Re: [PATCH v2 2/3] GDB: Bring back the handling of DW_CC_program

[Tom Tromey <tom@tromey.com>]

>>>>> "Maciej" == Maciej W Rozycki <macro@embecosm.com> writes:

Maciej> Original code was removed with commit 6209cde4ddb8 ("Delete DWARF 
Maciej> psymtab code"), and then an update to complement commit 81873cc81eff 
Maciej> ("[gdb/symtab] Support DW_AT_main_subprogram with -readnow.") has also 
Maciej> been included here.

Sorry about that.

Maciej> NB this was discovered with and is covered by the test case included with 
Maciej> 3/3; it has been verified to trigger with GCC 4.4.7 and will likely also 
Maciej> with other configurations.

Maciej> +/* Return TRUE if the given DIE is the program's "main".  */
Maciej> +
Maciej> +static bool
Maciej> +dwarf2_func_is_main_p (struct die_info *die, struct dwarf2_cu *cu)
Maciej> +{
Maciej> +  if (dwarf2_flag_true_p (die, DW_AT_main_subprogram, cu))
Maciej> +    return true;
Maciej> +  struct attribute *attr = dwarf2_attr (die, DW_AT_calling_convention, cu);
Maciej> +  return (attr != nullptr
Maciej> +	  && attr->constant_value (DW_CC_normal) == DW_CC_program);

Should the DW_AT_calling_convention thing only apply to Fortran CUs?  I
have only ever heard of this approach being used by Fortran compilers.

Maciej> +	case DW_AT_calling_convention:
Maciej> +	  if (attr.constant_value (DW_CC_normal) == DW_CC_program)
Maciej> +	    *flags |= IS_MAIN;

Both the spots doing this check could probably use a comment explaining
why.  Maybe someday we'll get to remove this again.

Tom