path: root/devdocs/gcc~13/instrumentation-options.html

<div class="section-level-extent" id="Instrumentation-Options"> <div class="nav-panel"> <p> Next: <a href="preprocessor-options" accesskey="n" rel="next">Options Controlling the Preprocessor</a>, Previous: <a href="optimize-options" accesskey="p" rel="prev">Options That Control Optimization</a>, Up: <a href="invoking-gcc" accesskey="u" rel="up">GCC Command Options</a> [<a href="index#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="indices" title="Index" rel="index">Index</a>]</p> </div>  <h1 class="section" id="Program-Instrumentation-Options"><span>3.12 Program Instrumentation Options<a class="copiable-link" href="#Program-Instrumentation-Options"> ¶</a></span></h1>        <p>GCC supports a number of command-line options that control adding run-time instrumentation to the code it normally generates. For example, one purpose of instrumentation is collect profiling statistics for use in finding program hot spots, code coverage analysis, or profile-guided optimizations. Another class of program instrumentation is adding run-time checking to detect programming errors like invalid pointer dereferences or out-of-bounds array accesses, as well as deliberately hostile attacks such as stack smashing or C++ vtable hijacking. There is also a general hook which can be used to implement other forms of tracing or function-level instrumentation for debug or program analysis purposes. </p> <dl class="table"> <dt>
   <span><code class="code">-p</code><a class="copiable-link" href="#index-prof"> ¶</a></span>
</dt> <dt><code class="code">-pg</code></dt> <dd>
<p>Generate extra code to write profile information suitable for the analysis program <code class="command">prof</code> (for <samp class="option">-p</samp>) or <code class="command">gprof</code> (for <samp class="option">-pg</samp>). You must use this option when compiling the source files you want data about, and you must also use it when linking. </p> <p>You can use the function attribute <code class="code">no_instrument_function</code> to suppress profiling of individual functions when compiling with these options. See <a class="xref" href="common-function-attributes">Common Function Attributes</a>. </p> </dd> <dt>
<span><code class="code">-fprofile-arcs</code><a class="copiable-link" href="#index-fprofile-arcs"> ¶</a></span>
</dt> <dd>
<p>Add code so that program flow <em class="dfn">arcs</em> are instrumented. During execution the program records how many times each branch and call is executed and how many times it is taken or returns. On targets that support constructors with priority support, profiling properly handles constructors, destructors and C++ constructors (and destructors) of classes which are used as a type of a global variable. </p> <p>When the compiled program exits it saves this data to a file called <samp class="file"><var class="var">auxname</var>.gcda</samp> for each source file. The data may be used for profile-directed optimizations (<samp class="option">-fbranch-probabilities</samp>), or for test coverage analysis (<samp class="option">-ftest-coverage</samp>). Each object file’s <var class="var">auxname</var> is generated from the name of the output file, if explicitly specified and it is not the final executable, otherwise it is the basename of the source file. In both cases any suffix is removed (e.g. <samp class="file">foo.gcda</samp> for input file <samp class="file">dir/foo.c</samp>, or <samp class="file">dir/foo.gcda</samp> for output file specified as <samp class="option">-o dir/foo.o</samp>). </p> <p>Note that if a command line directly links source files, the corresponding <var class="var">.gcda</var> files will be prefixed with the unsuffixed name of the output file. E.g. <code class="code">gcc a.c b.c -o binary</code> would generate <samp class="file">binary-a.gcda</samp> and <samp class="file">binary-b.gcda</samp> files. </p> <p>See <a class="xref" href="cross-profiling">Data File Relocation to Support Cross-Profiling</a>. </p> </dd> <dt>
 <span><code class="code">--coverage</code><a class="copiable-link" href="#index-gcov"> ¶</a></span>
</dt> <dd> <p>This option is used to compile and link code instrumented for coverage analysis. The option is a synonym for <samp class="option">-fprofile-arcs</samp> <samp class="option">-ftest-coverage</samp> (when compiling) and <samp class="option">-lgcov</samp> (when linking). See the documentation for those options for more details. </p> <ul class="itemize mark-bullet"> <li>Compile the source files with <samp class="option">-fprofile-arcs</samp> plus optimization and code generation options. For test coverage analysis, use the additional <samp class="option">-ftest-coverage</samp> option. You do not need to profile every source file in a program. </li>
<li>Compile the source files additionally with <samp class="option">-fprofile-abs-path</samp> to create absolute path names in the <samp class="file">.gcno</samp> files. This allows <code class="command">gcov</code> to find the correct sources in projects where compilations occur with different working directories. </li>
<li>Link your object files with <samp class="option">-lgcov</samp> or <samp class="option">-fprofile-arcs</samp> (the latter implies the former). </li>
<li>Run the program on a representative workload to generate the arc profile information. This may be repeated any number of times. You can run concurrent instances of your program, and provided that the file system supports locking, the data files will be correctly updated. Unless a strict ISO C dialect option is in effect, <code class="code">fork</code> calls are detected and correctly handled without double counting. <p>Moreover, an object file can be recompiled multiple times and the corresponding <samp class="file">.gcda</samp> file merges as long as the source file and the compiler options are unchanged. </p> </li>
<li>For profile-directed optimizations, compile the source files again with the same optimization and code generation options plus <samp class="option">-fbranch-probabilities</samp> (see <a class="pxref" href="optimize-options">Options that Control Optimization</a>). </li>
<li>For test coverage analysis, use <code class="command">gcov</code> to produce human readable information from the <samp class="file">.gcno</samp> and <samp class="file">.gcda</samp> files. Refer to the <code class="command">gcov</code> documentation for further information. </li>
</ul> <p>With <samp class="option">-fprofile-arcs</samp>, for each function of your program GCC creates a program flow graph, then finds a spanning tree for the graph. Only arcs that are not on the spanning tree have to be instrumented: the compiler adds code to count the number of times that these arcs are executed. When an arc is the only exit or only entrance to a block, the instrumentation code can be added to the block; otherwise, a new basic block must be created to hold the instrumentation code. </p> </dd> <dt>
<span><code class="code">-ftest-coverage</code><a class="copiable-link" href="#index-ftest-coverage"> ¶</a></span>
</dt> <dd>
<p>Produce a notes file that the <code class="command">gcov</code> code-coverage utility (see <a class="pxref" href="gcov"><code class="command">gcov</code>—a Test Coverage Program</a>) can use to show program coverage. Each source file’s note file is called <samp class="file"><var class="var">auxname</var>.gcno</samp>. Refer to the <samp class="option">-fprofile-arcs</samp> option above for a description of <var class="var">auxname</var> and instructions on how to generate test coverage data. Coverage data matches the source files more closely if you do not optimize. </p> </dd> <dt>
<span><code class="code">-fprofile-abs-path</code><a class="copiable-link" href="#index-fprofile-abs-path"> ¶</a></span>
</dt> <dd>
<p>Automatically convert relative source file names to absolute path names in the <samp class="file">.gcno</samp> files. This allows <code class="command">gcov</code> to find the correct sources in projects where compilations occur with different working directories. </p> </dd> <dt>
<span><code class="code">-fprofile-dir=<var class="var">path</var></code><a class="copiable-link" href="#index-fprofile-dir"> ¶</a></span>
</dt> <dd> <p>Set the directory to search for the profile data files in to <var class="var">path</var>. This option affects only the profile data generated by <samp class="option">-fprofile-generate</samp>, <samp class="option">-ftest-coverage</samp>, <samp class="option">-fprofile-arcs</samp> and used by <samp class="option">-fprofile-use</samp> and <samp class="option">-fbranch-probabilities</samp> and its related options. Both absolute and relative paths can be used. By default, GCC uses the current directory as <var class="var">path</var>, thus the profile data file appears in the same directory as the object file. In order to prevent the file name clashing, if the object file name is not an absolute path, we mangle the absolute path of the <samp class="file"><var class="var">sourcename</var>.gcda</samp> file and use it as the file name of a <samp class="file">.gcda</samp> file. See details about the file naming in <samp class="option">-fprofile-arcs</samp>. See similar option <samp class="option">-fprofile-note</samp>. </p> <p>When an executable is run in a massive parallel environment, it is recommended to save profile to different folders. That can be done with variables in <var class="var">path</var> that are exported during run-time: </p> <dl class="table"> <dt><code class="code">%p</code></dt> <dd>
<p>process ID. </p> </dd> <dt><code class="code">%q{VAR}</code></dt> <dd>
<p>value of environment variable <var class="var">VAR</var> </p> </dd> </dl> </dd> <dt>
<span><code class="code">-fprofile-generate</code><a class="copiable-link" href="#index-fprofile-generate"> ¶</a></span>
</dt> <dt><code class="code">-fprofile-generate=<var class="var">path</var></code></dt> <dd> <p>Enable options usually used for instrumenting application to produce profile useful for later recompilation with profile feedback based optimization. You must use <samp class="option">-fprofile-generate</samp> both when compiling and when linking your program. </p> <p>The following options are enabled: <samp class="option">-fprofile-arcs</samp>, <samp class="option">-fprofile-values</samp>, <samp class="option">-finline-functions</samp>, and <samp class="option">-fipa-bit-cp</samp>. </p> <p>If <var class="var">path</var> is specified, GCC looks at the <var class="var">path</var> to find the profile feedback data files. See <samp class="option">-fprofile-dir</samp>. </p> <p>To optimize the program based on the collected profile information, use <samp class="option">-fprofile-use</samp>. See <a class="xref" href="optimize-options">Options That Control Optimization</a>, for more information. </p> </dd> <dt>
<span><code class="code">-fprofile-info-section</code><a class="copiable-link" href="#index-fprofile-info-section"> ¶</a></span>
</dt> <dt><code class="code">-fprofile-info-section=<var class="var">name</var></code></dt> <dd> <p>Register the profile information in the specified section instead of using a constructor/destructor. The section name is <var class="var">name</var> if it is specified, otherwise the section name defaults to <code class="code">.gcov_info</code>. A pointer to the profile information generated by <samp class="option">-fprofile-arcs</samp> is placed in the specified section for each translation unit. This option disables the profile information registration through a constructor and it disables the profile information processing through a destructor. This option is not intended to be used in hosted environments such as GNU/Linux. It targets freestanding environments (for example embedded systems) with limited resources which do not support constructors/destructors or the C library file I/O. </p> <p>The linker could collect the input sections in a continuous memory block and define start and end symbols. A GNU linker script example which defines a linker output section follows: </p> <div class="example smallexample"> <pre class="example-preformatted" data-language="cpp">.gcov_info      :
{
  PROVIDE (__gcov_info_start = .);
  KEEP (*(.gcov_info))
  PROVIDE (__gcov_info_end = .);
}</pre>
</div> <p>The program could dump the profiling information registered in this linker set for example like this: </p> <div class="example smallexample"> <pre class="example-preformatted" data-language="cpp">#include &lt;gcov.h&gt;
#include &lt;stdio.h&gt;
#include &lt;stdlib.h&gt;

extern const struct gcov_info *const __gcov_info_start[];
extern const struct gcov_info *const __gcov_info_end[];

static void
dump (const void *d, unsigned n, void *arg)
{
  const unsigned char *c = d;

  for (unsigned i = 0; i &lt; n; ++i)
    printf ("%02x", c[i]);
}

static void
filename (const char *f, void *arg)
{
  __gcov_filename_to_gcfn (f, dump, arg );
}

static void *
allocate (unsigned length, void *arg)
{
  return malloc (length);
}

static void
dump_gcov_info (void)
{
  const struct gcov_info *const *info = __gcov_info_start;
  const struct gcov_info *const *end = __gcov_info_end;

  /* Obfuscate variable to prevent compiler optimizations.  */
  __asm__ ("" : "+r" (info));

  while (info != end)
  {
    void *arg = NULL;
    __gcov_info_to_gcda (*info, filename, dump, allocate, arg);
    putchar ('\n');
    ++info;
  }
}

int
main (void)
{
  dump_gcov_info ();
  return 0;
}</pre>
</div> <p>The <code class="command">merge-stream</code> subcommand of <code class="command">gcov-tool</code> may be used to deserialize the data stream generated by the <code class="code">__gcov_filename_to_gcfn</code> and <code class="code">__gcov_info_to_gcda</code> functions and merge the profile information into <samp class="file">.gcda</samp> files on the host filesystem. </p> </dd> <dt>
<span><code class="code">-fprofile-note=<var class="var">path</var></code><a class="copiable-link" href="#index-fprofile-note"> ¶</a></span>
</dt> <dd> <p>If <var class="var">path</var> is specified, GCC saves <samp class="file">.gcno</samp> file into <var class="var">path</var> location. If you combine the option with multiple source files, the <samp class="file">.gcno</samp> file will be overwritten. </p> </dd> <dt>
<span><code class="code">-fprofile-prefix-path=<var class="var">path</var></code><a class="copiable-link" href="#index-fprofile-prefix-path"> ¶</a></span>
</dt> <dd> <p>This option can be used in combination with <samp class="option">profile-generate=</samp><var class="var">profile_dir</var> and <samp class="option">profile-use=</samp><var class="var">profile_dir</var> to inform GCC where is the base directory of built source tree. By default <var class="var">profile_dir</var> will contain files with mangled absolute paths of all object files in the built project. This is not desirable when directory used to build the instrumented binary differs from the directory used to build the binary optimized with profile feedback because the profile data will not be found during the optimized build. In such setups <samp class="option">-fprofile-prefix-path=</samp><var class="var">path</var> with <var class="var">path</var> pointing to the base directory of the build can be used to strip the irrelevant part of the path and keep all file names relative to the main build directory. </p> </dd> <dt>
<span><code class="code">-fprofile-prefix-map=<var class="var">old</var>=<var class="var">new</var></code><a class="copiable-link" href="#index-fprofile-prefix-map"> ¶</a></span>
</dt> <dd>
<p>When compiling files residing in directory <samp class="file"><var class="var">old</var></samp>, record profiling information (with <samp class="option">--coverage</samp>) describing them as if the files resided in directory <samp class="file"><var class="var">new</var></samp> instead. See also <samp class="option">-ffile-prefix-map</samp> and <samp class="option">-fcanon-prefix-map</samp>. </p> </dd> <dt>
<span><code class="code">-fprofile-update=<var class="var">method</var></code><a class="copiable-link" href="#index-fprofile-update"> ¶</a></span>
</dt> <dd> <p>Alter the update method for an application instrumented for profile feedback based optimization. The <var class="var">method</var> argument should be one of ‘<samp class="samp">single</samp>’, ‘<samp class="samp">atomic</samp>’ or ‘<samp class="samp">prefer-atomic</samp>’. The first one is useful for single-threaded applications, while the second one prevents profile corruption by emitting thread-safe code. </p> <p><strong class="strong">Warning:</strong> When an application does not properly join all threads (or creates an detached thread), a profile file can be still corrupted. </p> <p>Using ‘<samp class="samp">prefer-atomic</samp>’ would be transformed either to ‘<samp class="samp">atomic</samp>’, when supported by a target, or to ‘<samp class="samp">single</samp>’ otherwise. The GCC driver automatically selects ‘<samp class="samp">prefer-atomic</samp>’ when <samp class="option">-pthread</samp> is present in the command line. </p> </dd> <dt>
<span><code class="code">-fprofile-filter-files=<var class="var">regex</var></code><a class="copiable-link" href="#index-fprofile-filter-files"> ¶</a></span>
</dt> <dd> <p>Instrument only functions from files whose name matches any of the regular expressions (separated by semi-colons). </p> <p>For example, <samp class="option">-fprofile-filter-files=main\.c;module.*\.c</samp> will instrument only <samp class="file">main.c</samp> and all C files starting with ’module’. </p> </dd> <dt>
<span><code class="code">-fprofile-exclude-files=<var class="var">regex</var></code><a class="copiable-link" href="#index-fprofile-exclude-files"> ¶</a></span>
</dt> <dd> <p>Instrument only functions from files whose name does not match any of the regular expressions (separated by semi-colons). </p> <p>For example, <samp class="option">-fprofile-exclude-files=/usr/.*</samp> will prevent instrumentation of all files that are located in the <samp class="file">/usr/</samp> folder. </p> </dd> <dt>
<span><code class="code">-fprofile-reproducible=<span class="r">[</span>multithreaded<span class="r">|</span>parallel-runs<span class="r">|</span>serial<span class="r">]</span></code><a class="copiable-link" href="#index-fprofile-reproducible"> ¶</a></span>
</dt> <dd>
<p>Control level of reproducibility of profile gathered by <code class="code">-fprofile-generate</code>. This makes it possible to rebuild program with same outcome which is useful, for example, for distribution packages. </p> <p>With <samp class="option">-fprofile-reproducible=serial</samp> the profile gathered by <samp class="option">-fprofile-generate</samp> is reproducible provided the trained program behaves the same at each invocation of the train run, it is not multi-threaded and profile data streaming is always done in the same order. Note that profile streaming happens at the end of program run but also before <code class="code">fork</code> function is invoked. </p> <p>Note that it is quite common that execution counts of some part of programs depends, for example, on length of temporary file names or memory space randomization (that may affect hash-table collision rate). Such non-reproducible part of programs may be annotated by <code class="code">no_instrument_function</code> function attribute. <code class="command">gcov-dump</code> with <samp class="option">-l</samp> can be used to dump gathered data and verify that they are indeed reproducible. </p> <p>With <samp class="option">-fprofile-reproducible=parallel-runs</samp> collected profile stays reproducible regardless the order of streaming of the data into gcda files. This setting makes it possible to run multiple instances of instrumented program in parallel (such as with <code class="code">make -j</code>). This reduces quality of gathered data, in particular of indirect call profiling. </p> </dd> <dt>
<span><code class="code">-fsanitize=address</code><a class="copiable-link" href="#index-fsanitize_003daddress"> ¶</a></span>
</dt> <dd>
<p>Enable AddressSanitizer, a fast memory error detector. Memory access instructions are instrumented to detect out-of-bounds and use-after-free bugs. The option enables <samp class="option">-fsanitize-address-use-after-scope</samp>. See <a class="uref" href="https://github.com/google/sanitizers/wiki/AddressSanitizer">https://github.com/google/sanitizers/wiki/AddressSanitizer</a> for more details. The run-time behavior can be influenced using the <code class="env">ASAN_OPTIONS</code> environment variable. When set to <code class="code">help=1</code>, the available options are shown at startup of the instrumented program. See <a class="url" href="https://github.com/google/sanitizers/wiki/AddressSanitizerFlags#run-time-flags">https://github.com/google/sanitizers/wiki/AddressSanitizerFlags#run-time-flags</a> for a list of supported options. The option cannot be combined with <samp class="option">-fsanitize=thread</samp> or <samp class="option">-fsanitize=hwaddress</samp>. Note that the only target <samp class="option">-fsanitize=hwaddress</samp> is currently supported on is AArch64. </p> <p>To get more accurate stack traces, it is possible to use options such as <samp class="option">-O0</samp>, <samp class="option">-O1</samp>, or <samp class="option">-Og</samp> (which, for instance, prevent most function inlining), <samp class="option">-fno-optimize-sibling-calls</samp> (which prevents optimizing sibling and tail recursive calls; this option is implicit for <samp class="option">-O0</samp>, <samp class="option">-O1</samp>, or <samp class="option">-Og</samp>), or <samp class="option">-fno-ipa-icf</samp> (which disables Identical Code Folding for functions). Since multiple runs of the program may yield backtraces with different addresses due to ASLR (Address Space Layout Randomization), it may be desirable to turn ASLR off. On Linux, this can be achieved with ‘<samp class="samp">setarch `uname -m` -R ./prog</samp>’. </p> </dd> <dt>
<span><code class="code">-fsanitize=kernel-address</code><a class="copiable-link" href="#index-fsanitize_003dkernel-address"> ¶</a></span>
</dt> <dd>
<p>Enable AddressSanitizer for Linux kernel. See <a class="uref" href="https://github.com/google/kernel-sanitizers">https://github.com/google/kernel-sanitizers</a> for more details. </p> </dd> <dt>
<span><code class="code">-fsanitize=hwaddress</code><a class="copiable-link" href="#index-fsanitize_003dhwaddress"> ¶</a></span>
</dt> <dd>
<p>Enable Hardware-assisted AddressSanitizer, which uses a hardware ability to ignore the top byte of a pointer to allow the detection of memory errors with a low memory overhead. Memory access instructions are instrumented to detect out-of-bounds and use-after-free bugs. The option enables <samp class="option">-fsanitize-address-use-after-scope</samp>. See <a class="uref" href="https://clang.llvm.org/docs/HardwareAssistedAddressSanitizerDesign.html">https://clang.llvm.org/docs/HardwareAssistedAddressSanitizerDesign.html</a> for more details. The run-time behavior can be influenced using the <code class="env">HWASAN_OPTIONS</code> environment variable. When set to <code class="code">help=1</code>, the available options are shown at startup of the instrumented program. The option cannot be combined with <samp class="option">-fsanitize=thread</samp> or <samp class="option">-fsanitize=address</samp>, and is currently only available on AArch64. </p> </dd> <dt>
<span><code class="code">-fsanitize=kernel-hwaddress</code><a class="copiable-link" href="#index-fsanitize_003dkernel-hwaddress"> ¶</a></span>
</dt> <dd>
<p>Enable Hardware-assisted AddressSanitizer for compilation of the Linux kernel. Similar to <samp class="option">-fsanitize=kernel-address</samp> but using an alternate instrumentation method, and similar to <samp class="option">-fsanitize=hwaddress</samp> but with instrumentation differences necessary for compiling the Linux kernel. These differences are to avoid hwasan library initialization calls and to account for the stack pointer having a different value in its top byte. </p> <p><em class="emph">Note:</em> This option has different defaults to the <samp class="option">-fsanitize=hwaddress</samp>. Instrumenting the stack and alloca calls are not on by default but are still possible by specifying the command-line options <samp class="option">--param hwasan-instrument-stack=1</samp> and <samp class="option">--param hwasan-instrument-allocas=1</samp> respectively. Using a random frame tag is not implemented for kernel instrumentation. </p> </dd> <dt>
<span><code class="code">-fsanitize=pointer-compare</code><a class="copiable-link" href="#index-fsanitize_003dpointer-compare"> ¶</a></span>
</dt> <dd>
<p>Instrument comparison operation (&lt;, &lt;=, &gt;, &gt;=) with pointer operands. The option must be combined with either <samp class="option">-fsanitize=kernel-address</samp> or <samp class="option">-fsanitize=address</samp> The option cannot be combined with <samp class="option">-fsanitize=thread</samp>. Note: By default the check is disabled at run time. To enable it, add <code class="code">detect_invalid_pointer_pairs=2</code> to the environment variable <code class="env">ASAN_OPTIONS</code>. Using <code class="code">detect_invalid_pointer_pairs=1</code> detects invalid operation only when both pointers are non-null. </p> </dd> <dt>
<span><code class="code">-fsanitize=pointer-subtract</code><a class="copiable-link" href="#index-fsanitize_003dpointer-subtract"> ¶</a></span>
</dt> <dd>
<p>Instrument subtraction with pointer operands. The option must be combined with either <samp class="option">-fsanitize=kernel-address</samp> or <samp class="option">-fsanitize=address</samp> The option cannot be combined with <samp class="option">-fsanitize=thread</samp>. Note: By default the check is disabled at run time. To enable it, add <code class="code">detect_invalid_pointer_pairs=2</code> to the environment variable <code class="env">ASAN_OPTIONS</code>. Using <code class="code">detect_invalid_pointer_pairs=1</code> detects invalid operation only when both pointers are non-null. </p> </dd> <dt>
<span><code class="code">-fsanitize=shadow-call-stack</code><a class="copiable-link" href="#index-fsanitize_003dshadow-call-stack"> ¶</a></span>
</dt> <dd>
