Can't boot Salix -Openbox-14.1 on AMD K6-2 machines

[junqueman]

http://www.ultimatebootcd.com/

this has the memtest tool etc. ..

And can save coasters..

Good luck.

Here's the irony of ironies: Ultimate Boot CD won't boot in these 2 machines! (The same CD works fine in other machines.)

Thanks, though.

And many thanks to westms for the long, informative response. Very busy right now with other demands on my time, but I'll answer his questions later. Just no time right now to provide the thorough response that his to me deserves.

So sorry for the late reply.

[junqueman]

But first, you write in a section / paragraph:

'Finally, the following distros don't make it past the boot: (or boot>) prompt:'

and also:

'For any of these, any parameters entered at the boot: prompt result in an "aborted." response.'

but that can not be true. You've described before, with the exception of three CDs, all the others have produced the total message: 'Loading /Kernels/huge.s/bzImage. aborted.' Is this true or do you correct your descriptions?

Apologies for the confusion. Perhaps I was generalizing too much.

If at the boot prompt I try to load the huge.s kernel, I can get one of three outcomes, depending on the distro:

(1) Loading /Kernels/huge.s/bzImage. aborted.

(2) aborted.

(3) Could not find kernel image: huge.s

In each case I am returned to the boot: (or boot>) prompt following the message. I am assuming (3), and possibly (2) as well, means the huge.s kernel isn't included in the distro.

The still desired information about the loader and versions, can be obtained through (mount and) browse the ISO images. Examples:

mkdir /media/image

and

strings Downloads/connochaetos-14.1.iso | grep 'LINUX [0-9]*[0-9]\.[0-9]*[0-9]'

or

mount -o loop connochaetos-14.1.iso /media/image
strings /media/image/isolinux/isolinux.bin | grep 'LINUX [0-9]*[0-9]\.[0-9]*[0-9]'
umount /media/image

or by inserting and mounting the CD/DVD, like:

strings /run/media/USERNAME/connochaetos-14.1/isolinux/isolinux.bin | grep 'LINUX [0-9]*[0-9]\.[0-9]*[0-9]'

then umount the CD/DVD. This allows finding ISOLINUX, SYSLINUX und PXELINUX texts and the following version numbers. Please add the new informations to the concentrated list of distributions

Thanks for the lesson! Here's the results:

10. connochaetos-14.1: ISOLINUX 4.06 0x513e7151
11. lupu-528.005 Lucid Puppy 5.2.8: ISOLINUX 4.03 2010-10-22
12. antiX-13.2-386: ISOLINUX 4.05 20120630
13. lubuntu 14.04.4 desktop: ISOLINUX 4.05 20140113
14. lubuntu 14.04 alternate-i386: ISOLINUX 4.05 2014 0113
15. lubuntu 12.04-alternate-i386: ISOLINUX 4.05 20120131
16. Bohdi-3.2.1-legacy: ISOLINUX 4.05 20140113

After all previously written, I can not rule out a hardware defect or other problems. So I ask:
Were all boot CD's created on the same optical disc drive and can this drive be trusted?

Yes. They were all written on my not-quite-one-year-old Macbook Pro. I have never had any other issues with CDs or DVD's burned on this drive, even at the maximum rate.

Was always paid attention to low write rate?

The first CD's I burned (including the "successful" Ubuntu 8.05, AntiX M8.5, and possibly other "successes") were burned at the maximum rate. After the suggestion was made to burn at a lower rate, I have used exclusively the lowest rate available (10x). This probably includes the majority of them. Regardless, in casual observation I can't identify any correlation between distros which boot vs. those that don't with burn rate.

Was always the optimization of the writing process used, e.g. BurnFree?

I have been using Disk Utility to burn the CDs & DVDs on the Macbook. I am not aware of any optimization methods that may be available. After burning, a verification test is performed that has always passed successfully.

Did you tested the memory of the computer with Memtest86+?

Yes, but not before yesterday. I found I could run Memtest86+ v 2.11 from the AntiX M8,5 live CD.I allowed it to complete 2 "passes." In each case I got "Pass complete, no errors, press Esc to exit." I am confining my work now to the newer of the 2 machines, the K6-III, which now has 524MB of RAM. (I added RAM to the maximum allowed by the motherboard.)

Please check if the BIOS has a setting for the handling of 'keyboard controller A20 gate' / 'A20 gate switching'. Optionally, please turn on. (It's not the reason, I think, but...)

