ACPI, Gigabyte GA-K8NS Ultra-939

[David McKenna]

I am running eCS 2.0 RC6a with ACPI 3.15 and use the kernel supplied with ACPI 3.15. I have installed Fixpak 5. Hardware is a Gigabyte GA-K8NS Ultra-939 motherboard (nForce3 Ultra chipset), 3GB RAM, Radeon 9600 Video 256MB, one 150GB SATA drive (one partition - no Boot Manager) with Danis 1.8.5, one SATA DVD-RW. I use UniAud 1.9.12 to drive the on-board ALC850 sound, and GenMac to drive the on-board Marvell 8001 gigabit LAN (see attached logs).

Using ACPI.PSD /SMP /APIC /TMR, system does not boot ~1 out of 3 tries. It will hang when the video tries to switch screen resolution for the desktop. The monitor loses sync and the 'no signal' message appears on screen. I need to hard reset the computer. I have not found any better switches for ACPI.PSD.

When it boots successfully it seems to work well (that is when I got the logs..:-).

ACPI 3.14 would not boot at all - it would Trap 0008 when trying to load UniAud (not heavily tested). I have not tried any earlier versions on this hardware.

[David McKenna]

Forgot to mention... Athlon 64 4000+ CPU....

[pasha]

When is running AcpiDaemon.exe? From config.sys or from startup.cmd?

[David McKenna]

ACPIDaemon is started detached from an object in the startup folder.

[David McKenna]

The situation is more complex than I originally reported.

This system can go for a few days and boot the first time, then it will take 5 or 6 attempts to get started. It will hang in one of three places:

1. At start of WPS (as originally reported)

2. At Boot logo (on any of the BASEDEV= drivers, but usually USBD.SYS, IBM1FLPY.ADD, or DANIS506.ADD).

3. Trap 8 (loading UNIAUD, or at end of loading all drivers)

I have tried all the interim releases of ACPI and there seems to be no difference. I have also tried every combination of switches I could think of, including every possible /DV:vector setting to no avail. Changing BIOS settings also does not seem to help.

FWIW I kept a log of boot attempts. Out of 132 boot attempts:

15 ended in Trap 8 (11%)

31 ended hung at the logo (23%)

40 ended hung at the WPS (30%)

HTH

[pasha]

Please download experimental ACPI build from Mensys site:

• Experimental build for you:

ACPI-06-16.ZIP

• Read instructions: ​http://ecomstation.ru/projects/acpitools/?action=testcase

Remove pls /OS="Microsoft Windows NT" from command line

[David McKenna]

Unfortunately, this new version does not change anything (only the APM.ADD and ACPIDAEMON.EXE files seems to be different). I will upload acpi.log.

[David McKenna]

Version 3.16 Full works the same as 3.15, except now shutdown does not work (with or without the /R switch).

[David McKenna]

acpica$ for 3.16 full

[David McKenna]

This may just be coincidence....

3.16 Full contains the files 'table.dat' and 'table.dsl'. It looks like they are SSDT files (which my BIOS does not have). I put 'table.dat' in the \OS2\BOOT folder and added the line 'FILE TABLE.DAT' to my ACPI.CFG file.

After 44 boots, I have had 2 hangs at the Logo and 2 Trap8 loading UNIAud. No hangs at the WPS. This seems like a huge improvement! Is this the intended use of 'table.dat'?

I still cannot power off properly - it hangs at a blue screen when normally the disk flushes. Reset boots with a check disk every time. This worked fine with 3.15.

Strangely, ACPI shutdown is 'greyed-out' on the XShutdown page in the desktop properties folder. Only APM shutdown is available. I tried re-installing XWP, but the result is the same.

[pasha]

acpi.log with "table.dat" ?

[David McKenna]

ACPICA$ with 3.16 full and table.dat

[David McKenna]

FWIW - After running this way a couple of days it seems the chance of a boot hang goes up the longer the computer has been turned off.

