[PATCH] D41438: [ELF] - Do not handle --defsym as regular linkerscript assignment.
Rafael Avila de Espindola via llvm-commits
llvm-commits at lists.llvm.org
Mon Dec 25 13:27:42 PST 2017
George Rimar <grimar at accesssoftek.com> writes:
>>I quite like the fact that defsym symbols have the same semantics as a
>>mini linker script.
>>
>>It looks like it is important for linker scripts to allow a symbol to
>>be redefined (see D27276).
>>
>>Why is it important for defsym to error?
>>
>>Thanks,
>>Rafael
>
> I thought you agreed with that behavior of -defsym described in my following comment
> is a bug of LLD and gold: https://reviews.llvm.org/D38239#958550.
Sorry, what I had in mind was producing an duplicated symbol error for
both cases, but I realize now we can't do that.
> If we choose defsym to have the same semantics as a mini linker script,
> it will be the gold style. We often selected bfd behavior when it was different from
> gold yearlier, though keeping defsym as mini linkerscript is indeed simpler.
>
> So, think we should leave it as is ?
I think so. I took a look at D38239 and I think it is possible to
implement that without this.
Cheers,
Rafael
More information about the llvm-commits
mailing list