path: root/devdocs/gcc~13/extended-asm.html

<div class="subsection-level-extent" id="Extended-Asm"> <div class="nav-panel"> <p> Next: <a href="constraints" accesskey="n" rel="next">Constraints for <code class="code">asm</code> Operands</a>, Previous: <a href="basic-asm" accesskey="p" rel="prev">Basic Asm — Assembler Instructions Without Operands</a>, Up: <a href="using-assembly-language-with-c" accesskey="u" rel="up">How to Use Inline Assembly Language in C Code</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="subsection" id="Extended-Asm---Assembler-Instructions-with-C-Expression-Operands"><span>6.47.2 Extended Asm - Assembler Instructions with C Expression Operands<a class="copiable-link" href="#Extended-Asm---Assembler-Instructions-with-C-Expression-Operands"> ¶</a></span></h1>   <p>With extended <code class="code">asm</code> you can read and write C variables from assembler and perform jumps from assembler code to C labels. Extended <code class="code">asm</code> syntax uses colons (‘<samp class="samp">:</samp>’) to delimit the operand parameters after the assembler template: </p> <div class="example"> <pre class="example-preformatted" data-language="cpp">asm <var class="var">asm-qualifiers</var> ( <var class="var">AssemblerTemplate</var> 
                 : <var class="var">OutputOperands</var> 
                 <span class="r">[</span> : <var class="var">InputOperands</var>
                 <span class="r">[</span> : <var class="var">Clobbers</var> <span class="r">]</span> <span class="r">]</span>)

asm <var class="var">asm-qualifiers</var> ( <var class="var">AssemblerTemplate</var> 
                      : <var class="var">OutputOperands</var>
                      : <var class="var">InputOperands</var>
                      : <var class="var">Clobbers</var>
                      : <var class="var">GotoLabels</var>)</pre>
</div> <p>where in the last form, <var class="var">asm-qualifiers</var> contains <code class="code">goto</code> (and in the first form, not). </p> <p>The <code class="code">asm</code> keyword is a GNU extension. When writing code that can be compiled with <samp class="option">-ansi</samp> and the various <samp class="option">-std</samp> options, use <code class="code">__asm__</code> instead of <code class="code">asm</code> (see <a class="pxref" href="alternate-keywords">Alternate Keywords</a>). </p> <h1 class="subsubheading" id="Qualifiers-2"><span>Qualifiers<a class="copiable-link" href="#Qualifiers-2"> ¶</a></span></h1> <dl class="table"> <dt><code class="code">volatile</code></dt> <dd>
<p>The typical use of extended <code class="code">asm</code> statements is to manipulate input values to produce output values. However, your <code class="code">asm</code> statements may also produce side effects. If so, you may need to use the <code class="code">volatile</code> qualifier to disable certain optimizations. See <a class="xref" href="#Volatile">Volatile</a>. </p> </dd> <dt><code class="code">inline</code></dt> <dd>
<p>If you use the <code class="code">inline</code> qualifier, then for inlining purposes the size of the <code class="code">asm</code> statement is taken as the smallest size possible (see <a class="pxref" href="size-of-an-asm">Size of an <code class="code">asm</code></a>). </p> </dd> <dt><code class="code">goto</code></dt> <dd><p>This qualifier informs the compiler that the <code class="code">asm</code> statement may perform a jump to one of the labels listed in the <var class="var">GotoLabels</var>. See <a class="xref" href="#GotoLabels">GotoLabels</a>. </p></dd> </dl> <h1 class="subsubheading" id="Parameters-1"><span>Parameters<a class="copiable-link" href="#Parameters-1"> ¶</a></span></h1> <dl class="table"> <dt><var class="var">AssemblerTemplate</var></dt> <dd>
<p>This is a literal string that is the template for the assembler code. It is a combination of fixed text and tokens that refer to the input, output, and goto parameters. See <a class="xref" href="#AssemblerTemplate">AssemblerTemplate</a>. </p> </dd> <dt><var class="var">OutputOperands</var></dt> <dd>
<p>A comma-separated list of the C variables modified by the instructions in the <var class="var">AssemblerTemplate</var>. An empty list is permitted. See <a class="xref" href="#OutputOperands">OutputOperands</a>. </p> </dd> <dt><var class="var">InputOperands</var></dt> <dd>
<p>A comma-separated list of C expressions read by the instructions in the <var class="var">AssemblerTemplate</var>. An empty list is permitted. See <a class="xref" href="#InputOperands">InputOperands</a>. </p> </dd> <dt><var class="var">Clobbers</var></dt> <dd>
<p>A comma-separated list of registers or other values changed by the <var class="var">AssemblerTemplate</var>, beyond those listed as outputs. An empty list is permitted. See <a class="xref" href="#Clobbers-and-Scratch-Registers">Clobbers and Scratch Registers</a>. </p> </dd> <dt><var class="var">GotoLabels</var></dt> <dd>
<p>When you are using the <code class="code">goto</code> form of <code class="code">asm</code>, this section contains the list of all C labels to which the code in the <var class="var">AssemblerTemplate</var> may jump. See <a class="xref" href="#GotoLabels">GotoLabels</a>. </p> <p><code class="code">asm</code> statements may not perform jumps into other <code class="code">asm</code> statements, only to the listed <var class="var">GotoLabels</var>. GCC’s optimizers do not know about other jumps; therefore they cannot take account of them when deciding how to optimize. </p>
</dd> </dl> <p>The total number of input + output + goto operands is limited to 30. </p> <h1 class="subsubheading" id="Remarks-1"><span>Remarks<a class="copiable-link" href="#Remarks-1"> ¶</a></span></h1> <p>The <code class="code">asm</code> statement allows you to include assembly instructions directly within C code. This may help you to maximize performance in time-sensitive code or to access assembly instructions that are not readily available to C programs. </p> <p>Note that extended <code class="code">asm</code> statements must be inside a function. Only basic <code class="code">asm</code> may be outside functions (see <a class="pxref" href="basic-asm">Basic Asm — Assembler Instructions Without Operands</a>). Functions declared with the <code class="code">naked</code> attribute also require basic <code class="code">asm</code> (see <a class="pxref" href="function-attributes">Declaring Attributes of Functions</a>). </p> <p>While the uses of <code class="code">asm</code> are many and varied, it may help to think of an <code class="code">asm</code> statement as a series of low-level instructions that convert input parameters to output parameters. So a simple (if not particularly useful) example for i386 using <code class="code">asm</code> might look like this: </p> <div class="example"> <pre class="example-preformatted" data-language="cpp">int src = 1;
int dst;   

asm ("mov %1, %0\n\t"
    "add $1, %0"
    : "=r" (dst) 
    : "r" (src));

printf("%d\n", dst);</pre>
</div> <p>This code copies <code class="code">src</code> to <code class="code">dst</code> and add 1 to <code class="code">dst</code>. </p> <ul class="mini-toc"> <li><a href="#Volatile-1" accesskey="1">Volatile</a></li> <li><a href="#Assembler-Template" accesskey="2">Assembler Template</a></li> <li><a href="#Output-Operands" accesskey="3">Output Operands</a></li> <li><a href="#Flag-Output-Operands" accesskey="4">Flag Output Operands</a></li> <li><a href="#Input-Operands" accesskey="5">Input Operands</a></li> <li><a href="#Clobbers-and-Scratch-Registers-1" accesskey="6">Clobbers and Scratch Registers</a></li> <li><a href="#Goto-Labels" accesskey="7">Goto Labels</a></li> <li><a href="#Generic-Operand-Modifiers" accesskey="8">Generic Operand Modifiers</a></li> <li><a href="#x86-Operand-Modifiers" accesskey="9">x86 Operand Modifiers</a></li> <li><a href="#x86-Floating-Point-asm-Operands">x86 Floating-Point <code class="code">asm</code> Operands</a></li> <li><a href="#MSP430-Operand-Modifiers">MSP430 Operand Modifiers</a></li> <li><a href="#LoongArch-Operand-Modifiers">LoongArch Operand Modifiers</a></li> </ul> <div class="subsubsection-level-extent" id="Volatile-1"> <h1 class="subsubsection"><span>6.47.2.1 Volatile<a class="copiable-link" href="#Volatile-1"> ¶</a></span></h1>   <p>GCC’s optimizers sometimes discard <code class="code">asm</code> statements if they determine there is no need for the output variables. Also, the optimizers may move code out of loops if they believe that the code will always return the same result (i.e. none of its input values change between calls). Using the <code class="code">volatile</code> qualifier disables these optimizations. <code class="code">asm</code> statements that have no output operands and <code class="code">asm goto</code> statements, are implicitly volatile. </p> <p>This i386 code demonstrates a case that does not use (or require) the <code class="code">volatile</code> qualifier. If it is performing assertion checking, this code uses <code class="code">asm</code> to perform the validation. Otherwise, <code class="code">dwRes</code> is unreferenced by any code. As a result, the optimizers can discard the <code class="code">asm</code> statement, which in turn removes the need for the entire <code class="code">DoCheck</code> routine. By omitting the <code class="code">volatile</code> qualifier when it isn’t needed you allow the optimizers to produce the most efficient code possible. </p> <div class="example"> <pre class="example-preformatted" data-language="cpp">void DoCheck(uint32_t dwSomeValue)
{
   uint32_t dwRes;

