tcg: Remove CONFIG_VECTOR16

The comment in tcg-runtime-gvec.c about CONFIG_VECTOR16 says that
tcg-op-gvec.c has eliminated size 8 vectors, and only passes on
multiples of 16. This may have been true of the first few operations,
but is not true of all operations.

In particular, multiply, shift by scalar, and compare of 8- and 16-bit
elements are not expanded inline if host vector operations are not
supported.

For an x86_64 host that does not support AVX, this means that we will
fall back to the helper, which will attempt to use SSE instructions,
which will SEGV on an invalid 8-byte aligned memory operation.

This patch simply removes the CONFIG_VECTOR16 code and configuration
without further simplification.

Buglink: https://bugs.launchpad.net/bugs/1863508

Backports commit 43d1ccd2a02fadf3c36b46a8c31125a28864d141 from qemu

qemu/accel/tcg/tcg-runtime-gvec.c

@@ -24,32 +24,6 @@
 #include "tcg-gvec-desc.h"
 
 
-/* Virtually all hosts support 16-byte vectors.  Those that don't can emulate
- * them via GCC's generic vector extension.  This turns out to be simpler and
- * more reliable than getting the compiler to autovectorize.
- *
- * In tcg-op-gvec.c, we asserted that both the size and alignment of the data
- * are multiples of 16.
- *
- * When the compiler does not support all of the operations we require, the
- * loops are written so that we can always fall back on the base types.
- */
-#ifdef CONFIG_VECTOR16
-typedef uint8_t vec8 __attribute__((vector_size(16)));
-typedef uint16_t vec16 __attribute__((vector_size(16)));
-typedef uint32_t vec32 __attribute__((vector_size(16)));
-typedef uint64_t vec64 __attribute__((vector_size(16)));
-
-typedef int8_t svec8 __attribute__((vector_size(16)));
-typedef int16_t svec16 __attribute__((vector_size(16)));
-typedef int32_t svec32 __attribute__((vector_size(16)));
-typedef int64_t svec64 __attribute__((vector_size(16)));
-
-#define DUP16(X)  { X, X, X, X, X, X, X, X, X, X, X, X, X, X, X, X }
-#define DUP8(X)   { X, X, X, X, X, X, X, X }
-#define DUP4(X)   { X, X, X, X }
-#define DUP2(X)   { X, X }
-#else
 typedef uint8_t vec8;
 typedef uint16_t vec16;
 typedef uint32_t vec32;
@@ -64,7 +38,6 @@ typedef int64_t svec64;
 #define DUP8(X)   X
 #define DUP4(X)   X
 #define DUP2(X)   X
-#endif /* CONFIG_VECTOR16 */
 
 static inline void clear_high(void *d, intptr_t oprsz, uint32_t desc)
 {
@@ -917,13 +890,7 @@ void HELPER(gvec_sar64v)(void *d, void *a, void *b, uint32_t desc)
     clear_high(d, oprsz, desc);
 }
 
-/* If vectors are enabled, the compiler fills in -1 for true.
+#define DO_CMP0(X)  -(X)
-   Otherwise, we must take care of this by hand.  */
-#ifdef CONFIG_VECTOR16
-# define DO_CMP0(X)  X
-#else
-# define DO_CMP0(X)  -(X)
-#endif
 
 #define DO_CMP1(NAME, TYPE, OP)                                            \
 void HELPER(NAME)(void *d, void *a, void *b, uint32_t desc)                \

qemu/configure

@@ -1174,58 +1174,6 @@ if compile_prog "" "" ; then
     int128=yes
 fi
 
-########################################
-# See if 16-byte vector operations are supported.
-# Even without a vector unit the compiler may expand these.
-# There is a bug in old GCC for PPC that crashes here.
-# Unfortunately it's the system compiler for Centos 7.
-
-cat > $TMPC << EOF
-typedef unsigned char U1 __attribute__((vector_size(16)));
-typedef unsigned short U2 __attribute__((vector_size(16)));
-typedef unsigned int U4 __attribute__((vector_size(16)));
-typedef unsigned long long U8 __attribute__((vector_size(16)));
-typedef signed char S1 __attribute__((vector_size(16)));
-typedef signed short S2 __attribute__((vector_size(16)));
-typedef signed int S4 __attribute__((vector_size(16)));
-typedef signed long long S8 __attribute__((vector_size(16)));
-static U1 a1, b1;
-static U2 a2, b2;
-static U4 a4, b4;
-static U8 a8, b8;
-static S1 c1;
-static S2 c2;
-static S4 c4;
-static S8 c8;
-static int i;
-void helper(void *d, void *a, int shift, int i);
-void helper(void *d, void *a, int shift, int i)
-{
-  *(U1 *)(d + i) = *(U1 *)(a + i) << shift;
-  *(U2 *)(d + i) = *(U2 *)(a + i) << shift;
-  *(U4 *)(d + i) = *(U4 *)(a + i) << shift;
-  *(U8 *)(d + i) = *(U8 *)(a + i) << shift;
-}
-int main(void)
-{
-  a1 += b1; a2 += b2; a4 += b4; a8 += b8;
-  a1 -= b1; a2 -= b2; a4 -= b4; a8 -= b8;
-  a1 *= b1; a2 *= b2; a4 *= b4; a8 *= b8;
-  a1 &= b1; a2 &= b2; a4 &= b4; a8 &= b8;
-  a1 |= b1; a2 |= b2; a4 |= b4; a8 |= b8;
-  a1 ^= b1; a2 ^= b2; a4 ^= b4; a8 ^= b8;
-  a1 <<= i; a2 <<= i; a4 <<= i; a8 <<= i;
-  a1 >>= i; a2 >>= i; a4 >>= i; a8 >>= i;
-  c1 >>= i; c2 >>= i; c4 >>= i; c8 >>= i;
-  return 0;
-}
-EOF
-
-vector16=no
-if compile_prog "" "" ; then
-  vector16=yes
-fi
-
 ##########################################
 # check for usable membarrier system call
 if test "$membarrier" = "yes"; then
@@ -1564,10 +1512,6 @@ if test "$atomic64" = "yes" ; then
   echo "CONFIG_ATOMIC64=y" >> $config_host_mak
 fi
 
-if test "$vector16" = "yes" ; then
-  echo "CONFIG_VECTOR16=y" >> $config_host_mak
-fi
-
 if test "$have_static_assert" = "yes" ; then
   echo "CONFIG_STATIC_ASSERT=y" >> $config_host_mak
 fi