<html>
<head>
<base href="https://bugs.llvm.org/">
</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 - Incorrect remapping of DIArgList for LocalAsMetadata"
href="https://bugs.llvm.org/show_bug.cgi?id=52552">52552</a>
</td>
</tr>
<tr>
<th>Summary</th>
<td>Incorrect remapping of DIArgList for LocalAsMetadata
</td>
</tr>
<tr>
<th>Product</th>
<td>libraries
</td>
</tr>
<tr>
<th>Version</th>
<td>trunk
</td>
</tr>
<tr>
<th>Hardware</th>
<td>PC
</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>Transformation Utilities
</td>
</tr>
<tr>
<th>Assignee</th>
<td>unassignedbugs@nondot.org
</td>
</tr>
<tr>
<th>Reporter</th>
<td>mark.p.mendell@intel.com
</td>
</tr>
<tr>
<th>CC</th>
<td>llvm-bugs@lists.llvm.org
</td>
</tr></table>
<p>
<div>
<pre>Created <span class=""><a href="attachment.cgi?id=25459" name="attach_25459" title="foo.ll - sample file containing dbg.value with DIArgList">attachment 25459</a> <a href="attachment.cgi?id=25459&action=edit" title="foo.ll - sample file containing dbg.value with DIArgList">[details]</a></span>
foo.ll - sample file containing dbg.value with DIArgList
Mapper::mapValue doesn't handle LocalAsMetadata in a DIArgList. It replaces it
with undef, which is not properly written/read by the BitCode handlers.
This can be shown using llvm-link which causes remapping to be done.
foo.ll contains:
call void @llvm.dbg.value(metadata !DIArgList(i32 0, i32 %A), metadata !5,
metadata !DIExpression(DW_OP_LLVM_arg, 0, DW_OP_LLVM_arg, 1, DW_OP_minus,
DW_OP_stack_value)), !dbg !10
If it is linked with foo2.ll (nothing interesting in it), the result is
call void @llvm.dbg.value(metadata !DIArgList(i32 0, i32 undef), metadata !5,
metadata !DIExpression(DW_OP_LLVM_arg, 0, DW_OP_LLVM_arg, 1, DW_OP_minus,
DW_OP_stack_value)), !dbg !10
If you then try running llvm-dis on the resulting BC file, it will crash
llvm-link -o result.bc foo.ll foo2.ll
llvm-dis result.bc # boom
The contents of foo.ll debug information is mostly garbage, just left over from
a cut down failing test case.
Tested fix:
ValueMapper.cpp ====
***************
*** 400,412 ****
// be mapped via mapValue (apart from constants when we have no
// module level changes, which have an identity mapping).
if ((Flags & RF_NoModuleLevelChanges) &&
isa<ConstantAsMetadata>(VAM)) {
MappedArgs.push_back(VAM);
} else if (Value *LV = mapValue(VAM->getValue())) {
MappedArgs.push_back(
! LV == VAM->getValue() ? VAM : ValueAsMetadata::get(LV));
} else {
// If we cannot map the value, set the argument as undef.
MappedArgs.push_back(ValueAsMetadata::get(
UndefValue::get(VAM->getValue()->getType())));
}
}
--- 400,414 ----
// be mapped via mapValue (apart from constants when we have no
// module level changes, which have an identity mapping).
if ((Flags & RF_NoModuleLevelChanges) &&
isa<ConstantAsMetadata>(VAM)) {
MappedArgs.push_back(VAM);
} else if (Value *LV = mapValue(VAM->getValue())) {
MappedArgs.push_back(
! LV == VAM->getValue() ? VAM : ValueAsMetadata::get(LV));
! } else if (isa<LocalAsMetadata>(VAM)) {
! MappedArgs.push_back(VAM);
} else {
// If we cannot map the value, set the argument as undef.
MappedArgs.push_back(ValueAsMetadata::get(
UndefValue::get(VAM->getValue()->getType())));
}
}</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>