   // Assumes dwSomeValue is not zero.
   asm ("bsfl %1,%0"
     : "=r" (dwRes)
     : "r" (dwSomeValue)
     : "cc");

   assert(dwRes &gt; 3);
}</pre>
</div> <p>The next example shows a case where the optimizers can recognize that the input (<code class="code">dwSomeValue</code>) never changes during the execution of the function and can therefore move the <code class="code">asm</code> outside the loop to produce more efficient code. Again, using the <code class="code">volatile</code> qualifier disables this type of optimization. </p> <div class="example"> <pre class="example-preformatted" data-language="cpp">void do_print(uint32_t dwSomeValue)
{
   uint32_t dwRes;

   for (uint32_t x=0; x &lt; 5; x++)
   {
      // Assumes dwSomeValue is not zero.
      asm ("bsfl %1,%0"
        : "=r" (dwRes)
        : "r" (dwSomeValue)
        : "cc");

      printf("%u: %u %u\n", x, dwSomeValue, dwRes);
   }
}</pre>
</div> <p>The following example demonstrates a case where you need to use the <code class="code">volatile</code> qualifier. It uses the x86 <code class="code">rdtsc</code> instruction, which reads the computer’s time-stamp counter. Without the <code class="code">volatile</code> qualifier, the optimizers might assume that the <code class="code">asm</code> block will always return the same value and therefore optimize away the second call. </p> <div class="example"> <pre class="example-preformatted" data-language="cpp">uint64_t msr;

asm volatile ( "rdtsc\n\t"    // Returns the time in EDX:EAX.
        "shl $32, %%rdx\n\t"  // Shift the upper bits left.
        "or %%rdx, %0"        // 'Or' in the lower bits.
        : "=a" (msr)
        : 
        : "rdx");

printf("msr: %llx\n", msr);

// Do other work...

// Reprint the timestamp
asm volatile ( "rdtsc\n\t"    // Returns the time in EDX:EAX.
        "shl $32, %%rdx\n\t"  // Shift the upper bits left.
        "or %%rdx, %0"        // 'Or' in the lower bits.
        : "=a" (msr)
        : 
        : "rdx");

printf("msr: %llx\n", msr);</pre>
</div> <p>GCC’s optimizers do not treat this code like the non-volatile code in the earlier examples. They do not move it out of loops or omit it on the assumption that the result from a previous call is still valid. </p> <p>Note that the compiler can move even <code class="code">volatile asm</code> instructions relative to other code, including across jump instructions. For example, on many targets there is a system register that controls the rounding mode of floating-point operations. Setting it with a <code class="code">volatile asm</code> statement, as in the following PowerPC example, does not work reliably. </p> <div class="example"> <pre class="example-preformatted" data-language="cpp">asm volatile("mtfsf 255, %0" : : "f" (fpenv));
sum = x + y;</pre>
</div> <p>The compiler may move the addition back before the <code class="code">volatile asm</code> statement. To make it work as expected, add an artificial dependency to the <code class="code">asm</code> by referencing a variable in the subsequent code, for example: </p> <div class="example"> <pre class="example-preformatted" data-language="cpp">asm volatile ("mtfsf 255,%1" : "=X" (sum) : "f" (fpenv));
sum = x + y;</pre>
</div> <p>Under certain circumstances, GCC may duplicate (or remove duplicates of) your assembly code when optimizing. This can lead to unexpected duplicate symbol errors during compilation if your <code class="code">asm</code> code defines symbols or labels. Using ‘<samp class="samp">%=</samp>’ (see <a class="pxref" href="#AssemblerTemplate">AssemblerTemplate</a>) may help resolve this problem. </p> 
</div> <div class="subsubsection-level-extent" id="Assembler-Template"> <h1 class="subsubsection"><span>6.47.2.2 Assembler Template<a class="copiable-link" href="#Assembler-Template"> ¶</a></span></h1>  <p>An assembler template is a literal string containing assembler instructions. The compiler replaces tokens in the template that refer to inputs, outputs, and goto labels, and then outputs the resulting string to the assembler. The string can contain any instructions recognized by the assembler, including directives. GCC does not parse the assembler instructions themselves and does not know what they mean or even whether they are valid assembler input. However, it does count the statements (see <a class="pxref" href="size-of-an-asm">Size of an <code class="code">asm</code></a>). </p> <p>You may place multiple assembler instructions together in a single <code class="code">asm</code> string, separated by the characters normally used in assembly code for the system. A combination that works in most places is a newline to break the line, plus a tab character to move to the instruction field (written as ‘<samp class="samp">\n\t</samp>’). Some assemblers allow semicolons as a line separator. However, note that some assembler dialects use semicolons to start a comment. </p> <p>Do not expect a sequence of <code class="code">asm</code> statements to remain perfectly consecutive after compilation, even when you are using the <code class="code">volatile</code> qualifier. If certain instructions need to remain consecutive in the output, put them in a single multi-instruction <code class="code">asm</code> statement. </p> <p>Accessing data from C programs without using input/output operands (such as by using global symbols directly from the assembler template) may not work as expected. Similarly, calling functions directly from an assembler template requires a detailed understanding of the target assembler and ABI. </p> <p>Since GCC does not parse the assembler template, it has no visibility of any symbols it references. This may result in GCC discarding those symbols as unreferenced unless they are also listed as input, output, or goto operands. </p> <h1 class="subsubheading" id="Special-format-strings"><span>Special format strings<a class="copiable-link" href="#Special-format-strings"> ¶</a></span></h1> <p>In addition to the tokens described by the input, output, and goto operands, these tokens have special meanings in the assembler template: </p> <dl class="table"> <dt>‘<samp class="samp">%%</samp>’</dt> <dd>
<p>Outputs a single ‘<samp class="samp">%</samp>’ into the assembler code. </p> </dd> <dt>‘<samp class="samp">%=</samp>’</dt> <dd>
<p>Outputs a number that is unique to each instance of the <code class="code">asm</code> statement in the entire compilation. This option is useful when creating local labels and referring to them multiple times in a single template that generates multiple assembler instructions. </p> </dd> <dt>‘<samp class="samp">%{</samp>’</dt> <dt>‘<samp class="samp">%|</samp>’</dt> <dt>‘<samp class="samp">%}</samp>’</dt> <dd><p>Outputs ‘<samp class="samp">{</samp>’, ‘<samp class="samp">|</samp>’, and ‘<samp class="samp">}</samp>’ characters (respectively) into the assembler code. When unescaped, these characters have special meaning to indicate multiple assembler dialects, as described below. </p></dd> </dl> <h1 class="subsubheading" id="Multiple-assembler-dialects-in-asm-templates"><span>Multiple assembler dialects in asm templates<a class="copiable-link" href="#Multiple-assembler-dialects-in-asm-templates"> ¶</a></span></h1> <p>On targets such as x86, GCC supports multiple assembler dialects. The <samp class="option">-masm</samp> option controls which dialect GCC uses as its default for inline assembler. The target-specific documentation for the <samp class="option">-masm</samp> option contains the list of supported dialects, as well as the default dialect if the option is not specified. This information may be important to understand, since assembler code that works correctly when compiled using one dialect will likely fail if compiled using another. See <a class="xref" href="x86-options">x86 Options</a>. </p> <p>If your code needs to support multiple assembler dialects (for example, if you are writing public headers that need to support a variety of compilation options), use constructs of this form: </p> <div class="example"> <pre class="example-preformatted" data-language="cpp">{ dialect0 | dialect1 | dialect2... }</pre>
</div> <p>This construct outputs <code class="code">dialect0</code> when using dialect #0 to compile the code, <code class="code">dialect1</code> for dialect #1, etc. If there are fewer alternatives within the braces than the number of dialects the compiler supports, the construct outputs nothing. </p> <p>For example, if an x86 compiler supports two dialects (‘<samp class="samp">att</samp>’, ‘<samp class="samp">intel</samp>’), an assembler template such as this: </p> <div class="example"> <pre class="example-preformatted" data-language="cpp">"bt{l %[Offset],%[Base] | %[Base],%[Offset]}; jc %l2"</pre>
</div> <p>is equivalent to one of </p> <div class="example"> <pre class="example-preformatted" data-language="cpp">"btl %[Offset],%[Base] ; jc %l2"   <span class="r">/* att dialect */</span>
"bt %[Base],%[Offset]; jc %l2"     <span class="r">/* intel dialect */</span></pre>
</div> <p>Using that same compiler, this code: </p> <div class="example"> <pre class="example-preformatted" data-language="cpp">"xchg{l}\t{%%}ebx, %1"</pre>
</div> <p>corresponds to either </p> <div class="example"> <pre class="example-preformatted" data-language="cpp">"xchgl\t%%ebx, %1"                 <span class="r">/* att dialect */</span>
"xchg\tebx, %1"                    <span class="r">/* intel dialect */</span></pre>
</div> <p>There is no support for nesting dialect alternatives. </p> 
</div> <div class="subsubsection-level-extent" id="Output-Operands"> <h1 class="subsubsection"><span>6.47.2.3 Output Operands<a class="copiable-link" href="#Output-Operands"> ¶</a></span></h1>  <p>An <code class="code">asm</code> statement has zero or more output operands indicating the names of C variables modified by the assembler code. </p> <p>In this i386 example, <code class="code">old</code> (referred to in the template string as <code class="code">%0</code>) and <code class="code">*Base</code> (as <code class="code">%1</code>) are outputs and <code class="code">Offset</code> (<code class="code">%2</code>) is an input: </p> <div class="example"> <pre class="example-preformatted" data-language="cpp">bool old;

