DPDK patches and discussions
 help / color / mirror / Atom feed
From: Liming Sun <lsun@ezchip.com>
To: <dev@dpdk.org>
Subject: [dpdk-dev] [PATCH 1/2] driver/net/mpipe: add rte_vect.h and enable CONFIG_RTE_LIBRTE_LPM
Date: Tue, 22 Dec 2015 14:49:49 -0500	[thread overview]
Message-ID: <1450813790-11874-2-git-send-email-lsun@ezchip.com> (raw)
In-Reply-To: <1450813790-11874-1-git-send-email-lsun@ezchip.com>

rte_vect.h was missing earlier thus LPM was disabled and l3fwd is
not able to compile. This commit implements the vector api and
enable LPM in the tilegx configuration by default. It also includes
a minor optimization to use __insn_fetchadd4() instead of
rte_atomic32_xxx() in mpipe_dp_enter/mpipe_dp_exit to avoid the
unnecessary memory fence.

Signed-off-by: Liming Sun <lsun@ezchip.com>
---
 config/defconfig_tile-tilegx-linuxapp-gcc          |    2 +-
 drivers/net/mpipe/mpipe_tilegx.c                   |   18 +++-
 lib/librte_eal/common/include/arch/tile/rte_vect.h |   93 ++++++++++++++++++++
 3 files changed, 107 insertions(+), 6 deletions(-)
 create mode 100644 lib/librte_eal/common/include/arch/tile/rte_vect.h

diff --git a/config/defconfig_tile-tilegx-linuxapp-gcc b/config/defconfig_tile-tilegx-linuxapp-gcc
index fb61bcd..39794f6 100644
--- a/config/defconfig_tile-tilegx-linuxapp-gcc
+++ b/config/defconfig_tile-tilegx-linuxapp-gcc
@@ -64,7 +64,7 @@ CONFIG_RTE_LIBRTE_ENIC_PMD=n
 
 # This following libraries are not available on the tile architecture.
 # So they're turned off.
-CONFIG_RTE_LIBRTE_LPM=n
+CONFIG_RTE_LIBRTE_LPM=y
 CONFIG_RTE_LIBRTE_ACL=n
 CONFIG_RTE_LIBRTE_SCHED=n
 CONFIG_RTE_LIBRTE_PORT=n
diff --git a/drivers/net/mpipe/mpipe_tilegx.c b/drivers/net/mpipe/mpipe_tilegx.c
index 5845511..8d006fa 100644
--- a/drivers/net/mpipe/mpipe_tilegx.c
+++ b/drivers/net/mpipe/mpipe_tilegx.c
@@ -451,13 +451,13 @@ static inline void
 mpipe_dp_enter(struct mpipe_dev_priv *priv)
 {
 	__insn_mtspr(SPR_DSTREAM_PF, 0);
-	rte_atomic32_inc(&priv->dp_count);
+	__insn_fetchadd4(&priv->dp_count, 1);
 }
 
 static inline void
 mpipe_dp_exit(struct mpipe_dev_priv *priv)
 {
-	rte_atomic32_dec(&priv->dp_count);
+	__insn_fetchadd4(&priv->dp_count, -1);
 }
 
 static inline void
@@ -484,12 +484,20 @@ mpipe_recv_mbuf(struct mpipe_dev_priv *priv, gxio_mpipe_idesc_t *idesc,
 	uint16_t size = gxio_mpipe_idesc_get_xfer_size(idesc);
 	struct rte_mbuf *mbuf = RTE_PTR_SUB(va, priv->rx_offset);
 
-	rte_pktmbuf_reset(mbuf);
 	mbuf->data_off = (uintptr_t)va - (uintptr_t)mbuf->buf_addr;
-	mbuf->port     = in_port;
-	mbuf->data_len = size;
+	mbuf->nb_segs = 1;
+	mbuf->port = in_port;
+	mbuf->ol_flags = 0;
+	if (gxio_mpipe_idesc_get_ethertype(idesc) == ETHER_TYPE_IPv4)
+		mbuf->packet_type = RTE_PTYPE_L3_IPV4 | RTE_PTYPE_L2_ETHER;
+	else if (gxio_mpipe_idesc_get_ethertype(idesc) == ETHER_TYPE_IPv6)
+		mbuf->packet_type = RTE_PTYPE_L3_IPV6 | RTE_PTYPE_L2_ETHER;
+	else
+		mbuf->packet_type = RTE_PTYPE_UNKNOWN;
 	mbuf->pkt_len  = size;
+	mbuf->data_len = size;
 	mbuf->hash.rss = gxio_mpipe_idesc_get_flow_hash(idesc);
+	mbuf->next = NULL;
 
