Despite that the RISC-V ISA Manual version 2.2 prohibited "RV32EF", later
versions beginning with the version
20190608-Base-Ratified removed this
restriction. Because the 'E' extension is still a draft, the author chose
to *just* remove the conflict (not checking the ISA version).
Note that, because RV32E is only used with a soft-float calling convention,
there's no valid official ABI for RV32EF. It means, even if we can assemble
a program with -march=rv32ef -mabi=ilp32e, floating-point registers are kept
in an unmanaged state (outside ABI management).
The purpose of this commit is to suppress unnecessary errors while parsing
an ISA string and/or disassembling, not to allow hard-float with RVE.
bfd/ChangeLog:
* elfxx-riscv.c (riscv_parse_check_conflicts): Accept RV32EF
because only older specifications disallowed it.
gas/ChangeLog:
* testsuite/gas/riscv/march-fail-rv32ef.d: Remove as not directly
prohibited.
* testsuite/gas/riscv/march-fail-rv32ef.l: Likewise.
rps->error_handler (_("rv%d does not support the `q' extension"), xlen);
no_conflict = false;
}
- if (riscv_lookup_subset (rps->subset_list, "e", &subset)
- && riscv_lookup_subset (rps->subset_list, "f", &subset))
- {
- rps->error_handler
- (_("rv32e does not support the `f' extension"));
- no_conflict = false;
- }
if (riscv_lookup_subset (rps->subset_list, "zfinx", &subset)
&& riscv_lookup_subset (rps->subset_list, "f", &subset))
{
+++ /dev/null
-#as: -march=rv32ef
-#source: empty.s
-#error_output: march-fail-rv32ef.l
+++ /dev/null
-.*Assembler messages:
-.*Error: .*rv32e does not support the `f' extension