[PATCH 2/2] asm inline

Richard Sandiford richard.sandiford@arm.com
Tue Dec 4 15:31:00 GMT 2018


Segher Boessenkool <segher@kernel.crashing.org> writes:
> diff --git a/gcc/c/c-typeck.c b/gcc/c/c-typeck.c
> index 81c520a..9b572d7 100644
> --- a/gcc/c/c-typeck.c
> +++ b/gcc/c/c-typeck.c
> @@ -10315,7 +10315,7 @@ build_asm_stmt (tree cv_qualifier, tree args)
>     are subtly different.  We use a ASM_EXPR node to represent this.  */
>  tree
>  build_asm_expr (location_t loc, tree string, tree outputs, tree inputs,
> -		tree clobbers, tree labels, bool simple)
> +		tree clobbers, tree labels, bool simple, bool is_inline)
>  {
>    tree tail;
>    tree args;

Function comment doesn't document the new parameter.

> diff --git a/gcc/cp/semantics.c b/gcc/cp/semantics.c
> index 733c42f..87d54b1 100644
> --- a/gcc/cp/semantics.c
> +++ b/gcc/cp/semantics.c
> @@ -1486,7 +1486,7 @@ finish_compound_stmt (tree stmt)
>  
>  tree
>  finish_asm_stmt (int volatile_p, tree string, tree output_operands,
> -		 tree input_operands, tree clobbers, tree labels)
> +		 tree input_operands, tree clobbers, tree labels, bool inline_p)
>  {
>    tree r;
>    tree t;

Same here.

> @@ -19608,6 +19610,17 @@ cp_parser_asm_definition (cp_parser* parser)
>  	  else
>  	    done = true;
>  	  break;
> +	case RID_INLINE:
> +	  if (!inline_p && parser->in_function_body)
> +	    {
> +	      /* Remember that we saw the `inline' keyword.  */
> +	      inline_p = true;
> +	      /* Consume the token.  */
> +	      cp_lexer_consume_token (parser->lexer);
> +	    }
> +	  else
> +	    done = true;
> +	  break;
>  	case RID_GOTO:
>  	  if (!goto_p && parser->in_function_body)
>  	    {

Hmm, so we allow top-level "asm volatile" in C++ but not C?

Probably worth having tests to show that we (intentionally) don't
allow top-level "asm inline".

> @@ -1640,6 +1640,7 @@ finish_asm_stmt (int volatile_p, tree string, tree output_operands,
>  		  output_operands, input_operands,
>  		  clobbers, labels);
>    ASM_VOLATILE_P (r) = volatile_p || noutputs == 0;
> +  ASM_INLINE_P (r) = inline_p;
>    r = maybe_cleanup_point_expr_void (r);
>    return add_stmt (r);
>  }
> diff --git a/gcc/doc/extend.texi b/gcc/doc/extend.texi
> index 2791f25..cebbfc7 100644
> --- a/gcc/doc/extend.texi
> +++ b/gcc/doc/extend.texi
> @@ -8382,6 +8382,10 @@ various @option{-std} options, use @code{__asm__} instead of
>  @item volatile
>  The optional @code{volatile} qualifier has no effect. 
>  All basic @code{asm} blocks are implicitly volatile.
> +
> +@item inline
> +If you use the @code{inline} qualifier, then for inlining purposes the size
> +of the asm is taken as the smallest size possible (@pxref{Size of an asm}).
>  @end table
>  
>  @subsubheading Parameters

You need to update the syntax above too, which currently reads:

@example
asm @r{[} volatile @r{]} ( @var{AssemblerInstructions} )
@end example

Same for the equivalent extended asm docs.

> @@ -3932,6 +3933,27 @@ gimple_asm_set_volatile (gasm *asm_stmt, bool volatile_p)
>  }
>  
>  
> +/* Return true ASM_STMT ASM_STMT is an asm statement marked inline.  */
> +
> +static inline bool
> +gimple_asm_inline_p (const gasm *asm_stmt)
> +{
> +  return (asm_stmt->subcode & GF_ASM_INLINE) != 0;
> +}

Return true if ASM_STMT is ...

(Or "Return true if asm statement ASM_STMT is marked inline", since gasm
forces it to be an asm statement.)

> diff --git a/gcc/tree-inline.c b/gcc/tree-inline.c
> index 5aa782b..7e9ed99 100644
> --- a/gcc/tree-inline.c
> +++ b/gcc/tree-inline.c
> @@ -4109,6 +4109,9 @@ estimate_num_insns (gimple *stmt, eni_weights *weights)
>  	   with very long asm statements.  */
>  	if (count > 1000)
>  	  count = 1000;
> +	/* If this asm is asm inline, count anything as minimum size.  */
> +	if (gimple_asm_inline_p (as_a <gasm *> (stmt)))
> +	  count = !!count;

FWIW, since Marc found it confusing too... I think MIN (count, 1) would
show the intent more clearerly.  But that's just personal preference,
no need to change.

OK with those changes, thanks.

Richard



More information about the Gcc-patches mailing list