[dpdk-dev,v2] hash: fix library compilation for CPU with no SSE4.1
Commit Message
From: "De Lara Guarch, Pablo" <pablo.de.lara.guarch@intel.com>
_mm_test_all_zeros is not available for CPUs with no SSE4.1,
therefore, DPDK would not build.
This patch adds an alternative for this, using _mm_cmpeq_epi32 and
_mm_movemask_epi8.
Fixes: 48a399119619 ("hash: replace with cuckoo hash implementation")
Signed-off-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>
---
Changes in v2:
- Fixed wrong ifndef (should be ifdef)
lib/librte_hash/rte_cuckoo_hash.c | 6 ++++++
1 file changed, 6 insertions(+)
Comments
2015-07-16 10:00, Pablo de Lara:
> From: "De Lara Guarch, Pablo" <pablo.de.lara.guarch@intel.com>
>
> _mm_test_all_zeros is not available for CPUs with no SSE4.1,
> therefore, DPDK would not build.
> This patch adds an alternative for this, using _mm_cmpeq_epi32 and
> _mm_movemask_epi8.
>
> Fixes: 48a399119619 ("hash: replace with cuckoo hash implementation")
>
> Signed-off-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>
Applied, thanks
@@ -1128,9 +1128,15 @@ rte_hash_k16_cmp_eq(const void *key1, const void *key2, size_t key_len __rte_unu
{
const __m128i k1 = _mm_loadu_si128((const __m128i *) key1);
const __m128i k2 = _mm_loadu_si128((const __m128i *) key2);
+#ifdef RTE_MACHINE_CPUFLAG_SSE4_1
const __m128i x = _mm_xor_si128(k1, k2);
return !_mm_test_all_zeros(x, x);
+#else
+ const __m128i x = _mm_cmpeq_epi32(k1, k2);
+
+ return (_mm_movemask_epi8(x) != 0xffff);
+#endif
}
static int