[llvm-bugs] [Bug 122602] [LoopVectorize] Miscompilation
Issue
122602
Summary
[LoopVectorize] Miscompilation
Labels
miscompilation,
loopoptim,
vectorizers
Assignees
Reporter
dtcxzyw
Reproducer: https://alive2.llvm.org/ce/z/PASME7
```
; bin/opt -passes=loop-vectorize reduced.ll -S -o opt.ll
target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-i128:128-f80:128-n8:16:32:64-S128"
target triple = "x86_64-unknown-linux-gnu"
@e = global [1 x [4 x i16]] zeroinitializer
@crc32_tab = global [256 x i32] zeroinitializer
define i32 @src() {
entry:
store i32 1, ptr @crc32_tab, align 4
br label %for.cond4.preheader.outer.i.i.split
for.cond4.preheader.outer.i.i.split: ; preds = %for.cond4.preheader.outer.i.i.split, %entry
%0 = phi i64 [ %1, %for.cond4.preheader.outer.i.i.split ], [ 3, %entry ]
%indvars.iv.i.i23 = phi i64 [ %indvars.iv.next.i.i, %for.cond4.preheader.outer.i.i.split ], [ 2, %entry ]
%arrayidx9.i.i = getelementptr i16, ptr @e, i64 %0
store i16 1, ptr %arrayidx9.i.i, align 2
%indvars.iv.next.i.i = add i64 %indvars.iv.i.i23, -1
%tobool7.not.i.i = icmp eq i64 %indvars.iv.next.i.i, 0
%1 = add i64 %indvars.iv.next.i.i, 1
br i1 %tobool7.not.i.i, label %for.end.thread.i.i.1, label %for.cond4.preheader.outer.i.i.split
for.end.thread.i.i.1: ; preds = %for.cond4.preheader.outer.i.i.split
%arrayidx947.i.i = getelementptr i16, ptr @e, i64 %1
store i16 0, ptr %arrayidx947.i.i, align 2
br label %for.body3
for.body3: ; preds = %for.body3, %for.end.thread.i.i.1
%indvars.iv = phi i64 [ 0, %for.end.thread.i.i.1 ], [ %indvars.iv.next, %for.body3 ]
%xor2.i70.i.i1718 = phi i32 [ 0, %for.end.thread.i.i.1 ], [ %3, %for.body3 ]
%arrayidx5 = getelementptr i16, ptr @e, i64 %indvars.iv
%2 = load i16, ptr %arrayidx5, align 2
%conv = zext i16 %2 to i64
%idxprom.i.i.i = and i64 %conv, 1
%arrayidx.i.i.i = getelementptr i32, ptr @crc32_tab, i64 %idxprom.i.i.i
%3 = load i32, ptr %arrayidx.i.i.i, align 4
%indvars.iv.next = add i64 %indvars.iv, 1
%exitcond.not = icmp eq i64 %indvars.iv.next, 4
br i1 %exitcond.not, label %for.end8, label %for.body3
for.end8: ; preds = %for.body3
ret i32 %xor2.i70.i.i1718
}
```
```
> bin/lli reduced.ll
> echo $?
0
> bin/lli opt.ll
> echo $?
1
```
llvm version: ab9a80a3ad78f611fd06cd6f7215bd828809310c
___
llvm-bugs mailing list
llvm-bugs@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs
[llvm-bugs] [Bug 122592] [clang-tidy] Add ClassIgnoreList option to misc-const-correctness
Issue
122592
Summary
[clang-tidy] Add ClassIgnoreList option to misc-const-correctness
Labels
clang-tidy
Assignees
Reporter
vbvictor
Sometimes in large codebases we want to ignore const-correctness of certain classes because they are immutable by nature, so there is no point in adding const to them.
Minimal example:
```cpp
#include
int main() {
std::mutex m;
std::lock_guard l(m); // check is triggered but we want to ignore it since lock_guard is already immutable.
}
```
Writing NOLINT for thousands of places bloats code and is not an appealing option.
___
llvm-bugs mailing list
llvm-bugs@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs
[llvm-bugs] [Bug 122584] clang crashes on valid code at -O{s, 2, 3} on x86_64-linux-gnu: Assertion `I1 == PoisonMaskElem && "Expected unused subvectors mask"' failed
Issue
122584
Summary
clang crashes on valid code at -O{s,2,3} on x86_64-linux-gnu: Assertion `I1 == PoisonMaskElem && "Expected unused subvectors mask"' failed
Labels
clang
Assignees
Reporter
zhendongsu
It appears to be a recent regression as it doesn't reproduce with 19.1.0 and earlier.