Unfortunately nothing close to that in the BIOS of either machine.

If you have read along up to this point yet, then you get a candy now. Candy: There is still a way to test for suitability for salixlive-13.0.1. For this purpose boot from DOS floppy disk, which must contain 'smartdrv'. Then change to the CD drive in which the salixlive CD must be inserted. Then enter 'cd \BOOT\DOS', then 'salixliv.bat'. This then trys by linld097.com to load the kernel.

I haven't tried this yet, but will give it a go at the next opportunity. Thanks for all the help and ideas.

You seem to be capable of suffering. 19 shiny coasters. This amount still offers another application. Perhaps you can hang the disks in your cherry tree.

Yes, I'm a patient man. Blank CDs/DVDs are pretty inexpensive for the entertainment/educational value derived from this exercise. I don't have a cherry tree, but do have a crabapple tree. Hmmm...

I wish success.

Me, too! In this case I define success as identification of the cause of the problem.

[junqueman]

If you have read along up to this point yet, then you get a candy now. Candy: There is still a way to test for suitability for salixlive-13.0.1. For this purpose boot from DOS floppy disk, which must contain 'smartdrv'. Then change to the CD drive in which the salixlive CD must be inserted. Then enter 'cd \BOOT\DOS', then 'salixliv.bat'. This then trys by linld097.com to load the kernel.

OK, tried this. I get:
"This kernel requires the following features not present on the CPU:
cmov
Unable to boot - please use a kernel appropriate for your CPU."

I have seen this message before, when attempting to run recent Ubuntu versions (above 10.04, at least).
I do know this hardware supports neither pae or cmov.

[westms]

Yes. They were all written on my not-quite-one-year-old Macbook Pro. I have never had any other issues with CDs or DVD's burned on this drive, even at the maximum rate.

And

I have been using Disk Utility to burn the CDs & DVDs on the Macbook. I am not aware of any optimization methods that may be available. After burning, a verification test is performed that has always passed successfully.

A few more words about that.

If a data CD/DVD is created and the data stream can reach the drive without buffer underflow, then everything is fine. But falter the data stream because the machine is too slow or the write speed is too high, then there are gaps of block size on disk. These gaps are in sequence with the used blocks and remain unused. As long as a full file system driver (ISO-9660 file system) is used to write and read, the only drawback lies in the reduction of the usable capacity of the CD/DVD.

But if ISOLINUX reads from a CD, then there is no file system driver. ISOLINUX uses a simple procedure to load e.g. a kernel. It recognizes the position and structure of the tabel of contents. File names with certain extensions, are not kernel names. All other file names come into consideration as kernels. The block number of the header of a kernel is taken from the table of contents. This block number is used to read the first block through a BIOS low-level function. Then the next block is read. This must immediately follow the first block. And so on, until all the blocks of the kernel have been read.

Empty blocks (= gaps) in this sequence are treated the same as filled blocks. There are only a few possibilities of checking available. ISOLINUX knows nothing about files when reading blocks and therefore nothing of gaps and therefore can not handle it.

One can see from this, buffer underflow must be prevented. A CD verification test, only shows whether file system and files are good.

You seem to be capable of suffering. 19 shiny coasters. This amount still offers another application. Perhaps you can hang the disks in your cherry tree.

Yes, I'm a patient man. Blank CDs/DVDs are pretty inexpensive for the entertainment/educational value derived from this exercise. I don't have a cherry tree, but do have a crabapple tree. Hmmm...

A crabapple tree is also beautiful but unsuitable. Probably there are no Starlings on the tree.

[westms]

If you have read along up to this point yet, then you get a candy now. Candy: There is still a way to test for suitability for salixlive-13.0.1. For this purpose boot from DOS floppy disk, which must contain 'smartdrv'. Then change to the CD drive in which the salixlive CD must be inserted. Then enter 'cd \BOOT\DOS', then 'salixliv.bat'. This then trys by linld097.com to load the kernel.

OK, tried this. I get:
"This kernel requires the following features not present on the CPU:
cmov
Unable to boot - please use a kernel appropriate for your CPU."

I have seen this message before, when attempting to run recent Ubuntu versions (above 10.04, at least).
I do know this hardware supports neither pae or cmov.

That's bad news. Of the cmov message, I have not read anything before.

There are AMD K6-CPU with and without "conditional move instructions", such as cmov. This is the result of a search. What a mess, if applicable. The AMD datasheets I could not download at AMD, to check that.

