[gambit-list] [RESOLVED] Failed to build Gambit v4.6.0 on i586 (builds fine on x86-64)

Paul Wolneykien manowar at altlinux.org
Fri May 6 07:41:18 EDT 2011


В Птн, 06/05/2011 в 01:20 +0400, Paul Wolneykien пишет:
> В Чтв, 05/05/2011 в 18:54 +0400, Paul Wolneykien пишет: 
> > Hi,
> > 
> > Wed May 04 2011 14:55 +0400, Paul Wolneykien wrote:
> > > Still get SEGFAULT trying to compile some code with the resulting
> > > build of gsc.
> > 
> >   I've fixed the problem: v4.6.0 builds fine from the following commit:
> 
>   Sorry, I was wrong — it doesn't. So I checked the differences between
> the *.c files generated by the bootstrap compiler with the packaged *.c
> files in the official Gambit *.tgz release package (diffs attached).
>   The main difference is the use of "IFD" instead of "OFD" in many
> places (may be everywhere). And there is no such difference when
> building on x86-64.
>   So I think the problem really exists and it is caused by the frame
> size calculations (gsc/_t-c-*.scm): where the official release *.c files
> contain "OFD" declarations, the *.c files generated from *.scm files on
> i586-machine contain "IFD" declarations instead. The same process when
> run on x86-64 machine produces *.c with "OFD" declarations nearly
> identical to the official release *.c files.

  I've resolved the problem applying the patch «Fix compiler bug which
caused out-of-line internal continuation frames to have incorrect frame
descriptors» (02c1427d40eb080a4ada916d55ecbb14cc191772) to the bootstrap
tree too. The original bootstrap tree marked with [COMPILER CHANGES
NEEDED FOR v4.6.0] produces compiler with the denoted bug so the release
compiler being built with that bootstrap is defective. With that patch
applied the bootstrap tree produces normal bootstrap compiler.


    Best regards,

      Paul.

> 
> 
>   Regards,
> 
>     Paul.
> 
> > 
> > ---
> > commit 02c1427d40eb080a4ada916d55ecbb14cc191772
> > Author: Marc Feeley <feeley at iro.umontreal.ca>
> > Date:   Thu Jan 14 23:06:37 2010 -0500
> > 
> >     Fix compiler bug which caused out-of-line internal continuation
> > frames to have incorrect frame descriptors.
> > ---
> > 
> >   Previously, I tried to build from the commit marked [RUNTIME CHANGES
> > NEEDED FOR v4.6.0] and got a segfault on test number 5.
> > 
> >   Marc, may I ask you to tag stable releases of Gambit in the git
> > repository?
> > 
> > 
> >     TIA,
> > 
> >       Paul.
> > 
> 
> 
> _______________________________________________
> Gambit-list mailing list
> Gambit-list at iro.umontreal.ca
> https://webmail.iro.umontreal.ca/mailman/listinfo/gambit-list





More information about the Gambit-list mailing list