The nuts and bolts => Installation & Updates => Topic started by: brodders on February 13, 2008, 01:27:43 am

Title: Two issues with 5.9 Gold
Post by: brodders on February 13, 2008, 01:27:43 am
Two issues with 5.9 Gold

These are for an install on an Intel BX chipset Abit BX6r2 mb with a 700MHz P-III cpu and 512M ram. Near 10 years old and goes like a train - usually runs Mandrake 9.1 (from 2003).

The system has been fitted with a SATA PCI card and a large SATA disk, plus 2 old PATA drives.

VL 5.9G is on sda5 (with Mandrake 9.1 on hda1 and Wolvix 1.1.0 on sda1) but the system boots into VL 5.9G

1. Samba Issue
The box serves it's 3 HD's onto the local LAN. The fastest way to set that up was for me to make a symbolic link in the server user home dir to the machine's /mnt, where all the HD's are mounted.

I make sure Samba was configured to serve / follow sym links.

But the share Samba serves does not give this!

What happens is that the user home dir is shared, but following the link takes you to the LOCAL machines /mnt, not the /mnt on the server :( I could not see the 3 remote HD's, just my local ones.

In retrospect – is this a security issue? After all, as a user I might not have permission to rummage around in /mnt on my local machine. But Samba lets me...

If this is a security issue, where is the fault? My local machine is using VL Soho 5.8.6. RC2

2. The same server – does not power-down. Seen this before on Slax builds on BX chipsets; Wolvix is the same. For some builds the BX does shutdown OK but AMD K6 boxes do not.

Anyhow, on this server halt / shutdown “works” - but the machine hangs with power-up at the point when all services, HDs etc are stopped and it should finally turn itself off. No error messages on screen.

Is there a boot-flag which will cause Intel BX chipset machines to shut-down properly?


Title: Re: Two issues with 5.9 Gold
Post by: nubcnubdo on February 13, 2008, 02:48:14 am
Try adding "acpi=force" to boot parameters. If that doesnt work, try appending LILO for APM power management.
Title: Re: Two issues with 5.9 Gold
Post by: lagagnon on February 13, 2008, 07:23:13 am
Is there a boot-flag which will cause Intel BX chipset machines to shut-down properly?

Yes. Read this, it might work:
Title: Re: Two issues with 5.9 Gold
Post by: brodders on February 14, 2008, 03:17:02 am
Hi guys,

:) what to say. The apm system works very well... too well.

This in conjunction with mb BIOS, which turns on cpu temp monitor hardware with APM Enabled.

There seems to be a module loaded early in booting which checks for CPU temps (my mb hardware supports - but does not have - an external thermocouple, which on some rigs of this era was pressed onto the underside of the CPU, most cpu's not having a temp sense diode).

Alas the lack of thermocouple is reported as "High CPU temp". The monitor s/w, during boot, sees the "high temp" and shuts itself down out of fright of frying something.

So the rig now never completes a boot. Good shutdown, though....

// that's what I had with Wolvix a year ago! Now I remember why I didn't use it! //

Here is a thought - if I could find a way to stop cpu temp monitor being loaded during boot, and get it loaded in the shut-down script - I would have a way to turn the rig off!

This is getting very OTT. Basically, the CPU temp monitor needs to be not loaded. Anyone know what it's called and how to stop it loading??

I need APM - it's proven to shut the rig down.

:( this feels like a loosing battle. I don't want to walk away from VL; at an application level it does everything I need.

Title: Re: Two issues with 5.9 Gold
Post by: nubcnubdo on February 14, 2008, 04:47:38 am
The protective temperature-related shutdown can be disabled in BIOS. On my computer, this setting is located in "PC Health Status" section of BIOS. If your computer is shutting down due to temperature, it is very likely to have this setting somewhere in BIOS. You can either disable the protection, or raise the shutdown temperature to 65C or 70C. (I hope your CPU doesn't get hotter than that!)

There is another circumstance that could be causing thermal shutdown: the CPU fan cuts off during boot. This happens occasionally when BIOS and kernel compete for power management duties. The solution is to disable both APM and ACPI in BIOS and then to append LILO for APM. So I recommend you remove the cover of your computer and observe the CPU fan during boot, to see if the fan cuts off during boot.
Title: Re: Two issues with 5.9 Gold
Post by: brodders on February 15, 2008, 06:34:35 am

A real game. I can turn ACPI on/off as well as APM on/off in the BIOS.

Any ACPI = Enabled => will not boot with Coldplugging "reached critical temperature (71F) - shutting down"

Any APM => will not shut down

This is with LILO appendflags for acpi, apm and both together.

I have control over "Temperature Warning" which presently has a number in it over the Coldplugging module alarm. Coldplugging seems to listen to itself, not this setting.

The other thing is that my BIOS reports "CPU (RT2) Temp: N/A - which I take it to mean the BIOS knows the sensor is not there.

I expect the hw does something like flag an impossible number to the BIOS (like 999?) anyhow, the BIOS knows the sensor is not fitted.

But Coldplug does not and it is shutting the rig down. I want Coldplug to be gone!

CPU fan - keeps turning all the time, apparently (by ear) - the only clear changes are in disk spinning sounds.

Think time to chase the Coldplug team. Hm. Why do other distros shut this rig down then without any issue??