This is the mail archive of the binutils@sources.redhat.com mailing list for the binutils project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: check mips abi x linker emulation compatibility


At 08 May 2003 15:19:39 -0300, Alexandre Oliva wrote:
> > There _isn't_ really a "neutral" ABI, though, is there?
> 
> What if none of the ABI bits are set?

The binary's ABI still isn't neutral; it's one of, and compatible with
only one of, the other ABIs.  we just don't know which one, so we
should allow it to be *linked* with any other.

IMO, though, it's not particularly desirable for the assembler to
*ever* create (elf32) binaries w/o ABI markings, these days.

(it may be reasonable to have a 'no (known) abi' used explicitly for
things like blobs turned into .o files by objdump or maybe even ld.)


> Except for the fact that we have to make a choice between elf32 or
> elf64 early on, it can remain neutral for longer...

well, the bits used in elf32 for ABI flags are currently not used at
all in elf64, and elf64 binaries are currently *always* n64, right?


> > It sounds like what you're asking for is a "no-abi-marking" flag.
> 
> Nope.  I want something that disables the assembler's default, so that
> it can figure it out as if the target configured for was a generic
> mips target.

even 'generic' MIPS targets assemblers might use some ABIs.  they
might (erroneously, IMO) not mark them on the binaries, though.  8-)


chris


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]