+2016-09-21 Richard Sandiford <richard.sandiford@arm.com>
+
+ * config/tc-aarch64.c (parse_aarch64_imm_float): Remove range check.
+ (parse_operands): Check the range of 8-bit FP immediates here instead.
+
2016-09-21 Richard Sandiford <richard.sandiford@arm.com>
* config/tc-aarch64.c (parse_aarch64_imm_float): Report a specific
hexadecimal representation is involved). REG_TYPE says which register
names should be treated as registers rather than as symbolic immediates.
- N.B. 0.0 is accepted by this function. */
+ This routine accepts any IEEE float; it is up to the callers to reject
+ invalid ones. */
static bfd_boolean
parse_aarch64_imm_float (char **ccp, int *immed, bfd_boolean dp_p,
}
}
- if (aarch64_imm_float_p (fpword) || fpword == 0)
- {
- *immed = fpword;
- *ccp = str;
- return TRUE;
- }
+ *immed = fpword;
+ *ccp = str;
+ return TRUE;
invalid_fp:
set_fatal_syntax_error (_("invalid floating-point constant"));
= (aarch64_get_qualifier_esize (inst.base.operands[0].qualifier)
== 8);
if (!parse_aarch64_imm_float (&str, &qfloat, dp_p, imm_reg_type)
- || qfloat == 0)
+ || !aarch64_imm_float_p (qfloat))
{
if (!error_p ())
set_fatal_syntax_error (_("invalid floating-point"