 	PMD_DEBUG_RX("%s: RX mbuf %p, buffer %p, buf_addr %p, size %d\n",
 		     mpipe_name(priv), mbuf, va, mbuf->buf_addr, size);
diff --git a/lib/librte_eal/common/include/arch/tile/rte_vect.h b/lib/librte_eal/common/include/arch/tile/rte_vect.h
new file mode 100644
index 0000000..32d768a
--- /dev/null
+++ b/lib/librte_eal/common/include/arch/tile/rte_vect.h
@@ -0,0 +1,93 @@
+/*
+ *   BSD LICENSE
+ *
+ *   Copyright (C) EZchip Semiconductor Ltd. 2015.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions
+ *   are met:
+ *
+ *     * Redistributions of source code must retain the above copyright
+ *       notice, this list of conditions and the following disclaimer.
+ *     * Redistributions in binary form must reproduce the above copyright
+ *       notice, this list of conditions and the following disclaimer in
+ *       the documentation and/or other materials provided with the
+ *       distribution.
+ *     * Neither the name of EZchip Semiconductor nor the names of its
+ *       contributors may be used to endorse or promote products derived
+ *       from this software without specific prior written permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#ifndef _RTE_VECT_H_
+#define _RTE_VECT_H_
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+typedef __int128 __m128i;
+
+#define	XMM_SIZE	sizeof(__m128i)
+#define	XMM_MASK	(XMM_SIZE - 1)
+
+typedef union rte_xmm {
+	__m128i x;
+	uint32_t u32[XMM_SIZE / sizeof(uint32_t)];
+	uint64_t u64[XMM_SIZE / sizeof(uint64_t)];
+} rte_xmm_t;
+
+/* Extracts the low order 64-bit integer. */
+#define _mm_cvtsi128_si64(a) ((rte_xmm_t*)&a)->u64[0]
+
+/* Sets the 2 signed 64-bit integer values. */
+#define _mm_set_epi64x(i1, i0) ({         \
+	rte_xmm_t m;                      \
+	m.u64[0] = i0;                    \
+	m.u64[1] = i1;                    \
+	(m.x);                            \
+})
+
+/* Sets the 4 signed 32-bit integer values. */
+#define _mm_set_epi32(i3, i2, i1, i0)  ({ \
+	rte_xmm_t m;                      \
+	m.u32[0] = i0;                    \
+	m.u32[1] = i1;                    \
+	m.u32[2] = i2;                    \
+	m.u32[3] = i3;                    \
+	(m.x);                            \
+})
+
+/* Shifts right the 4 32-bit integers by count bits with zeros. */
+#define _mm_srli_epi32(v, cnt) ({                  \
+	rte_xmm_t m;                                 \
+	m.u64[0] = __insn_v4shru(((rte_xmm_t*)&(v))->u64[0], cnt); \
+	m.u64[1] = __insn_v4shru(((rte_xmm_t*)&(v))->u64[1], cnt); \
+	(m.x);                                       \
+})
+
+/* Shifts the 128-bit value in a right by imm bytes. */
+#define _mm_srli_si128(v, imm) ((v) >> (imm * sizeof(uint8_t)))
+
+/* Bitwise AND of the 128-bit value in a and the 128-bit value in b. */
+#define _mm_and_si128(a, b) ((a) & (b))
+
+/* Loads 128-bit value. */
+#define _mm_loadu_si128(p) (*(p))
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /* _RTE_VECT_H_ */
-- 
1.7.1

  reply	other threads:[~2015-12-22 19:50 UTC|newest]

Thread overview: 20+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-12-22 19:49 [dpdk-dev] [PATCH 0/2] Some enhancement of the tilegx mpipe driver Liming Sun
2015-12-22 19:49 ` Liming Sun [this message]
2016-01-08  7:41   ` [dpdk-dev] [PATCH 1/2] driver/net/mpipe: add rte_vect.h and enable CONFIG_RTE_LIBRTE_LPM Tony Lu
2015-12-22 19:49 ` [dpdk-dev] [PATCH 2/2] driver/net/mpipe: fix the crash/hung issue when testpmd quits Liming Sun
2016-01-08  7:49   ` Tony Lu
2016-01-08 14:39   ` [dpdk-dev] [PATCH v2 1/2] driver/net/mpipe: add rte_vect.h and enable CONFIG_RTE_LIBRTE_LPM Liming Sun
2016-01-08 14:39     ` [dpdk-dev] [PATCH v2 2/2] driver/net/mpipe: fix the crash/hung issue when testpmd quits Liming Sun
2016-02-09 15:59       ` Bruce Richardson
2016-02-10 14:50         ` Liming Sun
2016-02-10  5:15       ` [dpdk-dev] [PATCH v3] " Liming Sun
2016-02-26 17:35         ` Bruce Richardson
2016-02-09 15:55     ` [dpdk-dev] [PATCH v2 1/2] driver/net/mpipe: add rte_vect.h and enable CONFIG_RTE_LIBRTE_LPM Bruce Richardson
2016-02-09 18:40       ` Liming Sun
2016-02-10 14:52       ` Liming Sun
2016-02-10  3:46     ` [dpdk-dev] [PATCH v3 1/2] eal/tile: " Liming Sun
2016-02-10  3:47       ` [dpdk-dev] [PATCH v3 2/2] driver/net/mpipe: some code optimization Liming Sun
2016-02-10  4:04     ` [dpdk-dev] [PATCH v4 1/2] eal/tile: add rte_vect.h and enable CONFIG_RTE_LIBRTE_LPM Liming Sun
2016-02-10  4:04       ` [dpdk-dev] [PATCH v4 2/2] driver/net/mpipe: some code optimization Liming Sun
2016-03-08 19:59       ` [dpdk-dev] [PATCH v4 1/2] eal/tile: add rte_vect.h and enable CONFIG_RTE_LIBRTE_LPM Thomas Monjalon
2016-03-17 14:04         ` Thomas Monjalon

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=1450813790-11874-2-git-send-email-lsun@ezchip.com \
    --to=lsun@ezchip.com \
    --cc=dev@dpdk.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).