[pasha]

acpi.psd don't read your table. It is under construction for SMP. Try latest uniaud pls. Or you must change LACI to another IRQ for boot. For example, add to acpi.cfg LINK LACI 5

[David McKenna]

OK, I removed TABLE.DAT. I am using UNIAud 1.9.16 - I believe that is the latest. I added LINK LACI 5 to ACPI.CFG and will run with this for awhile to see if there is any improvement. One thing I notice from acpica$ - the LINK appears to work under 'IRQ links found':

17:1:9 Set: 5 5: 'LACI' (F92D8588) P[ 3, 4,*5, 6, 7, 9, 10, 11, 12, 14, 15] <- 0:6.0

but not here:

17:1:9 0:6.0 10DE:EA IRQs: PCI:9 PIC:'LACI':9 APIC:'APCJ':20 Set: 20

This happens to be the sound card - maybe this is the root of the problems with this hardware? I'll upload acpica$

[David McKenna]

acpica$ with LINK LACI 5

[pasha]

yes, I need time for find this problem in code.

[David McKenna]

No improvement using LINK LACI 5. Also no improvement with the new (7-23) version of ACPI. acpica$ has some new information in it, so I will upload....

[David McKenna]

acpica$ from 7-23 version of ACPI

[David McKenna]

No change with 08-09 build of ver 3.16 ACPI....

[David McKenna]

No change with version 3.17 ACPI....

[pasha]

Sorry, as I understand from log, you have comport debugging? Say to me day, when we can set and resolve your problem. Be nice if we will connect online, irc or icq. My icq is 32400130. Resolve process will be - I send acpi.psd - you back log.

[David McKenna]

OK... I could spend some time this coming Saturday 8/29 if you are available. I am GMT +4 (east coast US DST) and could do this anytime between 9:00AM and 7:00PM my time. I would also have time on Thursday 8/27 7:00PM to 10:00PM if that is better (probably very late for you). Let me know...

I have not used ICQ, but use SIM with AIM so I will try to set up an ICQ account.

Thanks for this offering!

[pasha]

Sorry, I afraid I will absent near internet at this week end :( Any day in any week exclude this week end. I see your authorize, so ICQ must work.

[David McKenna]

No problem... the next time I will have an opportunity during the daytime will be 9/4 (Friday) between 10:00AM and 7:00PM (= 2:00PM to 11:00PM GMT). Can we make this work?

I did set up an ICQ account: #596909246.

[pasha]

Ok. I'll reseved for you.When begin your time I send ICQ message for you

[David McKenna]

I'll be waiting. Thanks again...

[David McKenna]

No change with 3.18. I'll upload acpica$.

[pasha]

I remember, that we want going to online session for debug your problem? Say day in this or next week.

[David McKenna]

I'm sorry to say I will be very busy these next few weeks and will only have time to set aside for this on Sundays from 10:00AM to 4:00PM (2:00PM to 8:00PM GMT) both 10/18/09 and 10/25/09. If you can do that, let me know. If not, I understand...

[pasha]

Ok. I wait message in ICQ 18 october

[David McKenna]

No change with 3.18S package. There is new information in acpica$ so I will upload.

BTW, I tried to contact you by ICQ this past Sunday - it showed you were online when I did.... is everything alright?

[David McKenna]

acpica$ with ver. 3.18S

[pasha]

I here and wait your ICQ message

[David McKenna]

I started up my system this morning (using the version of ACPI you gave me yesterday)and it hung right away just before the WPS blue screen would normally appear. I captured a comport log for you and will upload...

[David McKenna]

Just got home from work, started the computer, and got a Trap0008 after loading drivers. I will upload the COM port log....

[pasha]

add pls /Ox to acpi.psd string in config.sys, where x - your comport.

[David McKenna]

OK... I will upload a couple of boot logo hangs, this time with /O1 (sorry about that!).

I will update the others as they occur...

[David McKenna]

Hang at Boot Logo #1

[David McKenna]

Hang at Boot Logo #2

[David McKenna]

Here is a new log after a Trap0008 and also another hang at boot logo....

[David McKenna]

COM port log during Trap0008

[David McKenna]

Hang at Boot Logo #3

[David McKenna]

Here is a new log for the hang right before the WPS appears....

[David McKenna]

ComPort log during hang right before WPS appears

[pasha]

Debug kernel is better for this. But I understand. Wait new version, then continue.

[David McKenna]

OK, I installed the debug kernel and the 318FB version of ACPI. The only change I see is there are no more Trap0008 - instead it just hangs where there used to be a Trap0008.

I will upload a bunch of logs...

[David McKenna]

First Hang at boot logo 3.18FB

[David McKenna]

Second hang at boot logo 3.18GB

[David McKenna]

Third hang at boot logo 3.18FB

[David McKenna]

Fourth hang at boot logo 3.18FB

[David McKenna]

Hang before WPS appears

[David McKenna]

Hang in middle of loading drivers

[David McKenna]

Successful boot 3.18FB

[pasha]

Ok. I see 1 trouble , tomorrow change and place version for you

[pasha]

Try ACPI318FN.ZIP from betazone. Next, create kdb.ini as:

G

ln

<empty string>

ln need for my look where was stop.

[David McKenna]

I got the new ACPI (thanks!) and created kbd.ini as you describe (I hope - I will upload it in case there is a problem). I was able to test booting many times and so far I have only had hangs at the Logo. I will upload logs for them. If any others occur I will upload them as well...

[David McKenna]

First hang at logo 318FN

[David McKenna]

Second hang at Logo 318FN

[David McKenna]

Third hang at Logo 318FN

[David McKenna]

Fourth hang at Logo 318FN

[David McKenna]

Successful Boot 318FN

[David McKenna]

KBD.INI used with 318FN

[pasha]

kdb.ini is right. Tomorrow we can repeat session in ICQ

[David McKenna]

Successful boot with 20091108

[David McKenna]

This morning my system hung right before the WPS would appear.... I will upload the log.

[David McKenna]

Hang before WPS using 20091108

[David McKenna]

FYI - since installing 20091108 I have not had any hangs at boot logo, or Trap0008's. Shutdown works and ACPI appears on the XShutdown settings page too. The only problem is the hang before the WPS (always the same). It will happen after the computer has been off for > 1 hour or so. The next attempt will always boot. Reboot attempts after that will always work as well....

[pasha]

try acpi318APIC.zip pls. APICOnly in acpi.cfg. Only for os4 kernel now. If will success - I do patch for IBM kernels too

[David McKenna]

COMport log 318APIC OS4 kernel with APICOnly

[David McKenna]

COMport log 318APIC OS4 kernel no APICOnly

[David McKenna]

I tried this new version 318APIC. If I add APICOnly to ACPI.CFG boot hangs at the logo with both the OS4 and IBM kernels. If I remove APICOnly from ACPI.CFG I can boot both the OS4 and IBM kernels. I have uploaded COMport logs for the OS4 kernel. (I have logs for the IBM kernel too if you want them).

The real test will come after I power off this computer for at least 1 hour....

[David McKenna]

I still get a hang right before the WPS would appear after the machine has been turned off for awhile.... Next attempt to boot works.

[David McKenna]

No change with 318Full....

[David McKenna]

No change with 318ND....

[David McKenna]

No change with 318APIC1stCall.... except a little more information in ACPICA$ (debug APM?). I'll post it....

I tried with the SVN1920_Unoff kernel too - no difference.

[David McKenna]

ACPICA$ from 318ACPI1stCall

[David McKenna]

ACPICA$ using ACPI318PREREL

[David McKenna]

No change again with ACPI318PREREL. ACPICA$ has new info so I posted it. I tried APICOnly and it would hang at the logo (both OS2 and OS4 kernels) but I could CAD. I still have to boot twice after a power off toget to the desktop....

I wonder if having two APM devices with the same name can cause problems (in my case "")?...

[pasha]

APICOnly don't tested now. Wait next version. "" usially mean USB drivers

[David McKenna]

ACPI318ESR is a step backwards - I can not boot at all with IBM loader and Kernel. The boot hangs at the point the video would normally switch resolution. I will upload a COMPort log.

I can boot with OS4 loader and kernel though...

[David McKenna]

COMPort log ACPI318ESR

[pasha]

Can you try debug IBM kenel and os2ldr?

[David McKenna]

COMPort log ACPI318ESR with debug kernel

[David McKenna]

OK... it seems to hang a little sooner with the debug kernel...

[pasha]

Log with trap?

[David McKenna]

No trap, just a hang. Or do I misunderstand your question?

[David McKenna]

I experimented with 318ESR some more and found I can also boot with OS4 loader and IBM kernel. I will upload acpica$.

[David McKenna]

ACPICA$ with OS4 loader + IBM kernel

[pasha]

try acpi318ESR2.zip pls with IBM kernel and os2ldr

[David McKenna]

No change with acpi318ESR2 - with IBM loader and kernel, it hangs in the exact same place. I can only boot if I use the OS4 loader.

Do you want a new COMPort log?

[David McKenna]

The latest OS4 loader (2278) with ACPI318ESR2 and IBM kernel will allow this computer to boot the first time every time. Now it behaves like a real computer!

Time to get a new one :-)

[David McKenna]

No change with 318SW. I will upload a COMPort log (using IBM loader and IBM kernel) and ACPICA$ (using OS4 loader and IBM kernel).

[David McKenna]

COM Port log of ACPI318sw boot hang

[David McKenna]

ACPICA$ of 318sw booted with OS4 loader

[pasha]

Please download experimental ACPI build from Mensys site:

• Experimental build for you:

acpi319ForDesktop.zip

• Read instructions: ​http://ecomstation.ru/projects/acpitools/?action=testcase

[David McKenna]

I am now using eCS 2.0 GA with latest DANIS (1.8.7) and UNIAud (1.9.5) drivers.

There is no change. I cannot boot this computer using the IBM loader and this new ACPI. It hangs in the same place as before (just before the screen switches resolution for SNAP).The same goes for the ACPI that came with eCS 2.0.

If I use the OS4 loader, I can boot this machine and it works well. ACPI.PSD /SMP /APIC /TMR /R.

[pasha]

Create and attach pls comport log. I understand, that for you all is same, each log can say some for resolving your problem. Better choice - both log. os2 loader and os4 loader.

[David McKenna]

No problem, I will upload now.

[David McKenna]

ACPI 319 for Desktop with IBM Loader

[David McKenna]

ACPI 3.19 for Desktop with OS4 loader

[pasha]

Try pls UNIAUDATIRQ.ZIP and IBM loader and attach log (acpi log)

[David McKenna]

3.19 for Desktop with Uniaudatirq

[pasha]

can you find and install debug kernel?

[David McKenna]

Just like before, it hangs a little sooner in the boot process with the debug kernel.

[David McKenna]

COMPort log with debug kernel and ACPI 3.19 for desktop and Uniaudatirq

[pasha]

If your machine has 2 comport, try disable com2 in BIOS? Kernels debug output now going via com2. acpi via com1

[pasha]

to add: You can create kdb.ini at root on boot drive with next context:

.b 115200t 1

g

[David McKenna]

OK... now there is more info in the file...

[David McKenna]

ACPI 3.19 For Desktop with UniAudatirq and debug kernel with COM2 turned off.

[pasha]

add pls ln to kdb.ini and repeat log. Best for kdb.ini

.b 115200t 1

g

ln

k

ks

kb

[pasha]

to add. Also you cat type ln<enter>, ks<enter> e.t.c. from your comport terminal

[David McKenna]

Debug kernel, ACPI3.19 for desktop, UniAudatirq and no COM2 with new kbd.ini

[David McKenna]

OK... I changed kbd.ini but it did not make any difference. Typing the commands on the terminal did get extra info. Uploaded...

[pasha]

Do you sure, that os2krnl.sym is right? I don't see int 3 at this label. Check pls os2krnl.sym, if this is right, type 10 times u<enter> in your terminal.

[pasha]

Ok. you have not compared os2krnl.sym for debug os2krnl 104a_SMP. I find where this problem - I need think

[David McKenna]

debug kernel and .sym + ACPI 3.19 For Desktop + Uniaudatirq + No COM2

[David McKenna]

You are correct, I did not use the debug SYM file. Now I have installed it and uploaded the COMPort log, plus the kbd.ini commands and 10x u.

[pasha]

Please download experimental ACPI build from Mensys site:

• Experimental build for you:

AcpiFor405.zip

• Read instructions: ​http://ecomstation.ru/projects/acpitools/?action=testcase

Need log + ln + u if has int 3

[David McKenna]

debug kernel + ACPIfor405 +uniaudatirq + No COM2 + ln + u

[pasha]

Please download experimental ACPI build from Mensys site:

• Experimental build for you:

acpifor405-1.zip

• Read instructions: ​http://ecomstation.ru/projects/acpitools/?action=testcase

if you have trap, I need in log + ln + u. Next. you need change kernel to retail (non debug) make photo of trap and attach here.

[David McKenna]

debug kernel+ACPIfor405-1+Uniaudatirq+NoCOM2+ln+u

[David McKenna]

This one also will not boot with the OS4 loader...

[David McKenna]

Replying to David McKenna:

This one also will not boot with the OS4 loader...

Turns out this is not true... I CAN boot the retail kernel with the OS4 loader (I still had the debug kernel with ACPI.PSD /O1 but the null modem was removed).

No need for a photo... with the IBM loader, retail kernel and ACPIfor405-1 boot hangs on a blank screen right before it would switch to the desktop. This has been the case all along.

[Steven Levine]

My analysis of debug+ACPIfor405+ln+u.2.TXT says you are using the wrong .sym file.

My best guess is that the trap in right before BeginIntHook in intIRQRouter.

The correct .sym file is

8-11-05 9:50 192,148 0 os2krnl.sym

with md5

a8f73100668f6c610867079ee44f33bd *os2krnl.sym

Please check this.

[David McKenna]

You are correct, that is not the SYM file I am using. I am using:

6-11-09 4:07 173,188 0 os2krnl.sym

found in the \OS2IMAGE\DEBUG\SMP directory on the eCS 2.0 GA disc 1. There is only one other os2krnl.sym file on that disc and it is in the \UNI directory also dated 6-11-09.

Where can I find the SYM file you require?

[David McKenna]

OK... I found the correct debug SYM file. I will install and re-do the last COM port log.

I notice also that the kernel in the DEBUG\SMP directory on the eCS CD is different... it is also dated 6-11-09 and is 1,000,533 in size. Or maybe these are not debug files?

Sorry for all this confusion...

[Steven Levine]

IBM has odd naming conventions, at times.

I assume you found \OS2IMAGE\DEBUG\SMP\OS2KRNLB.SYM. Just in case you don't know, you need to rename it when you copy it to the root of you boot volume. PMDF can use it without renaming.

[David McKenna]

Corrected debug files + ACPI405-1 + ln +u

[David McKenna]

No.. I did not know that. I googled 'OS/2 SMP debug kernel' and found a site that had IBM OS/2 kernels (retail and debug) going back a decade. I got 'smp0811d.zip' which contained an OS2KRNL which matches my current debug kernel, and an OS2KRNL.SYM which matches the one in your description. I installed and the COMport log is uploaded.

I hope everything is in sync now....

[David McKenna]

Tried the new ACPI319CR4 - here are the results:

SMP kernel IBM loader and ACPI.PSD /Q : system boots

UNI kernel IBM loader and ACPI.PSD /Q : system boots

SMP kernel IBM loader and ACPI.PSD /SMP /APIC /TMR /Q : system hangs before screen res switch

SMP kernel OS4 loader and ACPI.PSD /SMP /APIC /TMR /Q : system boots

I will upload logs...

[David McKenna]

ACPI319CR4, SMP kernel, IBM loader, ACPI.PSD /Q

[David McKenna]

ACPI319CR4, UNI kernel, IBM loader, ACPI.PSD /Q

[David McKenna]

ACPI319CR4, SMP debug kernel, IBM loader, /SMP/APIC COM log

[David McKenna]

ACPI319CR4, SMP kernel, OS4 loader, /SMP/APIC

[pasha]

Please download experimental ACPI build from Mensys site:

• Experimental build for you:

acpi319PollingFullOnly.zip

• Read instructions: ​http://ecomstation.ru/projects/acpitools/?action=testcase

[David McKenna]

I tried the PollingFull version and the results are similar to CR4 - I can boot in PIC mode, but not in APIC mode unless I use the OS4 loader. I will upload logs...

[David McKenna]

ACPICA$ with UNI kernel ACPI.PSD /Q

[David McKenna]

ACPICA$ with SMP kernel ACPI.PSD /Q

[David McKenna]

COM log SMP debug kernel ACPI.PSD /SMP /APIC /TMR /O1 /Q

[David McKenna]

ACPICA$ SMP kernel ACPI.PSD /SMP /APIC /TMR /Q + OS4 load

[pasha]

Please download experimental ACPI build from Mensys site:

• Experimental build for you:

acpi319PollingFullOnly.zip

• Read instructions: ​http://ecomstation.ru/projects/acpitools/?action=testcase

[David McKenna]

Hunh? This is the version I posted logs for 4 weeks ago...

