vhdl - signed to std_logic_vector, slice results

Question

I need to take the absolute value of a result and I am only interested in the most significant bits. This is what I have done:

data_ram_h <= std_logic_vector(abs(signed(resize(r4(calc_cnt - 2), data_ram_h'length) + r4(calc_cnt - 1) +
                    r4(calc_cnt) + r4(calc_cnt + 1) + r4(calc_cnt + 2) -
                    r2(calc_cnt - 2) - r2(calc_cnt - 1) - r2(calc_cnt) -
                    r2(calc_cnt + 1) - r2(calc_cnt + 2))))(11 downto 4);

I try to check the syntax and I get this error:

type conversion std_logic_vector is not allowed as a prefix for an slice name.

data_ram_h is a std_logic_vector of the right dimension, and the abs function returns a signed, to there shouldn't be problem in the conversion to std_logic_vector. The library I am using is use ieee.numeric_std.all.

Where am I wrong? Thanks in advance c:

Answer 1

A type conversion is a basic operation that happens to require parentheses around it's operand expression. And there's the rub, it's use is not a function call, so it can't be used as a prefix for a slice name.

A prefix for a slice name is either a function_call or a name. (IEEE Std 1076-2008, 5 Types, 5.1 General, explicit type conversion, 8 Names, 8.1 General, 8.5 Slice names).

If it was a function call you could slice the result.

On the other hand you can slice `"abs", so slice that and then do the type conversion:

library ieee;
use ieee.std_logic_1164.all;
use ieee.numeric_std.all;

entity slice is
end entity;

architecture foo of slice is
    signal h_tmp: signed (11 downto 0);
    signal h_tmp_vec: std_logic_vector (11 downto 0);
    signal data_ram_h: std_logic_vector(7 downto 0);
    signal calc_cnt:     integer := 3;
    type r_array is array (0 to 15) of unsigned(15 downto 0);
    signal r2, r4: r_array := (others => (others => '0'));
begin


    data_ram_h<= std_logic_vector (
                     "abs"(signed(resize(r4(calc_cnt - 2), data_ram_h'length) + r4(calc_cnt - 1) +
                      r4(calc_cnt) + r4(calc_cnt + 1) + r4(calc_cnt + 2) -
                      r2(calc_cnt - 2) - r2(calc_cnt - 1) - r2(calc_cnt) -
                      r2(calc_cnt + 1) - r2(calc_cnt + 2)))(11 downto 4)
                      );

end architecture;

Using abs as a function call requires you use it's declared name which is "abs".

I'm just guessing at some declarations here, so I can't guarantee this works in your code. The above example does analyze, elaborate and run which says the subtype ranges are compatible.