OK, changing the optimize to -g made it work, so I am going to leave
it that way -- not going to try -O1 because it takes so long -- its
still going, but at least it got past our problem.
Thanks.
on 10/12/2004 Nicholas Clark via RT([EMAIL PROTECTED]) wrote
> On Tue, Oct 12, 2004 at 05:25:46PM -0000, covici @ ccs. covici. com wrote:
>
> > osname=linux, osvers=2.4.19-rmk6-ads3, archname=armv4l-linux
>
> > cc='gcc', ccflags ='-fno-strict-aliasing -D_LARGEFILE_SOURCE
> > -D_FILE_OFFSET_BITS=64',
> > optimize='-O2',
>
> > ./miniperl -w -Ilib -MExporter -e '<?>' || make minitest
> > Attempt to free unreferenced scalar: SV 0xeeeec.
> > make: [extra.pods] Error 1 (ignored)
> > ./miniperl -Ilib configpm configpm.tmp
> > make: *** [lib/Config.pm] Segmentation fault
>
> Well, clearly it shouldn't do that.
>
> I've built 5.8.4 on 2.4.18-ish on ARM no problem. I suspect a gcc bug.
> Can you try building with less optimisation, say -O1 ?
>
> Either
>
> edit config.sh, replacing the line
>
> optimize='-O2'
>
> with
>
> optimize='-O1'
>
> then run
>
> ./Configure -S
>
> then re-run
>
> make
>
> which should cause a re-compile of everything. Or alternatively, and a bit
> slower, restart with a clean source tree and run the Configure script with
> -Doptimize=-O1 added to its arguments, then re-run make
>
>
> I hope that this is sufficient to get past the segfault and build all the
> extensions. All tests should pass.
>
> Nicholas Clark
>
--
John Covici
[EMAIL PROTECTED]
