How do YOU (TC team) build core.gz?

[theYinYeti]

Dear TC team,
I have this unresolved issue about a custom “core.gz” having trouble with multiple processors:
http://forum.tinycorelinux.net/index.php/topic,15591.0.html
I intend to get to the _core_ of the problem! 

So, I would like to know exactly (or have pointers to) how the TC team is building packaging their core.gz, as this one seems to have much less trouble than user-built ones.
Here are some random questions that come to mind:
— Do you build on a machine with one or more processors?
— Which tools are from GNU and which from Busybox in the build process?
— On what version of what distribution do you build core.gz?
— Which commands do you use?
— Is the host machine where core.gz is built 32 or 64 bits? And the OS?
…

[Juanito]

— Do you build on a machine with one or more processors?

more than one processor

— Which tools are from GNU and which from Busybox in the build process?

compiletc and the balance from busybox

— On what version of what distribution do you build core.gz?

The most recent version of tinycore

— Is the host machine where core.gz is built 32 or 64 bits? And the OS?

64bit, most recent version of tinycore

Keep in mind that core.gz is a collection of tools compiled separately

[theYinYeti]

Thank you Juanito. That does answer some questions

Your last remark makes me realize I should be more specific in my query: I’m actually interested in the packaging of core.gz, not the actual build of the included programs.
My problem, and that of many other TC users, is that for some reason, we take a core.gz, explode it into a directory following the Wiki instructions, then repackage it (cpio→gzip→advdef) without any change in any binary (only scripts, if even that in case of a test), and the resulting core-new.gz fails with this error:
http://yeti.selfip.net/public/www/TC-new+syslinux.png
(Fixing recursive fault but reboot is needed!)

I’ll change the first post to clarify a bit.

[coreplayer2]

My problem, and that of many other TC users, is that for some reason, we take a core.gz, explode it into a directory following the Wiki instructions, then repackage it (cpio→gzip→advdef) without any change in any binary (only scripts, if even that in case of a test), and the resulting core-new.gz fails with this error:
http://yeti.selfip.net/public/www/TC-new+syslinux.png
(Fixing recursive fault but reboot is needed!)

I must say,  while i've experienced this error once in the distant past and iirc it was not because of my repackaging core.gz.

I would like for you to test this script on your hardware to check if the issue repeats itself.
http://forum.tinycorelinux.net/index.php/topic,15493.0.html

remember if you change anything in the archive during the test you're not testing the process of unpackaging or repackaging core, which void's the test, right?

but I'm confused here, because this thread begins with multiprocessor questions which I am curious about, then morphs into another issue?

 

[theYinYeti]

Hey coreplayer2!
I tried your script this morning and all worked smoothly (although I could not test the part with kernel modules because the computer I have access to right now does not have the specs for compiling a kernel, nor do I have the time). That said, your script basically does the same thing mine does.

However, I tried to boot on this computer the core-new.gz attempts that failed on the computer I used before: all boot attempts worked! The former PC was your standard core-duo- or amd-based 4-core tower PC, whereas the computer I use now is a low-power Shuttle PC…

That will certainly complicate things, because it will be months, if not a year, before I can access the PC where the tests failed again
In a week or so, I should be able to use a computer more similar to the PC where the boot attempts failed; I’ll update the topic then.

As for the number of processors/cores, I mentioned it because the problem I experienced before disappears when boot is forced in mono-processor mode, either with maxcpus=0, or with ioapic=0 iolapic=0.