Commit 7b0579df authored by Ben Walker's avatar Ben Walker Committed by Jim Harris
Browse files

nvme: Use a non-temporal move instruction when copying NVMe cmd 



When copying the NVMe command from the request to the actual
submission queue slot, use a non-temporal move instruction.

The submission queue slots are never read by software - only
written to. So don't pollute the CPU cache with their contents.

Change-Id: I112f721abfac03bd7b33ec9ddf783d4bf2952b42
Signed-off-by: default avatarBen Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/450193


Tested-by: default avatarSPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: default avatarChangpeng Liu <changpeng.liu@intel.com>
Reviewed-by: default avatarXiaodong Liu <xiaodong.liu@intel.com>
Reviewed-by: default avatarJim Harris <james.r.harris@intel.com>
Reviewed-by: default avatarShuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
parent 794b2533

lib/nvme/nvme_pcie.c

+6 −6
Original line number Diff line number Diff line
@@ -1079,16 +1079,16 @@ nvme_pcie_copy_command(struct spdk_nvme_cmd *dst, const struct spdk_nvme_cmd *sr 
	__m256i *d256 = (__m256i *)dst;

	const __m256i *s256 = (const __m256i *)src;



	_mm256_store_si256(&d256[0], _mm256_load_si256(&s256[0]));

	_mm256_store_si256(&d256[1], _mm256_load_si256(&s256[1]));

	_mm256_stream_si256(&d256[0], _mm256_load_si256(&s256[0]));

	_mm256_stream_si256(&d256[1], _mm256_load_si256(&s256[1]));

#elif defined(__SSE2__)

	__m128i *d128 = (__m128i *)dst;

	const __m128i *s128 = (const __m128i *)src;



	_mm_store_si128(&d128[0], _mm_load_si128(&s128[0]));

	_mm_store_si128(&d128[1], _mm_load_si128(&s128[1]));

	_mm_store_si128(&d128[2], _mm_load_si128(&s128[2]));

	_mm_store_si128(&d128[3], _mm_load_si128(&s128[3]));

	_mm_stream_si128(&d128[0], _mm_load_si128(&s128[0]));

	_mm_stream_si128(&d128[1], _mm_load_si128(&s128[1]));

	_mm_stream_si128(&d128[2], _mm_load_si128(&s128[2]));

	_mm_stream_si128(&d128[3], _mm_load_si128(&s128[3]));

#else

	*dst = *src;

#endif