mingw/info/stabs/Negative-Type-Numbers.html

   1 <html lang="en">
   2 <head>
   3 <title>STABS</title>
   4 <meta http-equiv="Content-Type" content="text/html">
   5 <meta name="description" content="STABS">
   6 <meta name="generator" content="makeinfo 4.3">
   7 <link href="http://www.gnu.org/software/texinfo/" rel="generator-home">
   8 </head>
   9 <body>
  10 <div class="node">
  11 <p>
  12 Node:<a name="Negative%20Type%20Numbers">Negative Type Numbers</a>,
  13 Previous:<a rel="previous" accesskey="p" href="Builtin-Type-Descriptors.html#Builtin%20Type%20Descriptors">Builtin Type Descriptors</a>,
  14 Up:<a rel="up" accesskey="u" href="Builtin-Types.html#Builtin%20Types">Builtin Types</a>
  15 <hr><br>
  16 </div>
  17
  18 <h4 class="subsection">Negative Type Numbers</h4>
  19
  20    <p>This is the method used in XCOFF for defining builtin types.
  21 Since the debugger knows about the builtin types anyway, the idea of
  22 negative type numbers is simply to give a special type number which
  23 indicates the builtin type.  There is no stab defining these types.
  24
  25    <p>There are several subtle issues with negative type numbers.
  26
  27    <p>One is the size of the type.  A builtin type (for example the C types
  28 <code>int</code> or <code>long</code>) might have different sizes depending on
  29 compiler options, the target architecture, the ABI, etc.  This issue
  30 doesn't come up for IBM tools since (so far) they just target the
  31 RS/6000; the sizes indicated below for each size are what the IBM
  32 RS/6000 tools use.  To deal with differing sizes, either define separate
  33 negative type numbers for each size (which works but requires changing
  34 the debugger, and, unless you get both AIX dbx and GDB to accept the
  35 change, introduces an incompatibility), or use a type attribute
  36 (see <a href="String-Field.html#String%20Field">String Field</a>) to define a new type with the appropriate size
  37 (which merely requires a debugger which understands type attributes,
  38 like AIX dbx or GDB).  For example,
  39
  40 <pre class="example">     .stabs "boolean:t10=@s8;-16",128,0,0,0
  41      </pre>
  42
  43    <p>defines an 8-bit boolean type, and
  44
  45 <pre class="example">     .stabs "boolean:t10=@s64;-16",128,0,0,0
  46      </pre>
  47
  48    <p>defines a 64-bit boolean type.
  49
  50    <p>A similar issue is the format of the type.  This comes up most often for
  51 floating-point types, which could have various formats (particularly
  52 extended doubles, which vary quite a bit even among IEEE systems).
  53 Again, it is best to define a new negative type number for each
  54 different format; changing the format based on the target system has
  55 various problems.  One such problem is that the Alpha has both VAX and
  56 IEEE floating types.  One can easily imagine one library using the VAX
  57 types and another library in the same executable using the IEEE types.
  58 Another example is that the interpretation of whether a boolean is true
  59 or false can be based on the least significant bit, most significant
  60 bit, whether it is zero, etc., and different compilers (or different
  61 options to the same compiler) might provide different kinds of boolean.
  62
  63    <p>The last major issue is the names of the types.  The name of a given
  64 type depends <em>only</em> on the negative type number given; these do not
  65 vary depending on the language, the target system, or anything else.
  66 One can always define separate type numbers--in the following list you
  67 will see for example separate <code>int</code> and <code>integer*4</code> types
  68 which are identical except for the name.  But compatibility can be
  69 maintained by not inventing new negative type numbers and instead just
  70 defining a new type with a new name.  For example:
  71
  72 <pre class="example">     .stabs "CARDINAL:t10=-8",128,0,0,0
  73      </pre>
  74
  75    <p>Here is the list of negative type numbers.  The phrase <dfn>integral
  76 type</dfn> is used to mean twos-complement (I strongly suspect that all
  77 machines which use stabs use twos-complement; most machines use
  78 twos-complement these days).
  79
  80      <dl>
  81 <dt><code>-1</code>
  82      <dd><code>int</code>, 32 bit signed integral type.
  83
  84      <br><dt><code>-2</code>
  85      <dd><code>char</code>, 8 bit type holding a character.   Both GDB and dbx on AIX
  86 treat this as signed.  GCC uses this type whether <code>char</code> is signed
  87 or not, which seems like a bad idea.  The AIX compiler (<code>xlc</code>) seems to
  88 avoid this type; it uses -5 instead for <code>char</code>.
  89
  90      <br><dt><code>-3</code>
  91      <dd><code>short</code>, 16 bit signed integral type.
  92
  93      <br><dt><code>-4</code>
  94      <dd><code>long</code>, 32 bit signed integral type.
