Syslinux - Oct 2017 - ldlinux.elf: Not enough room for program headers (Was: Re: Patches from Debian)

Hi,

On Mon, 23 Oct 2017 17:19:43 +0000
Ady Ady via Syslinux <syslinux at zytor.com> wrote:
> When you say "new syslinux version from git", do you mean the git
> head in anonscm.debian.org/git... (Debian's source code for
> Syslinux-related packages), or rather the upstream Syslinux git head
> (in repo.or.cz... or in git.zytor.com...)?
Sorry for the confusion: I mean the syslinux git head from
git.zytor.com.
> Either way, your test's results would suggest that the problem might
> be avoided when/if:
> 
> _ updating the gnu-efi submodule in Syslinux;
It fails in the `make bios` part, which does not use gnu-efi at all.
> _ changing the sort (order) within Makefiles so as to make the build 
> more reproducible.
Both test builds had Debian's patches applied, so that's not it either.
> If I'm not mistaken, these 2 are the relevant differences between the 
> code that triggers the FTBFS and the code that doesn't.
I've bisected the issue (building on Debian "testing" but without
any
patches applied).  It is fixed by the following commit:
http://git.zytor.com/syslinux/syslinux.git/commit/?id=ff859050fa4e6535cae098dc35d88a265466448d

Regards
Lukas

> Hi,
> 
> On Mon, 23 Oct 2017 17:19:43 +0000
> Ady Ady via Syslinux <syslinux at zytor.com> wrote:
> 
> > When you say "new syslinux version from git", do you mean
the git
> > head in anonscm.debian.org/git... (Debian's source code for
> > Syslinux-related packages), or rather the upstream Syslinux git head
> > (in repo.or.cz... or in git.zytor.com...)?
> 
> Sorry for the confusion: I mean the syslinux git head from
> git.zytor.com.
> 
> > Either way, your test's results would suggest that the problem
might
> > be avoided when/if:
> > 
> > _ updating the gnu-efi submodule in Syslinux;
> 
> It fails in the `make bios` part, which does not use gnu-efi at all.
> 
> > _ changing the sort (order) within Makefiles so as to make the build 
> > more reproducible.
> 
> Both test builds had Debian's patches applied, so that's not it
either.
 
 
Let's not forget that other similar changes to Makefiles were performed 
upstream after 6.03, so part of the changes that are now performed in 
Debian's Syslinux 6.03 are already in the upstream git head.
 
 
> 
> > If I'm not mistaken, these 2 are the relevant differences between
the
> > code that triggers the FTBFS and the code that doesn't.
> 
> I've bisected the issue (building on Debian "testing" but
without any
> patches applied).  It is fixed by the following commit:
>
http://git.zytor.com/syslinux/syslinux.git/commit/?id=ff859050fa4e6535cae098dc35d88a265466448d
 
 
Reminder: partial revert of that commit (already in 6.04-pre1):

 
repo.or.cz/syslinux.git/commit/8079602cfd8b9e52091a900ce500a2a2e4c90f00 


Quoting Peter (hpa) in a related discussion about those 2 commits:
 "OK, so something is still an issue."
 bugzilla.syslinux.org/show_bug.cgi?id=71#c16 

That "pending issue" was not resolved, even though the bug report was 
closed (because the OP got the conflict, in his specific 
OS/environment, resolved by those commits).
 
 
> 
> Regards
> Lukas
Thank you for the tests and feedback.

Regards,
Ady.

On Tue, 24 Oct 2017 00:58:46 +0000
Ady Ady via Syslinux <syslinux at zytor.com>
wrote:
> > I've bisected the issue (building on Debian "testing"
but without
> > any patches applied).  It is fixed by the following commit:
> >
http://git.zytor.com/syslinux/syslinux.git/commit/?id=ff859050fa4e6535cae098dc35d88a265466448d
>  
>  
> Reminder: partial revert of that commit (already in 6.04-pre1):
> repo.or.cz/syslinux.git/commit/8079602cfd8b9e52091a900ce500a2a2e4c90f00 
I had noticed that and explicitly checked that commit:  It does not
re-introduce the problem.

Regards
Lukas