[release-branch.r60] documentation: Debugging Go code with GDB tutorial.

««« CL 5168046 / 320d83ed794a
documentation: Debugging Go code with GDB tutorial.

R=adg, cw, lvd
CC=golang-dev
https://golang.org/cl/5168046

»»»

R=dsymonds
CC=golang-dev
https://golang.org/cl/5172053

2 files changed, 484 insertions, 0 deletions

diff --git a/doc/debugging_with_gdb.html b/doc/debugging_with_gdb.html
new file mode 100644
index 0000000000..04850c0266
--- /dev/null
+++ b/doc/debugging_with_gdb.html
@@ -0,0 +1,479 @@
+<!-- title Debugging Go Code with GDB -->
+
+<p><i>
+This applies to the 6g toolchain. Gccgo has native gdb support. Besides this
+overview you might want to consult the
+<a href="http://sourceware.org/gdb/current/onlinedocs/gdb/">GDB manual</a>.
+</i></p>
+
+<h2 id="Introduction">Introduction</h2>
+
+<p>
+When you compile and link your Go programs with the 6g/6l or 8g/8l toolchains
+on Linux, Mac OSX or FreeBSD, the resulting binaries contain DWARFv3
+debugging information that recent versions (>7.1) of the GDB debugger can
+use to inspect a live process or a core dump.
+</p>
+
+<p>
+Pass the <code>'-s'</code> flag to the linker to omit the debug information.
+</p>
+
+
+<h3 id="Common_Operations">Common Operations</h3>
+
+<ul>
+<li>
+Show file and line number for code 
+and set breakpoints:
+<pre>(gdb) <b>list</b>
+(gdb) <b>list <i>line</i></b>
+(gdb) <b>list <i>file.go</i>:<i>line</i></b>
+(gdb) <b>break <i>line</i></b>
+(gdb) <b>break <i>file.go</i>:<i>line</i></b>
+(gdb) <b>disas</b></pre>
+</li>
+<li>
+Unwind stack frames:
+<pre>(gdb) <b>bt</b>
+(gdb) <b>frame <i>n</i></b></pre>
+</li>
+<li>
+Show the name, type and location on the stack frame of local variables,
+arguments and return values:
+<pre>(gdb) <b>info locals</b>
+(gdb) <b>info args</b>
+(gdb) <b>p variable</b>
+(gdb) <b>whatis variable</b></pre>
+</li>
+<li>
+Show the name, type and location of global variables:
+<pre>(gdb) <b>info variables <i>regexp</i></b></pre>
+</li>
+</ul>
+
+
+<h3 id="Go_Extensions">Go Extensions</h3>
+
+<p>
+A recent extension mechanism to GDB allows it to load extension scripts for a
+given binary. The tool chain uses this to extend GDB with a handful of
+commands to inspect internals of the runtime code (such as goroutines) and to
+pretty print the built-in map, slice and channel types.
+</p>
+
+<ul>
+<li>
+Pretty printing a string, slice, map, channel or interface:
+<pre>(gdb) <b>p <i>var</i></b></pre>
+</li>
+<li>
+A $len() and $cap() function for strings, slices and maps:
+<pre>(gdb) <b>p $len(<i>var</i>)</b></pre>
+</li>
+<li>
+A function to cast interfaces to their dynamic types:
+<pre>(gdb) <b>p $dtype(<i>var</i>)</b>
+(gdb) <b>iface <i>var</i></b></pre>
+<p class="detail"><b>Known issue:</b> GDB can’t automatically find the dynamic
+type of an interface value if its long name differs from its short name
+(annoying when printing stacktraces, the pretty printer falls back to printing
+the short type name and a pointer).</p>
+</li>
+<li>
+Inspecting goroutines:
+<pre>(gdb) <b>info goroutines</b>
+(gdb) <b>goroutine <i>n</i> <i>cmd</i></b>
+(gdb) <b>help goroutine</b></pre>
+For example:
+<pre>(gdb) <b>goroutine 12 bt</b></pre>
+</li>
+</ul>
+
+<p>
+If you'd like to see how this works, or want to extend it, take a look at <a
+href="/src/pkg/runtime/runtime-gdb.py">src/pkg/runtime/runtime-gdb.py</a> in
+the Go source distribution. It depends on some special magic types
+(<code>hash&lt;T,U&gt;</code>) and variables (<code>runtime.m</code> and
+<code>runtime.g</code>) that the linker
+(<a href="/src/cmd/ld/dwarf.c">src/cmd/ld/dwarf.c</a>) ensures are described in
+the DWARF code.
+</ines
+
+<p>
+If you're interested in what the debugging information looks like, run
+'<code>objdump -W 6.out</code>' and browse through the <code>.debug_*</code>
+sections.
+</p>
+
+
+<h3 id="Known_Issues">Known Issues</h3>
+
+<ol>
+<li>String pretty printing only triggers for type string, not for types derived
+from it.</li>
+<li>Type information is missing for the C parts of the runtime library.</li>
+<li>GDB does not understand Go’s name qualifications and treats
+<code>"fmt.Print"</code> as an unstructured literal with a <code>"."</code>
+that needs to be quoted.  It objects even more strongly to method names of
+the form <code>pkg.(*MyType).Meth</code>.
+<li>All global variables are lumped into package <code>"main"</code>.</li>
+</ol>
+
+<h2 id="Tutorial">Tutorial</h2>
+
+<p>
+In this tutorial we will inspect the binary of the
+<a href="/pkg/regexp/">regexp</a> package's unit tests. To build the binary,
+change to <code>$GOROOT/src/pkg/regexp</code> and run <code>gotest</code>.
+This should produce an executable file named <code>6.out</code>.
+</p>
+
+
+<h3 id="Getting_Started">Getting Started</h3>
+
+<p>
+Launch GDB, debugging <code>6.out</code>:
+</p>
+
+<pre>
+$ <b>gdb 6.out</b>
+GNU gdb (GDB) 7.2-gg8
+Copyright (C) 2010 Free Software Foundation, Inc.
+License GPLv  3+: GNU GPL version 3 or later &lt;http://gnu.org/licenses/gpl.html&gt;
+Type "show copying" and "show warranty" for licensing/warranty details.
+This GDB was configured as "x86_64-linux".
+
+Reading symbols from  /home/user/go/src/pkg/regexp/6.out...
+done.
+Loading Go Runtime support.
+(gdb) 
+</pre>
+
+<p>
+The message <code>"Loading Go Runtime support"</code> means that GDB loaded the
+extension from <code>$GOROOT/src/pkg/runtime/runtime-gdb.py</code>.
+</p>
+
+<p>
+To help GDB find the Go runtime sources and the accompanying support script,
+pass your <code>$GOROOT</code> with the <code>'-d'</code> flag:
+</p>
+
+<pre>
+$ <b>gdb 6.out -d $GOROOT</b>
+</pre>
+
+<p>
+If for some reason GDB still can't find that directory or that script, you can load
+it by hand by telling gdb (assuming you have the go sources in
+<code>~/go/</code>):
+<p>
+
+<pre>
+(gdb) <b>source ~/go/src/pkg/runtime/runtime-gdb.py</b>
+Loading Go Runtime support.
+</pre>
+
+<h3 id="Inspecting_the_source">Inspecting the source</h3>
+
+<p>
+Use the <code>"l"</code> or <code>"list"</code> command to inspect source code.
+</p>
+
+<pre>
+(gdb) <b>l</b>
+</pre>
+
+<p>
+List a specific part of the source parametrizing <code>"list"</code> with a
+function name (it must be qualified with its package name).
+</p>
+
+<pre>
+(gdb) <b>l main.main</b>
+</pre>
+
+<p>
+List a specific file and line number:
+</p>
+
+<pre>
+(gdb) <b>l regexp.go:1</b>
+(gdb) <i># Hit enter to repeat last command. Here, this lists next 10 lines.</i>
+</pre>
+
+
+<h3 id="Naming">Naming</h3>
+
+<p>
+Variable and function names must be qualified with the name of the packages
+they belong to. The <code>Compile</code> function from the <code>regexp</code>
+package is known to GDB as <code>'regexp.Compile'</code>. 
