[llvm-bugs] [Bug 136457] Fix crash lowering stack guard on OpenBSD/aarch64
Issue 136457 Summary Fix crash lowering stack guard on OpenBSD/aarch64 Labels new issue Assignees Reporter brad0 /cherry-pick c180e249d0013474d502cd779ec65b33cf7e9468 ___ llvm-bugs mailing list llvm-bugs@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs
[llvm-bugs] [Bug 136464] Clang checks impossible cast
Issue
136464
Summary
Clang checks impossible cast
Labels
clang
Assignees
Reporter
vzze
### Issue
Clang tries to check if cast of `const char*` is ill-formed even though the context of the cast is reduced to only `int` by `if constexpr`
Code
```c++
// main.cc
#include
#include
int main() {
using type = std::variant;
type a = 12;
type b = 42;
int res = std::visit([&](auto a1) -> int {
return std::visit([&](auto b1) -> int {
if constexpr(
std::is_same_v, std::decay_t> &&
std::is_same_v, int>
)
return int(a1) + int(b1);
else
throw "Wrong type!";
}, b);
}, a);
}
```
Unix Error
```
main.cc:16:24: error: cast from pointer to smaller type 'int' loses information
16 | return int(a1) + int(b1);
|^~~
main.cc:11:47: note: while substituting into a lambda _expression_ here
11 | return std::visit([&](auto b1) -> int {
| ^
/usr/bin/../lib64/gcc/x86_64-pc-linux-gnu/14.2.1/../../../../include/c++/14.2.1/bits/invoke.h:61:14: note: in instantiation of function template specialization 'main():
:(anonymous class)::operator()' requested here
61 | { return std::forward<_Fn>(__f)(std::forward<_Args>(__args)...); }
| ^
/usr/bin/../lib64/gcc/x86_64-pc-linux-gnu/14.2.1/../../../../include/c++/14.2.1/bits/invoke.h:96:19: note: in instantiation of function template specialization 'std::__
invoke_impl' requested here
96 | return std::__invoke_impl<__type>(__tag{}, std::forward<_Callable>(__fn),
| ^
/usr/bin/../lib64/gcc/x86_64-pc-linux-gnu/14.2.1/../../../../include/c++/14.2.1/variant:1060:16: note: in instantiation of function template specialization 'std::__invo
ke<(lambda at main.cc:10:26), const char *&>' requested here
1060 | return std::__invoke(std::forward<_Visitor>(__visitor),
| ^
/usr/bin/../lib64/gcc/x86_64-pc-linux-gnu/14.2.1/../../../../include/c++/14.2.1/variant:1820:5: note: in instantiation of member function 'std::__detail::__variant::__g
en_vtable_impl (*)((lambda at main.cc:10:26) &&, std::variant &)>, std::integer_sequence>::__visit_invoke' requested here
1820 | _GLIBCXX_VISIT_CASE(1)
| ^
/usr/bin/../lib64/gcc/x86_64-pc-linux-gnu/14.2.1/../../../../include/c++/14.2.1/variant:1811:4: note: expanded from macro '_GLIBCXX_VISIT_CASE'
1811 | __visit_invoke(std::forward<_Visitor>(__visitor), \
| ^
/usr/bin/../lib64/gcc/x86_64-pc-linux-gnu/14.2.1/../../../../include/c++/14.2.1/variant:1882:18: note: in instantiation of function template specialization 'std::__do_v
isit, (lambda at main.cc:10:26), std::variant &>' requested here
1882 | return std::__do_visit<_Tag>(
| ^
main.cc:10:20: note: in instantiation of function template specialization 'std::visit<(lambda at main.cc:10:26), std::variant &>' requested here
10 | int res = std::visit([&](auto a1) -> int {
|^
1 error generated.