[pasha]

Please download experimental ACPI build from Mensys site:

• Experimental build for you:

ACPIFIXVW-2011-04-12.ZIP

• Read instructions: ​http://ecomstation.ru/projects/acpitools/?action=testcase

[David McKenna]

Since my last report I have changed my hardware by adding an Adaptec AIC160 SCSI adapter and two (C:=9GB and D:=80GB) drives and removed the SATA drive. I had to add 'REMAP 18 TO 11' to ACPI.CFG to get the SCSI drive to boot with OS4 loader in APIC mode. I still could not boot with the IBM loader in APIC mode (PIC works fine). This was using the 3.18 version of ACPI that comes with eCS 2.0.

With this new ACPIFIXVW version, I can now boot using the IBM loader in APIC mode. Thanks! It seems to boot a little faster too.

Now I am having trouble shutting down - either APM or ACPI shutdown (in the eWP shutdown dialog) hangs after closing filesystems at a blue screen. Even CAD doesn't work. I can shut down using the 'Poweroff' utility in C:\ecs\ACPI\Util. Not sure if this is ACPI related or not... it happens even with version 3.18.

I'll upload ACPICA$

[David McKenna]

ACPICA$ using ACPIFIXVW-2011-04-12 in APIC mode with IBM

[pasha]

Please download experimental ACPI build from Mensys site:

• Experimental build for you:

acpi319for105-2011-05-30.zip

• Read instructions: ​http://ecomstation.ru/projects/acpitools/?action=testcase

[David McKenna]

I have not installed eCS 2.1 (which has the 14.105 kernel) on this hardware yet. Should I wait to test this ACPI version after installing 2.1 or do it now (with 14.104a kernel)?

[pasha]

Please download experimental ACPI build from Mensys site:

• Experimental build for you:

acpi319APICFix-20110605.zip

• Read instructions: ​http://ecomstation.ru/projects/acpitools/?action=testcase

[David McKenna]

I have installed (migrated) eCS 2.1 onto this hardware. I also installed acpi319APICFix2. I can boot in PIC mode, but not in APIC mode (IBM loader nor OS4 loader). I have added 'GPEDisable 0x00' to ACPI.CFG in order to be able to shut down completely. I have collected COMport logs, but not with debug kernel because I can not find version 14.105 debug. I will upload...

[David McKenna]

ACPI.PSD /SMP /TMR /Q ACPICA$ log

[David McKenna]

ACPI.PSD /SMP /APIC /TMR /O1 /Q COMPort log

[David McKenna]

ACPI.PSD /SMP /APIC /TMR /O1 /Q COMPort log with REMAP 18 TO 11

[David McKenna]

I should add:

I used to be able to add 'REMAP 18 TO 11' in ACPI.CFG and boot in APIC mode using the OS4 loader, but this does not work with acpi319APICfix2. See log...

[pasha]

You can't use IRQ11 for remap, because IRQ11 is using by another device. Try do next:

LINK LNK3 4

REMAP 18 TO 4

I mean, 1st you change PIC IRQ link to free value of IRQ, then remap to this IRQ. Attach comport log after doing pls.

[pasha]

To add. In some BIOS IRQ link can save changes value, check this too

[David McKenna]

What other device is using IRQ11? I don't see it...

You ask to LINK LNK3 4, but COM1 uses IRQ4. I need that for COMPort log :-)

REMAP 18 TO 11 worked with 3.18, now it doesn't.

There is no way to assign IRQ in BIOS. I disabled IDE and SATA (I don't use them now) in BIOS to free IRQ's. Now everything has it's own IRQ (except USB ports all share).

Anyway, I will try your idea and report...

[David McKenna]

OK... I finally got a chance to do this. I added:

LINK LNK3 4

REMAP 18 TO 4

to ACPI.CFG.

Boot hangs at the boot logo. I will upload COMPort log...

[David McKenna]

ACPI3.19FIX2 /SMP /APIC with LINK+REMAP

[David McKenna]

This hardware works fine with version 3.19.14 using the /VW switch (because of the Adaptec U160 driver). Go ahead and close this ticket. Thanks!

[David Azarewicz]

Great! Thanks for testing.