https://gcc.gnu.org/bugzilla/show_bug.cgi?id=70042
Bug ID: 70042
Summary: Room for optimization of x+1>y vs x>=y
Product: gcc
Version: 6.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: c
Assignee: unassigned at gcc dot gnu.org
Reporter: jengelh at inai dot de
Target Milestone: ---
$ gcc -v
Using built-in specs.
COLLECT_GCC=gcc-6
COLLECT_LTO_WRAPPER=/usr/lib64/gcc/x86_64-suse-linux/6/lto-wrapper
Target: x86_64-suse-linux
Configured with: ../configure --prefix=/usr --infodir=/usr/share/info
--mandir=/usr/share/man --libdir=/usr/lib64 --libexecdir=/usr/lib64
--enable-languages=c,c++,objc,fortran,obj-c++,java,ada,go --enable-checking=yes
--with-gxx-include-dir=/usr/include/c++/6 --enable-ssp --disable-libssp
--disable-libvtv --disable-plugin --with-bugurl=http://bugs.opensuse.org/
--with-pkgversion='SUSE Linux' --disable-libgcj --with-slibdir=/lib64
--with-system-zlib --enable-__cxa_atexit --enable-libstdcxx-allocator=new
--disable-libstdcxx-pch --with-default-libstdcxx-abi=gcc4-compatible
--enable-version-specific-runtime-libs --enable-linker-build-id
--enable-linux-futex --program-suffix=-6 --without-system-libunwind
--enable-multilib --with-arch-32=x86-64 --with-tune=generic
--build=x86_64-suse-linux --host=x86_64-suse-linux
Thread model: posix
gcc version 6.0.0 20160202 (experimental) [trunk revision 233076] (SUSE Linux)
$ cat g1.c geq.c
#include <string.h>
int main(int argc, char **argv) { return strlen(argv[0])+1 > 108; }
#include <string.h>
int main(int argc, char **argv) { return strlen(argv[0]) >= 108; }
$ gcc-6 g1.c -Wall -O2 -c geq.c
$ objdump -d g1.o geq.o
0000000000000000 <main>: //g1
0: 48 83 ec 08 sub $0x8,%rsp
4: 48 8b 3e mov (%rsi),%rdi
7: e8 00 00 00 00 callq c <main+0xc>
c: 48 83 c0 01 add $0x1,%rax
10: 48 83 f8 6c cmp $0x6c,%rax
14: 0f 97 c0 seta %al
17: 48 83 c4 08 add $0x8,%rsp
1b: 0f b6 c0 movzbl %al,%eax
1e: c3 retq
0000000000000000 <main>: //geq
0: 48 83 ec 08 sub $0x8,%rsp
4: 48 8b 3e mov (%rsi),%rdi
7: e8 00 00 00 00 callq c <main+0xc>
c: 48 83 f8 6b cmp $0x6b,%rax
10: 0f 97 c0 seta %al
13: 48 83 c4 08 add $0x8,%rsp
17: 0f b6 c0 movzbl %al,%eax
1a: c3 retq
The x+1>=y variant uses more code, but does not have to.
In contrast, with -Os:
0000000000000000 <main>: //g1
0: 48 8b 3e mov (%rsi),%rdi
3: 31 c0 xor %eax,%eax
5: 48 83 c9 ff or $0xffffffffffffffff,%rcx
9: f2 ae repnz scas %es:(%rdi),%al
b: 31 c0 xor %eax,%eax
d: 48 f7 d1 not %rcx
10: 48 83 f9 6c cmp $0x6c,%rcx
14: 0f 97 c0 seta %al
17: c3 retq
0000000000000000 <main>: //geq
0: 48 8b 3e mov (%rsi),%rdi
3: 31 c0 xor %eax,%eax
5: 48 83 c9 ff or $0xffffffffffffffff,%rcx
9: f2 ae repnz scas %es:(%rdi),%al
b: 48 89 c8 mov %rcx,%rax
e: 48 f7 d0 not %rax
11: 48 ff c8 dec %rax
14: 48 83 f8 6b cmp $0x6b,%rax
18: 0f 97 c0 seta %al
1b: 0f b6 c0 movzbl %al,%eax
1e: c3 retq
the reverse is true, x>=y using more code but does not have to.
