xilinx synthesis debugging foxes

configure

@@ -65,7 +65,7 @@ copy_files() {
                     filename_no_ext="${filename%.in}"
                     dest_file="$dest_dir/$filename_no_ext"
                     mkdir -p "$dest_dir"
-                    sed "s|@VORTEX_HOME@|$SCRIPT_DIR|g; s|@XLEN@|$XLEN|g; s|@TOOLDIR@|$TOOLDIR|g; s|@OSVERSION@|$OSVERSION|g; s|@INSTALLDIR@|$PREFIX|g" "$file" > "$dest_file"
+                    sed "s|@VORTEX_HOME@|$SCRIPT_DIR|g; s|@XLEN@|$XLEN|g; s|@TOOLDIR@|$TOOLDIR|g; s|@OSVERSION@|$OSVERSION|g; s|@INSTALLDIR@|$PREFIX|g; s|@CURRENTDIR@|$CURRENT_DIR|g" "$file" > "$dest_file"
                     # apply permissions to bash scripts
                     read -r firstline < "$dest_file"
                     if [[ "$firstline" =~ ^#!.*bash ]]; then

hw/rtl/VX_platform.vh

@@ -56,8 +56,25 @@
 `define UNUSED_PIN(x) . x ()
 `define UNUSED_ARG(x) x
 
+`define __SCOPE (* mark_debug="true" *)
+
+`define __SCOPE_X
+
+`define __SCOPE_ON  \
+    `undef __SCOPE_X \
+    `define __SCOPE_X `__SCOPE
+
+`define __SCOPE_OFF  \
+    `undef __SCOPE_X \
+    `define __SCOPE_X
+
 `else // not SYNTHESIS
 
+`define __SCOPE
+`define __SCOPE_X
+`define __SCOPE_ON
+`define __SCOPE_OFF
+
 `ifdef VERILATOR
 
 `ifndef TRACING_ALL

hw/rtl/afu/xrt/VX_afu_wrap.sv

@@ -21,8 +21,8 @@ module VX_afu_wrap #(
 	  parameter C_M_AXI_MEM_DATA_WIDTH  = `VX_MEM_DATA_WIDTH
 ) (
     // System signals
-    input wire ap_clk,
+    input wire clk,
-    input wire ap_rst_n,
+    input wire reset,
 
     // AXI4 master interface
 	`REPEAT (`M_AXI_MEM_NUM_BANKS, GEN_AXI_MEM, REPEAT_COMMA),
@@ -82,8 +82,6 @@ module VX_afu_wrap #(
 	// convert memory interface to array
 	`REPEAT (`M_AXI_MEM_NUM_BANKS, AXI_MEM_TO_ARRAY, REPEAT_SEMICOLON);
 
-	wire reset = ~ap_rst_n;
-
 	reg [`CLOG2(`RESET_DELAY+1)-1:0] vx_reset_ctr;
 	reg [15:0] vx_pending_writes;
 	reg vx_busy_wait;
@@ -122,7 +120,7 @@ module VX_afu_wrap #(
 		end
 	end
 
-	always @(posedge ap_clk) begin
+	always @(posedge clk) begin
 		if (reset || ap_reset) begin
 			state             <= STATE_IDLE;
 			vx_pending_writes <= '0;
@@ -187,7 +185,7 @@ module VX_afu_wrap #(
 		.AXI_DATA_WIDTH (C_S_AXI_CTRL_DATA_WIDTH),
 		.AXI_NUM_BANKS  (C_M_AXI_MEM_NUM_BANKS)
 	) afu_ctrl (
-		.clk       		(ap_clk),
+		.clk       		(clk),
 		.reset     		(reset),
 
 		.s_axi_awvalid  (s_axi_ctrl_awvalid),
@@ -245,7 +243,7 @@ module VX_afu_wrap #(
 	) vortex_axi (
 		`SCOPE_IO_BIND  (1)
 
-		.clk			(ap_clk),
+		.clk			(clk),
 		.reset			(vx_reset),
 
 		.m_axi_awvalid	(m_axi_mem_awvalid_a),
@@ -301,9 +299,32 @@ module VX_afu_wrap #(
 
     // SCOPE //////////////////////////////////////////////////////////////////////
 
+`ifdef CHIPSCOPE
+    ila_afu ila_afu_inst (
+      	.clk (clk),
+		.probe0 ({
+			ap_reset,
+        	ap_start,
+        	ap_done,
+			ap_idle,
+			interrupt
+		}),
+		.probe1 ({
+        	vx_pending_writes,
+			vx_busy_wait,
+			vx_busy,
+			vx_reset,
+			dcr_wr_valid,
+			dcr_wr_addr,
+			dcr_wr_data
+		})
+    );
+`endif
+
 `ifdef DBG_SCOPE_AFU
 	`define TRIGGERS { \
 		reset, \
+		ap_reset, \
 		ap_start, \
 		ap_done, \
 		ap_idle, \
@@ -343,7 +364,7 @@ module VX_afu_wrap #(
 	initial begin
 		$assertoff(0, vortex_axi);
 	end
-	always @(posedge ap_clk) begin
+	always @(posedge clk) begin
 		if (reset) begin
 			assert_delay_ctr <= '0;
 			assert_enabled   <= 0;
@@ -362,7 +383,7 @@ module VX_afu_wrap #(
 `endif
 
 `ifdef DBG_TRACE_AFU
-    always @(posedge ap_clk) begin
+    always @(posedge clk) begin
 		for (integer i = 0; i < C_M_AXI_MEM_NUM_BANKS; ++i) begin
 			if (m_axi_mem_awvalid_a[i] && m_axi_mem_awready_a[i]) begin
 				`TRACE(2, ("%t: AFU Wr Req [%0d]: addr=0x%0h, tag=0x%0h\n", $time, i, m_axi_mem_awaddr_a[i], m_axi_mem_awid_a[i]))

hw/rtl/afu/xrt/vortex_afu.v

@@ -56,8 +56,8 @@ module vortex_afu #(
 		.C_M_AXI_MEM_ADDR_WIDTH  (C_M_AXI_MEM_ADDR_WIDTH),
 		.C_M_AXI_MEM_DATA_WIDTH  (C_M_AXI_MEM_DATA_WIDTH)
 	) afu_wrap (
-		.ap_clk             (ap_clk),
+		.clk             	(ap_clk),
-		.ap_rst_n           (ap_rst_n),
+		.reset           	(~ap_rst_n),
 
 		`REPEAT (`M_AXI_MEM_NUM_BANKS, AXI_MEM_ARGS, REPEAT_COMMA),
 

hw/scripts/ila_insert.tcl

@@ -0,0 +1,231 @@
+######################################################################
+# Automatically inserts ILA instances in a batch flow, and calls "implement_debug_core".   Can also be used in a GUI flow
+# This should ONLY be invoked after synthesis, and before opt_design.   If opt_design is called first, marked nets may be missing and not found
+# Warning: Currently will skip a net if it has no obvious clock domain on the driver.  Nets connected to input buffers will be dropped unless "mark_debug_clock" is attached to the net.
+# Nets attached to VIO cores have the "mark_debug" attribute, and will be filtered out unless the "mark_debug_valid" attribute is attached.
+# Supports the following additional attributes beyond "mark_debug"
+# attribute mark_debug_valid of X : signal is "true";   -- Marks a net for ILA capture, even if net is also attached to a VIO core
+# attribute mark_debug_clock of X : signal is "inst1_bufg/clock";  -- Specifies clock net to use for capturing this net.  May create a new ILA core for that clock domain
+# attribute mark_debug_depth of X : signal is "4096";              -- overrides default depth for this ILA core. valid values: 1024, 2048, ... 132072.   Last attribute that is scanned will win.
+# attribute mark_debug_adv_trigger of X : signal is "true";        -- specifies that advanced trigger capability will be added to ILA core
+# Engineer:  J. McCluskey
+proc insert_ila { depth } {
+    # sequence through debug nets and organize them by clock in the
+    # clock_list array. Also create max and min array for bus indices
+    set dbgs [get_nets -hierarchical -filter {MARK_DEBUG}]
+    if {[llength $dbgs] == 0} {
+        puts "No debug net found. No ILA cores created"
+        return
+    }
+
+    # process list of nets to find and reject nets that are attached to VIO cores.
+    # This has a side effect that VIO nets can't be monitored with an ILA
+    # This can be overridden by using the attribute "mark_debug_valid" = "true" on a net like this.
+    set net_list {}
+    foreach net $dbgs {
+        if { [get_property -quiet MARK_DEBUG_VALID $net] != "true" } {
+            set pin_list [get_pins -of_objects [get_nets -segments $net]]
+            set not_vio_net 1
+            foreach pin $pin_list {
+                if { [get_property IS_DEBUG_CORE [get_cells -of_object $pin]] == 1 } {
+                    # It seems this net is attached to a debug core (i.e. VIO core) already, so we should skip adding it to the netlist
+                    set not_vio_net 0
+                    break
+                }
+            }
+            if { $not_vio_net == 1 } { lappend net_list $net; }
+        } else {
+            lappend net_list $net
+        }
+    }
+
+    # check again to see if we have any nets left now
+    if {[llength $net_list] == 0} {
+        puts "All nets with MARK_DEBUG are already connected to VIO cores. No ILA cores created"
+        return
+    }
+
+    # Now that the netlist has been filtered, determine bus names and clock domains
+    foreach d $net_list {
+        # name is root name of a bus, index is the bit index in the bus
+        set name [regsub {\[[[:digit:]]+\]$} $d {}]
+        set index [regsub {^.*\[([[:digit:]]+)\]$} $d {\1}]
+        if {[string is integer -strict $index]} {
+            if {![info exists max($name)]} {
+                set max($name) $index
+                set min($name) $index
+            } elseif {$index > $max($name)} {
+                set max($name) $index
+            } elseif {$index < $min($name)} {
+                set min($name) $index
+            }
+        } else {
+            set max($name) -1
+        }
+        # Now we search for the local clock net associated with the target net.
+        # There may be ambiguities or no answer in some cases
+        if {![info exists clocks($name)]} {
+            # does MARK_DEBUG_CLOCK decorate this net? If not, then search backwards to the driver cell
+            set clk_name [get_property -quiet MARK_DEBUG_CLOCK $d]
+            if {  [llength $clk_name] == 0 } {
+                # trace to the clock net, tracing backwards via the driver pin.
+                set driver_pin [get_pins -filter {DIRECTION == "OUT" && IS_LEAF == TRUE } -of_objects [ get_nets -segments $d ]]
+                set driver_cell [get_cells -of_objects $driver_pin]
+                if { [get_property IS_SEQUENTIAL $driver_cell] == 1 } {
+                    set timing_arc [get_timing_arcs -to $driver_pin]
+                    set cell_clock_pin [get_pins -filter {IS_CLOCK} [get_property FROM_PIN $timing_arc]]
+                    if { [llength $cell_clock_pin] > 1 } {
+                        puts "Error: in insert_ila. Found more than 1 clock pin in driver cell $driver_cell with timing arc $timing_arc for net $d"
+                        continue
+                    }
+                } else {
+                    # our driver cell is a LUT or LUTMEM in combinatorial mode, we need to trace further.
+                    set paths [get_timing_paths -quiet -through $driver_pin ]
+                    if { [llength $paths] > 0 } {
+                        # note that here we arbitrarily select the start point of the FIRST timing path... there might be multiple clocks with timing paths for this net.
+                        # use MARK_DEBUG_CLOCK to specify another clock in this case.
+                        set cell_clock_pin [get_pins [get_property STARTPOINT_PIN [lindex $paths 0]]]
+                    } else {
+                        # Can't find any timing path, so skip the net, and warn the user.
+                        puts "Critical Warning: from insert_ila.tcl Can't trace any clock domain on driver of net $d"
+                        puts "Please attach the attribute MARK_DEBUG_CLOCK with a string containing the net name of the desired sampling clock, .i.e."
+                        puts "attribute mark_debug_clock of $d : signal is \"inst_bufg/clk\";"
+                        continue
+                    }
+                }
+                # clk_net will usually be a list of net segments, which needs filtering to determine the net connected to the driver pin
+                set clk_net [get_nets -segments -of_objects $cell_clock_pin]
+            } else {
+                set clk_net [get_nets -segments $clk_name]
+                if { [llength $clk_net] == 0 } { puts "MARK_DEBUG_CLOCK attribute on net $d does not match any known net. Please fix."; continue; }
+            }
+            # trace forward to net actually connected to clock buffer output, not any of the lower level segment names
+            set clocks($name) [get_nets -of_objects [get_pins -filter {DIRECTION == "OUT" && IS_LEAF == TRUE } -of_objects $clk_net]]
+            if { [llength $clocks($name)] == 0 } {
+                puts "Critical Warning: from insert_ila.tcl Can't trace any clock domain on driver of net $d"
+                puts "Please attach the attribute MARK_DEBUG_CLOCK with a string containing the net name of the desired sampling clock, .i.e."
+                puts "attribute mark_debug_clock of $d : signal is \"inst_bufg/clk\";"
+                continue
+            }
+            if {![info exists clock_list($clocks($name))]} {
+              # found a new clock
+              puts "New clock found is $clocks($name)"
+              set clock_list($clocks($name)) [list $name]
+              set ila_depth($clocks($name)) $depth
+              set ila_adv_trigger($clocks($name)) false
+            } else {
+              lappend clock_list($clocks($name)) $name
+            }
+            # Does this net have a "MARK_DEBUG_DEPTH" attribute attached?
+            set clk_depth [get_property -quiet MARK_DEBUG_DEPTH $d]
+            if { [llength $clk_depth] != 0 } {
+                set ila_depth($clocks($name)) $clk_depth
+            }
+            # Does this net have a "MARK_DEBUG_ADV_TRIGGER" attribute attached?
+            set trigger [get_property -quiet MARK_DEBUG_ADV_TRIGGER $d]
+            if { $trigger == "true" } {
+                set ila_adv_trigger($clocks($name)) true
+            }
+        }
+    }
+
+    set ila_count 0
+    set trig_out ""
+    set trig_out_ack ""
+
+    if { [llength [array names clock_list]] > 1 } {
+        set enable_trigger true
+    } else {
+        set enable_trigger false
+    }
+
+    foreach c [array names clock_list] {
+        # Now build and connect an ILA core for each clock domain
+        [incr ila_count ]
+        set ila_inst "ila_$ila_count"
+        # first verify if depth is a member of the set, 1024, 2048, 4096, 8192, ... 131072
+        if { $ila_depth($c) < 1024 || [expr $ila_depth($c) & ($ila_depth($c) - 1)] || $ila_depth($c) > 131072 } {
+            # Depth is not right...  lets fix it, and continue
+            if { $ila_depth($c) < 1024 } {
+                set new_depth 1024
+            } elseif { $ila_depth($c) > 131072 } {
+                set new_depth 131072
+            } else {
+                # round value to next highest power of 2, (in log space)
+                set new_depth [expr 1 << int( log($ila_depth($c))/log(2) + .9999 )]
+            }
+            puts "Can't create ILA core $ila_inst with depth of $ila_depth($c)! Changed capture depth to $new_depth"
+            set ila_depth($c) $new_depth
+        }
+        # create ILA and connect its clock
+        puts "Creating ILA $ila_inst with clock $c, capture depth $ila_depth($c) and advanced trigger = $ila_adv_trigger($c)"
+        create_debug_core $ila_inst ila
+        if { $ila_adv_trigger($c) } { set mu_cnt 4; } else { set mu_cnt 2; }
+        set_property    C_DATA_DEPTH   $ila_depth($c) [get_debug_cores $ila_inst]
+        set_property    C_TRIGIN_EN    $enable_trigger [get_debug_cores $ila_inst]
+        set_property    C_TRIGOUT_EN   $enable_trigger [get_debug_cores $ila_inst]
+        set_property    C_ADV_TRIGGER  $ila_adv_trigger($c) [get_debug_cores $ila_inst]
+        set_property    C_INPUT_PIPE_STAGES 1 [get_debug_cores $ila_inst]
+        set_property    C_EN_STRG_QUAL true [get_debug_cores $ila_inst]
+        set_property    ALL_PROBE_SAME_MU true [get_debug_cores $ila_inst]
+        set_property    ALL_PROBE_SAME_MU_CNT $mu_cnt [get_debug_cores $ila_inst]
+        set_property    port_width 1 [get_debug_ports $ila_inst/clk]
+        connect_debug_port $ila_inst/clk $c
+        # hookup trigger ports in a circle if more than one ILA is created
+        if { $enable_trigger == true } {
+            create_debug_port $ila_inst trig_in
+            create_debug_port $ila_inst trig_in_ack
+            create_debug_port $ila_inst trig_out
+            create_debug_port $ila_inst trig_out_ack
+            if { $trig_out != "" } {
+                connect_debug_port $ila_inst/trig_in [get_nets $trig_out]
+            }
+            if { $trig_out_ack != "" } {
+                connect_debug_port $ila_inst/trig_in_ack [get_nets $trig_out_ack]
+            }
+            set trig_out ${ila_inst}_trig_out_$ila_count
+            create_net $trig_out
+            connect_debug_port  $ila_inst/trig_out [get_nets $trig_out]
+            set trig_out_ack ${ila_inst}_trig_out_ack_$ila_count
+            create_net $trig_out_ack
+            connect_debug_port  $ila_inst/trig_out_ack [get_nets $trig_out_ack]
+        }
+        # add probes
+        set nprobes 0
+        foreach n [lsort $clock_list($c)] {
+            set nets {}
+            if {$max($n) < 0} {
+                lappend nets [get_nets $n]
+            } else {
+                # n is a bus name
+                for {set i $min($n)} {$i <= $max($n)} {incr i} {
+                    lappend nets [get_nets $n[$i]]
+                }
+            }
+            set prb probe$nprobes
+            if {$nprobes > 0} {
+                create_debug_port $ila_inst probe
+            }
+            set_property port_width [llength $nets] [get_debug_ports $ila_inst/$prb]
+            connect_debug_port $ila_inst/$prb $nets
+            incr nprobes
+        }
+    }
+
+    # at this point, we need to complete the circular connection of trigger outputs and acks
+    if { $enable_trigger == true } {
+        connect_debug_port ila_1/trig_in [get_nets $trig_out]
+        connect_debug_port ila_1/trig_in_ack [get_nets $trig_out_ack]
+    }
+    set project_found [get_projects -quiet]
+    if { $project_found != "New Project" } {
+        puts "Saving constraints now in project [current_project -quiet]"
+        save_constraints_as debug_constraints.xdc
+    }
+
+    # run ILA cores implementation
+    implement_debug_core
+
+    # write out probe info file
+    write_debug_probes -force debug_nets.ltx
+}

hw/syn/altera/dut/Makefile

@@ -13,7 +13,7 @@ IP_CACHE_DIR := $(ROOT_DIR)/hw/syn/altera/ip_cache/$(DEVICE_FAMILY)
 
 ip-gen: $(IP_CACHE_DIR)/ip_gen.log
 $(IP_CACHE_DIR)/ip_gen.log:
-	$(SCRIPT_DIR)/ip_gen.sh $(IP_CACHE_DIR)
+	$(SCRIPT_DIR)/altera_ip_gen.sh $(IP_CACHE_DIR)
 
 unittest:
 	mkdir -p unittest/$(BUILD_DIR)

hw/syn/altera/opae/Makefile

@@ -102,7 +102,7 @@ all: swconfig ip-gen setup build
 
 ip-gen: $(IP_CACHE_DIR)/ip-gen.log
 $(IP_CACHE_DIR)/ip-gen.log:
-	$(SCRIPT_DIR)/ip_gen.sh $(IP_CACHE_DIR)
+	$(SCRIPT_DIR)/altera_ip_gen.sh $(IP_CACHE_DIR)
 
 swconfig: vortex_afu.h
 vortex_afu.h: $(SRC_DIR)/vortex_afu.json

hw/syn/xilinx/README

@@ -5,7 +5,7 @@ platforminfo -l
 xbutil validate --device 0000:09:00.1 --verbose
 
 # generate FPU IPs
-vivado -mode batch -source scripts/gen_ip.tcl -tclargs ip/xilinx_u50_gen3x16_xdma_5_202210_1
+vivado -mode batch -source xilinx_ip_gen.tcl -tclargs ip/xilinx_u50_gen3x16_xdma_5_202210_1
 
 # build FPGA
 PREFIX=build_base_1c NUM_CORES=1 TARGET=hw_emu PLATFORM=xilinx_u50_gen3x16_xdma_5_202210_1 make > build_u50_hw_emu_base_1c.log 2>&1 &

hw/syn/xilinx/dut/common.mk

@@ -31,9 +31,9 @@ project_1/sources.txt:
 build: $(PROJECT).xpr
 $(PROJECT).xpr: project_1/sources.txt
 ifdef FPU_IP
-	MAX_JOBS=$(JOBS) FPU_IP=project_1/ip $(VIVADO) -mode batch -source $(SRC_DIR)/project.tcl -tclargs $(TOP_LEVEL_ENTITY) $(DEVICE) project_1/sources.txt $(SRC_DIR)/project.xdc $(SCRIPT_DIR) $(SRC_DIR)/../scripts
+	MAX_JOBS=$(JOBS) FPU_IP=project_1/ip $(VIVADO) -mode batch -source $(SRC_DIR)/project.tcl -tclargs $(TOP_LEVEL_ENTITY) $(DEVICE) project_1/sources.txt $(SRC_DIR)/project.xdc $(SCRIPT_DIR)
 else
-	MAX_JOBS=$(JOBS) $(VIVADO) -mode batch -source $(SRC_DIR)/project.tcl -tclargs $(TOP_LEVEL_ENTITY) $(DEVICE) project_1/sources.txt $(SRC_DIR)/project.xdc $(SCRIPT_DIR) $(SRC_DIR)/../scripts
+	MAX_JOBS=$(JOBS) $(VIVADO) -mode batch -source $(SRC_DIR)/project.tcl -tclargs $(TOP_LEVEL_ENTITY) $(DEVICE) project_1/sources.txt $(SRC_DIR)/project.xdc $(SCRIPT_DIR)
 endif
 
 clean:

hw/syn/xilinx/dut/project.tcl

@@ -14,9 +14,9 @@
 # Start time
 set start_time [clock seconds]
 
-if { $::argc != 6 } {
+if { $::argc != 5 } {
   puts "ERROR: Program \"$::argv0\" requires 5 arguments!\n"
-  puts "Usage: $::argv0 <top_module> <device_part> <vcs_file> <xdc_file> <tool_dir> <script_dir>\n"
+  puts "Usage: $::argv0 <top_module> <device_part> <vcs_file> <xdc_file> <tool_dir>\n"
   exit
 }
 
@@ -28,14 +28,12 @@ set device_part [lindex $::argv 1]
 set vcs_file [lindex $::argv 2]
 set xdc_file [lindex $::argv 3]
 set tool_dir [lindex $::argv 4]
-set script_dir [lindex $::argv 5]
 
 puts "Using top_module=$top_module"
 puts "Using device_part=$device_part"
 puts "Using vcs_file=$vcs_file"
 puts "Using xdc_file=$xdc_file"
 puts "Using tool_dir=$tool_dir"
-puts "Using script_dir=$script_dir"
 
 # Set the number of jobs based on MAX_JOBS environment variable
 if {[info exists ::env(MAX_JOBS)]} {
@@ -50,7 +48,7 @@ if {[info exists ::env(FPU_IP)]} {
   set ip_dir $::env(FPU_IP)
   set argv [list $ip_dir $device_part]
   set argc 2
-  source ${script_dir}/gen_ip.tcl
+  source ${tool_dir}/xilinx_ip_gen.tcl
 }
 
 source "${tool_dir}/parse_vcs_list.tcl"

hw/syn/xilinx/gen_xo.tcl

@@ -12,7 +12,7 @@
 # limitations under the License.
 
 if { $::argc != 5 } {
-    puts "ERROR: Program \"$::argv0\" requires 4 arguments!\n"
+    puts "ERROR: Program \"$::argv0\" requires 5 arguments!\n"
     puts "Usage: $::argv0 <xoname> <krnl_name> <vcs_file> <tool_dir> <build_dir>\n"
     exit
 }

hw/syn/xilinx/package_kernel.tcl

@@ -71,65 +71,60 @@ set_property -verbose -name "top" -value ${krnl_name} -objects $obj
 
 if { $chipscope == 1 } {
     # hw debugging
-    create_ip -name axis_ila -vendor xilinx.com -library ip -version 1.1 -module_name ila_afu
+    create_ip -name ila -vendor xilinx.com -library ip -version 6.2 -module_name ila_afu
     set_property -dict [list CONFIG.C_ADV_TRIGGER {true} \
                              CONFIG.C_EN_STRG_QUAL {1} \
-                             CONFIG.C_DATA_DEPTH {4096} \
+                             CONFIG.C_DATA_DEPTH {8192} \
                              CONFIG.C_NUM_OF_PROBES {2} \
                              CONFIG.C_PROBE0_WIDTH {8} \
-                             CONFIG.C_PROBE1_WIDTH {24} \
+                             CONFIG.C_PROBE1_WIDTH {64} \
+                             CONFIG.ALL_PROBE_SAME_MU {false} \
+                             CONFIG.ALL_PROBE_SAME_MU_CNT {2} \
                         ] [get_ips ila_afu]
     generate_target {instantiation_template} [get_files ila_afu.xci]
     set_property generate_synth_checkpoint false [get_files ila_afu.xci]
 
-    create_ip -name axis_ila -vendor xilinx.com -library ip -version 1.1 -module_name ila_fetch
+    create_ip -name ila -vendor xilinx.com -library ip -version 6.2 -module_name ila_fetch
     set_property -dict [list CONFIG.C_ADV_TRIGGER {true} \
                              CONFIG.C_EN_STRG_QUAL {1} \
-                             CONFIG.C_DATA_DEPTH {4096} \
+                             CONFIG.C_DATA_DEPTH {8192} \
                              CONFIG.C_NUM_OF_PROBES {3} \
                              CONFIG.C_PROBE0_WIDTH {128} \
                              CONFIG.C_PROBE1_WIDTH {128} \
                              CONFIG.C_PROBE2_WIDTH {128} \
+                             CONFIG.ALL_PROBE_SAME_MU {false} \
+                             CONFIG.ALL_PROBE_SAME_MU_CNT {2} \
                         ] [get_ips ila_fetch]
     generate_target {instantiation_template} [get_files ila_fetch.xci]
     set_property generate_synth_checkpoint false [get_files ila_fetch.xci]
 
-    create_ip -name axis_ila -vendor xilinx.com -library ip -version 1.1 -module_name ila_issue
+    create_ip -name ila -vendor xilinx.com -library ip -version 6.2 -module_name ila_issue
     set_property -dict [list CONFIG.C_ADV_TRIGGER {true} \
                              CONFIG.C_EN_STRG_QUAL {1} \
-                             CONFIG.C_DATA_DEPTH {4096} \
+                             CONFIG.C_DATA_DEPTH {8192} \
                              CONFIG.C_NUM_OF_PROBES {2} \
                              CONFIG.C_PROBE0_WIDTH {256} \
                              CONFIG.C_PROBE1_WIDTH {128} \
+                             CONFIG.ALL_PROBE_SAME_MU {false} \
+                             CONFIG.ALL_PROBE_SAME_MU_CNT {2} \
                         ] [get_ips ila_issue]
     generate_target {instantiation_template} [get_files ila_issue.xci]
     set_property generate_synth_checkpoint false [get_files ila_issue.xci]
 
-    create_ip -name axis_ila -vendor xilinx.com -library ip -version 1.1 -module_name ila_lsu
+    create_ip -name ila -vendor xilinx.com -library ip -version 6.2 -module_name ila_lsu
     set_property -dict [list CONFIG.C_ADV_TRIGGER {true} \
                              CONFIG.C_EN_STRG_QUAL {1} \
-                             CONFIG.C_DATA_DEPTH {4096} \
+                             CONFIG.C_DATA_DEPTH {8192} \
                              CONFIG.C_NUM_OF_PROBES {4} \
                              CONFIG.C_PROBE0_WIDTH {256} \
                              CONFIG.C_PROBE1_WIDTH {128} \
                              CONFIG.C_PROBE2_WIDTH {288} \
                              CONFIG.C_PROBE3_WIDTH {256} \
+                             CONFIG.ALL_PROBE_SAME_MU {false} \
+                             CONFIG.ALL_PROBE_SAME_MU_CNT {2} \
                         ] [get_ips ila_lsu]
     generate_target {instantiation_template} [get_files ila_lsu.xci]
     set_property generate_synth_checkpoint false [get_files ila_lsu.xci]
-
-    create_ip -name axis_ila -vendor xilinx.com -library ip -version 1.1 -module_name ila_msched
-    set_property -dict [list CONFIG.C_ADV_TRIGGER {true} \
-                             CONFIG.C_EN_STRG_QUAL {1} \
-                             CONFIG.C_DATA_DEPTH {4096} \
-                             CONFIG.C_NUM_OF_PROBES {4} \
-                             CONFIG.C_PROBE0_WIDTH {128} \
-                             CONFIG.C_PROBE1_WIDTH {128} \
-                             CONFIG.C_PROBE2_WIDTH {128} \
-                             CONFIG.C_PROBE3_WIDTH {128} \
-                        ] [get_ips ila_msched]
-    generate_target {instantiation_template} [get_files ila_msched.xci]
-    set_property generate_synth_checkpoint false [get_files ila_msched.xci]
 }
 
 update_compile_order -fileset sources_1

hw/syn/xilinx/sandbox/Makefile

@@ -19,16 +19,16 @@ KERNEL ?= fibonacci
 NCPUS := $(shell lscpu | grep "^Core(s) per socket:" | awk '{print $$4}')
 JOBS ?= $(shell echo $$(( $(NCPUS) > $(MAX_JOBS) ? $(MAX_JOBS) : $(NCPUS) )))
 
