cpusupport: detect a missing _mm_loadu_si64

Old compilers (clang before 2016, gcc before 2019) didn't implement the _mm_loadu_si64() function. As for the compiler flags: in some cases, all we need is -maes -DBROKEN_MM_LOADU_SI64 but sometimes we might need -maes -Wno-cast-align -DBROKEN_MM_LOADU_SI64 or -maes -Wno-cast-align -Wno-cast-qual -DBROKEN_MM_LOADU_SI64 or even -maes -Wno-cast-align -Wno-cast-qual -Wno-missing-prototypes -DBROKEN_MM_LOADU_SI64 However, the more tests we add to cpusupport.sh, the longer it takes. I figured that since this is only relevant to old compilers, we might as well give it the full set of flags, so that newer compilers don't spend time checking it.
Tarsnap · Feb 23, 2021 · 4349f54 · 4349f54
1 parent 59d05dc
commit 4349f54
Show file tree

Hide file tree

Showing 2 changed files with 9 additions and 2 deletions.
diff --git a/libcperciva/cpusupport/Build/cpusupport-X86-AESNI.c b/libcperciva/cpusupport/Build/cpusupport-X86-AESNI.c
@@ -23,7 +23,12 @@ main(void)
 	uint8_t a[16];
 
 	x = load_128(a);
-	y = _mm_aesenc_si128(x, x);
+#ifdef BROKEN_MM_LOADU_SI64
+	y = _mm_loadu_si128(a);
+#else
+	y = _mm_loadu_si64(a);
+#endif
+	y = _mm_aesenc_si128(x, y);
 	_mm_storeu_si128((__m128i *)&a[0], y);
 	return (a[0]);
 }
diff --git a/libcperciva/cpusupport/Build/cpusupport.sh b/libcperciva/cpusupport/Build/cpusupport.sh
@@ -60,7 +60,9 @@ feature X86 CPUID_COUNT ""
 feature X86 AESNI "" "-maes"						\
     "-maes -Wno-cast-align"						\
     "-maes -Wno-missing-prototypes -Wno-cast-qual"			\
-    "-maes -Wno-missing-prototypes -Wno-cast-qual -Wno-cast-align"
+    "-maes -Wno-missing-prototypes -Wno-cast-qual -Wno-cast-align"	\
+    "-maes -Wno-missing-prototypes -Wno-cast-qual -Wno-cast-align	\
+    -DBROKEN_MM_LOADU_SI64"
 feature X86 CRC32_64 "" "-msse4.2"					\
     "-msse4.2 -Wno-cast-align"						\
     "-msse4.2 -Wno-cast-align -fno-strict-aliasing"			\