[prev in list] [next in list] [prev in thread] [next in thread]
List: llvm-bugs
Subject: [LLVMbugs] [Bug 19909] New: va_arg fails with __int128
From: bugzilla-daemon () llvm ! org
Date: 2014-05-31 22:58:27
Message-ID: bug-19909-206 () http ! llvm ! org/bugs/
[Download RAW message or body]
--1401577108.3eecF10.4210
Date: Sat, 31 May 2014 17:58:28 -0500
MIME-Version: 1.0
Content-Type: text/plain; charset="UTF-8"
http://llvm.org/bugs/show_bug.cgi?id=19909
Bug ID: 19909
Summary: va_arg fails with __int128
Product: new-bugs
Version: trunk
Hardware: All
OS: All
Status: NEW
Severity: normal
Priority: P
Component: new bugs
Assignee: unassignedbugs@nondot.org
Reporter: nenad@intrepid.com
CC: llvmbugs@cs.uiuc.edu
Classification: Unclassified
Created attachment 12583
--> http://llvm.org/bugs/attachment.cgi?id=12583&action=edit
Demonstrate __int128 va_arg failure
The attached example fails on Clang/LLVM trunk (or 3.4 or Apple 3.4) and passes
when compiled with gcc 4.8.2. The code simple tries to pass multiple __int128
after two ints and a pointer. I debugged it a bit and it seems that the
"overflow_arg_area" pointer in va_list is wrong by 8 bytes, which might
indicate some alignment problem.
I also checked how __int128 is passed to the function. It seems that Clang is
capable of splitting the __int128 where lower part is passed in a register and
upper part on the stack. GCC does not split __int128 if it does not have 2
registers available. This might explain the wrong 8 bytes offset for the
"overflow_arg_area" as va_start does not adjust for the split register.
--
You are receiving this mail because:
You are on the CC list for the bug.
--1401577108.3eecF10.4210
Date: Sat, 31 May 2014 17:58:28 -0500
MIME-Version: 1.0
Content-Type: text/html; charset="UTF-8"
<html>
<head>
<base href="http://llvm.org/bugs/" />
</head>
<body><table border="1" cellspacing="0" cellpadding="8">
<tr>
<th>Bug ID</th>
<td><a class="bz_bug_link
bz_status_NEW "
title="NEW --- - va_arg fails with __int128"
href="http://llvm.org/bugs/show_bug.cgi?id=19909">19909</a>
</td>
</tr>
<tr>
<th>Summary</th>
<td>va_arg fails with __int128
</td>
</tr>
<tr>
<th>Product</th>
<td>new-bugs
</td>
</tr>
<tr>
<th>Version</th>
<td>trunk
</td>
</tr>
<tr>
<th>Hardware</th>
<td>All
</td>
</tr>
<tr>
<th>OS</th>
<td>All
</td>
</tr>
<tr>
<th>Status</th>
<td>NEW
</td>
</tr>
<tr>
<th>Severity</th>
<td>normal
</td>
</tr>
<tr>
<th>Priority</th>
<td>P
</td>
</tr>
<tr>
<th>Component</th>
<td>new bugs
</td>
</tr>
<tr>
<th>Assignee</th>
<td>unassignedbugs@nondot.org
</td>
</tr>
<tr>
<th>Reporter</th>
<td>nenad@intrepid.com
</td>
</tr>
<tr>
<th>CC</th>
<td>llvmbugs@cs.uiuc.edu
</td>
</tr>
<tr>
<th>Classification</th>
<td>Unclassified
</td>
</tr></table>
<p>
<div>
<pre>Created <span class=""><a href="attachment.cgi?id=12583" \
name="attach_12583" title="Demonstrate __int128 va_arg failure">attachment 12583</a> \
<a href="attachment.cgi?id=12583&action=edit" title="Demonstrate __int128 va_arg \
failure">[details]</a></span> Demonstrate __int128 va_arg failure
The attached example fails on Clang/LLVM trunk (or 3.4 or Apple 3.4) and passes
when compiled with gcc 4.8.2. The code simple tries to pass multiple __int128
after two ints and a pointer. I debugged it a bit and it seems that the
"overflow_arg_area" pointer in va_list is wrong by 8 bytes, which might
indicate some alignment problem.
I also checked how __int128 is passed to the function. It seems that Clang is
capable of splitting the __int128 where lower part is passed in a register and
upper part on the stack. GCC does not split __int128 if it does not have 2
registers available. This might explain the wrong 8 bytes offset for the
"overflow_arg_area" as va_start does not adjust for the split \
register.</pre> </div>
</p>
<hr>
<span>You are receiving this mail because:</span>
<ul>
<li>You are on the CC list for the bug.</li>
</ul>
</body>
</html>
--1401577108.3eecF10.4210--
_______________________________________________
LLVMbugs mailing list
LLVMbugs@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvmbugs
[prev in list] [next in list] [prev in thread] [next in thread]
Configure |
About |
News |
Add a list |
Sponsored by KoreLogic