Hi,

I found that some people complain about too high frequency
of the kernel timer (1000kHz). So, I have two questions:

1) what is it kernel timer? where is it used?

2) is this may cause turning off power?

Recently I had unexpected shutdown - power goes on - and I want to
know if I may blame for this the kernel timer.

My hardware: motherboard MSI MS-7211, Pentium 4 HT 2,99 Ghz, (6 years old)

Kernel timer may cause higher cpu temps and more psu current draw. See bios health settings and any other logs that may lead to causes.

Run diag cd such as ultimate boot cd, start with memtest.

Kernel timer I believe how the OS checks interrupts and uses clock cycles. A finer or faster speed would help some setups that may need closer to real time kernel. I could be wrong on that.

I've never heard of damage to hardware being done by high timer frequency ... if someone has heard of it, please report, because I'm running at 1000

Timer frequency mostly has to do with latency vs throughput. If you're running a desktop system you will probably choose 1000 Hz, if you're running a server or something that requires maximum performance and throughput and you don't need to run video or audio apps, then you'd go for tho other lower frequencies. As an experiment try running multimedia apps with the lower frequencies, the latency will be too high, and will make the video unwatchable.

Perhaps just as important and dealing with the same issue is Preemption Model, read the Help on each of them in 'make menuconfig' as for the above as well. If you're running a desktop there are two choices, if server only one. You would choose the Voluntary for a normal desktop, Preemptible for low-latency desktop, and none for server.

No, none of this can cause your machine to suddenly shut down, it is much more likely that it is something completely different.

Post at least the 'dmesg' or look through the logs at '/var/log/syslog' and messages to see what happens right before the shutdown. Like jefro suggests, it's also a good idea to run some diagnostics, especially memtest.

One possibility: your 6 year-old power supply is going, and the shutdowns happen when the power demands are more than it can supply.

Yes this can result in hardware damage.

Thank you all,
I think that the general impression about this thread is that something is wrong with my computer (hardware). It may be true. So it is a time for a new one...But "new system new computer" .. Where did I hear that? Mmmm....

Stay cool, love your wives/husbands...

It could also be a kernel bug in one of the drivers. I told you, try and post the output of 'dmesg' or maybe the logs.

@H_TeXMeX_X

I hope I understood you well. Here is an output of dmesg

I am still using huge-smp kernel.

Attached Files

dmesg.txt (30.1 KB, 19 views)

Yes, that may help. I notice:

The first may indicate a possible bug in the BIOS, but it's not for sure. There's always risk involved in updating the BIOS, and I would only do so if you have to, and only with a DOS boot disk. First I would eliminate other possible causes, such as failing PSU or bad RAM, etc.

The kernel you are running is the huge kernel in slackware, and it's not a good idea to run it on a production system. It is recommended that you use the generic kernel + initrd, or you build your own kernel (making sure to build-in support for filesystem drivers and HDD drivers, otherwise it will not boot). Try the generic + initrd first.

I also notice that you're using the sata_via and pata_via. There aren't particularly stable, and I would use AHCI instead, but if your HDDs run ok, then you can leave it the way it is.

I updated BIOS about year ago. I also noticed both of these messages.
I will follow your advice and will try the generic kernel.
AHCI - it is supposed to be set in BIOS, is it? Are there any particular driver for AHCI?

Yes, you have to set AHCI in the BIOS, and use ahci sata driver (if you compile a kernel, you should build this in, or it won't find the HDD).

Hi,

I switched to the generic-kernel. The first thing I noticed the quality of image display improved much. But I have no idea for what reasons.

I have only two options in BIOS for SATA disk drive: IDE and RAID. So, according what you wrote I cannot use ahci drivers - or is there a workaround?

Again, thanks.

Sometimes RAID = AHCI, at least according to the BIOS, maybe check the manual and see. You can try it and see if it works, if it doesn't change it back.

igadoter, it seems you have just have data corruption. Using the generic or vanilla kernel should not improve graphics quality unless the generic is a more recent kernel which could have more features to handle your graphics card.

You may want to increase the memory voltage by 0.1 volt to increase stability. All data that is written and read from the hard drive is handle first in memory. DDR memory can handle 2.6 volts, but be careful using more than this. The normal voltage for DDR is 2.5 volts.

Like someone said, replace the power supply if you having problems like this and it is at least 2 to 3 years old. The brands that I suggest are Seasonic, Enermax, and FSP.


This is wrong to say. Slackware is used on servers too. It is just a distribution. It depends on the setup and the distribution really does not matter unless the user does not know what he or she is doing. Saying to use a vanilla kernel will not help much.

I disagree with this too. AHCI is not a substitute for the real driver for your storage controller. Use AHCI as the last resort when the operating system does not support your storage controller. AHCI performance is poor because its specification is based on layers compared to the actual drivers that is specifically for your storage controller. AHCI is like VESA, it uses standards in order to get the device to work, but it does not utilize its full potential.

VIA storage controllers works just fine. I would be more alarmed about JMicron and Silicon Image controllers.

@Electro,
Thank for your post. Very interesting. How to increase memory voltage? I am rather hardware ignorant. Memory on my computer is "Auto by SPD". For me it is pointless
to buy new power supply. It is about 70% or more of today price of the computer!

@H_TeXMeX_H,
I am interesting in your opinion about the @Electro post.

I wouldn't change the RAM voltage, it is dangerous, could fry the RAM. If the PSU is the cause, then change the PSU, but it is hard to know, you'd have to have a spare PSU to try. I would take it to a repair shop, they can tell you if it is a bad PSU. But, first, I would eliminate other possible causes.

As for the rest, he is wrong, and have him on my ignore list, so I'd rather not even reply. But for your benefit, I suppose I should address them.

I always recommend AHCI and ahci driver over any other sata driver in the kernel. This is because it is a better driver, with less bugs (in fact no bugs that I know of), and it provides BETTER performance than the other SATA drivers. It allows for advanced functions like NCQ and hotplugging, so it is MORE advanced and BETTER than any regular sata driver. Not to mention that a number of sata drivers have known bugs.

See the wiki:
http://en.wikipedia.org/wiki/AHCI

Maybe it's just personal preference, but I've found ahci to be a much better option in most cases. In some rare cases ahci is worse than the sata drivers, but this is rare.

Elecro doesn't even use Slackware, so he has no clue about the difference between the huge and generic kernels. Even Pat V (the creator and maintainer of slackware) suggests that you use the huge kernel only for installs, and switch to generic + initrd for a production system. This is because (this is my interpretation of what is going on) the huge kernel has everything built-in, and although this is great for an install kernel, where it has to detect as many possible hardware configs, it is not great for a production system where it could cause conflicts among drivers and strange bugs. If drivers are build-in, you can't remove them from the kernel if they cause problems, you can't 'rmmod' them. Sometimes they start probing for devices that do not exist, and thus may cause similar devices to freak out, which may cause system hang or kernel panic, etc. It's true that this isn't that common, but it does happen, and to clear up any possible connection to this, you should try the generic + initrd, or just compile your own.

You have no idea how much a custom kernel means, it can mean 100% stability if you configure it right. However, it is a lot of work, and technically in many cases the stability is the same as generic + initrd.