This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [wwwdocs] Document changes in 3.4 for SPARC (rev2)
> I just have few and very minor comments; please consider them and then
> commit an updated patch.
Thanks for the review.
> Index: gcc-3.4/changes.html
> ===================================================================
> + <li>A small structure is passed or returned in a register; and</li>
> + <li>it contains a unique field of type 'float' .</li>
>
> <code>float</code>?
Done.
> + <p><code>x</code> is passed in floating-point register <code>%f0</code>,
> + instead of <code>%f1</code> previously.</p></td>
>
> I'd omit "previously" in this case; and I believe there are one or two
> instances later on as well.
All occurences removed.
> + <li>A small union is passed or returned in a register; and</li>
> + <li>its size in bytes is lesser than 8.</li>
>
> "less"
Sure? Is "lesser than" not the standard reading for "<"?
I've also added a few lines to the "Caveats" section, right after those added
for the ABI changes on MIPS.
> And I do think this really ought to be an entry in the news section of
> our main page.
I don't see such an entry for MIPS. So my egalitarian French mind :-) is
very reluctant to adding one for SPARC.
--
Eric Botcazou
Index: gcc-3.4/changes.html
===================================================================
RCS file: /cvs/gcc/wwwdocs/htdocs/gcc-3.4/changes.html,v
retrieving revision 1.92
diff -u -r1.92 changes.html
--- gcc-3.4/changes.html 28 Jan 2004 22:50:11 -0000 1.92
+++ gcc-3.4/changes.html 2 Feb 2004 21:36:00 -0000
@@ -35,6 +35,9 @@
<li>The implementation of the <a href="mips-abi.html">MIPS ABIs has
changed</a>. As a result, the code generated for certain MIPS
targets will not be binary compatible with earlier releases.</li>
+ <li>The implementation of the <a href="sparc-abi.html">SPARC ABIs has
+ changed</a>. As a result, the code generated will not be binary
+ compatible with earlier releases in certain cases.</li>
<li>The configure option <code>--enable-threads=pthreads</code> has
been removed; use <code>--enable-threads=posix</code> instead,
which should have the same effect.</li>
@@ -635,6 +638,18 @@
<li>Libgcj is enabled by default. On systems older than <code>
powerpc-apple-darwin7.0.0</code> you need to install <a
href="http://www.opendarwin.org/projects/dlcompat";> dlcompat</a>. </li>
+ </ul>
+
+<h3>SPARC</h3>
+ <ul>
+ <li>The option <code>-mflat</code> is deprecated.</li>
+
+ <li>Support for large (> 2GB) frames has been added to the
+ 64-bit port.</li>
+
+ <li>Several <a href="sparc-abi.html">ABI bugs</a> have been fixed.
+ Unfortunately, these changes will break binary compatibility with
+ earlier releases.</li>
</ul>
<h3>SuperH</h3>
--- /dev/null Thu Apr 11 16:25:15 2002
+++ gcc-3.4/sparc-abi.html Mon Feb 2 22:33:40 2004
@@ -0,0 +1,390 @@
+<html>
+
+<head>
+<title>SPARC ABI Changes in GCC 3.4</title>
+</head>
+
+<body>
+
+<h1>SPARC ABI Changes in GCC 3.4</h1>
+
+<p>GCC 3.4 fixes several cases in which earlier releases would not
+ follow the SPARC calling conventions. This document describes
+ each fix and the kind of code it will affect. In each case,
+ GCC 3.4 will not be binary compatible with earlier releases.</p>
+
+<p>Most of the fixes are related to the handling of small structure
+ and union types in 64-bit mode.</p>
+
+<h2>A. Small structure arguments and return values (1)</h2>
+
+<table cellpadding="4">
+ <tr valign="top">
+ <th align="right">Affected ABI</th>
+ <td>64-bit</td>
+ </tr>
+
+ <tr valign="top">
+ <th align="right">Conditions</th>
+ <td><ul>
+ <li>A small structure is passed or returned in a register; and</li>
+ <li>it contains a unique field of type <code>float</code> .</li>
+ </ul></td>
+ </tr>
+
+ <tr valign="top">
+ <th align="right">Old behavior</th>
+ <td>The register was odd-numbered.</td>
+ </tr>
+
+ <tr valign="top">
+ <th align="right">New behavior</th>
+ <td>The register is even-numbered.</td>
+ </tr>
+
+ <tr valign="top">
+ <th align="right">Example</th>
+ <td><pre>struct s { float f; };
+void g (struct s x);</pre>
+
+ <p><code>x</code> is passed in floating-point register <code>%f0</code>,
+ instead of <code>%f1</code>.</p></td>
+ </tr>
+</table>
+
+<h2>B. Small structure arguments and return values (2)</h2>
+
+<table cellpadding="4">
+ <tr valign="top">
+ <th align="right">Affected ABI</th>
+ <td>64-bit</td>
+ </tr>
+
+ <tr valign="top">
+ <th align="right">Conditions</th>
+ <td><ul>
+ <li>A small structure is passed or returned in registers;</li>
+ <li>its size in bytes is not a multiple of 8 and is greater than 8;</li>
+ <li>its last field is of integral type; and</li>
+ <li>its last but one field is smaller than a word.</li>
+ </ul></td>
+ </tr>
+
+ <tr valign="top">
+ <th align="right">Old behavior</th>
+ <td>The last used register was padded at the most significant end.</td>
+ </tr>
+
+ <tr valign="top">
+ <th align="right">New behavior</th>
+ <td>The last used register is padded at the least significant end.</td>
+ </tr>
+
+ <tr valign="top">
+ <th align="right">Example</th>
+ <td><pre>struct s { float f; int i; int j; };
+void g (struct s x);</pre>
+
+ <p><code>x</code> is passed in several registers, which are laid
+ out as follows:</p>
+
+ <table cellpadding="4">
+ <tr align="center">
+ <th></th>
+ <th>%f0</th> <th>%o0 (high)</th> <th>%o0 (low)</th>
+ <th>%o1 (high)</th> <th>%o1 (low)</th>
+ </tr>
+
+ <tr align="center">
+ <th align="left">Old behavior</th>
+ <td>f</td> <td>padding</td> <td>i</td>
+ <td>padding</td> <td>j</td>
+ </tr>
+
+ <tr align="center">
+ <th align="left">New behavior</th>
+ <td>f</td> <td>padding</td> <td>i</td>
+ <td>j</td> <td>padding</td>
+ </tr>
+ </table></td>
+ </tr>
+</table>
+
+<h2>C. Small unions arguments and return values</h2>
+
+<table cellpadding="4">
+ <tr valign="top">
+ <th align="right">Affected ABI</th>
+ <td>64-bit</td>
+ </tr>
+
+ <tr valign="top">
+ <th align="right">Conditions</th>
+ <td><ul>
+ <li>A small union is passed or returned in a register; and</li>
+ <li>its size in bytes is lesser than 8.</li>
+ </ul></td>
+ </tr>
+
+ <tr valign="top">
+ <th align="right">Old behavior</th>
+ <td>The register was padded at the most significant end.</td>
+ </tr>
+
+ <tr valign="top">
+ <th align="right">New behavior</th>
+ <td>The register is padded at the least significant end.</td>
+ </tr>
+
+ <tr valign="top">
+ <th align="right">Example</th>
+ <td><pre>union u { int i; float f; };
+void g (union u x);</pre>
+
+ <p><code>x</code> is passed in register <code>%o0</code>, which is
+ laid out as follows:</p>
+
+ <table cellpadding="4">
+ <tr align="center">
+ <th></th>
+ <th>%o0 (high)</th> <th>%o0 (low)</th>
+ </tr>
+
+ <tr align="center">
+ <th align="left">Old behavior</th>
+ <td>padding</td> <td>i/f</td>
+ </tr>
+
+ <tr align="center">
+ <th align="left">New behavior</th>
+ <td>i/f</td> <td>padding</td>
+ </tr>
+ </table></td>
+ </tr>
+</table>
+
+<h2>D. Small structure arguments (1)</h2>
+
+<table cellpadding="4">
+ <tr valign="top">
+ <th align="right">Affected ABI</th>
+ <td>64-bit</td>
+ </tr>
+
+ <tr valign="top">
+ <th align="right">Conditions</th>
+ <td><ul>
+ <li>A small structure is passed past the 6th argument slot and
+ prior to the last one; and</li>
+ <li>it contains a complex floating-point field.</li>
+ </ul></td>
+ </tr>
+
+ <tr valign="top">
+ <th align="right">Old behavior</th>
+ <td>The complex floating-point field was passed in memory.</td>
+ </tr>
+
+ <tr valign="top">
+ <th align="right">New behavior</th>
+ <td>The complex floating-point field is passed in registers.</td>
+ </tr>
+
+ <tr valign="top">
+ <th align="right">Example</th>
+ <td><pre>struct s { _Complex float cf; };
+void g (struct s x1, struct s x2, struct s x3, struct s x4, struct s x5, struct s x6, struct s x7);</pre>
+
+ <p><code>x7</code> is passed in floating-point registers <code>%f12-%f13</code>,
+ instead of in memory.</p></td>
+ </tr>
+</table>
+
+<h2>E. Small structure arguments (2)</h2>
+
+<table cellpadding="4">
+ <tr valign="top">
+ <th align="right">Affected ABI</th>
+ <td>64-bit</td>
+ </tr>
+
+ <tr valign="top">
+ <th align="right">Conditions</th>
+ <td><ul>
+ <li>A small structure is passed past the 6th argument slot and
+ prior to the last one;</li>
+ <li>it contains a nested structure; and</li>
+ <li>the nested structure contains a floating-point field.</li>
+ </ul></td>
+ </tr>
+
+ <tr valign="top">
+ <th align="right">Old behavior</th>
+ <td>The floating-point field was passed in memory.</td>
+ </tr>
+
+ <tr valign="top">
+ <th align="right">New behavior</th>
+ <td>The floating-point field is passed in registers.</td>
+ </tr>
+
+ <tr valign="top">
+ <th align="right">Example</th>
+ <td><pre>struct s { struct { double d; } ns; };
+void g (struct s x1, struct s x2, struct s x3, struct s x4, struct s x5, struct s x6, struct s x7);</pre>
+
+ <p><code>x7</code> is passed in floating-point registers <code>%f12-%f13</code>,
+ instead of in memory.</p></td>
+ </tr>
+</table>
+
+<h2>F. Complex floating-point arguments and return values</h2>
+
+<table cellpadding="4">
+ <tr valign="top">
+ <th align="right">Affected ABI</th>
+ <td>32-bit</td>
+ </tr>
+
+ <tr valign="top">
+ <th align="right">Conditions</th>
+ <td>A complex floating-point value is passed to or returned from a
+ function.</td>
+ </tr>
+
+ <tr valign="top">
+ <th align="right">Old behavior</th>
+ <td><p>The complex floating-point value was passed or returned according
+ to the following table:</p>
+
+ <table cellpadding="4">
+ <tr align="center">
+ <th></th>
+ <th>argument</th> <th>return value</th>
+ </tr>
+
+ <tr align="center">
+ <th align="left">_Complex float</th>
+ <td>integer registers</td> <td>integer registers</td>
+ </tr>
+
+ <tr align="center">
+ <th align="left">_Complex double</th>
+ <td>integer registers</td> <td>integer registers</td>
+ </tr>
+
+ <tr align="center">
+ <th align="left">_Complex long double</th>
+ <td>memory</td> <td>memory</td>
+ </tr>
+ </table></td>
+ </tr>
+
+ <tr valign="top">
+ <th align="right">New behavior</th>
+ <td><p>The complex floating-point value is passed or returned according
+ to the following table:</p>
+
+ <table cellpadding="4">
+ <tr align="center">
+ <th></th>
+ <th>argument</th> <th>return value</th>
+ </tr>
+
+ <tr align="center">
+ <th align="left">_Complex float</th>
+ <td>memory</td> <td>floating-point registers</td>
+ </tr>
+
+ <tr align="center">
+ <th align="left">_Complex double</th>
+ <td>memory</td> <td>floating-point registers</td>
+ </tr>
+
+ <tr align="center">
+ <th align="left">_Complex long double</th>
+ <td>memory</td> <td>floating-point registers</td>
+ </tr>
+ </table></td>
+ </tr>
+
+ <tr valign="top">
+ <th align="right">Example</th>
+ <td><pre>_Complex float g (void);</pre>
+
+ <p>The return value is returned in floating-point registers <code>%f0-%f1</code>,
+ instead of registers <code>%o0-%o1</code>.</p></td>
+ </tr>
+</table>
+
+<h2>G. TI mode integral arguments (GCC extension)</h2>
+
+<table cellpadding="4">
+ <tr valign="top">
+ <th align="right">Affected ABI</th>
+ <td>64-bit</td>
+ </tr>
+
+ <tr valign="top">
+ <th align="right">Conditions</th>
+ <td>A TI mode integral value is passed to a function.</td>
+ </tr>
+
+ <tr valign="top">
+ <th align="right">Old behavior</th>
+ <td>The TI mode integral value was not aligned on a 16-byte boundary in
+ the parameter array.</td>
+ </tr>
+
+ <tr valign="top">
+ <th align="right">New behavior</th>
+ <td>The TI mode integral value is aligned on a 16-byte boundary in
+ the parameter array.</td>
+ </tr>
+
+ <tr valign="top">
+ <th align="right">Example</th>
+ <td><pre>typedef int TItype __attribute__ ((mode (TI)));
+void g (int x1, TItype x2);</pre>
+
+ <p><code>x2</code> is passed in registers <code>%o2-%o3</code>, instead
+ of <code>%o1-%o2</code>.</p></td>
+ </tr>
+</table>
+
+<h2>H. Complex integral arguments (GCC extension)</h2>
+
+<table cellpadding="4">
+ <tr valign="top">
+ <th align="right">Affected ABI</th>
+ <td>64-bit</td>
+ </tr>
+
+ <tr valign="top">
+ <th align="right">Conditions</th>
+ <td>A complex integral value is passed in registers.</td>
+ </tr>
+
+ <tr valign="top">
+ <th align="right">Old behavior</th>
+ <td>Two consecutive registers were reserved, regardless of the size of
+ the complex integral value.</td>
+ </tr>
+
+ <tr valign="top">
+ <th align="right">New behavior</th>
+ <td>Only one register is reserved if the complex integral value can fit
+ in a single register.</td>
+ </tr>
+
+ <tr valign="top">
+ <th align="right">Example</th>
+ <td><pre>void g (_Complex int x1, _Complex int x2);</pre>
+
+ <p><code>x2</code> is passed in register <code>%o1</code>, instead
+ of <code>%o2</code>.</p></td>
+ </tr>
+</table>
+
+</body>
+</html>