unfortunately the ffs/clz/ctz/popcount/parity builtins are defined to return
signed types despite the fact they are either undefined or return a
non-negative integer. as a result unnecessary sign extension occurs. for
example in this case the result of bsf could be used directly in the mask[]
index.
-dean
% cat ctz.c
unsigned mask[8];
unsigned foo(unsigned y, unsigned char x)
{
return y & mask[__builtin_ctz(x)];
}
% ~/gcc/bin/gcc -g -O3 -Wall -c ctz.c
% objdump -dr ctz.o
ctz.o: file format elf64-x86-64
Disassembly of section .text:
0000000000000000 <foo>:
0: 40 0f b6 f6 movzbl %sil,%esi
4: 0f bc f6 bsf %esi,%esi
7: 48 63 f6 movslq %esi,%rsi
a: 23 3c b5 00 00 00 00 and 0x0(,%rsi,4),%edi
d: R_X86_64_32S mask
11: 89 f8 mov %edi,%eax
13: c3 retq
% ~/gcc/bin/gcc -v
Using built-in specs.
Target: x86_64-unknown-linux-gnu
Configured with: ../gcc/configure --prefix=/home/odo/gcc --enable-languages=c
--enable-targets=x86_64-unknown-linux-gnu x86_64-unknown-linux-gnu :
(reconfigured) ../gcc/configure --prefix=/home/odo/gcc --enable-languages=c
--without-mudflap --disable-biarch x86_64-unknown-linux-gnu : (reconfigured)
../gcc/configure --prefix=/home/odo/gcc --enable-languages=c --disable-multilib
--disable-biarch x86_64-unknown-linux-gnu
Thread model: posix
gcc version 4.3.0 20061104 (experimental)
--
Summary: result of ffs/clz/ctz/popcount/parity are already sign-
extended
Product: gcc
Version: 4.3.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: target
AssignedTo: unassigned at gcc dot gnu dot org
ReportedBy: dean at arctic dot org
GCC build triplet: x86_64-unknown-linux-gnu
GCC host triplet: x86_64-unknown-linux-gnu
GCC target triplet: x86_64-unknown-linux-gnu
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=29776
