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: BFD_ASSEMBLER, bignum questions


Ian Lance Taylor wrote:

Stan Shebs <shebs@apple.com> writes:


A couple random GAS questions that the current sources are not
answering for me:

* Are there any non-BFD configurations still in use? It looks like
configure can generate some, but they mostly seem really obscure,
maybe they're no longer in use.


Do you mean non-BFD, or non-BFD_ASSEMBLER?


And here I thought they were the same thing...


BFD_ASSEMBLER is always used for alpha, arm, i386, ia64, mips, ns32k, pdp11, ppc, and sparc. It is always used for elf, ecoff, and som. What's left is a29k, h8300, h8500, i960, m68k, m88k, ns32k, or32, sh, tahoe, tic80, vax, w65, z8k for COFF and aout targets.

BFD is used for all COFF targets, even those which do not use
BFD_ASSEMBLER (such targets define BFD_HEADERS).  There are some aout
targets which do not use BFD at all.  I do not know whether any of
these are in common use.


Hmm, they mostly seem ancient. Has there been any program of
obsoleting unused configs, or do people actually test that all
of these still work? Non-BFD would seem hard to make work as a
cross for instance.

Every time I see #ifdef BFD_ASSEMBLER, I get a gigantic urge
to drop everything else and hack it out.

* Is the bignum code actually needed? 64-bit values seem to depend
on 64-bit BFD and do 64-bit arithmetic, so I don't see how bignum
code would ever need to kick in.


The bignum code is not limited to 64-bit values. For example, it is used to support the .octa pseudo-op.

Ah.

Stan


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