【转】SYNOPSYS VCS Makefile文件编写与研究
原文地址:http://www.cnblogs.com/zhtxwd/archive/2012/03/30/2425180.html
YNOPSYS VCS Makefile文件编写与研究 这个Makefile是synopsys提供的模板,看上去非常好用,你只要按部就班提供实际项目的参数就可以了。我们来看这个文件的头部说明: makefile 其实完全可以用csh或其他脚本来编写,只是VCS使用的linux内置的make命令定义了一个标准的仿真脚本,make命令是专门用来 做项目的源文件管理和编译控制的命令。这篇文章重点看synpsys的标准仿真脚本都做了哪些操作,然后使用其他脚本来实现。这里主要是自己 写的一点东西,有些地方是猜测的或者不准确。 #--------------------------------------------------------------------------------------------------------------------------- # SYNOPSYS CONFIDENTIAL - This is an unpublished, proprietary work of # Synopsys, Inc., and is fully protected under copyright and trade secret # laws. You may not view, use, disclose, copy, or distribute this file or # any information contained herein except pursuant to a valid written # license from Synopsys. # SYNOPSYS公司的版权声明,没有权限不可使用 #----------------------------------------------------------------------------------------------------------------------------- # Filename : $Id: Makefile,v 1.0 2006/07/18 23:59:59 vangundy Exp $ # Created by : Synopsys Inc. 07/17/2006 # $Author : vangundy $ # Description : Demonstrates Verilog DUT and SVTB using VCS # makefile文件头 #--------------------------------------------------------------------------------------------------------------------------- # The Makefile works on two seperate flows. The DEBUG flow is intended to be used # During debugging of a testcase and/or the DUT. The REGRESSION flow is used # During regression runs and collects coverage data. # 该makefile模版包括两部分流程,debug(查错)流程和regress(回归测试)流程,两个流程大致步骤都相同都是:Compile,SIM(urg,覆盖 # 率的分析和采集),debug时主要是跑一个pattern,并dump VPD文件,SIM的同时可以打开DVE视图界面,结束后观察波形,regress主要用 # 于采集覆盖率,一般要跑多个pattern,这时就无需dump VPD文件(节约时间),由于是debug后有进行的重复运行,所以叫regress(回归)。 # 在我们的验证平台中,若不做代码覆盖率的功能,可以不写regress,只要写debug的流程和跑多个pattern的脚本就好了。 #--------------------------------------------------------------------------------------------------------------------- # The DEBUG flow turns on VPD dumping and turns off coverage collection. After # building a testcase using the debug targets, you can debug the TB and the DUT # source code using the testbench debugger and DVE. Of course, you can turn on # coverage metrics and run in debug mode by changing compile and runtime options # in the makefile. These changes are independent of the regression flow so that # the regressions will still run optimally without the interference of VPD dumping. # debug流程打开VPD文件的dump并关闭覆盖率在build了一个包含DUT的testcase后,可以使用VCS的debugger和DVE进行debug。当 # 然,你也以通过改变makefile文件中的compile和runtime选项参数来开启覆盖率功能。Debug流程和regress流程是各自独立的,regression # 流程一般不生成VPD。 # -------------------------------------------------------------------------------------------------------------------------------- # The REGRESSION flow turns off VPD dumping and turns on Coverage Metrics and TB # coverage collection. This flow is intended to support verification engineers who # are working through the regression process and are interested in coverage # collection and urg. # REGRESSION流程关闭VPD dump并打开Coverage collection功能,该流程是为了支持验证引擎进行“流水线验证“(跑多个testcase)和 # 代码覆盖率功能。??在验证平台中可以将运行多个testcase的脚本命名为regress,运行单个testcase的脚本命名为regone??,这只是 # synopsys的模版,我们不必完全遵守,可以不区分debug和regress,然后将是否打开波形和coverage设置成参数。 # ------------------------------------------------------------------------------------------------------------------------------- # Command Line make命令行 # ----------------- # The Makefile supports the following command line # makefile支持下列命令行 # % make target_name_* <SEED=xxx> <DEFINES=xxxx> # makefile文件放在哪?放在仿真路径。make [-f makefile文件名][选项][宏定义][目标] # -f 指定makefile 若没有则make程序首先在当前目录查找名为makefile的文件,如果没有找到,它就会转而查找名为Makefile的文件。 # Where target_name is the name of a testcase located in the test directory. Every # test in the test directory is named using test_{test_name}. All of the test targets # are listed in the TEST TARGETS section of the makefile. # target_name是test路径下的一个testcase的名字,test路径下的testcase的名字使用test_{test_name}来命名,例如test_1 # 所有的test target 都在makefile文件中的TEST TARGETS部分列出 # --------------------------------------------------------------------------------------------------------------------------- # Compile and Run Testcases 编译与运行testcase # ------------------------------- # To compile and run a tescase use the test_* and regress_test_* targets. # 编译与运行testcase,(test_1 就是执行了下面的两个命令先编译在运行) test_1 ==> compile_1 run_1 详见下面命令定义 # % make test_1 // Builds and runs test 1 with VPD dumping 其实就是debug的前边的流程 # % make regress_test_1 // Builds and runs test 1 with coverage turned on # ------------------------------------------------------------------------------------------------------------------------- # Debugging Testcases Debug 实在上面命令之后在进行的 # ------------------------ # You can use DVE and the testbench debugger to visualize waveforms and testbench # execution. You must first build the testbench using the make compile_* command. # dubug必须是在DVE(VCS的debug工具,与debussy一样的功能)下进行,因为要看波形嘛,但是debug之前必须先compile # % make compile_1 // Builds test 1 for debugging //需要重新编译一次吗? # Once you have built the environment with the proper debug switches, you can use DVE and the testbench debugger. # # testbench debugger 是否是指编译后的那个simv可执行文件呢? 其实gui_1 和上面test_1中的run_1是一样的只是增加了-gui项 # 即增加了打开gui界面的参数,其他雷同 # % make gui_1 // Debug test 1 with DVE # % make tb_gui_1 // Debug test 1 with the testbench debugger # % make both_guis_1 // Debug using both guis # % make pp_1 // Debug using the VPD file VPD文件要在执行simv之后才有吧? # If you want, you can turn on coverage for the DEBUG flow by uncommenting the # coverage flag in the makefile. If you do this, you can still look at coverage. # This may be useful in helping those who are debugging coverage related issues. # 如果在makefile中的debug流程中使用了coverage功能,那么可以使用下面命令观察覆盖率 # % make urg // Visualize coverage data from debug runs # ----------------------------------------------------------------------------------------------------------------------------- # Regression Testcases # -------------------- # Regression tests are used to collect coverage information. To build a testcase # for coverage collection use a command similar to the following. # regress流程主要是为了收集代码覆盖率信息,在执行regress之前需要重新build testcase 类似debug时的compile # % make regress_build_1 // Build and run a regression test with a default seed # Once the test has been built, you can run it again with a new seed. # 与debug不同的是regress需要重新run(使用新的SEED)一下,【还是debug的时候也要run一下?】 # % make regress_run_1 SEED=1234 # After running one or more regression runs, you can visualize the coverage data # using urg and the following command # run完之后可以用下面命令看代码覆盖率 # % make regress_urg #---------------------------------------------------------------------------------------------------------------------------- # HOW TO REUSE THIS FILE ON ANOTHER DUT //如何重用该模版 # STEP 1: Update the file locations as required //设置file所属的路径 # STEP 2: Update the DUT section with directory and source location info//更新模版中DUT部分,指定DUT的路径和include的路径 # STEP 3: Update the TB section with directory and source location info//更新模版中TB部分,指定TB的路径和include的路径 # STEP 4: Update the Coverage section with name of dut top (eg top.dut) //跟新模版中Coverage部分,指定要测试代码覆盖率的dut的top # STEP 5: Add test targets to the debug and regression targets section//将debug和regress的target加入模版中对应的部分 # STEP 5: Adjust the debug and regression compile and run time arguments//调整debug和regress的compile和runtime的命令参数 # STEP 7: Adjust command line options as required//调整命令行命令(后边带百分号和冒号的就表示可以在make命令行中使用的命令) # STEP 8: Update the env class so that it extends dkm_env//更新env class(环境类)使得它可以提供dkm_env # You will need to have a copy of the dkm directory and it should //dkm是什么? # be located at $(TB_SRC_DIR)/dkm # a) Add [`include "dkm_env.sv"] # b) Add [extends dkm_env] to the environment class definition # c) Call the super.new("name") from the constructor # STEP 9: Run the debug and regression targets # % make testbench_target_* // testbench_target_* 是指test_1之类的testcase #----------------------------------------------------------------------------- 看了上文,大家应该可以简单了解这个Makefile的功能了。接下来就按照step1~9来填空即可: .PHONY : default help clean regress_clean default: help #----------------------------------------------------------------------------- # DIRECTORIES 总路径 #----------------------------------------------------------------------------- OUTPUT_DIR = ./output //为什么debug和regress没有分开? debug和regress的Coverage在COV_DIR下是有区分的 COV_DIR = ./coverage //为什么没有VPD的路径? LOG_DIR = ./logs //output是做什么的? VPD文件 simv文件 还有一写其他文件在这里 # Set this to the location where you installed the designware models. This # depends on whether you ran the setup_vip_dw_home to install the models or # the setup_vip_here script. #DW_MODELS_DIR = $(DESIGNWARE_HOME) DW_MODELS_DIR = /user/synopsys/designware //软件路径 VCS的路径? #DESIGNWARE_HOME = ~synopsys/bk/designware #DW_MODELS_DIR = ./designware #---------------------------------------------------------------------------- # DEVICE UNDER TEST DUT路径 #----------------------------------------------------------------------------- DUT_SRC_DIR = ./source/Verilog //SRC source DUT_SRC = -f $(DUT_SRC_DIR)/rtl_list.f //待编译的rtl文件列表文件 SRC是什么意思? DUT_INC += +incdir+/user/myproj/PROJECT/RTL/SRC/mymodule/ DUT_INC += +incdir+/user/myproj/PROJECT/RTL/SRC/mymodule/mymodule_inc.v DUT_CMP_OPTIONS += +libext+.v+.V //这个参数是干什么的?指定VCS搜索文件时的文件后缀.v #DUT_CMP_OPTIONS += -timescale=1ps/1ps //CMP是compile的意思 不是compare #DUT_CMP_OPTIONS += -override_timescale=1ps/1ps #----------------------------------------------------------------------------- # TESTBENCH TB路径设置 #----------------------------------------------------------------------------- TB_SRC_DIR = ./source/svtb # AXI TESTBENCH, VIP Sources first #TB_SRC += -f $(TB_SRC_DIR)/mac_if_tb/vip/gslv_model_package.f TB_SRC += $(TB_SRC_DIR)/mpdu_trx_tb/tests/mpdu_tb_top.sv TB_SRC += $(TB_SRC_DIR)/mpdu_trx_tb/tests/$(TB_TEST).sv //为什么有两个SV TB_INC += +incdir+$(TB_SRC_DIR)/mpdu_trx_tb/vip TB_INC += +incdir+$(TB_SRC_DIR)/mpdu_trx_tb/env TB_INC += +incdir+$(TB_SRC_DIR)/mpdu_trx_tb/tests TB_INC += +incdir+$(DW_MODELS_DIR)/include/svtb //这部分为了支持sv吗? TB_INC += +incdir+$(DW_MODELS_DIR)/include/verilog TB_INC += +incdir+$(DW_MODELS_DIR)/svtb #TB_CMP_OPTIONS += -tb_timescale=1ns/1ps #TB_CMP_OPTIONS += -lca Y-2006.06-SP2 TB_CMP_OPTIONS += +pkgdir+$(DW_MODELS_DIR)/include/svtb //TB的编译选项和DUT不同?sv 和verilog的区别吗? TB_CMP_OPTIONS += -ntb_incdir $(DW_MODELS_DIR)/include/vera TB_CMP_OPTIONS += -ntb_incdir $(DESIGNWARE_HOME)/vip/vmt/latest/vera/src TB_CMP_OPTIONS += -ntb_incdir $(DESIGNWARE_HOME)/vip/amba/latest/vera/src TB_CMP_OPTIONS += -ntb_incdir $(DESIGNWARE_HOME)/vip/amba/latest/axi_master_vmt/vera/src //TB中使用的一些模型 TB_CMP_OPTIONS += -ntb_incdir $(DESIGNWARE_HOME)/vip/amba/latest/axi_master_rvm_vera_vmt/vera/src TB_CMP_OPTIONS += -ntb_incdir $(DESIGNWARE_HOME)/vip/amba/latest/axi_slave_vmt/vera/src TB_CMP_OPTIONS += -ntb_incdir $(DESIGNWARE_HOME)/vip/amba/latest/axi_slave_rvm_vera_vmt/vera/src TB_CMP_OPTIONS += -ntb_incdir $(DESIGNWARE_HOME)/vip/amba/latest/axi_monitor_vmt/vera/src TB_CMP_OPTIONS += -ntb_incdir $(DESIGNWARE_HOME)/vip/amba/latest/axi_monitor_rvm_vera_vmt/vera/src TB_CMP_OPTIONS += -ntb_incdir $(DESIGNWARE_HOME)/vip/amba/latest/axi_port_monitor_vmt/vera/src TB_CMP_OPTIONS += -ntb_incdir $(DESIGNWARE_HOME)/vip/amba/latest/axi_port_monitor_rvm_vera_vmt/vera/src TB_CMP_OPTIONS += -ntb_incdir $(DESIGNWARE_HOME)/vip/amba/latest/axi_interconnect_vmt/vera/src TB_CMP_OPTIONS += -ntb_incdir $(DESIGNWARE_HOME)/vip/amba/latest/axi_interconnect_rvm_vera_vmt/vera/src TB_CMP_OPTIONS += -ntb_incdir ${DESIGNWARE_HOME}/vip/amba/latest/ahb_master_vmt/vera/src TB_CMP_OPTIONS += -ntb_incdir ${DESIGNWARE_HOME}/vip/amba/latest/ahb_slave_vmt/vera/src TB_CMP_OPTIONS += -ntb_incdir ${DESIGNWARE_HOME}/vip/amba/latest/ahb_monitor_vmt/vera/src TB_CMP_OPTIONS += -ntb_incdir ${DESIGNWARE_HOME}/vip/amba/latest/ahb_bus_vmt/vera/src TB_CMP_OPTIONS += -ntb_incdir ${DESIGNWARE_HOME}/vip/amba/latest/ahb_master_rvm_vera_vmt/vera/src TB_CMP_OPTIONS += -ntb_incdir ${DESIGNWARE_HOME}/vip/amba/latest/ahb_slave_rvm_vera_vmt/vera/src TB_CMP_OPTIONS += -ntb_incdir ${DESIGNWARE_HOME}/vip/amba/latest/ahb_monitor_rvm_vera_vmt/vera/src TB_CMP_OPTIONS += -ntb_incdir ${DESIGNWARE_HOME}/vip/amba/latest/ahb_bus_rvm_vera_vmt/vera/src TB_CMP_OPTIONS += -ntb_define NTB TB_CMP_OPTIONS += -ntb_define DW_VIP_AXI_MAX_NO_MSTRS=6 TB_CMP_OPTIONS += -ntb_define DW_VIP_AXI_MAX_NO_SLVS=2 TB_CMP_OPTIONS += +define+DW_VIP_AXI_MAX_NO_MSTRS_6 TB_CMP_OPTIONS += +define+DW_VIP_AXI_MAX_NO_SLVS_2 TB_CMP_OPTIONS += -ntb_opts rvm TB_CMP_OPTIONS += -ntb_opts dtm TB_CMP_OPTIONS += -ntb_opts use_sigprop TB_CMP_OPTIONS += -ntb_opts interop TB_CMP_OPTIONS += -ntb_opts dw_vip TB_CMP_OPTIONS += +define+NT # AIP Related files and compilation options #TB_CMP_OPTIONS += +incdir+../BP062-BU-01000-r0p0-00rel0/sva \ +incdir+../BP062-BU-01000-r0p0-00rel0/verilog \ ../BP062-BU-01000-r0p0-00rel0/sva/AxiPC.sv \ ../BP062-BU-01000-r0p0-00rel0/verilog/Axi.v \ ./source/svtb/platform_tb/env/Snps_ARMAXI_CheckerBind.sv #${VCS_HOME}/packages/aip/DDR2_AIP/src/Snps_DDR2_Checker.sv \ -assert enable_diag \ +incdir+.+${VCS_HOME}/packages/aip/DDR2_AIP/src/ \ ./source/svtb/platform_tb/env/Snps_DDR2_Bind.sv \ +incdir+../BP062-BU-01000-r0p0-00rel0/sva \ +incdir+../BP062-BU-01000-r0p0-00rel0/verilog \ ../BP062-BU-01000-r0p0-00rel0/sva/AxiPC.sv \ ../BP062-BU-01000-r0p0-00rel0/verilog/Axi.v \ ./source/svtb/platform_tb/env/Snps_ARMAXI_CheckerBind.sv #----------------------------------------------------------------------------- # COVERAGE 覆盖率的设置 #----------------------------------------------------------------------------- COV_TREE += '+tree mpdu_tb_top' COV_CM_OPTIONS += -cm line+cond+fsm+assert 注意CM和CMP不一样 #----------------------------------------------------------------------------- # TEST TARGETS 总命令 #----------------------------------------------------------------------------- # debug targets test_1: compile_1 run_1 与前边对应test_1 就是debug流程(debug还可以将run_1 换成 gui_1);regress_test_1就是regress流程 test_11: compile_11 run_11 test_12: compile_12 run_12 test_13: compile_13 run_13 test_14: compile_14 run_14 test_2: compile_2 run_2 test_perf: compile_perf run_perf # regression targets regress_test_1: regress_build_1 regress_run_1 regress_test_11: regress_build_11 regress_run_11 regress_test_12: regress_build_12 regress_run_12 regress_test_13: regress_build_13 regress_run_13 regress_test_14: regress_build_14 regress_run_14 regress_test_2: regress_build_2 regress_run_2 regress_test_perf: regress_build_perf regress_run_perf #----------------------------------------------------------------------------- # COMPILE AND RUN TIME ARGUMENTS 编译与运行时的参数设置(run和sim可以看成一个意思,run就是run simv) #----------------------------------------------------------------------------- # Debug compile time arguments DBG_CMP += $(COV_CMP_OPTIONS) //debug 编译的参数 DBG_CMP += -debug_all //使能DVE debugging (包括 line stepping) //DBG_CMP += -debug_pp //使能VPD dump 和assertion debug DBG_CMP += +define+VPD_ON //debug compile的时候定义一个VPD_ON的宏,注意VPD是SIM时生成的 #DBG_CMP += +define+VPD_OFF //若CMP时参数把VPD关了,但是在SIM时输出一个VPD会怎么样? #DBG_CMP += +define+LOG_FMT_OFF //应该是这样,在verilog代码中将VPD dump的代码写在 ifdefine VPD_ON 后面 # Debug run time arguments DBG_RUN += $(COV_SIM_OPTIONS) //COV_SIM_OPTION 和 COV_CMP_OPTION的区别 # Regression compile time arguments REG_CMP += $(COV_CMP_OPTIONS) REG_CMP += +define+VPD_OFF //regress compile的时候定义了VPD_OFF, debug和regress的区别其实主要就是这, //因为debug时也可以做urg,所以在CMP和SIM参数中关于覆盖率实际上是一致的 # Regression run time arguments //注意在debug和regress各自的流程中urg命令(make 命令行命令)是不同的 REG_RUN += $(COV_SIM_OPTIONS) # Define where the coverage data is for URG //覆盖率数据,这个是给后边urg命令用的,产生覆盖率实在CMP和SIM之后进行的 COV_DBG_DATA += -dir $(COV_DIR)/debug/simv.vdb -dir $(COV_DIR)/debug/simv.cm COV_REG_DATA += -dir $(COV_DIR)/regress/simv.vdb -dir $(COV_DIR)/debug/simv.cm #----------------------------------------------------------------------------- # COMMAND LINE ARGUMENTS make命令行参数 #----------------------------------------------------------------------------- SEED = 766 #234567 #DEFINES = "+rvm_log_default=DEBUG" DEFINES = "+vmm_log_default=DEBUG" #DEFINES = "+vmm_log_default=NOTE" #DEFINES = "+rvm_log_default=WARNING" #DEFINES = "+vmm_log_default=ERROR" ############################################################################## ############################################################################## # PRIVATE //私有部分,用户可以不改 # You should not need to modify anything below this point # The following code supports a SV DUT and SVTB. ############################################################################## ############################################################################## DIR = $(/user/synopsys/Gaon/Platform) ########################################################################## # DEVICE UNDER TEST DUT CMP和SIM参数设置,之前已经设置过debug和regress在Compile时的参数 ########################################################################## DUT_CMP_OPTIONS += -sverilog +v2k DUT_CMP_OPTIONS += -o $(DUT_SIM_EXEC) DUT_CMP_OPTIONS += -Mdir=$(OUTPUT_DIR)/$(TB_TEST_ID)_csrc DUT_CMP_OPTIONS += -l $(LOG_DIR)/$(TB_TEST).cmp_log DUT_CMP_OPTIONS += +vcs+lic+wait +plusarg_save DUT_CMP_OPTIONS += $(DUT_INC) DUT_SIM_OPTIONS += -l $(LOG_DIR)/$(TB_TEST_ID).run_log DUT_SIM_OPTIONS += +vcs+lic+wait DUT_SIM_OPTIONS += +vpdfile+$(OUTPUT_DIR)/$(TB_TEST_ID).vpd //vpd实在执行sim后生成的,但是在debugcompile的时候为什么有个vdp_on的参数呢? #DUT_SIM_OPTIONS += +ntb_random_seed=$(SEED) DUT_SIM_OPTIONS += +ntb_random_seed_automatic DUT_SIM_OPTIONS += -assert nopostproc+report=$(LOG_DIR)/$(TB_TEST_ID).sva_log DUT_SIM_OPTIONS += -cm_assert_name $(TB_TEST_ID) DUT_SIM_OPTIONS += $(DEFINES) DUT_SIM_EXEC += $(OUTPUT_DIR)/$(TB_TEST)_simv ########################################################################## # TESTBENCH TB CMP和SIM参数设置,之前已经设置过debug和regress在Compile时的参数 ########################################################################## TB_TEST += test_$* //$*是什么意思,$* :去掉后缀的当前目标名(?)。例如,若当前目标是pro.o,则$*表示pro。 