d6
/
xz-analysis-mirror
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

tuklib_integer: Use __builtin_clz() with Clang. 

Clang has support for __builtin_clz(), but previously Clang would
fallback to either the MSVC intrinsic or the regular C code. This was
discovered due to a bug where a new version of Clang required the
<intrin.h> header file in order to use the MSVC intrinsics.

Thanks to Anton Kochkov for notifying us about the bug.
This commit is contained in:
Jia Tan 2023-04-19 21:59:03 +08:00
parent 3938718ce3
commit 7062348bf3
1 changed files with 3 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
src/common/tuklib_integer.h
View File

@ -630,7 +630,7 @@ bsr32(uint32_t n)
#if defined(__INTEL_COMPILER) #if defined(__INTEL_COMPILER)
return _bit_scan_reverse(n); return _bit_scan_reverse(n);
#elif TUKLIB_GNUC_REQ(3, 4) && UINT_MAX == UINT32_MAX #elif (TUKLIB_GNUC_REQ(3, 4) || defined(__clang__)) && UINT_MAX == UINT32_MAX
// GCC >= 3.4 has __builtin_clz(), which gives good results on // GCC >= 3.4 has __builtin_clz(), which gives good results on
// multiple architectures. On x86, __builtin_clz() ^ 31U becomes // multiple architectures. On x86, __builtin_clz() ^ 31U becomes
// either plain BSR (so the XOR gets optimized away) or LZCNT and // either plain BSR (so the XOR gets optimized away) or LZCNT and
@ -684,7 +684,7 @@ clz32(uint32_t n)
#if defined(__INTEL_COMPILER) #if defined(__INTEL_COMPILER)
return _bit_scan_reverse(n) ^ 31U; return _bit_scan_reverse(n) ^ 31U;
#elif TUKLIB_GNUC_REQ(3, 4) && UINT_MAX == UINT32_MAX #elif (TUKLIB_GNUC_REQ(3, 4) || defined(__clang__)) && UINT_MAX == UINT32_MAX
return (uint32_t)__builtin_clz(n); return (uint32_t)__builtin_clz(n);
#elif defined(__GNUC__) && (defined(__i386__) || defined(__x86_64__)) #elif defined(__GNUC__) && (defined(__i386__) || defined(__x86_64__))
@ -739,7 +739,7 @@ ctz32(uint32_t n)
#elif TUKLIB_GNUC_REQ(3, 4) && UINT_MAX >= UINT32_MAX #elif TUKLIB_GNUC_REQ(3, 4) && UINT_MAX >= UINT32_MAX
return (uint32_t)__builtin_ctz(n); return (uint32_t)__builtin_ctz(n);
#elif defined(__GNUC__) && (defined(__i386__) || defined(__x86_64__)) #elif (TUKLIB_GNUC_REQ(3, 4) || defined(__clang__)) && UINT_MAX == UINT32_MAX
uint32_t i; uint32_t i;
__asm__("bsfl %1, %0" : "=r" (i) : "rm" (n)); __asm__("bsfl %1, %0" : "=r" (i) : "rm" (n));
return i; return i;