[nos-bbs] Early note on RedHat FC-5

[George (Skip) VerDuin]

I find this failure fascinating...
Clearly my FC-5 platform will not permit jnos compilation because of
infrastructure change over FC-4 - glibc.
The failure is central to the kernel context switch design by Phil KA9Q
and Rob PE1CHL on multiple platforms - mine Linux.
Present jnos2 is to support DOS, MAC, Linux, (SUN?), all requiring
adjustment for library support details.
Future support for Windows (without DOS constraints) may introduce
another variation?

Those of you who modify the source are truly gifted with the ability to
juggle.  
Thankfully pre-compiled modules are (more?) tolerant of variation...
I can hope the glibc design change only requires a SMALL modification to
jnos source - I'm sad I can't do it myself.
Yet I find __GNU_LIBRARY__ is deprecated in favor of __GLIBC__ as an
example, and that's only one of many.
Certainly the desire to run jnos on top of DOS the old with the same
source as Linux the new causes stress.

Those of us who apply the code can not demand that jnos work on just any
platform, we must fit within constraints.
As for myself, I'll be rolling jnos compilation back to another older
platform for the time being (after installing gcc).
That will be my approach to remaining on the leading (not bleeding) edge
of jnos development.
In addition - It makes little sense to fill up this reflector talking to
myself...

On Mon, 2006-05-01 at 19:04 -0400, George (Skip) VerDuin wrote:

> OK - I am over my head and can not repair the FC-5 compile failure...
> It occurs only when ifdefined linux and __GNU_LIBRARY__ > 1 
> 
> I have found FC-5 did change to gcc 4.1.0-3.i386 and (buggy) glibc
> 2.4-4.i686.
> A design change intentionally no longer defines JB_SP etc.
> It sounds like setjump & jmpbuf are modified (may be deprecated?) and
> need to be replaced with __builtin_frame_address (etc..) for security?
> I looked around gcc documentation for a while and did not land on the
> answer.
> This smells like a non-trivial change for linux platform memory
> management some time?
> See:
> http://www.redhat.com/archives/fedora-devel-list/2006-April/msg00571.html
> 
> On the memory subject - jnos log reports that jnos has 3.05GB
> available memory...
> Can that be intended?
> It could be right if jnos augments swap space by also using free space
> on the root (maybe /tmp?).
> Come on recursive qsort - I dare you to try to use up 3gb...! (of
> course I won't wait all night...)
> Thrashing comes to mind.
> 
> In any case, Maiko thanks for putting FC-5 on your fix-it list.
> I'll be OK with pre-compiled executables for quite a while since it
> has sufficient features for my needs as distributed.
> 
> On Wed, 2006-03-29 at 09:22 -0500, George (Skip) VerDuin wrote:
> 
> > Greetings,
> > 
> > This is my recent experience that seems worth sharing at an early
> > stage...
> > The jnos executeable from FC-4 seems normal in operation over the
> > past 24hrs.
> > The jnos compile "make" fails in ksubr.c over "JB_SP", "JB_BP",
> > "JB_PC",  undeclared.
> > 
> > Sorry - no solution at this moment.
> > I might suggest to avoid FC-5 development until the differences
> > become clear?
> > Something in "#define ... JB_SP ..." has changed in FC-5 over FC-4.
> > I will follow up...
> > 
> > 73
> > de Skip k8rra k
> > 
> > 
> > 
> > 
> > 
> > _______________________________________________
> > nos-bbs mailing list
> > nos-bbs at lists.tapr.org
> > https://lists.tapr.org/cgi-bin/mailman/listinfo/nos-bbs
> 
> 
> 73
> de Skip k8rra k
> 
> 
> 
> 
> _______________________________________________
> nos-bbs mailing list
> nos-bbs at lists.tapr.org
> https://lists.tapr.org/cgi-bin/mailman/listinfo/nos-bbs


73
de Skip k8rra k


-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.tapr.org/pipermail/nos-bbs/attachments/20060505/1ef6ac62/attachment.htm