TB_TEST_ID += $(TB_TEST)_$(SEED) # VK ENVIRONMENT TB_INC += +incdir+$(TB_SRC_DIR)/vk TB_CMP_OPTIONS += $(TB_INC) ########################################################################## # COVERAGE 覆盖率设置 ########################################################################## #COV_CM_OPTIONS += +tb_cov_db_name=$(TB_TEST_ID) COV_CM_OPTIONS += -cm_name $(TB_TEST_ID) COV_CMP_OPTIONS += $(COV_CM_OPTIONS) -cm_hier $(COV_HIER) COV_SIM_OPTIONS += $(COV_CM_OPTIONS) COV_SIM_OPTIONS += -cm_log $(LOG_DIR)/$(TB_TEST_ID).cm_log COV_HIER += $(OUTPUT_DIR)/vcm.cfg # Coverage options for build and run with debug COV_CM_DBG += -cm_dir $(COV_DIR)/debug/simv.cm #COV_CM_DBG += -ova_dir $(COV_DIR)/debug/simv.vdb #COV_CM_DBG += +tb_cov_db_dir=$(COV_DIR)/debug/simv.vdb # Coverage options for build and run with regressions COV_CM_REG += -cm_dir $(COV_DIR)/regress/simv.cm COV_CM_REG += -ova_dir $(COV_DIR)/regress/simv.vdb COV_CM_REG += +tb_cov_db_dir=$(COV_DIR)/regress/simv.vdb ########################################################################## # DEBUG TARGETS ########################################################################## compile_%: echo $(COV_TREE) > $(COV_HIER); //debug 编译时主要有下面几个参数 TB_CMP DUT_CMP DBG_CMP COV_CM_DBG vcs $(TB_CMP_OPTIONS) \ //请详细查看上面几个变量的设置 $(DUT_CMP_OPTIONS) \ $(DUT_SRC) \ $(TB_SRC) \ $(SVA_SRC) \ $(SVA_OPTIONS) \ $(COV_CM_DBG) \ $(DBG_CMP) run_%: $(DUT_SIM_EXEC) $(DUT_SIM_OPTIONS) $(DBG_RUN) $(COV_CM_DBG) //run 是上面说的test_1中的第二步,属于debug流程,其实就是执行simv,参数有DUT_SIM和DBG_RUN //注意TB在run(sim)时没有相关参数 gui_%: //gui_1和run_1的区别就是打开了视图界面,他们都是执行sim $(DUT_SIM_EXEC) $(DUT_SIM_OPTIONS) $(DBG_RUN) $(COV_CM_DBG) \ -gui tb_gui_%: //-tb_gui 和-gui的区别是什么(上文提到DVE和testbenchdebugeer的含义),猜测-gui是打开DVE软件,-tb_gui就不知道是神马 $(DUT_SIM_EXEC) $(DUT_SIM_OPTIONS) $(DBG_RUN) $(COV_CM_DBG) \ -tb_gui +ntb_debug_on_start both_guis_%: $(DUT_SIM_EXEC) $(DUT_SIM_OPTIONS) $(DBG_RUN) $(COV_CM_DBG) \ -gui \ -tb_gui +ntb_debug_on_start new_gui_%: //打开一个新的 DVE软件窗口? $(DUT_SIM_EXEC) $(DUT_SIM_OPTIONS) $(DBG_RUN) $(COV_CM_DBG) \ -gui \ -tbug pp_%: //这个命令应该就是打开VPD波形 dve -vpd $(OUTPUT_DIR)/$(TB_TEST_ID).vpd urg: //执行代码覆盖率操作? 代码覆盖律不是在sim的时候产生的吗(在CMP和SIM的时候都有COV的参数啊)? 这个要具体查一下 urg $(COV_DBG_DATA) -report $(COV_DIR)/debug/urgReport -lca mozilla $(DIR)/$(COV_DIR)/debug/urgReport/dashboard.html & dve_cov: //该命令应该是使用DVE软件查看coverage结果 @echo "" @echo "WARNING: Did you run this command?" @echo "" @echo " % source ./utils/setup_dve_cov" @echo "" dve -cov & ########################################################################## # REGRESSION TARGETS ########################################################################## regress_clean: clean @rm -rf $(COV_DIR)/* @mkdir -p $(COV_DIR)/debug //怎么把debug的路径也给删除了? @mkdir -p $(COV_DIR)/regress @mkdir -p $(LOG_DIR) @mkdir -p $(OUTPUT_DIR) regress_build_%: //regress compile 的时候就$(REG_CMP)和debug不同仔细检查两者的差异 echo $(COV_TREE) > $(COV_HIER); vcs $(TB_CMP_OPTIONS) $(DUT_CMP_OPTIONS) \ $(DUT_SRC) \ $(TB_SRC) \ $(SVA_SRC) \ $(SVA_OPTIONS) \ $(COV_CM_REG) \ $(REG_CMP) regress_run_%: //观察各参数和debug流程的有什么差异 $(DUT_SIM_EXEC) $(DUT_SIM_OPTIONS) $(REG_RUN) $(COV_CM_REG) regress_urg: //看来VPD是SIM时产生的,但是覆盖率不是SIM时产生的,可能因为覆盖律要跑多个case才有意义 urg $(COV_REG_DATA) -grade -report $(COV_DIR)/regress/urgReport mozilla $(DIR)/$(COV_DIR)/regress/urgReport/dashboard.html & regress_dve_cov: @echo "" @echo "WARNING: Did you run this command?" @echo "" @echo " % source ./utils/setup_dve_cov" @echo "" dve -cov & # 综上,debug和regress流程类似,都是四步,compile,sim,urg,dve_cov ########################################################################## # ADMINISTRATIVE 管理命令 ########################################################################## help: @echo ======================================================================= @echo " " @echo " USAGE: %make target_name_* <SEED=xxx> <DEFINES=xxxx> " @echo " " @echo " ------------------------ DEBUG TARGETS ----------------------------" @echo " test_* => Compile TB and DUT files, runs the simulation. " @echo " clean => Clean the intermediate files. " @echo " compile_* => Compile the TB and DUT. " @echo " run_* => Run the simulation. " @echo " gui_* => Run simulation interactively with DVE. " @echo " tb_gui_* => Runs simulation interactively with TB Debugger. " @echo " both_guis_* => Run both debuggers. " @echo " new_gui_* => Run new integrated debuggers. " @echo " pp_* => Post process VPD with DVE. " @echo " urg => Make a coverage report for debug runs. " @echo " dve_cov => Brings up DVE for coverage reporting. " @echo " " @echo " ----------------------- REGRESSION TARGETS ------------------------" @echo " regress_test_* => Compile and run with coverage. " @echo " regress_clean => Remove all coverage files. " @echo " regress_build_* => Build test_*. " @echo " regress_run_* => Run test * collecting coverage information. " @echo " regress_urg => Make a coverage report for regression runs. " @echo " regress_dve_cov => Brings up DVE for coverage reporting. " @echo " " @echo " -------------------- ADMINISTRATIVE TARGETS -----------------------" @echo " help => Displays this message. " @echo " init => Clean all files, including coverage files. " @echo " tar => Tar and zip kit and place at ../ " @echo " " @echo " e.g. gmake test_1 " @echo ======================================================================= tar: clean cd ..; \ tar cvf ${DIR}.tar ${DIR}; \ rm -f ${DIR}.tgz; \ gzip ${DIR}.tar; \ mv ${DIR}.tar.gz ${DIR}.tgz clean: @rm -rf $(OUTPUT_DIR)/* $(COV_DIR)/debug/* $(LOG_DIR)/* ./DVEfiles @rm -rf urgReport *.tcl *.tcl.old vc_hdrs.h testbench_debugger_rc @rm -rf ucli.key vcs.key vera_debugger_rc .vera_debugger_rc.lock @rm -rf .test* .vlog* .dummyDir *.db *.vdb verilog.dump @rm -rf ._* .dw* *.log init: regress_clean @rm -rf include @rm -rf examples 总结: 经过研究,首先使用VCS验证的流程是: 1. 对各个源文件(包括DUT和TB)进行编译,生成可执行文件(默认名字为simv) 2. 执行simv文件进行仿真,这时会将testcase的仿真结果输出,包括VPD(若开启了VPD dumpping)文件等输出文件 3. 若进行代码覆盖率检查,在simv后可以使用urg命令收集覆盖率信息(须跑多个case) Makefile 完成的功能是: 1. 对源文件的管理:主要是将DUT文件和TB文件以及一些必要的文件的路径写入变量($DUT_SRC_DIR $TB_SRC_DIR等),还有include的路径(供VCS搜索)。在使用VCS命令进行编译时,可以将该路径加入命令中。我们可以使用脚本将需要编译的文件输出为一个file_list.f文件,然后用vcs –f file_list.f 进行编译。 2. 对路径的管理:将源文件路径和输出文件路径写入各自的变量,一遍在后边使用vcs命令时调用。 3. 对VCS命令参数的调整:这部分是重要的步骤,根据需要的不同(debug和regress的需要就不相同),参数设置也有不同,但有一些参数是不管怎样的流程都需要的,在我们的验证平台中,我们可以将参数分为两部分,一部分是都需要的可以公用的,一部分是可以灵活使用的,可以更改的。 4. 对VCS命令的调整:这部分就是将VCS的几个主要命令(其实就是VCS 和 SIMV)加上对应的参数设置。我们的验证平台可以将不同的命令写成不同的脚本,然后写一个总的脚本进行调用。 5. 其实需要灵活配置的主要就是VPD文件和覆盖率的设置(VPD 在rtl级实际上应该每次都生成)
YNOPSYS VCS Makefile文件编写与研究
这个Makefile是synopsys提供的模板,看上去非常好用,你只要按部就班提供实际项目的参数就可以了。我们来看这个文件的头部说明:
makefile 其实完全可以用csh或其他脚本来编写,只是VCS使用的linux内置的make命令定义了一个标准的仿真脚本,make命令是专门用来
做项目的源文件管理和编译控制的命令。这篇文章重点看synpsys的标准仿真脚本都做了哪些操作,然后使用其他脚本来实现。这里主要是自己
写的一点东西,有些地方是猜测的或者不准确。
#---------------------------------------------------------------------------------------------------------------------------
