Merge branch 'bump-rocket' into 'master'

Bump rocket

See merge request pard/labeled-RISC-V!230

.gitignore

@@ -10,3 +10,14 @@ ControlPlanes.v
 src/main/scala/.idea/
 src/main/scala/build.sbt
 src/main/scala/project/
+
+.idea/
+boom-ether.err
+bootrom/bootrom.txt
+emulator/boom-failed.md
+emulator/log.txt
+emulator/serial6*
+fpga/.Xil/
+fpga/bbl.elf.txt
+fpga/vivado_*.str
+fpga/vmlinux.txt

.gitmodules

@@ -1,6 +1,3 @@
-[submodule "riscv-tools"]
-	path = riscv-tools
-	url = https://github.com/riscv/riscv-tools.git
 [submodule "hardfloat"]
 	path = hardfloat
 	url = https://github.com/ucb-bar/berkeley-hardfloat.git
@@ -13,3 +10,6 @@
 [submodule "firrtl"]
 	path = firrtl
 	url = https://github.com/ucb-bar/firrtl.git
+[submodule "src/main/scala/boom"]
+	path = src/main/scala/boom
+	url = https://github.com/shinezyy/BOOM-inside-LvNA.git

.travis.yml

@@ -70,8 +70,8 @@ jobs:
     - &test
       stage: Test
       script:
-        - travis_wait 80 make emulator-ndebug -C regression SUITE=UnittestSuite JVM_MEMORY=3G
-        - travis_wait 80 make emulator-regression-tests -C regression SUITE=UnittestSuite JVM_MEMORY=3G
+        - travis_wait 100 make emulator-ndebug -C regression SUITE=UnittestSuite JVM_MEMORY=3G VERILATOR_THREADS=1
+        - travis_wait 100 make emulator-regression-tests -C regression SUITE=UnittestSuite JVM_MEMORY=3G VERILATOR_THREADS=1
     - <<: *test
       script:
         - travis_wait 80 make emulator-ndebug -C regression SUITE=JtagDtmSuite JVM_MEMORY=3G

bootrom/Makefile

@@ -13,3 +13,4 @@ all: $(bootrom_img)
 
 %.elf: %.S linker.ld
 	$(GCC) -Tlinker.ld $< -nostdlib -static -Wl,--no-gc-sections -o $@
+	riscv64-unknown-elf-objdump -d $@ >bootrom.txt

bootrom/bootrom.S

@@ -1,18 +1,20 @@
+// #define DRAM_BASE 0x80000000
 #define DRAM_BASE 0x100000000
 
 .section .text.start, "ax", @progbits
 .globl _start
 _start:
-  csrwi 0x7c1, 0 // disable chicken bits
+  // csrwi 0x7c1, 0 // disable chicken bits
   li s0, DRAM_BASE
   csrr a0, mhartid
   la a1, _dtb
+
   jr s0
 
 .section .text.hang, "ax", @progbits
 .globl _hang
 _hang:
-  csrwi 0x7c1, 0 // disable chicken bits
+  // csrwi 0x7c1, 0 // disable chicken bits
   csrr a0, mhartid
   la a1, _dtb
   csrwi mie, 0

build.sbt

@@ -16,7 +16,7 @@ lazy val commonSettings = Seq(
   traceLevel   := 15,
   scalacOptions ++= Seq("-deprecation","-unchecked","-Xsource:2.11"),
   libraryDependencies ++= Seq("org.scala-lang" % "scala-reflect" % scalaVersion.value),
-  libraryDependencies ++= Seq("org.json4s" %% "json4s-jackson" % "3.5.3"),
+  libraryDependencies ++= Seq("org.json4s" %% "json4s-jackson" % "3.6.1"),
   addCompilerPlugin("org.scalamacros" % "paradise" % "2.1.0" cross CrossVersion.full)
 )
 

emulator/Makefile

@@ -14,11 +14,13 @@ LDFLAGS := $(LDFLAGS) -L$(RISCV)/lib -Wl,-rpath,$(RISCV)/lib -L$(abspath $(sim_d
 
 emu = emulator-$(PROJECT)-$(CONFIG)
 emu_debug = emulator-$(PROJECT)-$(CONFIG)-debug
+origin_emu = emulator-$(PROJECT)-$(CONFIG)-origin # without init mem
 
 include $(sim_dir)/Makefrag-verilator
 
 all: $(emu)
 debug: $(emu_debug)
+emu_origin: $(origin_emu)
 
 clean:
 	rm -rf *.o *.a emulator-* $(generated_dir) $(generated_dir_debug) DVEfiles $(output_dir)
@@ -33,11 +35,11 @@ ifneq ($(filter run% %.run %.out %.vpd %.vcd,$(MAKECMDGOALS)),)
 -include $(generated_dir)/$(long_name).d
 endif
 
-$(output_dir)/%.run: $(output_dir)/% $(emu)
-	./$(emu) +max-cycles=$(timeout_cycles) $< 2> /dev/null 2> $@ && [ $$PIPESTATUS -eq 0 ]
+$(output_dir)/%.run: $(output_dir)/% $(origin_emu)
+	./$(origin_emu) +max-cycles=$(timeout_cycles) $< 2> /dev/null 2> $@ && [ $$PIPESTATUS -eq 0 ]
 
-$(output_dir)/%.out: $(output_dir)/% $(emu)
-	./$(emu) +max-cycles=$(timeout_cycles) +verbose $< $(disasm) $@ && [ $$PIPESTATUS -eq 0 ]
+$(output_dir)/%.out: $(output_dir)/% $(origin_emu)
+	./$(origin_emu) +max-cycles=$(timeout_cycles) +verbose $< $(disasm) $@ && [ $$PIPESTATUS -eq 0 ]
 
 $(output_dir)/%.vcd: $(output_dir)/% $(emu_debug)
 	./$(emu_debug) +max-cycles=$(timeout_cycles) +verbose -v$@ $< $(disasm) $(patsubst %.vcd,%.out,$@) && [ $$PIPESTATUS -eq 0 ]

emulator/Makefrag-verilator

@@ -8,7 +8,11 @@ verilog = \
   $(generated_dir)/$(long_name).v \
   $(generated_dir)/$(long_name).behav_srams.v \
 
-.SECONDARY: $(firrtl) $(verilog)
+origin_verilog = \
+  $(generated_dir)/$(long_name).v \
+  $(generated_dir)/$(long_name).origin_behav_srams.v \
+
+.SECONDARY: $(firrtl) $(verilog) $(origin_verilog)
 
 $(generated_dir)/%.fir $(generated_dir)/%.d: $(FIRRTL_JAR) $(chisel_srcs) $(bootrom_img)
 	mkdir -p $(dir $@)
@@ -24,8 +28,12 @@ $(generated_dir)/$(long_name).behav_srams.v : $(generated_dir)/$(long_name).conf
   $(mem_init) [email protected] mem_ext bin.txt $(generated_dir)/$(long_name).json > $@
 	rm -f [email protected]
 
+$(generated_dir)/$(long_name).origin_behav_srams.v : $(generated_dir)/$(long_name).conf $(VLSI_MEM_GEN)
+	cd $(generated_dir) && \
+	$(VLSI_MEM_GEN) $(generated_dir)/$(long_name).conf > $@
+
 # Build and install our own Verilator, to work around versionining issues.
-VERILATOR_VERSION=4.004
+VERILATOR_VERSION=4.008
 VERILATOR_SRCDIR ?= verilator/src/verilator-$(VERILATOR_VERSION)
 VERILATOR_TARGET := $(abspath verilator/install/bin/verilator)
 INSTALLED_VERILATOR ?= $(VERILATOR_TARGET)
@@ -55,13 +63,14 @@ verilator: $(INSTALLED_VERILATOR)
 