```
Win11 Error
```
.\main.cc:16:24: warning: cast to smaller integer type 'int' from 'const char *' [-Wpointer-to-int-cast]
16 | return int(a1) + int(b1);
| ^~~
.\main.cc:11:47: note: while substituting into a lambda _expression_ here
11 | return std::visit([&](auto b1) -> int {
| ^
C:\Program Files\Microsoft Visual Studio\2022\Community\VC\Tools\MSVC\14.43.34808\include\type_traits:1706:16: note: in
instantiation of function template specialization 'main()::(anonymous class)::operator()' requested
here
1706 | return static_cast<_Callable&&>(_Obj)(static_cast<_Ty1&&>(_Arg1), static_cast<_Types2&&>(_Args2)...);
|^
C:\Program Files\Microsoft Visual Studio\2022\Community\VC\Tools\MSVC\14.43.34808\include\variant:1422:25: note: in
instantiation of function template specialization 'std::invoke<(lambda at ./main.cc:10:26), const char *&>'
requested here
1422 | return _STD invoke(static_cast<_Callable&&>(_Obj),
| ^
C:\Program Files\Microsoft Visual Studio\2022\Community\VC\Tools\MSVC\14.43.34808\include\variant:1492:23: note: in
instantiation of function template specialization 'std::_Variant_dispatcher>::_Dispatch2 &, false>' requested here
1492 | _STL_STAMP(4, _STL_VISIT_STAMP);
| ^
C:\Program Files\Microsoft Visual Studio\2022\Community\VC\Tools\MSVC\14.43.34808\include\variant:1546:49: note: in
instantiation of function template specialization 'std::_Visit_strategy<1>::_Visit2, std::integer_sequence,
std::integer_sequence>, (lambda at ./main.cc:10:26), std::variant &>'
requested here
1546 | return _Visit_strategy<_Strategy>::template _Visit2<_Ret, _ListOfIndexVectors>(
|
[llvm-bugs] [Bug 136472] Clang errors for out-of-line definition with deducing-this of non-templated type nested within templated type
Issue
136472
Summary
Clang errors for out-of-line definition with deducing-this of non-templated type nested within templated type
Labels
clang
Assignees
Reporter
scrossuk
Given this code:
```
template
struct Type final
{
struct Nested
{
void method(this auto& self);
};
};
template
void Type::Nested::method(this auto& self) {}
```
Clang reports ([Godbolt](https://godbolt.org/z/9MqhEGPsK)):
```
:11:30: error: an explicit object parameter cannot appear in a non-member function
11 | void Type::Nested::method(this auto& self) {}
```
Note that a work around is to make the nested type also be a template ([Godbolt](https://godbolt.org/z/WWjKe5WPv)):
```
template
struct Type final
{
template
struct Nested
{
void method(this auto& self);
};
};
template
template
void Type::Nested::method(this auto& self) {}
```
___
llvm-bugs mailing list
llvm-bugs@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs
[llvm-bugs] [Bug 136432] Compiler crash, fails to compile
Issue 136432 Summary Compiler crash, fails to compile Labels new issue Assignees Reporter troldal Clang crashes when compiling my code. As instructed in the failure message, here is the backtrace, preprocessed source, and associated run script: [Flash-90296d.zip](https://github.com/user-attachments/files/19821407/Flash-90296d.zip) ___ llvm-bugs mailing list llvm-bugs@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs
[llvm-bugs] [Bug 136473] Missed optimization: switch with GEP inside cannot generate lookup table
Issue
136473
Summary
Missed optimization: switch with GEP inside cannot generate lookup table
Labels
new issue
Assignees
Reporter
oxalica
[C godbolt](https://godbolt.org/z/exoPTKKPa)
I'm expecting `selector1` to be optimized to `selector2`. But the optimizer cannot treat constant GEP as add in order to hoist the common `s + _` out of switch.
[IIUC](https://llvm.org/docs/GetElementPtr.html#how-is-gep-different-from-ptrtoint-arithmetic-and-inttoptr), `getelementptr` carries more (aliasing) information and should be more efficient than direct pointer arithmetic, but it seems to be worse than simple pointer arithmetic in this case.
Similar issue: #33633 is about homogeneous struct, while this issue is about heterogeneous struct. I'm not sure if this is considered the same, because in this case, we cannot simply cast the pointer into `[4 x i32]`, a lookup table is necessary.
Origin: I run into this issue on Rust [godbolt](https://godbolt.org/z/6s8jbYhEP), where it's hard to do `offset_of` workaround even with `unsafe`, because of the need to attach type-related vtable to each pointer.
```c
#include
#include
#include
typedef struct {
uint8_t a;
uint8_t b;
uint16_t c;
uint32_t d;
uint64_t e;
float f;
double g;
} S;
/* indirect jump, with LEA+RET in each branch */
void *selector1(S *s, int i) {
switch (i) {
case 0: return &s->a;
case 1: return &s->b;
case 2: return &s->c;
case 3: return &s->d;
case 4: return &s->e;
case 5: return &s->f;
case 6: return &s->g;
default: return NULL;
}
}
/* indirect load, ADD */
void *selector2(S *s, int i) {
size_t off;
switch (i) {
case 0: off = offsetof(S, a); break;
case 1: off = offsetof(S, b); break;
case 2: off = offsetof(S, c); break;
case 3: off = offsetof(S, d); break;
case 4: off = offsetof(S, e); break;
case 5: off = offsetof(S, f); break;
case 6: off = offsetof(S, g); break;
default: return NULL;
}
return (uint8_t *)s + off;
}
```
___
llvm-bugs mailing list
llvm-bugs@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs
[llvm-bugs] [Bug 136425] Widened pointers not optimized back to scalar for vectorized interleaved accesses on RISC-V
Issue 136425 Summary Widened pointers not optimized back to scalar for vectorized interleaved accesses on RISC-V Labels backend:RISC-V, missed-optimization Assignees Reporter lukel97 https://compiler-explorer.com/z/bcnocK139 In the loop f, the loop vectorizer is able to use a scalar address throughout the segmented accesses. In g, where not every segmented field is written back to, it's kept around as a widened vector of pointers, even though we only use the first element of it. `legalizeAndOptimizeInductions` in VPlanTransforms.cpp may not be catching this case. ___ llvm-bugs mailing list llvm-bugs@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs
[llvm-bugs] [Bug 136430] [InstCombine] ninf should not be propagated
Issue
136430
Summary
[InstCombine] ninf should not be propagated
Labels
miscompilation,
llvm:instcombine,
floating-point
Assignees
dtcxzyw
Reporter
dtcxzyw
Reproducer: https://alive2.llvm.org/ce/z/cmneUX
```
define half @test_fcmp_select_maxnum(half %x) {
#0:
%cmp2 = fcmp ogt half %x, 0x5bf8
%sel2 = select nnan ninf nsz i1 %cmp2, half %x, half 0x5bf8
ret half %sel2
}
=>
define half @test_fcmp_select_maxnum(half %x) {
#0:
%sel2 = fmax ninf nsz half %x, 0x5bf8
ret half %sel2
}
Transformation doesn't verify!
