https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108869
Bug ID: 108869
Summary: compiling an intrinsic wrapper : gives internal
compiler error: in
dwarf2out_register_main_translation_unit
Product: gcc
Version: 7.5.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: c
Assignee: unassigned at gcc dot gnu.org
Reporter: jaydesh9 at gmail dot com
Target Milestone: ---
Hi,
I have a generically built binary that needs to include a lookup routine which
gets compiled into vectorized instructions or otherwise based upon whether the
cpu supports avx/avx2.
The lookup routine is same as that explained here :
https://stackoverflow.com/questions/54897297/check-all-bytes-of-a-m128i-for-a-match-of-a-single-byte-using-sse-avx-avx2
Here the (_mm_set1_epi8, __mm_cmpeq_epi8,_mm_movemask_epi8) intrinsic set will
compile into either vectorized instructions if avx/avx2 is supported by the cpu
or just sse based instructions, otherwise.
in a oversimplified main.c : compiled without mavx/mavx2 and with -msse3
-msse4 -o 3
#define __SSE2__
#define SSE_Lookup() \
_mm_set1_epi8; \
__mm_cmpeq_epi8; \
match_bitmap=_mm_movemask_epi8
#endif
static inline __attribute__((always_inline))
uint64_t foo()
{
unsigned int a=1,b,c,d;
uint64_t match_bitmap;
__cpuid(1,a,b,c,d);
if(c & bit_AVX)
{
match_bitmap= avx_lookup();
}else
{
#if __SSE__
SSE_Lookup();
#endif
}
}
foo_avx.c
#include <emmintrin.h>
//mimicing an intrinsic wrapper
//don't want to create any new stack frames so keeping it inline
extern __inline int __attribute__((__gnu_inline__, __always_inline__,
__artificial__))
__avx_lookup (char h, __m128i h)
{
__m128i k = _mm_set1_epi8(h);
__m128i r = _mm_cmpeq_epi8(k,h);
return _mm_movemask_epi8(r);
}
GCC Bug report : Bugzilla :
x86_64_gcc-7.5.0_glibc/bin/x86_64-openwrt-linux-gnu-gcc
Compiled as follows :
/build/openwrt/toolchain-x86_64_gcc-7.5.0_glibc/bin/x86_64-openwrt-linux-gnu-gcc
-fsigned-char -fmerge-constants -fPIC -pipe -MD -MP -Werror -Wall -Wextra
-Wpointer-arith -Wreturn-type -Wwrite-strings -Wno-unused-parameter
-Wunused-variable -Wno-unused-result -Wformat -Wformat-security -Winit-self
-Wno-implicit-fallthrough -Wno-format-truncation -Wno-deprecated-declarations
-Wno-error=deprecated-declarations -mtune=generic -msse3 -msse4 -mpclmul -maes
-mcx16 -g -O0 -DBUILD_UT=1 -std=gnu99 -Wno-missing-field-initializers
-Wmissing-prototypes -mavx -o 2 -o /build/x86_64/common/foo_avx.o foo_avx.c
foo_avx.c:19:1: internal compiler error: in
dwarf2out_register_main_translation_unit, at dwarf2out.c:26628
}
^
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://bugs.openwrt.org/> for instructions.
Makefile:72: recipe for target '/build/x86_64/common/foo_avx.o' failed
make[3]: *** [/build/x86_64/common/foo_avx.o] Error 1
So question are :
1. Is this a genuine bug ?
2. If not, is the approach correct in defining the intrinsic wrapper
3. Is there a better way of doing this ? goal is to have an executable with
code for sse , avx as well as avx2 avx512 embedded that can be invoked based
upon the cpu support at the run time.
Thanks in advance.
-J
