Author Topic: eCS install problems on a recent mainboard  (Read 21283 times)

mcs_5

  • Guest
eCS install problems on a recent mainboard
« on: August 24, 2016, 03:28:32 am »
I have just attempted, unsuccessfully, to install eCS on a system based on an Asrock mini-itx mainboard. The board is a Q1900-ITX with a 4-core Celeron. The disk is a Samsung SSD connected to a SATA port controlled by the Intel chipset.

If I select the ACPI option in the boot menu, the system boots to the installer without complaints. The disk is detected with either the AHCI driver or the Dani driver, depending on the mode I select in the UEFI setup. But if I use AHCI, I'm unable to format partitions in HPFS format. If I try in a command line, I get the error that not enough memory is available to run format Formatting with JFS works. If I use ATA-mode, both HPFS and JFS formatting work.

But after the first install phase has completed, the disk in not bootable. Or not detected as bootable by the UEFI firmware?

I have tried with both eCS 2.1 and 2.2 beta, and the same thing happens. No error messages, but the system won't boot from the disk and complete the install. Any advice?

rdunkle

  • Guest
Re: eCS install problems on a recent mainboard
« Reply #1 on: August 24, 2016, 10:14:24 am »
Perhaps if you enter the BIOS setup program and enable the CSM (Compatibility Support Module).  It is under the Boot icon in the BIOS setup. Then start the install again.

mcs_5

  • Guest
Re: eCS install problems on a recent mainboard
« Reply #2 on: August 24, 2016, 12:51:44 pm »
CSM is enabled - it was enabled in the default settings.

And I guess CSM would have to be enabled, for the eCS boot CD to work?

Martin Iturbide

  • OS2World NewsMaster
  • Global Moderator
  • Hero Member
  • *****
  • Posts: 4713
  • Karma: +41/-1
  • Your Friend Wil Declares...
    • View Profile
    • Martin's Personal Blog
Re: eCS install problems on a recent mainboard
« Reply #3 on: August 24, 2016, 03:01:30 pm »
Hi Mikkel, welcome to the forums.

eCS cd boots on your  Asrock Q1900-ITX, which seems a positive thing.

It is very strange that if you enable AHCI (on the BIOS and on the eCS boot menu) it does not want to format the drive.

After the first install phase you don't get anything on OS/2-eCS? The little square on top? You can not do an Alt+F1? or it shows the common message like there is no OS installed on the HDD and the BIOS  tries to boot from other source?

One wild guess would be to double check your install media. Sometimes I got very strange behavior from corrupted CDs.

My other guess will be to try if there any things on the UEFI secure boot on the BIOS and try to turn it off.

Maybe someone with more experience can share more thoughts.
Regards
Martin Iturbide
OS2World NewsMaster
... just share the dream.

mcs_5

  • Guest
Re: eCS install problems on a recent mainboard
« Reply #4 on: August 24, 2016, 05:32:35 pm »
I did some more testing. I found an old HP system in the junk pile at work (Core2 Duo CPU). I installed the Samsung disk, and it was still unbootable. If I force the system to boot from disk, all I get is a flashing cursor in the upper left corner.

I then tried re-installing eCS on the HP system, and I got the same thing. First phase of the install completes, but the disk remains unbootable.

Next step was to wipe the first MB of the disk, and create partitions using DFSee. DFSee did warn that the disk was not bootable, before I re-partitioned it. I then ran the installation on the HP system again, and now the install completes.

So whatever went wrong during the first install (something in the MBR I guess), was not fixed by subsequent installs.

But now it runs on the HP system, so I just have to try moving the disk back to the mini-ITX system...

mcs_5

  • Guest
Re: eCS install problems on a recent mainboard
« Reply #5 on: August 24, 2016, 06:48:12 pm »
Another update. I installed the latest ACPI, AHCI and USB drivers from Arca Noae while the disk was in the HP system. I then moved it to the system with the Asrock board. And it boots!

All four CPU cores are detected, and after installing the Realtek NIC driver, that seems to be working also.

But there is a weird problem I have never seen before. If I try to open a command line (windowed or fullscreen) it exits immediately. What could cause this?

If you try to open a DOS window, you get complaints about missing video support, but I have read about that somewhere before. That is not a huge problem, but having no command line access is ;)

mcs_5

  • Guest
Re: eCS install problems on a recent mainboard
« Reply #6 on: August 24, 2016, 10:53:54 pm »
It's not looking good...

If I enable AHCI mode, the system hangs for a long time after printing the line "Unicode translate table for CP 850 loaded."  If I leave the system alone at that point, I get a trap screen after some minutes.

In IDE/ATA mode the system works better, but very slowly. Danis506 runs the disk in PIO4 mode (because the controller is not recognized?), which results in the stellar read/write speed of 2MB/s according to SysBench...

I thought the NIC was working, but the driver is just installed without error messages. But if I try to transfer data (ping etc.) no packets actually get through. I have tried the MMRE, R8169 and R8110 drivers. They all install, but none of them actually work. According to PCI.EXE, the NIC is a 10EC:8168.

No USB ports work. The drivers report that no USB controllers are found during boot. According to the docs the board should include both USB2 and USB3 ports, but the USB2 ports are not discovered .

But some change I made (removing the command line options of ACPI.PSD? changing NIC drivers?), made the command line work again.

I guess I should look for a board that works better ;)

But are there any relevant tests I should perform, to get some feedback for the driver developers?

Lars

  • Hero Member
  • *****
  • Posts: 1271
  • Karma: +65/-0
    • View Profile
Re: eCS install problems on a recent mainboard
« Reply #7 on: August 25, 2016, 06:40:57 pm »
For AHCI:
1) make sure it's set to AHCI in BIOS
2) use OS2AHCI.ADD (of course)
3) try the /!R for OS2AHCI.ADD. I already had to specify this switch on installation. It prevents reset of the SATA ports on system start.

For OS2AHCI.ADD you can also try /F and /N switches. These switches should improve performance (enable command queuing, force usage of disk HW cache).

Lars

Roderick Klein

  • Hero Member
  • *****
  • Posts: 655
  • Karma: +14/-0
    • View Profile
Re: eCS install problems on a recent mainboard
« Reply #8 on: August 27, 2016, 10:49:33 pm »
It's not looking good...

If I enable AHCI mode, the system hangs for a long time after printing the line "Unicode translate table for CP 850 loaded."  If I leave the system alone at that point, I get a trap screen after some minutes.

In IDE/ATA mode the system works better, but very slowly. Danis506 runs the disk in PIO4 mode (because the controller is not recognized?), which results in the stellar read/write speed of 2MB/s according to SysBench...

I thought the NIC was working, but the driver is just installed without error messages. But if I try to transfer data (ping etc.) no packets actually get through. I have tried the MMRE, R8169 and R8110 drivers. They all install, but none of them actually work. According to PCI.EXE, the NIC is a 10EC:8168.

No USB ports work. The drivers report that no USB controllers are found during boot. According to the docs the board should include both USB2 and USB3 ports, but the USB2 ports are not discovered .

But some change I made (removing the command line options of ACPI.PSD? changing NIC drivers?), made the command line work again.

I guess I should look for a board that works better ;)

But are there any relevant tests I should perform, to get some feedback for the driver developers?

It's not looking good...

If I enable AHCI mode, the system hangs for a long time after printing the line "Unicode translate table for CP 850 loaded."  If I leave the system alone at that point, I get a trap screen after some minutes.

In IDE/ATA mode the system works better, but very slowly. Danis506 runs the disk in PIO4 mode (because the controller is not recognized?), which results in the stellar read/write speed of 2MB/s according to SysBench...

I thought the NIC was working, but the driver is just installed without error messages. But if I try to transfer data (ping etc.) no packets actually get through. I have tried the MMRE, R8169 and R8110 drivers. They all install, but none of them actually work. According to PCI.EXE, the NIC is a 10EC:8168.

No USB ports work. The drivers report that no USB controllers are found during boot. According to the docs the board should include both USB2 and USB3 ports, but the USB2 ports are not discovered .

But some change I made (removing the command line options of ACPI.PSD? changing NIC drivers?), made the command line work again.

I guess I should look for a board that works better ;)

But are there any relevant tests I should perform, to get some feedback for the driver developers?

The first question that comes to mind ? Post *what* the trap screen mentioned. Did it it mention what driver (module the kernel posted when you had the OS2ACHI driver installed. Since netbind.exe has already loaded (I think) it should have already written some info to disc OS/2. Do you have earlymeminit=true in your config.sys.
If not your kernel memory to load drivers might get exhausted to fast.

Did a DOS command line not work or an OS/2 command line. You could try loading ACPI.PSD with the option /MAXCPU=1 and see if that fixes the problem. Infact removing the PSD and loading the system in PIC mode might introduce some other issue's if this computer only operated correctly running in APIC mode.

Just some off my thoughts.

Roderick Klein

Andreas Schnellbacher

  • Hero Member
  • *****
  • Posts: 827
  • Karma: +14/-0
    • View Profile
Re: eCS install problems on a recent mainboard
« Reply #9 on: August 28, 2016, 01:06:50 am »
Additionally to that what Roderick wrote, you could test the parameter /VW (virtual wire) for ACPI. That is required for quite some mainboards to work.

mcs_5

  • Guest
Re: eCS install problems on a recent mainboard
« Reply #10 on: August 30, 2016, 05:38:23 pm »
The first question that comes to mind ? Post *what* the trap screen mentioned. Did it it mention what driver (module the kernel posted when you had the OS2ACHI driver installed.

It's not one of the "usual" trap screens I have seen many times. It does not mention a driver. Here's what's printed:

Code: [Select]
The system detected an internal processing error at
location ##1800:1b4e - 0003:1b4e
60000, 2008

10860673
Internal revision 14.105_SMP


Quote
Since netbind.exe has already loaded (I think) it should have already written some info to disc OS/2.

Where should I look for that info?


Quote
Do you have earlymeminit=true in your config.sys.
If not your kernel memory to load drivers might get exhausted to fast.

Yes. I have not changed the default settings.


Quote
Did a DOS command line not work or an OS/2 command line. You could try loading ACPI.PSD with the option /MAXCPU=1 and see if that fixes the problem. Infact removing the PSD and loading the system in PIC mode might introduce some other issue's if this computer only operated correctly running in APIC mode.

It was the OS/2 command line that did not work at all. But some change I made fixed it. I will try some of the other suggestions here also.

Mikkel

mcs_5

  • Guest
Re: eCS install problems on a recent mainboard
« Reply #11 on: August 30, 2016, 10:34:25 pm »
I have tried some different AHCI/ACPI combinations now.

/VW on ACPI.PSD works in ATA mode. In AHCI mode the system still hangs, with the same trap screen.

In PIC mode, both ATA and AHCI mode works - but without SMP of course.

I got performance in ATA mode improved also. I noticed that the driver does not recognize the controller, but If I add /GBM to the DANIS506.ADD line, I get much improved performance. Throughput increases from about 2MB/s to about 150MB/s.

I ran some SysBench tests. PM, DIVE and memory benchmarks stay about the same in PIC, VW and default ACPI mode. But the CPU tests get a remarkable boost in ACPI/SMP mode. In PIC mode I get 1738/133 (integer/float) and in ACPI/SMP mode 6791/466! Does SysBench run the tests multithreaded, or does ACPI do some "magic" tuning?

But without USB and networking, the system is still not entirely useful ;)

Neil Waldhauer

  • Hero Member
  • *****
  • Posts: 1024
  • Karma: +24/-0
    • View Profile
    • Blonde Guy
Re: eCS install problems on a recent mainboard
« Reply #12 on: September 02, 2016, 03:44:03 pm »
Ouch. That is pretty slow for a modern computer.

Yes, Sysbench runs CPU test multi-threaded. Most OS/2 usage is single threaded, so the higher speed for four cores can be pretty misleading.

My three generations back (but still available new) computer is running more than twice the speed for i7 and nearly twice for dual core celeron. By the way, Celeron all the way for OS/2. i7 gave very little additional performance for a lot of additional money.
Expert consulting for ArcaOS, OS/2 and eComStation
http://www.blondeguy.com

OS4User

  • Sr. Member
  • ****
  • Posts: 406
  • Karma: +10/-0
    • View Profile
Re: eCS install problems on a recent mainboard
« Reply #13 on: September 02, 2016, 06:05:04 pm »
... i7 gave very little additional performance...

Try to compare DIVE -> Video bus bandwidth on Celeron and on i7.
This mark is most important for screen update speed in Panorama.

Roderick Klein

  • Hero Member
  • *****
  • Posts: 655
  • Karma: +14/-0
    • View Profile
Re: eCS install problems on a recent mainboard
« Reply #14 on: September 02, 2016, 09:35:25 pm »
The first question that comes to mind ? Post *what* the trap screen mentioned. Did it it mention what driver (module the kernel posted when you had the OS2ACHI driver installed.

It's not one of the "usual" trap screens I have seen many times. It does not mention a driver. Here's what's printed:

Code: [Select]
The system detected an internal processing error at
location ##1800:1b4e - 0003:1b4e
60000, 2008

10860673
Internal revision 14.105_SMP


Quote
Since netbind.exe has already loaded (I think) it should have already written some info to disc OS/2.

Where should I look for that info?


Quote
Do you have earlymeminit=true in your config.sys.
If not your kernel memory to load drivers might get exhausted to fast.

Yes. I have not changed the default settings.


Quote
Did a DOS command line not work or an OS/2 command line. You could try loading ACPI.PSD with the option /MAXCPU=1 and see if that fixes the problem. Infact removing the PSD and loading the system in PIC mode might introduce some other issue's if this computer only operated correctly running in APIC mode.

It was the OS/2 command line that did not work at all. But some change I made fixed it. I will try some of the other suggestions here also.

Mikkel

Oke with such a TRAP I guess you need open a support ticket at Arca Noae. While  a module name in a TRAP is not always a reliable method, but you might get a hint which module potentially caused the trap.

Which category the ticket would need to be opened at AN I do not. But this trap occurs when you boot with OS2AHCI driver ? Open a ticket then for that, that is if you want to see if this defect can be fixed.

I guess the TRAP is beyond the skillset of the members in this forum. In fact I would not be surprised if a memory dump MIGHT be needed to find out why this TRAP occurs.

Roderick Klein