Re: Linking to Gt.m object files

[Mike Dupont]

Just a note,
there is noone who says that we should be able to link against this
code, I am just learning about the gt.m system
maybe they patch the files up before loading them.

Bug report here :
http://sourceware.org/bugzilla/show_bug.cgi?id=13730

I tested this on the latest compiler on the build farm gcc10
GNU ld (GNU Binutils) 2.22.51.20120122

address@hidden:~/test$ /home/schwab/src/binutils/bin/bin/ld
attachment.cgi\?id\=6238
/home/schwab/src/binutils/bin/bin/ld: i386 architecture of input file
`attachment.cgi?id=6238' is incompatible with i386:x86-64 output
/home/schwab/src/binutils/bin/bin/ld: warning: cannot find entry
symbol _start; defaulting to 00000000004000b0
attachment.cgi?id=6238:attachment.cgi?id=6238:(.text+0x67): undefined
reference to `_Serenji.SHELL'
attachment.cgi?id=6238:attachment.cgi?id=6238:(.text+0x6c): undefined
reference to `_Serenji'
/home/schwab/src/binutils/bin/bin/ld: BFD (GNU Binutils)
2.22.51.20120122 internal error, aborting at
../../binutils/bfd/reloc.c line 6394 in
bfd_generic_get_relocated_section_contents

/home/schwab/src/binutils/bin/bin/ld: Please report this bug.



On Thu, Feb 23, 2012 at 5:41 PM, nick clifton <address@hidden> wrote:
> Hi Mike,
>
>>  but the linker fails.
>
>
> Please could you open a bug report about this at:
>
>  http://sourceware.org/bugzilla/
>
> Including the testcase will be really helpful.  Itwill also be helpful to
> know that command line that g++ used to invoke the linker, as well as the
> target architecture for which this particular linker was configured.  (I am
> currently assuming i386-pc-go32).
>
>
>
>> /usr/bin/ld.bfd.real: BFD (GNU Binutils for Ubuntu) 2.21.53.20110810
>
>
> Note: 2.21... is an old binutils release.  We are now on version 2.22, and
> the development sources are in advance of that.  Not that it matters in this
> particular case as the problem appears to be reproducible with a linker
> built from today's sources.  But I just thought that I should mention it, as
> the first thing anyone else looking at this bug is likely to try is using
> the latest linker.
>
>
>
>> internal error, aborting at ../../bfd/reloc.c line 6281 in
>> bfd_generic_get_relocated_section_contents
>
>
> The problem in general is that the generic part of the linker is receiving
> an out-of-range error from that target specific backend when it is not
> expecting one.  It is relatively easy to add some code to produce a warning
> message when this happens (rather than aborting), but I am a little bit
> concerned as to why the outofrange error is happening in the first place.
>  If you fix the references to the undefined symbols (_Serenji and
> _Serenji.SHELL) does the abort go away ?
>
> Cheers
>  Nick



-- 
James Michael DuPont
Member of Free Libre Open Source Software Kosova http://flossk.org