For this error message linld097.com must examine the CPU flags or test this command group directly. Your previous "inxi" call also returned CPU information: AMD-K6 3D+, but that has aroused no suspicion. Unfortunately, there are contained no information about the CPU flags. Examine the CPU flags you can do also. For this you could call "inxi -Cf" on the used machine, if installed, and show the result here.

[junqueman]

There are AMD K6-CPU with and without "conditional move instructions", such as cmov. This is the result of a search. What a mess, if applicable. The AMD datasheets I could not download at AMD, to check that.

For this error message linld097.com must examine the CPU flags or test this command group directly. Your previous "inxi" call also returned CPU information: AMD-K6 3D+, but that has aroused no suspicion. Unfortunately, there are contained no information about the CPU flags. Examine the CPU flags you can do also. For this you could call "inxi -Cf" on the used machine, if installed, and show the result here.

Here is the resulting output:

CPU: Single core AMD-K6 3D+ (UP) cache 256 KB clocked at 400.840 MHz
CPU flags fpu vme de pse tsc msr cx8 pge syscall 3dnow k6_mtrr

Thanks also for the information on CD burning, contiguous blocks, and all. If this was what was causing me problems, I would expect it would also fail when setting up a virtual machine on my Mac host from the same CD; am I correct?

[coyotl]

https://www.plop.at/en/bootmanager/plpbt.bin.html

It's very easy to make a Plop boot manager floppy disk.

[westms]

Here is the resulting output:

CPU: Single core AMD-K6 3D+ (UP) cache 256 KB clocked at 4000.840 MHz
CPU flags fpu vme de pse tsc msr cx8 pge syscall 3dnow k6_mtrr

With the question of the CPU flags (more correctly: (Extended) Feature Identifiers) I wanted to see if the CMOV flag exists. Your output shown does not include this flag. The PAE and NX identifiers are not shown, too. If the characteristics, for which are these identifiers really, are missing, then the hardware is too old to reasonably modern operating systems. To represent the feature identifiers, inxi evaluates /proc/cpuinfo. The /proc directory reflects internal states of the running kernel.

In Syslinux Project version 4.06 the conditional move instructions are only be used in module com32 explicitly. In the execution flow in the assembler source code, checked by me, ISOLINUX has no CMOV instructions. But there is C source code too, which I can not say which OP code is compiled therefrom. Basically, CMOV instructions could come to pass, so that the execution on CPU AMD-K6 3D+ is not safe. To obtain an accurate statement, more effort would be necessary.

I had studied the ISOLINUX sources and am now of the opinion that the BIOS when attempting to read the second block (Extended Read Sectors From Drive, INT 13h, AH=42h), reports an error by setting the carry flag. This sign is then by ISOLINUX converted to the already known termination message 'aborted.'

There could exist two problems. First, the old BIOS is faulty, which is not uncommon. Socond, the CPU is not sufficiently equipped for modern software. But you could still use distributions, which are developed for old hardware and their loaders do not use the extensions in the BIOS (like: INT 13h, AH=42h).

Syslinux Project version 3.07 loads the kernel and Syslinux Project version 4.06 can not. Version 3.07 includes pretty sure no CMOV instructions, I have been looking for it. Version 4.06 has some in any case. What should we conclude?

[junqueman]

https://www.plop.at/en/bootmanager/plpbt.bin.html

It's very easy to make a Plop boot manager floppy disk.

Yes, I know. I made one a couple weeks ago, when you (or someone else?) first suggested it.

However, it doesn't behave very well on my HOH (that's Hopeless Old Hardware, in case you forgot ) The keyboard doesn't appear to be operative with it: I get the splash screen, but arrow keys, tab, space bar, any other key I can think of trying don't work to select between the menu choices. It remains stuck on "HDA Partition 1," and hitting "Enter" does nothing. The Plop boot manager CD behaves similarly, except at least hitting "Enter" on "HDA Partition 1" takes me to the next step.

The "All-in-One Boot Floppy" works fine, though.

[coyotl]

Well, I'm deeply impressed with the effort put into this 'HOH' here.
I had just such a machine until 2004 or 2005. Had lots of fun with it, mandrake 7.1, debian and possibly Libranet (best Linux ever) and windows 98
to play Halflife etc. After that it served as office pc for my sister a year or so. died at some point, long gone now. AT form factor so even cabinet no use.