+</p>
+
+<p>
+Methods must be qualified with the name of their receiver types. For example,
+the <code>*Regexp</code> type’s <code>doParse</code> method is known as
+<code>'regexp.*Regexp.doParse'</code>. (Note that the second dot is a "middot,"
+an artifact of Go’s internal representation of methods.)
+</p>
+
+<p>
+Variables that shadow other variables are magically suffixed with a number in the debug info.
+Variables referenced by closures will appear as pointers magically prefixed with '&amp'.
+</p>
+
+<h3 id="Setting_breakpoints">Setting breakpoints</h3>
+
+<p>
+Set a breakpoint at the <code>TestFind</code> function:
+</p>
+
+<pre>
+(gdb) <b>b 'regexp.TestFind'</b>
+Breakpoint 1 at 0x424908: file /home/user/go/src/pkg/regexp/find_test.go, line 148.
+</pre>
+
+<p>
+Run the program:
+</p>
+
+<pre>
+(gdb) <b>run</b>
+Starting program: /home/lvd/g/src/pkg/regexp/6.out 
+
+Breakpoint 1, regexp.TestFind (t=0xf8404a89c0) at /home/user/go/src/pkg/regexp/find_test.go:148
+148	func TestFind(t *testing.T) {
+</pre>
+
+<p>
+Execution has paused at the breakpoint.
+See which goroutines are running, and what they're doing:
+</p>
+
+<pre>
+(gdb) <b>info goroutines</b>
+  1  waiting runtime.gosched
+* 13  running runtime.goexit
+</pre>
+
+<p>
+the one marked with the <code>*</code> is the current goroutine.
+</p>
+
+<h3 id="Inspecting_the_stack">Inspecting the stack</h3>
+
+<p>
+Look at the stack trace for where we’ve paused the program:
+</p>
+
+<pre>
+(gdb) <b>bt</b>  <i># backtrace</i>
+#0  regexp.TestFind (t=0xf8404a89c0) at /home/user/go/src/pkg/regexp/find_test.go:148
+#1  0x000000000042f60b in testing.tRunner (t=0xf8404a89c0, test=0x573720) at /home/user/go/src/pkg/testing/testing.go:156
+#2  0x000000000040df64 in runtime.initdone () at /home/user/go/src/pkg/runtime/proc.c:242
+#3  0x000000f8404a89c0 in ?? ()
+#4  0x0000000000573720 in ?? ()
+#5  0x0000000000000000 in ?? ()
+</pre>
+
+<p>
+The other goroutine, number 1, is stuck in <code>runtime.gosched</code>, blocked on a channel receive:
+</p>
+
+<pre>
+(gdb) <b>goroutine 1 bt</b>
+#0  0x000000000040facb in runtime.gosched () at /home/lvd/g/src/pkg/runtime/proc.c:873
+#1  0x00000000004031c9 in runtime.chanrecv (c=void, ep=void, selected=void, received=void)
+ at  /home/lvd/g/src/pkg/runtime/chan.c:342
+#2  0x0000000000403299 in runtime.chanrecv1 (t=void, c=void) at/home/lvd/g/src/pkg/runtime/chan.c:423
+#3  0x000000000043075b in testing.RunTests (matchString={void (struct string, struct string, bool *, os.Error *)} 0x7ffff7f9ef60, tests=  []testing.InternalTest = {...}) at /home/lvd/g/src/pkg/testing/testing.go:201
+#4  0x00000000004302b1 in testing.Main (matchString={void (struct string, struct string, bool *, os.Error *)} 0x7ffff7f9ef80, tests= []testing.InternalTest = {...}, benchmarks= []testing.InternalBenchmark = {...})
+    at /home/lvd/g/src/pkg/testing/testing.go:168
+#5  0x0000000000400dc1 in main.main () at /home/lvd/g/src/pkg/regexp/_testmain.go:98
+#6  0x00000000004022e7 in runtime.mainstart () at /home/lvd/g/src/pkg/runtime/amd64/asm.s:78
+#7  0x000000000040ea6f in runtime.initdone () at /home/lvd/g/src/pkg/runtime/proc.c:243
+#8  0x0000000000000000 in ?? ()
+</pre>
+
+<p>
+The stack frame shows we’re currently executing the <code>regexp.TestFind</code> function, as expected.
+</p>
+
+<pre>
+(gdb) <b>info frame</b>
+Stack level 0, frame at 0x7ffff7f9ff88:
+ rip = 0x425530 in regexp.TestFind (/home/lvd/g/src/pkg/regexp/find_test.go:148); 
+    saved rip 0x430233
+ called by frame at 0x7ffff7f9ffa8
+ source language minimal.
+ Arglist at 0x7ffff7f9ff78, args: t=0xf840688b60
+ Locals at 0x7ffff7f9ff78, Previous frame's sp is 0x7ffff7f9ff88
+ Saved registers:
+  rip at 0x7ffff7f9ff80
+</pre>
+
+<p>
+The command <code>info locals</code> lists all variables local to the function and their values, but is a bit
+dangerous to use, since it will also try to print uninitialized variables. Uninitialized slices may cause gdb to try
+to print arbitrary large arrays.
+</p>
+
+<p>
+The function’s arguments:
+</p>
+
+<pre>
+(gdb) <b>info args</b>
+t = 0xf840688b60
+</pre>
+
+<p>
+When printing the argument, notice that it’s a pointer to a
+<code>Regexp</code> value. Note that GDB has incorrectly put the <code>*</code>
+on the right-hand side of the type name and made up a 'struct' keyword, in traditional C style.
+</p>
+
+<pre>
+(gdb) <b>p re</b>
+(gdb) p t
+$1 = (struct testing.T *) 0xf840688b60
+(gdb) p t
+$1 = (struct testing.T *) 0xf840688b60
+(gdb) p *t
+$2 = {errors = "", failed = false, ch = 0xf8406f5690}
+(gdb) p *t->ch
+$3 = struct hchan<*testing.T>
+</pre>
+
+<p>
+That <code>struct hchan<*testing.T></code> is the runtime-internal represntation of a channel.  It is currently empty, or gdb would have pretty-printed it's contents.
+</p>
+
+<p>
+Stepping forward:
+</p>
+
+<pre>
+(gdb) <b>n</b>  <i># execute next line</i>
+149             for _, test := range findTests {
+(gdb)    <i># enter is repeat</i>
+150                     re := MustCompile(test.pat)
+(gdb) <b>p test.pat</b>
+$4 = ""
+(gdb) <b>p re</b>
+$5 = (struct regexp.Regexp *) 0xf84068d070
+(gdb) <b>p *re</b>
+$6 = {expr = "", prog = 0xf840688b80, prefix = "", prefixBytes =  []uint8, prefixComplete = true, 
+  prefixRune = 0, cond = 0 '\000', numSubexp = 0, longest = false, mu = {state = 0, sema = 0}, 
+  machine =  []*regexp.machine}
+(gdb) <b>p *re->prog</b>
+$7 = {Inst =  []regexp/syntax.Inst = {{Op = 5 '\005', Out = 0, Arg = 0, Rune =  []int}, {Op = 
+    6 '\006', Out = 2, Arg = 0, Rune =  []int}, {Op = 4 '\004', Out = 0, Arg = 0, Rune =  []int}}, 
+  Start = 1, NumCap = 2}
+</pre>
+
+
+<p>
+We can step into the <code>String</code>function call with <code>"s"</code>:
+</p>
+
+<pre>
+(gdb) <b>s</b>
+regexp.(*Regexp).String (re=0xf84068d070, noname=void) at /home/lvd/g/src/pkg/regexp/regexp.go:97
+97      func (re *Regexp) String() string {
+</pre>
+
+<p>
+Get a stack trace to see where we are:
+</p>
+
+<pre>
+(gdb) <b>bt</b>
+(gdb) bt
+#0  regexp.(*Regexp).String (re=0xf84068d070, noname=void)
+    at /home/lvd/g/src/pkg/regexp/regexp.go:97
+#1  0x0000000000425615 in regexp.TestFind (t=0xf840688b60)
+    at /home/lvd/g/src/pkg/regexp/find_test.go:151
+#2  0x0000000000430233 in testing.tRunner (t=0xf840688b60, test=0x5747b8)
+    at /home/lvd/g/src/pkg/testing/testing.go:156
+#3  0x000000000040ea6f in runtime.initdone () at /home/lvd/g/src/pkg/runtime/proc.c:243
+....
+</pre>
+
+<p>
+Look at the source code:
+</p>
+
+<pre>
+(gdb) <b>l</b>
+92              mu      sync.Mutex
+93              machine []*machine
+94      }
+95
+96      // String returns the source text used to compile the regular expression.
+97      func (re *Regexp) String() string {
+98              return re.expr
+99      }
+100
+101     // Compile parses a regular expression and returns, if successful,
+</pre>
+
+<h3 id="Pretty_Printing">Pretty Printing</h3>
+
+<p>
+GDB's pretty printing mechanism is triggered by regexp matches on type names.  An example for slices:
+</p>
+
+<pre>
+(gdb) <b>p utf</b>
+$22 =  []uint8 = {0 '\000', 0 '\000', 0 '\000', 0 '\000'}
+</pre>
+
+<p>
+Since slices, arrays and strings are not C pointers, GDB can't interpret the subscripting operation for you, but
+you can look inside the runtime representation to do that (tab completion helps here):
+</p>
+<pre>
+
+(gdb) <b>p slc</b>
+$11 =  []int = {0, 0}
+(gdb) <b>p slc-&gt</b><i>&ltTAB&gt</i>
+array  slc    len    
+(gdb) <b>p slc->array</b>
+$12 = (int *) 0xf84057af00
+(gdb) <b>p slc->array[1]</b>
+$13 = 0</pre>
+
+
+
+<p>
+The extension functions $len and $cap work on strings, arrays and slices:
+</p>
+
+<pre>
+(gdb) <b>p $len(utf)</b>
+$23 = 4
+(gdb) <b>p $cap(utf)</b>
+$24 = 4
+</pre>
+
+<p>
+Channels and maps are 'reference' types, which gdb shows as pointers to C++-like types <code>hash&ltint,string&gt*</code>.  Dereferencing will trigger prettyprinting
+</p>
+
+<p>
+Interfaces are represented in the runtime as a pointer to a type descriptor and a pointer to a value.  The Go GDB runtime extension decodes this and automatically triggers pretty printing for the runtime type.  The extension function <code>$dtype</code> decodes the dynamic type for you (examples are taken from a breakpoint at <code>regexp.go</code> line 293.)
+</p>
+
+<pre>
+(gdb) <b>p i</b>
+$4 = {str = "cbb"}
+(gdb) <b>whatis i</b>
+type = regexp.input
+(gdb) <b>p $dtype(i)</b>
+$26 = (struct regexp.inputBytes *) 0xf8400b4930
+(gdb) <b>iface i</b>
+regexp.input: struct regexp.inputBytes *
+</pre>
diff --git a/doc/docs.html b/doc/docs.html
index 288cfd0bd0..7bc5c77d8e 100644
--- a/doc/docs.html
+++ b/doc/docs.html
@@ -181,6 +181,11 @@ one goroutine can be guaranteed to observe values produced by writes to the
 same variable in a different goroutine.
 </p>
 
+<h3 id="debugging_with_gdb"><a href="debugging_with_gdb.html">Debugging Go Code with GDB</a></h3>
+<p>
+Using GDB to debug Go programs.
+</p>
+
 <h2 id="videos_talks">Videos and Talks</h2>
 
 <h3 id="writing_web_apps"><a href="http://www.youtube.com/watch?v=-i0hat7pdpk">Writing Web Apps in Go</a></h3>