[PATCH 2/3] Add check for 8-bit old registers in EVEX format

Jan Beulich jbeulich@suse.com
Tue May 21 12:24:45 GMT 2024


On 20.05.2024 08:22, Cui, Lili wrote:
> --- a/gas/config/tc-i386.c
> +++ b/gas/config/tc-i386.c
> @@ -4311,10 +4311,10 @@ static void establish_rex (void)
>        && !is_apx_rex2_encoding () && !is_any_vex_encoding (&i.tm) && !i.rex)
>      i.rex |= REX_OPCODE;
>  
> -  /* For REX/REX2 prefix instructions, we need to convert old registers
> +  /* For REX/REX2/EVEX prefix instructions, we need to convert old registers
>       (AL, CL, DL and BL) to new ones (AXL, CXL, DXL and BXL) and report bad
>       for AH, CH, DH and BH.  */
> -  if (i.rex || i.rex2)
> +  if (i.rex || i.rex2 || i.tm.opcode_modifier.evex)
>      {
>        for (unsigned int x = first; x <= last; x++)
>  	{
> @@ -4326,7 +4326,7 @@ static void establish_rex (void)
>  	      /* In case it is "hi" register, give up.  */
>  	      if (i.op[x].regs->reg_num > 3)
>  		as_bad (_("can't encode register '%s%s' in an "
> -			  "instruction requiring REX/REX2 prefix"),
> +			  "instruction requiring REX/REX2/EVEX prefix"),
>  			register_prefix, i.op[x].regs->reg_name);
>  
>  	      /* Otherwise it is equivalent to the extended register.
> --- a/gas/testsuite/gas/i386/rex-bad.l
> +++ b/gas/testsuite/gas/i386/rex-bad.l
> @@ -3,8 +3,8 @@
>  .*:5: Error: same .*
>  .*:6: Error: same .*
>  .*:7: Error: same .*
> -.*:9: Error: .* REX/REX2 .*
> -.*:10: Error: .* REX/REX2 .*
> -.*:12: Error: .* REX/REX2 .*
> -.*:13: Error: .* REX/REX2 .*
> +.*:9: Error: .* REX/REX2/EVEX .*
> +.*:10: Error: .* REX/REX2/EVEX .*
> +.*:12: Error: .* REX/REX2/EVEX .*
> +.*:13: Error: .* REX/REX2/EVEX .*

To avoid these diagnostics getting yet more diffuse, can you please switch
to

		as_bad (_("can't encode register '%s%s' in an "
			  "instruction requiring %s prefix"),
			register_prefix, i.op[x].regs->reg_name,
			i.tm.opcode_modifier.evex ? "EVEX" : "REX/REX2");

Okay with that (or a substantially similar; I didn't test the above, after
all) adjustment.

Jan


More information about the Binutils mailing list