swcore: bugfix: lossing rtu decisions in case of very short stuck input

2dfa0b8e · Maciej Lipinski · Tomasz Wlostowski · 676c5b4a · 2dfa0b8e
Commit 2dfa0b8e authored Mar 15, 2012 by Maciej Lipinski Committed by Tomasz Wlostowski Mar 15, 2012
Hide whitespace changes
Inline Side-by-side

Showing with 5 additions and 7 deletions

xswc_input_block.vhd modules/wrsw_swcore/xswc_input_block.vhd +5 -7

No files found.
--- a/modules/wrsw_swcore/xswc_input_block.vhd
+++ b/modules/wrsw_swcore/xswc_input_block.vhd
@@ -311,7 +311,6 @@ architecture syn of xswc_input_block is
  signal pckstart_usecnt_write    : std_logic_vector(g_usecount_width - 1 downto 0);
  -- previously set usecnt (not necessarly the same as pckstart_usecnt !!!)
  signal pckstart_usecnt_prev     : std_logic_vector(g_usecount_width - 1 downto 0);
-  signal pckstart_usecnt_next     : std_logic_vector(g_usecount_width - 1 downto 0);
  signal pckstart_usecnt_pgaddr   : std_logic_vector(g_page_addr_width - 1 downto 0);

  -- interpck page allocation in advance
@@ -396,7 +395,6 @@ architecture syn of xswc_input_block is
  -- used to produce delayed in_pck_sof -- basically, this is illegal by pWB standard, but 
  -- if it happens, we loose, pck, why not to take care of this?
  signal in_pck_sof_on_stall : std_logic;
-  signal in_pck_delayed_sof  : std_logic;

  -- indicates that the reception of pck finishes due to :
  -- (1) eof, error    (by in_pck_*)
@@ -631,7 +629,7 @@ begin  --arch
        -- tracks start of frame on the stall (not allowed, but sometimes happen)
        if(snk_cyc_int = '1' and snk_cyc_d0 = '0' and snk_stall_int = '1') then
          in_pck_sof_on_stall <= '1';
-        elsif(in_pck_delayed_sof = '1' and in_pck_sof_on_stall = '1') then
+        elsif(in_pck_sof_delayed = '1' and in_pck_sof_on_stall = '1') then
          in_pck_sof_on_stall <= '0';
        end if;

@@ -959,6 +957,8 @@ begin  --arch
            if(tp_stuck = '0') then     -- un-stuck the input :)

              s_rcv_pck         <= S_IDLE;
+              snk_stall_force_h <= '1';
+              snk_stall_force_l <= '1';
              rp_drop_on_stuck  <= '0';
              
            else                        -- still stuck
@@ -1222,8 +1222,7 @@ end process p_page_alloc_fsm;
 --------------------------------------------------------------------------------------------------
 --================================================================================================

--rtu_dst_port_mask_tmp <= rtu_dst_port_mask_i and (not f_gen_mask(g_port_index, current_mask'length));
-rtu_dst_port_mask_tmp <= rtu_dst_port_mask_i;
+rtu_dst_port_mask_tmp <= rtu_dst_port_mask_i; -- and (not f_gen_mask(g_port_index, current_mask'length));

 p_register_rtu_rsp : process(clk_i)
 begin
@@ -1241,7 +1240,7 @@ begin
      -- remember input rtu decision
      if(rtu_rsp_valid_i = '1' and rtu_rsp_ack = '0' and rp_accept_rtu = '1') then
        -- make sure we're not forwarding packets to ourselves. 
-        current_mask   <= rtu_dst_port_mask_tmp;--rtu_dst_port_mask_i and (not f_gen_mask(g_port_index, current_mask'length));
+        current_mask   <= rtu_dst_port_mask_tmp;
        current_prio   <= rtu_prio_i;
        current_drop   <= rtu_drop_i;
        current_usecnt <= rtu_dst_port_usecnt;
@@ -1896,7 +1895,6 @@ rp_in_pck_error <= '1' when (rp_in_pck_err = '1' or in_pck_err = '1') else '0';
 --================================================================================================
 -- Input signals
 --================================================================================================
--rtu_dst_port_usecnt <= std_logic_vector(to_unsigned(cnt(rtu_dst_port_mask_i), g_usecount_width));
 rtu_dst_port_usecnt <= std_logic_vector(to_unsigned(cnt(rtu_dst_port_mask_tmp), g_usecount_width));

 -- generating output STALL: fifo_full or stall_after_err or stall_when_stuck;