[Bug libstdc++/105720] New: std::views::split_view wrong behaviour in case of partial match
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105720
Bug ID: 105720
Summary: std::views::split_view wrong behaviour in case of
partial match
Product: gcc
Version: 10.3.1
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: libstdc++
Assignee: unassigned at gcc dot gnu.org
Reporter: andij.cr at gmail dot com
Target Milestone: ---
compiled with
g++-10 -std=c++20 split_view_wrong.cpp -lfmt
godbolt link https://gcc.godbolt.org/z/47TxWovd4
fmtlib used for exposition only
#include
#include
#include
#include
auto words_no_bug = std::string_view{"Hello-_-C++-_-20-_-!-_-"};
auto words_bug = std::string_view{"Hello--_-C++-_-20-_-!-_-"};
auto delim = std::string_view{"-_-"};
// needed because split_view is lazy in gcc 10.3
auto range_to_str = [](auto &) {
return fmt::format("{}", fmt::join(r, ""));
};
int main() {
fmt::print("no bug: '{}' tokens: {}\n", words_no_bug,
words_no_bug | std::views::split(delim) |
std::views::transform(range_to_str));
fmt::print("bug: '{}' tokens: {}\n", words_bug,
words_bug | std::views::split(delim) |
std::views::transform(range_to_str));
}
this code applies split to tokenize a text
compiled with gcc-10.3 it wrongly produces
no bug: 'Hello-_-C++-_-20-_-!-_-' tokens: ["Hello", "C++", "20", "!"]
bug: 'Hello--_-C++-_-20-_-!-_-' tokens: ["Hello-", "20", "!"]
while compiled with gcc-11.3 is correctly produces
no bug: 'Hello-_-C++-_-20-_-!-_-' tokens: ["Hello", "C++", "20", "!"]
bug: 'Hello--_-C++-_-20-_-!-_-' tokens: ["Hello-", "C++", "20", "!"]
notice how the substring "--_-C++" instead of being split in ["-", "C++"]
is split as ["-"], skipping the "C++" token.
it's fixed from gcc-11, but i couldn't find a mention in the release notes
about it
[Bug tree-optimization/104959] New: nested lambda capture pack by ref will load from nullptr
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104959
Bug ID: 104959
Summary: nested lambda capture pack by ref will load from
nullptr
Product: gcc
Version: 10.3.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: tree-optimization
Assignee: unassigned at gcc dot gnu.org
Reporter: andij.cr at gmail dot com
Target Milestone: ---
testcase:
#include
template
auto line = [](Ts &&...args) {
if constexpr (sizeof...(Ts) != 0) {
([&] { assert( != nullptr); }(), ...);
}
};
int main() { line<10>(false); }
compiling and executing this with
g++ 10.3 -std=c++20 -O1 -fsanitize=undefined
will trigger the assertion.
this code is a reduction of a more complex code, where the bug caused a crash.
compiling with -O0 or with GCC 11 will not trigger the assertion.
each template, lambda, if constexpr (sizeof...) seems to be necessary
to trigger the bug
the assert needs to be here to trigger the load of args
using a different method (e.g. using args in an expression)
will also trigger -Wuninitialized
compiler explorer link:
https://gcc.godbolt.org/z/W7EMTP4W8
note that in the assembly __assert_fail is called directly
this seems similar to
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=68177
and
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=97938
[Bug tree-optimization/104275] New: Os does not apply return value optimization while O2 and O3 does
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104275
Bug ID: 104275
Summary: Os does not apply return value optimization while O2
and O3 does
Product: gcc
Version: 12.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: tree-optimization
Assignee: unassigned at gcc dot gnu.org
Reporter: andij.cr at gmail dot com
Target Milestone: ---
tested from gcc 8 to gcc 11
an identity function (mark) interposed in a call stack that ends in a complex
type is reasonably elided in O2 and O3, but at Os it creates a somewhat strange
assembly.
tested on arm32 and x86_64.
for a less artificial example, where the problem still appears:
https://gcc.godbolt.org/z/GbKrGKa6f
code:
https://godbolt.org/z/v95jEvvzc
// condensed result of a constexpr trasformation.
// in this form, it would be nice if it was transparent to the value
template
auto mark(Ts&& head) noexcept -> decltype(auto) {
return static_cast(head);
}
#include
// generic producer of a complex type
auto generate() -> std::vector;
// here is a stack of functions using mark
namespace {
// in an anonymous namespace to nudge the compiler to inline them
auto user_base() { return mark(generate()); }
auto user_mark() { return mark(user_base()); }
auto user_mark2() { return mark(user_mark()); }
auto user_mark3() { return mark(user_mark2()); }
} // namespace
// this function has a normal assembly at O2 and O3
// but a silly one at Os
auto user_mark4() { return mark(user_mark3()); }
compiled with
-std=c++17 -O2
user_mark4():
pushr12
mov r12, rdi
sub rsp, 32
mov rdi, rsp
callgenerate()
mov rax, QWORD PTR [rsp]
mov QWORD PTR [r12], rax
mov rax, QWORD PTR [rsp+8]
mov QWORD PTR [r12+8], rax
mov rax, QWORD PTR [rsp+16]
mov QWORD PTR [r12+16], rax
add rsp, 32
mov rax, r12
pop r12
ret
compiled with
-std=c++17 -Os
user_mark4():
pushr13
pushr12
mov r12, rdi
pushrbp
pushrbx
sub rsp, 40
lea rdi, [rsp+8]
callgenerate()
lea rdi, [rsp+8]
mov r13, QWORD PTR [rsp+8]
mov rbp, QWORD PTR [rsp+16]
mov QWORD PTR [rsp+8], 0
mov rbx, QWORD PTR [rsp+24]
mov QWORD PTR [rsp+16], 0
mov QWORD PTR [rsp+24], 0
callstd::_Vector_base
>::~_Vector_base() [base object destructor]
lea rdi, [rsp+8]
mov QWORD PTR [rsp+24], 0
mov QWORD PTR [rsp+16], 0
mov QWORD PTR [rsp+8], 0
callstd::_Vector_base
>::~_Vector_base() [base object destructor]
lea rdi, [rsp+8]
mov QWORD PTR [rsp+24], 0
mov QWORD PTR [rsp+16], 0
mov QWORD PTR [rsp+8], 0
callstd::_Vector_base
>::~_Vector_base() [base object destructor]
lea rdi, [rsp+8]
mov QWORD PTR [rsp+24], 0
mov QWORD PTR [rsp+16], 0
mov QWORD PTR [rsp+8], 0
callstd::_Vector_base
>::~_Vector_base() [base object destructor]
mov QWORD PTR [r12], r13
lea rdi, [rsp+8]
mov QWORD PTR [r12+8], rbp
mov QWORD PTR [r12+16], rbx
mov QWORD PTR [rsp+24], 0
mov QWORD PTR [rsp+16], 0
mov QWORD PTR [rsp+8], 0
callstd::_Vector_base
>::~_Vector_base() [base object destructor]
add rsp, 40
mov rax, r12
pop rbx
pop rbp
pop r12
pop r13
ret
[Bug c++/93513] New: internal compiler error internal compiler error: unexpected expression ‘(char)(e)’ of kind cast_expr
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93513
Bug ID: 93513
Summary: internal compiler error internal compiler error:
unexpected expression ‘(char)(e)’ of kind cast_expr
Product: gcc
Version: 9.2.1
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: c++
Assignee: unassigned at gcc dot gnu.org
Reporter: andij.cr at gmail dot com
Target Milestone: ---
compiling this c++ code
enum class error {};
template
void afunction(F) {
error{char(0)};
}
with g++ 9.2
with std=c++17 or std=c++20
will give
internal compiler error: unexpected expression ‘(char)(0)’ of kind cast_expr
4 | error{char(0)};
|^
in contrast, with std=c++14:
error: cannot convert ‘char’ to ‘error’ in initialization
4 | error{char(0)};
| ^~~
| |
| char
checking with compiler explorer, it seems that gcc 8.3 does not generate this
error: https://gcc.godbolt.org/z/yZ9ckH
[Bug middle-end/91674] New: [ARM/thumb] redundant memcpy does not get optimized away on thumb
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91674
Bug ID: 91674
Summary: [ARM/thumb] redundant memcpy does not get optimized
away on thumb
Product: gcc
Version: 8.2.1
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: middle-end
Assignee: unassigned at gcc dot gnu.org
Reporter: andij.cr at gmail dot com
Target Milestone: ---
consider this c++ function
#include
#include
#include
auto to_bytes(uint32_t arg){
std::array out{};
std::memcpy(out.data(), , sizeof(arg));
return out;
}
on a little endian arch this function could be no-op.
compiled with g++ -Os we get:
to_bytes(unsigned int):
mov eax, edi
ret
on arm this somewhat works:
compiled with arm-none-eabi-g++ -Os
to_bytes(unsigned int):
sub sp, sp, #8
add sp, sp, #8
bx lr
notice the redundant sub followed by an add
but if if thumb is forced, the full optimization is not performed
compiled with arm-none-eabi-g++ -Os -march=armv7-m -mtune=cortex-m3
to_bytes(unsigned int):
mov r3, r0
movsr0, #0
uxtbr2, r3
bfi r0, r2, #0, #8
ubfxr2, r3, #8, #8
bfi r0, r2, #8, #8
ubfxr2, r3, #16, #8
bfi r0, r2, #16, #8
lsrsr3, r3, #24
sub sp, sp, #8
bfi r0, r3, #24, #8
add sp, sp, #8
bx lr
in contrast, cross compiling with clang7 produces the desired optimization:
compiled with clang++7 --target=arm-none-eabi -march=armv7-m -mtune=cortex-m3
to_bytes(unsigned int):
bx lr
notice also how there is no redundant stack pointer manipulation