__asm__ ("btsl %2,%1\n\t" // Turn on zero-based bit #Offset in Base.
         "sbb %0,%0"      // Use the CF to calculate old.
   : "=r" (old), "+rm" (*Base)
   : "Ir" (Offset)
   : "cc");

return old;</pre>
</div> <p>Operands are separated by commas. Each operand has this format: </p> <div class="example"> <pre class="example-preformatted" data-language="cpp"><span class="r">[</span> [<var class="var">asmSymbolicName</var>] <span class="r">]</span> <var class="var">constraint</var> (<var class="var">cvariablename</var>)</pre>
</div> <dl class="table"> <dt><var class="var">asmSymbolicName</var></dt> <dd>
<p>Specifies a symbolic name for the operand. Reference the name in the assembler template by enclosing it in square brackets (i.e. ‘<samp class="samp">%[Value]</samp>’). The scope of the name is the <code class="code">asm</code> statement that contains the definition. Any valid C variable name is acceptable, including names already defined in the surrounding code. No two operands within the same <code class="code">asm</code> statement can use the same symbolic name. </p> <p>When not using an <var class="var">asmSymbolicName</var>, use the (zero-based) position of the operand in the list of operands in the assembler template. For example if there are three output operands, use ‘<samp class="samp">%0</samp>’ in the template to refer to the first, ‘<samp class="samp">%1</samp>’ for the second, and ‘<samp class="samp">%2</samp>’ for the third. </p> </dd> <dt><var class="var">constraint</var></dt> <dd>
<p>A string constant specifying constraints on the placement of the operand; See <a class="xref" href="constraints">Constraints for <code class="code">asm</code> Operands</a>, for details. </p> <p>Output constraints must begin with either ‘<samp class="samp">=</samp>’ (a variable overwriting an existing value) or ‘<samp class="samp">+</samp>’ (when reading and writing). When using ‘<samp class="samp">=</samp>’, do not assume the location contains the existing value on entry to the <code class="code">asm</code>, except when the operand is tied to an input; see <a class="pxref" href="#InputOperands">Input Operands</a>. </p> <p>After the prefix, there must be one or more additional constraints (see <a class="pxref" href="constraints">Constraints for <code class="code">asm</code> Operands</a>) that describe where the value resides. Common constraints include ‘<samp class="samp">r</samp>’ for register and ‘<samp class="samp">m</samp>’ for memory. When you list more than one possible location (for example, <code class="code">"=rm"</code>), the compiler chooses the most efficient one based on the current context. If you list as many alternates as the <code class="code">asm</code> statement allows, you permit the optimizers to produce the best possible code. If you must use a specific register, but your Machine Constraints do not provide sufficient control to select the specific register you want, local register variables may provide a solution (see <a class="pxref" href="local-register-variables">Specifying Registers for Local Variables</a>). </p> </dd> <dt><var class="var">cvariablename</var></dt> <dd>
<p>Specifies a C lvalue expression to hold the output, typically a variable name. The enclosing parentheses are a required part of the syntax. </p> </dd> </dl> <p>When the compiler selects the registers to use to represent the output operands, it does not use any of the clobbered registers (see <a class="pxref" href="#Clobbers-and-Scratch-Registers">Clobbers and Scratch Registers</a>). </p> <p>Output operand expressions must be lvalues. The compiler cannot check whether the operands have data types that are reasonable for the instruction being executed. For output expressions that are not directly addressable (for example a bit-field), the constraint must allow a register. In that case, GCC uses the register as the output of the <code class="code">asm</code>, and then stores that register into the output. </p> <p>Operands using the ‘<samp class="samp">+</samp>’ constraint modifier count as two operands (that is, both as input and output) towards the total maximum of 30 operands per <code class="code">asm</code> statement. </p> <p>Use the ‘<samp class="samp">&amp;</samp>’ constraint modifier (see <a class="pxref" href="modifiers">Constraint Modifier Characters</a>) on all output operands that must not overlap an input. Otherwise, GCC may allocate the output operand in the same register as an unrelated input operand, on the assumption that the assembler code consumes its inputs before producing outputs. This assumption may be false if the assembler code actually consists of more than one instruction. </p> <p>The same problem can occur if one output parameter (<var class="var">a</var>) allows a register constraint and another output parameter (<var class="var">b</var>) allows a memory constraint. The code generated by GCC to access the memory address in <var class="var">b</var> can contain registers which <em class="emph">might</em> be shared by <var class="var">a</var>, and GCC considers those registers to be inputs to the asm. As above, GCC assumes that such input registers are consumed before any outputs are written. This assumption may result in incorrect behavior if the <code class="code">asm</code> statement writes to <var class="var">a</var> before using <var class="var">b</var>. Combining the ‘<samp class="samp">&amp;</samp>’ modifier with the register constraint on <var class="var">a</var> ensures that modifying <var class="var">a</var> does not affect the address referenced by <var class="var">b</var>. Otherwise, the location of <var class="var">b</var> is undefined if <var class="var">a</var> is modified before using <var class="var">b</var>. </p> <p><code class="code">asm</code> supports operand modifiers on operands (for example ‘<samp class="samp">%k2</samp>’ instead of simply ‘<samp class="samp">%2</samp>’). <a class="ref" href="#GenericOperandmodifiers">Generic Operand modifiers</a> lists the modifiers that are available on all targets. Other modifiers are hardware dependent. For example, the list of supported modifiers for x86 is found at <a class="ref" href="#x86Operandmodifiers">x86 Operand modifiers</a>. </p> <p>If the C code that follows the <code class="code">asm</code> makes no use of any of the output operands, use <code class="code">volatile</code> for the <code class="code">asm</code> statement to prevent the optimizers from discarding the <code class="code">asm</code> statement as unneeded (see <a class="ref" href="#Volatile">Volatile</a>). </p> <p>This code makes no use of the optional <var class="var">asmSymbolicName</var>. Therefore it references the first output operand as <code class="code">%0</code> (were there a second, it would be <code class="code">%1</code>, etc). The number of the first input operand is one greater than that of the last output operand. In this i386 example, that makes <code class="code">Mask</code> referenced as <code class="code">%1</code>: </p> <div class="example"> <pre class="example-preformatted" data-language="cpp">uint32_t Mask = 1234;
uint32_t Index;

  asm ("bsfl %1, %0"
     : "=r" (Index)
     : "r" (Mask)
     : "cc");</pre>
</div> <p>That code overwrites the variable <code class="code">Index</code> (‘<samp class="samp">=</samp>’), placing the value in a register (‘<samp class="samp">r</samp>’). Using the generic ‘<samp class="samp">r</samp>’ constraint instead of a constraint for a specific register allows the compiler to pick the register to use, which can result in more efficient code. This may not be possible if an assembler instruction requires a specific register. </p> <p>The following i386 example uses the <var class="var">asmSymbolicName</var> syntax. It produces the same result as the code above, but some may consider it more readable or more maintainable since reordering index numbers is not necessary when adding or removing operands. The names <code class="code">aIndex</code> and <code class="code">aMask</code> are only used in this example to emphasize which names get used where. It is acceptable to reuse the names <code class="code">Index</code> and <code class="code">Mask</code>. </p> <div class="example"> <pre class="example-preformatted" data-language="cpp">uint32_t Mask = 1234;
uint32_t Index;

  asm ("bsfl %[aMask], %[aIndex]"
     : [aIndex] "=r" (Index)
     : [aMask] "r" (Mask)
     : "cc");</pre>
</div> <p>Here are some more examples of output operands. </p> <div class="example"> <pre class="example-preformatted" data-language="cpp">uint32_t c = 1;
uint32_t d;
uint32_t *e = &amp;c;

asm ("mov %[e], %[d]"
   : [d] "=rm" (d)
   : [e] "rm" (*e));</pre>
</div> <p>Here, <code class="code">d</code> may either be in a register or in memory. Since the compiler might already have the current value of the <code class="code">uint32_t</code> location pointed to by <code class="code">e</code> in a register, you can enable it to choose the best location for <code class="code">d</code> by specifying both constraints. </p> 
</div> <div class="subsubsection-level-extent" id="Flag-Output-Operands"> <h1 class="subsubsection"><span>6.47.2.4 Flag Output Operands<a class="copiable-link" href="#Flag-Output-Operands"> ¶</a></span></h1>  <p>Some targets have a special register that holds the “flags” for the result of an operation or comparison. Normally, the contents of that register are either unmodifed by the asm, or the <code class="code">asm</code> statement is considered to clobber the contents. </p> <p>On some targets, a special form of output operand exists by which conditions in the flags register may be outputs of the asm. The set of conditions supported are target specific, but the general rule is that the output variable must be a scalar integer, and the value is boolean. When supported, the target defines the preprocessor symbol <code class="code">__GCC_ASM_FLAG_OUTPUTS__</code>. </p> <p>Because of the special nature of the flag output operands, the constraint may not include alternatives. </p> <p>Most often, the target has only one flags register, and thus is an implied operand of many instructions. In this case, the operand should not be referenced within the assembler template via <code class="code">%0</code> etc, as there’s no corresponding text in the assembly language. </p> <dl class="table"> <dt>ARM</dt> <dt>AArch64</dt> <dd>
<p>The flag output constraints for the ARM family are of the form ‘<samp class="samp">=@cc<var class="var">cond</var></samp>’ where <var class="var">cond</var> is one of the standard conditions defined in the ARM ARM for <code class="code">ConditionHolds</code>. </p> <dl class="table"> <dt><code class="code">eq</code></dt> <dd><p>Z flag set, or equal </p></dd> <dt><code class="code">ne</code></dt> <dd><p>Z flag clear or not equal </p></dd> <dt><code class="code">cs</code></dt> <dt><code class="code">hs</code></dt> <dd><p>C flag set or unsigned greater than equal </p></dd> <dt><code class="code">cc</code></dt> <dt><code class="code">lo</code></dt> <dd><p>C flag clear or unsigned less than </p></dd> <dt><code class="code">mi</code></dt> <dd><p>N flag set or “minus” </p></dd> <dt><code class="code">pl</code></dt> <dd><p>N flag clear or “plus” </p></dd> <dt><code class="code">vs</code></dt> <dd><p>V flag set or signed overflow </p></dd> <dt><code class="code">vc</code></dt> <dd><p>V flag clear </p></dd> <dt><code class="code">hi</code></dt> <dd><p>unsigned greater than </p></dd> <dt><code class="code">ls</code></dt> <dd><p>unsigned less than equal </p></dd> <dt><code class="code">ge</code></dt> <dd><p>signed greater than equal </p></dd> <dt><code class="code">lt</code></dt> <dd><p>signed less than </p></dd> <dt><code class="code">gt</code></dt> <dd><p>signed greater than </p></dd> <dt><code class="code">le</code></dt> <dd><p>signed less than equal </p></dd> </dl> <p>The flag output constraints are not supported in thumb1 mode. </p> </dd> <dt>x86 family</dt> <dd>
<p>The flag output constraints for the x86 family are of the form ‘<samp class="samp">=@cc<var class="var">cond</var></samp>’ where <var class="var">cond</var> is one of the standard conditions defined in the ISA manual for <code class="code">j<var class="var">cc</var></code> or <code class="code">set<var class="var">cc</var></code>. </p> <dl class="table"> <dt><code class="code">a</code></dt> <dd><p>“above” or unsigned greater than </p></dd> <dt><code class="code">ae</code></dt> <dd><p>“above or equal” or unsigned greater than or equal </p></dd> <dt><code class="code">b</code></dt> <dd><p>“below” or unsigned less than </p></dd> <dt><code class="code">be</code></dt> <dd><p>“below or equal” or unsigned less than or equal </p></dd> <dt><code class="code">c</code></dt> <dd><p>carry flag set </p></dd> <dt><code class="code">e</code></dt> <dt><code class="code">z</code></dt> <dd><p>“equal” or zero flag set </p></dd> <dt><code class="code">g</code></dt> <dd><p>signed greater than </p></dd> <dt><code class="code">ge</code></dt> <dd><p>signed greater than or equal </p></dd> <dt><code class="code">l</code></dt> <dd><p>signed less than </p></dd> <dt><code class="code">le</code></dt> <dd><p>signed less than or equal </p></dd> <dt><code class="code">o</code></dt> <dd><p>overflow flag set </p></dd> <dt><code class="code">p</code></dt> <dd><p>parity flag set </p></dd> <dt><code class="code">s</code></dt> <dd><p>sign flag set </p></dd> <dt><code class="code">na</code></dt> <dt><code class="code">nae</code></dt> <dt><code class="code">nb</code></dt> <dt><code class="code">nbe</code></dt> <dt><code class="code">nc</code></dt> <dt><code class="code">ne</code></dt> <dt><code class="code">ng</code></dt> <dt><code class="code">nge</code></dt> <dt><code class="code">nl</code></dt> <dt><code class="code">nle</code></dt> <dt><code class="code">no</code></dt> <dt><code class="code">np</code></dt> <dt><code class="code">ns</code></dt> <dt><code class="code">nz</code></dt> <dd><p>“not” <var class="var">flag</var>, or inverted versions of those above </p></dd> </dl> </dd> </dl> 
</div> <div class="subsubsection-level-extent" id="Input-Operands"> <h1 class="subsubsection"><span>6.47.2.5 Input Operands<a class="copiable-link" href="#Input-Operands"> ¶</a></span></h1>   <p>Input operands make values from C variables and expressions available to the assembly code. </p> <p>Operands are separated by commas. Each operand has this format: </p> <div class="example"> <pre class="example-preformatted" data-language="cpp"><span class="r">[</span> [<var class="var">asmSymbolicName</var>] <span class="r">]</span> <var class="var">constraint</var> (<var class="var">cexpression</var>)</pre>
</div> <dl class="table"> <dt><var class="var">asmSymbolicName</var></dt> <dd>
<p>Specifies a symbolic name for the operand. Reference the name in the assembler template by enclosing it in square brackets (i.e. ‘<samp class="samp">%[Value]</samp>’). The scope of the name is the <code class="code">asm</code> statement that contains the definition. Any valid C variable name is acceptable, including names already defined in the surrounding code. No two operands within the same <code class="code">asm</code> statement can use the same symbolic name. </p> <p>When not using an <var class="var">asmSymbolicName</var>, use the (zero-based) position of the operand in the list of operands in the assembler template. For example if there are two output operands and three inputs, use ‘<samp class="samp">%2</samp>’ in the template to refer to the first input operand, ‘<samp class="samp">%3</samp>’ for the second, and ‘<samp class="samp">%4</samp>’ for the third. </p> </dd> <dt><var class="var">constraint</var></dt> <dd>
<p>A string constant specifying constraints on the placement of the operand; See <a class="xref" href="constraints">Constraints for <code class="code">asm</code> Operands</a>, for details. </p> <p>Input constraint strings may not begin with either ‘<samp class="samp">=</samp>’ or ‘<samp class="samp">+</samp>’. When you list more than one possible location (for example, ‘<samp class="samp">"irm"</samp>’), the compiler chooses the most efficient one based on the current context. If you must use a specific register, but your Machine Constraints do not provide sufficient control to select the specific register you want, local register variables may provide a solution (see <a class="pxref" href="local-register-variables">Specifying Registers for Local Variables</a>). </p> <p>Input constraints can also be digits (for example, <code class="code">"0"</code>). This indicates that the specified input must be in the same place as the output constraint at the (zero-based) index in the output constraint list. When using <var class="var">asmSymbolicName</var> syntax for the output operands, you may use these names (enclosed in brackets ‘<samp class="samp">[]</samp>’) instead of digits. </p> </dd> <dt><var class="var">cexpression</var></dt> <dd>
<p>This is the C variable or expression being passed to the <code class="code">asm</code> statement as input. The enclosing parentheses are a required part of the syntax. </p> </dd> </dl> <p>When the compiler selects the registers to use to represent the input operands, it does not use any of the clobbered registers (see <a class="pxref" href="#Clobbers-and-Scratch-Registers">Clobbers and Scratch Registers</a>). </p> <p>If there are no output operands but there are input operands, place two consecutive colons where the output operands would go: </p> <div class="example"> <pre class="example-preformatted" data-language="cpp">__asm__ ("some instructions"
   : /* No outputs. */
   : "r" (Offset / 8));</pre>
</div> <p><strong class="strong">Warning:</strong> Do <em class="emph">not</em> modify the contents of input-only operands (except for inputs tied to outputs). The compiler assumes that on exit from the <code class="code">asm</code> statement these operands contain the same values as they had before executing the statement. It is <em class="emph">not</em> possible to use clobbers to inform the compiler that the values in these inputs are changing. One common work-around is to tie the changing input variable to an output variable that never gets used. Note, however, that if the code that follows the <code class="code">asm</code> statement makes no use of any of the output operands, the GCC optimizers may discard the <code class="code">asm</code> statement as unneeded (see <a class="ref" href="#Volatile">Volatile</a>). </p> <p><code class="code">asm</code> supports operand modifiers on operands (for example ‘<samp class="samp">%k2</samp>’ instead of simply ‘<samp class="samp">%2</samp>’). <a class="ref" href="#GenericOperandmodifiers">Generic Operand modifiers</a> lists the modifiers that are available on all targets. Other modifiers are hardware dependent. For example, the list of supported modifiers for x86 is found at <a class="ref" href="#x86Operandmodifiers">x86 Operand modifiers</a>. </p> <p>In this example using the fictitious <code class="code">combine</code> instruction, the constraint <code class="code">"0"</code> for input operand 1 says that it must occupy the same location as output operand 0. Only input operands may use numbers in constraints, and they must each refer to an output operand. Only a number (or the symbolic assembler name) in the constraint can guarantee that one operand is in the same place as another. The mere fact that <code class="code">foo</code> is the value of both operands is not enough to guarantee that they are in the same place in the generated assembler code. </p> <div class="example"> <pre class="example-preformatted" data-language="cpp">asm ("combine %2, %0" 
   : "=r" (foo) 
   : "0" (foo), "g" (bar));</pre>
</div> <p>Here is an example using symbolic names. </p> <div class="example"> <pre class="example-preformatted" data-language="cpp">asm ("cmoveq %1, %2, %[result]" 
   : [result] "=r"(result) 
   : "r" (test), "r" (new), "[result]" (old));</pre>
</div> 
</div> <div class="subsubsection-level-extent" id="Clobbers-and-Scratch-Registers-1"> <h1 class="subsubsection"><span>6.47.2.6 Clobbers and Scratch Registers<a class="copiable-link" href="#Clobbers-and-Scratch-Registers-1"> ¶</a></span></h1>   <p>While the compiler is aware of changes to entries listed in the output operands, the inline <code class="code">asm</code> code may modify more than just the outputs. For example, calculations may require additional registers, or the processor may overwrite a register as a side effect of a particular assembler instruction. In order to inform the compiler of these changes, list them in the clobber list. Clobber list items are either register names or the special clobbers (listed below). Each clobber list item is a string constant enclosed in double quotes and separated by commas. </p> <p>Clobber descriptions may not in any way overlap with an input or output operand. For example, you may not have an operand describing a register class with one member when listing that register in the clobber list. Variables declared to live in specific registers (see <a class="pxref" href="explicit-register-variables">Variables in Specified Registers</a>) and used as <code class="code">asm</code> input or output operands must have no part mentioned in the clobber description. In particular, there is no way to specify that input operands get modified without also specifying them as output operands. </p> <p>When the compiler selects which registers to use to represent input and output operands, it does not use any of the clobbered registers. As a result, clobbered registers are available for any use in the assembler code. </p> <p>Another restriction is that the clobber list should not contain the stack pointer register. This is because the compiler requires the value of the stack pointer to be the same after an <code class="code">asm</code> statement as it was on entry to the statement. However, previous versions of GCC did not enforce this rule and allowed the stack pointer to appear in the list, with unclear semantics. This behavior is deprecated and listing the stack pointer may become an error in future versions of GCC. </p> <p>Here is a realistic example for the VAX showing the use of clobbered registers: </p> <div class="example"> <pre class="example-preformatted" data-language="cpp">asm volatile ("movc3 %0, %1, %2"
                   : /* No outputs. */
                   : "g" (from), "g" (to), "g" (count)
                   : "r0", "r1", "r2", "r3", "r4", "r5", "memory");</pre>
</div> <p>Also, there are two special clobber arguments: </p> <dl class="table"> <dt><code class="code">"cc"</code></dt> <dd>
<p>The <code class="code">"cc"</code> clobber indicates that the assembler code modifies the flags register. On some machines, GCC represents the condition codes as a specific hardware register; <code class="code">"cc"</code> serves to name this register. On other machines, condition code handling is different, and specifying <code class="code">"cc"</code> has no effect. But it is valid no matter what the target. </p> </dd> <dt><code class="code">"memory"</code></dt> <dd>
<p>The <code class="code">"memory"</code> clobber tells the compiler that the assembly code performs memory reads or writes to items other than those listed in the input and output operands (for example, accessing the memory pointed to by one of the input parameters). To ensure memory contains correct values, GCC may need to flush specific register values to memory before executing the <code class="code">asm</code>. Further, the compiler does not assume that any values read from memory before an <code class="code">asm</code> remain unchanged after that <code class="code">asm</code>; it reloads them as needed. Using the <code class="code">"memory"</code> clobber effectively forms a read/write memory barrier for the compiler. </p> <p>Note that this clobber does not prevent the <em class="emph">processor</em> from doing speculative reads past the <code class="code">asm</code> statement. To prevent that, you need processor-specific fence instructions. </p> </dd> </dl> <p>Flushing registers to memory has performance implications and may be an issue for time-sensitive code. You can provide better information to GCC to avoid this, as shown in the following examples. At a minimum, aliasing rules allow GCC to know what memory <em class="emph">doesn’t</em> need to be flushed. </p> <p>Here is a fictitious sum of squares instruction, that takes two pointers to floating point values in memory and produces a floating point register output. Notice that <code class="code">x</code>, and <code class="code">y</code> both appear twice in the <code class="code">asm</code> parameters, once to specify memory accessed, and once to specify a base register used by the <code class="code">asm</code>. You won’t normally be wasting a register by doing this as GCC can use the same register for both purposes. However, it would be foolish to use both <code class="code">%1</code> and <code class="code">%3</code> for <code class="code">x</code> in this <code class="code">asm</code> and expect them to be the same. In fact, <code class="code">%3</code> may well not be a register. It might be a symbolic memory reference to the object pointed to by <code class="code">x</code>. </p> <div class="example smallexample"> <pre class="example-preformatted" data-language="cpp">asm ("sumsq %0, %1, %2"
     : "+f" (result)
     : "r" (x), "r" (y), "m" (*x), "m" (*y));</pre>
</div> <p>Here is a fictitious <code class="code">*z++ = *x++ * *y++</code> instruction. Notice that the <code class="code">x</code>, <code class="code">y</code> and <code class="code">z</code> pointer registers must be specified as input/output because the <code class="code">asm</code> modifies them. </p> <div class="example smallexample"> <pre class="example-preformatted" data-language="cpp">asm ("vecmul %0, %1, %2"
     : "+r" (z), "+r" (x), "+r" (y), "=m" (*z)
     : "m" (*x), "m" (*y));</pre>
</div> <p>An x86 example where the string memory argument is of unknown length. </p> <div class="example smallexample"> <pre class="example-preformatted" data-language="cpp">asm("repne scasb"
    : "=c" (count), "+D" (p)
    : "m" (*(const char (*)[]) p), "0" (-1), "a" (0));</pre>
</div> <p>If you know the above will only be reading a ten byte array then you could instead use a memory input like: <code class="code">"m" (*(const char (*)[10]) p)</code>. </p> <p>Here is an example of a PowerPC vector scale implemented in assembly, complete with vector and condition code clobbers, and some initialized offset registers that are unchanged by the <code class="code">asm</code>. </p> <div class="example smallexample"> <pre class="example-preformatted" data-language="cpp">void
dscal (size_t n, double *x, double alpha)
{
  asm ("/* lots of asm here */"
       : "+m" (*(double (*)[n]) x), "+&amp;r" (n), "+b" (x)
       : "d" (alpha), "b" (32), "b" (48), "b" (64),
         "b" (80), "b" (96), "b" (112)
       : "cr0",
         "vs32","vs33","vs34","vs35","vs36","vs37","vs38","vs39",
         "vs40","vs41","vs42","vs43","vs44","vs45","vs46","vs47");
}</pre>
</div> <p>Rather than allocating fixed registers via clobbers to provide scratch registers for an <code class="code">asm</code> statement, an alternative is to define a variable and make it an early-clobber output as with <code class="code">a2</code> and <code class="code">a3</code> in the example below. This gives the compiler register allocator more freedom. You can also define a variable and make it an output tied to an input as with <code class="code">a0</code> and <code class="code">a1</code>, tied respectively to <code class="code">ap</code> and <code class="code">lda</code>. Of course, with tied outputs your <code class="code">asm</code> can’t use the input value after modifying the output register since they are one and the same register. What’s more, if you omit the early-clobber on the output, it is possible that GCC might allocate the same register to another of the inputs if GCC could prove they had the same value on entry to the <code class="code">asm</code>. This is why <code class="code">a1</code> has an early-clobber. Its tied input, <code class="code">lda</code> might conceivably be known to have the value 16 and without an early-clobber share the same register as <code class="code">%11</code>. On the other hand, <code class="code">ap</code> can’t be the same as any of the other inputs, so an early-clobber on <code class="code">a0</code> is not needed. It is also not desirable in this case. An early-clobber on <code class="code">a0</code> would cause GCC to allocate a separate register for the <code class="code">"m" (*(const double (*)[]) ap)</code> input. Note that tying an input to an output is the way to set up an initialized temporary register modified by an <code class="code">asm</code> statement. An input not tied to an output is assumed by GCC to be unchanged, for example <code class="code">"b" (16)</code> below sets up <code class="code">%11</code> to 16, and GCC might use that register in following code if the value 16 happened to be needed. You can even use a normal <code class="code">asm</code> output for a scratch if all inputs that might share the same register are consumed before the scratch is used. The VSX registers clobbered by the <code class="code">asm</code> statement could have used this technique except for GCC’s limit on the number of <code class="code">asm</code> parameters. </p> <div class="example smallexample"> <pre class="example-preformatted" data-language="cpp">static void
dgemv_kernel_4x4 (long n, const double *ap, long lda,
                  const double *x, double *y, double alpha)
{
  double *a0;
  double *a1;
  double *a2;
  double *a3;

  __asm__
    (
     /* lots of asm here */
     "#n=%1 ap=%8=%12 lda=%13 x=%7=%10 y=%0=%2 alpha=%9 o16=%11\n"
     "#a0=%3 a1=%4 a2=%5 a3=%6"
     :
       "+m" (*(double (*)[n]) y),
       "+&amp;r" (n),	// 1
       "+b" (y),	// 2
       "=b" (a0),	// 3
       "=&amp;b" (a1),	// 4
       "=&amp;b" (a2),	// 5
       "=&amp;b" (a3)	// 6
     :
       "m" (*(const double (*)[n]) x),
       "m" (*(const double (*)[]) ap),
       "d" (alpha),	// 9
       "r" (x),		// 10
       "b" (16),	// 11
       "3" (ap),	// 12
       "4" (lda)	// 13
     :
       "cr0",
       "vs32","vs33","vs34","vs35","vs36","vs37",
       "vs40","vs41","vs42","vs43","vs44","vs45","vs46","vs47"
     );
}</pre>
</div> 
</div> <div class="subsubsection-level-extent" id="Goto-Labels"> <h1 class="subsubsection"><span>6.47.2.7 Goto Labels<a class="copiable-link" href="#Goto-Labels"> ¶</a></span></h1>  <p><code class="code">asm goto</code> allows assembly code to jump to one or more C labels. The <var class="var">GotoLabels</var> section in an <code class="code">asm goto</code> statement contains a comma-separated list of all C labels to which the assembler code may jump. GCC assumes that <code class="code">asm</code> execution falls through to the next statement (if this is not the case, consider using the <code class="code">__builtin_unreachable</code> intrinsic after the <code class="code">asm</code> statement). Optimization of <code class="code">asm goto</code> may be improved by using the <code class="code">hot</code> and <code class="code">cold</code> label attributes (see <a class="pxref" href="label-attributes">Label Attributes</a>). </p> <p>If the assembler code does modify anything, use the <code class="code">"memory"</code> clobber to force the optimizers to flush all register values to memory and reload them if necessary after the <code class="code">asm</code> statement. </p> <p>Also note that an <code class="code">asm goto</code> statement is always implicitly considered volatile. </p> <p>Be careful when you set output operands inside <code class="code">asm goto</code> only on some possible control flow paths. If you don’t set up the output on given path and never use it on this path, it is okay. Otherwise, you should use ‘<samp class="samp">+</samp>’ constraint modifier meaning that the operand is input and output one. With this modifier you will have the correct values on all possible paths from the <code class="code">asm goto</code>. </p> <p>To reference a label in the assembler template, prefix it with ‘<samp class="samp">%l</samp>’ (lowercase ‘<samp class="samp">L</samp>’) followed by its (zero-based) position in <var class="var">GotoLabels</var> plus the number of input and output operands. Output operand with constraint modifier ‘<samp class="samp">+</samp>’ is counted as two operands because it is considered as one output and one input operand. For example, if the <code class="code">asm</code> has three inputs, one output operand with constraint modifier ‘<samp class="samp">+</samp>’ and one output operand with constraint modifier ‘<samp class="samp">=</samp>’ and references two labels, refer to the first label as ‘<samp class="samp">%l6</samp>’ and the second as ‘<samp class="samp">%l7</samp>’). </p> <p>Alternately, you can reference labels using the actual C label name enclosed in brackets. For example, to reference a label named <code class="code">carry</code>, you can use ‘<samp class="samp">%l[carry]</samp>’. The label must still be listed in the <var class="var">GotoLabels</var> section when using this approach. It is better to use the named references for labels as in this case you can avoid counting input and output operands and special treatment of output operands with constraint modifier ‘<samp class="samp">+</samp>’. </p> <p>Here is an example of <code class="code">asm goto</code> for i386: </p> <div class="example"> <pre class="example-preformatted" data-language="cpp">asm goto (
    "btl %1, %0\n\t"
    "jc %l2"
    : /* No outputs. */
    : "r" (p1), "r" (p2) 
    : "cc" 
    : carry);

return 0;

carry:
return 1;</pre>
</div> <p>The following example shows an <code class="code">asm goto</code> that uses a memory clobber. </p> <div class="example"> <pre class="example-preformatted" data-language="cpp">int frob(int x)
{
  int y;
  asm goto ("frob %%r5, %1; jc %l[error]; mov (%2), %%r5"
            : /* No outputs. */
            : "r"(x), "r"(&amp;y)
            : "r5", "memory" 
            : error);
  return y;
error:
  return -1;
}</pre>
</div> <p>The following example shows an <code class="code">asm goto</code> that uses an output. </p> <div class="example"> <pre class="example-preformatted" data-language="cpp">int foo(int count)
{
  asm goto ("dec %0; jb %l[stop]"
            : "+r" (count)
            :
            :
            : stop);
  return count;
stop:
  return 0;
}</pre>
</div> <p>The following artificial example shows an <code class="code">asm goto</code> that sets up an output only on one path inside the <code class="code">asm goto</code>. Usage of constraint modifier <code class="code">=</code> instead of <code class="code">+</code> would be wrong as <code class="code">factor</code> is used on all paths from the <code class="code">asm goto</code>. </p> <div class="example"> <pre class="example-preformatted" data-language="cpp">int foo(int inp)
{
  int factor = 0;
  asm goto ("cmp %1, 10; jb %l[lab]; mov 2, %0"
            : "+r" (factor)
            : "r" (inp)
            :
            : lab);
lab:
  return inp * factor; /* return 2 * inp or 0 if inp &lt; 10 */
}</pre>
</div> 
</div> <div class="subsubsection-level-extent" id="Generic-Operand-Modifiers"> <h1 class="subsubsection"><span>6.47.2.8 Generic Operand Modifiers<a class="copiable-link" href="#Generic-Operand-Modifiers"> ¶</a></span></h1> <p>The following table shows the modifiers supported by all targets and their effects: </p> <table class="multitable"> <thead><tr>
<th width="15%">Modifier</th>
<th width="70%">Description</th>
<th width="15%">Example</th>
</tr></thead> <tbody>
<tr>
<td width="15%"><code class="code">c</code></td>
<td width="70%">Require a constant operand and print the constant expression with no punctuation.</td>
<td width="15%"><code class="code">%c0</code></td>
</tr> <tr>
<td width="15%"><code class="code">n</code></td>
<td width="70%">Like ‘<samp class="samp">%c</samp>’ except that the value of the constant is negated before printing.</td>
<td width="15%"><code class="code">%n0</code></td>
</tr> <tr>
<td width="15%"><code class="code">a</code></td>
<td width="70%">Substitute a memory reference, with the actual operand treated as the address. This may be useful when outputting a “load address” instruction, because often the assembler syntax for such an instruction requires you to write the operand as if it were a memory reference.</td>
<td width="15%"><code class="code">%a0</code></td>
</tr> <tr>
<td width="15%"><code class="code">l</code></td>
<td width="70%">Print the label name with no punctuation.</td>
<td width="15%"><code class="code">%l0</code></td>
</tr> </tbody> </table> 
</div> <div class="subsubsection-level-extent" id="x86-Operand-Modifiers"> <h1 class="subsubsection"><span>6.47.2.9 x86 Operand Modifiers<a class="copiable-link" href="#x86-Operand-Modifiers"> ¶</a></span></h1> <p>References to input, output, and goto operands in the assembler template of extended <code class="code">asm</code> statements can use modifiers to affect the way the operands are formatted in the code output to the assembler. For example, the following code uses the ‘<samp class="samp">h</samp>’ and ‘<samp class="samp">b</samp>’ modifiers for x86: </p> <div class="example"> <pre class="example-preformatted" data-language="cpp">uint16_t  num;
asm volatile ("xchg %h0, %b0" : "+a" (num) );</pre>
</div> <p>These modifiers generate this assembler code: </p> <div class="example"> <pre class="example-preformatted" data-language="cpp">xchg %ah, %al</pre>
</div> <p>The rest of this discussion uses the following code for illustrative purposes. </p> <div class="example"> <pre class="example-preformatted" data-language="cpp">int main()
{
   int iInt = 1;

top:

   asm volatile goto ("some assembler instructions here"
   : /* No outputs. */
   : "q" (iInt), "X" (sizeof(unsigned char) + 1), "i" (42)
   : /* No clobbers. */
   : top);
}</pre>
</div> <p>With no modifiers, this is what the output from the operands would be for the ‘<samp class="samp">att</samp>’ and ‘<samp class="samp">intel</samp>’ dialects of assembler: </p> <table class="multitable"> <thead><tr>
<th>Operand</th>
<th>‘<samp class="samp">att</samp>’</th>
<th>‘<samp class="samp">intel</samp>’</th>
</tr></thead> <tbody>
<tr>
<td><code class="code">%0</code></td>
<td><code class="code">%eax</code></td>
<td><code class="code">eax</code></td>
</tr> <tr>
<td><code class="code">%1</code></td>
<td><code class="code">$2</code></td>
<td><code class="code">2</code></td>
</tr> <tr>
<td><code class="code">%3</code></td>
<td><code class="code">$.L3</code></td>
<td><code class="code">OFFSET FLAT:.L3</code></td>
</tr> <tr>
<td><code class="code">%4</code></td>
<td><code class="code">$8</code></td>
<td><code class="code">8</code></td>
</tr> <tr>
<td><code class="code">%5</code></td>
<td><code class="code">%xmm0</code></td>
<td><code class="code">xmm0</code></td>
</tr> <tr>
<td><code class="code">%7</code></td>
<td><code class="code">$0</code></td>
<td><code class="code">0</code></td>
</tr> </tbody> </table> <p>The table below shows the list of supported modifiers and their effects. </p> <table class="multitable"> <thead><tr>
<th>Modifier</th>
<th>Description</th>
<th>Operand</th>
<th>‘<samp class="samp">att</samp>’</th>
<th>‘<samp class="samp">intel</samp>’</th>
</tr></thead> <tbody>
<tr>
<td><code class="code">A</code></td>
<td>Print an absolute memory reference.</td>
<td><code class="code">%A0</code></td>
<td><code class="code">*%rax</code></td>
<td><code class="code">rax</code></td>
</tr> <tr>
<td><code class="code">b</code></td>
<td>Print the QImode name of the register.</td>
<td><code class="code">%b0</code></td>
<td><code class="code">%al</code></td>
<td><code class="code">al</code></td>
</tr> <tr>
<td><code class="code">B</code></td>
<td>print the opcode suffix of b.</td>
<td><code class="code">%B0</code></td>
<td><code class="code">b</code></td>
<td></td>
</tr> <tr>
<td><code class="code">c</code></td>
<td>Require a constant operand and print the constant expression with no punctuation.</td>
<td><code class="code">%c1</code></td>
<td><code class="code">2</code></td>
<td><code class="code">2</code></td>
</tr> <tr>
<td><code class="code">d</code></td>
<td>print duplicated register operand for AVX instruction.</td>
<td><code class="code">%d5</code></td>
<td><code class="code">%xmm0, %xmm0</code></td>
<td><code class="code">xmm0, xmm0</code></td>
</tr> <tr>
<td><code class="code">E</code></td>
<td>Print the address in Double Integer (DImode) mode (8 bytes) when the target is 64-bit. Otherwise mode is unspecified (VOIDmode).</td>
<td><code class="code">%E1</code></td>
<td><code class="code">%(rax)</code></td>
<td><code class="code">[rax]</code></td>
</tr> <tr>
<td><code class="code">g</code></td>
<td>Print the V16SFmode name of the register.</td>
<td><code class="code">%g0</code></td>
<td><code class="code">%zmm0</code></td>
<td><code class="code">zmm0</code></td>
</tr> <tr>
<td><code class="code">h</code></td>
<td>Print the QImode name for a “high” register.</td>
<td><code class="code">%h0</code></td>
<td><code class="code">%ah</code></td>
<td><code class="code">ah</code></td>
</tr> <tr>
<td><code class="code">H</code></td>
<td>Add 8 bytes to an offsettable memory reference. Useful when accessing the high 8 bytes of SSE values. For a memref in (%rax), it generates</td>
<td><code class="code">%H0</code></td>
<td><code class="code">8(%rax)</code></td>
<td><code class="code">8[rax]</code></td>
</tr> <tr>
<td><code class="code">k</code></td>
<td>Print the SImode name of the register.</td>
<td><code class="code">%k0</code></td>
<td><code class="code">%eax</code></td>
<td><code class="code">eax</code></td>
</tr> <tr>
<td><code class="code">l</code></td>
<td>Print the label name with no punctuation.</td>
<td><code class="code">%l3</code></td>
<td><code class="code">.L3</code></td>
<td><code class="code">.L3</code></td>
</tr> <tr>
<td><code class="code">L</code></td>
<td>print the opcode suffix of l.</td>
<td><code class="code">%L0</code></td>
<td><code class="code">l</code></td>
<td></td>
</tr> <tr>
<td><code class="code">N</code></td>
<td>print maskz.</td>
<td><code class="code">%N7</code></td>
<td><code class="code">{z}</code></td>
<td><code class="code">{z}</code></td>
</tr> <tr>
<td><code class="code">p</code></td>
<td>Print raw symbol name (without syntax-specific prefixes).</td>
<td><code class="code">%p2</code></td>
<td><code class="code">42</code></td>
<td><code class="code">42</code></td>
</tr> <tr>
<td><code class="code">P</code></td>
<td>If used for a function, print the PLT suffix and generate PIC code. For example, emit <code class="code">foo@PLT</code> instead of ’foo’ for the function foo(). If used for a constant, drop all syntax-specific prefixes and issue the bare constant. See <code class="code">p</code> above.</td>
</tr> <tr>
<td><code class="code">q</code></td>
<td>Print the DImode name of the register.</td>
<td><code class="code">%q0</code></td>
<td><code class="code">%rax</code></td>
<td><code class="code">rax</code></td>
</tr> <tr>
<td><code class="code">Q</code></td>
<td>print the opcode suffix of q.</td>
<td><code class="code">%Q0</code></td>
<td><code class="code">q</code></td>
<td></td>
</tr> <tr>
<td><code class="code">R</code></td>
<td>print embedded rounding and sae.</td>
<td><code class="code">%R4</code></td>
<td><code class="code">{rn-sae}, </code></td>
<td><code class="code">, {rn-sae}</code></td>
</tr> <tr>
<td><code class="code">r</code></td>
<td>print only sae.</td>
<td><code class="code">%r4</code></td>
<td><code class="code">{sae}, </code></td>
<td><code class="code">, {sae}</code></td>
</tr> <tr>
<td><code class="code">s</code></td>
<td>print a shift double count, followed by the assemblers argument delimiterprint the opcode suffix of s.</td>
<td><code class="code">%s1</code></td>
<td><code class="code">$2, </code></td>
<td><code class="code">2, </code></td>
</tr> <tr>
<td><code class="code">S</code></td>
<td>print the opcode suffix of s.</td>
<td><code class="code">%S0</code></td>
<td><code class="code">s</code></td>
<td></td>
</tr> <tr>
<td><code class="code">t</code></td>
<td>print the V8SFmode name of the register.</td>
<td><code class="code">%t5</code></td>
<td><code class="code">%ymm0</code></td>
<td><code class="code">ymm0</code></td>
</tr> <tr>
<td><code class="code">T</code></td>
<td>print the opcode suffix of t.</td>
<td><code class="code">%T0</code></td>
<td><code class="code">t</code></td>
<td></td>
</tr> <tr>
<td><code class="code">V</code></td>
<td>print naked full integer register name without %.</td>
<td><code class="code">%V0</code></td>
<td><code class="code">eax</code></td>
<td><code class="code">eax</code></td>
</tr> <tr>
<td><code class="code">w</code></td>
<td>Print the HImode name of the register.</td>
<td><code class="code">%w0</code></td>
<td><code class="code">%ax</code></td>
<td><code class="code">ax</code></td>
</tr> <tr>
<td><code class="code">W</code></td>
<td>print the opcode suffix of w.</td>
<td><code class="code">%W0</code></td>
<td><code class="code">w</code></td>
<td></td>
</tr> <tr>
<td><code class="code">x</code></td>
<td>print the V4SFmode name of the register.</td>
<td><code class="code">%x5</code></td>
<td><code class="code">%xmm0</code></td>
<td><code class="code">xmm0</code></td>
</tr> <tr>
<td><code class="code">y</code></td>
<td>print "st(0)" instead of "st" as a register.</td>
<td><code class="code">%y6</code></td>
<td><code class="code">%st(0)</code></td>
<td><code class="code">st(0)</code></td>
</tr> <tr>
<td><code class="code">z</code></td>
<td>Print the opcode suffix for the size of the current integer operand (one of <code class="code">b</code>/<code class="code">w</code>/<code class="code">l</code>/<code class="code">q</code>).</td>
<td><code class="code">%z0</code></td>
<td><code class="code">l</code></td>
<td></td>
</tr> <tr>
<td><code class="code">Z</code></td>
<td>Like <code class="code">z</code>, with special suffixes for x87 instructions.</td>
</tr> </tbody> </table> 
</div> <div class="subsubsection-level-extent" id="x86-Floating-Point-asm-Operands"> <h1 class="subsubsection"><span>6.47.2.10 x86 Floating-Point asm Operands<a class="copiable-link" href="#x86-Floating-Point-asm-Operands"> ¶</a></span></h1> <p>On x86 targets, there are several rules on the usage of stack-like registers in the operands of an <code class="code">asm</code>. These rules apply only to the operands that are stack-like registers: </p> <ol class="enumerate"> <li> Given a set of input registers that die in an <code class="code">asm</code>, it is necessary to know which are implicitly popped by the <code class="code">asm</code>, and which must be explicitly popped by GCC. <p>An input register that is implicitly popped by the <code class="code">asm</code> must be explicitly clobbered, unless it is constrained to match an output operand. </p> </li>
<li> For any input register that is implicitly popped by an <code class="code">asm</code>, it is necessary to know how to adjust the stack to compensate for the pop. If any non-popped input is closer to the top of the reg-stack than the implicitly popped register, it would not be possible to know what the stack looked like—it’s not clear how the rest of the stack “slides up”. <p>All implicitly popped input registers must be closer to the top of the reg-stack than any input that is not implicitly popped. </p> <p>It is possible that if an input dies in an <code class="code">asm</code>, the compiler might use the input register for an output reload. Consider this example: </p> <div class="example smallexample"> <pre class="example-preformatted" data-language="cpp">asm ("foo" : "=t" (a) : "f" (b));</pre>
</div> <p>This code says that input <code class="code">b</code> is not popped by the <code class="code">asm</code>, and that the <code class="code">asm</code> pushes a result onto the reg-stack, i.e., the stack is one deeper after the <code class="code">asm</code> than it was before. But, it is possible that reload may think that it can use the same register for both the input and the output. </p> <p>To prevent this from happening, if any input operand uses the ‘<samp class="samp">f</samp>’ constraint, all output register constraints must use the ‘<samp class="samp">&amp;</samp>’ early-clobber modifier. </p> <p>The example above is correctly written as: </p> <div class="example smallexample"> <pre class="example-preformatted" data-language="cpp">asm ("foo" : "=&amp;t" (a) : "f" (b));</pre>
</div> </li>
<li> Some operands need to be in particular places on the stack. All output operands fall in this category—GCC has no other way to know which registers the outputs appear in unless you indicate this in the constraints. <p>Output operands must specifically indicate which register an output appears in after an <code class="code">asm</code>. ‘<samp class="samp">=f</samp>’ is not allowed: the operand constraints must select a class with a single register. </p> </li>
<li> Output operands may not be “inserted” between existing stack registers. Since no 387 opcode uses a read/write operand, all output operands are dead before the <code class="code">asm</code>, and are pushed by the <code class="code">asm</code>. It makes no sense to push anywhere but the top of the reg-stack. <p>Output operands must start at the top of the reg-stack: output operands may not “skip” a register. </p> </li>
<li> Some <code class="code">asm</code> statements may need extra stack space for internal calculations. This can be guaranteed by clobbering stack registers unrelated to the inputs and outputs. </li>
</ol> <p>This <code class="code">asm</code> takes one input, which is internally popped, and produces two outputs. </p> <div class="example smallexample"> <pre class="example-preformatted" data-language="cpp">asm ("fsincos" : "=t" (cos), "=u" (sin) : "0" (inp));</pre>
</div> <p>This <code class="code">asm</code> takes two inputs, which are popped by the <code class="code">fyl2xp1</code> opcode, and replaces them with one output. The <code class="code">st(1)</code> clobber is necessary for the compiler to know that <code class="code">fyl2xp1</code> pops both inputs. </p> <div class="example smallexample"> <pre class="example-preformatted" data-language="cpp">asm ("fyl2xp1" : "=t" (result) : "0" (x), "u" (y) : "st(1)");</pre>
</div> 
</div> <div class="subsubsection-level-extent" id="MSP430-Operand-Modifiers"> <h1 class="subsubsection"><span>6.47.2.11 MSP430 Operand Modifiers<a class="copiable-link" href="#MSP430-Operand-Modifiers"> ¶</a></span></h1> <p>The list below describes the supported modifiers and their effects for MSP430. </p> <table class="multitable"> <thead><tr>
<th width="10%">Modifier</th>
<th width="90%">Description</th>
</tr></thead> <tbody>
<tr>
<td width="10%"><code class="code">A</code></td>
<td width="90%">Select low 16-bits of the constant/register/memory operand.</td>
</tr> <tr>
<td width="10%"><code class="code">B</code></td>
<td width="90%">Select high 16-bits of the constant/register/memory operand.</td>
</tr> <tr>
<td width="10%"><code class="code">C</code></td>
<td width="90%">Select bits 32-47 of the constant/register/memory operand.</td>
</tr> <tr>
<td width="10%"><code class="code">D</code></td>
<td width="90%">Select bits 48-63 of the constant/register/memory operand.</td>
</tr> <tr>
<td width="10%"><code class="code">H</code></td>
<td width="90%">Equivalent to <code class="code">B</code> (for backwards compatibility).</td>
</tr> <tr>
<td width="10%"><code class="code">I</code></td>
<td width="90%">Print the inverse (logical <code class="code">NOT</code>) of the constant value.</td>
</tr> <tr>
<td width="10%"><code class="code">J</code></td>
<td width="90%">Print an integer without a <code class="code">#</code> prefix.</td>
</tr> <tr>
<td width="10%"><code class="code">L</code></td>
<td width="90%">Equivalent to <code class="code">A</code> (for backwards compatibility).</td>
</tr> <tr>
<td width="10%"><code class="code">O</code></td>
<td width="90%">Offset of the current frame from the top of the stack.</td>
</tr> <tr>
<td width="10%"><code class="code">Q</code></td>
<td width="90%">Use the <code class="code">A</code> instruction postfix.</td>
</tr> <tr>
<td width="10%"><code class="code">R</code></td>
<td width="90%">Inverse of condition code, for unsigned comparisons.</td>
</tr> <tr>
<td width="10%"><code class="code">W</code></td>
<td width="90%">Subtract 16 from the constant value.</td>
</tr> <tr>
<td width="10%"><code class="code">X</code></td>
<td width="90%">Use the <code class="code">X</code> instruction postfix.</td>
</tr> <tr>
<td width="10%"><code class="code">Y</code></td>
<td width="90%">Subtract 4 from the constant value.</td>
</tr> <tr>
<td width="10%"><code class="code">Z</code></td>
<td width="90%">Subtract 1 from the constant value.</td>
</tr> <tr>
<td width="10%"><code class="code">b</code></td>
<td width="90%">Append <code class="code">.B</code>, <code class="code">.W</code> or <code class="code">.A</code> to the instruction, depending on the mode.</td>
</tr> <tr>
<td width="10%"><code class="code">d</code></td>
<td width="90%">Offset 1 byte of a memory reference or constant value.</td>
</tr> <tr>
<td width="10%"><code class="code">e</code></td>
<td width="90%">Offset 3 bytes of a memory reference or constant value.</td>
</tr> <tr>
<td width="10%"><code class="code">f</code></td>
<td width="90%">Offset 5 bytes of a memory reference or constant value.</td>
</tr> <tr>
<td width="10%"><code class="code">g</code></td>
<td width="90%">Offset 7 bytes of a memory reference or constant value.</td>
</tr> <tr>
<td width="10%"><code class="code">p</code></td>
<td width="90%">Print the value of 2, raised to the power of the given constant. Used to select the specified bit position.</td>
</tr> <tr>
<td width="10%"><code class="code">r</code></td>
<td width="90%">Inverse of condition code, for signed comparisons.</td>
</tr> <tr>
<td width="10%"><code class="code">x</code></td>
<td width="90%">Equivialent to <code class="code">X</code>, but only for pointers.</td>
</tr> </tbody> </table> 
</div> <div class="subsubsection-level-extent" id="LoongArch-Operand-Modifiers"> <h1 class="subsubsection"><span>6.47.2.12 LoongArch Operand Modifiers<a class="copiable-link" href="#LoongArch-Operand-Modifiers"> ¶</a></span></h1> <p>The list below describes the supported modifiers and their effects for LoongArch. </p> <table class="multitable"> <thead><tr>
<th width="10%">Modifier</th>
<th width="90%">Description</th>
</tr></thead> <tbody>
<tr>
<td width="10%"><code class="code">d</code></td>
<td width="90%">Same as <code class="code">c</code>.</td>
</tr> <tr>
<td width="10%"><code class="code">i</code></td>
<td width="90%">Print the character ”<code class="code">i</code>” if the operand is not a register.</td>
</tr> <tr>
<td width="10%"><code class="code">m</code></td>
<td width="90%">Same as <code class="code">c</code>, but the printed value is <code class="code">operand - 1</code>.</td>
</tr> <tr>
<td width="10%"><code class="code">X</code></td>
<td width="90%">Print a constant integer operand in hexadecimal.</td>
</tr> <tr>
<td width="10%"><code class="code">z</code></td>
<td width="90%">Print the operand in its unmodified form, followed by a comma.</td>
</tr> </tbody> </table> </div> </div>  <div class="nav-panel"> <p> Next: <a href="constraints">Constraints for <code class="code">asm</code> Operands</a>, Previous: <a href="basic-asm">Basic Asm — Assembler Instructions Without Operands</a>, Up: <a href="using-assembly-language-with-c">How to Use Inline Assembly Language in C Code</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/Extended-Asm.html" class="_attribution-link">https://gcc.gnu.org/onlinedocs/gcc-13.1.0/gcc/Extended-Asm.html</a>
  </p>
</div>