ERROR: Target is more poisonous than source
Example:
half %x = #xfc00 (-oo)
Source:
i1 %cmp2 = #x0 (0)
half %sel2 = #x5bf8 (255)
Target:
half %sel2 = poison
Source value: #x5bf8 (255)
Target value: poison
Summary:
0 correct transformations
1 incorrect transformations
0 failed-to-prove transformations
0 Alive2 errors
```
___
llvm-bugs mailing list
llvm-bugs@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs
[llvm-bugs] [Bug 136428] Clang uses wrong ABI for __atomic_store_16 on Windows
Issue
136428
Summary
Clang uses wrong ABI for __atomic_store_16 on Windows
Labels
clang
Assignees
Reporter
Alcaro
```c++
unsigned __int128 g;
extern "C" void k(unsigned __int128* a, unsigned __int128 b, int c);
extern "C" void r()
{
k(&g, 1234, 5);
}
extern "C" void p()
{
__atomic_store_n(&g, 1234, __ATOMIC_SEQ_CST);
}
```
Expected: Pass the int128 by reference, and place __ATOMIC_SEQ_CST=5 in r8d, for both calls, like GCC.
Actual: Clang calls __atomic_store_16 by passing the int128 by value in r8:rdx, and the 5 goes in r9d. https://godbolt.org/z/49Gzso4ax
Originally reported at https://github.com/brechtsanders/winlibs_mingw/issues/256
___
llvm-bugs mailing list
llvm-bugs@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs
[llvm-bugs] [Bug 136418] [mlir] error: Dialect `ml_program' not found for custom op 'ml_program.global'
Issue
136418
Summary
[mlir] error: Dialect `ml_program' not found for custom op 'ml_program.global'
Labels
mlir
Assignees
ryan-holt-1
Reporter
vfdff
The MR [[mlir][MLProgram] Add MLProgram to MemRef bufferization pass](https://github.com/llvm/llvm-project/pull/75103) has landed for a long time, but I still get the compile error, does I some miss something ?
> mlir-translate -mlir-to-llvmir mlp.out.mlir
> error: Dialect `ml_program' not found for custom op 'ml_program.global'
* test: https://godbolt.org/z/d8MGqPGb9
> (py311-source) root@998ee80b761b:/home/zhongyunde/source/test/simple_mlp# cat mlp.out.mlir
```
module attributes {torch.debug_module_name = "MlpModel"} {
ml_program.global private mutable @global_seed(dense<0> : tensor) : tensor
}
```
___
llvm-bugs mailing list
llvm-bugs@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs
[llvm-bugs] [Bug 136435] clang_equalCursors not work as expect
Issue
136435
Summary
clang_equalCursors not work as expect
Labels
new issue
Assignees
Reporter
neko-para
I'm using my own nodejs binding of libclang to parse my source to generate a simplified ast.
```typescript
// dfs logic, using cursors as a stack
function parseAnnotate(tu: CTranslationUnit) {
const root = tu.cursor
const cursors: CCursor[] = [root]
tu.cursor.visitChildren((cursor, parent) => {
while (cursors.length > 0 && !cursors[0].equal(parent)) {
cursors.shift()
}
if (cursors.length === 0) {
console.warn('jump out root', cursor, parent)
}
cursors.unshift(cursor)
console.log(' '.repeat(cursors.length - 2), cursor.spelling, cursor.kindStr)
return CXChildVisitResult.Recurse
})
}
```
The code above works quite well, but when parsing the code below, the cursors that should be equal become different.
```
_Const_lvalue_cond_oper ClassTemplatePartialSpecialization
_Ty1 TemplateTypeParameter
_Ty2 TemplateTypeParameter
RequiresExpr
> Here should be a TemplateRef with parent of RequiresExpr, but my program crashed
```
https://github.com/microsoft/STL/blob/5762e6bcaf7f5f8b5dba0a9aabf0acbd0e335e80/stl/inc/type_traits#L1288
**After dumping the CXCursor data, it shows that the parent.data[0] is non-zero, while cursors[0].data[0] (previously stored one) is zero, causing clang_equalCursors return false.**
I think there should be some problem either on my binding code or the way I use libclang, but I'm not sure.
___
llvm-bugs mailing list
llvm-bugs@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs
