debug TCL script with free tools

面临的问题

tcl脚本被广泛使用于EDA工具中，像Cadence， Synopys和mentor的工具脚本都是tcl脚本，可以在里面嵌入tcl脚本以实现比较复杂的设计流程和自动化工作。

目前tcl的调试主要依靠插入打印信息，这样需要叠代的次数比较，代码里会充满了打印语句也不太美观。

可选方案

在网上尝试了几种方案，最好的是activestate的Komodo， 它是一个IDE， 可以直接GUI下的各种调试手段，无奈它是一个收费软件，可以免费尝试21天。所以作者把目光投向了一些其他免费的方案。

GUI下的调试手段还是比较友好的，所以找到了, 这个工具虽然支持的功能很简单，但都非常实用，唯一不足就是速度太慢了，所以又尝试了类似工具RamDebugger.另外一个工具是一个简单的脚本stepsource.tcl, 它也可以实现类似的功能，只是没有GUI，但速度飞快，作者本人更喜欢使用它。下面就简单介绍一下他们的安装和使用

DDT

介绍

可以支持Tcl8.5或更高版本的动态调试。它主要提供了step， breakpoint, variable display功能。它有一个简单的界面如下：

安装

tkcon

它依赖于tkcon, 所以需要先安装tkcon.

下载地址:

BTW：也可以用git clone https://github.com/wjoye/tkcon.git来下载补丁版本。

下载后可以在自己home目录下新建一个.tcllib目录，然后把下载的tkcon解压缩到里面。

然后设置环境变量export TCLLIBPATH="/home/harriszh/.tcllib"

下面是测试是否安装好的方法：

tclsh% package require tkcon2.7% tkcon show

会调出tkcon的窗口

ddt

使用git下载ddt源码：

git clone https://github.com/Drolla/ddt.git

可以将它放到某个地方，比如作者放到了/home/harriszh/bin/

然后在bash下设置如下别名, 就可以调起ddt。

alias ddt="curdir=$(pwd); cd /home/harriszh/bin/ddt/ddt_debugger && ./ddt_debugger.tcl && cd ${curdir}"

如果是csh, 则需要

alias ddt="curdir=`pwd`; cd /home/harriszh/bin/ddt/ddt_debugger && ./ddt_debugger.tcl && cd ${curdir}"

使用

通过File->load来加载要调试的tcl文件

如果这个脚本依赖输入参数，那么可以通过config->Define initilization variables and script来提供调用参数或者可以直接在脚本的最上面写上

set argc 5set argv {1010 -- out 10 100}

所有的功能都在Debug菜单下， 和一般调试器相似， F5是run/continue, F8是设置breakpoints， shift-F5是stop， F7F7是增加watch

在运行时，右半边窗口会列出所有变量

缺点

• 运行速度非常慢
• 当脚本有问题时，容易hang， refresh都没用

总结

非常友好的GUI和使用方法。但因为速度问题使得作者寻找其他解决方案，如stepsource

---

stepsource

介绍

提供按行执行tcl脚本的功能

安装

把下面文件保存为stepsource.tcl

#!/usr/bin/tclsh#===============================================================================# FILE               :  stepsource.tcl# USAGE              :  ./stepsource.tcl 
    
     # DESCRIPTION        :  ---# REQUIREMENT        :  ---# AUTHOR             :  Harris Zhu (harriszh), harriszh@cadence.com# Created On         :  2018-06-30 21:16# Last Modified      :  2018-06-30 21:16# Update Count       :  1# REVISION           :  ---#=============================================================================== namespace eval ::stepsource { variable VERSION "1.0" proc StepCommand {stepcommand} {    switch -regexp -- $stepcommand {        ^\[0-9\]+$ {            set ::stepsource::currentBreakPoint $stepcommand            return        }        ^b\ *-*[0-9?]*$ {            if {$stepcommand == "b -"} {set ::stepsource::breakPoints {} ; return}            set bOption [lindex $stepcommand 1]            if {$bOption == "?"} {puts $::stepsource::outChannel "breakpoints: $::stepsource::breakPoints" ; return}            if {![string first - $bOption]} {                set eraseBreakPoint [lsearch $::stepsource::breakPoints [expr abs($bOption)]]                if {$eraseBreakPoint > -1} {                    set ::stepsource::breakPoints [lreplace $::stepsource::breakPoints $eraseBreakPoint $eraseBreakPoint]                }                puts $::stepsource::outChannel "breakpoints: $::stepsource::breakPoints"                return            }            set ::stepsource::breakPoints "$::stepsource::breakPoints $bOption"            set ::stepsource::breakPoints [lsort -unique $::stepsource::breakPoints]            if {$bOption == {}} {set ::stepsource::currentBreakPoint $bOption}            return        }        ^l\ *[0-9]*\ *[0-9]*$ {            regexp {l ([0-9]+) *([0-9]*)} $stepcommand trash listStart listEnd            if ![info exists listStart] {                set listStart 1                set listEnd $::stepsource::lineCount            } else {                if {(![string is integer -strict $listEnd]) || ($listEnd < $listStart)} {set listEnd $listStart}            }            for {set i $listStart} {$i <= $listEnd} {incr i} {                if ![info exists ::stepsource::lineArray($i)] {return}                puts -nonewline $::stepsource::outChannel "$::stepsource::lineArray($i)"            }            return        }    }    switch -- $stepcommand {        a {            foreach var [uplevel 3 info vars] {                if ![uplevel 3 array exists [list $var]] {continue}                puts $::stepsource::outChannel "-----------------------------------"                uplevel 3 parray [list $var]            }        }        c {            foreach var [lsort [uplevel 3 info vars]] {                if {$var == "errorInfo"} {continue}                if ![uplevel 3 info exists [list $var]] {continue}                if [uplevel 3 array exists [list $var]] {continue}                set changeIcon "=="                catch {if {$::stepsource::varValues($var) != [uplevel 3 set [list $var]]} {set changeIcon "->"}}                if {![info exists ::stepsource::varValues($var)]} {set changeIcon "->"}                if {$changeIcon == "->"} {puts $::stepsource::outChannel [format "%-30s %s %s" $var $changeIcon "[uplevel 3 set [list $var]]"]}                set varrayadd $var ; lappend varrayadd [uplevel 3 set [list $var]] ; array set ::stepsource::currentValues $varrayadd            }            set ::stepsource::varDefault c        }        e {            set level [expr [info level] - 1]            set ::stepsource::watchLevel $level            if {$level <= $::stepsource::highestLevel} {unset ::stepsource::watchLevel}        }        g {            foreach var [lsort [info globals]] {                if [array exists ::$var] {puts $::stepsource::outChannel [format "%-27s %s" $var Array:] ; continue}                set changeIcon "=="                catch {if {$::stepsource::varValues($var) != [set $var]} {set changeIcon "->"}}                puts $::stepsource::outChannel [format "%-30s %s %s" $var $changeIcon "[set ::$var]"]                set varrayadd $var ; lappend varrayadd [set ::$var] ; array set ::stepsource::currentValues $varrayadd            }        }        h {            puts $::stepsource::outChannel {\ 
     
          run until line number 
      
           run next line a        list array values b        run until next breakpoint b ?        list breakpoints b 
       
            set breakpoint b -
        
             unset breakpoint b -        unset all breakpoints c        list changed variable values e        run to end of current procedure g        list global variables h        help l        list all instrumented lines l 
         
           [
          
           ] list line numbers v list variable values x abort execution 
           
             execute as tcl command } } v { foreach var [lsort [uplevel 3 info vars]] { if ![uplevel 3 info exists [list $var]] {continue} if [uplevel 3 array exists [list $var]] {puts $::stepsource::outChannel [format "%-27s %s" $var Array:] ; continue} set changeIcon "==" catch {if {$::stepsource::varValues($var) != [uplevel 3 set [list $var]]} {set changeIcon "->"}} if {![info exists ::stepsource::varValues($var)]} {set changeIcon "->"} puts $::stepsource::outChannel [format "%-30s %s %s" $var $changeIcon "[uplevel 3 set [list $var]]"] set varrayadd $var ; lappend varrayadd [uplevel 3 set [list $var]] ; array set ::stepsource::currentValues $varrayadd } set ::stepsource::varDefault v } x { error "abort" } {} { set ::stepsource::currentBreakPoint 0 } default { catch {uplevel 3 $stepcommand} result puts $::stepsource::outChannel $result } } } proc StepNumber {linenumber} { set level [info level] if ![info exists ::stepsource::highestLevel] {set ::stepsource::highestLevel $level} if {$level < $::stepsource::highestLevel} {set $::stepsource::highestLevel $level} if ![info exists ::stepsource::currentBreakPoint] {set ::stepsource::currentBreakPoint 0} if {$::stepsource::currentBreakPoint > $::stepsource::lineCount} {set ::stepsource::currentBreakPoint $::stepsource::lineCount} set returnOK 1 catch { if {[info level] < $::stepsource::watchLevel} { unset ::stepsource::watchLevel set returnOK 0 } else { set ::stepsource::currentBreakPoint {} } } if {$::stepsource::currentBreakPoint == 0} {set returnOK 0} if {$linenumber == $::stepsource::currentBreakPoint} {unset ::stepsource::currentBreakPoint ; set returnOK 0} if {[lsearch -exact $::stepsource::breakPoints $linenumber] > -1} {set returnOK 0} if $returnOK {return} catch { set currentProcedure [lindex [info level -2] 0] if {[uplevel 2 info procs $currentProcedure] == {}} {set currentProcedure {}} } if ![info exists ::stepsource::lastProcedure] {set ::stepsource::lastProcedure {}} if ![info exists currentProcedure] {set currentProcedure {}} if {($level != $::stepsource::highestLevel) && ($::stepsource::lastProcedure != $currentProcedure)} {puts $::stepsource::outChannel "||||current procedure: $currentProcedure"} set ::stepsource::lastProcedure $currentProcedure set stepCommand $::stepsource::varDefault StepCommand $stepCommand while {$stepCommand != {}} { puts $::stepsource::outChannel "\n-----------------------------------" puts $::stepsource::outChannel $::stepsource::lineArray($linenumber)\n puts -nonewline $::stepsource::outChannel > set stepCommand [gets $::stepsource::inChannel] StepCommand $stepCommand if {([string is integer -strict $stepCommand]) || ($stepCommand == "b") || ($stepCommand == {}) || ($stepCommand == "e")} { catch {array set ::stepsource::varValues [array get ::stepsource::currentValues]} catch {array unset ::stepsource::currentValues} break } } } proc StepSource {filename} { namespace eval ::stepsource {} set ::stepsource::filename $filename namespace eval ::stepsource { if {[info procs original_unknown] == {}} { rename ::unknown original_unknown proc ::unknown {args} { if [string is integer -strict $args] { ::stepsource::StepNumber $args } else { set ::stepsource::unk_args $args uplevel 1 ::stepsource::original_unknown $::stepsource::unk_args } } } if ![info exists inChannel] {set inChannel stdin} if ![info exists outChannel] {set outChannel stdout} if ![info exists breakPoints] {set breakPoints {}} if ![info exists varDefault] {set varDefault v} if ![info exists sourcedFiles] {set sourcedFiles {}} if {[lsearch -exact $sourcedFiles $filename] < 0} {lappend sourcedFiles $filename} if ![info exists ::stepsource::sourceProcs] {set ::stepsource::sourceProcs {}} set mtime [file mtime $filename] set oldMtime 0 catch {set oldMtime $mtimes($filename)} array unset lineArray set lineCount 1 foreach sF $sourcedFiles { set $sF {} set f [open $sF r] set noNumberLine {} while {![eof $f]} { set line [gets $f] set firstWord [string trim [string range [string trim $line] 0 [expr [string wordend [string trim $line] 0] - 1]]] set secondWord [string trim [string range [string trim $line] [string length $firstWord] [string wordend [string trim $line] [expr [string length $firstWord] + 1]]]] if ![regexp {(::[^ ]+)(\ |$)} $line trash firstNameSpace] {set firstNameSpace {}} if {$firstWord == ":"} {set firstWord $firstNameSpace} if {[string index $secondWord 0] == ":"} {set secondWord $firstNameSpace} if {$firstWord == "proc"} {lappend ::stepsource::sourceProcs $secondWord} if {([info commands $firstWord] != {}) || ([lsearch -exact $::stepsource::sourceProcs $firstWord] > -1)} { set $sF "[set $sF]$noNumberLine[set lineCount]\;\t$line\n" set arrayadd $lineCount ; lappend arrayadd $noNumberLine$lineCount\;\t$line\n ; array set lineArray $arrayadd set noNumberLine {} incr lineCount } elseif {($firstWord == "\{") && (([info commands $secondWord] != {}) || ([lsearch -exact $::stepsource::sourceProcs $secondWord] > -1))} { set arrayadd $lineCount ; lappend arrayadd $noNumberLine$lineCount\;\t$line\n ; array set lineArray $arrayadd regsub {\{} $line "\{$lineCount\;" line set $sF "[set $sF]$noNumberLine\t$line\n" set noNumberLine {} incr lineCount } else { set noNumberLine $noNumberLine\t$line\n } } close $f if {$noNumberLine != {}} {set $sF "[set $sF]$noNumberLine"} } } set ::stepsource::sourceProcs {} uplevel 1 eval \$\{::stepsource::$::stepsource::filename\} } } # end namespace eval ::stepsource proc ::ss {args} { catch {unset ::stepsource::watchLevel} catch {unset ::stepsource::currentBreakPoint} catch {array unset ::stepsource::varValues} uplevel 1 $args } package provide stepsource $::stepsource::VERSION
           
          
         
        
       
      
     
    

使用

在tclsh下先source stepsource.tcl, 然后::stepsource::StepSource <your_tcl_script>就可以进行调试模式

它的命令如下：

command	Usage
<line#>	run until line number
<return>	run next line
a	list array values
b	run until next breakpoint
b ?	list breakpoints
b <line#>	set breakpoint
b -<line#>	unset breakpoint
b -	unset all breakpoints
c	list changed variable values
e	run to end of current procedure
g	list global variables
h	help
l	list all instrumented lines
l <line#> [<line#>]	list line numbers
v	list variae values
x	abort execion
<anything else>	execute as tcl command

启动后界面发下

直接按回车键就可以执行下一行可以输入puts $argv等任意tcl命令输入b 53在第53行加入breakpoint取消用b -53

---

RamDebugger

安装

RamDebugger需要下面四个库：

tcllib

tcllib下载地址：

用tclsh ./installer.tcl来启动安装界面

tklib

tklib下载地址：

直接解压缩到$TCLLIBPATH里就好

Img

Img下载地址：

同tklib, 直接解压缩到$TCLLIBPATH

tktreectrl

tktreectrl下载地址：

解压缩后， 在相应目录里执行./configure --prefix=/home/harriszh/.tcllib && make 在$TCLLIBPATH下新建目录treectrl2.4.1,然后把library里的两个tcl文件， libtreectrl2.4.so 和pkgIndex.tcl拷到treectrl2.4.11就可以了

RamDebugger

下载文件源文件后解压缩

在解压缩后的目录里执行tclsh ./RamDebugger.tcl就可以看到下面界面

使用

建议打开变量状态框，如上面图右侧

方案是点击下面红框的选项基本功能和其他相似，也是run/stop, add breakpoints, step/next等功能

比较

功能和DDT相似，但速度快了很多，一样没法停止正在执行的程序。

工具本身依赖较多库，所以安装比较麻烦如果要使用GUI来debug tcl，这个还是最优的免费解决方案

总结

作者比较使用了三个调试器后，最满意的还是stepsource, 安装,速度和功能都能满足需要， 不依赖其他库，功能简洁又满足需求。