# SYNOPSYS CONFIDENTIAL - This is an unpublished, proprietary work of
# Synopsys, Inc., and is fully protected under copyright and trade secret
# laws. You may not view, use, disclose, copy, or distribute this file or
# any information contained herein except pursuant to a valid written
# license from Synopsys.
# SYNOPSYS公司的版权声明,没有权限不可使用
#-----------------------------------------------------------------------------------------------------------------------------
# Filename : $Id: Makefile,v 1.0 2006/07/18 23:59:59 vangundy Exp $
# Created by : Synopsys Inc. 07/17/2006
# $Author : vangundy $
# Description : Demonstrates Verilog DUT and SVTB using VCS
# makefile文件头
#---------------------------------------------------------------------------------------------------------------------------
# The Makefile works on two seperate flows. The DEBUG flow is intended to be used
# During debugging of a testcase and/or the DUT. The REGRESSION flow is used
# During regression runs and collects coverage data.
# 该makefile模版包括两部分流程,debug(查错)流程和regress(回归测试)流程,两个流程大致步骤都相同都是:Compile,SIM(urg,覆盖
# 率的分析和采集),debug时主要是跑一个pattern,并dump VPD文件,SIM的同时可以打开DVE视图界面,结束后观察波形,regress主要用
# 于采集覆盖率,一般要跑多个pattern,这时就无需dump VPD文件(节约时间),由于是debug后有进行的重复运行,所以叫regress(回归)。
# 在我们的验证平台中,若不做代码覆盖率的功能,可以不写regress,只要写debug的流程和跑多个pattern的脚本就好了。
#---------------------------------------------------------------------------------------------------------------------
# The DEBUG flow turns on VPD dumping and turns off coverage collection. After
# building a testcase using the debug targets, you can debug the TB and the DUT
# source code using the testbench debugger and DVE. Of course, you can turn on
# coverage metrics and run in debug mode by changing compile and runtime options
# in the makefile. These changes are independent of the regression flow so that
# the regressions will still run optimally without the interference of VPD dumping.
# debug流程打开VPD文件的dump并关闭覆盖率在build了一个包含DUT的testcase后,可以使用VCS的debugger和DVE进行debug。当
# 然,你也以通过改变makefile文件中的compile和runtime选项参数来开启覆盖率功能。Debug流程和regress流程是各自独立的,regression
# 流程一般不生成VPD。
# --------------------------------------------------------------------------------------------------------------------------------
# The REGRESSION flow turns off VPD dumping and turns on Coverage Metrics and TB
# coverage collection. This flow is intended to support verification engineers who
# are working through the regression process and are interested in coverage
# collection and urg.
# REGRESSION流程关闭VPD dump并打开Coverage collection功能,该流程是为了支持验证引擎进行“流水线验证“(跑多个testcase)和
# 代码覆盖率功能。??在验证平台中可以将运行多个testcase的脚本命名为regress,运行单个testcase的脚本命名为regone??,这只是
# synopsys的模版,我们不必完全遵守,可以不区分debug和regress,然后将是否打开波形和coverage设置成参数。
# -------------------------------------------------------------------------------------------------------------------------------
# Command Line make命令行
# -----------------
# The Makefile supports the following command line
# makefile支持下列命令行
# % make target_name_* <SEED=xxx> <DEFINES=xxxx>
# makefile文件放在哪?放在仿真路径。make [-f makefile文件名][选项][宏定义][目标]
# -f 指定makefile 若没有则make程序首先在当前目录查找名为makefile的文件,如果没有找到,它就会转而查找名为Makefile的文件。
# Where target_name is the name of a testcase located in the test directory. Every
# test in the test directory is named using test_{test_name}. All of the test targets
# are listed in the TEST TARGETS section of the makefile.
# target_name是test路径下的一个testcase的名字,test路径下的testcase的名字使用test_{test_name}来命名,例如test_1
# 所有的test target 都在makefile文件中的TEST TARGETS部分列出
# ---------------------------------------------------------------------------------------------------------------------------
# Compile and Run Testcases 编译与运行testcase
# -------------------------------
# To compile and run a tescase use the test_* and regress_test_* targets.
# 编译与运行testcase,(test_1 就是执行了下面的两个命令先编译在运行) test_1 ==> compile_1 run_1 详见下面命令定义
# % make test_1 // Builds and runs test 1 with VPD dumping 其实就是debug的前边的流程
# % make regress_test_1 // Builds and runs test 1 with coverage turned on
# -------------------------------------------------------------------------------------------------------------------------
# Debugging Testcases Debug 实在上面命令之后在进行的
# ------------------------
# You can use DVE and the testbench debugger to visualize waveforms and testbench
# execution. You must first build the testbench using the make compile_* command.
# dubug必须是在DVE(VCS的debug工具,与debussy一样的功能)下进行,因为要看波形嘛,但是debug之前必须先compile
# % make compile_1 // Builds test 1 for debugging //需要重新编译一次吗?
# Once you have built the environment with the proper debug switches, you can use DVE and the testbench debugger.
# # testbench debugger 是否是指编译后的那个simv可执行文件呢? 其实gui_1 和上面test_1中的run_1是一样的只是增加了-gui项
# 即增加了打开gui界面的参数,其他雷同
# % make gui_1 // Debug test 1 with DVE
# % make tb_gui_1 // Debug test 1 with the testbench debugger
# % make both_guis_1 // Debug using both guis
# % make pp_1 // Debug using the VPD file VPD文件要在执行simv之后才有吧?
# If you want, you can turn on coverage for the DEBUG flow by uncommenting the
# coverage flag in the makefile. If you do this, you can still look at coverage.
# This may be useful in helping those who are debugging coverage related issues.
# 如果在makefile中的debug流程中使用了coverage功能,那么可以使用下面命令观察覆盖率
# % make urg // Visualize coverage data from debug runs
# -----------------------------------------------------------------------------------------------------------------------------
# Regression Testcases
# --------------------
# Regression tests are used to collect coverage information. To build a testcase
# for coverage collection use a command similar to the following.
# regress流程主要是为了收集代码覆盖率信息,在执行regress之前需要重新build testcase 类似debug时的compile
# % make regress_build_1 // Build and run a regression test with a default seed
# Once the test has been built, you can run it again with a new seed.
# 与debug不同的是regress需要重新run(使用新的SEED)一下,【还是debug的时候也要run一下?】
# % make regress_run_1 SEED=1234
# After running one or more regression runs, you can visualize the coverage data
# using urg and the following command
# run完之后可以用下面命令看代码覆盖率
# % make regress_urg
#----------------------------------------------------------------------------------------------------------------------------
# HOW TO REUSE THIS FILE ON ANOTHER DUT //如何重用该模版
# STEP 1: Update the file locations as required //设置file所属的路径
# STEP 2: Update the DUT section with directory and source location info//更新模版中DUT部分,指定DUT的路径和include的路径
# STEP 3: Update the TB section with directory and source location info//更新模版中TB部分,指定TB的路径和include的路径
# STEP 4: Update the Coverage section with name of dut top (eg top.dut) //跟新模版中Coverage部分,指定要测试代码覆盖率的dut的top
# STEP 5: Add test targets to the debug and regression targets section//将debug和regress的target加入模版中对应的部分
# STEP 5: Adjust the debug and regression compile and run time arguments//调整debug和regress的compile和runtime的命令参数
# STEP 7: Adjust command line options as required//调整命令行命令(后边带百分号和冒号的就表示可以在make命令行中使用的命令)
# STEP 8: Update the env class so that it extends dkm_env//更新env class(环境类)使得它可以提供dkm_env
# You will need to have a copy of the dkm directory and it should //dkm是什么?
# be located at $(TB_SRC_DIR)/dkm
# a) Add [`include "dkm_env.sv"]
# b) Add [extends dkm_env] to the environment class definition
# c) Call the super.new("name") from the constructor
# STEP 9: Run the debug and regression targets
# % make testbench_target_* // testbench_target_* 是指test_1之类的testcase
#-----------------------------------------------------------------------------
看了上文,大家应该可以简单了解这个Makefile的功能了。接下来就按照step1~9来填空即可:
.PHONY : default help clean regress_clean
default: help
#-----------------------------------------------------------------------------
# DIRECTORIES 总路径
#-----------------------------------------------------------------------------
OUTPUT_DIR = ./output //为什么debug和regress没有分开? debug和regress的Coverage在COV_DIR下是有区分的
COV_DIR = ./coverage //为什么没有VPD的路径?
