https://bugs.llvm.org/show_bug.cgi?id=45302
Bug ID: 45302
Summary: llc: ../lib/CodeGen/IfConversion.cpp:2056: [...]
Assertion `(TailBB || !TrueBBI.IsBrAnalyzable) &&
"Unexpected!"' failed.
Product: libraries
Version: trunk
Hardware: PC
OS: Linux
Status: NEW
Severity: enhancement
Priority: P
Component: Common Code Generator Code
Assignee: unassignedb...@nondot.org
Reporter: mikael.hol...@ericsson.com
CC: llvm-bugs@lists.llvm.org
Created attachment 23278
--> https://bugs.llvm.org/attachment.cgi?id=23278&action=edit
bbi-40556_arm.mir reproducer.
Reproduce with:
llc -mtriple=thumbv7-apple-ios -o /dev/null bbi-40556_arm.mir -run-pass
if-converter
which results in
llc: ../lib/CodeGen/IfConversion.cpp:2056: bool (anonymous
namespace)::IfConverter::IfConvertDiamond((anonymous
namespace)::IfConverter::BBInfo &, (anonymous
namespace)::IfConverter::IfcvtKind, unsigned int, unsigned int, bool, bool):
Assertion `(TailBB || !TrueBBI.IsBrAnalyzable) && "Unexpected!"' failed.
The reproducer is quite small:
--- |
define void @func() minsize {
ret void
}
...
---
name: func
body: |
bb.0:
tBcc %bb.3, 1, $cpsr
bb.1:
tBcc %bb.2, 1, $cpsr
tB %bb.2, 14, $noreg
bb.2:
bb.3:
successors:
tBX_RET 14, _
...
I assume the interesting part is the two branches from bb.1 to bb.2.
This started failing with 4f454b22754:
[IfCvt][ARM] Optimise diamond if-conversion for code size
If revert that commit the crash goes away.
--
You are receiving this mail because:
You are on the CC list for the bug._______________________________________________
llvm-bugs mailing list
llvm-bugs@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs
