Mail Archives: djgpp/1998/08/25/03:09:04
On Mon, 24 Aug 1998, Troy VanHorn wrote:
> libraries built with gcc-2.8.0 with gcc-2.8.1 (warning: there will be
> similar problems with upcomming egcs-1.1)
>
>
> Andris
>
>
> ============================================================================
>
> Speaking of egcs, (or gcc for that matter), is there any chance that a
> future release of gcc or egcs for djgpp will be properly configured to
> produce local lables that will really be local? As it is now with gcc
> 2.8.1, if I compile some C++ with exceptions enabled, the resulting object
> files are really bloated, and it doesn't need to be that way. I reallize I
> could probably recompile GAS to support the old method of local labels which
> gcc now produces, but it would be better if the compiler itself was properly
> configured.
The problem really is with binutils-2.8.1:
AS.EXE for DJGPP should discard symbols beginning with 'L' (local
labels) and not place info in object file unless explicitly asked
to keep them. Unfortunatelly current official release of
binutils-2.8.1 for DJGPP expects .L for local labels and as the
result info about local labels goes into object files.
There were discussion about this topic some months ago.
I myself am using patched version of binutils-2.8.1 that doesn't have this
problem (thanks Andrew Crabtree for patch, I had to build from bnu281s.zip
which was not very easy task)
I think it should be fixed in DJGPP port of binutils-2.9.1
>
> I've also considered writing a program which would "fix" the output of
> gcc before being assembled, but I don't know if there is any way for it to
> be automatically inserted between the compiler and assembly stages of
> gcc.exe.
>
Andris
- Raw text -