Re: [PATCH 2/2] gdb: Resolve dynamic target types of pointers.

[Tom Tromey <tom@tromey.com>]

>>>>> Nils-Christian Kempke via Gdb-patches <gdb-patches@sourceware.org> writes:

Hi.  Thanks for the patch.

> +	  if (type->code () == TYPE_CODE_PTR && is_dynamic_type (type))
> +	    {
> +	      CORE_ADDR addr;
> +	      if (nullptr != TYPE_DATA_LOCATION (TYPE_TARGET_TYPE (type)))
> +		addr = value_address (val);
> +	      else
> +		addr = value_as_address (val);

This seems weird to me.  When does value_as_address fail?
It seems to me that a value whose type is TYPE_CODE_PTR should be more
easily convertible to a CORE_ADDR without examining TYPE_DATA_LOCATION.

The same thing applies in a couple more spots in the patch.

I'm curious why the code added here is not also needed in f-valprint.c.

> --- a/gdb/testsuite/gdb.cp/vla-cxx.exp
> +++ b/gdb/testsuite/gdb.cp/vla-cxx.exp
> @@ -23,6 +23,36 @@ if ![runto_main] {
>      return -1
>  }
 
> +gdb_breakpoint [gdb_get_line_number "Before pointer assignment"]
> +gdb_continue_to_breakpoint "Before pointer assignment"
> +
> +set test_name "ptype ptr, Before pointer assignment"
> +gdb_test_multiple "ptype ptr" $test_name {
> +  # gfortran
> +  -re "= int \\(\\*\\)\\\[variable length\\\]\r\n$gdb_prompt $" {
> +    pass $test_name
> +  }
> +  # ifort/ifx

This references fortran compilers but is a c++ test.

> +gdb_test "ptype ptr" "int \\(\\*\\)\\\[3\\\]"
> +gdb_test "print ptr" "\\(int \\(\\*\\)\\\[3\\\]\\) $hex"
> +gdb_test "print *ptr" " = \\{5, 7, 9\\}"

Do these pass with all compilers or do they also need to be conditional?

thanks,
Tom