  95
  96      <br><dt><code>-5</code>
  97      <dd><code>unsigned char</code>, 8 bit unsigned integral type.
  98
  99      <br><dt><code>-6</code>
 100      <dd><code>signed char</code>, 8 bit signed integral type.
 101
 102      <br><dt><code>-7</code>
 103      <dd><code>unsigned short</code>, 16 bit unsigned integral type.
 104
 105      <br><dt><code>-8</code>
 106      <dd><code>unsigned int</code>, 32 bit unsigned integral type.
 107
 108      <br><dt><code>-9</code>
 109      <dd><code>unsigned</code>, 32 bit unsigned integral type.
 110
 111      <br><dt><code>-10</code>
 112      <dd><code>unsigned long</code>, 32 bit unsigned integral type.
 113
 114      <br><dt><code>-11</code>
 115      <dd><code>void</code>, type indicating the lack of a value.
 116
 117      <br><dt><code>-12</code>
 118      <dd><code>float</code>, IEEE single precision.
 119
 120      <br><dt><code>-13</code>
 121      <dd><code>double</code>, IEEE double precision.
 122
 123      <br><dt><code>-14</code>
 124      <dd><code>long double</code>, IEEE double precision.  The compiler claims the size
 125 will increase in a future release, and for binary compatibility you have
 126 to avoid using <code>long double</code>.  I hope when they increase it they
 127 use a new negative type number.
 128
 129      <br><dt><code>-15</code>
 130      <dd><code>integer</code>.  32 bit signed integral type.
 131
 132      <br><dt><code>-16</code>
 133      <dd><code>boolean</code>.  32 bit type.  GDB and GCC assume that zero is false,
 134 one is true, and other values have unspecified meaning.  I hope this
 135 agrees with how the IBM tools use the type.
 136
 137      <br><dt><code>-17</code>
 138      <dd><code>short real</code>.  IEEE single precision.
 139
 140      <br><dt><code>-18</code>
 141      <dd><code>real</code>.  IEEE double precision.
 142
 143      <br><dt><code>-19</code>
 144      <dd><code>stringptr</code>.  See <a href="Strings.html#Strings">Strings</a>.
 145
 146      <br><dt><code>-20</code>
 147      <dd><code>character</code>, 8 bit unsigned character type.
 148
 149      <br><dt><code>-21</code>
 150      <dd><code>logical*1</code>, 8 bit type.  This Fortran type has a split
 151 personality in that it is used for boolean variables, but can also be
 152 used for unsigned integers.  0 is false, 1 is true, and other values are
 153 non-boolean.
 154
 155      <br><dt><code>-22</code>
 156      <dd><code>logical*2</code>, 16 bit type.  This Fortran type has a split
 157 personality in that it is used for boolean variables, but can also be
 158 used for unsigned integers.  0 is false, 1 is true, and other values are
 159 non-boolean.
 160
 161      <br><dt><code>-23</code>
 162      <dd><code>logical*4</code>, 32 bit type.  This Fortran type has a split
 163 personality in that it is used for boolean variables, but can also be
 164 used for unsigned integers.  0 is false, 1 is true, and other values are
 165 non-boolean.
 166
 167      <br><dt><code>-24</code>
 168      <dd><code>logical</code>, 32 bit type.  This Fortran type has a split
 169 personality in that it is used for boolean variables, but can also be
 170 used for unsigned integers.  0 is false, 1 is true, and other values are
 171 non-boolean.
 172
 173      <br><dt><code>-25</code>
 174      <dd><code>complex</code>.  A complex type consisting of two IEEE single-precision
 175 floating point values.
 176
 177      <br><dt><code>-26</code>
 178      <dd><code>complex</code>.  A complex type consisting of two IEEE double-precision
 179 floating point values.
 180
 181      <br><dt><code>-27</code>
 182      <dd><code>integer*1</code>, 8 bit signed integral type.
 183
 184      <br><dt><code>-28</code>
 185      <dd><code>integer*2</code>, 16 bit signed integral type.
 186
 187      <br><dt><code>-29</code>
 188      <dd><code>integer*4</code>, 32 bit signed integral type.
 189
 190      <br><dt><code>-30</code>
 191      <dd><code>wchar</code>.  Wide character, 16 bits wide, unsigned (what format?
 192 Unicode?).
 193
 194      <br><dt><code>-31</code>
 195      <dd><code>long long</code>, 64 bit signed integral type.
 196
 197      <br><dt><code>-32</code>
 198      <dd><code>unsigned long long</code>, 64 bit unsigned integral type.
 199
 200      <br><dt><code>-33</code>
 201      <dd><code>logical*8</code>, 64 bit unsigned integral type.
 202
 203      <br><dt><code>-34</code>
 204      <dd><code>integer*8</code>, 64 bit signed integral type.
 205 </dl>
 206
 207    </body></html>
 208