<p>Enable ShadowCallStack, a security enhancement mechanism used to protect programs against return address overwrites (e.g. stack buffer overflows.) It works by saving a function’s return address to a separately allocated shadow call stack in the function prologue and restoring the return address from the shadow call stack in the function epilogue. Instrumentation only occurs in functions that need to save the return address to the stack. </p> <p>Currently it only supports the aarch64 platform. It is specifically designed for linux kernels that enable the CONFIG_SHADOW_CALL_STACK option. For the user space programs, runtime support is not currently provided in libc and libgcc. Users who want to use this feature in user space need to provide their own support for the runtime. It should be noted that this may cause the ABI rules to be broken. </p> <p>On aarch64, the instrumentation makes use of the platform register <code class="code">x18</code>. This generally means that any code that may run on the same thread as code compiled with ShadowCallStack must be compiled with the flag <samp class="option">-ffixed-x18</samp>, otherwise functions compiled without <samp class="option">-ffixed-x18</samp> might clobber <code class="code">x18</code> and so corrupt the shadow stack pointer. </p> <p>Also, because there is no userspace runtime support, code compiled with ShadowCallStack cannot use exception handling. Use <samp class="option">-fno-exceptions</samp> to turn off exceptions. </p> <p>See <a class="uref" href="https://clang.llvm.org/docs/ShadowCallStack.html">https://clang.llvm.org/docs/ShadowCallStack.html</a> for more details. </p> </dd> <dt>
<span><code class="code">-fsanitize=thread</code><a class="copiable-link" href="#index-fsanitize_003dthread"> ¶</a></span>
</dt> <dd>
<p>Enable ThreadSanitizer, a fast data race detector. Memory access instructions are instrumented to detect data race bugs. See <a class="uref" href="https://github.com/google/sanitizers/wiki#threadsanitizer">https://github.com/google/sanitizers/wiki#threadsanitizer</a> for more details. The run-time behavior can be influenced using the <code class="env">TSAN_OPTIONS</code> environment variable; see <a class="url" href="https://github.com/google/sanitizers/wiki/ThreadSanitizerFlags">https://github.com/google/sanitizers/wiki/ThreadSanitizerFlags</a> for a list of supported options. The option cannot be combined with <samp class="option">-fsanitize=address</samp>, <samp class="option">-fsanitize=leak</samp>. </p> <p>Note that sanitized atomic builtins cannot throw exceptions when operating on invalid memory addresses with non-call exceptions (<samp class="option">-fnon-call-exceptions</samp>). </p> </dd> <dt>
<span><code class="code">-fsanitize=leak</code><a class="copiable-link" href="#index-fsanitize_003dleak"> ¶</a></span>
</dt> <dd>
<p>Enable LeakSanitizer, a memory leak detector. This option only matters for linking of executables. The executable is linked against a library that overrides <code class="code">malloc</code> and other allocator functions. See <a class="uref" href="https://github.com/google/sanitizers/wiki/AddressSanitizerLeakSanitizer">https://github.com/google/sanitizers/wiki/AddressSanitizerLeakSanitizer</a> for more details. The run-time behavior can be influenced using the <code class="env">LSAN_OPTIONS</code> environment variable. The option cannot be combined with <samp class="option">-fsanitize=thread</samp>. </p> </dd> <dt>
<span><code class="code">-fsanitize=undefined</code><a class="copiable-link" href="#index-fsanitize_003dundefined"> ¶</a></span>
</dt> <dd>
<p>Enable UndefinedBehaviorSanitizer, a fast undefined behavior detector. Various computations are instrumented to detect undefined behavior at runtime. See <a class="uref" href="https://clang.llvm.org/docs/UndefinedBehaviorSanitizer.html">https://clang.llvm.org/docs/UndefinedBehaviorSanitizer.html</a> for more details. The run-time behavior can be influenced using the <code class="env">UBSAN_OPTIONS</code> environment variable. Current suboptions are: </p> <dl class="table"> <dt>
<span><code class="code">-fsanitize=shift</code><a class="copiable-link" href="#index-fsanitize_003dshift"> ¶</a></span>
</dt> <dd>
<p>This option enables checking that the result of a shift operation is not undefined. Note that what exactly is considered undefined differs slightly between C and C++, as well as between ISO C90 and C99, etc. This option has two suboptions, <samp class="option">-fsanitize=shift-base</samp> and <samp class="option">-fsanitize=shift-exponent</samp>. </p> </dd> <dt>
<span><code class="code">-fsanitize=shift-exponent</code><a class="copiable-link" href="#index-fsanitize_003dshift-exponent"> ¶</a></span>
</dt> <dd>
<p>This option enables checking that the second argument of a shift operation is not negative and is smaller than the precision of the promoted first argument. </p> </dd> <dt>
<span><code class="code">-fsanitize=shift-base</code><a class="copiable-link" href="#index-fsanitize_003dshift-base"> ¶</a></span>
</dt> <dd>
<p>If the second argument of a shift operation is within range, check that the result of a shift operation is not undefined. Note that what exactly is considered undefined differs slightly between C and C++, as well as between ISO C90 and C99, etc. </p> </dd> <dt>
<span><code class="code">-fsanitize=integer-divide-by-zero</code><a class="copiable-link" href="#index-fsanitize_003dinteger-divide-by-zero"> ¶</a></span>
</dt> <dd>
<p>Detect integer division by zero. </p> </dd> <dt>
<span><code class="code">-fsanitize=unreachable</code><a class="copiable-link" href="#index-fsanitize_003dunreachable"> ¶</a></span>
</dt> <dd>
<p>With this option, the compiler turns the <code class="code">__builtin_unreachable</code> call into a diagnostics message call instead. When reaching the <code class="code">__builtin_unreachable</code> call, the behavior is undefined. </p> </dd> <dt>
<span><code class="code">-fsanitize=vla-bound</code><a class="copiable-link" href="#index-fsanitize_003dvla-bound"> ¶</a></span>
</dt> <dd>
<p>This option instructs the compiler to check that the size of a variable length array is positive. </p> </dd> <dt>
<span><code class="code">-fsanitize=null</code><a class="copiable-link" href="#index-fsanitize_003dnull"> ¶</a></span>
</dt> <dd>
<p>This option enables pointer checking. Particularly, the application built with this option turned on will issue an error message when it tries to dereference a NULL pointer, or if a reference (possibly an rvalue reference) is bound to a NULL pointer, or if a method is invoked on an object pointed by a NULL pointer. </p> </dd> <dt>
<span><code class="code">-fsanitize=return</code><a class="copiable-link" href="#index-fsanitize_003dreturn"> ¶</a></span>
</dt> <dd>
<p>This option enables return statement checking. Programs built with this option turned on will issue an error message when the end of a non-void function is reached without actually returning a value. This option works in C++ only. </p> </dd> <dt>
<span><code class="code">-fsanitize=signed-integer-overflow</code><a class="copiable-link" href="#index-fsanitize_003dsigned-integer-overflow"> ¶</a></span>
</dt> <dd>
<p>This option enables signed integer overflow checking. We check that the result of <code class="code">+</code>, <code class="code">*</code>, and both unary and binary <code class="code">-</code> does not overflow in the signed arithmetics. This also detects <code class="code">INT_MIN / -1</code> signed division. Note, integer promotion rules must be taken into account. That is, the following is not an overflow: </p>
<div class="example smallexample"> <pre class="example-preformatted" data-language="cpp">signed char a = SCHAR_MAX;
a++;</pre>
</div> </dd> <dt>
<span><code class="code">-fsanitize=bounds</code><a class="copiable-link" href="#index-fsanitize_003dbounds"> ¶</a></span>
</dt> <dd>
<p>This option enables instrumentation of array bounds. Various out of bounds accesses are detected. Flexible array members, flexible array member-like arrays, and initializers of variables with static storage are not instrumented, with the exception of flexible array member-like arrays for which <code class="code">-fstrict-flex-arrays</code> or <code class="code">-fstrict-flex-arrays=</code> options or <code class="code">strict_flex_array</code> attributes say they shouldn’t be treated like flexible array member-like arrays. </p> </dd> <dt>
<span><code class="code">-fsanitize=bounds-strict</code><a class="copiable-link" href="#index-fsanitize_003dbounds-strict"> ¶</a></span>
</dt> <dd>
<p>This option enables strict instrumentation of array bounds. Most out of bounds accesses are detected, including flexible array member-like arrays. Initializers of variables with static storage are not instrumented. </p> </dd> <dt>
