ***
  • Thank You
  • -Given: 3
  • -Receive: 3
  • Posts: 104
    • View Profile
Re: Final Request for Help for the Final Warp 4 Image Clone
« Reply #1 on: April 05, 2013, 12:35:27 pm »
Hello eComStation users

The user continues make damage to eComStation (and eComStation.RU)

Here is the description of the conflict:
http://ecomstation.ru/temp/201301/201301-attack.html

Sigurd Fastenrath

  • Sr. Member
  • ****
  • Thank You
  • -Given: 13
  • -Receive: 35
  • Posts: 337
    • View Profile
Re: Final Request for Help for the Final Warp 4 Image Clone
« Reply #2 on: April 05, 2013, 12:44:48 pm »
Hello eComStation users

The user continues make damage to eComStation (and eComStation.RU)

Here is the description of the conflict:
http://ecomstation.ru/temp/201301/201301-attack.html

Hihi,

the truth is sometimes hard to stand, even in your world. For more information just wait for the next - and final video at youtube  8)

 ;) :'(
« Last Edit: April 05, 2013, 12:48:40 pm by Sigurd Fastenrath »

walking_x

  • Jr. Member
  • **
  • Thank You
  • -Given: 3
  • -Receive: 6
  • Posts: 93
    • View Profile
Re: Final Request for Help for the Final Warp 4 Image Clone
« Reply #3 on: April 05, 2013, 03:20:52 pm »
Using of +512Mb memory on modern chipsets requires a large patching of OS2LDR - or alternative loader ;).
Now you can get this memory only with non-IBM OS2LDR loader and /VW key in new ACPIs (or with ACPIs 3.1x).
The problem is simple - IBM (eCS) OS2LDR gets only 1st block of memory from system, but EFI BIOS split memory on 2-3 huge blocks...

Roderick Klein

  • Full Member
  • ***
  • Thank You
  • -Given: 0
  • -Receive: 8
  • Posts: 179
    • View Profile
Re: Final Request for Help for the Final Warp 4 Image Clone
« Reply #4 on: April 05, 2013, 04:10:50 pm »
Using of +512Mb memory on modern chipsets requires a large patching of OS2LDR - or alternative loader ;).
Now you can get this memory only with non-IBM OS2LDR loader and /VW key in new ACPIs (or with ACPIs 3.1x).
The problem is simple - IBM (eCS) OS2LDR gets only 1st block of memory from system, but EFI BIOS split memory on 2-3 huge blocks...

I have somebody who opened a ticket. He had the same problem. But when a installed a new video card the memory hole went away!
It seems it is ot related to EFI but rather how some onboard video chipsets map video memory.
There is an eCS 2.x ticket in the eCS bugtracker that describes.

Because if this would be an EFI issue and you insert a different video card and the memory hole then goes away!?

Roderick Klein
Mensys B.V.

walking_x

  • Jr. Member
  • **
  • Thank You
  • -Given: 3
  • -Receive: 6
  • Posts: 93
    • View Profile
Re: Final Request for Help for the Final Warp 4 Image Clone
« Reply #5 on: April 05, 2013, 04:36:02 pm »
This problem has ages, actually :) But now it occurs much often.
Some years ago I saw report about old HP notebook with 64 + 96 Mb onboard with a small hole between of them... And the user was very grateful to OS/4 loader ;)

Doug Bissett

  • Hero Member
  • *****
  • Thank You
  • -Given: 1
  • -Receive: 38
  • Posts: 938
    • View Profile
Re: Final Request for Help for the Final Warp 4 Image Clone
« Reply #6 on: April 05, 2013, 06:19:01 pm »
Quote
- ACPI - I have to use the /VW switch on the Samsung Slate, otherwise the boot process stops with "Can not load OS2AHCI.ADD because IRQ 19 allready claimed" -> so obvious for me tha the handling of hig IRQ is not working as it should as eCS 2.1 with ACPI 3.21.07 does not have this problem (but some more I do not have with Warp 4). I only have the ACPI32107.wpi file installed, no more ACPI related files. Do I need something else to handle the high IRQ in the right way?

ACPI32107 seems to work very well. Your problem is probably one of the system drivers causing problems, or it may be something in the BIOS that is claiming IRQ 19 before ACPI gets started. Either you aren't starting ACPI properly (at the right time), or you need to figure out what is holding IRQ 19, and do something about it. On the other hand, is there really any reason to not use /VW?

