Toshiba Tecra T9000 boot hang with "HD busy" on

[Chuck McKinnis]

Having upgraded now (manually) to the latest ACPI/APM drivers, I am getting a problem again I already had end 2007 with the then current version.

Many times the notebook does start and shutdown ok with ACPI (although I did not test any sleep modes), and it solves a power management problem I have otherwise with the old APM ("low power" if battery less than 50%, making the display so dark that normal work is not possible any more).

However sometimes it hangs on startup, with the eCS boot logo in the screen and the HD light constantly turned on, but no HD activity to be heard. Some months ago I did some testing with ALT-F2 and looking which drivers are currently being loaded on hang, but it looked like it could be the one or the other, no systematic result.

Also trying to find out any pattern about what sometimes really means, and I am getting the impression that it may have to do with the docking station. At this moment it looks as follows:

• Several boots and shutdowns just fine, without docking station
• One (several?) reboots and shutdowns in the docking station ok
• Next reboot without docking station hangs, also trying several times
• Booting into Suse Linux 9.3 works, but shutdown takes several attemps and finally a C-A-D (but I do not know whether the C-A-D did the shutdown or one of the attempts finally succeeded!)
• Now also eCS with ACPI booted again just fine (still without docking station)

So it COULD be that there is a problem with power down when attached to the docking station with the result that the one or other device does not want to wake up afterwards.

Attaching acpi.log from a copy acpica$ operation, and a acpi-log-...ziq file from the logs collector, taken without the docking station attached. If that would be of interest, I could send the same files also once more, with docking station attached.

[Chuck McKinnis]

log from copy acpica$

[Chuck McKinnis]

from logs collector

[Chuck McKinnis]

e-co:

Pasha talks:

try remove OS2LOGO (copy it to backup dir), if you have a problem then make photo of the screen using digitall still camera

​http://ecomstation.ru/projects/acpitools/?action=logs

[Chuck McKinnis]

The story with these screen fotos was that I had to do what you propose, then shutdown in the docking station, then undock, then try a reboot to get the hang.

Actually this seems now "almost" to be the regular pattern, because with one exception boot went through ok if either in the docking station or outside, but then the last shutdown had to be outside already. The one exception was outside the docking station, but also the last shutdown was outside, so it seems not to be 100% correct anyway.

This the information as far as I can provide it currently!

[pasha]

I need in output of GPEDevice.exe, when you boot with dockstation and without

[Chuck McKinnis]

Replying to pasha:

I need in output of GPEDevice.exe, when you boot with dockstation and without

I am uploading both, although I see that they are identical!

Furthermore I am uploading another screen fotograph. I took it during another boot hang. I had the computer in the docking station running, and in order to avoid a hang on next reboot, I removed it before shutdown. I had a hang anyway, with some differences compared to the first, so I am sending the second version screen foto. After "killing" the system with a long press on the power button a next retry to boot worked ok again.

[pasha]

Ok. Problem not in event. Can you boot without acpi.psd, then do pci.exe >pci.out and attach log here.

[Chuck McKinnis]

output of pci -D -B -P -R -M

[Chuck McKinnis]

Today I experienced another hang, this time when attached with the docking station. And since I am running now constantly with the debug driver, I could immediately take another screen image, see Variant3.jpg.

Sequence of events this time:

• Shutdown in docking station

• Reboot in docking station -> hang as mentioned above

• System "killed" with a LONG press on power button

• Reboot again -> again the same hang

• Power button pressed, but too short for a "kill" - and "miraculously" the Debug output said "PowerButton" - and went on with booting!

• Undock computer, shutdown

• Reboot -> again a hang, but "type 1" this time (my first screen images)

• Short press power button - no reaction. Long press to "kill"

• Reboot, now works fine

No idea whether this helps or makes the story only more confusing!?

One more pecularity of that notebook computer: Sound does not seem to work, i.e.: I could not get it running with neither eCS nor Suse Linux nor Windows XP. Years ago it worked with Windows (when the machine was still my office computer), but most of the time I try to make computers as quiet as possible anyway, so I could not say from when on it failed; I once just tried to get it running with any of the three systems, but no success any more.

But anyway, Linux and Windows seem to boot just fine, and I assume they both work with ACPI, so it must be feasible somehow!?

And one more thing: I am normally ONLY booting with eCS, not with the other systems, so the chance that they interfere in between two boots is not given, certainly not since I am posting for this "ticket".