-COE_FILE := $(shell realpath kernel.bin.coe)
-ESCAPED_COE_FILE := $(shell echo "$(COE_FILE)" | sed -e 's/[\/&]/\\&/g')
-
 # include paths
 FPU_INCLUDE = -I$(RTL_DIR)/fpu
 ifneq (,$(findstring FPU_FPNEW,$(CONFIGS)))
 	FPU_INCLUDE += -J$(THIRD_PARTY_DIR)/cvfpu/src/common_cells/include -J$(THIRD_PARTY_DIR)/cvfpu/src/common_cells/src -J$(THIRD_PARTY_DIR)/cvfpu/src/fpu_div_sqrt_mvp/hdl -J$(THIRD_PARTY_DIR)/cvfpu/src
 endif
+TEX_INCLUDE = -I$(RTL_DIR)/tex
+RASTER_INCLUDE = -I$(RTL_DIR)/raster
+OM_INCLUDE = -I$(RTL_DIR)/om
 RTL_INCLUDE = -I$(RTL_DIR) -I$(RTL_DIR)/libs -I$(RTL_DIR)/interfaces -I$(RTL_DIR)/core -I$(RTL_DIR)/mem -I$(RTL_DIR)/cache
-RTL_INCLUDE += $(FPU_INCLUDE)
+RTL_INCLUDE += $(FPU_INCLUDE) $(TEX_INCLUDE) $(RASTER_INCLUDE) $(OM_INCLUDE)
 RTL_INCLUDE += -I$(SRC_DIR)
 
 # compilation flags
@@ -43,9 +43,6 @@ CFLAGS += -DSTACK_BASE_ADDR=32\'hFF000
 
 all: build
 
-project2.tcl: project.tcl
-	@sed -e "s/@COE_FILE@/$(ESCAPED_COE_FILE)/g" $< > $@
-
 $(KERNEL).bin:
 	$(MAKE) -C $(ROOT_DIR)/kernel clean
 	STACK_BASE_ADDR=0xFF000 $(MAKE) -C $(ROOT_DIR)/kernel
@@ -61,13 +58,14 @@ project_1/sources.txt:
 	mkdir -p project_1
 	$(SCRIPT_DIR)/gen_sources.sh $(CFLAGS) -P -Cproject_1/src -Oproject_1/sources.txt
 
-build: project_1/project_1.xpr
+build: done.dcp
-project_1/project_1.xpr: project_1/sources.txt kernel.bin.coe project2.tcl
+done.dcp: project_1/sources.txt kernel.bin.coe project.tcl
-	MAX_JOBS=$(JOBS) $(VIVADO) -mode batch -source project2.tcl -tclargs $(DEVICE) project_1/sources.txt $(SCRIPT_DIR)
+	MAX_JOBS=$(JOBS) $(VIVADO) -mode batch -source project.tcl -tclargs $(DEVICE) project_1/sources.txt $(SCRIPT_DIR)
+	echo done > done.dcp
 
 run: project_1/project_1.xpr
 	$(VIVADO) project_1/project_1.xpr &
 
 clean:
-	rm -rf project_1 project2.tcl $(KERNEL).bin kernel.bin.coe
+	rm -rf project_1 project1.tcl $(KERNEL).bin kernel.bin.coe
-	rm -rf .Xil *.log *.jou
+	rm -rf .Xil *.log *.jou *.dcp *.rpt

hw/syn/xilinx/sandbox/project.tcl.in

@@ -11,9 +11,6 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-# Start time
-set start_time [clock seconds]
-
 if { $::argc != 3 } {
     puts "ERROR: Program \"$::argv0\" requires 3 arguments!\n"
     puts "Usage: $::argv0 <device_part> <vcs_file> <tool_dir>\n"
@@ -24,7 +21,7 @@ set device_part [lindex $::argv 0]
 set vcs_file [lindex $::argv 1]
 set tool_dir [lindex $::argv 2]
 
-uuts "Using device_part=$device_part"
+puts "Using device_part=$device_part"
 puts "Using vcs_file=$vcs_file"
 puts "Using tool_dir=$tool_dir"
 
@@ -37,12 +34,8 @@ if {[info exists ::env(MAX_JOBS)]} {
   set num_jobs 0
 }
 
-set origin_dir [file normalize "."]
+proc run_setup {} {
-
+  global device_part vcs_file tool_dir
-# Use origin directory path location variable, if specified in the tcl shell
-if { [info exists ::origin_dir_loc] } {
-  set origin_dir $::origin_dir_loc
-}
 
   # Set the project name
   set project_name "project_1"
@@ -110,12 +103,18 @@ if {[string equal [get_filesets -quiet sim_1] ""]} {
     create_fileset -simset sim_1
   }
 
+  set testbench_file ""
+  foreach file ${vsources_list} {
+    if {[string match "*testbench.v" $file]} {
+        set testbench_file [file normalize $file]
+        break
+    }
+  }
+
   # Set 'sim_1' fileset object
   set obj [get_filesets sim_1]
   # Import local files from the original project
-set files [list \
+  set files [list $testbench_file]
- [file normalize "testbench.v" ]\
-]
   set imported_files [import_files -fileset sim_1 $files]
 
