simon burge writes:

>What confused me was this from magdir/elf:
>
>>>18    leshort         8               MIPS R3000_BE - invalid byte order,
>>>18    leshort         9               Amdahl - invalid byte order,
>>>18    leshort         10              MIPS R3000_LE,
>
>and NetBSD/pmax ELF files have 8 at offset 18.  I dunno if changing
>"MIPS R3000_BE - invalid byte order," to "MIPS R3000_LE," would break
>something else (SGIish?).
>
>Would you like a PR for this?

ISTR Christos looked at this and came to the conclusion that
BFD/gas/gld was not using the correct value for mipsel ELF binaries,
but instead using teh byteswapped version of the SGI mipseb value.
Does anyone have official ELF docs to check? (Jason?)

Again, ISRT talking with Ian Taylor about this, and his opinion was
that the gas/gld behaviour was what he wanted to stick with
(possibly for internal BFD reasons, but I forget the details).

I could be completely misremembering, but if the byteswapping
diagnosis is accurate, it may be worth double-checking with Ian.