<span><code class="code">-fsanitize=alignment</code><a class="copiable-link" href="#index-fsanitize_003dalignment"> ¶</a></span>
</dt> <dd> <p>This option enables checking of alignment of pointers when they are dereferenced, or when a reference is bound to insufficiently aligned target, or when a method or constructor is invoked on insufficiently aligned object. </p> </dd> <dt>
<span><code class="code">-fsanitize=object-size</code><a class="copiable-link" href="#index-fsanitize_003dobject-size"> ¶</a></span>
</dt> <dd>
<p>This option enables instrumentation of memory references using the <code class="code">__builtin_dynamic_object_size</code> function. Various out of bounds pointer accesses are detected. </p> </dd> <dt>
<span><code class="code">-fsanitize=float-divide-by-zero</code><a class="copiable-link" href="#index-fsanitize_003dfloat-divide-by-zero"> ¶</a></span>
</dt> <dd>
<p>Detect floating-point division by zero. Unlike other similar options, <samp class="option">-fsanitize=float-divide-by-zero</samp> is not enabled by <samp class="option">-fsanitize=undefined</samp>, since floating-point division by zero can be a legitimate way of obtaining infinities and NaNs. </p> </dd> <dt>
<span><code class="code">-fsanitize=float-cast-overflow</code><a class="copiable-link" href="#index-fsanitize_003dfloat-cast-overflow"> ¶</a></span>
</dt> <dd>
<p>This option enables floating-point type to integer conversion checking. We check that the result of the conversion does not overflow. Unlike other similar options, <samp class="option">-fsanitize=float-cast-overflow</samp> is not enabled by <samp class="option">-fsanitize=undefined</samp>. This option does not work well with <code class="code">FE_INVALID</code> exceptions enabled. </p> </dd> <dt>
<span><code class="code">-fsanitize=nonnull-attribute</code><a class="copiable-link" href="#index-fsanitize_003dnonnull-attribute"> ¶</a></span>
</dt> <dd> <p>This option enables instrumentation of calls, checking whether null values are not passed to arguments marked as requiring a non-null value by the <code class="code">nonnull</code> function attribute. </p> </dd> <dt>
<span><code class="code">-fsanitize=returns-nonnull-attribute</code><a class="copiable-link" href="#index-fsanitize_003dreturns-nonnull-attribute"> ¶</a></span>
</dt> <dd> <p>This option enables instrumentation of return statements in functions marked with <code class="code">returns_nonnull</code> function attribute, to detect returning of null values from such functions. </p> </dd> <dt>
<span><code class="code">-fsanitize=bool</code><a class="copiable-link" href="#index-fsanitize_003dbool"> ¶</a></span>
</dt> <dd> <p>This option enables instrumentation of loads from bool. If a value other than 0/1 is loaded, a run-time error is issued. </p> </dd> <dt>
<span><code class="code">-fsanitize=enum</code><a class="copiable-link" href="#index-fsanitize_003denum"> ¶</a></span>
</dt> <dd> <p>This option enables instrumentation of loads from an enum type. If a value outside the range of values for the enum type is loaded, a run-time error is issued. </p> </dd> <dt>
<span><code class="code">-fsanitize=vptr</code><a class="copiable-link" href="#index-fsanitize_003dvptr"> ¶</a></span>
</dt> <dd> <p>This option enables instrumentation of C++ member function calls, member accesses and some conversions between pointers to base and derived classes, to verify the referenced object has the correct dynamic type. </p> </dd> <dt>
<span><code class="code">-fsanitize=pointer-overflow</code><a class="copiable-link" href="#index-fsanitize_003dpointer-overflow"> ¶</a></span>
</dt> <dd> <p>This option enables instrumentation of pointer arithmetics. If the pointer arithmetics overflows, a run-time error is issued. </p> </dd> <dt>
<span><code class="code">-fsanitize=builtin</code><a class="copiable-link" href="#index-fsanitize_003dbuiltin"> ¶</a></span>
</dt> <dd> <p>This option enables instrumentation of arguments to selected builtin functions. If an invalid value is passed to such arguments, a run-time error is issued. E.g. passing 0 as the argument to <code class="code">__builtin_ctz</code> or <code class="code">__builtin_clz</code> invokes undefined behavior and is diagnosed by this option. </p> </dd> </dl> <p>Note that sanitizers tend to increase the rate of false positive warnings, most notably those around <samp class="option">-Wmaybe-uninitialized</samp>. We recommend against combining <samp class="option">-Werror</samp> and [the use of] sanitizers. </p> <p>While <samp class="option">-ftrapv</samp> causes traps for signed overflows to be emitted, <samp class="option">-fsanitize=undefined</samp> gives a diagnostic message. This currently works only for the C family of languages. </p> </dd> <dt>
<span><code class="code">-fno-sanitize=all</code><a class="copiable-link" href="#index-fno-sanitize_003dall"> ¶</a></span>
</dt> <dd> <p>This option disables all previously enabled sanitizers. <samp class="option">-fsanitize=all</samp> is not allowed, as some sanitizers cannot be used together. </p> </dd> <dt>
<span><code class="code">-fasan-shadow-offset=<var class="var">number</var></code><a class="copiable-link" href="#index-fasan-shadow-offset"> ¶</a></span>
</dt> <dd>
<p>This option forces GCC to use custom shadow offset in AddressSanitizer checks. It is useful for experimenting with different shadow memory layouts in Kernel AddressSanitizer. </p> </dd> <dt>
<span><code class="code">-fsanitize-sections=<var class="var">s1</var>,<var class="var">s2</var>,...</code><a class="copiable-link" href="#index-fsanitize-sections"> ¶</a></span>
</dt> <dd>
<p>Sanitize global variables in selected user-defined sections. <var class="var">si</var> may contain wildcards. </p> </dd> <dt>
 <span><code class="code">-fsanitize-recover<span class="r">[</span>=<var class="var">opts</var><span class="r">]</span></code><a class="copiable-link" href="#index-fsanitize-recover"> ¶</a></span>
</dt> <dd>
<p><samp class="option">-fsanitize-recover=</samp> controls error recovery mode for sanitizers mentioned in comma-separated list of <var class="var">opts</var>. Enabling this option for a sanitizer component causes it to attempt to continue running the program as if no error happened. This means multiple runtime errors can be reported in a single program run, and the exit code of the program may indicate success even when errors have been reported. The <samp class="option">-fno-sanitize-recover=</samp> option can be used to alter this behavior: only the first detected error is reported and program then exits with a non-zero exit code. </p> <p>Currently this feature only works for <samp class="option">-fsanitize=undefined</samp> (and its suboptions except for <samp class="option">-fsanitize=unreachable</samp> and <samp class="option">-fsanitize=return</samp>), <samp class="option">-fsanitize=float-cast-overflow</samp>, <samp class="option">-fsanitize=float-divide-by-zero</samp>, <samp class="option">-fsanitize=bounds-strict</samp>, <samp class="option">-fsanitize=kernel-address</samp> and <samp class="option">-fsanitize=address</samp>. For these sanitizers error recovery is turned on by default, except <samp class="option">-fsanitize=address</samp>, for which this feature is experimental. <samp class="option">-fsanitize-recover=all</samp> and <samp class="option">-fno-sanitize-recover=all</samp> is also accepted, the former enables recovery for all sanitizers that support it, the latter disables recovery for all sanitizers that support it. </p> <p>Even if a recovery mode is turned on the compiler side, it needs to be also enabled on the runtime library side, otherwise the failures are still fatal. The runtime library defaults to <code class="code">halt_on_error=0</code> for ThreadSanitizer and UndefinedBehaviorSanitizer, while default value for AddressSanitizer is <code class="code">halt_on_error=1</code>. This can be overridden through setting the <code class="code">halt_on_error</code> flag in the corresponding environment variable. </p> <p>Syntax without an explicit <var class="var">opts</var> parameter is deprecated. It is equivalent to specifying an <var class="var">opts</var> list of: </p> <div class="example smallexample"> <pre class="example-preformatted" data-language="cpp">undefined,float-cast-overflow,float-divide-by-zero,bounds-strict</pre>
</div> </dd> <dt>
<span><code class="code">-fsanitize-address-use-after-scope</code><a class="copiable-link" href="#index-fsanitize-address-use-after-scope"> ¶</a></span>
</dt> <dd>
<p>Enable sanitization of local variables to detect use-after-scope bugs. The option sets <samp class="option">-fstack-reuse</samp> to ‘<samp class="samp">none</samp>’. </p> </dd> <dt>
 <span><code class="code">-fsanitize-trap<span class="r">[</span>=<var class="var">opts</var><span class="r">]</span></code><a class="copiable-link" href="#index-fsanitize-trap"> ¶</a></span>
</dt> <dd>
<p>The <samp class="option">-fsanitize-trap=</samp> option instructs the compiler to report for sanitizers mentioned in comma-separated list of <var class="var">opts</var> undefined behavior using <code class="code">__builtin_trap</code> rather than a <code class="code">libubsan</code> library routine. If this option is enabled for certain sanitizer, it takes precedence over the <samp class="option">-fsanitizer-recover=</samp> for that sanitizer, <code class="code">__builtin_trap</code> will be emitted and be fatal regardless of whether recovery is enabled or disabled using <samp class="option">-fsanitize-recover=</samp>. </p> <p>The advantage of this is that the <code class="code">libubsan</code> library is not needed and is not linked in, so this is usable even in freestanding environments. </p> <p>Currently this feature works with <samp class="option">-fsanitize=undefined</samp> (and its suboptions except for <samp class="option">-fsanitize=vptr</samp>), <samp class="option">-fsanitize=float-cast-overflow</samp>, <samp class="option">-fsanitize=float-divide-by-zero</samp> and <samp class="option">-fsanitize=bounds-strict</samp>. <code class="code">-fsanitize-trap=all</code> can be also specified, which enables it for <code class="code">undefined</code> suboptions, <samp class="option">-fsanitize=float-cast-overflow</samp>, <samp class="option">-fsanitize=float-divide-by-zero</samp> and <samp class="option">-fsanitize=bounds-strict</samp>. If <code class="code">-fsanitize-trap=undefined</code> or <code class="code">-fsanitize-trap=all</code> is used and <code class="code">-fsanitize=vptr</code> is enabled on the command line, the instrumentation is silently ignored as the instrumentation always needs <code class="code">libubsan</code> support, <samp class="option">-fsanitize-trap=vptr</samp> is not allowed. </p> </dd> <dt>
<span><code class="code">-fsanitize-undefined-trap-on-error</code><a class="copiable-link" href="#index-fsanitize-undefined-trap-on-error"> ¶</a></span>
</dt> <dd>
<p>The <samp class="option">-fsanitize-undefined-trap-on-error</samp> option is deprecated equivalent of <samp class="option">-fsanitize-trap=all</samp>. </p> </dd> <dt>
<span><code class="code">-fsanitize-coverage=trace-pc</code><a class="copiable-link" href="#index-fsanitize-coverage_003dtrace-pc"> ¶</a></span>
</dt> <dd>
<p>Enable coverage-guided fuzzing code instrumentation. Inserts a call to <code class="code">__sanitizer_cov_trace_pc</code> into every basic block. </p> </dd> <dt>
<span><code class="code">-fsanitize-coverage=trace-cmp</code><a class="copiable-link" href="#index-fsanitize-coverage_003dtrace-cmp"> ¶</a></span>
</dt> <dd>
<p>Enable dataflow guided fuzzing code instrumentation. Inserts a call to <code class="code">__sanitizer_cov_trace_cmp1</code>, <code class="code">__sanitizer_cov_trace_cmp2</code>, <code class="code">__sanitizer_cov_trace_cmp4</code> or <code class="code">__sanitizer_cov_trace_cmp8</code> for integral comparison with both operands variable or <code class="code">__sanitizer_cov_trace_const_cmp1</code>, <code class="code">__sanitizer_cov_trace_const_cmp2</code>, <code class="code">__sanitizer_cov_trace_const_cmp4</code> or <code class="code">__sanitizer_cov_trace_const_cmp8</code> for integral comparison with one operand constant, <code class="code">__sanitizer_cov_trace_cmpf</code> or <code class="code">__sanitizer_cov_trace_cmpd</code> for float or double comparisons and <code class="code">__sanitizer_cov_trace_switch</code> for switch statements. </p> </dd> <dt>
<span><code class="code">-fcf-protection=<span class="r">[</span>full<span class="r">|</span>branch<span class="r">|</span>return<span class="r">|</span>none<span class="r">|</span>check<span class="r">]</span></code><a class="copiable-link" href="#index-fcf-protection"> ¶</a></span>
</dt> <dd>
<p>Enable code instrumentation of control-flow transfers to increase program security by checking that target addresses of control-flow transfer instructions (such as indirect function call, function return, indirect jump) are valid. This prevents diverting the flow of control to an unexpected target. This is intended to protect against such threats as Return-oriented Programming (ROP), and similarly call/jmp-oriented programming (COP/JOP). </p> <p>The value <code class="code">branch</code> tells the compiler to implement checking of validity of control-flow transfer at the point of indirect branch instructions, i.e. call/jmp instructions. The value <code class="code">return</code> implements checking of validity at the point of returning from a function. The value <code class="code">full</code> is an alias for specifying both <code class="code">branch</code> and <code class="code">return</code>. The value <code class="code">none</code> turns off instrumentation. </p> <p>The value <code class="code">check</code> is used for the final link with link-time optimization (LTO). An error is issued if LTO object files are compiled with different <samp class="option">-fcf-protection</samp> values. The value <code class="code">check</code> is ignored at the compile time. </p> <p>The macro <code class="code">__CET__</code> is defined when <samp class="option">-fcf-protection</samp> is used. The first bit of <code class="code">__CET__</code> is set to 1 for the value <code class="code">branch</code> and the second bit of <code class="code">__CET__</code> is set to 1 for the <code class="code">return</code>. </p> <p>You can also use the <code class="code">nocf_check</code> attribute to identify which functions and calls should be skipped from instrumentation (see <a class="pxref" href="function-attributes">Declaring Attributes of Functions</a>). </p> <p>Currently the x86 GNU/Linux target provides an implementation based on Intel Control-flow Enforcement Technology (CET) which works for i686 processor or newer. </p> </dd> <dt>
<span><code class="code">-fharden-compares</code><a class="copiable-link" href="#index-fharden-compares"> ¶</a></span>
</dt> <dd>
<p>For every logical test that survives gimple optimizations and is <em class="emph">not</em> the condition in a conditional branch (for example, conditions tested for conditional moves, or to store in boolean variables), emit extra code to compute and verify the reversed condition, and to call <code class="code">__builtin_trap</code> if the results do not match. Use with ‘<samp class="samp">-fharden-conditional-branches</samp>’ to cover all conditionals. </p> </dd> <dt>
<span><code class="code">-fharden-conditional-branches</code><a class="copiable-link" href="#index-fharden-conditional-branches"> ¶</a></span>
</dt> <dd>
<p>For every non-vectorized conditional branch that survives gimple optimizations, emit extra code to compute and verify the reversed condition, and to call <code class="code">__builtin_trap</code> if the result is unexpected. Use with ‘<samp class="samp">-fharden-compares</samp>’ to cover all conditionals. </p> </dd> <dt>
<span><code class="code">-fstack-protector</code><a class="copiable-link" href="#index-fstack-protector"> ¶</a></span>
</dt> <dd>
<p>Emit extra code to check for buffer overflows, such as stack smashing attacks. This is done by adding a guard variable to functions with vulnerable objects. This includes functions that call <code class="code">alloca</code>, and functions with buffers larger than or equal to 8 bytes. The guards are initialized when a function is entered and then checked when the function exits. If a guard check fails, an error message is printed and the program exits. Only variables that are actually allocated on the stack are considered, optimized away variables or variables allocated in registers don’t count. </p> </dd> <dt>
<span><code class="code">-fstack-protector-all</code><a class="copiable-link" href="#index-fstack-protector-all"> ¶</a></span>
</dt> <dd>
<p>Like <samp class="option">-fstack-protector</samp> except that all functions are protected. </p> </dd> <dt>
<span><code class="code">-fstack-protector-strong</code><a class="copiable-link" href="#index-fstack-protector-strong"> ¶</a></span>
</dt> <dd>
<p>Like <samp class="option">-fstack-protector</samp> but includes additional functions to be protected — those that have local array definitions, or have references to local frame addresses. Only variables that are actually allocated on the stack are considered, optimized away variables or variables allocated in registers don’t count. </p> </dd> <dt>
<span><code class="code">-fstack-protector-explicit</code><a class="copiable-link" href="#index-fstack-protector-explicit"> ¶</a></span>
</dt> <dd>
<p>Like <samp class="option">-fstack-protector</samp> but only protects those functions which have the <code class="code">stack_protect</code> attribute. </p> </dd> <dt>
<span><code class="code">-fstack-check</code><a class="copiable-link" href="#index-fstack-check"> ¶</a></span>
</dt> <dd>
<p>Generate code to verify that you do not go beyond the boundary of the stack. You should specify this flag if you are running in an environment with multiple threads, but you only rarely need to specify it in a single-threaded environment since stack overflow is automatically detected on nearly all systems if there is only one stack. </p> <p>Note that this switch does not actually cause checking to be done; the operating system or the language runtime must do that. The switch causes generation of code to ensure that they see the stack being extended. </p> <p>You can additionally specify a string parameter: ‘<samp class="samp">no</samp>’ means no checking, ‘<samp class="samp">generic</samp>’ means force the use of old-style checking, ‘<samp class="samp">specific</samp>’ means use the best checking method and is equivalent to bare <samp class="option">-fstack-check</samp>. </p> <p>Old-style checking is a generic mechanism that requires no specific target support in the compiler but comes with the following drawbacks: </p> <ol class="enumerate"> <li> Modified allocation strategy for large objects: they are always allocated dynamically if their size exceeds a fixed threshold. Note this may change the semantics of some code. </li>
<li> Fixed limit on the size of the static frame of functions: when it is topped by a particular function, stack checking is not reliable and a warning is issued by the compiler. </li>
<li> Inefficiency: because of both the modified allocation strategy and the generic implementation, code performance is hampered. </li>
</ol> <p>Note that old-style stack checking is also the fallback method for ‘<samp class="samp">specific</samp>’ if no target support has been added in the compiler. </p> <p>‘<samp class="samp">-fstack-check=</samp>’ is designed for Ada’s needs to detect infinite recursion and stack overflows. ‘<samp class="samp">specific</samp>’ is an excellent choice when compiling Ada code. It is not generally sufficient to protect against stack-clash attacks. To protect against those you want ‘<samp class="samp">-fstack-clash-protection</samp>’. </p> </dd> <dt>
<span><code class="code">-fstack-clash-protection</code><a class="copiable-link" href="#index-fstack-clash-protection"> ¶</a></span>
</dt> <dd>
<p>Generate code to prevent stack clash style attacks. When this option is enabled, the compiler will only allocate one page of stack space at a time and each page is accessed immediately after allocation. Thus, it prevents allocations from jumping over any stack guard page provided by the operating system. </p> <p>Most targets do not fully support stack clash protection. However, on those targets <samp class="option">-fstack-clash-protection</samp> will protect dynamic stack allocations. <samp class="option">-fstack-clash-protection</samp> may also provide limited protection for static stack allocations if the target supports <samp class="option">-fstack-check=specific</samp>. </p> </dd> <dt>
  <span><code class="code">-fstack-limit-register=<var class="var">reg</var></code><a class="copiable-link" href="#index-fstack-limit-register"> ¶</a></span>
</dt> <dt><code class="code">-fstack-limit-symbol=<var class="var">sym</var></code></dt> <dt><code class="code">-fno-stack-limit</code></dt> <dd>
<p>Generate code to ensure that the stack does not grow beyond a certain value, either the value of a register or the address of a symbol. If a larger stack is required, a signal is raised at run time. For most targets, the signal is raised before the stack overruns the boundary, so it is possible to catch the signal without taking special precautions. </p> <p>For instance, if the stack starts at absolute address ‘<samp class="samp">0x80000000</samp>’ and grows downwards, you can use the flags <samp class="option">-fstack-limit-symbol=__stack_limit</samp> and <samp class="option">-Wl,--defsym,__stack_limit=0x7ffe0000</samp> to enforce a stack limit of 128KB. Note that this may only work with the GNU linker. </p> <p>You can locally override stack limit checking by using the <code class="code">no_stack_limit</code> function attribute (see <a class="pxref" href="function-attributes">Declaring Attributes of Functions</a>). </p> </dd> <dt>
<span><code class="code">-fsplit-stack</code><a class="copiable-link" href="#index-fsplit-stack"> ¶</a></span>
</dt> <dd>
<p>Generate code to automatically split the stack before it overflows. The resulting program has a discontiguous stack which can only overflow if the program is unable to allocate any more memory. This is most useful when running threaded programs, as it is no longer necessary to calculate a good stack size to use for each thread. This is currently only implemented for the x86 targets running GNU/Linux. </p> <p>When code compiled with <samp class="option">-fsplit-stack</samp> calls code compiled without <samp class="option">-fsplit-stack</samp>, there may not be much stack space available for the latter code to run. If compiling all code, including library code, with <samp class="option">-fsplit-stack</samp> is not an option, then the linker can fix up these calls so that the code compiled without <samp class="option">-fsplit-stack</samp> always has a large stack. Support for this is implemented in the gold linker in GNU binutils release 2.21 and later. </p> </dd> <dt>
<span><code class="code">-fvtable-verify=<span class="r">[</span>std<span class="r">|</span>preinit<span class="r">|</span>none<span class="r">]</span></code><a class="copiable-link" href="#index-fvtable-verify"> ¶</a></span>
</dt> <dd>
<p>This option is only available when compiling C++ code. It turns on (or off, if using <samp class="option">-fvtable-verify=none</samp>) the security feature that verifies at run time, for every virtual call, that the vtable pointer through which the call is made is valid for the type of the object, and has not been corrupted or overwritten. If an invalid vtable pointer is detected at run time, an error is reported and execution of the program is immediately halted. </p> <p>This option causes run-time data structures to be built at program startup, which are used for verifying the vtable pointers. The options ‘<samp class="samp">std</samp>’ and ‘<samp class="samp">preinit</samp>’ control the timing of when these data structures are built. In both cases the data structures are built before execution reaches <code class="code">main</code>. Using <samp class="option">-fvtable-verify=std</samp> causes the data structures to be built after shared libraries have been loaded and initialized. <samp class="option">-fvtable-verify=preinit</samp> causes them to be built before shared libraries have been loaded and initialized. </p> <p>If this option appears multiple times in the command line with different values specified, ‘<samp class="samp">none</samp>’ takes highest priority over both ‘<samp class="samp">std</samp>’ and ‘<samp class="samp">preinit</samp>’; ‘<samp class="samp">preinit</samp>’ takes priority over ‘<samp class="samp">std</samp>’. </p> </dd> <dt>
<span><code class="code">-fvtv-debug</code><a class="copiable-link" href="#index-fvtv-debug"> ¶</a></span>
</dt> <dd>
<p>When used in conjunction with <samp class="option">-fvtable-verify=std</samp> or <samp class="option">-fvtable-verify=preinit</samp>, causes debug versions of the runtime functions for the vtable verification feature to be called. This flag also causes the compiler to log information about which vtable pointers it finds for each class. This information is written to a file named <samp class="file">vtv_set_ptr_data.log</samp> in the directory named by the environment variable <code class="env">VTV_LOGS_DIR</code> if that is defined or the current working directory otherwise. </p> <p>Note: This feature <em class="emph">appends</em> data to the log file. If you want a fresh log file, be sure to delete any existing one. </p> </dd> <dt>
<span><code class="code">-fvtv-counts</code><a class="copiable-link" href="#index-fvtv-counts"> ¶</a></span>
</dt> <dd>
<p>This is a debugging flag. When used in conjunction with <samp class="option">-fvtable-verify=std</samp> or <samp class="option">-fvtable-verify=preinit</samp>, this causes the compiler to keep track of the total number of virtual calls it encounters and the number of verifications it inserts. It also counts the number of calls to certain run-time library functions that it inserts and logs this information for each compilation unit. The compiler writes this information to a file named <samp class="file">vtv_count_data.log</samp> in the directory named by the environment variable <code class="env">VTV_LOGS_DIR</code> if that is defined or the current working directory otherwise. It also counts the size of the vtable pointer sets for each class, and writes this information to <samp class="file">vtv_class_set_sizes.log</samp> in the same directory. </p> <p>Note: This feature <em class="emph">appends</em> data to the log files. To get fresh log files, be sure to delete any existing ones. </p> </dd> <dt>
<span><code class="code">-finstrument-functions</code><a class="copiable-link" href="#index-finstrument-functions"> ¶</a></span>
</dt> <dd>
<p>Generate instrumentation calls for entry and exit to functions. Just after function entry and just before function exit, the following profiling functions are called with the address of the current function and its call site. (On some platforms, <code class="code">__builtin_return_address</code> does not work beyond the current function, so the call site information may not be available to the profiling functions otherwise.) </p> <div class="example smallexample"> <pre class="example-preformatted" data-language="cpp">void __cyg_profile_func_enter (void *this_fn,
                               void *call_site);
void __cyg_profile_func_exit  (void *this_fn,
                               void *call_site);</pre>
</div> <p>The first argument is the address of the start of the current function, which may be looked up exactly in the symbol table. </p> <p>This instrumentation is also done for functions expanded inline in other functions. The profiling calls indicate where, conceptually, the inline function is entered and exited. This means that addressable versions of such functions must be available. If all your uses of a function are expanded inline, this may mean an additional expansion of code size. If you use <code class="code">extern inline</code> in your C code, an addressable version of such functions must be provided. (This is normally the case anyway, but if you get lucky and the optimizer always expands the functions inline, you might have gotten away without providing static copies.) </p> <p>A function may be given the attribute <code class="code">no_instrument_function</code>, in which case this instrumentation is not done. This can be used, for example, for the profiling functions listed above, high-priority interrupt routines, and any functions from which the profiling functions cannot safely be called (perhaps signal handlers, if the profiling routines generate output or allocate memory). See <a class="xref" href="common-function-attributes">Common Function Attributes</a>. </p> </dd> <dt>
<span><code class="code">-finstrument-functions-once</code><a class="copiable-link" href="#index-finstrument-functions-once"> ¶</a></span>
</dt> <dd>
<p>This is similar to <samp class="option">-finstrument-functions</samp>, but the profiling functions are called only once per instrumented function, i.e. the first profiling function is called after the first entry into the instrumented function and the second profiling function is called before the exit corresponding to this first entry. </p> <p>The definition of <code class="code">once</code> for the purpose of this option is a little vague because the implementation is not protected against data races. As a result, the implementation only guarantees that the profiling functions are called at <em class="emph">least</em> once per process and at <em class="emph">most</em> once per thread, but the calls are always paired, that is to say, if a thread calls the first function, then it will call the second function, unless it never reaches the exit of the instrumented function. </p> </dd> <dt>
<span><code class="code">-finstrument-functions-exclude-file-list=<var class="var">file</var>,<var class="var">file</var>,…</code><a class="copiable-link" href="#index-finstrument-functions-exclude-file-list"> ¶</a></span>
</dt> <dd> <p>Set the list of functions that are excluded from instrumentation (see the description of <samp class="option">-finstrument-functions</samp>). If the file that contains a function definition matches with one of <var class="var">file</var>, then that function is not instrumented. The match is done on substrings: if the <var class="var">file</var> parameter is a substring of the file name, it is considered to be a match. </p> <p>For example: </p> <div class="example smallexample"> <pre class="example-preformatted" data-language="cpp">-finstrument-functions-exclude-file-list=/bits/stl,include/sys</pre>
</div> <p>excludes any inline function defined in files whose pathnames contain <samp class="file">/bits/stl</samp> or <samp class="file">include/sys</samp>. </p> <p>If, for some reason, you want to include letter ‘<samp class="samp">,</samp>’ in one of <var class="var">sym</var>, write ‘<samp class="samp">\,</samp>’. For example, <samp class="option">-finstrument-functions-exclude-file-list='\,\,tmp'</samp> (note the single quote surrounding the option). </p> </dd> <dt>
<span><code class="code">-finstrument-functions-exclude-function-list=<var class="var">sym</var>,<var class="var">sym</var>,…</code><a class="copiable-link" href="#index-finstrument-functions-exclude-function-list"> ¶</a></span>
</dt> <dd> <p>This is similar to <samp class="option">-finstrument-functions-exclude-file-list</samp>, but this option sets the list of function names to be excluded from instrumentation. The function name to be matched is its user-visible name, such as <code class="code">vector&lt;int&gt; blah(const vector&lt;int&gt; &amp;)</code>, not the internal mangled name (e.g., <code class="code">_Z4blahRSt6vectorIiSaIiEE</code>). The match is done on substrings: if the <var class="var">sym</var> parameter is a substring of the function name, it is considered to be a match. For C99 and C++ extended identifiers, the function name must be given in UTF-8, not using universal character names. </p> </dd> <dt>
<span><code class="code">-fpatchable-function-entry=<var class="var">N</var>[,<var class="var">M</var>]</code><a class="copiable-link" href="#index-fpatchable-function-entry"> ¶</a></span>
</dt> <dd>
<p>Generate <var class="var">N</var> NOPs right at the beginning of each function, with the function entry point before the <var class="var">M</var>th NOP. If <var class="var">M</var> is omitted, it defaults to <code class="code">0</code> so the function entry points to the address just at the first NOP. The NOP instructions reserve extra space which can be used to patch in any desired instrumentation at run time, provided that the code segment is writable. The amount of space is controllable indirectly via the number of NOPs; the NOP instruction used corresponds to the instruction emitted by the internal GCC back-end interface <code class="code">gen_nop</code>. This behavior is target-specific and may also depend on the architecture variant and/or other compilation options. </p> <p>For run-time identification, the starting addresses of these areas, which correspond to their respective function entries minus <var class="var">M</var>, are additionally collected in the <code class="code">__patchable_function_entries</code> section of the resulting binary. </p> <p>Note that the value of <code class="code">__attribute__ ((patchable_function_entry
(N,M)))</code> takes precedence over command-line option <samp class="option">-fpatchable-function-entry=N,M</samp>. This can be used to increase the area size or to remove it completely on a single function. If <code class="code">N=0</code>, no pad location is recorded. </p> <p>The NOP instructions are inserted at—and maybe before, depending on <var class="var">M</var>—the function entry address, even before the prologue. On PowerPC with the ELFv2 ABI, for a function with dual entry points, the local entry point is this function entry address. </p> <p>The maximum value of <var class="var">N</var> and <var class="var">M</var> is 65535. On PowerPC with the ELFv2 ABI, for a function with dual entry points, the supported values for <var class="var">M</var> are 0, 2, 6 and 14. </p>
</dd> </dl> </div>  <div class="nav-panel"> <p> Next: <a href="preprocessor-options">Options Controlling the Preprocessor</a>, Previous: <a href="optimize-options">Options That Control Optimization</a>, Up: <a href="invoking-gcc">GCC Command Options</a> [<a href="index#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="indices" title="Index" rel="index">Index</a>]</p> </div><div class="_attribution">
  <p class="_attribution-p">
    &copy; Free Software Foundation<br>Licensed under the GNU Free Documentation License, Version 1.3.<br>
    <a href="https://gcc.gnu.org/onlinedocs/gcc-13.1.0/gcc/Instrumentation-Options.html" class="_attribution-link">https://gcc.gnu.org/onlinedocs/gcc-13.1.0/gcc/Instrumentation-Options.html</a>
  </p>
</div>