execute1: Implement trace interrupts

Trace interrupts occur when the MSR[TE] field is non-zero and an
instruction other than rfid has been successfully completed.  A trace
interrupt occurs before the next instruction is executed or any
asynchronous interrupt is taken.

Since the trace interrupt is defined to set SRR1 bits depending on
whether the traced instruction is a load or an instruction treated as
a load, or a store or an instruction treated as a store, we need to
make sure the treated-as-a-load instructions (icbi, icbt, dcbt, dcbst,
dcbf) and the treated-as-a-store instructions (dcbtst, dcbz) have the
correct opcodes in decode1.  Several of them were previously marked as
OP_NOP.

We don't yet implement the SIAR or SDAR registers, which should be set
by trace interrupts.

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

common.vhdl

@@ -13,6 +13,8 @@ package common is
     constant MSR_SF  : integer := (63 - 0);     -- Sixty-Four bit mode
     constant MSR_EE  : integer := (63 - 48);    -- External interrupt Enable
     constant MSR_PR  : integer := (63 - 49);    -- PRoblem state
+    constant MSR_SE  : integer := (63 - 53);    -- Single-step bit of TE field
+    constant MSR_BE  : integer := (63 - 54);    -- Branch trace bit of TE field
     constant MSR_IR  : integer := (63 - 58);    -- Instruction Relocation
     constant MSR_DR  : integer := (63 - 59);    -- Data Relocation
     constant MSR_RI  : integer := (63 - 62);    -- Recoverable Interrupt

