⚠️ remove Wishbone tag signal (#845)

stnolting · Mar 11, 2024 · e893ddb · e893ddb
2 parents 46baf5a + 5a90534
commit e893ddb
Show file tree

Hide file tree

Showing 16 changed files with 120 additions and 177 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -30,6 +30,7 @@ mimpid = 0x01040312 -> Version 01.04.03.12 -> v1.4.3.12
 
 | Date | Version | Comment | Link |
 |:----:|:-------:|:--------|:----:|
+| 11.03.2024 | 1.9.6.3 | :warning: remove Wishbone tag signal; minor rtl edits and optimizations | [#845](https://github.com/stnolting/neorv32/pull/845) |
 | 10.03.2024 | 1.9.6.2 | minor rtl clean-ups, optimizations and fixes | [#843](https://github.com/stnolting/neorv32/pull/843) |
 | 09.03.2024 | 1.9.6.1 | add generic cache module (not used yet) | [#842](https://github.com/stnolting/neorv32/pull/842) |
 | 01.03.2024 | [**:rocket:1.9.6**](https://github.com/stnolting/neorv32/releases/tag/v1.9.6) | **New release** | |

diff --git a/docs/datasheet/soc.adoc b/docs/datasheet/soc.adoc
@@ -71,7 +71,6 @@ bits/channels are hardwired to zero.
 [NOTE]
 Some interfaces (like the TWI and the 1-Wire bus) require tri-state drivers in the designs top module.
 
-
 .NEORV32 Processor Signal List
 [cols="<3,^1,^1,^1,<8"]
 [options="header",grid="rows"]
@@ -87,7 +86,6 @@ Some interfaces (like the TWI and the 1-Wire bus) require tri-state drivers in t
 | `jtag_tdo_o`     |  1 | out |   -   | serial data output
 | `jtag_tms_i`     |  1 |  in | `'L'` | mode select
 5+^| **<<_processor_external_memory_interface_wishbone>>**
-| `wb_tag_o`       |  3 | out |   -   |  tag (access type identifier)
 | `wb_adr_o`       | 32 | out |   -   |  destination address
 | `wb_dat_i`       | 32 |  in | `'L'` |  write data
 | `wb_dat_o`       | 32 | out |   -   |  read data

diff --git a/docs/datasheet/soc_wishbone.adoc b/docs/datasheet/soc_wishbone.adoc
@@ -7,8 +7,7 @@
 |=======================
 | Hardware source file(s): | neorv32_wishbone.vhd |
 | Software driver file(s): | none                 | _implicitly used_
-| Top entity port:         | `wb_tag_o`  | request tag output (3-bit)
-|                          | `wb_adr_o`  | address output (32-bit)
+| Top entity port:         | `wb_adr_o`  | address output (32-bit)
 |                          | `wb_dat_i`  | data input (32-bit)
 |                          | `wb_dat_o`  | data output (32-bit)
 |                          | `wb_we_o`   | write enable (1-bit)
@@ -17,14 +16,12 @@
 |                          | `wb_cyc_o`  | valid cycle (1-bit)
 |                          | `wb_ack_i`  | acknowledge (1-bit)
 |                          | `wb_err_i`  | bus error (1-bit)
-|                          | `fence_o`   | an executed `fence` instruction
-|                          | `fencei_o`  | an executed `fence.i` instruction
 | Configuration generics:  | `MEM_EXT_EN`         | enable external memory interface when `true`
 |                          | `MEM_EXT_TIMEOUT`    | number of clock cycles after which an unacknowledged external bus access will auto-terminate (0 = disabled)
 |                          | `MEM_EXT_PIPE_MODE`  | when `false` (default): classic/standard Wishbone protocol; when `true`: pipelined Wishbone protocol
 |                          | `MEM_EXT_BIG_ENDIAN` | byte-order (Endianness) of external memory interface; `true`=BIG, `false`=little (default)
 |                          | `MEM_EXT_ASYNC_RX`   | use registered RX path when `false` (default); use async/direct RX path when `true`
-|                          | `MEM_EXT_ASYNC_TX_`  | use registered TX path when `false` (default); use async/direct TX path when `true`
+|                          | `MEM_EXT_ASYNC_TX`   | use registered TX path when `false` (default); use async/direct TX path when `true`
 | CPU interrupts:          | none |
 |=======================
 
@@ -85,15 +82,6 @@ or terminate (via `wb_err_i`) the transfer within `MEM_EXT_TIMEOUT` clock cycles
 setting `wb_cyc_o` low again and a CPU load/store/instruction fetch bus access fault exception is raised.
 