[pasha]

1. What a devices are ppovides by your dock?

2. Place pls file AltF2On.$$$ to root at boot drive and do photo of freeze againg

[Chuck McKinnis]

Attaching one more screen image, with a hang of "variant 1" and AltF2On, but this time it happened in the docking station (and after shutting down also there). I leave that file now there, and if I am getting other variants, I may continue to take screenshots.

The docking station is mostly a port replicator and offers:

2 x USB 1 x mouse 1 x kbd 1 x modem 1 x network 1 x monitor analog 1 x monitor digital (?) 1 x parallel port 1 x sound in 1 x sound out 1 x ??? (little plug, don't know what that is...)

I am only using the network and the USB ports

[Chuck McKinnis]

New hang variant, with F2

[Chuck McKinnis]

Again variant 1, with F2, but another driver this time

[Chuck McKinnis]

Just attached two more hang variants that occurred yesterday evening, both with the computer attached to the docking station, and immediately following each other, i.e., first "variant4...", then hard reset and reboot, leading to "variant1...", then followed by a successful hard reset and reboot.

After all this, the following conclusions may be drawn:

1. Although it looked like for a while (and I wrote it even in the original title of this "ticket"), there doesn't seem to be a strict "pattern" concerning the notebook being attached to the docking station or not

2. In terms of ACPI debug output, several types of "hangs" are possible, but some come back repeatedly, like the "variant1"

3. What I already found during earlier tests with an earlier version of ACPI: It is NOT always the same driver that is being loaded during a hang

But I am now getting more and more confident that there is at least one thing that is constant with all hangs: The "HD busy" control lamp is ON during the hang, but no audible HD activity can be detected.

Maybe this latter observation leads to a more useful track!?

[Chuck McKinnis]

Variant 1, with more info from driver 080423 (!?)

[Chuck McKinnis]

Yesterday I upgraded to the new ACPI driver version. Manually, by replacing the files one by one. For some replacements, I had to boot to a "management" partition (because I was too lazy to find out again how to replace DLLs and EXEs that are currently running). The first reboot went fine then.

However, now the first shutdown did hang: I do it with the XCenter normally, and when I selected "Shutdown", and before being able to choose between "shutdown", "reboot" etc., the whole desktop was greyed and nothing happened any more. Mouse and keyboard were still "intact", so a double CAD had still the desired effect. Today this same thing happened again, so I may have to try the remedies you propose: do it with shutdown.exe from the ACPI tools, and see whether there might be BIOS updates available from Toshiba.

One of the reboots also showed a hang, which looked like a new variant of "variant 1", displaying somehow more information now with the new driver (!?). In any case I thought it might not hurt to upload the screen photograph.

[Chuck McKinnis]

Now the computer is being booted several times per day since many weeks, always with the ACPI debug output turned on. One more precise "pattern" can be reported:

Every time if the OS2LVM.DMD is reached, the boot process has definitely succeeded''

Hangs may occur at different points during the period before that, and always this is with a running, but inactive HD (= control light on, but no movements to hear).

The whole thing looks like some kind of deadlock between an ACPI function waiting for the HD and the HD waiting for maybe the same function to end.

I am attaching my CONFIG.SYS, which may give a hint about with drivers are being loaded before the OS2LVM.DMD.

[Chuck McKinnis]

Current CONFIG.SYS file

[eco]

Dock station..

can you join irc-chat? start openchat -- ​http://ecomstation.ru/download/openchat-20070217.zip irc channel: ​irc://irc.ecomstation.com/ecolabs server: irc.ecomstation.ru:6667, channel: #ecolabs

it's necessary hang 7 days x 8 hours and watch what happens. and spend 4 hours for real work if the developer is ready for research.

[cbockem]

Replying to stevenhl: Only now I realize that this issue has now "feedback pending" - and this even because my own intervention had reactivated the issue as "not solved" shortly before that! The reason why I did not react any more is simply the fact that my Toshiba Tecra T9000 has "died" completely (i.e., something is broken so that I cannot start it any more) already round the time that this "feedback pending" was set - and I was more busy with trying to retrieve the data from the HD (which was possible) than with further following this issue! And of course I must now apologize for the fact that I am not any more in a state to do any tests on a Toshiba Tecra T9000.

There is even some theoretical possibility that the computer already had some hardware flaw that showed up only in connection with the ACPI driver (not with the APM drivers), so the problem would be "solved" now the hard way ;-)

[eco]

Hardware is not available.

We keep in mind that Toshiba, HP have some problems..

We keep in mind that you are using Dock station.