LOG_DIR = ./logs //output是做什么的? VPD文件 simv文件 还有一写其他文件在这里
# Set this to the location where you installed the designware models. This
# depends on whether you ran the setup_vip_dw_home to install the models or
# the setup_vip_here script.
#DW_MODELS_DIR = $(DESIGNWARE_HOME)
DW_MODELS_DIR = /user/synopsys/designware //软件路径 VCS的路径?
#DESIGNWARE_HOME = ~synopsys/bk/designware
#DW_MODELS_DIR = ./designware
#----------------------------------------------------------------------------
# DEVICE UNDER TEST DUT路径
#-----------------------------------------------------------------------------
DUT_SRC_DIR = ./source/Verilog //SRC source
DUT_SRC = -f $(DUT_SRC_DIR)/rtl_list.f //待编译的rtl文件列表文件 SRC是什么意思?
DUT_INC += +incdir+/user/myproj/PROJECT/RTL/SRC/mymodule/
DUT_INC += +incdir+/user/myproj/PROJECT/RTL/SRC/mymodule/mymodule_inc.v
DUT_CMP_OPTIONS += +libext+.v+.V //这个参数是干什么的?指定VCS搜索文件时的文件后缀.v
#DUT_CMP_OPTIONS += -timescale=1ps/1ps //CMP是compile的意思 不是compare
#DUT_CMP_OPTIONS += -override_timescale=1ps/1ps
#-----------------------------------------------------------------------------
# TESTBENCH TB路径设置
#-----------------------------------------------------------------------------
TB_SRC_DIR = ./source/svtb
# AXI TESTBENCH, VIP Sources first
#TB_SRC += -f $(TB_SRC_DIR)/mac_if_tb/vip/gslv_model_package.f
TB_SRC += $(TB_SRC_DIR)/mpdu_trx_tb/tests/mpdu_tb_top.sv
TB_SRC += $(TB_SRC_DIR)/mpdu_trx_tb/tests/$(TB_TEST).sv //为什么有两个SV
TB_INC += +incdir+$(TB_SRC_DIR)/mpdu_trx_tb/vip
TB_INC += +incdir+$(TB_SRC_DIR)/mpdu_trx_tb/env
TB_INC += +incdir+$(TB_SRC_DIR)/mpdu_trx_tb/tests
TB_INC += +incdir+$(DW_MODELS_DIR)/include/svtb //这部分为了支持sv吗?
TB_INC += +incdir+$(DW_MODELS_DIR)/include/verilog
TB_INC += +incdir+$(DW_MODELS_DIR)/svtb
#TB_CMP_OPTIONS += -tb_timescale=1ns/1ps
#TB_CMP_OPTIONS += -lca Y-2006.06-SP2
TB_CMP_OPTIONS += +pkgdir+$(DW_MODELS_DIR)/include/svtb //TB的编译选项和DUT不同?sv 和verilog的区别吗?
TB_CMP_OPTIONS += -ntb_incdir $(DW_MODELS_DIR)/include/vera
TB_CMP_OPTIONS += -ntb_incdir $(DESIGNWARE_HOME)/vip/vmt/latest/vera/src
TB_CMP_OPTIONS += -ntb_incdir $(DESIGNWARE_HOME)/vip/amba/latest/vera/src
TB_CMP_OPTIONS += -ntb_incdir $(DESIGNWARE_HOME)/vip/amba/latest/axi_master_vmt/vera/src //TB中使用的一些模型
TB_CMP_OPTIONS += -ntb_incdir $(DESIGNWARE_HOME)/vip/amba/latest/axi_master_rvm_vera_vmt/vera/src
TB_CMP_OPTIONS += -ntb_incdir $(DESIGNWARE_HOME)/vip/amba/latest/axi_slave_vmt/vera/src
TB_CMP_OPTIONS += -ntb_incdir $(DESIGNWARE_HOME)/vip/amba/latest/axi_slave_rvm_vera_vmt/vera/src
TB_CMP_OPTIONS += -ntb_incdir $(DESIGNWARE_HOME)/vip/amba/latest/axi_monitor_vmt/vera/src
TB_CMP_OPTIONS += -ntb_incdir $(DESIGNWARE_HOME)/vip/amba/latest/axi_monitor_rvm_vera_vmt/vera/src
TB_CMP_OPTIONS += -ntb_incdir $(DESIGNWARE_HOME)/vip/amba/latest/axi_port_monitor_vmt/vera/src
TB_CMP_OPTIONS += -ntb_incdir $(DESIGNWARE_HOME)/vip/amba/latest/axi_port_monitor_rvm_vera_vmt/vera/src
TB_CMP_OPTIONS += -ntb_incdir $(DESIGNWARE_HOME)/vip/amba/latest/axi_interconnect_vmt/vera/src
TB_CMP_OPTIONS += -ntb_incdir $(DESIGNWARE_HOME)/vip/amba/latest/axi_interconnect_rvm_vera_vmt/vera/src
TB_CMP_OPTIONS += -ntb_incdir ${DESIGNWARE_HOME}/vip/amba/latest/ahb_master_vmt/vera/src
TB_CMP_OPTIONS += -ntb_incdir ${DESIGNWARE_HOME}/vip/amba/latest/ahb_slave_vmt/vera/src
TB_CMP_OPTIONS += -ntb_incdir ${DESIGNWARE_HOME}/vip/amba/latest/ahb_monitor_vmt/vera/src
TB_CMP_OPTIONS += -ntb_incdir ${DESIGNWARE_HOME}/vip/amba/latest/ahb_bus_vmt/vera/src
TB_CMP_OPTIONS += -ntb_incdir ${DESIGNWARE_HOME}/vip/amba/latest/ahb_master_rvm_vera_vmt/vera/src
TB_CMP_OPTIONS += -ntb_incdir ${DESIGNWARE_HOME}/vip/amba/latest/ahb_slave_rvm_vera_vmt/vera/src
TB_CMP_OPTIONS += -ntb_incdir ${DESIGNWARE_HOME}/vip/amba/latest/ahb_monitor_rvm_vera_vmt/vera/src
TB_CMP_OPTIONS += -ntb_incdir ${DESIGNWARE_HOME}/vip/amba/latest/ahb_bus_rvm_vera_vmt/vera/src
TB_CMP_OPTIONS += -ntb_define NTB
TB_CMP_OPTIONS += -ntb_define DW_VIP_AXI_MAX_NO_MSTRS=6
TB_CMP_OPTIONS += -ntb_define DW_VIP_AXI_MAX_NO_SLVS=2
TB_CMP_OPTIONS += +define+DW_VIP_AXI_MAX_NO_MSTRS_6
TB_CMP_OPTIONS += +define+DW_VIP_AXI_MAX_NO_SLVS_2
TB_CMP_OPTIONS += -ntb_opts rvm
TB_CMP_OPTIONS += -ntb_opts dtm
TB_CMP_OPTIONS += -ntb_opts use_sigprop
TB_CMP_OPTIONS += -ntb_opts interop
TB_CMP_OPTIONS += -ntb_opts dw_vip
TB_CMP_OPTIONS += +define+NT
# AIP Related files and compilation options
#TB_CMP_OPTIONS += +incdir+../BP062-BU-01000-r0p0-00rel0/sva \
+incdir+../BP062-BU-01000-r0p0-00rel0/verilog \
../BP062-BU-01000-r0p0-00rel0/sva/AxiPC.sv \
../BP062-BU-01000-r0p0-00rel0/verilog/Axi.v \
./source/svtb/platform_tb/env/Snps_ARMAXI_CheckerBind.sv
#${VCS_HOME}/packages/aip/DDR2_AIP/src/Snps_DDR2_Checker.sv \
-assert enable_diag \
+incdir+.+${VCS_HOME}/packages/aip/DDR2_AIP/src/ \
./source/svtb/platform_tb/env/Snps_DDR2_Bind.sv \
+incdir+../BP062-BU-01000-r0p0-00rel0/sva \
+incdir+../BP062-BU-01000-r0p0-00rel0/verilog \
../BP062-BU-01000-r0p0-00rel0/sva/AxiPC.sv \
../BP062-BU-01000-r0p0-00rel0/verilog/Axi.v \
./source/svtb/platform_tb/env/Snps_ARMAXI_CheckerBind.sv
#-----------------------------------------------------------------------------
# COVERAGE 覆盖率的设置
#-----------------------------------------------------------------------------
COV_TREE += '+tree mpdu_tb_top'
COV_CM_OPTIONS += -cm line+cond+fsm+assert 注意CM和CMP不一样
#-----------------------------------------------------------------------------
# TEST TARGETS 总命令
#-----------------------------------------------------------------------------
# debug targets
test_1: compile_1 run_1 与前边对应test_1 就是debug流程(debug还可以将run_1 换成 gui_1);regress_test_1就是regress流程
test_11: compile_11 run_11
test_12: compile_12 run_12
test_13: compile_13 run_13
test_14: compile_14 run_14
test_2: compile_2 run_2
test_perf: compile_perf run_perf