 
-**Wishbone Tag**
-
-The 3-bit wishbone `wb_tag_o` signal provides additional information regarding the access type:
-
-* `wb_tag_o(0)`: `1` = privileged access (CPU is in machine mode); `0` = unprivileged access (CPU is not in machine mode)
-* `wb_tag_o(1)`: always zero
-* `wb_tag_o(2)`: `1` = instruction fetch access, `0` = data access
-
-
 **Endianness**
 
 The NEORV32 CPU and the Processor setup are *little-endian* architectures. To allow direct connection

diff --git a/docs/figures/wishbone_classic_read.png b/docs/figures/wishbone_classic_read.png
diff --git a/docs/figures/wishbone_pipelined_write.png b/docs/figures/wishbone_pipelined_write.png
diff --git a/docs/sources/wishbone_classic_read.json b/docs/sources/wishbone_classic_read.json
@@ -1,14 +1,12 @@
 {signal: [
   {name: 'clk',       wave: 'p....|...'},
-  {name: 'wb_tag_o',  wave: 'x3...|.x.', data: ['Tag']},
   {name: 'wb_adr_o',  wave: 'x3...|.x.', data: ['Address']},
   {name: 'wb_dat_i',  wave: 'x....|3x.', data: ['rdata']},
   {name: 'wb_dat_o',  wave: 'x....|.x.', data: ['Wdata']},
   {name: 'wb_we_o',   wave: 'x0...|.x.'},
   {name: 'wb_sel_o',  wave: 'x....|.x.', data: ['Byte_enable']},
   {name: 'wb_stb_o',  wave: '01...|.0.'},
   {name: 'wb_cyc_o',  wave: '01...|.0.'},
-  {name: 'wb_lock_o', wave: '0....|...'},
   {name: 'wb_ack_i',  wave: '0....|10.'},
   {name: 'wb_err_i',  wave: '0....|...'},
 ]}
diff --git a/docs/sources/wishbone_pipelined_write.json b/docs/sources/wishbone_pipelined_write.json
@@ -1,14 +1,12 @@
 {signal: [
   {name: 'clk',       wave: 'p....|...'},
-  {name: 'wb_tag_o',  wave: 'x3...|.x.', data: ['Tag']},
   {name: 'wb_adr_o',  wave: 'x3...|.x.', data: ['Address']},
   {name: 'wb_dat_i',  wave: 'x....|.x.'},
   {name: 'wb_dat_o',  wave: 'x3...|.x.', data: ['Wdata']},
   {name: 'wb_we_o',   wave: 'x1...|.x.'},
   {name: 'wb_sel_o',  wave: 'x3...|.x.', data: ['Byte_enable']},
   {name: 'wb_stb_o',  wave: '010..|...'},
   {name: 'wb_cyc_o',  wave: '01...|.0.'},
-  {name: 'wb_lock_o', wave: '0....|...'},
   {name: 'wb_ack_i',  wave: '0....|10.'},
   {name: 'wb_err_i',  wave: '0....|...'},
 ]}
diff --git a/rtl/core/neorv32_cache.vhd b/rtl/core/neorv32_cache.vhd
@@ -830,18 +830,16 @@ architecture neorv32_cache_bus_rtl of neorv32_cache_bus is
   constant index_size_c  : natural := index_size_f(NUM_BLOCKS);
   constant tag_size_c    : natural := 32 - (offset_size_c + index_size_c + 2);
 
-  -- host request buffer --
-  signal hreq : bus_req_t;
-
   -- control fsm --
-  type state_t is (S_IDLE, S_CHECK, S_DOWNLOAD_REQ, S_DOWNLOAD_RSP, S_UPLOAD_GET, S_UPLOAD_REQ, S_UPLOAD_RSP, S_FLUSH_START, S_FLUSH_READ, S_FLUSH_CHECK);
+  type state_t is (S_IDLE, S_CHECK, S_DOWNLOAD_REQ, S_DOWNLOAD_RSP, S_UPLOAD_GET,
+                   S_UPLOAD_REQ, S_UPLOAD_RSP, S_FLUSH_START, S_FLUSH_READ, S_FLUSH_CHECK);
   signal state, upret, state_nxt, upret_nxt: state_t;
 
   -- address generator --
   type addr_t is record
     tag : std_ulogic_vector(tag_size_c-1 downto 0);
-    ind : std_ulogic_vector(index_size_c-1 downto 0);
-    off : std_ulogic_vector(offset_size_c-1 downto 0); -- WORD offset!
+    idx : std_ulogic_vector(index_size_c-1 downto 0);
+    ofs : std_ulogic_vector(offset_size_c-1 downto 0); -- WORD offset!
   end record;
   signal haddr, baddr, addr, addr_nxt : addr_t;
 
@@ -851,13 +849,13 @@ begin
   -- -------------------------------------------------------------------------------------------
   -- base address of original host access --
   haddr.tag <= host_req_i.addr(31 downto (32-tag_size_c));
-  haddr.ind <= host_req_i.addr((offset_size_c+2+index_size_c)-1 downto offset_size_c+2);
-  haddr.off <= (others => '0'); -- unused
+  haddr.idx <= (others => '0'); -- unused
+  haddr.ofs <= (others => '0'); -- unused
 
   -- base address of indexed cache block --
   baddr.tag <= base_i(31 downto (32-tag_size_c));
-  baddr.ind <= base_i((offset_size_c+2+index_size_c)-1 downto offset_size_c+2);
-  baddr.off <= (others => '0'); -- unused
+  baddr.idx <= base_i((offset_size_c+2+index_size_c)-1 downto offset_size_c+2);
+  baddr.ofs <= (others => '0'); -- unused
 
 
   -- Control Engine FSM Sync ----------------------------------------------------------------
@@ -868,29 +866,27 @@ begin
       state    <= S_IDLE;
       upret    <= S_IDLE;
       addr.tag <= (others => '0');
-      addr.ind <= (others => '0');
-      addr.off <= (others => '0');
-      hreq     <= req_terminate_c;
+      addr.idx <= (others => '0');
+      addr.ofs <= (others => '0');
     elsif rising_edge(clk_i) then
       state <= state_nxt;
       upret <= upret_nxt;
       addr  <= addr_nxt;
-      hreq  <= host_req_i;
     end if;
   end process ctrl_engine_sync;
 
 
   -- Control Engine FSM Comb ----------------------------------------------------------------
   -- -------------------------------------------------------------------------------------------
-  ctrl_engine_comb: process(state, upret, addr, hreq, haddr, baddr, bus_rsp_i, cmd_sync_i, cmd_miss_i, rdata_i, dirty_i)
+  ctrl_engine_comb: process(state, upret, addr, haddr, baddr, bus_rsp_i, cmd_sync_i, cmd_miss_i, rdata_i, dirty_i)
   begin
     -- control engine defaults --
     state_nxt <= state;
     upret_nxt <= upret;
     addr_nxt  <= addr;
 
     -- cache defaults --
-    addr_o  <= addr.tag & addr.ind & addr.off & "00"; -- always word-aligned
+    addr_o  <= addr.tag & addr.idx & addr.ofs & "00"; -- always word-aligned
     we_o    <= (others => '0');
     swe_o   <= '0';
     wdata_o <= bus_rsp_i.data;
@@ -902,17 +898,18 @@ begin
 
     -- bus interface defaults --
     bus_req_o      <= req_terminate_c; -- all-zero
-    bus_req_o.addr <= addr.tag & addr.ind & addr.off & "00"; -- always word-aligned
+    bus_req_o.addr <= addr.tag & addr.idx & addr.ofs & "00"; -- always word-aligned
     bus_req_o.data <= rdata_i;
     bus_req_o.ben  <= (others => '1'); -- full-word writes only
-    bus_req_o.priv <= hreq.priv; -- keep original privilege level
+    bus_req_o.src  <= '0'; -- cache accesses are always "data" accesses
+    bus_req_o.priv <= '0'; -- cache accesses are always "unprivileged" accesses
 
     -- fsm --
     case state is
 
       when S_IDLE => -- wait for request
       -- ------------------------------------------------------------
-        addr_nxt.off <= (others => '0'); -- align block base address for upload/download (and flush)
+        addr_nxt.ofs <= (others => '0'); -- align block base address for upload/download (and flush)
         if (cmd_sync_i = '1') then -- cache sync
           state_nxt <= S_FLUSH_START;
         elsif (cmd_miss_i = '1') then -- cache miss
@@ -921,14 +918,13 @@ begin
 
       when S_CHECK => -- check if accessed block is dirty (cache address is still applied by host controller!)
       -- ------------------------------------------------------------
-        upret_nxt <= S_DOWNLOAD_REQ; -- go straight to S_DOWNLOAD_REQ when S_UPLOAD_GET has completed (if executed)
+        upret_nxt    <= S_DOWNLOAD_REQ; -- go straight to S_DOWNLOAD_REQ when S_UPLOAD_GET has completed (if executed)
+        addr_nxt.idx <= baddr.idx; -- index of reference cache block
         if (dirty_i = '1') then -- block is dirty, upload first
           addr_nxt.tag <= baddr.tag; -- base address (tag + index) of accessed block
-          addr_nxt.ind <= baddr.ind;
           state_nxt    <= S_UPLOAD_GET;
-        else -- block is clean, download new block and override
+        else -- block is clean, download new block
           addr_nxt.tag <= haddr.tag; -- base address (tag + index) of requested block
-          addr_nxt.ind <= haddr.ind;
           state_nxt    <= S_DOWNLOAD_REQ;
         end if;
 
@@ -946,8 +942,8 @@ begin
         swe_o        <= '1'; -- cache: write status bit (bus error response)
         new_o        <= '1'; -- set new block (set tag, make valid, make clean)
         if (bus_rsp_i.ack = '1') or (bus_rsp_i.err = '1') then -- wait for response
-          addr_nxt.off <= std_ulogic_vector(unsigned(addr.off) + 1);
-          if (and_reduce_f(addr.off) = '1') then -- block completed? offset will be all-zero again after block completion
+          addr_nxt.ofs <= std_ulogic_vector(unsigned(addr.ofs) + 1);
+          if (and_reduce_f(addr.ofs) = '1') then -- block completed? offset will be all-zero again after block completion
             state_nxt <= S_IDLE;
           else -- get next word
             state_nxt <= S_DOWNLOAD_REQ;
@@ -971,8 +967,8 @@ begin
         bus_req_o.rw <= '1'; -- write access
         new_o        <= '1'; -- set new block (set tag, make valid, make clean)
         if (bus_rsp_i.ack = '1') or (bus_rsp_i.err = '1') then -- wait for response
-          addr_nxt.off <= std_ulogic_vector(unsigned(addr.off) + 1);
-          if (and_reduce_f(addr.off) = '1') then -- block completed? offset will be all-zero again after block completion
+          addr_nxt.ofs <= std_ulogic_vector(unsigned(addr.ofs) + 1);
+          if (and_reduce_f(addr.ofs) = '1') then -- block completed? offset will be all-zero again after block completion
             state_nxt <= upret; -- go back to "upload-done return state"
           else -- get next word
             state_nxt <= S_UPLOAD_GET;
@@ -982,7 +978,7 @@ begin
 
       when S_FLUSH_START => -- start checking for dirty blocks
       -- ------------------------------------------------------------
-        addr_nxt.ind <= (others => '0'); -- start with index 0
+        addr_nxt.idx <= (others => '0'); -- start with index 0
         upret_nxt    <= S_FLUSH_CHECK; -- come back to S_FLUSH_CHECK after block upload
         state_nxt    <= S_FLUSH_READ;
 
@@ -997,8 +993,8 @@ begin
         if (dirty_i = '1') then -- block dirty?
           state_nxt    <= S_UPLOAD_GET;
         else -- move on to next block
-          addr_nxt.ind <= std_ulogic_vector(unsigned(addr.ind) + 1);
-          if (and_reduce_f(addr.ind) = '1') then -- all blocks done?
+          addr_nxt.idx <= std_ulogic_vector(unsigned(addr.idx) + 1);
+          if (and_reduce_f(addr.idx) = '1') then -- all blocks done?
             bus_req_o.fence <= '1'; -- forward fence request to downstream memories
             state_nxt       <= S_IDLE;
           else -- go to next block

diff --git a/rtl/core/neorv32_package.vhd b/rtl/core/neorv32_package.vhd
@@ -53,7 +53,7 @@ package neorv32_package is
 
   -- Architecture Constants -----------------------------------------------------------------
   -- -------------------------------------------------------------------------------------------
-  constant hw_version_c : std_ulogic_vector(31 downto 0) := x"01090602"; -- hardware version
+  constant hw_version_c : std_ulogic_vector(31 downto 0) := x"01090603"; -- hardware version
   constant archid_c     : natural := 19; -- official RISC-V architecture ID
   constant XLEN         : natural := 32; -- native data path width
 
@@ -843,7 +843,6 @@ package neorv32_package is
       jtag_tdo_o     : out std_ulogic;
       jtag_tms_i     : in  std_ulogic := 'L';
       -- Wishbone bus interface (available if MEM_EXT_EN = true) --
-      wb_tag_o       : out std_ulogic_vector(02 downto 0);
       wb_adr_o       : out std_ulogic_vector(31 downto 0);
       wb_dat_i       : in  std_ulogic_vector(31 downto 0) := (others => 'L');
       wb_dat_o       : out std_ulogic_vector(31 downto 0);