LinuxQuestions.org

LinuxQuestions.org (/questions/)
-   Linux - General (https://www.linuxquestions.org/questions/linux-general-1/)
-   -   INIT: PANIC Segmentation Error (https://www.linuxquestions.org/questions/linux-general-1/init-panic-segmentation-error-3266/)

cornell 06-11-2001 08:10 AM

When booting Linux (Mandrake 6.0), I get past the SCSI scans (no devices) and get to a point where it says something about freeing memory and then I get:

INIT: PANIC Segmentation error...

I've been running this box for a few months and yesterday I upgraded the memory on the machine, added a 64 Meg DIMM to two 32 Meg DIMMs. POST reports 80 Meg (don't ask why it's 16 Meg per module short, I don't know, that's another question, I think/hope. Prior to the additional 64 Meg, it reported 32 Meg, again 16 per module short, but it worked.)

After the upgrade, I turned on the machine and it failed to boot, couldn't find the hard drive. Apparently, I'd loosened the data cable at the motherboard when putting in the modules. Seated the cable. Turned on the machine, it finds the drive and starts to boot, then I get the PANIC message.

Tried to boot from the Mandrake Boot Floppy, same results.
I can get the Mandrake installation to start up, offering to upgrade / install Mandrake, as long as I don't lose my data on the hard-drive, this is not an unacceptable path. But I'd rather not.

Any ideas, suggestions?

Thanks in advance
Cornell

unSpawn 06-11-2001 02:35 PM

since u get memory/BIOS failures I would look into that first. test with replacement memory if it reports it correctly and see if ure BIOS needs upgrading.
older LILO rely on the BIOS to tell it the amount of RAM, so if the BIOS "lies" to LILO itll happily use the value provided. a way around that is adding an "append mem=128m" to lilo.conf, or get a newer LILO, but since ure not getting there Id troubleshoot the HW first.

cornell 06-11-2001 03:01 PM

Thank you...

But if BIOS lies by telling too small a number, what harm would there be? Any memory accesses will be against actual memory. If the lie was telling too big, that would be a problem, memory accesses could occur against non-existant memory. Besides, as I said, prior to the upgrade, the memory was being under-reported also, and Linux booted up just fine.

Can't make any changes to lilo.conf, until I get the machine up :-(

unSpawn 06-11-2001 03:41 PM

I cant guess at what failures can occur, it just doesnt seem right to me.
Hmm. if u arent willing to explore a BIOS/RAM check I suggest u take out the added RAM, make the necessary changes to LILO & then put it back in..

mcleodnine 06-11-2001 09:15 PM

Quote:

Originally posted by cornell
Thank you...

But if BIOS lies by telling too small a number, what harm would there be? Any memory accesses will be against actual memory. If the lie was telling too big, that would be a problem, memory accesses could occur against non-existant memory. Besides, as I said, prior to the upgrade, the memory was being under-reported also, and Linux booted up just fine.

Can't make any changes to lilo.conf, until I get the machine up :-(


Bad Idea(tm). Harm. Mayhem. It will lead to drugs, guns, and ****ing in the streets.

If BIOS reports that the memory size is different to what is installed it is merely a SYMPTOM of something WRONG, very wrong. It's jusk plain KOOKY to run on RAM that the BIOS reports as suspect. Either the RAM is mis-matched, the Memory controller is ouched, or a combination of the two. What kind of RAM is it? What speed is it? Is it error correcting? Is it EDO? Buffered? Is one stick defective? Whta about the voltage? MOBO or RAM has gotta go.

Sure, you can put brake fluid in your car's engine oil and the 'Check Oil' lamp will go out. It doesn't mean the car will get you very far. Ever seen what happens to a filesystem (Windows OR Linux) where the CPU overheated during a busy i/o duty cycle?

You might as well mount a set of Firestones on there as well.

This is NOT a flame. Strong opinions were voiced here in order to underline the severity that this configuration WILL have on a 'running' system.

Had this reply been a real flame it would have implemented at least several uses of body appendages and references to geneology and magic markers. Frequent sightings of the phrases 'RMS' and 'Moores Law' have also been reported in past flames.


Please take the content of this reply as was intended - a warning of dire consequences.

[Edited by mcleodnine on 06-11-2001 at 10:24 PM]

cornell 06-12-2001 07:38 AM

Thank you both.

In my defense, I'd like to point out that I thought it had 32M. I pulled the memory to take to a computer show so as to get more of whatever it is I got. And they do trades. I was surprised when he tested my DIMMs as 32M 66. Which was when I found out that it should've been reporting 64M. Anyway, I bought a 64M 66, installed it and began experiencing the problem.

Last evening, I was checking out the PC's setup to see if I could find anything memory related. Under the "Chipset" section, I stumbled upon a list:

C000, 16K Shadow Cached
C400, 16K Shadow Cached
C800, 16K Shadow Disabled
CC00, 16K Shadow Disabled
etc.

Well, I don't know what that is, but it says 16K, and it can't get worse, so I changed the third to Cached (the other option is "Enabled") and rebooted. Linux still no boot. So I shutdown and removed the new DIMM and rebooted. The POST now says 64M, Linux boots up and the initial logon screen that reports BIOS and version and stuff, reports 64M (actually it says 63).

So now I have 64M and the machine and OS know it. That's good. But what do I have to do to put in the additional 64M? And what is 16K Shadow and what's it do? Does Linux have a memory limit? (I'm sure it does, but I doubt that it's that small).

I expect that if I install the extra DIMM, and do whatever it is I have to do to the machine (whatever that is) that Linux will be alright.



unSpawn 06-13-2001 05:41 AM

now lets establish the facts again, watson.
Quote:

1. In my defense, I'd like to point out that I thought it had 32M.
2. I bought a 64M 66, installed it and began experiencing the problem.
3. The POST now says 64M, Linux boots up.
So (1) there *was* something wrong the whole time.
If under (3) it is the *old* RAM that reported as 32Mb and now reports as 64Mb, there is either something flakey like corrosion on the pins but then they must be a few yrs old. If its not that, then theres something wrong with ure mobo/BIOS and/or settings. Now if it aint the old RAM but the new block, its crystal clear the 3 blocks wont operate together well so u gotta match em up. some boards are known to have difficulties with using different brands of RAM together. I always buy the same brand & I always split the value by 2, so I still got half of it if one module dies.

About the shadow options, this is for taking routines from slow ROM and caching em into RAM. If it's BIOS, go right ahead, that's always good, if it's video ROM u gotta check, not all HW combinations will let this work flawlessly, u may experience lockups etc. The 16K refers to the cachable blocksize, so that aint in relation to the RAM question.
Linux has a max physical memory limit as compile-time choice, somewhere between 1 and 2Gb.

Quote:

I expect that if I install the extra DIMM, and do whatever it is I have to do to the machine (whatever that is) that Linux will be alright.
Again, check ure hardware, like BIOS patches & settings, (mis)matched/faulty memory modules before thinking it's a Linux problem. U may want to check the excellent http://www.pcguide.com on troubleshooting HW problems and/or BIOS settings.


All times are GMT -5. The time now is 01:09 PM.