Re: [PATCH 05/10] Add handler for more i386_cpu_flags

[Jan Beulich <jbeulich@suse.com>]

On 14.10.2022 11:12, Haochen Jiang wrote:
> From: Kong Lingling <lingling.kong@intel.com>
> 
> gas/ChangeLog:
> 
> 	* config/tc-i386.c (cpu_flags_all_zero): Add new ARRAY_SIZE handle.
> 	(cpu_flags_equal): Ditto.
> 	(cpu_flags_and): Ditto.
> 	(cpu_flags_or): Ditto.
> 	(cpu_flags_and_not): Ditto.
> ---
>  gas/config/tc-i386.c | 17 +++++++++++++++++
>  1 file changed, 17 insertions(+)
> 
> diff --git a/gas/config/tc-i386.c b/gas/config/tc-i386.c
> index 0d161d60ef..7045e18cff 100644
> --- a/gas/config/tc-i386.c
> +++ b/gas/config/tc-i386.c
> @@ -1618,6 +1618,10 @@ cpu_flags_all_zero (const union i386_cpu_flags *x)
>  {
>    switch (ARRAY_SIZE(x->array))
>      {
> +    case 5:
> +      if (x->array[4])
> +	return 0;
> +      /* Fall through.  */
>      case 4:
>        if (x->array[3])
>  	return 0;
> @@ -1643,6 +1647,10 @@ cpu_flags_equal (const union i386_cpu_flags *x,
>  {
>    switch (ARRAY_SIZE(x->array))
>      {
> +    case 5:
> +      if (x->array[4] != y->array[4])
> +	return 0;
> +      /* Fall through.  */
>      case 4:
>        if (x->array[3] != y->array[3])
>  	return 0;
> @@ -1675,6 +1683,9 @@ cpu_flags_and (i386_cpu_flags x, i386_cpu_flags y)
>  {
>    switch (ARRAY_SIZE (x.array))
>      {
> +    case 5:
> +      x.array [4] &= y.array [4];
> +      /* Fall through.  */
>      case 4:
>        x.array [3] &= y.array [3];
>        /* Fall through.  */
> @@ -1698,6 +1709,9 @@ cpu_flags_or (i386_cpu_flags x, i386_cpu_flags y)
>  {
>    switch (ARRAY_SIZE (x.array))
>      {
> +    case 5:
> +      x.array [4] |= y.array [4];
> +      /* Fall through.  */
>      case 4:
>        x.array [3] |= y.array [3];
>        /* Fall through.  */
> @@ -1721,6 +1735,9 @@ cpu_flags_and_not (i386_cpu_flags x, i386_cpu_flags y)
>  {
>    switch (ARRAY_SIZE (x.array))
>      {
> +    case 5:
> +      x.array [4] &= ~y.array [4];
> +      /* Fall through.  */
>      case 4:
>        x.array [3] &= ~y.array [3];
>        /* Fall through.  */

Not a request for action on your part, but it is high time for us to sort
this requirement of widening the bitmap by one bit for each and every new
ISA extension. I have some vague plan, but I need to find time to actually
try it out.

Jan