[PATCH 0/3] x86: improve handling of equates to registers

[Jan Beulich]

Fuzzed input led to Alan addressing PR gas/28977 in, I think, a
sub-optimal way. An alternative is proposed in patch 1 alongside
the intention to then revert the original workaround.

In the course of discussing things with Alan he pointed out another
problematic construct, which patch 2 makes work. Variants of that
involving forward references, which are e.g. not resolvable at the
time an insn using it is being processed, would cause fatal and/or
internal errors, which patch 3 "softens" to an ordinary error.

1: don't attempt to resolve equates and alike from i386_parse_name()
2: improve resolution of register equates
3: reject relocations involving registers

Jan