loadstore1: Simplify address generation in OP_FETCH_FAILED case

Instead of having a multiplexer in loadstore1 in order to be able to
put the instruction address into v.addr, we now set decode.input_reg_a
to CIA in the decode table entry for OP_FETCH_FAILED.  That means that
the operand selection machinery in decode2 will supply the instruction
address to loadstore1 on the lv.addr1 input and no special case is
needed in loadstore1.  This saves a few LUTs (~40 on the Artix-7).

Signed-off-by: Paul Mackerras <paulus@ozlabs.org>

common.vhdl

@@ -476,7 +476,6 @@ package common is
     type Execute1ToLoadstore1Type is record
 	valid : std_ulogic;
         op : insn_type_t;                               -- what ld/st or m[tf]spr or TLB op to do
-        nia : std_ulogic_vector(63 downto 0);
         insn : std_ulogic_vector(31 downto 0);
         instr_tag : instr_tag_t;
 	addr1 : std_ulogic_vector(63 downto 0);
@@ -504,7 +503,7 @@ package common is
         (valid => '0', op => OP_ILLEGAL, ci => '0', byte_reverse => '0',
          sign_extend => '0', update => '0', xerc => xerc_init,
          reserve => '0', rc => '0', virt_mode => '0', priv_mode => '0',
-         nia => (others => '0'), insn => (others => '0'),
+         insn => (others => '0'),
          instr_tag => instr_tag_init,
          addr1 => (others => '0'), addr2 => (others => '0'), data => (others => '0'),
          write_reg => (others => '0'),

decode1.vhdl

@@ -527,7 +527,7 @@ architecture behaviour of decode1 is
     --                                        unit   fac   internal         in1         in2          in3   out   CR   CR   inv  inv  cry   cry  ldst  BR   sgn  upd  rsrv 32b  sgn  rc    lk   sgl  rpt
     --                                                           op                                              in   out   A   out  in    out  len        ext                                 pipe
     constant nop_instr      : decode_rom_t := (ALU,  NONE, OP_NOP,          NONE,       NONE,        NONE, NONE, '0', '0', '0', '0', ZERO, '0', NONE, '0', '0', '0', '0', '0', '0', NONE, '0', '0', NONE);
-    constant fetch_fail_inst: decode_rom_t := (LDST, NONE, OP_FETCH_FAILED, NONE,       NONE,        NONE, NONE, '0', '0', '0', '0', ZERO, '0', NONE, '0', '0', '0', '0', '0', '0', NONE, '0', '0', NONE);
+    constant fetch_fail_inst: decode_rom_t := (LDST, NONE, OP_FETCH_FAILED, CIA,        NONE,        NONE, NONE, '0', '0', '0', '0', ZERO, '0', NONE, '0', '0', '0', '0', '0', '0', NONE, '0', '0', NONE);
 
     function decode_ram_spr(sprn : spr_num_t) return ram_spr_info is
         variable ret : ram_spr_info;

execute1.vhdl

@@ -1554,7 +1554,6 @@ begin
 
         -- Outputs to loadstore1 (async)
         lv.op := e_in.insn_type;
-        lv.nia := e_in.nia;
         lv.instr_tag := e_in.instr_tag;
         lv.addr1 := a_in;
         lv.addr2 := b_in;

loadstore1.vhdl

@@ -497,7 +497,6 @@ begin
             when OP_FETCH_FAILED =>
                 -- send it to the MMU to do the radix walk
                 v.instr_fault := '1';
-                v.addr := l_in.nia;
                 v.mmu_op := '1';
             when others =>
         end case;