decode1.vhdl

@@ -201,10 +201,10 @@ architecture behaviour of decode1 is
         2#1000111010#  =>       (ALU,    OP_CNTZ,      NONE,       NONE,        RS,   RA,   '0', '0', '0', '0', ZERO, '0', NONE, '0', '0', '0', '0', '0', '0', RC,   '0', '0'), -- cnttzd
         2#1000011010#  =>       (ALU,    OP_CNTZ,      NONE,       NONE,        RS,   RA,   '0', '0', '0', '0', ZERO, '0', NONE, '0', '0', '0', '0', '1', '0', RC,   '0', '0'), -- cnttzw
         2#1011110011#  =>       (ALU,    OP_DARN,      NONE,       NONE,        NONE, RT,   '0', '0', '0', '0', ZERO, '0', NONE, '0', '0', '0', '0', '0', '0', NONE, '0', '0'), -- darn
-        2#0001010110#  =>       (ALU,    OP_NOP,       NONE,       NONE,        NONE, NONE, '0', '0', '0', '0', ZERO, '0', NONE, '0', '0', '0', '0', '0', '0', NONE, '0', '1'), -- dcbf
-        2#0000110110#  =>       (ALU,    OP_NOP,       NONE,       NONE,        NONE, NONE, '0', '0', '0', '0', ZERO, '0', NONE, '0', '0', '0', '0', '0', '0', NONE, '0', '1'), -- dcbst
-        2#0100010110#  =>       (ALU,    OP_NOP,       NONE,       NONE,        NONE, NONE, '0', '0', '0', '0', ZERO, '0', NONE, '0', '0', '0', '0', '0', '0', NONE, '0', '1'), -- dcbt
-        2#0011110110#  =>       (ALU,    OP_NOP,       NONE,       NONE,        NONE, NONE, '0', '0', '0', '0', ZERO, '0', NONE, '0', '0', '0', '0', '0', '0', NONE, '0', '1'), -- dcbtst
+        2#0001010110#  =>       (ALU,    OP_DCBF,      NONE,       NONE,        NONE, NONE, '0', '0', '0', '0', ZERO, '0', NONE, '0', '0', '0', '0', '0', '0', NONE, '0', '1'), -- dcbf
+        2#0000110110#  =>       (ALU,    OP_DCBST,     NONE,       NONE,        NONE, NONE, '0', '0', '0', '0', ZERO, '0', NONE, '0', '0', '0', '0', '0', '0', NONE, '0', '1'), -- dcbst
+        2#0100010110#  =>       (ALU,    OP_DCBT,      NONE,       NONE,        NONE, NONE, '0', '0', '0', '0', ZERO, '0', NONE, '0', '0', '0', '0', '0', '0', NONE, '0', '1'), -- dcbt
+        2#0011110110#  =>       (ALU,    OP_DCBTST,    NONE,       NONE,        NONE, NONE, '0', '0', '0', '0', ZERO, '0', NONE, '0', '0', '0', '0', '0', '0', NONE, '0', '1'), -- dcbtst
         2#1111110110#  =>       (LDST,   OP_DCBZ,      RA_OR_ZERO, RB,          NONE, NONE, '0', '0', '0', '0', ZERO, '0', NONE, '0', '0', '0', '0', '0', '0', NONE, '0', '0'), -- dcbz
         2#0110001001#  =>       (ALU,    OP_DIVE,      RA,         RB,          NONE, RT,   '0', '0', '0', '0', ZERO, '0', NONE, '0', '0', '0', '0', '0', '0', RC,   '0', '0'), -- divdeu
         2#1110001001#  =>       (ALU,    OP_DIVE,      RA,         RB,          NONE, RT,   '0', '0', '0', '0', ZERO, '0', NONE, '0', '0', '0', '0', '0', '0', RC,   '0', '0'), -- divdeuo
@@ -230,7 +230,7 @@ architecture behaviour of decode1 is
         2#1101111010#  =>       (ALU,    OP_EXTSWSLI,  NONE,       CONST_SH,    RS,   RA,   '0', '0', '0', '0', ZERO, '0', NONE, '0', '0', '0', '0', '0', '0', RC,   '0', '0'), -- extswsli
         2#1101111011#  =>       (ALU,    OP_EXTSWSLI,  NONE,       CONST_SH,    RS,   RA,   '0', '0', '0', '0', ZERO, '0', NONE, '0', '0', '0', '0', '0', '0', RC,   '0', '0'), -- extswsli
         2#1111010110#  =>       (ALU,    OP_ICBI,      NONE,       NONE,        NONE, NONE, '0', '0', '0', '0', ZERO, '0', NONE, '0', '0', '0', '0', '0', '0', NONE, '0', '1'), -- icbi
-        2#0000010110#  =>       (ALU,    OP_NOP,       NONE,       NONE,        NONE, NONE, '0', '0', '0', '0', ZERO, '0', NONE, '0', '0', '0', '0', '0', '0', NONE, '0', '1'), -- icbt
+        2#0000010110#  =>       (ALU,    OP_ICBT,      NONE,       NONE,        NONE, NONE, '0', '0', '0', '0', ZERO, '0', NONE, '0', '0', '0', '0', '0', '0', NONE, '0', '1'), -- icbt
         2#0000001111#  =>       (ALU,    OP_ISEL,      RA_OR_ZERO, RB,          NONE, RT,   '1', '0', '0', '0', ZERO, '0', NONE, '0', '0', '0', '0', '0', '0', NONE, '0', '1'), -- isel
         2#0000101111#  =>       (ALU,    OP_ISEL,      RA_OR_ZERO, RB,          NONE, RT,   '1', '0', '0', '0', ZERO, '0', NONE, '0', '0', '0', '0', '0', '0', NONE, '0', '0'), -- isel
         2#0001001111#  =>       (ALU,    OP_ISEL,      RA_OR_ZERO, RB,          NONE, RT,   '1', '0', '0', '0', ZERO, '0', NONE, '0', '0', '0', '0', '0', '0', NONE, '0', '0'), -- isel

execute1.vhdl

@@ -53,6 +53,8 @@ architecture behaviour of execute1 is
         f : Execute1ToFetch1Type;
         busy: std_ulogic;
         terminate: std_ulogic;
+        trace_next : std_ulogic;
+        prev_op : insn_type_t;
 	lr_update : std_ulogic;
 	next_lr : std_ulogic_vector(63 downto 0);
 	mul_in_progress : std_ulogic;
@@ -69,7 +71,7 @@ architecture behaviour of execute1 is
     end record;
     constant reg_type_init : reg_type :=
         (e => Execute1ToWritebackInit, f => Execute1ToFetch1Init,
-         busy => '0', lr_update => '0', terminate => '0',
+         busy => '0', lr_update => '0', terminate => '0', trace_next => '0', prev_op => OP_ILLEGAL,
          mul_in_progress => '0', mul_finish => '0', div_in_progress => '0', cntz_in_progress => '0',
          slow_op_insn => OP_ILLEGAL, slow_op_rc => '0', slow_op_oe => '0', slow_op_xerc => xerc_init,
          next_lr => (others => '0'), last_nia => (others => '0'), others => (others => '0'));
@@ -330,6 +332,7 @@ begin
         variable abs_branch : std_ulogic;
         variable spr_val : std_ulogic_vector(63 downto 0);
         variable addend : std_ulogic_vector(127 downto 0);
+        variable do_trace : std_ulogic;
     begin
 	result := (others => '0');
 	sum_with_carry := (others => '0');
@@ -525,6 +528,11 @@ begin
 
         v.e.mode_32bit := not ctrl.msr(MSR_SF);
 
+        do_trace := valid_in and ctrl.msr(MSR_SE);
+        if valid_in = '1' then
+            v.prev_op := e_in.insn_type;
+        end if;
+
  	if ctrl.irq_state = WRITE_SRR1 then
  	    v.e.exc_write_reg := fast_spr_num(SPR_SRR1);
  	    v.e.exc_write_data := ctrl.srr1;
@@ -532,13 +540,29 @@ begin
             ctrl_tmp.msr(MSR_SF) <= '1';
             ctrl_tmp.msr(MSR_EE) <= '0';
             ctrl_tmp.msr(MSR_PR) <= '0';
+            ctrl_tmp.msr(MSR_SE) <= '0';
+            ctrl_tmp.msr(MSR_BE) <= '0';
             ctrl_tmp.msr(MSR_IR) <= '0';
             ctrl_tmp.msr(MSR_DR) <= '0';
             ctrl_tmp.msr(MSR_RI) <= '0';
             ctrl_tmp.msr(MSR_LE) <= '1';
             v.e.valid := '1';
+            v.trace_next := '0';
 	    report "Writing SRR1: " & to_hstring(ctrl.srr1);
 
+        elsif r.trace_next = '1' and valid_in = '1' then
+            -- Generate a trace interrupt rather than executing the next instruction
+            -- or taking any asynchronous interrupt
+            v.f.redirect_nia := std_logic_vector(to_unsigned(16#d00#, 64));
+            ctrl_tmp.srr1(63 - 33) <= '1';
+            if r.prev_op = OP_LOAD or r.prev_op = OP_ICBI or r.prev_op = OP_ICBT or
+                r.prev_op = OP_DCBT or r.prev_op = OP_DCBST or r.prev_op = OP_DCBF then
+                ctrl_tmp.srr1(63 - 35) <= '1';
+            elsif r.prev_op = OP_STORE or r.prev_op = OP_DCBZ or r.prev_op = OP_DCBTST then
+                ctrl_tmp.srr1(63 - 36) <= '1';
+            end if;
+            exception := '1';
+
 	elsif irq_valid = '1' and valid_in = '1' then
 	    -- we need two cycles to write srr0 and 1
 	    -- will need more when we have to write HEIR
@@ -594,7 +618,7 @@ begin
                 else
                     illegal := '1';
                 end if;
-	    when OP_NOP =>
+	    when OP_NOP | OP_DCBF | OP_DCBST | OP_DCBT | OP_DCBTST | OP_ICBT =>
 		-- Do nothing
 	    when OP_ADD | OP_CMP | OP_TRAP =>
 		result := sum_with_carry(63 downto 0);
@@ -715,6 +739,9 @@ begin
                 is_branch := '1';
                 taken_branch := '1';
                 abs_branch := insn_aa(e_in.insn);
+                if ctrl.msr(MSR_BE) = '1' then
+                    do_trace := '1';
+                end if;
 	    when OP_BC =>
 		-- read_data1 is CTR
 		bo := insn_bo(e_in.insn);
@@ -727,6 +754,9 @@ begin
                 is_branch := '1';
 		taken_branch := ppc_bc_taken(bo, bi, cr_in, a_in);
                 abs_branch := insn_aa(e_in.insn);
+                if ctrl.msr(MSR_BE) = '1' then
+                    do_trace := '1';
+                end if;
 	    when OP_BCREG =>
 		-- read_data1 is CTR
 		-- read_data2 is target register (CTR, LR or TAR)
@@ -740,6 +770,9 @@ begin
                 is_branch := '1';
 		taken_branch := ppc_bc_taken(bo, bi, cr_in, a_in);
                 abs_branch := '1';
+                if ctrl.msr(MSR_BE) = '1' then
+                    do_trace := '1';
+                end if;
 
 	    when OP_RFID =>
                 v.f.virt_mode := a_in(MSR_IR) or a_in(MSR_PR);
@@ -760,6 +793,7 @@ begin
                 is_branch := '1';
                 taken_branch := '1';
                 abs_branch := '1';
+                do_trace := '0';
 
             when OP_CNTZ =>
                 v.e.valid := '0';
@@ -1135,6 +1169,10 @@ begin
             end if;
 	end if;
 
+        if do_trace = '1' then
+            v.trace_next := '1';
+        end if;
+
 	v.e.write_data := result;
 	v.e.write_enable := result_en and not exception;