Cold-Air-Networks/mirror-llvm-project
2
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions 7

[llubi] Truncate struct field offsets exceeding the pointer index width (#213471)

Browse source 
Update computeGEP to implicitly truncate struct field offsets. This
fixes crashes in the new test, where the field offset does not fit in
the index width.

Related https://github.com/llvm/llvm-project/pull/213436

PR: https://github.com/llvm/llvm-project/pull/213471
This commit is contained in:
Florian Hahn 2026-08-01 21:10:45 +02:00 committed by GitHub
parent 4671437fb0
commit 8cb8e872e7
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
2 changed files with 20 additions and 1 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

18
llvm/test/tools/llubi/gep-16-bit-addrspace.ll Normal file
View file

@ -0,0 +1,18 @@
; NOTE: Assertions have been autogenerated by utils/update_llubi_test_checks.py UTC_ARGS: --version 6
; RUN: llubi --verbose < %s 2>&1 | FileCheck %s
target datalayout = "e-m:e-p:16:16"
%s = type { [65536 x i8], i8 }
; Offset of the struct's second field does not fit into a 16 bit pointer.
define void @main() {
%p = alloca i32
%a = getelementptr %s, ptr %p, i32 0, i32 1
ret void
}
; CHECK: Entering function: main
; CHECK-NEXT: %p = alloca i32, align 4 => ptr 0x8 [p]
; CHECK-NEXT: %a = getelementptr %s, ptr %p, i32 0, i32 1 => ptr 0x8 [p]
; CHECK-NEXT: ret void
; CHECK-NEXT: Exiting function: main

3
llvm/tools/llubi/lib/Context.cpp
View file

@ -1054,7 +1054,8 @@ Context::computeGEP(GEPOperator &GEP,
unsigned ElementIdx = cast<ConstantInt>(V)->getZExtValue();
const StructLayout *SL = DL.getStructLayout(STy);
// Element offset is in bytes.
ApplyScaledOffset(APInt(IndexBitWidth, SL->getElementOffset(ElementIdx)),
ApplyScaledOffset(APInt(IndexBitWidth, SL->getElementOffset(ElementIdx),
/*isSigned=*/false, /*implicitTrunc=*/true),
APInt(IndexBitWidth, 1));
continue;
}