Quote
- Atheros WLAN - I still try to figure out to run the Atheros WLAN 9285 chipset. I already opened the Slate and could change the MiniPCIE Option GM67 card, and put an Intel inside, but I would loose Bluetooth or 3G,a s there is no Intel chipset Mini PCIE card around with Bluetooth, Wlan and 3G - I do use and need all three for windows. And I am not sure if the WLAN Multimac for Intel will see the light of the day (hopefully it will). Someone posted to me a mail with some hints for this card (I already got the driver loading in UNSUPPORTED mode but the boot process stops - I guess because of netbind.exe or an IRQ conflict) - but unfortunately I lost this mail. May be thsi person is reading here and can send it to me once again?

If you want to use an UNSUPPORTED wireless LAN device (any of them), you need to patch \IBMCOM\macs\GENM32W.OS2. Apparently, GENMAC does not enable the piece of code that XWLAN needs to interface with the windows driver, if you use UNSUPPORTED mode. See the GENMU_TEST_README.TXT file at ftp://genmac@ftp.os2voice.org/ (use the word "unsupported", lower case, and no quotes, as a password if asked), for more information. Also check that for the presence of an already prepared driver. I don't know what the PCI ID for that device is, and it may already be included. If not, supply the parts, and I will add it to the TEST version. The instructions are in the README. I can't guarantee that it will actually work, but it should get installed properly that way (including the required patch).

Quote
- Memory management - the Installation "only" sees 512MB of the internal 4GB of RAM - according to the xworkplace program "OS/2 kernel" - as it is an Intel Core i5 second generation cpu the same problems occur with eCS 2.2 beta. Is there a program that can show me the amount of space OS/2 "really" uses or sees?

It seems that all UEFI machines are going to have this problem, to some extent. Somebody decided to allow more than one memory hole, and OS2LDR can't handle that. It also seems that 512 meg seems to be a popular spot to put the second hole. That means that OS/2 (eCS) can't assign memory above that address. OS2LDR needs to be fixed, somehow. From what I can tell, OS/2 "sees" 512 meg, and no more. Since OS/2 can actually operate reasonably well with that amount of memory, you will probably find that the machine will work okay, and hope they can fix the problem, eventually. Changing video cards is not a viable work around for most machines (especially laptops). I expect that what Roderick describes simply moved the memory hole to another spot. In my case, on my Lenovo ThinkPad L530, I think it is one of the NICs that sets up shop at 512 meg, not the video card.

Quote
The user continues make damage to eComStation (and eComStation.RU)

eComStation.RU does their own damage. They don't need any help from Sigurd.

Quote
The problem is simple - IBM (eCS) OS2LDR gets only 1st block of memory from system, but EFI BIOS split memory on 2-3 huge blocks...

Apparently, it is not always "huge" blocks, and it can be more than 2-3. It depends on the specific system, and what I/O they have installed.

Quote
Because if this would be an EFI issue and you insert a different video card and the memory hole then goes away!?

It is definitely a UEFI problem. The UEFI specs allow for I/O devices (not just video cards) to put holes in memory, and it seems that those holes can be put anywhere they please. Logically, they should be at the end of memory, but that is not in the specs. It also seems that Microsoft defined the specs, and saw an opportunity to mess up other operating systems, again. Linux has, apparently, found a work around, but it is not that easy in OS/2 since they do not have the source for OS2LDR (or, if they do, it is not common knowledge).

I have yet to see any way to "insert a different video card" in a laptop. It can also be difficult in a desktop, sometimes. Not to mention it is an extra expense that many people aren't willing to pay (eCS is not cheap to start with). The problem needs to be fixed, soon, because UEFI seems to be the way that all new hardware is going.

Roderick Klein

  • Full Member
  • ***
  • Thank You
  • -Given: 0
  • -Receive: 8
  • Posts: 179
    • View Profile
Re: Final Request for Help for the Final Warp 4 Image Clone
« Reply #7 on: April 05, 2013, 06:57:31 pm »

It is definitely a UEFI problem. The UEFI specs allow for I/O devices (not just video cards) to put holes in memory, and it seems that those holes can be put anywhere they please. Logically, they should be at the end of memory, but that is not in the specs. It also seems that Microsoft defined the specs, and saw an opportunity to mess up other operating systems, again. Linux has, apparently, found a work around, but it is not that easy in OS/2 since they do not have the source for OS2LDR (or, if they do, it is not common knowledge).

I have yet to see any way to "insert a different video card" in a laptop. It can also be difficult in a desktop, sometimes. Not to mention it is an extra expense that many people aren't willing to pay (eCS is not cheap to start with). The problem needs to be fixed, soon, because UEFI seems to be the way that all new hardware is going.

Well I don't think this is specific for the UEFI standard. Remember the old ISA cards video boards. They also had a memory hole!
So I still don't see why this is specificly related to the UEFI standard.

And I know sticking in a video card in a laptop is not possible.

And certainly network cards do not require a memory hole...

Roderick

walking_x

  • Jr. Member
  • **
  • Thank You
  • -Given: 3
  • -Receive: 6
  • Posts: 93
    • View Profile
Re: Final Request for Help for the Final Warp 4 Image Clone
« Reply #8 on: April 05, 2013, 07:23:24 pm »
Apparently, it is not always "huge" blocks, and it can be more than 2-3. It depends on the specific system, and what I/O they have installed.
Yes, but there is no reason to use tiny blocks... Whole list can hold up to 20 entries and most of them - small in size. OS/4 loader just ignore blocks smaller 64k, as I know.
And - OS2LDR still can be rewritten from scratch ;) This is possible if you know how organized kernel loading process.

Sigurd Fastenrath

  • Sr. Member
  • ****
  • Thank You
  • -Given: 13
  • -Receive: 35
  • Posts: 337
    • View Profile
Re: Final Request for Help for the Final Warp 4 Image Clone
« Reply #9 on: April 05, 2013, 07:41:12 pm »
First I would like to thank for answering and , to avoid double messages, I would like to point to the similar thread at the yahoo group, for those not already monotoring both forums:

http://tech.groups.yahoo.com/group/eComStation/message/88015



If you want to use an UNSUPPORTED wireless LAN device (any of them), you need to patch \IBMCOM\macs\GENM32W.OS2. Apparently, GENMAC does not enable the piece of code that XWLAN needs to interface with the windows driver, if you use UNSUPPORTED mode. See the GENMU_TEST_README.TXT file at ftp://genmac@ftp.os2voice.org/ (use the word "unsupported", lower case, and no quotes, as a password if asked), for more information. Also check that for the presence of an already prepared driver. I don't know what the PCI ID for that device is, and it may already be included. If not, supply the parts, and I will add it to the TEST version. The instructions are in the README. I can't guarantee that it will actually work, but it should get installed properly that way (including the required patch).

Yes, I already used your genmac files from the genmac list  ;). And I already have the XP drivers for the Atheros and I changed the protocol.ini and some more. I will come back to this once Finger support is finished. Wim just handed a one finger driver and this is working great!! He may have the chance not only to support two but also FIVE finger support ;D But we have to see and all credits to Wim!

By the way, for me one can not top using Warp 4 (or ecomstation if you like) with this tablet and finger + stylus support.....

 8)
« Last Edit: April 05, 2013, 07:43:30 pm by Sigurd Fastenrath »

Sigurd Fastenrath

  • Sr. Member
  • ****
  • Thank You
  • -Given: 13
  • -Receive: 35
  • Posts: 337
    • View Profile
Re: Final Request for Help for the Final Warp 4 Image Clone
« Reply #10 on: April 06, 2013, 01:39:50 pm »
Apparently, it is not always "huge" blocks, and it can be more than 2-3. It depends on the specific system, and what I/O they have installed.
Yes, but there is no reason to use tiny blocks... Whole list can hold up to 20 entries and most of them - small in size. OS/4 loader just ignore blocks smaller 64k, as I know.
And - OS2LDR still can be rewritten from scratch ;) This is possible if you know how organized kernel loading process.

Hm, I just replaced the eCS 2.2 OS2LDR with the one from the OS4 project

(NO other files replaced, kernel or whatever, version is from the "os2krnlSVN2814_unoff")

 and guess what -

the whole RAM is there (despite the one that is used by the Graphic card)!

No additional problems occur while using this one so far - but testing still pending.
The file size of this OS2LDR is about 49kb compared with the 43kb from eCS.

I wonder - Is there a way to merge the experience of OS4 with the one from ecomstation and David Azariewicz as this seems to may solve more actual eCS problems?!


Andreas Schnellbacher

  • Sr. Member
  • ****
  • Thank You
  • -Given: 1
  • -Receive: 15
  • Posts: 259
    • View Profile
Re: Final Request for Help for the Final Warp 4 Image Clone
« Reply #11 on: April 06, 2013, 02:13:25 pm »
Hm, I just replaced the eCS 2.2 OS2LDR with the one from the OS4 project

(NO other files replaced, kernel or whatever, version is from the "os2krnlSVN2814_unoff")

 and guess what -

the whole RAM is there (despite the one that is used by the Graphic card)!

That's what walking_x said. Thanks for the info. I understand now
finally why they wrote a new loader.

I doubt that will not effect the eCS development due to the license
agreement with IBM. I still wonder what is possible and what isn't:

o  Replacing the kernel is not allowed, but patching it in memory is.

o  Replacing the IBM installation is not allowed, but creating heavy
   frameworks around it is.
   
o  Replacing the loader: ?

Andreas

Andreas Schnellbacher

  • Sr. Member
  • ****
  • Thank You
  • -Given: 1
  • -Receive: 15
  • Posts: 259
    • View Profile
http://www.os2world.com/forum/index.php?action=dlattach;topic=101.0;attach=92

Sigurd, I would change VIRTUALADDRESSLIMIT to 1536. Otherwise you'll
run out of shared memory too soon. That depends on the apps you use.

Andreas

guzzi

  • Sr. Member
  • ****
  • Thank You
  • -Given: 0
  • -Receive: 13
  • Posts: 263
    • View Profile
Re: Final Request for Help for the Final Warp 4 Image Clone
« Reply #13 on: April 06, 2013, 10:53:44 pm »
I am no expert on copyright law. I think however that rewriting the loader needs reverse engineering and that would make it a copyright infringement. They put a lot of work in the os/4 loader and kernel and it's basically been a waste of effort because of legal issues. A shame really.

Martin Iturbide

  • OS2World NewsMaster
  • Global Moderator
  • Hero Member
  • *****
  • Thank You
  • -Given: 255
  • -Receive: 49
  • Posts: 1708
  • Your Friend Wil Declares...
    • View Profile
    • Martin's Personal Blog
Re: Final Request for Help for the Final Warp 4 Image Clone
« Reply #14 on: April 07, 2013, 02:10:30 am »
Hi Guzzi.

I'm not expert also in copyright law, but I saw there are some examples that are working outside our community. For example Linux (GNU Linux) is some kind of reverse engineering of Unix. It does not mean that they grab Unix binaries and make a inverse-compile process which will be a copyright infringement. ReactOS is a clone of Windows (32bits) and it is not marked as an "illegal project".

Studying how OS2Loader works and create a clone of it, without using any kind of leaked IBM source code, is legal according to my knowledge. For example, I don't consider illegal this: http://www.edm2.com/0705/freeldr/freeldr.html

I think that replacing the kernel is allowed, only if the kernel is a 100% original work, without any leaked source code from IBM. If you hack the binaries and share them, or use a leaked IBM source code you will doing a copyright infringement. Again, studying the OS/2 Kernel and creating an 100% original clone is not a copyright infringement.

OS/4 Loader
About OS/4 Loader.  I really don't have the knowledge to know if it is a 100% original work, and I will love if someone outside the OS/4 team can check the source code can confirm this, but I consider it an interesting project and I'm happy that someone coded it.

But the license of use is completely against open source software and I'm against it.  "Source code can be available by request, for non-commercial use only." In other words, you don't know which are the rules to keep developing that loader.

I will like to recommend to the OS/4 team (my personal opinion) to use an open source license for their development and expand their project reach. Any open source license they like or choose , but a formal open source license for their project (http://opensource.org/licenses). With that there will be clear rules how to use their source code.

Open Source is not evil. Open Source is not capitalist (and not communist). Open source is not focus on helping big companies to get money. Open Source is a way to license software, and if you choose a strict (copyleft) license like GNU GPL you will guarantee that any improvement to the source code will be released back to the public, no mater if it is an individual or big company.

Regards
« Last Edit: April 07, 2013, 02:21:31 am by Martin Iturbide »
Martin Iturbide
OS2World NewsMaster
... just share the dream.