# regression targets
regress_test_1: regress_build_1 regress_run_1
regress_test_11: regress_build_11 regress_run_11
regress_test_12: regress_build_12 regress_run_12
regress_test_13: regress_build_13 regress_run_13
regress_test_14: regress_build_14 regress_run_14
regress_test_2: regress_build_2 regress_run_2
regress_test_perf: regress_build_perf regress_run_perf
#-----------------------------------------------------------------------------
# COMPILE AND RUN TIME ARGUMENTS 编译与运行时的参数设置(run和sim可以看成一个意思,run就是run simv)
#-----------------------------------------------------------------------------
# Debug compile time arguments
DBG_CMP += $(COV_CMP_OPTIONS) //debug 编译的参数
DBG_CMP += -debug_all //使能DVE debugging (包括 line stepping)
//DBG_CMP += -debug_pp //使能VPD dump 和assertion debug
DBG_CMP += +define+VPD_ON //debug compile的时候定义一个VPD_ON的宏,注意VPD是SIM时生成的
#DBG_CMP += +define+VPD_OFF //若CMP时参数把VPD关了,但是在SIM时输出一个VPD会怎么样?
#DBG_CMP += +define+LOG_FMT_OFF //应该是这样,在verilog代码中将VPD dump的代码写在 ifdefine VPD_ON 后面
# Debug run time arguments
DBG_RUN += $(COV_SIM_OPTIONS) //COV_SIM_OPTION 和 COV_CMP_OPTION的区别
# Regression compile time arguments
REG_CMP += $(COV_CMP_OPTIONS)
REG_CMP += +define+VPD_OFF //regress compile的时候定义了VPD_OFF, debug和regress的区别其实主要就是这,
//因为debug时也可以做urg,所以在CMP和SIM参数中关于覆盖率实际上是一致的
# Regression run time arguments //注意在debug和regress各自的流程中urg命令(make 命令行命令)是不同的
REG_RUN += $(COV_SIM_OPTIONS)
# Define where the coverage data is for URG //覆盖率数据,这个是给后边urg命令用的,产生覆盖率实在CMP和SIM之后进行的
COV_DBG_DATA += -dir $(COV_DIR)/debug/simv.vdb -dir $(COV_DIR)/debug/simv.cm
COV_REG_DATA += -dir $(COV_DIR)/regress/simv.vdb -dir $(COV_DIR)/debug/simv.cm
#-----------------------------------------------------------------------------
# COMMAND LINE ARGUMENTS make命令行参数
#-----------------------------------------------------------------------------
SEED = 766
#234567
#DEFINES = "+rvm_log_default=DEBUG"
DEFINES = "+vmm_log_default=DEBUG"
#DEFINES = "+vmm_log_default=NOTE"
#DEFINES = "+rvm_log_default=WARNING"
#DEFINES = "+vmm_log_default=ERROR"
##############################################################################
##############################################################################
# PRIVATE //私有部分,用户可以不改
# You should not need to modify anything below this point
# The following code supports a SV DUT and SVTB.
##############################################################################
##############################################################################
DIR = $(/user/synopsys/Gaon/Platform)
##########################################################################
# DEVICE UNDER TEST DUT CMP和SIM参数设置,之前已经设置过debug和regress在Compile时的参数
##########################################################################
DUT_CMP_OPTIONS += -sverilog +v2k
DUT_CMP_OPTIONS += -o $(DUT_SIM_EXEC)
DUT_CMP_OPTIONS += -Mdir=$(OUTPUT_DIR)/$(TB_TEST_ID)_csrc
DUT_CMP_OPTIONS += -l $(LOG_DIR)/$(TB_TEST).cmp_log
DUT_CMP_OPTIONS += +vcs+lic+wait +plusarg_save
DUT_CMP_OPTIONS += $(DUT_INC)
DUT_SIM_OPTIONS += -l $(LOG_DIR)/$(TB_TEST_ID).run_log
DUT_SIM_OPTIONS += +vcs+lic+wait
DUT_SIM_OPTIONS += +vpdfile+$(OUTPUT_DIR)/$(TB_TEST_ID).vpd
//vpd实在执行sim后生成的,但是在debugcompile的时候为什么有个vdp_on的参数呢?
#DUT_SIM_OPTIONS += +ntb_random_seed=$(SEED)
DUT_SIM_OPTIONS += +ntb_random_seed_automatic
DUT_SIM_OPTIONS += -assert nopostproc+report=$(LOG_DIR)/$(TB_TEST_ID).sva_log
DUT_SIM_OPTIONS += -cm_assert_name $(TB_TEST_ID)
DUT_SIM_OPTIONS += $(DEFINES)
DUT_SIM_EXEC += $(OUTPUT_DIR)/$(TB_TEST)_simv
##########################################################################
# TESTBENCH TB CMP和SIM参数设置,之前已经设置过debug和regress在Compile时的参数
##########################################################################
TB_TEST += test_$* //$*是什么意思,$* :去掉后缀的当前目标名(?)。例如,若当前目标是pro.o,则$*表示pro。
TB_TEST_ID += $(TB_TEST)_$(SEED)
# VK ENVIRONMENT
TB_INC += +incdir+$(TB_SRC_DIR)/vk
TB_CMP_OPTIONS += $(TB_INC)
##########################################################################
# COVERAGE 覆盖率设置
##########################################################################
#COV_CM_OPTIONS += +tb_cov_db_name=$(TB_TEST_ID)
COV_CM_OPTIONS += -cm_name $(TB_TEST_ID)
COV_CMP_OPTIONS += $(COV_CM_OPTIONS) -cm_hier $(COV_HIER)
COV_SIM_OPTIONS += $(COV_CM_OPTIONS)
COV_SIM_OPTIONS += -cm_log $(LOG_DIR)/$(TB_TEST_ID).cm_log
COV_HIER += $(OUTPUT_DIR)/vcm.cfg
# Coverage options for build and run with debug
COV_CM_DBG += -cm_dir $(COV_DIR)/debug/simv.cm
#COV_CM_DBG += -ova_dir $(COV_DIR)/debug/simv.vdb
#COV_CM_DBG += +tb_cov_db_dir=$(COV_DIR)/debug/simv.vdb
# Coverage options for build and run with regressions
COV_CM_REG += -cm_dir $(COV_DIR)/regress/simv.cm
COV_CM_REG += -ova_dir $(COV_DIR)/regress/simv.vdb
COV_CM_REG += +tb_cov_db_dir=$(COV_DIR)/regress/simv.vdb
##########################################################################
# DEBUG TARGETS
##########################################################################
compile_%:
echo $(COV_TREE) > $(COV_HIER); //debug 编译时主要有下面几个参数 TB_CMP DUT_CMP DBG_CMP COV_CM_DBG
vcs $(TB_CMP_OPTIONS) \ //请详细查看上面几个变量的设置
$(DUT_CMP_OPTIONS) \
$(DUT_SRC) \
$(TB_SRC) \
$(SVA_SRC) \
$(SVA_OPTIONS) \
$(COV_CM_DBG) \
$(DBG_CMP)
run_%:
$(DUT_SIM_EXEC) $(DUT_SIM_OPTIONS) $(DBG_RUN) $(COV_CM_DBG)
//run 是上面说的test_1中的第二步,属于debug流程,其实就是执行simv,参数有DUT_SIM和DBG_RUN
//注意TB在run(sim)时没有相关参数
gui_%: //gui_1和run_1的区别就是打开了视图界面,他们都是执行sim
$(DUT_SIM_EXEC) $(DUT_SIM_OPTIONS) $(DBG_RUN) $(COV_CM_DBG) \
-gui
tb_gui_%: //-tb_gui 和-gui的区别是什么(上文提到DVE和testbenchdebugeer的含义),猜测-gui是打开DVE软件,-tb_gui就不知道是神马
$(DUT_SIM_EXEC) $(DUT_SIM_OPTIONS) $(DBG_RUN) $(COV_CM_DBG) \
-tb_gui +ntb_debug_on_start
both_guis_%:
$(DUT_SIM_EXEC) $(DUT_SIM_OPTIONS) $(DBG_RUN) $(COV_CM_DBG) \
-gui \
-tb_gui +ntb_debug_on_start
new_gui_%: //打开一个新的 DVE软件窗口?