 # Run Verilator to produce a fast binary to emulate this circuit.
 VERILATOR := $(INSTALLED_VERILATOR) --cc --exe
+VERILATOR_THREADS ?= 2
 VERILATOR_FLAGS := --top-module $(MODEL) \
   +define+PRINTF_COND=\$$c\(\"verbose\",\"\&\&\"\,\"done_reset\"\) \
   +define+RANDOMIZE_GARBAGE_ASSIGN \
   +define+STOP_COND=\$$c\(\"done_reset\"\) --assert \
   --output-split 20000 \
   --output-split-cfuncs 20000 \
-  --threads 4 \
+  --threads $(VERILATOR_THREADS) \
 	-Wno-STMTDLY --x-assign unique \
   -I$(vsrc) \
   -O3 -CFLAGS "$(CXXFLAGS) -DVERILATOR -DTEST_HARNESS=V$(MODEL) -include $(csrc)/verilator.h -include $(generated_dir)/$(PROJECT).$(CONFIG).plusArgs"
@@ -84,3 +93,12 @@ $(emu_debug): $(verilog) $(cppfiles) $(headers) $(generated_dir)/$(long_name).d
 	-o $(abspath $(sim_dir))/$@ $(verilog) $(cppfiles) -LDFLAGS "$(LDFLAGS)" \
 	-CFLAGS "-I$(generated_dir_debug) -include $(model_header_debug)"
 	$(MAKE) VM_PARALLEL_BUILDS=1 -C $(generated_dir_debug)/$(long_name) -f V$(MODEL).mk
+
+$(origin_emu): $(origin_verilog) $(cppfiles) $(headers) $(INSTALLED_VERILATOR)
+	mkdir -p $(generated_dir)/$(long_name)
+	$(VERILATOR) $(VERILATOR_FLAGS) -Mdir $(generated_dir)/$(long_name) \
+	-o $(abspath $(sim_dir))/$@ $(origin_verilog) $(cppfiles) -LDFLAGS "$(LDFLAGS)" \
+	-CFLAGS "-I$(generated_dir) -include $(model_header)"
+	$(MAKE) VM_PARALLEL_BUILDS=1 -C $(generated_dir)/$(long_name) -f V$(MODEL).mk
+
+

fpga/emu/Makefile

@@ -5,6 +5,8 @@ build_dir = $(fpga_dir)/build
 generated_dir = $(build_dir)/generated-src
 
 CONFIG ?= LvNAConfigemu
+NUMA_CONFIG ?=
+SEED ?=
 
 -include $(base_dir)/Makefrag
 
@@ -19,6 +21,8 @@ emu_gen_script = gen_bin.sh
 emu_bin_hex_file = $(build_dir)/bin.txt
 nohype_dtb = $(build_dir)/c0.dtb
 nohype_dtb_hex_file = $(build_dir)/c0.dtb.txt
+# max_cycles = 40000000
+max_cycles = 2209069
 
 $(emu): $(original_emu)
 	ln -sf $< $@
@@ -46,7 +50,8 @@ DEBUG_ARGS = +jtag_rbb_enable=1 -r 4040
 endif
 
 run-emu: $(emu) $(emu_bin_hex_file) $(nohype_dtb_hex_file)
-	cd $(dir $(emu)) && $(emu) $(DEBUG_ARGS) +verbose . 3>&1 1>&2 2>&3 | spike-dasm > $(dir $(emu))/emu.log
+	cd $(dir $(emu)) && time $< $(DEBUG_ARGS) $(SEED) +verbose -m $(max_cycles) . 3>&1 1>&2 2>&3 \
+		| spike-dasm > $(dir $(emu))/emu.log
 
 dtb-clean:
 	-rm -f $(build_dir)/c*.dtb{,.txt}

fpga/emu/gen_dtb_bin.sh

@@ -1,5 +1,7 @@
 #!/usr/bin/env bash
+
 BUILD_DIR=$1
+
 for i in $(ls $1/c*.dtb)
 do
     sh gen_bin.sh $i $i.txt

fpga/emu/python/.gitignore

@@ -0,0 +1,2 @@
+ts
+__pycache__

fpga/emu/python/common.py

@@ -0,0 +1,115 @@
+import sh
+import os
+import errno
+from pathlib import Path
+import binascii as ba
+from os.path import join as pjoin
+
+
+# hex_dump == hexdump -ve '2/ "%08x " "\n"' $1 | awk '{print $2$1}' > $2
+def hex_dump(in_file, out_file):
+    assert os.path.isfile(in_file)
+    print(in_file)
+    with open(in_file, 'rb') as f, open(out_file, 'w') as out_f:
+        chunks = iter(lambda : f.read(1), b'')
+        hex_bytes = map(ba.hexlify, chunks)
+
+        buffer = []
+        for b in hex_bytes:
+            buffer.append(b.decode())
+            if len(buffer) == 8:
+                buffer.reverse()
+                out_f.write(''.join(buffer) + '\n')
+                buffer.clear()
+
+
+def safe_mkdir(dir: str):
+    if not os.path.exists(dir):
+        try:
+            os.makedirs(dir)
+        except OSError as e:
+            # when another thread create this directory
+            assert e.errno == errno.EEXIST
+
+
+def is_older(left, right):
+    if not os.path.isfile(right):
+        return False
+    assert os.path.isfile(left)
+    return os.path.getmtime(left) < os.path.getmtime(right)
+
+
+def avoid_repeating(
+        emu: str, input_file: str, output_dir: str,
+        force_run: bool, ok, func):
+    # this function avoid repeated run for most common situations
+    input_file_n = os.path.basename(input_file)
+    emu_n = os.path.basename(emu)
+
+    # python 3.6 please
+    running_lock_name = f'running-{func.__name__}-{emu_n}-{input_file_n}'
+    running_lock_file = pjoin(output_dir, running_lock_name)
+
+    if os.path.isfile(running_lock_file):
+        print('running lock found in {}, skip!'.format(output_dir))
+        return
+    else:
+        sh.touch(running_lock_file)
+
+    script_ts_dir = pjoin(str(Path(__file__).resolve().parent), 'ts')
+    safe_mkdir(script_ts_dir)
+    # python 3.6 please
+    ts_name = f'ts-{func.__name__}-{emu_n}-{input_file_n}'
+
+    out_ts_file = pjoin(output_dir, ts_name)
+    if os.path.isfile(out_ts_file):
+        out_ts = os.path.getmtime(out_ts_file)
+    else:
+        out_ts = 0.0
+
+    script_ts_file = pjoin(script_ts_dir, ts_name)
+    if not os.path.isfile(script_ts_file):
+        sh.touch(script_ts_file)
+    script_ts = os.path.getmtime(script_ts_file)
+
+    program_ts = os.path.getmtime(emu)
+    input_binary_ts = os.path.getmtime(input_file)
+
+    cmd_ts = max(script_ts, program_ts, input_binary_ts)
+
+    if out_ts < cmd_ts or force_run:
+        try:
+            func()
+            sh.touch(out_ts_file)
+        except Exception as e:
+            if (ok()):
+                sh.touch(out_ts_file)
+            else:
+                print(e)
+        sh.rm(running_lock_file)
+    else:
+        print('none of the inputs is older than output, skip!')
+        if os.path.isfile(running_lock_file):
+            sh.rm(running_lock_file)
+
+
+def get_am_apps():
+    assert 'AM_HOME' in os.environ.keys()
+    app_dir = pjoin(os.environ['AM_HOME'], 'apps')
+    apps = set(os.listdir(app_dir))
+    with open('./am_app_filter.txt') as f:
+        filters = set([line.strip() for line in f.readlines()])
+    return list(apps & filters)
+
+
+def get_am_app_dir(app: str):
+    assert 'AM_HOME' in os.environ.keys()
+    app_base_dir = pjoin(os.environ['AM_HOME'], 'apps')
+    app_dir = pjoin(app_base_dir, app)
+    assert os.path.isdir(app_dir)
+    return app_dir
+
+
+if __name__ == '__main__':
+    hex_dump('/home/zyy/projects/nexus-am/apps/cache-flush/build/cache-flush'
+             '-riscv64-rocket.bin', 'test.txt')