This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
[Bug c++/70094] New: Missed optimization when passing a constant struct argument by value
- From: "robryk at gmail dot com" <gcc-bugzilla at gcc dot gnu dot org>
- To: gcc-bugs at gcc dot gnu dot org
- Date: Sat, 05 Mar 2016 13:47:28 +0000
- Subject: [Bug c++/70094] New: Missed optimization when passing a constant struct argument by value
- Auto-submitted: auto-generated
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=70094
Bug ID: 70094
Summary: Missed optimization when passing a constant struct
argument by value
Product: gcc
Version: 5.3.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: c++
Assignee: unassigned at gcc dot gnu.org
Reporter: robryk at gmail dot com
Target Milestone: ---
Function baz in the code listing below gets compiled into something that writes
to the stack. This is unnecessary: one can just load the argument into rdi with
movabs and get rid of stack adjustments and memory accesses:
---snip---
[robryk@sharya-rana gccbug]$ cat > bug.cc
struct foo {
int a;
int b;
int c;
};
void bar(foo);
void baz() {
foo f;
f.a = 1;
f.b = 2;
f.c = 3;
bar(f);
}
[robryk@sharya-rana gccbug]$ g++ -v
Using built-in specs.
COLLECT_GCC=g++
COLLECT_LTO_WRAPPER=/usr/lib/gcc/x86_64-unknown-linux-gnu/5.3.0/lto-wrapper
Target: x86_64-unknown-linux-gnu
Configured with: /build/gcc-multilib/src/gcc-5.3.0/configure --prefix=/usr
--libdir=/usr/lib --libexecdir=/usr/lib --mandir=/usr/share/man
--infodir=/usr/share/info --with-bugurl=https://bugs.archlinux.org/
--enable-languages=c,c++,ada,fortran,go,lto,objc,obj-c++ --enable-shared
--enable-threads=posix --enable-libmpx --with-system-zlib --with-isl
--enable-__cxa_atexit --disable-libunwind-exceptions --enable-clocale=gnu
--disable-libstdcxx-pch --disable-libssp --enable-gnu-unique-object
--enable-linker-build-id --enable-lto --enable-plugin
--enable-install-libiberty --with-linker-hash-style=gnu
--enable-gnu-indirect-function --enable-multilib --disable-werror
--enable-checking=release
Thread model: posix
gcc version 5.3.0 (GCC)
[robryk@sharya-rana gccbug]$ g++ bug.cc -O2 -c -o bug.o
[robryk@sharya-rana gccbug]$ objdump -d bug.o
bug.o: file format elf64-x86-64
Disassembly of section .text:
0000000000000000 <_Z3bazv>:
0: 48 83 ec 18 sub $0x18,%rsp
4: be 03 00 00 00 mov $0x3,%esi
9: c7 04 24 01 00 00 00 movl $0x1,(%rsp)
10: c7 44 24 04 02 00 00 movl $0x2,0x4(%rsp)
17: 00
18: 48 8b 3c 24 mov (%rsp),%rdi
1c: e8 00 00 00 00 callq 21 <_Z3bazv+0x21>
21: 48 83 c4 18 add $0x18,%rsp
25: c3 retq
---snip---
I've verified that Clang performs the optimization I was talking about and
that, according to gcc.godbolt.org, a snapshot of gcc 6 misses this
optimization too.
For comparison, the code Clang produces (according to godbolt):
movabsq $8589934593, %rdi # imm = 0x200000001
movl $3, %esi
jmp bar(foo) # TAILCALL