Compiler Explorer: https://godbolt.org/z/vWPhsMfoz
```
[510] % clangtk -v
clang version 20.0.0git (https://github.com/llvm/llvm-project.git 24bd9bc0b59d51c82e9a4d84c21d86d58d0ef6ce)
Target: x86_64-unknown-linux-gnu
Thread model: posix
InstalledDir: /local/home/suz/suz-local/software/local/clang-trunk/bin
Build config: +assertions
Found candidate GCC installation: /usr/lib/gcc/x86_64-linux-gnu/11
Found candidate GCC installation: /usr/lib/gcc/x86_64-linux-gnu/12
Selected GCC installation: /usr/lib/gcc/x86_64-linux-gnu/12
Candidate multilib: .;@m64
Selected multilib: .;@m64
Found CUDA installation: /usr/local/cuda, version 12.1
[511] %
[511] % clangtk -O2 small.c
clang-20: /local/suz-local/software/clangbuild/llvm-project/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp:10936: llvm::InstructionCost llvm::slpvectorizer::BoUpSLP::ShuffleCostEstimator::finalize(llvm::ArrayRef, llvm::ArrayRef >, llvm::ArrayRef, unsigned int, llvm::function_ref&)>): Assertion `I1 == PoisonMaskElem && "Expected unused subvectors mask"' failed.
PLEASE submit a bug report to https://github.com/llvm/llvm-project/issues/ and include the crash backtrace, preprocessed source, and associated run script.
Stack dump:
0. Program arguments: /local/home/suz/suz-local/software/local/clang-trunk/bin/clang-20 -cc1 -triple x86_64-unknown-linux-gnu -emit-obj -dumpdir a- -disable-free -clear-ast-before-backend -main-file-name small.c -mrelocation-model pic -pic-level 2 -pic-is-pie -mframe-pointer=none -fmath-errno -ffp-contract=on -fno-rounding-math -mconstructor-aliases -funwind-tables=2 -target-cpu x86-64 -tune-cpu generic -debugger-tuning=gdb -fdebug-compilation-dir=/local/suz-local/software/emitesting/bugs/20250111-clangtk-m64-Os-build-025751/delta -fcoverage-compilation-dir=/local/suz-local/software/emitesting/bugs/20250111-clangtk-m64-Os-build-025751/delta -resource-dir /local/home/suz/suz-local/software/local/clang-trunk/lib/clang/20 -I /usr/local/include -I /local/suz-local/software/local/include -internal-isystem /local/home/suz/suz-local/software/local/clang-trunk/lib/clang/20/include -internal-isystem /usr/local/include -internal-isystem /usr/lib/gcc/x86_64-linux-gnu/12/../../../../x86_64-linux-gnu/include -internal-externc-isystem /usr/include/x86_64-linux-gnu -internal-externc-isystem /include -internal-externc-isystem /usr/include -O2 -ferror-limit 19 -fgnuc-version=4.2.1 -fskip-odr-check-in-gmf -fcolor-diagnostics -vectorize-loops -vectorize-slp -faddrsig -D__GCC_HAVE_DWARF2_CFI_ASM=1 -o /tmp/small-2771d7.o -x c small.c
1. parser at end of file
2. Optimizer
3. Running pass "function(float2int,lower-constant-intrinsics,loop(loop-rotate,loop-deletion),loop-distribute,inject-tli-mappings,loop-vectorize,infer-alignment,loop-load-elim,instcombine,simplifycfg,slp-vectorizer,vector-combine,instcombine,loop-unroll,transform-warning,sroa,infer-alignment,instcombine,loop-mssa(licm),alignment-from-assumptions,loop-sink,instsimplify,div-rem-pairs,tailcallelim,simplifycfg)" on module "small.c"
4. Running pass "slp-vectorizer" on function "main"
#0 0x562bd646d500 llvm::sys::PrintStackTrace(llvm::raw_ostream&, int) (/local/home/suz/suz-local/software/local/clang-trunk/bin/clang-20+0x4576500)
#1 0x562bd646a91f llvm::sys::RunSignalHandlers() (/local/home/suz/suz-local/software/local/clang-trunk/bin/clang-20+0x457391f)
#2 0x562bd646aa75 SignalHandler(int) Signals.cpp:0:0
#3 0x7f13d1771520 (/lib/x86_64-linux-gnu/libc.so.6+0x42520)
#4 0x7f13d17c5a7c pthread_kill (/lib/x86_64-linux-gnu/libc.so.6+0x96a7c)
#5 0x7f13d1771476 gsignal (/lib/x86_64-linux-gnu/libc.so.6+0x42476)
#6 0x7f13d17577f3 abort (/lib/x86_64-linux-gnu/libc.so.6+0x287f3)
#7 0x7f13d175771b (/lib/x86_64-linux-gnu/libc.so.6+0x2871b)
#8 0x7f13d1768e96 (/lib/x86_64-linux-gnu/libc.so.6+0x39e96)
#9 0x562bd7f25a4f llvm::slpvectorizer::BoUpSLP::ShuffleCostEstimator::finalize(llvm::ArrayRef, llvm::ArrayRef>, llvm::ArrayRef, unsigned int, llvm::function_ref&)>) (/local/home/suz/suz-local/software/local/clang-trunk/bin/clang-20+0x602ea4f)
#10 0x562bd7f9164b llvm::InstructionCost llvm::slpvectorizer::BoUpSLP::processBuildVector, llvm::slpvectorizer::BoUpSLP, llvm::SmallPtrSetImpl>(llvm::slpvectorizer::BoUpSLP::TreeEntry const*, llvm::Type*, llvm::TargetTransformInfo&, llvm::ArrayRef&, llvm::slpvectorizer::BoUpSLP&,
[llvm-bugs] [Bug 122624] [AArch64] Failure to CSE `add` with commuted `adds`
Issue
122624
Summary
[AArch64] Failure to CSE `add` with commuted `adds`
Labels
backend:AArch64,
missed-optimization
Assignees
Reporter
Kmeakin
https://godbolt.org/z/zxfEn17d8
The code for `u32::checked_add()` produces two `add`s on AArch64:
```rust
pub fn checked_add(x: u32, y: u32) -> Option {
x.checked_add(y)
}
```
```asm
example::checked_add::h2758207837cc4075:
cmn w0, w1
add w1, w1, w0
csetw8, lo
mov w0, w8
ret
```
`cmn` is an alias for `adds wzr, w0, w1`. And since `add` is commutative, it should be CSE'd with the next instruction, to give:
```asm
example::checked_add::h2758207837cc4075:
addsw1, w1, w0
csetw8, lo
mov w0, w8
ret
```
The same issue does not occur on x86_64 or RISC-V
___
llvm-bugs mailing list
llvm-bugs@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs
[llvm-bugs] [Bug 122631] Request Commit Access For Sharjeel-Khan
Issue 122631 Summary Request Commit Access For Sharjeel-Khan Labels infra:commit-access-request Assignees Reporter Sharjeel-Khan ### Why Are you requesting commit access ? I recently joined Android LLVM toolchain team at Google. I need commit access to land PRs myself after approval for Android changes. ___ llvm-bugs mailing list llvm-bugs@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs
[llvm-bugs] [Bug 122612] [clangd] Add an option to disable special handling of stanadard library code
Issue 122612 Summary [clangd] Add an option to disable special handling of stanadard library code Labels enhancement, clangd Assignees Reporter philnik777 While there are good reasons for adding some special magic for standard libraries, this really gets in the way when actually working on a standard library. For example, the include cleaner is completely broken, since it's always mapping public symbols to the public headers, even though library-internal headers should actually be used. It would be awesome if there was a flag for clangd to disable such special magic, allowing it to be used when working on a standard library without patching your own version together. ___ llvm-bugs mailing list llvm-bugs@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs
[llvm-bugs] [Bug 122616] MCA fails to parse movsx intsructions
Issue 122616 Summary MCA fails to parse movsx intsructions Labels new issue Assignees Reporter TiborGY See: https://gcc.godbolt.org/z/z4nWWTj6E Trying to analyze some assembly from gfortran results in errors when MCA sees some movsx instructions: ``` :105:2: error: invalid operand for instruction movsx rax, r10d ^ :253:2: error: invalid operand for instruction movsx rdi, esi ``` ___ llvm-bugs mailing list llvm-bugs@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs
