diff options
Diffstat (limited to 'www/formats.html')
| -rwxr-xr-x | www/formats.html | 288 |
1 files changed, 288 insertions, 0 deletions
diff --git a/www/formats.html b/www/formats.html new file mode 100755 index 000000000000..58db831d596f --- /dev/null +++ b/www/formats.html @@ -0,0 +1,288 @@ +<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"> +<html xmlns="http://www.w3.org/1999/xhtml"> +<head> +<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1" /> +<link href="style.css" rel="stylesheet" type="text/css" /> +<title>LLDB Stack and Frame Formats</title> +</head> + +<body> + <div class="www_title"> + The <strong>LLDB</strong> Debugger + </div> + +<div id="container"> + <div id="content"> + <!--#include virtual="sidebar.incl"--> + + <div id="middle"> + <div class="post"> + <h1 class ="postheader">Stack Frame and Thread Format</h1> + <div class="postcontent"> + <p>LLDB has a facility to allow users to define the + format of the information that generates the descriptions + for threads and stack frames. Typically when your program stops + at a breakpoint you will get a line that describes why + your thread stopped:</p> + + <p><b><code>* thread #1: tid = 0x2e03, 0x0000000100000e85 a.out`main + 4, stop reason = breakpoint 1.1</code></b></p> + + <p>Stack backtraces frames also have a similar information line:</p> + + <p><code><b>(lldb)</b> thread backtrace + <br><b>thread #1: tid = 0x2e03, stop reason = breakpoint 1.1 + <br> frame #0: 0x0000000100000e85 a.out`main + 4 at test.c:19 + <br> frame #1: 0x0000000100000e40 a.out`start + 52 + </code></b></p> + + <p>The two format strings can currently be set using the <b>settings set</b> command:</p> + <p><code><b>(lldb)</b> settings set frame-format STRING + <br><b>(lldb)</b> settings set thread-format STRING + </p></code> + + </div> + <div class="postfooter"></div> + </div> + + <div class="post"> + <h1 class ="postheader">Format Strings</h1> + <div class="postcontent"> + + <p>So what is the format of the format strings? Format strings can + contain plain text, control characters and variables that have access + to the current program state.</p> + + <p>Normal characters are any text that doesn't contain a <code><b>'{'</b></code>, <code><b>'}'</b></code>, <code><b>'$'</b></code>, + or <code><b>'\'</b></code> character.</p> + + <p>Variable names are found in between a <code><b>"${"</b></code> prefix, and + end with a <code><b>"}"</b></code> suffix. In other words, a variable looks like + <code>"<b>${frame.pc}</b>"</code>.</p> + + </div> + <div class="postfooter"></div> + </div> + + <div class="post"> + <h1 class ="postheader">Variables</h1> + <div class="postcontent"> + + <p>A complete list of currently supported format string variables is listed below:</p> + + <table border="1"> + <tr valign=top><td><b>Variable Name</b></td><td><b>Description</b></td></tr> + <tr valign=top><td><b>file.basename</b></td><td>The current compile unit file basename for the current frame.</td></tr> + <tr valign=top><td><b>file.fullpath</b></td><td>The current compile unit file fullpath for the current frame.</td></tr> + <tr valign=top><td><b>language</b></td><td>The current compile unit language for the current frame.</td></tr> + <tr valign=top><td><b>frame.index</b></td><td>The frame index (0, 1, 2, 3...)</td></tr> + <tr valign=top><td><b>frame.pc</b></td><td>The generic frame register for the program counter.</td></tr> + <tr valign=top><td><b>frame.sp</b></td><td>The generic frame register for the stack pointer.</td></tr> + <tr valign=top><td><b>frame.fp</b></td><td>The generic frame register for the frame pointer.</td></tr> + <tr valign=top><td><b>frame.flags</b></td><td>The generic frame register for the flags register.</td></tr> + <tr valign=top><td><b>frame.reg.NAME</b></td><td>Access to any platform specific register by name (replace <b>NAME</b> with the name of the desired register).</td></tr> + <tr valign=top><td><b>function.name</b></td><td>The name of the current function or symbol.</td></tr> + <tr valign=top><td><b>function.name-with-args</b></td><td>The name of the current function with arguments and values or the symbol name.</td></tr> + <tr valign=top><td><b>function.name-without-args</b></td><td>The name of the current function without arguments and values (used to include a function name in-line in the <tt>disassembly-format</tt>)</td></tr> + <tr valign=top><td><b>function.pc-offset</b></td><td>The program counter offset within the current function or symbol</td></tr> + <tr valign=top><td><b>function.addr-offset</b></td><td>The offset in bytes of the current function, formatted as " + dddd"</td></tr> + <tr valign=top><td><b>function.concrete-only-addr-offset-no-padding</b></td><td>Similar to <b>function.addr-offset</b> except that there are no spaces in the output (e.g. "+dddd") and the offset is computed from the nearest concrete function -- inlined functions are not included</td></tr> + <tr valign=top><td><b>function.changed</b></td><td>Will evaluate to true when the line being formatted is a different symbol context from the previous line (may be used in <tt>disassembly-format</tt> to print the new function name on a line by itself at the start of a new function). Inlined functions are not considered for this variable</td></tr> + <tr valign=top><td><b>function.initial-function</b></td><td>Will evaluate to true if this is the start of the first function, as opposed to a change of functions (may be used in <tt>disassembly-format</tt> to print the function name for the first function being disassembled)</td></tr> + <tr valign=top><td><b>line.file.basename</b></td><td>The line table entry basename to the file for the current line entry in the current frame.</td></tr> + <tr valign=top><td><b>line.file.fullpath</b></td><td>The line table entry fullpath to the file for the current line entry in the current frame.</td></tr> + <tr valign=top><td><b>line.number</b></td><td>The line table entry line number for the current line entry in the current frame.</td></tr> + <tr valign=top><td><b>line.start-addr</b></td><td>The line table entry start address for the current line entry in the current frame.</td></tr> + <tr valign=top><td><b>line.end-addr</b></td><td>The line table entry end address for the current line entry in the current frame.</td></tr> + <tr valign=top><td><b>module.file.basename</b></td><td>The basename of the current module (shared library or executable)</td></tr> + <tr valign=top><td><b>module.file.fullpath</b></td><td>The basename of the current module (shared library or executable)</td></tr> + <tr valign=top><td><b>process.file.basename</b></td><td>The basename of the file for the process</td></tr> + <tr valign=top><td><b>process.file.fullpath</b></td><td>The fullname of the file for the process</td></tr> + <tr valign=top><td><b>process.id</b></td><td>The process ID native to the system on which the inferior runs.</td></tr> + <tr valign=top><td><b>process.name</b></td><td>The name of the process at runtime</td></tr> + <tr valign=top><td><b>thread.id</b></td><td>The thread identifier for the current thread</td></tr> + <tr valign=top><td><b>thread.index</b></td><td>The unique one based thread index ID which is guaranteed to be unique as threads come and go.</td></tr> + <tr valign=top><td><b>thread.name</b></td><td>The name of the thread if the target OS supports naming threads</td></tr> + <tr valign=top><td><b>thread.queue</b></td><td>The queue name of the thread if the target OS supports dispatch queues</td></tr> + <tr valign=top><td><b>thread.stop-reason</b></td><td>A textual reason each thread stopped</td></tr> + <tr valign=top><td><b>thread.return-value</b></td><td>The return value of the latest step operation (currently only for step-out.)</td></tr> + <tr valign=top><td><b>thread.completed-expression</b></td><td>The expression result for a thread that just finished an interrupted expression evaluation.</td></tr> + <tr valign=top><td><b>target.arch</b></td><td>The architecture of the current target</td></tr> + <tr valign=top><td><b>target.script:<i>python_func</i></b></td><td>Use a Python function to generate a piece of textual output</td></tr> + <tr valign=top><td><b>process.script:<i>python_func</i></b></td><td>Use a Python function to generate a piece of textual output</td></tr> + <tr valign=top><td><b>thread.script:<i>python_func</i></b></td><td>Use a Python function to generate a piece of textual output</td></tr> + <tr valign=top><td><b>frame.script:<i>python_func</i></b></td><td>Use a Python function to generate a piece of textual output</td></tr> + <tr valign=top><td><b>current-pc-arrow</b></td><td>Prints either '<tt>-> </tt>' or '<tt> </tt>' if the current pc value is matched (used in <tt>disassembly-format</tt>)</td></tr> + <tr valign=top><td><b>addr-file-or-load</b></td><td>Formats an address either as a load address, or if process has not yet been launched, as a load address (used in <tt>disassembly-format</tt>)</td></tr> + </table> + + </div> + <div class="postfooter"></div> + </div> + + <div class="post"> + <h1 class ="postheader">Control Characters</h1> + <div class="postcontent"> + + <p>Control characters include <b><code>'{'</code></b>, + <b><code>'}'</code></b>, and <b><code>'\'</code></b>.</p> + + <p>The '{' and '}' are used for scoping blocks, and the '\' character + allows you to desensitize control characters and also emit non-printable + characters. + + </div> + <div class="postfooter"></div> + </div> + + <div class="post"> + <h1 class ="postheader">Desensitizing Characters in the format string</h1> + <div class="postcontent"> + <p>The backslash control character allows your to enter the typical + <b><code>"\a"</code></b>, <b><code>"\b"</code></b>, <b><code>"\f"</code></b>, <b><code>"\n"</code></b>, + <b><code>"\r"</code></b>, <b><code>"\t"</code></b>, <b><code>"\v"</code></b>, <b><code>"\\"</code></b>, characters + and along with the standard octal representation <b><code>"\0123"</code></b> + and hex <b><code>"\xAB"</code></b> characters. This allows you to enter + escape characters into your format strings and will + allow colorized output for terminals that support color. + + </div> + <div class="postfooter"></div> + </div> + + <div class="post"> + <h1 class ="postheader">Scoping</h1> + <div class="postcontent"> + <p>Many times the information that you might have in your prompt might not be + available and you won't want it to print out if it isn't valid. To take care + of this you can enclose everything that <b>must</b> resolve into a scope. A scope + is starts with <code><b>'{'</code></b> and ends with + <code><b>'}'</code></b>. For example in order to only display + the current frame line table entry basename and line number when the information + is available for the current frame: + + <p><b><code>"{ at {$line.file.basename}:${line.number}}"</code></b></p> + + <p>Broken down this is: + <ul> + <li>The start the scope <p><b><code>"{"</code></b></p></li> + <li> format whose content will only be displayed if all information is available: + <p><b><code>"at {$line.file.basename}:${line.number}"</code></b></p></li> + <li>end the scope: <p><b><code>"}"</code></b></p></li> + </ul> + + </div> + <div class="postfooter"></div> + </div> + + <div class="post"> + <h1 class ="postheader">Making the Frame Format</h1> + <div class="postcontent"> + <p>The information that we see when stopped in a frame: + + <p><b><code>frame #0: 0x0000000100000e85 a.out`main + 4 at test.c:19</code></b></p> + + <p>can be displayed with the following format:</p> + + <p><b><code>"frame #${frame.index}: ${frame.pc}{ ${module.file.basename}`${function.name}{${function.pc-offset}}}{ at ${line.file.basename}:${line.number}}\n"</code></b></p> + + <p>This breaks down to: + + <ul> + <li>Always print the frame index and frame PC: + + <p><b><code>"frame #${frame.index}: ${frame.pc}"</code></b></p> + + <li>only print the module followed by a tick if there is a valid + module for the current frame: + + <p><b><code>"{ ${module.file.basename}`}"</code></b></p> + + <li>print the function name with optional offset:</p> + <p><b><code>"{${function.name}{${function.pc-offset}}}"</code></b></p> + + <li>print the line info if it is available:</p> + + <p><b><code>"{ at ${line.file.basename}:${line.number}}"</code></b></p> + + <li>then finish off with a newline:</p> + + <p><b><code>"\n"</code></b></p> + </ul> + + </div> + <div class="postfooter"></div> + </div> + + <div class="post"> + <h1 class ="postheader">Making Your Own Formats</h1> + <div class="postcontent"> + + <p>When modifying your own format strings, it is useful + to start with the default values for the frame and + thread format strings. These can be accessed with the + <b><code>"settings show"</code></b> command: + + <p><b><code>(lldb)</b> settings show thread-format + <br>thread-format (string) = 'thread #${thread.index}: tid = ${thread.id}{, ${frame.pc}}{ ${module.file.basename}`${function.name}{${function.pc-offset}}}{, stop reason = ${thread.stop-reason}}{, name = ${thread.name}}{, queue = ${thread.queue}}\n' + <br><b>(lldb)</b> settings show frame-format + <br>frame-format (string) = 'frame #${frame.index}: ${frame.pc}{ ${module.file.basename}`${function.name}{${function.pc-offset}}}{ at ${line.file.basename}:${line.number}}\n' + </code></p> + + + <p>When making thread formats, you will need surround any + of the information that comes from a stack frame with scopes (<b>{</b> <i>frame-content</i> <b>}</b>) + as the thread format doesn't always want to show frame information. + When displaying the backtrace for a thread, we don't need to duplicate + the information for frame zero in the thread information: + + <p><code><b>(lldb)</b> thread backtrace + <br>thread #1: tid = 0x2e03, stop reason = breakpoint 1.1 2.1 + <br> frame #0: 0x0000000100000e85 a.out`main + 4 at test.c:19 + <br> frame #1: 0x0000000100000e40 a.out`start + 52 + </code> + </p> + + <p>The frame related variables are: + <ul> + <li><code><b>${file.*}</b></code></li> + <li><code><b>${frame.*}</b></code></li> + <li><code><b>${function.*}</b></code></li> + <li><code><b>${line.*}</b></code></li> + <li><code><b>${module.*}</b></code></li> + </ul> + </p> + + <p>Looking at the default format for the thread, and underlining + the frame information: + <p><code>'thread #${thread.index}: tid = ${thread.id}<u><b>{</b>, ${frame.pc}<b>}{</b> ${module.file.basename}`${function.name}{${function.pc-offset}}<b>}</b></u>{, stop reason = ${thread.stop-reason}}{, name = ${thread.name}}{, queue = ${thread.queue}}\n' + </code></p> + <p>We can see that all frame information is contained in scopes so + that when the thread information is displayed in a context where + we only want to show thread information, we can do so. + + <p>For both thread and frame formats, you can use ${target.script:<i>python_func</i>}, ${process.script:<i>python_func</i>} and ${thread.script:<i>python_func</i>} + (and of course ${frame.script:<i>python_func</i>} for frame formats)<br/> + In all cases, the signature of <i>python_func</i> is expected to be:<br/> + <p><code> + def <i>python_func</i>(<i>object</i>,unused):<br/> + ...<br/> + return <i>string</i><br/></code> + <p>Where <i>object</i> is an instance of the SB class associated to the keyword you are using. + + <p>e.g. Assuming your function looks like<br/><code><p> + def thread_printer_func (thread,unused):<br/> + return "Thread %s has %d frames\n" % (thread.name, thread.num_frames)<br/></code><p> + + And you set it up with <code><br/><b>(lldb)</b> settings set thread-format "${thread.script:thread_printer_func}"<br/></code> + you would see output like: + </p> + <code>* Thread main has 21 frames</code> + </div> + <div class="postfooter"></div> + </div> + + </div> + </div> +</div> +</body> +</html> |