[syslinux] The "core32" development branch

[Dag Wieers]

On Mon, 18 May 2009, Sebastian Herbszt wrote:

> H. Peter Anvin wrote:
>>  Sebastian Herbszt wrote:
>> >  H. Peter Anvin wrote:
>> > >  Sebastian Herbszt wrote:
>> > > > 
>> > > >  I am getting
>> > > > 
>> > > >  ld -m elf_i386  -T syslinux.ld -M -o ldlinux.elf ldlinux.o libcore.a
>> > > >  ../com32/lib/libcom32.a /usr/lib/gcc/i586-suse-linux/4.2.1/libgcc.a 
>> > > >  >
>> > > >  ldlinux.map
>> > > >  ld: ldlinux.o(.stab+0x0): Stabs entry has invalid string index.
>> > > >  ldlinux.o: could not read symbols: Bad value
>> > > >  make[1]: *** [ldlinux.elf] Error 1
>> > > 
>> > >  What version of NASM are you using?
>> > 
>> >  NASM version 0.98.39 compiled on Sep 21 2007
>>
>>  0.98.39 is *really* old by now (the current official release is 2.05.01,
>>  with 2.06 on the near horizon.)  It's quite plausible 0.98.39 produces
>>  bad stabs which makes the linker unhappy.
>>
>>  I have moved the debugging options to a separate NASMDEBUG variable.  I
>>  guess it would entail figuring out which versions of NASM actually work.
>>  One possibility, too, would be to specify only -F stabs, newer NASM has
>>  that as implies -g, whereas older ones don't.
>
> Works now with syslinux-3.81-pre3-31-gf482648. Thanks.
>
> I don't mind making core32 require NASM > 2.x or something more recent
> in case fixing the code to support older versions becomes painful.
> Newer openSUSE ships NASM 2.x, so it's not a problem for _me_ to upgrade.
> I am not sure about old RH / EL which Dag uses tho.

Well, a small look at the list below makes it quite obvious why I think 
NASM <= 2.x might be important.

Works currently on:

 	RHEL5:	nasm-0.98.39-3.2.2	end-of-life on 31/mar/2014
 	RHEL4:	nasm-0.98.38-3.EL4	end-of-life on 29/feb/2012
 	RHEL3:	nasm-0.98.35-3.EL3	end-of-life on 31/oct/2010
 	RH9:	nasm-0.98.35-3		end-of-life now

Fails already on:

 	RH7:	nasm-0.98.22-2		end-of-life now
 	RHEL2:	nasm-0.98-8.EL21	end-of-life on 31/may/2009

We use syslinux to generate new bootable ISO images on the fly with 
thousands of system's boot parameters hardcoded (not being able to use PXE 
can be hard :-))

Being able to use a new syslinux on normal Enterprise Linux servers is 
certainly a plus. I can only hope that RHEL6 will have a newer nasm by 
default, but RHEL5 will be around for some time before RHEL6 becomes an 
acceptable solution for a lot of companies.

In the end it's up to the project to determine its scope.

-- 
--   dag wieers,  dag at wieers.com,  http://dag.wieers.com/   --
[Any errors in spelling, tact or fact are transmission errors]