<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 - X86 ISel gets stuck in infinite loop"
href="https://bugs.llvm.org/show_bug.cgi?id=40793">40793</a>
</td>
</tr>
<tr>
<th>Summary</th>
<td>X86 ISel gets stuck in infinite loop
</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>Backend: X86
</td>
</tr>
<tr>
<th>Assignee</th>
<td>unassignedbugs@nondot.org
</td>
</tr>
<tr>
<th>Reporter</th>
<td>andrew@scheidecker.net
</td>
</tr>
<tr>
<th>CC</th>
<td>craig.topper@gmail.com, llvm-bugs@lists.llvm.org, llvm-dev@redking.me.uk, spatel+llvm@rotateright.com
</td>
</tr></table>
<p>
<div>
<pre>Created <span class=""><a href="attachment.cgi?id=21498" name="attach_21498" title="IR that triggers bug">attachment 21498</a> <a href="attachment.cgi?id=21498&action=edit" title="IR that triggers bug">[details]</a></span>
IR that triggers bug
Using SVN revision 354401 (from 2019/02/19), llc gets stuck in a loop on the
attached input, when run with the command-line:
llc llvm-lock.ll -mcpu=skylake-avx512
Using llc from the LLVM 6.0 release does not get stuck in a loop.
Running it with e.g. -mcpu=skylake also does not get stuck in a loop. Based on
this, I've assumed this is an X86/AVX512 target specific bug.
Passing -debug prints the following in a loop:
Promoting t1036: i16 = and t46, Constant:i16<15>
Creating new node: t1038: i32 = any_extend t46
Creating constant: t1039: i32 = Constant<15>
Creating new node: t1040: i32 = and t1038, Constant:i32<15>
Creating new node: t1041: i16 = truncate t1040
Replacing.1 t1036: i16 = and t46, Constant:i16<15>
With: t1041: i16 = truncate t1040
and 0 other values
Combining: t1037: i32 = srl Constant:i32<32733>, t1041
Creating new node: t1042: i16 = and t46, Constant:i16<15>
Creating new node: t1043: i32 = srl Constant:i32<32733>, t1042
... into: t1043: i32 = srl Constant:i32<32733>, t1042
Combining: t39: i32 = Constant<32733>
Combining: t43: i64 = any_extend t1043
Combining: t1043: i32 = srl Constant:i32<32733>, t1042
Combining: t1042: i16 = and t46, Constant:i16<15>
Breaking into the debugger shows this call stack:
`anonymous namespace'::DAGCombiner::CombineTo
`anonymous namespace'::DAGCombiner::CombineTo
`anonymous namespace'::DAGCombiner::PromoteIntBinOp
`anonymous namespace'::DAGCombiner::combine
`anonymous namespace'::DAGCombiner::Run
llvm::SelectionDAG::Combine
llvm::SelectionDAGISel::CodeGenAndEmitDAG
llvm::SelectionDAGISel::SelectBasicBlock
llvm::SelectionDAGISel::SelectAllBasicBlocks
llvm::SelectionDAGISel::runOnMachineFunction
`anonymous namespace'::X86DAGToDAGISel::runOnMachineFunction
llvm::MachineFunctionPass::runOnFunction
llvm::FPPassManager::runOnFunction
llvm::FPPassManager::runOnModule</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>