This is the mail archive of the
binutils@sourceware.org
mailing list for the binutils project.
Re: [patch] Add fido support (Part 5/n)
IMHO it would be better to fix the opcode table so that the
unsupported
insns don't match in the first place.
Hmm. That may make m68k-opc.c a bit ugly. Each CPU core is
defined as an OR of one or more flags like m68000, cpu32, fido_a,
etc in m68k_archs in tc-m68k.c. If we were to reject tbl
instructions with the opcode table, I would have to treat fido as a
new core. That is, I would have to define fido as fido_a, not
cpu32|fido_a. In addition, I have to put fido_a everywhere we see
cpu32 in m68k-opc.c. Normally, we don't have to worry about this
problem because because people keep adding new instructions and do
not remove instructions. I guess the m68k world is different.
Neither CPU32 or ColdFire is a superset of m68000. Fido is not a
strict superset of CPU32 either.
Thoughts?
We had this problem in mips as well my preferred way there was the
like Andreas describes since it made it easier to look at the table
and understand just what each processor had, but the m68k maintainers
may like your style instead.
-eric