$(DUT_SIM_EXEC) $(DUT_SIM_OPTIONS) $(DBG_RUN) $(COV_CM_DBG) \
-gui \
-tbug
pp_%: //这个命令应该就是打开VPD波形
dve -vpd $(OUTPUT_DIR)/$(TB_TEST_ID).vpd
urg: //执行代码覆盖率操作? 代码覆盖律不是在sim的时候产生的吗(在CMP和SIM的时候都有COV的参数啊)? 这个要具体查一下
urg $(COV_DBG_DATA) -report $(COV_DIR)/debug/urgReport -lca
mozilla $(DIR)/$(COV_DIR)/debug/urgReport/dashboard.html &
dve_cov: //该命令应该是使用DVE软件查看coverage结果
@echo ""
@echo "WARNING: Did you run this command?"
@echo ""
@echo " % source ./utils/setup_dve_cov"
@echo ""
dve -cov &
##########################################################################
# REGRESSION TARGETS
##########################################################################
regress_clean: clean
@rm -rf $(COV_DIR)/*
@mkdir -p $(COV_DIR)/debug //怎么把debug的路径也给删除了?
@mkdir -p $(COV_DIR)/regress
@mkdir -p $(LOG_DIR)
@mkdir -p $(OUTPUT_DIR)
regress_build_%: //regress compile 的时候就$(REG_CMP)和debug不同仔细检查两者的差异
echo $(COV_TREE) > $(COV_HIER);
vcs $(TB_CMP_OPTIONS) $(DUT_CMP_OPTIONS) \
$(DUT_SRC) \
$(TB_SRC) \
$(SVA_SRC) \
$(SVA_OPTIONS) \
$(COV_CM_REG) \
$(REG_CMP)
regress_run_%: //观察各参数和debug流程的有什么差异
$(DUT_SIM_EXEC) $(DUT_SIM_OPTIONS) $(REG_RUN) $(COV_CM_REG)
regress_urg: //看来VPD是SIM时产生的,但是覆盖率不是SIM时产生的,可能因为覆盖律要跑多个case才有意义
urg $(COV_REG_DATA) -grade -report $(COV_DIR)/regress/urgReport
mozilla $(DIR)/$(COV_DIR)/regress/urgReport/dashboard.html &
regress_dve_cov:
@echo ""
@echo "WARNING: Did you run this command?"
@echo ""
@echo " % source ./utils/setup_dve_cov"
@echo ""
dve -cov &
# 综上,debug和regress流程类似,都是四步,compile,sim,urg,dve_cov
##########################################################################
# ADMINISTRATIVE 管理命令
##########################################################################
help:
@echo =======================================================================
@echo " "
@echo " USAGE: %make target_name_* <SEED=xxx> <DEFINES=xxxx> "
@echo " "
@echo " ------------------------ DEBUG TARGETS ----------------------------"
@echo " test_* => Compile TB and DUT files, runs the simulation. "
@echo " clean => Clean the intermediate files. "
@echo " compile_* => Compile the TB and DUT. "
@echo " run_* => Run the simulation. "
@echo " gui_* => Run simulation interactively with DVE. "
@echo " tb_gui_* => Runs simulation interactively with TB Debugger. "
@echo " both_guis_* => Run both debuggers. "
@echo " new_gui_* => Run new integrated debuggers. "
@echo " pp_* => Post process VPD with DVE. "
@echo " urg => Make a coverage report for debug runs. "
@echo " dve_cov => Brings up DVE for coverage reporting. "
@echo " "
@echo " ----------------------- REGRESSION TARGETS ------------------------"
@echo " regress_test_* => Compile and run with coverage. "
@echo " regress_clean => Remove all coverage files. "
@echo " regress_build_* => Build test_*. "
@echo " regress_run_* => Run test * collecting coverage information. "
@echo " regress_urg => Make a coverage report for regression runs. "
@echo " regress_dve_cov => Brings up DVE for coverage reporting. "
@echo " "
@echo " -------------------- ADMINISTRATIVE TARGETS -----------------------"
@echo " help => Displays this message. "
@echo " init => Clean all files, including coverage files. "
@echo " tar => Tar and zip kit and place at ../ "
@echo " "
@echo " e.g. gmake test_1 "
@echo =======================================================================
tar: clean
cd ..; \
tar cvf ${DIR}.tar ${DIR}; \
rm -f ${DIR}.tgz; \
gzip ${DIR}.tar; \
mv ${DIR}.tar.gz ${DIR}.tgz
clean:
@rm -rf $(OUTPUT_DIR)/* $(COV_DIR)/debug/* $(LOG_DIR)/* ./DVEfiles
@rm -rf urgReport *.tcl *.tcl.old vc_hdrs.h testbench_debugger_rc
@rm -rf ucli.key vcs.key vera_debugger_rc .vera_debugger_rc.lock
@rm -rf .test* .vlog* .dummyDir *.db *.vdb verilog.dump
@rm -rf ._* .dw* *.log
init: regress_clean
@rm -rf include
@rm -rf examples
总结:
经过研究,首先使用VCS验证的流程是:
1. 对各个源文件(包括DUT和TB)进行编译,生成可执行文件(默认名字为simv)
2. 执行simv文件进行仿真,这时会将testcase的仿真结果输出,包括VPD(若开启了VPD dumpping)文件等输出文件
3. 若进行代码覆盖率检查,在simv后可以使用urg命令收集覆盖率信息(须跑多个case)
Makefile 完成的功能是:
1. 对源文件的管理:主要是将DUT文件和TB文件以及一些必要的文件的路径写入变量($DUT_SRC_DIR $TB_SRC_DIR等),还有include的路径(供VCS搜索)。在使用VCS命令进行编译时,可以将该路径加入命令中。我们可以使用脚本将需要编译的文件输出为一个file_list.f文件,然后用vcs –f file_list.f 进行编译。
2. 对路径的管理:将源文件路径和输出文件路径写入各自的变量,一遍在后边使用vcs命令时调用。
3. 对VCS命令参数的调整:这部分是重要的步骤,根据需要的不同(debug和regress的需要就不相同),参数设置也有不同,但有一些参数是不管怎样的流程都需要的,在我们的验证平台中,我们可以将参数分为两部分,一部分是都需要的可以公用的,一部分是可以灵活使用的,可以更改的。
4. 对VCS命令的调整:这部分就是将VCS的几个主要命令(其实就是VCS 和 SIMV)加上对应的参数设置。我们的验证平台可以将不同的命令写成不同的脚本,然后写一个总的脚本进行调用。
5. 其实需要灵活配置的主要就是VPD文件和覆盖率的设置(VPD 在rtl级实际上应该每次都生成)