<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 --- - [AArch64] Assertion when lowering LOAD_STACK_GUARD"
href="http://llvm.org/bugs/show_bug.cgi?id=20558">20558</a>
</td>
</tr>
<tr>
<th>Summary</th>
<td>[AArch64] Assertion when lowering LOAD_STACK_GUARD
</td>
</tr>
<tr>
<th>Product</th>
<td>tools
</td>
</tr>
<tr>
<th>Version</th>
<td>trunk
</td>
</tr>
<tr>
<th>Hardware</th>
<td>PC
</td>
</tr>
<tr>
<th>OS</th>
<td>Windows NT
</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>llvmc
</td>
</tr>
<tr>
<th>Assignee</th>
<td>unassignedbugs@nondot.org
</td>
</tr>
<tr>
<th>Reporter</th>
<td>apazos@codeaurora.org
</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=12859" name="attach_12859" title="Fix assertion when lowering LOAD-STACK-GUARD">attachment 12859</a> <a href="attachment.cgi?id=12859&action=edit" title="Fix assertion when lowering LOAD-STACK-GUARD">[details]</a></span>
Fix assertion when lowering LOAD-STACK-GUARD
Hi Akira,
We hit an assertion in expandPostRAPseudo when lowering LOAD_STACK_GUARD.
The call to cast<GlobalValue> asserts with
/llvm/include/llvm/Support/Casting.h:237: typename llvm::cast_retty<X,
Y*>::ret_type llvm::cast(Y*) [with X = llvm::GlobalValue; Y = const
llvm::Value; typename llvm::cast_retty<X, Y*>::ret_type = const
llvm::GlobalValue*]: Assertion `isa<X>(Val) && "cast<Ty>() argument of
incompatible type!"' failed.
The memory operand expected in the exapandPostRAPseudo is a GlobalValue but
instead in the failing test case we have a ConstantExpr.
I tried to get a simplified test case that reproduces the issue. Here it is:
define i32 @test_stack_guard_remat2() {
entry:
%StackGuardSlot = alloca i8*
%StackGuard = load i8** bitcast ([0 x i32]* @__stack_chk_guard to i8**)
call void @llvm.stackprotector(i8* %StackGuard, i8** %StackGuardSlot)
%container = alloca [32 x i8], align 1
call void @llvm.stackprotectorcheck(i8** bitcast ([0 x i32]*
@__stack_chk_guard to i8**))
ret i32 -1
}
I also tried to fix the issue. See the patch below.
--- a/lib/Target/AArch64/AArch64InstrInfo.cpp
+++ b/lib/Target/AArch64/AArch64InstrInfo.cpp
@@ -901,8 +901,12 @@
AArch64InstrInfo::expandPostRAPseudo(MachineBasicBlock::iterator MI) const {
MachineBasicBlock &MBB = *MI->getParent();
DebugLoc DL = MI->getDebugLoc();
unsigned Reg = MI->getOperand(0).getReg();
- const GlobalValue *GV =
- cast<GlobalValue>((*MI->memoperands_begin())->getValue());
+ const GlobalValue *GV = nullptr;
+
+ const Value *Val = (*MI->memoperands_begin())->getValue();
+ if (const ConstantExpr *CE = dyn_cast<const ConstantExpr>(Val))
+ Val = CE->getOperand(0);
+ GV = cast<GlobalValue>(Val);
const TargetMachine &TM = MBB.getParent()->getTarget();
unsigned char OpFlags = Subtarget.ClassifyGlobalReference(GV, TM);
const unsigned char MO_NC = AArch64II::MO_NC;
Let me know what you think.
Thanks,
Ana.</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>