   # Set 'sim_1' fileset file properties for remote files
@@ -301,7 +300,7 @@ set_property -dict [ list \
     CONFIG.Assume_Synchronous_Clk {true} \
     CONFIG.Byte_Size {8} \
     CONFIG.Load_Init_File {true} \
-  CONFIG.Coe_File {@COE_FILE@} \
+    CONFIG.Coe_File {@CURRENTDIR@/hw/syn/xilinx/sandbox/kernel.bin.coe} \
     CONFIG.EN_SAFETY_CKT {true} \
     CONFIG.Enable_32bit_Address {true} \
     CONFIG.Fill_Remaining_Memory_Locations {false} \
@@ -401,7 +400,10 @@ set wrapper_path [make_wrapper -fileset sources_1 -files [ get_files -norecurse
   add_files -norecurse -fileset sources_1 $wrapper_path
 
   update_compile_order -fileset sources_1
+}
 
+proc run_synthesis {} {
+  global num_jobs
   # Synthesis
   if {$num_jobs != 0} {
     launch_runs synth_1 -jobs $num_jobs
@@ -410,9 +412,16 @@ if {$num_jobs != 0} {
   }
   wait_on_run synth_1
   open_run synth_1
-write_checkpoint -force post_synth.dcp
   report_utilization -file utilization.rpt -hierarchical -hierarchical_percentages
 
+  write_checkpoint -force post_synth.dcp
+}
+
+proc run_implementation {} {
+  global tool_dir num_jobs
+  source "${tool_dir}/ila_insert.tcl"
+  insert_ila 8192
+
   # Implementation
   if {$num_jobs != 0} {
     launch_runs impl_1 -jobs $num_jobs
@@ -421,14 +430,40 @@ if {$num_jobs != 0} {
   }
   wait_on_run impl_1
   open_run impl_1
-write_checkpoint -force post_impl.dcp
-
-# Generate reports
   report_place_status -file place.rpt
   report_route_status -file route.rpt
+  write_checkpoint -force post_impl.dcp
+}
+
+proc run_report {} {
+  # Generate reports
   report_timing_summary -file timing.rpt
   report_power -file power.rpt
   report_drc -file drc.rpt
+}
+
+###############################################################################
+
+# Start time
+set start_time [clock seconds]
+
+# Check if the post-implementation checkpoint exists
+if { [file exists post_impl.dcp] } {
+  puts "Resuming from post-implementation checkpoint: post_impl.dcp"
+  open_checkpoint post_impl.dcp
+  run_report
+} elseif { [file exists post_synth.dcp] } {
+  puts "Resuming from post-synthesis checkpoint: post_synth.dcp"
+  open_checkpoint post_synth.dcp
+  run_implementation
+  run_report
+} else {
+  # execute full pipeline
+  run_setup
+  run_synthesis
+  run_implementation
+  run_report
+}
 
 # End time and calculation
 set elapsed_time [expr {[clock seconds] - $start_time}]

hw/syn/xilinx/xrt/Makefile

@@ -122,7 +122,8 @@ ifdef DEBUG
 		VPP_FLAGS += --vivado.prop fileset.sim_1.xsim.elaborate.debug_level=all
 		CFLAGS += -DDEBUG_LEVEL=$(DEBUG) $(DBG_TRACE_FLAGS)
 	else
-		CFLAGS += -DNDEBUG
+		VPP_FLAGS += --debug.chipscope vortex_afu_1
+		CFLAGS += -DNDEBUG -DCHIPSCOPE $(DBG_SCOPE_FLAGS)
 	endif
 else
 	VPP_FLAGS += --optimize 3
@@ -167,7 +168,7 @@ $(BUILD_DIR)/scope.json: $(BUILD_DIR)/vortex.xml
 
 gen-xo: $(XO_CONTAINER)
 $(XO_CONTAINER): $(BUILD_DIR)/sources.txt
-	mkdir -p $(BUILD_DIR); cd $(BUILD_DIR);	$(VIVADO) -mode batch -source $(SRC_DIR)/../scripts/gen_xo.tcl -tclargs ../$(XO_CONTAINER) vortex_afu sources.txt $(SCRIPT_DIR) ../$(BUILD_DIR)
+	mkdir -p $(BUILD_DIR); cd $(BUILD_DIR);	$(VIVADO) -mode batch -source $(SRC_DIR)/gen_xo.tcl -tclargs ../$(XO_CONTAINER) vortex_afu sources.txt $(SCRIPT_DIR) ../$(BUILD_DIR)
 
 gen-bin: $(XCLBIN_CONTAINER)
 $(XCLBIN_CONTAINER): $(XO_CONTAINER) $(SCOPE_JSON)

sim/simx/cache_sim.cpp

@@ -514,6 +514,7 @@ public:
 				bank_req.type  = bank_req_t::Core;
 				bank_req.write = core_req.write;
 				pipeline_req   = bank_req;
+				DT(3, simobject_->name() << " core-req: " << core_req);
 			}
 
 			if (core_req.write)