From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mails.dpdk.org (mails.dpdk.org [217.70.189.124]) by inbox.dpdk.org (Postfix) with ESMTP id B06A9A0093; Tue, 3 May 2022 14:03:53 +0200 (CEST) Received: from [217.70.189.124] (localhost [127.0.0.1]) by mails.dpdk.org (Postfix) with ESMTP id 607C940C35; Tue, 3 May 2022 14:03:53 +0200 (CEST) Received: from out4-smtp.messagingengine.com (out4-smtp.messagingengine.com [66.111.4.28]) by mails.dpdk.org (Postfix) with ESMTP id BBC4F40691 for ; Tue, 3 May 2022 14:03:51 +0200 (CEST) Received: from compute2.internal (compute2.nyi.internal [10.202.2.46]) by mailout.nyi.internal (Postfix) with ESMTP id 055E85C0144; Tue, 3 May 2022 08:03:51 -0400 (EDT) Received: from mailfrontend1 ([10.202.2.162]) by compute2.internal (MEProxy); Tue, 03 May 2022 08:03:51 -0400 DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=monjalon.net; h= cc:cc:content-transfer-encoding:date:date:from:from:in-reply-to :message-id:mime-version:reply-to:sender:subject:subject:to:to; s=fm1; t=1651579430; x=1651665830; bh=C1/BZoPojfDdLehdC+4JYeEp8 EKkaSdWElaVbiLF2aI=; b=S1nYZzmcJ9Sfpn+wy2m060uZPWg7UNIzSZn6O4n2u 2eLcZ46rFj6NDLlwCFWPbW5GZR4nPhZyYxMP46otF37C1aL5XvNW/t9OFIE5C0lH sFZpzarCgsQ6SH6ggKlYK5cy+hMu3PICX42quw/nAr40fHSAyZeeWDSMw5mD8NgH mozgTz61zv7F1VevPTcd+P30j2pIcOo7ijnquBgUNdByQxGcaAvfeMAtLz+KKLXB sQnSpxVEpQutOMPe89k4Mx+FQuJ3460qVMD68CR5wzP6A/fPGdwJhbSztfYhyYgT K0bhxlb8hugn+BQZIyjgIqvjYGTb8U4f6C66vJ8R5/T/g== DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d= messagingengine.com; h=cc:cc:content-transfer-encoding:date:date :from:from:in-reply-to:message-id:mime-version:reply-to:sender :subject:subject:to:to:x-me-proxy:x-me-proxy:x-me-sender :x-me-sender:x-sasl-enc; s=fm1; t=1651579430; x=1651665830; bh=C 1/BZoPojfDdLehdC+4JYeEp8EKkaSdWElaVbiLF2aI=; b=bvbuY2I2Jlb4efU8x q6G1IMZYuEXN/Jp8sVGmvLWr6PCln7is1At1cD1fDr1CsZOfNF0Hl5I4PITipDIe 0ZtGijUxnzYSbCUr7oCriPwusPLE6ROdPO1Gvb4lEC1iL1XBt9vhfSEZ24oVMuGJ 1hJL6g7CBOsF4W9WLZGPOh5YPgQ9T1CJTlfGBQW6rX7ROm3dYywIobSnD1bkJkTL NZj1LnxZAqkiyeeLa9+Kts/8akrLT9O7dq7/vD5t3DheBmBd31AVn6dRrYgc7sVD s4SKbFmB9ALpN2VJ2BJCu3TlDZ9i+6LPB1M319d2y7o1cLuS6NuRX1cTlF2sz11F cT+9w== X-ME-Sender: X-ME-Received: X-ME-Proxy-Cause: gggruggvucftvghtrhhoucdtuddrgedvfedrvdejgdegkecutefuodetggdotefrodftvf curfhrohhfihhlvgemucfhrghsthforghilhdpqfgfvfdpuffrtefokffrpgfnqfghnecu uegrihhlohhuthemuceftddtnecusecvtfgvtghiphhivghnthhsucdlqddutddtmdenuc fjughrpefhvfevufffkffoggfgsedtkeertdertddtnecuhfhrohhmpefvhhhomhgrshcu ofhonhhjrghlohhnuceothhhohhmrghssehmohhnjhgrlhhonhdrnhgvtheqnecuggftrf grthhtvghrnhepveevfeevieeihfetudekgeekleeigeffueekveduteeuffeiudevteei udekfeelnecuvehluhhsthgvrhfuihiivgeptdenucfrrghrrghmpehmrghilhhfrhhomh epthhhohhmrghssehmohhnjhgrlhhonhdrnhgvth X-ME-Proxy: Received: by mail.messagingengine.com (Postfix) with ESMTPA; Tue, 3 May 2022 08:03:49 -0400 (EDT) From: Thomas Monjalon To: dev@dpdk.org Cc: David Christensen , Beilei Xing , Matan Azrad , Viacheslav Ovsiienko , Maxime Coquelin , Chenbo Xia , Bruce Richardson , Vladimir Medvedkin Subject: [PATCH] avoid AltiVec keyword vector Date: Tue, 3 May 2022 14:03:21 +0200 Message-Id: <20220503120321.2438479-1-thomas@monjalon.net> X-Mailer: git-send-email 2.35.3 MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: DPDK patches and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dev-bounces@dpdk.org The keyword "vector" may conflict easily. As a rule, it is better to use the alternative keyword "__vector". Signed-off-by: Thomas Monjalon --- drivers/net/i40e/i40e_rxtx_vec_altivec.c | 214 ++-- drivers/net/mlx5/mlx5_rxtx_vec_altivec.h | 1078 ++++++++--------- .../net/virtio/virtio_rxtx_simple_altivec.c | 70 +- examples/l3fwd/l3fwd_lpm_altivec.h | 14 +- lib/eal/ppc/include/rte_vect.h | 2 +- lib/lpm/rte_lpm_altivec.h | 6 +- 6 files changed, 692 insertions(+), 692 deletions(-) diff --git a/drivers/net/i40e/i40e_rxtx_vec_altivec.c b/drivers/net/i40e/i40e_rxtx_vec_altivec.c index 825834c5cd..2dfa04599c 100644 --- a/drivers/net/i40e/i40e_rxtx_vec_altivec.c +++ b/drivers/net/i40e/i40e_rxtx_vec_altivec.c @@ -27,10 +27,10 @@ i40e_rxq_rearm(struct i40e_rx_queue *rxq) struct i40e_rx_entry *rxep = &rxq->sw_ring[rxq->rxrearm_start]; struct rte_mbuf *mb0, *mb1; - vector unsigned long hdr_room = (vector unsigned long){ + __vector unsigned long hdr_room = (__vector unsigned long){ RTE_PKTMBUF_HEADROOM, RTE_PKTMBUF_HEADROOM}; - vector unsigned long dma_addr0, dma_addr1; + __vector unsigned long dma_addr0, dma_addr1; rxdp = rxq->rx_ring + rxq->rxrearm_start; @@ -40,11 +40,11 @@ i40e_rxq_rearm(struct i40e_rx_queue *rxq) RTE_I40E_RXQ_REARM_THRESH) < 0) { if (rxq->rxrearm_nb + RTE_I40E_RXQ_REARM_THRESH >= rxq->nb_rx_desc) { - dma_addr0 = (vector unsigned long){}; + dma_addr0 = (__vector unsigned long){}; for (i = 0; i < RTE_I40E_DESCS_PER_LOOP; i++) { rxep[i].mbuf = &rxq->fake_mbuf; vec_st(dma_addr0, 0, - (vector unsigned long *)&rxdp[i].read); + (__vector unsigned long *)&rxdp[i].read); } } rte_eth_devices[rxq->port_id].data->rx_mbuf_alloc_failed += @@ -54,7 +54,7 @@ i40e_rxq_rearm(struct i40e_rx_queue *rxq) /* Initialize the mbufs in vector, process 2 mbufs in one loop */ for (i = 0; i < RTE_I40E_RXQ_REARM_THRESH; i += 2, rxep += 2) { - vector unsigned long vaddr0, vaddr1; + __vector unsigned long vaddr0, vaddr1; uintptr_t p0, p1; mb0 = rxep[0].mbuf; @@ -72,8 +72,8 @@ i40e_rxq_rearm(struct i40e_rx_queue *rxq) *(uint64_t *)p1 = rxq->mbuf_initializer; /* load buf_addr(lo 64bit) and buf_iova(hi 64bit) */ - vaddr0 = vec_ld(0, (vector unsigned long *)&mb0->buf_addr); - vaddr1 = vec_ld(0, (vector unsigned long *)&mb1->buf_addr); + vaddr0 = vec_ld(0, (__vector unsigned long *)&mb0->buf_addr); + vaddr1 = vec_ld(0, (__vector unsigned long *)&mb1->buf_addr); /* convert pa to dma_addr hdr/data */ dma_addr0 = vec_mergel(vaddr0, vaddr0); @@ -84,8 +84,8 @@ i40e_rxq_rearm(struct i40e_rx_queue *rxq) dma_addr1 = vec_add(dma_addr1, hdr_room); /* flush desc with pa dma_addr */ - vec_st(dma_addr0, 0, (vector unsigned long *)&rxdp++->read); - vec_st(dma_addr1, 0, (vector unsigned long *)&rxdp++->read); + vec_st(dma_addr0, 0, (__vector unsigned long *)&rxdp++->read); + vec_st(dma_addr1, 0, (__vector unsigned long *)&rxdp++->read); } rxq->rxrearm_start += RTE_I40E_RXQ_REARM_THRESH; @@ -102,32 +102,32 @@ i40e_rxq_rearm(struct i40e_rx_queue *rxq) } static inline void -desc_to_olflags_v(vector unsigned long descs[4], struct rte_mbuf **rx_pkts) +desc_to_olflags_v(__vector unsigned long descs[4], struct rte_mbuf **rx_pkts) { - vector unsigned int vlan0, vlan1, rss, l3_l4e; + __vector unsigned int vlan0, vlan1, rss, l3_l4e; /* mask everything except RSS, flow director and VLAN flags * bit2 is for VLAN tag, bit11 for flow director indication * bit13:12 for RSS indication. */ - const vector unsigned int rss_vlan_msk = (vector unsigned int){ + const __vector unsigned int rss_vlan_msk = (__vector unsigned int){ (int32_t)0x1c03804, (int32_t)0x1c03804, (int32_t)0x1c03804, (int32_t)0x1c03804}; /* map rss and vlan type to rss hash and vlan flag */ - const vector unsigned char vlan_flags = (vector unsigned char){ + const __vector unsigned char vlan_flags = (__vector unsigned char){ 0, 0, 0, 0, RTE_MBUF_F_RX_VLAN | RTE_MBUF_F_RX_VLAN_STRIPPED, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0}; - const vector unsigned char rss_flags = (vector unsigned char){ + const __vector unsigned char rss_flags = (__vector unsigned char){ 0, RTE_MBUF_F_RX_FDIR, 0, 0, 0, 0, RTE_MBUF_F_RX_RSS_HASH, RTE_MBUF_F_RX_RSS_HASH | RTE_MBUF_F_RX_FDIR, 0, 0, 0, 0, 0, 0, 0, 0}; - const vector unsigned char l3_l4e_flags = (vector unsigned char){ + const __vector unsigned char l3_l4e_flags = (__vector unsigned char){ 0, RTE_MBUF_F_RX_IP_CKSUM_BAD, RTE_MBUF_F_RX_L4_CKSUM_BAD, @@ -139,23 +139,23 @@ desc_to_olflags_v(vector unsigned long descs[4], struct rte_mbuf **rx_pkts) | RTE_MBUF_F_RX_IP_CKSUM_BAD, 0, 0, 0, 0, 0, 0, 0, 0}; - vlan0 = (vector unsigned int)vec_mergel(descs[0], descs[1]); - vlan1 = (vector unsigned int)vec_mergel(descs[2], descs[3]); - vlan0 = (vector unsigned int)vec_mergeh(vlan0, vlan1); + vlan0 = (__vector unsigned int)vec_mergel(descs[0], descs[1]); + vlan1 = (__vector unsigned int)vec_mergel(descs[2], descs[3]); + vlan0 = (__vector unsigned int)vec_mergeh(vlan0, vlan1); vlan1 = vec_and(vlan0, rss_vlan_msk); - vlan0 = (vector unsigned int)vec_perm(vlan_flags, - (vector unsigned char){}, - *(vector unsigned char *)&vlan1); + vlan0 = (__vector unsigned int)vec_perm(vlan_flags, + (__vector unsigned char){}, + *(__vector unsigned char *)&vlan1); - rss = vec_sr(vlan1, (vector unsigned int){11, 11, 11, 11}); - rss = (vector unsigned int)vec_perm(rss_flags, (vector unsigned char){}, - *(vector unsigned char *)&rss); + rss = vec_sr(vlan1, (__vector unsigned int){11, 11, 11, 11}); + rss = (__vector unsigned int)vec_perm(rss_flags, (__vector unsigned char){}, + *(__vector unsigned char *)&rss); - l3_l4e = vec_sr(vlan1, (vector unsigned int){22, 22, 22, 22}); - l3_l4e = (vector unsigned int)vec_perm(l3_l4e_flags, - (vector unsigned char){}, - *(vector unsigned char *)&l3_l4e); + l3_l4e = vec_sr(vlan1, (__vector unsigned int){22, 22, 22, 22}); + l3_l4e = (__vector unsigned int)vec_perm(l3_l4e_flags, + (__vector unsigned char){}, + *(__vector unsigned char *)&l3_l4e); vlan0 = vec_or(vlan0, rss); vlan0 = vec_or(vlan0, l3_l4e); @@ -169,23 +169,23 @@ desc_to_olflags_v(vector unsigned long descs[4], struct rte_mbuf **rx_pkts) #define PKTLEN_SHIFT 10 static inline void -desc_to_ptype_v(vector unsigned long descs[4], struct rte_mbuf **rx_pkts, +desc_to_ptype_v(__vector unsigned long descs[4], struct rte_mbuf **rx_pkts, uint32_t *ptype_tbl) { - vector unsigned long ptype0 = vec_mergel(descs[0], descs[1]); - vector unsigned long ptype1 = vec_mergel(descs[2], descs[3]); + __vector unsigned long ptype0 = vec_mergel(descs[0], descs[1]); + __vector unsigned long ptype1 = vec_mergel(descs[2], descs[3]); - ptype0 = vec_sr(ptype0, (vector unsigned long){30, 30}); - ptype1 = vec_sr(ptype1, (vector unsigned long){30, 30}); + ptype0 = vec_sr(ptype0, (__vector unsigned long){30, 30}); + ptype1 = vec_sr(ptype1, (__vector unsigned long){30, 30}); rx_pkts[0]->packet_type = - ptype_tbl[(*(vector unsigned char *)&ptype0)[0]]; + ptype_tbl[(*(__vector unsigned char *)&ptype0)[0]]; rx_pkts[1]->packet_type = - ptype_tbl[(*(vector unsigned char *)&ptype0)[8]]; + ptype_tbl[(*(__vector unsigned char *)&ptype0)[8]]; rx_pkts[2]->packet_type = - ptype_tbl[(*(vector unsigned char *)&ptype1)[0]]; + ptype_tbl[(*(__vector unsigned char *)&ptype1)[0]]; rx_pkts[3]->packet_type = - ptype_tbl[(*(vector unsigned char *)&ptype1)[8]]; + ptype_tbl[(*(__vector unsigned char *)&ptype1)[8]]; } /** @@ -204,17 +204,17 @@ _recv_raw_pkts_vec(struct i40e_rx_queue *rxq, struct rte_mbuf **rx_pkts, uint16_t nb_pkts_recd; int pos; uint64_t var; - vector unsigned char shuf_msk; + __vector unsigned char shuf_msk; uint32_t *ptype_tbl = rxq->vsi->adapter->ptype_tbl; - vector unsigned short crc_adjust = (vector unsigned short){ + __vector unsigned short crc_adjust = (__vector unsigned short){ 0, 0, /* ignore pkt_type field */ rxq->crc_len, /* sub crc on pkt_len */ 0, /* ignore high-16bits of pkt_len */ rxq->crc_len, /* sub crc on data_len */ 0, 0, 0 /* ignore non-length fields */ }; - vector unsigned long dd_check, eop_check; + __vector unsigned long dd_check, eop_check; /* nb_pkts has to be floor-aligned to RTE_I40E_DESCS_PER_LOOP */ nb_pkts = RTE_ALIGN_FLOOR(nb_pkts, RTE_I40E_DESCS_PER_LOOP); @@ -240,15 +240,15 @@ _recv_raw_pkts_vec(struct i40e_rx_queue *rxq, struct rte_mbuf **rx_pkts, return 0; /* 4 packets DD mask */ - dd_check = (vector unsigned long){0x0000000100000001ULL, + dd_check = (__vector unsigned long){0x0000000100000001ULL, 0x0000000100000001ULL}; /* 4 packets EOP mask */ - eop_check = (vector unsigned long){0x0000000200000002ULL, + eop_check = (__vector unsigned long){0x0000000200000002ULL, 0x0000000200000002ULL}; /* mask to shuffle from desc. to mbuf */ - shuf_msk = (vector unsigned char){ + shuf_msk = (__vector unsigned char){ 0xFF, 0xFF, /* pkt_type set as unknown */ 0xFF, 0xFF, /* pkt_type set as unknown */ 14, 15, /* octet 15~14, low 16 bits pkt_len */ @@ -274,35 +274,35 @@ _recv_raw_pkts_vec(struct i40e_rx_queue *rxq, struct rte_mbuf **rx_pkts, for (pos = 0, nb_pkts_recd = 0; pos < nb_pkts; pos += RTE_I40E_DESCS_PER_LOOP, rxdp += RTE_I40E_DESCS_PER_LOOP) { - vector unsigned long descs[RTE_I40E_DESCS_PER_LOOP]; - vector unsigned char pkt_mb1, pkt_mb2, pkt_mb3, pkt_mb4; - vector unsigned short staterr, sterr_tmp1, sterr_tmp2; - vector unsigned long mbp1, mbp2; /* two mbuf pointer + __vector unsigned long descs[RTE_I40E_DESCS_PER_LOOP]; + __vector unsigned char pkt_mb1, pkt_mb2, pkt_mb3, pkt_mb4; + __vector unsigned short staterr, sterr_tmp1, sterr_tmp2; + __vector unsigned long mbp1, mbp2; /* two mbuf pointer * in one XMM reg. */ /* B.1 load 2 mbuf point */ - mbp1 = *(vector unsigned long *)&sw_ring[pos]; + mbp1 = *(__vector unsigned long *)&sw_ring[pos]; /* Read desc statuses backwards to avoid race condition */ /* A.1 load desc[3] */ - descs[3] = *(vector unsigned long *)(rxdp + 3); + descs[3] = *(__vector unsigned long *)(rxdp + 3); rte_compiler_barrier(); /* B.2 copy 2 mbuf point into rx_pkts */ - *(vector unsigned long *)&rx_pkts[pos] = mbp1; + *(__vector unsigned long *)&rx_pkts[pos] = mbp1; /* B.1 load 2 mbuf point */ - mbp2 = *(vector unsigned long *)&sw_ring[pos + 2]; + mbp2 = *(__vector unsigned long *)&sw_ring[pos + 2]; /* A.1 load desc[2-0] */ - descs[2] = *(vector unsigned long *)(rxdp + 2); + descs[2] = *(__vector unsigned long *)(rxdp + 2); rte_compiler_barrier(); - descs[1] = *(vector unsigned long *)(rxdp + 1); + descs[1] = *(__vector unsigned long *)(rxdp + 1); rte_compiler_barrier(); - descs[0] = *(vector unsigned long *)(rxdp); + descs[0] = *(__vector unsigned long *)(rxdp); /* B.2 copy 2 mbuf point into rx_pkts */ - *(vector unsigned long *)&rx_pkts[pos + 2] = mbp2; + *(__vector unsigned long *)&rx_pkts[pos + 2] = mbp2; if (split_packet) { rte_mbuf_prefetch_part2(rx_pkts[pos]); @@ -315,78 +315,78 @@ _recv_raw_pkts_vec(struct i40e_rx_queue *rxq, struct rte_mbuf **rx_pkts, rte_compiler_barrier(); /* pkt 3,4 shift the pktlen field to be 16-bit aligned*/ - const vector unsigned int len3 = vec_sl( - vec_ld(0, (vector unsigned int *)&descs[3]), - (vector unsigned int){0, 0, 0, PKTLEN_SHIFT}); + const __vector unsigned int len3 = vec_sl( + vec_ld(0, (__vector unsigned int *)&descs[3]), + (__vector unsigned int){0, 0, 0, PKTLEN_SHIFT}); - const vector unsigned int len2 = vec_sl( - vec_ld(0, (vector unsigned int *)&descs[2]), - (vector unsigned int){0, 0, 0, PKTLEN_SHIFT}); + const __vector unsigned int len2 = vec_sl( + vec_ld(0, (__vector unsigned int *)&descs[2]), + (__vector unsigned int){0, 0, 0, PKTLEN_SHIFT}); /* merge the now-aligned packet length fields back in */ - descs[3] = (vector unsigned long)len3; - descs[2] = (vector unsigned long)len2; + descs[3] = (__vector unsigned long)len3; + descs[2] = (__vector unsigned long)len2; /* D.1 pkt 3,4 convert format from desc to pktmbuf */ - pkt_mb4 = vec_perm((vector unsigned char)descs[3], - (vector unsigned char){}, shuf_msk); - pkt_mb3 = vec_perm((vector unsigned char)descs[2], - (vector unsigned char){}, shuf_msk); + pkt_mb4 = vec_perm((__vector unsigned char)descs[3], + (__vector unsigned char){}, shuf_msk); + pkt_mb3 = vec_perm((__vector unsigned char)descs[2], + (__vector unsigned char){}, shuf_msk); /* C.1 4=>2 filter staterr info only */ - sterr_tmp2 = vec_mergel((vector unsigned short)descs[3], - (vector unsigned short)descs[2]); + sterr_tmp2 = vec_mergel((__vector unsigned short)descs[3], + (__vector unsigned short)descs[2]); /* C.1 4=>2 filter staterr info only */ - sterr_tmp1 = vec_mergel((vector unsigned short)descs[1], - (vector unsigned short)descs[0]); + sterr_tmp1 = vec_mergel((__vector unsigned short)descs[1], + (__vector unsigned short)descs[0]); /* D.2 pkt 3,4 set in_port/nb_seg and remove crc */ - pkt_mb4 = (vector unsigned char)vec_sub( - (vector unsigned short)pkt_mb4, crc_adjust); - pkt_mb3 = (vector unsigned char)vec_sub( - (vector unsigned short)pkt_mb3, crc_adjust); + pkt_mb4 = (__vector unsigned char)vec_sub( + (__vector unsigned short)pkt_mb4, crc_adjust); + pkt_mb3 = (__vector unsigned char)vec_sub( + (__vector unsigned short)pkt_mb3, crc_adjust); /* pkt 1,2 shift the pktlen field to be 16-bit aligned*/ - const vector unsigned int len1 = vec_sl( - vec_ld(0, (vector unsigned int *)&descs[1]), - (vector unsigned int){0, 0, 0, PKTLEN_SHIFT}); - const vector unsigned int len0 = vec_sl( - vec_ld(0, (vector unsigned int *)&descs[0]), - (vector unsigned int){0, 0, 0, PKTLEN_SHIFT}); + const __vector unsigned int len1 = vec_sl( + vec_ld(0, (__vector unsigned int *)&descs[1]), + (__vector unsigned int){0, 0, 0, PKTLEN_SHIFT}); + const __vector unsigned int len0 = vec_sl( + vec_ld(0, (__vector unsigned int *)&descs[0]), + (__vector unsigned int){0, 0, 0, PKTLEN_SHIFT}); /* merge the now-aligned packet length fields back in */ - descs[1] = (vector unsigned long)len1; - descs[0] = (vector unsigned long)len0; + descs[1] = (__vector unsigned long)len1; + descs[0] = (__vector unsigned long)len0; /* D.1 pkt 1,2 convert format from desc to pktmbuf */ - pkt_mb2 = vec_perm((vector unsigned char)descs[1], - (vector unsigned char){}, shuf_msk); - pkt_mb1 = vec_perm((vector unsigned char)descs[0], - (vector unsigned char){}, shuf_msk); + pkt_mb2 = vec_perm((__vector unsigned char)descs[1], + (__vector unsigned char){}, shuf_msk); + pkt_mb1 = vec_perm((__vector unsigned char)descs[0], + (__vector unsigned char){}, shuf_msk); /* C.2 get 4 pkts staterr value */ - staterr = (vector unsigned short)vec_mergeh( + staterr = (__vector unsigned short)vec_mergeh( sterr_tmp1, sterr_tmp2); /* D.3 copy final 3,4 data to rx_pkts */ vec_st(pkt_mb4, 0, - (vector unsigned char *)&rx_pkts[pos + 3] + (__vector unsigned char *)&rx_pkts[pos + 3] ->rx_descriptor_fields1 ); vec_st(pkt_mb3, 0, - (vector unsigned char *)&rx_pkts[pos + 2] + (__vector unsigned char *)&rx_pkts[pos + 2] ->rx_descriptor_fields1 ); /* D.2 pkt 1,2 set in_port/nb_seg and remove crc */ - pkt_mb2 = (vector unsigned char)vec_sub( - (vector unsigned short)pkt_mb2, crc_adjust); - pkt_mb1 = (vector unsigned char)vec_sub( - (vector unsigned short)pkt_mb1, crc_adjust); + pkt_mb2 = (__vector unsigned char)vec_sub( + (__vector unsigned short)pkt_mb2, crc_adjust); + pkt_mb1 = (__vector unsigned char)vec_sub( + (__vector unsigned short)pkt_mb1, crc_adjust); /* C* extract and record EOP bit */ if (split_packet) { - vector unsigned char eop_shuf_mask = - (vector unsigned char){ + __vector unsigned char eop_shuf_mask = + (__vector unsigned char){ 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, @@ -394,19 +394,19 @@ _recv_raw_pkts_vec(struct i40e_rx_queue *rxq, struct rte_mbuf **rx_pkts, }; /* and with mask to extract bits, flipping 1-0 */ - vector unsigned char eop_bits = vec_and( - (vector unsigned char)vec_nor(staterr, staterr), - (vector unsigned char)eop_check); + __vector unsigned char eop_bits = vec_and( + (__vector unsigned char)vec_nor(staterr, staterr), + (__vector unsigned char)eop_check); /* the staterr values are not in order, as the count * of dd bits doesn't care. However, for end of * packet tracking, we do care, so shuffle. This also * compresses the 32-bit values to 8-bit */ - eop_bits = vec_perm(eop_bits, (vector unsigned char){}, + eop_bits = vec_perm(eop_bits, (__vector unsigned char){}, eop_shuf_mask); /* store the resulting 32-bit value */ *split_packet = (vec_ld(0, - (vector unsigned int *)&eop_bits))[0]; + (__vector unsigned int *)&eop_bits))[0]; split_packet += RTE_I40E_DESCS_PER_LOOP; /* zero-out next pointers */ @@ -417,22 +417,22 @@ _recv_raw_pkts_vec(struct i40e_rx_queue *rxq, struct rte_mbuf **rx_pkts, } /* C.3 calc available number of desc */ - staterr = vec_and(staterr, (vector unsigned short)dd_check); + staterr = vec_and(staterr, (__vector unsigned short)dd_check); /* D.3 copy final 1,2 data to rx_pkts */ vec_st(pkt_mb2, 0, - (vector unsigned char *)&rx_pkts[pos + 1] + (__vector unsigned char *)&rx_pkts[pos + 1] ->rx_descriptor_fields1 ); vec_st(pkt_mb1, 0, - (vector unsigned char *)&rx_pkts[pos]->rx_descriptor_fields1 + (__vector unsigned char *)&rx_pkts[pos]->rx_descriptor_fields1 ); desc_to_ptype_v(descs, &rx_pkts[pos], ptype_tbl); desc_to_olflags_v(descs, &rx_pkts[pos]); /* C.4 calc available number of desc */ var = __builtin_popcountll((vec_ld(0, - (vector unsigned long *)&staterr)[0])); + (__vector unsigned long *)&staterr)[0])); nb_pkts_recd += var; if (likely(var != RTE_I40E_DESCS_PER_LOOP)) break; @@ -533,9 +533,9 @@ vtx1(volatile struct i40e_tx_desc *txdp, ((uint64_t)flags << I40E_TXD_QW1_CMD_SHIFT) | ((uint64_t)pkt->data_len << I40E_TXD_QW1_TX_BUF_SZ_SHIFT)); - vector unsigned long descriptor = (vector unsigned long){ + __vector unsigned long descriptor = (__vector unsigned long){ pkt->buf_iova + pkt->data_off, high_qw}; - *(vector unsigned long *)txdp = descriptor; + *(__vector unsigned long *)txdp = descriptor; } static inline void diff --git a/drivers/net/mlx5/mlx5_rxtx_vec_altivec.h b/drivers/net/mlx5/mlx5_rxtx_vec_altivec.h index f6e434c165..683a8f9a6c 100644 --- a/drivers/net/mlx5/mlx5_rxtx_vec_altivec.h +++ b/drivers/net/mlx5/mlx5_rxtx_vec_altivec.h @@ -47,11 +47,11 @@ rxq_copy_mbuf_v(struct rte_mbuf **elts, struct rte_mbuf **pkts, uint16_t n) uint16_t p = n & -2; for (pos = 0; pos < p; pos += 2) { - vector unsigned char mbp; + __vector unsigned char mbp; - mbp = (vector unsigned char)vec_vsx_ld(0, + mbp = (__vector unsigned char)vec_vsx_ld(0, (signed int const *)&elts[pos]); - *(vector unsigned char *)&pkts[pos] = mbp; + *(__vector unsigned char *)&pkts[pos] = mbp; } if (n & 1) pkts[pos] = elts[pos]; @@ -78,15 +78,15 @@ rxq_cq_decompress_v(struct mlx5_rxq_data *rxq, volatile struct mlx5_cqe *cq, { volatile struct mlx5_mini_cqe8 *mcq = (void *)&(cq + 1)->pkt_info; struct rte_mbuf *t_pkt = elts[0]; /* Title packet is pre-built. */ - const vector unsigned char zero = (vector unsigned char){0}; + const __vector unsigned char zero = (__vector unsigned char){0}; /* Mask to shuffle from extracted mini CQE to mbuf. */ - const vector unsigned char shuf_mask1 = (vector unsigned char){ + const __vector unsigned char shuf_mask1 = (__vector unsigned char){ -1, -1, -1, -1, /* skip packet_type */ 7, 6, -1, -1, /* bswap16, pkt_len */ 7, 6, /* bswap16, data_len */ -1, -1, /* skip vlan_tci */ 3, 2, 1, 0}; /* bswap32, rss */ - const vector unsigned char shuf_mask2 = (vector unsigned char){ + const __vector unsigned char shuf_mask2 = (__vector unsigned char){ -1, -1, -1, -1, /* skip packet_type */ 15, 14, -1, -1, /* bswap16, pkt_len */ 15, 14, /* data_len, bswap16 */ @@ -95,30 +95,30 @@ rxq_cq_decompress_v(struct mlx5_rxq_data *rxq, volatile struct mlx5_cqe *cq, /* Restore the compressed count. Must be 16 bits. */ const uint16_t mcqe_n = t_pkt->data_len + (rxq->crc_present * RTE_ETHER_CRC_LEN); - const vector unsigned char rearm = - (vector unsigned char)vec_vsx_ld(0, + const __vector unsigned char rearm = + (__vector unsigned char)vec_vsx_ld(0, (signed int const *)&t_pkt->rearm_data); - const vector unsigned char rxdf = - (vector unsigned char)vec_vsx_ld(0, + const __vector unsigned char rxdf = + (__vector unsigned char)vec_vsx_ld(0, (signed int const *)&t_pkt->rx_descriptor_fields1); - const vector unsigned char crc_adj = - (vector unsigned char)(vector unsigned short){ + const __vector unsigned char crc_adj = + (__vector unsigned char)(__vector unsigned short){ 0, 0, rxq->crc_present * RTE_ETHER_CRC_LEN, 0, rxq->crc_present * RTE_ETHER_CRC_LEN, 0, 0, 0}; - const vector unsigned short rxdf_sel_mask = - (vector unsigned short){ + const __vector unsigned short rxdf_sel_mask = + (__vector unsigned short){ 0xffff, 0xffff, 0, 0, 0, 0xffff, 0, 0}; - vector unsigned char ol_flags = (vector unsigned char){0}; - vector unsigned char ol_flags_mask = (vector unsigned char){0}; + __vector unsigned char ol_flags = (__vector unsigned char){0}; + __vector unsigned char ol_flags_mask = (__vector unsigned char){0}; unsigned int pos; unsigned int i; unsigned int inv = 0; #ifdef MLX5_PMD_SOFT_COUNTERS - const vector unsigned char ones = vec_splat_u8(-1); + const __vector unsigned char ones = vec_splat_u8(-1); uint32_t rcvd_byte = 0; /* Mask to shuffle byte_cnt to add up stats. Do bswap16 for all. */ - const vector unsigned char len_shuf_mask = (vector unsigned char){ + const __vector unsigned char len_shuf_mask = (__vector unsigned char){ 3, 2, 11, 10, 7, 6, 15, 14, -1, -1, -1, -1, @@ -133,125 +133,125 @@ rxq_cq_decompress_v(struct mlx5_rxq_data *rxq, volatile struct mlx5_cqe *cq, * E. store flow tag (rte_flow mark). */ for (pos = 0; pos < mcqe_n; ) { - vector unsigned char mcqe1, mcqe2; - vector unsigned char rxdf1, rxdf2; + __vector unsigned char mcqe1, mcqe2; + __vector unsigned char rxdf1, rxdf2; #ifdef MLX5_PMD_SOFT_COUNTERS - const vector unsigned short mcqe_sel_mask = - (vector unsigned short){0, 0, 0xffff, 0xffff, + const __vector unsigned short mcqe_sel_mask = + (__vector unsigned short){0, 0, 0xffff, 0xffff, 0, 0, 0xfff, 0xffff}; - const vector unsigned char lower_half = { + const __vector unsigned char lower_half = { 0, 1, 4, 5, 8, 9, 12, 13, 16, 17, 20, 21, 24, 25, 28, 29}; - const vector unsigned char upper_half = { + const __vector unsigned char upper_half = { 2, 3, 6, 7, 10, 11, 14, 15, 18, 19, 22, 23, 26, 27, 30, 31}; - vector unsigned short left, right; - vector unsigned char byte_cnt, invalid_mask; - vector unsigned long lshift; + __vector unsigned short left, right; + __vector unsigned char byte_cnt, invalid_mask; + __vector unsigned long lshift; __attribute__((altivec(vector__))) __attribute__((altivec(bool__))) unsigned long long shmask; - const vector unsigned long shmax = {64, 64}; + const __vector unsigned long shmax = {64, 64}; #endif for (i = 0; i < MLX5_VPMD_DESCS_PER_LOOP; ++i) if (likely(pos + i < mcqe_n)) rte_prefetch0((void *)(cq + pos + i)); /* A.1 load mCQEs into a 128bit register. */ - mcqe1 = (vector unsigned char)vec_vsx_ld(0, + mcqe1 = (__vector unsigned char)vec_vsx_ld(0, (signed int const *)&mcq[pos % 8]); - mcqe2 = (vector unsigned char)vec_vsx_ld(0, + mcqe2 = (__vector unsigned char)vec_vsx_ld(0, (signed int const *)&mcq[pos % 8 + 2]); /* B.1 store rearm data to mbuf. */ - *(vector unsigned char *) + *(__vector unsigned char *) &elts[pos]->rearm_data = rearm; - *(vector unsigned char *) + *(__vector unsigned char *) &elts[pos + 1]->rearm_data = rearm; /* C.1 combine data from mCQEs with rx_descriptor_fields1. */ rxdf1 = vec_perm(mcqe1, zero, shuf_mask1); rxdf2 = vec_perm(mcqe1, zero, shuf_mask2); - rxdf1 = (vector unsigned char) - ((vector unsigned short)rxdf1 - - (vector unsigned short)crc_adj); - rxdf2 = (vector unsigned char) - ((vector unsigned short)rxdf2 - - (vector unsigned short)crc_adj); - rxdf1 = (vector unsigned char) - vec_sel((vector unsigned short)rxdf1, - (vector unsigned short)rxdf, rxdf_sel_mask); - rxdf2 = (vector unsigned char) - vec_sel((vector unsigned short)rxdf2, - (vector unsigned short)rxdf, rxdf_sel_mask); + rxdf1 = (__vector unsigned char) + ((__vector unsigned short)rxdf1 - + (__vector unsigned short)crc_adj); + rxdf2 = (__vector unsigned char) + ((__vector unsigned short)rxdf2 - + (__vector unsigned short)crc_adj); + rxdf1 = (__vector unsigned char) + vec_sel((__vector unsigned short)rxdf1, + (__vector unsigned short)rxdf, rxdf_sel_mask); + rxdf2 = (__vector unsigned char) + vec_sel((__vector unsigned short)rxdf2, + (__vector unsigned short)rxdf, rxdf_sel_mask); /* D.1 store rx_descriptor_fields1. */ - *(vector unsigned char *) + *(__vector unsigned char *) &elts[pos]->rx_descriptor_fields1 = rxdf1; - *(vector unsigned char *) + *(__vector unsigned char *) &elts[pos + 1]->rx_descriptor_fields1 = rxdf2; /* B.1 store rearm data to mbuf. */ - *(vector unsigned char *) + *(__vector unsigned char *) &elts[pos + 2]->rearm_data = rearm; - *(vector unsigned char *) + *(__vector unsigned char *) &elts[pos + 3]->rearm_data = rearm; /* C.1 combine data from mCQEs with rx_descriptor_fields1. */ rxdf1 = vec_perm(mcqe2, zero, shuf_mask1); rxdf2 = vec_perm(mcqe2, zero, shuf_mask2); - rxdf1 = (vector unsigned char) - ((vector unsigned short)rxdf1 - - (vector unsigned short)crc_adj); - rxdf2 = (vector unsigned char) - ((vector unsigned short)rxdf2 - - (vector unsigned short)crc_adj); - rxdf1 = (vector unsigned char) - vec_sel((vector unsigned short)rxdf1, - (vector unsigned short)rxdf, rxdf_sel_mask); - rxdf2 = (vector unsigned char) - vec_sel((vector unsigned short)rxdf2, - (vector unsigned short)rxdf, rxdf_sel_mask); + rxdf1 = (__vector unsigned char) + ((__vector unsigned short)rxdf1 - + (__vector unsigned short)crc_adj); + rxdf2 = (__vector unsigned char) + ((__vector unsigned short)rxdf2 - + (__vector unsigned short)crc_adj); + rxdf1 = (__vector unsigned char) + vec_sel((__vector unsigned short)rxdf1, + (__vector unsigned short)rxdf, rxdf_sel_mask); + rxdf2 = (__vector unsigned char) + vec_sel((__vector unsigned short)rxdf2, + (__vector unsigned short)rxdf, rxdf_sel_mask); /* D.1 store rx_descriptor_fields1. */ - *(vector unsigned char *) + *(__vector unsigned char *) &elts[pos + 2]->rx_descriptor_fields1 = rxdf1; - *(vector unsigned char *) + *(__vector unsigned char *) &elts[pos + 3]->rx_descriptor_fields1 = rxdf2; #ifdef MLX5_PMD_SOFT_COUNTERS - invalid_mask = (vector unsigned char)(vector unsigned long){ + invalid_mask = (__vector unsigned char)(__vector unsigned long){ (mcqe_n - pos) * sizeof(uint16_t) * 8, 0}; lshift = - vec_splat((vector unsigned long)invalid_mask, 0); + vec_splat((__vector unsigned long)invalid_mask, 0); shmask = vec_cmpgt(shmax, lshift); - invalid_mask = (vector unsigned char) - vec_sl((vector unsigned long)ones, lshift); - invalid_mask = (vector unsigned char) - vec_sel((vector unsigned long)shmask, - (vector unsigned long)invalid_mask, shmask); + invalid_mask = (__vector unsigned char) + vec_sl((__vector unsigned long)ones, lshift); + invalid_mask = (__vector unsigned char) + vec_sel((__vector unsigned long)shmask, + (__vector unsigned long)invalid_mask, shmask); - byte_cnt = (vector unsigned char) - vec_sel((vector unsigned short) - vec_sro((vector unsigned short)mcqe1, - (vector unsigned char){32}), - (vector unsigned short)mcqe2, mcqe_sel_mask); + byte_cnt = (__vector unsigned char) + vec_sel((__vector unsigned short) + vec_sro((__vector unsigned short)mcqe1, + (__vector unsigned char){32}), + (__vector unsigned short)mcqe2, mcqe_sel_mask); byte_cnt = vec_perm(byte_cnt, zero, len_shuf_mask); - byte_cnt = (vector unsigned char) - vec_andc((vector unsigned long)byte_cnt, - (vector unsigned long)invalid_mask); - left = vec_perm((vector unsigned short)byte_cnt, - (vector unsigned short)zero, lower_half); - right = vec_perm((vector unsigned short)byte_cnt, - (vector unsigned short)zero, upper_half); - byte_cnt = (vector unsigned char)vec_add(left, right); - left = vec_perm((vector unsigned short)byte_cnt, - (vector unsigned short)zero, lower_half); - right = vec_perm((vector unsigned short)byte_cnt, - (vector unsigned short)zero, upper_half); - byte_cnt = (vector unsigned char)vec_add(left, right); - rcvd_byte += ((vector unsigned long)byte_cnt)[0]; + byte_cnt = (__vector unsigned char) + vec_andc((__vector unsigned long)byte_cnt, + (__vector unsigned long)invalid_mask); + left = vec_perm((__vector unsigned short)byte_cnt, + (__vector unsigned short)zero, lower_half); + right = vec_perm((__vector unsigned short)byte_cnt, + (__vector unsigned short)zero, upper_half); + byte_cnt = (__vector unsigned char)vec_add(left, right); + left = vec_perm((__vector unsigned short)byte_cnt, + (__vector unsigned short)zero, lower_half); + right = vec_perm((__vector unsigned short)byte_cnt, + (__vector unsigned short)zero, upper_half); + byte_cnt = (__vector unsigned char)vec_add(left, right); + rcvd_byte += ((__vector unsigned long)byte_cnt)[0]; #endif if (rxq->mark) { @@ -265,99 +265,99 @@ rxq_cq_decompress_v(struct mlx5_rxq_data *rxq, volatile struct mlx5_cqe *cq, elts[pos + 2]->hash.fdir.hi = flow_tag; elts[pos + 3]->hash.fdir.hi = flow_tag; } else { - const vector unsigned char flow_mark_adj = - (vector unsigned char) - (vector unsigned int){ + const __vector unsigned char flow_mark_adj = + (__vector unsigned char) + (__vector unsigned int){ -1, -1, -1, -1}; - const vector unsigned char flow_mark_shuf = - (vector unsigned char){ + const __vector unsigned char flow_mark_shuf = + (__vector unsigned char){ -1, -1, -1, -1, -1, -1, -1, -1, 12, 8, 9, -1, 4, 0, 1, -1}; - const vector unsigned char ft_mask = - (vector unsigned char) - (vector unsigned int){ + const __vector unsigned char ft_mask = + (__vector unsigned char) + (__vector unsigned int){ 0xffffff00, 0xffffff00, 0xffffff00, 0xffffff00}; - const vector unsigned char fdir_flags = - (vector unsigned char) - (vector unsigned int){ + const __vector unsigned char fdir_flags = + (__vector unsigned char) + (__vector unsigned int){ RTE_MBUF_F_RX_FDIR, RTE_MBUF_F_RX_FDIR, RTE_MBUF_F_RX_FDIR, RTE_MBUF_F_RX_FDIR}; - const vector unsigned char fdir_all_flags = - (vector unsigned char) - (vector unsigned int){ + const __vector unsigned char fdir_all_flags = + (__vector unsigned char) + (__vector unsigned int){ RTE_MBUF_F_RX_FDIR | RTE_MBUF_F_RX_FDIR_ID, RTE_MBUF_F_RX_FDIR | RTE_MBUF_F_RX_FDIR_ID, RTE_MBUF_F_RX_FDIR | RTE_MBUF_F_RX_FDIR_ID, RTE_MBUF_F_RX_FDIR | RTE_MBUF_F_RX_FDIR_ID}; - vector unsigned char fdir_id_flags = - (vector unsigned char) - (vector unsigned int){ + __vector unsigned char fdir_id_flags = + (__vector unsigned char) + (__vector unsigned int){ RTE_MBUF_F_RX_FDIR_ID, RTE_MBUF_F_RX_FDIR_ID, RTE_MBUF_F_RX_FDIR_ID, RTE_MBUF_F_RX_FDIR_ID}; /* Extract flow_tag field. */ - vector unsigned char ftag0 = vec_perm(mcqe1, + __vector unsigned char ftag0 = vec_perm(mcqe1, zero, flow_mark_shuf); - vector unsigned char ftag1 = vec_perm(mcqe2, + __vector unsigned char ftag1 = vec_perm(mcqe2, zero, flow_mark_shuf); - vector unsigned char ftag = - (vector unsigned char) - vec_mergel((vector unsigned int)ftag0, - (vector unsigned int)ftag1); - vector unsigned char invalid_mask = - (vector unsigned char) - vec_cmpeq((vector unsigned int)ftag, - (vector unsigned int)zero); + __vector unsigned char ftag = + (__vector unsigned char) + vec_mergel((__vector unsigned int)ftag0, + (__vector unsigned int)ftag1); + __vector unsigned char invalid_mask = + (__vector unsigned char) + vec_cmpeq((__vector unsigned int)ftag, + (__vector unsigned int)zero); - ol_flags_mask = (vector unsigned char) - vec_or((vector unsigned long) + ol_flags_mask = (__vector unsigned char) + vec_or((__vector unsigned long) ol_flags_mask, - (vector unsigned long)fdir_all_flags); + (__vector unsigned long)fdir_all_flags); /* Set RTE_MBUF_F_RX_FDIR if flow tag is non-zero. */ - invalid_mask = (vector unsigned char) - vec_cmpeq((vector unsigned int)ftag, - (vector unsigned int)zero); - ol_flags = (vector unsigned char) - vec_or((vector unsigned long)ol_flags, - (vector unsigned long) - vec_andc((vector unsigned long) + invalid_mask = (__vector unsigned char) + vec_cmpeq((__vector unsigned int)ftag, + (__vector unsigned int)zero); + ol_flags = (__vector unsigned char) + vec_or((__vector unsigned long)ol_flags, + (__vector unsigned long) + vec_andc((__vector unsigned long) fdir_flags, - (vector unsigned long)invalid_mask)); - ol_flags_mask = (vector unsigned char) - vec_or((vector unsigned long) + (__vector unsigned long)invalid_mask)); + ol_flags_mask = (__vector unsigned char) + vec_or((__vector unsigned long) ol_flags_mask, - (vector unsigned long)fdir_flags); + (__vector unsigned long)fdir_flags); /* Mask out invalid entries. */ - fdir_id_flags = (vector unsigned char) - vec_andc((vector unsigned long) + fdir_id_flags = (__vector unsigned char) + vec_andc((__vector unsigned long) fdir_id_flags, - (vector unsigned long)invalid_mask); + (__vector unsigned long)invalid_mask); /* Check if flow tag MLX5_FLOW_MARK_DEFAULT. */ - ol_flags = (vector unsigned char) - vec_or((vector unsigned long)ol_flags, - (vector unsigned long) - vec_andc((vector unsigned long) + ol_flags = (__vector unsigned char) + vec_or((__vector unsigned long)ol_flags, + (__vector unsigned long) + vec_andc((__vector unsigned long) fdir_id_flags, - (vector unsigned long) - vec_cmpeq((vector unsigned int)ftag, - (vector unsigned int)ft_mask))); + (__vector unsigned long) + vec_cmpeq((__vector unsigned int)ftag, + (__vector unsigned int)ft_mask))); - ftag = (vector unsigned char) - ((vector unsigned int)ftag + - (vector unsigned int)flow_mark_adj); + ftag = (__vector unsigned char) + ((__vector unsigned int)ftag + + (__vector unsigned int)flow_mark_adj); elts[pos]->hash.fdir.hi = - ((vector unsigned int)ftag)[0]; + ((__vector unsigned int)ftag)[0]; elts[pos + 1]->hash.fdir.hi = - ((vector unsigned int)ftag)[1]; + ((__vector unsigned int)ftag)[1]; elts[pos + 2]->hash.fdir.hi = - ((vector unsigned int)ftag)[2]; + ((__vector unsigned int)ftag)[2]; elts[pos + 3]->hash.fdir.hi = - ((vector unsigned int)ftag)[3]; + ((__vector unsigned int)ftag)[3]; } } if (unlikely(rxq->mcqe_format != MLX5_CQE_RESP_FORMAT_HASH)) { @@ -373,37 +373,37 @@ rxq_cq_decompress_v(struct mlx5_rxq_data *rxq, volatile struct mlx5_cqe *cq, mcq[pos % 8 + 2].hdr_type; const uint8_t pkt_hdr3 = mcq[pos % 8 + 3].hdr_type; - const vector unsigned char vlan_mask = - (vector unsigned char) - (vector unsigned int) { + const __vector unsigned char vlan_mask = + (__vector unsigned char) + (__vector unsigned int) { (RTE_MBUF_F_RX_VLAN | RTE_MBUF_F_RX_VLAN_STRIPPED), (RTE_MBUF_F_RX_VLAN | RTE_MBUF_F_RX_VLAN_STRIPPED), (RTE_MBUF_F_RX_VLAN | RTE_MBUF_F_RX_VLAN_STRIPPED), (RTE_MBUF_F_RX_VLAN | RTE_MBUF_F_RX_VLAN_STRIPPED)}; - const vector unsigned char cv_mask = - (vector unsigned char) - (vector unsigned int) { + const __vector unsigned char cv_mask = + (__vector unsigned char) + (__vector unsigned int) { MLX5_CQE_VLAN_STRIPPED, MLX5_CQE_VLAN_STRIPPED, MLX5_CQE_VLAN_STRIPPED, MLX5_CQE_VLAN_STRIPPED}; - vector unsigned char pkt_cv = - (vector unsigned char) - (vector unsigned int) { + __vector unsigned char pkt_cv = + (__vector unsigned char) + (__vector unsigned int) { pkt_hdr0 & 0x1, pkt_hdr1 & 0x1, pkt_hdr2 & 0x1, pkt_hdr3 & 0x1}; - ol_flags_mask = (vector unsigned char) - vec_or((vector unsigned long) + ol_flags_mask = (__vector unsigned char) + vec_or((__vector unsigned long) ol_flags_mask, - (vector unsigned long)vlan_mask); - ol_flags = (vector unsigned char) - vec_or((vector unsigned long)ol_flags, - (vector unsigned long) - vec_and((vector unsigned long)vlan_mask, - (vector unsigned long) - vec_cmpeq((vector unsigned int)pkt_cv, - (vector unsigned int)cv_mask))); + (__vector unsigned long)vlan_mask); + ol_flags = (__vector unsigned char) + vec_or((__vector unsigned long)ol_flags, + (__vector unsigned long) + vec_and((__vector unsigned long)vlan_mask, + (__vector unsigned long) + vec_cmpeq((__vector unsigned int)pkt_cv, + (__vector unsigned int)cv_mask))); elts[pos]->packet_type = mlx5_ptype_table[(pkt_hdr0 >> 2) | pkt_info]; @@ -431,36 +431,36 @@ rxq_cq_decompress_v(struct mlx5_rxq_data *rxq, volatile struct mlx5_cqe *cq, pkt_info) & (1 << 6)); } } - const vector unsigned char hash_mask = - (vector unsigned char)(vector unsigned int) { + const __vector unsigned char hash_mask = + (__vector unsigned char)(__vector unsigned int) { RTE_MBUF_F_RX_RSS_HASH, RTE_MBUF_F_RX_RSS_HASH, RTE_MBUF_F_RX_RSS_HASH, RTE_MBUF_F_RX_RSS_HASH}; - const vector unsigned char rearm_flags = - (vector unsigned char)(vector unsigned int) { + const __vector unsigned char rearm_flags = + (__vector unsigned char)(__vector unsigned int) { (uint32_t)t_pkt->ol_flags, (uint32_t)t_pkt->ol_flags, (uint32_t)t_pkt->ol_flags, (uint32_t)t_pkt->ol_flags}; - ol_flags_mask = (vector unsigned char) - vec_or((vector unsigned long)ol_flags_mask, - (vector unsigned long)hash_mask); - ol_flags = (vector unsigned char) - vec_or((vector unsigned long)ol_flags, - (vector unsigned long) - vec_andc((vector unsigned long)rearm_flags, - (vector unsigned long)ol_flags_mask)); + ol_flags_mask = (__vector unsigned char) + vec_or((__vector unsigned long)ol_flags_mask, + (__vector unsigned long)hash_mask); + ol_flags = (__vector unsigned char) + vec_or((__vector unsigned long)ol_flags, + (__vector unsigned long) + vec_andc((__vector unsigned long)rearm_flags, + (__vector unsigned long)ol_flags_mask)); elts[pos]->ol_flags = - ((vector unsigned int)ol_flags)[0]; + ((__vector unsigned int)ol_flags)[0]; elts[pos + 1]->ol_flags = - ((vector unsigned int)ol_flags)[1]; + ((__vector unsigned int)ol_flags)[1]; elts[pos + 2]->ol_flags = - ((vector unsigned int)ol_flags)[2]; + ((__vector unsigned int)ol_flags)[2]; elts[pos + 3]->ol_flags = - ((vector unsigned int)ol_flags)[3]; + ((__vector unsigned int)ol_flags)[3]; elts[pos]->hash.rss = 0; elts[pos + 1]->hash.rss = 0; elts[pos + 2]->hash.rss = 0; @@ -524,13 +524,13 @@ rxq_cq_decompress_v(struct mlx5_rxq_data *rxq, volatile struct mlx5_cqe *cq, */ static inline void rxq_cq_to_ptype_oflags_v(struct mlx5_rxq_data *rxq, - vector unsigned char cqes[4], vector unsigned char op_err, + __vector unsigned char cqes[4], __vector unsigned char op_err, struct rte_mbuf **pkts) { - vector unsigned char pinfo0, pinfo1; - vector unsigned char pinfo, ptype; - vector unsigned char ol_flags = (vector unsigned char) - (vector unsigned int){ + __vector unsigned char pinfo0, pinfo1; + __vector unsigned char pinfo, ptype; + __vector unsigned char ol_flags = (__vector unsigned char) + (__vector unsigned int){ rxq->rss_hash * RTE_MBUF_F_RX_RSS_HASH | rxq->hw_timestamp * rxq->timestamp_rx_flag, rxq->rss_hash * RTE_MBUF_F_RX_RSS_HASH | @@ -539,25 +539,25 @@ rxq_cq_to_ptype_oflags_v(struct mlx5_rxq_data *rxq, rxq->hw_timestamp * rxq->timestamp_rx_flag, rxq->rss_hash * RTE_MBUF_F_RX_RSS_HASH | rxq->hw_timestamp * rxq->timestamp_rx_flag}; - vector unsigned char cv_flags; - const vector unsigned char zero = (vector unsigned char){0}; - const vector unsigned char ptype_mask = - (vector unsigned char)(vector unsigned int){ + __vector unsigned char cv_flags; + const __vector unsigned char zero = (__vector unsigned char){0}; + const __vector unsigned char ptype_mask = + (__vector unsigned char)(__vector unsigned int){ 0x0000fd06, 0x0000fd06, 0x0000fd06, 0x0000fd06}; - const vector unsigned char ptype_ol_mask = - (vector unsigned char)(vector unsigned int){ + const __vector unsigned char ptype_ol_mask = + (__vector unsigned char)(__vector unsigned int){ 0x00000106, 0x00000106, 0x00000106, 0x00000106}; - const vector unsigned char pinfo_mask = - (vector unsigned char)(vector unsigned int){ + const __vector unsigned char pinfo_mask = + (__vector unsigned char)(__vector unsigned int){ 0x00000003, 0x00000003, 0x00000003, 0x00000003}; - const vector unsigned char cv_flag_sel = (vector unsigned char){ + const __vector unsigned char cv_flag_sel = (__vector unsigned char){ 0, (uint8_t)(RTE_MBUF_F_RX_VLAN | RTE_MBUF_F_RX_VLAN_STRIPPED), (uint8_t)(RTE_MBUF_F_RX_IP_CKSUM_GOOD >> 1), 0, (uint8_t)(RTE_MBUF_F_RX_L4_CKSUM_GOOD >> 1), 0, (uint8_t)((RTE_MBUF_F_RX_IP_CKSUM_GOOD | RTE_MBUF_F_RX_L4_CKSUM_GOOD) >> 1), 0, 0, 0, 0, 0, 0, 0, 0, 0}; - const vector unsigned char cv_mask = - (vector unsigned char)(vector unsigned int){ + const __vector unsigned char cv_mask = + (__vector unsigned char)(__vector unsigned int){ RTE_MBUF_F_RX_IP_CKSUM_GOOD | RTE_MBUF_F_RX_L4_CKSUM_GOOD | RTE_MBUF_F_RX_VLAN | RTE_MBUF_F_RX_VLAN_STRIPPED, RTE_MBUF_F_RX_IP_CKSUM_GOOD | RTE_MBUF_F_RX_L4_CKSUM_GOOD | @@ -566,77 +566,77 @@ rxq_cq_to_ptype_oflags_v(struct mlx5_rxq_data *rxq, RTE_MBUF_F_RX_VLAN | RTE_MBUF_F_RX_VLAN_STRIPPED, RTE_MBUF_F_RX_IP_CKSUM_GOOD | RTE_MBUF_F_RX_L4_CKSUM_GOOD | RTE_MBUF_F_RX_VLAN | RTE_MBUF_F_RX_VLAN_STRIPPED}; - const vector unsigned char mbuf_init = - (vector unsigned char)vec_vsx_ld - (0, (vector unsigned char *)&rxq->mbuf_initializer); - const vector unsigned short rearm_sel_mask = - (vector unsigned short){0, 0, 0, 0, 0xffff, 0xffff, 0, 0}; - vector unsigned char rearm0, rearm1, rearm2, rearm3; + const __vector unsigned char mbuf_init = + (__vector unsigned char)vec_vsx_ld + (0, (__vector unsigned char *)&rxq->mbuf_initializer); + const __vector unsigned short rearm_sel_mask = + (__vector unsigned short){0, 0, 0, 0, 0xffff, 0xffff, 0, 0}; + __vector unsigned char rearm0, rearm1, rearm2, rearm3; uint8_t pt_idx0, pt_idx1, pt_idx2, pt_idx3; /* Extract pkt_info field. */ - pinfo0 = (vector unsigned char) - vec_mergeh((vector unsigned int)cqes[0], - (vector unsigned int)cqes[1]); - pinfo1 = (vector unsigned char) - vec_mergeh((vector unsigned int)cqes[2], - (vector unsigned int)cqes[3]); - pinfo = (vector unsigned char) - vec_mergeh((vector unsigned long)pinfo0, - (vector unsigned long)pinfo1); + pinfo0 = (__vector unsigned char) + vec_mergeh((__vector unsigned int)cqes[0], + (__vector unsigned int)cqes[1]); + pinfo1 = (__vector unsigned char) + vec_mergeh((__vector unsigned int)cqes[2], + (__vector unsigned int)cqes[3]); + pinfo = (__vector unsigned char) + vec_mergeh((__vector unsigned long)pinfo0, + (__vector unsigned long)pinfo1); /* Extract hdr_type_etc field. */ - pinfo0 = (vector unsigned char) - vec_mergel((vector unsigned int)cqes[0], - (vector unsigned int)cqes[1]); - pinfo1 = (vector unsigned char) - vec_mergel((vector unsigned int)cqes[2], - (vector unsigned int)cqes[3]); - ptype = (vector unsigned char) - vec_mergeh((vector unsigned long)pinfo0, - (vector unsigned long)pinfo1); + pinfo0 = (__vector unsigned char) + vec_mergel((__vector unsigned int)cqes[0], + (__vector unsigned int)cqes[1]); + pinfo1 = (__vector unsigned char) + vec_mergel((__vector unsigned int)cqes[2], + (__vector unsigned int)cqes[3]); + ptype = (__vector unsigned char) + vec_mergeh((__vector unsigned long)pinfo0, + (__vector unsigned long)pinfo1); if (rxq->mark) { - const vector unsigned char pinfo_ft_mask = - (vector unsigned char)(vector unsigned int){ + const __vector unsigned char pinfo_ft_mask = + (__vector unsigned char)(__vector unsigned int){ 0xffffff00, 0xffffff00, 0xffffff00, 0xffffff00}; - const vector unsigned char fdir_flags = - (vector unsigned char)(vector unsigned int){ + const __vector unsigned char fdir_flags = + (__vector unsigned char)(__vector unsigned int){ RTE_MBUF_F_RX_FDIR, RTE_MBUF_F_RX_FDIR, RTE_MBUF_F_RX_FDIR, RTE_MBUF_F_RX_FDIR}; - vector unsigned char fdir_id_flags = - (vector unsigned char)(vector unsigned int){ + __vector unsigned char fdir_id_flags = + (__vector unsigned char)(__vector unsigned int){ RTE_MBUF_F_RX_FDIR_ID, RTE_MBUF_F_RX_FDIR_ID, RTE_MBUF_F_RX_FDIR_ID, RTE_MBUF_F_RX_FDIR_ID}; - vector unsigned char flow_tag, invalid_mask; + __vector unsigned char flow_tag, invalid_mask; - flow_tag = (vector unsigned char) - vec_and((vector unsigned long)pinfo, - (vector unsigned long)pinfo_ft_mask); + flow_tag = (__vector unsigned char) + vec_and((__vector unsigned long)pinfo, + (__vector unsigned long)pinfo_ft_mask); /* Check if flow tag is non-zero then set RTE_MBUF_F_RX_FDIR. */ - invalid_mask = (vector unsigned char) - vec_cmpeq((vector unsigned int)flow_tag, - (vector unsigned int)zero); - ol_flags = (vector unsigned char) - vec_or((vector unsigned long)ol_flags, - (vector unsigned long) - vec_andc((vector unsigned long)fdir_flags, - (vector unsigned long)invalid_mask)); + invalid_mask = (__vector unsigned char) + vec_cmpeq((__vector unsigned int)flow_tag, + (__vector unsigned int)zero); + ol_flags = (__vector unsigned char) + vec_or((__vector unsigned long)ol_flags, + (__vector unsigned long) + vec_andc((__vector unsigned long)fdir_flags, + (__vector unsigned long)invalid_mask)); /* Mask out invalid entries. */ - fdir_id_flags = (vector unsigned char) - vec_andc((vector unsigned long)fdir_id_flags, - (vector unsigned long)invalid_mask); + fdir_id_flags = (__vector unsigned char) + vec_andc((__vector unsigned long)fdir_id_flags, + (__vector unsigned long)invalid_mask); /* Check if flow tag MLX5_FLOW_MARK_DEFAULT. */ - ol_flags = (vector unsigned char) - vec_or((vector unsigned long)ol_flags, - (vector unsigned long) - vec_andc((vector unsigned long)fdir_id_flags, - (vector unsigned long) - vec_cmpeq((vector unsigned int)flow_tag, - (vector unsigned int)pinfo_ft_mask))); + ol_flags = (__vector unsigned char) + vec_or((__vector unsigned long)ol_flags, + (__vector unsigned long) + vec_andc((__vector unsigned long)fdir_id_flags, + (__vector unsigned long) + vec_cmpeq((__vector unsigned int)flow_tag, + (__vector unsigned int)pinfo_ft_mask))); } /* * Merge the two fields to generate the following: @@ -649,39 +649,39 @@ rxq_cq_to_ptype_oflags_v(struct mlx5_rxq_data *rxq, * bit[16] = tunneled * bit[17] = outer_l3_type */ - ptype = (vector unsigned char) - vec_and((vector unsigned long)ptype, - (vector unsigned long)ptype_mask); - pinfo = (vector unsigned char) - vec_and((vector unsigned long)pinfo, - (vector unsigned long)pinfo_mask); - pinfo = (vector unsigned char) - vec_sl((vector unsigned int)pinfo, - (vector unsigned int){16, 16, 16, 16}); + ptype = (__vector unsigned char) + vec_and((__vector unsigned long)ptype, + (__vector unsigned long)ptype_mask); + pinfo = (__vector unsigned char) + vec_and((__vector unsigned long)pinfo, + (__vector unsigned long)pinfo_mask); + pinfo = (__vector unsigned char) + vec_sl((__vector unsigned int)pinfo, + (__vector unsigned int){16, 16, 16, 16}); /* Make pinfo has merged fields for ol_flags calculation. */ - pinfo = (vector unsigned char) - vec_or((vector unsigned long)ptype, - (vector unsigned long)pinfo); - ptype = (vector unsigned char) - vec_sr((vector unsigned int)pinfo, - (vector unsigned int){10, 10, 10, 10}); - ptype = (vector unsigned char) - vec_packs((vector unsigned int)ptype, - (vector unsigned int)zero); + pinfo = (__vector unsigned char) + vec_or((__vector unsigned long)ptype, + (__vector unsigned long)pinfo); + ptype = (__vector unsigned char) + vec_sr((__vector unsigned int)pinfo, + (__vector unsigned int){10, 10, 10, 10}); + ptype = (__vector unsigned char) + vec_packs((__vector unsigned int)ptype, + (__vector unsigned int)zero); /* Errored packets will have RTE_PTYPE_ALL_MASK. */ - op_err = (vector unsigned char) - vec_sr((vector unsigned short)op_err, - (vector unsigned short){8, 8, 8, 8, 8, 8, 8, 8}); - ptype = (vector unsigned char) - vec_or((vector unsigned long)ptype, - (vector unsigned long)op_err); + op_err = (__vector unsigned char) + vec_sr((__vector unsigned short)op_err, + (__vector unsigned short){8, 8, 8, 8, 8, 8, 8, 8}); + ptype = (__vector unsigned char) + vec_or((__vector unsigned long)ptype, + (__vector unsigned long)op_err); - pt_idx0 = (uint8_t)((vector unsigned char)ptype)[0]; - pt_idx1 = (uint8_t)((vector unsigned char)ptype)[2]; - pt_idx2 = (uint8_t)((vector unsigned char)ptype)[4]; - pt_idx3 = (uint8_t)((vector unsigned char)ptype)[6]; + pt_idx0 = (uint8_t)((__vector unsigned char)ptype)[0]; + pt_idx1 = (uint8_t)((__vector unsigned char)ptype)[2]; + pt_idx2 = (uint8_t)((__vector unsigned char)ptype)[4]; + pt_idx3 = (uint8_t)((__vector unsigned char)ptype)[6]; pkts[0]->packet_type = mlx5_ptype_table[pt_idx0] | !!(pt_idx0 & (1 << 6)) * rxq->tunnel; @@ -693,63 +693,63 @@ rxq_cq_to_ptype_oflags_v(struct mlx5_rxq_data *rxq, !!(pt_idx3 & (1 << 6)) * rxq->tunnel; /* Fill flags for checksum and VLAN. */ - pinfo = (vector unsigned char) - vec_and((vector unsigned long)pinfo, - (vector unsigned long)ptype_ol_mask); + pinfo = (__vector unsigned char) + vec_and((__vector unsigned long)pinfo, + (__vector unsigned long)ptype_ol_mask); pinfo = vec_perm(cv_flag_sel, zero, pinfo); /* Locate checksum flags at byte[2:1] and merge with VLAN flags. */ - cv_flags = (vector unsigned char) - vec_sl((vector unsigned int)pinfo, - (vector unsigned int){9, 9, 9, 9}); - cv_flags = (vector unsigned char) - vec_or((vector unsigned long)pinfo, - (vector unsigned long)cv_flags); + cv_flags = (__vector unsigned char) + vec_sl((__vector unsigned int)pinfo, + (__vector unsigned int){9, 9, 9, 9}); + cv_flags = (__vector unsigned char) + vec_or((__vector unsigned long)pinfo, + (__vector unsigned long)cv_flags); /* Move back flags to start from byte[0]. */ - cv_flags = (vector unsigned char) - vec_sr((vector unsigned int)cv_flags, - (vector unsigned int){8, 8, 8, 8}); + cv_flags = (__vector unsigned char) + vec_sr((__vector unsigned int)cv_flags, + (__vector unsigned int){8, 8, 8, 8}); /* Mask out garbage bits. */ - cv_flags = (vector unsigned char) - vec_and((vector unsigned long)cv_flags, - (vector unsigned long)cv_mask); + cv_flags = (__vector unsigned char) + vec_and((__vector unsigned long)cv_flags, + (__vector unsigned long)cv_mask); /* Merge to ol_flags. */ - ol_flags = (vector unsigned char) - vec_or((vector unsigned long)ol_flags, - (vector unsigned long)cv_flags); + ol_flags = (__vector unsigned char) + vec_or((__vector unsigned long)ol_flags, + (__vector unsigned long)cv_flags); /* Merge mbuf_init and ol_flags. */ - rearm0 = (vector unsigned char) - vec_sel((vector unsigned short)mbuf_init, - (vector unsigned short) - vec_slo((vector unsigned short)ol_flags, - (vector unsigned char){64}), rearm_sel_mask); - rearm1 = (vector unsigned char) - vec_sel((vector unsigned short)mbuf_init, - (vector unsigned short) - vec_slo((vector unsigned short)ol_flags, - (vector unsigned char){32}), rearm_sel_mask); - rearm2 = (vector unsigned char) - vec_sel((vector unsigned short)mbuf_init, - (vector unsigned short)ol_flags, rearm_sel_mask); - rearm3 = (vector unsigned char) - vec_sel((vector unsigned short)mbuf_init, - (vector unsigned short) - vec_sro((vector unsigned short)ol_flags, - (vector unsigned char){32}), rearm_sel_mask); + rearm0 = (__vector unsigned char) + vec_sel((__vector unsigned short)mbuf_init, + (__vector unsigned short) + vec_slo((__vector unsigned short)ol_flags, + (__vector unsigned char){64}), rearm_sel_mask); + rearm1 = (__vector unsigned char) + vec_sel((__vector unsigned short)mbuf_init, + (__vector unsigned short) + vec_slo((__vector unsigned short)ol_flags, + (__vector unsigned char){32}), rearm_sel_mask); + rearm2 = (__vector unsigned char) + vec_sel((__vector unsigned short)mbuf_init, + (__vector unsigned short)ol_flags, rearm_sel_mask); + rearm3 = (__vector unsigned char) + vec_sel((__vector unsigned short)mbuf_init, + (__vector unsigned short) + vec_sro((__vector unsigned short)ol_flags, + (__vector unsigned char){32}), rearm_sel_mask); /* Write 8B rearm_data and 8B ol_flags. */ vec_vsx_st(rearm0, 0, - (vector unsigned char *)&pkts[0]->rearm_data); + (__vector unsigned char *)&pkts[0]->rearm_data); vec_vsx_st(rearm1, 0, - (vector unsigned char *)&pkts[1]->rearm_data); + (__vector unsigned char *)&pkts[1]->rearm_data); vec_vsx_st(rearm2, 0, - (vector unsigned char *)&pkts[2]->rearm_data); + (__vector unsigned char *)&pkts[2]->rearm_data); vec_vsx_st(rearm3, 0, - (vector unsigned char *)&pkts[3]->rearm_data); + (__vector unsigned char *)&pkts[3]->rearm_data); } /** @@ -788,31 +788,31 @@ rxq_cq_process_v(struct mlx5_rxq_data *rxq, volatile struct mlx5_cqe *cq, uint64_t comp_idx = MLX5_VPMD_DESCS_PER_LOOP; uint16_t nocmp_n = 0; unsigned int ownership = !!(rxq->cq_ci & (q_mask + 1)); - const vector unsigned char zero = (vector unsigned char){0}; - const vector unsigned char ones = vec_splat_u8(-1); - const vector unsigned char owner_check = - (vector unsigned char)(vector unsigned long){ + const __vector unsigned char zero = (__vector unsigned char){0}; + const __vector unsigned char ones = vec_splat_u8(-1); + const __vector unsigned char owner_check = + (__vector unsigned char)(__vector unsigned long){ 0x0100000001000000LL, 0x0100000001000000LL}; - const vector unsigned char opcode_check = - (vector unsigned char)(vector unsigned long){ + const __vector unsigned char opcode_check = + (__vector unsigned char)(__vector unsigned long){ 0xf0000000f0000000LL, 0xf0000000f0000000LL}; - const vector unsigned char format_check = - (vector unsigned char)(vector unsigned long){ + const __vector unsigned char format_check = + (__vector unsigned char)(__vector unsigned long){ 0x0c0000000c000000LL, 0x0c0000000c000000LL}; - const vector unsigned char resp_err_check = - (vector unsigned char)(vector unsigned long){ + const __vector unsigned char resp_err_check = + (__vector unsigned char)(__vector unsigned long){ 0xe0000000e0000000LL, 0xe0000000e0000000LL}; #ifdef MLX5_PMD_SOFT_COUNTERS uint32_t rcvd_byte = 0; /* Mask to shuffle byte_cnt to add up stats. Do bswap16 for all. */ - const vector unsigned char len_shuf_mask = (vector unsigned char){ + const __vector unsigned char len_shuf_mask = (__vector unsigned char){ 1, 0, 5, 4, 9, 8, 13, 12, -1, -1, -1, -1, -1, -1, -1, -1}; #endif /* Mask to shuffle from extracted CQE to mbuf. */ - const vector unsigned char shuf_mask = (vector unsigned char){ + const __vector unsigned char shuf_mask = (__vector unsigned char){ 5, 4, /* bswap16, pkt_len */ -1, -1, /* zero out 2nd half of pkt_len */ 5, 4, /* bswap16, data_len */ @@ -821,22 +821,22 @@ rxq_cq_process_v(struct mlx5_rxq_data *rxq, volatile struct mlx5_cqe *cq, 1, 2, 3, -1}; /* fdir.hi */ /* Mask to blend from the last Qword to the first DQword. */ /* Mask to blend from the last Qword to the first DQword. */ - const vector unsigned char blend_mask = (vector unsigned char){ + const __vector unsigned char blend_mask = (__vector unsigned char){ -1, 0, 0, 0, 0, 0, 0, 0, -1, -1, -1, -1, -1, -1, -1, -1}; - const vector unsigned char crc_adj = - (vector unsigned char)(vector unsigned short){ + const __vector unsigned char crc_adj = + (__vector unsigned char)(__vector unsigned short){ rxq->crc_present * RTE_ETHER_CRC_LEN, 0, rxq->crc_present * RTE_ETHER_CRC_LEN, 0, 0, 0, 0, 0}; - const vector unsigned char flow_mark_adj = - (vector unsigned char)(vector unsigned int){ + const __vector unsigned char flow_mark_adj = + (__vector unsigned char)(__vector unsigned int){ 0, 0, 0, rxq->mark * (-1)}; - const vector unsigned short cqe_sel_mask1 = - (vector unsigned short){0, 0, 0, 0, 0xffff, 0xffff, 0, 0}; - const vector unsigned short cqe_sel_mask2 = - (vector unsigned short){0, 0, 0xffff, 0, 0, 0, 0, 0}; + const __vector unsigned short cqe_sel_mask1 = + (__vector unsigned short){0, 0, 0, 0, 0xffff, 0xffff, 0, 0}; + const __vector unsigned short cqe_sel_mask2 = + (__vector unsigned short){0, 0, 0xffff, 0, 0, 0, 0, 0}; /* * A. load first Qword (8bytes) in one loop. @@ -861,30 +861,30 @@ rxq_cq_process_v(struct mlx5_rxq_data *rxq, volatile struct mlx5_cqe *cq, for (pos = 0; pos < pkts_n; pos += MLX5_VPMD_DESCS_PER_LOOP) { - vector unsigned char cqes[MLX5_VPMD_DESCS_PER_LOOP]; - vector unsigned char cqe_tmp1, cqe_tmp2; - vector unsigned char pkt_mb0, pkt_mb1, pkt_mb2, pkt_mb3; - vector unsigned char op_own, op_own_tmp1, op_own_tmp2; - vector unsigned char opcode, owner_mask, invalid_mask; - vector unsigned char comp_mask; - vector unsigned char mask; + __vector unsigned char cqes[MLX5_VPMD_DESCS_PER_LOOP]; + __vector unsigned char cqe_tmp1, cqe_tmp2; + __vector unsigned char pkt_mb0, pkt_mb1, pkt_mb2, pkt_mb3; + __vector unsigned char op_own, op_own_tmp1, op_own_tmp2; + __vector unsigned char opcode, owner_mask, invalid_mask; + __vector unsigned char comp_mask; + __vector unsigned char mask; #ifdef MLX5_PMD_SOFT_COUNTERS - const vector unsigned char lower_half = { + const __vector unsigned char lower_half = { 0, 1, 4, 5, 8, 9, 12, 13, 16, 17, 20, 21, 24, 25, 28, 29}; - const vector unsigned char upper_half = { + const __vector unsigned char upper_half = { 2, 3, 6, 7, 10, 11, 14, 15, 18, 19, 22, 23, 26, 27, 30, 31}; - const vector unsigned long shmax = {64, 64}; - vector unsigned char byte_cnt; - vector unsigned short left, right; - vector unsigned long lshift; - vector __attribute__((altivec(bool__))) + const __vector unsigned long shmax = {64, 64}; + __vector unsigned char byte_cnt; + __vector unsigned short left, right; + __vector unsigned long lshift; + __vector __attribute__((altivec(bool__))) unsigned long shmask; #endif - vector unsigned char mbp1, mbp2; - vector unsigned char p = - (vector unsigned char)(vector unsigned short){ + __vector unsigned char mbp1, mbp2; + __vector unsigned char p = + (__vector unsigned char)(__vector unsigned short){ 0, 1, 2, 3, 0, 0, 0, 0}; unsigned int p1, p2, p3; @@ -897,295 +897,295 @@ rxq_cq_process_v(struct mlx5_rxq_data *rxq, volatile struct mlx5_cqe *cq, } /* A.0 do not cross the end of CQ. */ - mask = (vector unsigned char)(vector unsigned long){ + mask = (__vector unsigned char)(__vector unsigned long){ (pkts_n - pos) * sizeof(uint16_t) * 8, 0}; { - vector unsigned long lshift; - vector __attribute__((altivec(bool__))) + __vector unsigned long lshift; + __vector __attribute__((altivec(bool__))) unsigned long shmask; - const vector unsigned long shmax = {64, 64}; + const __vector unsigned long shmax = {64, 64}; - lshift = vec_splat((vector unsigned long)mask, 0); + lshift = vec_splat((__vector unsigned long)mask, 0); shmask = vec_cmpgt(shmax, lshift); - mask = (vector unsigned char) - vec_sl((vector unsigned long)ones, lshift); - mask = (vector unsigned char) - vec_sel((vector unsigned long)shmask, - (vector unsigned long)mask, shmask); + mask = (__vector unsigned char) + vec_sl((__vector unsigned long)ones, lshift); + mask = (__vector unsigned char) + vec_sel((__vector unsigned long)shmask, + (__vector unsigned long)mask, shmask); } - p = (vector unsigned char) - vec_andc((vector unsigned long)p, - (vector unsigned long)mask); + p = (__vector unsigned char) + vec_andc((__vector unsigned long)p, + (__vector unsigned long)mask); /* A.1 load cqes. */ - p3 = (unsigned int)((vector unsigned short)p)[3]; - cqes[3] = (vector unsigned char)(vector unsigned long){ + p3 = (unsigned int)((__vector unsigned short)p)[3]; + cqes[3] = (__vector unsigned char)(__vector unsigned long){ *(__rte_aligned(8) unsigned long *) &cq[pos + p3].sop_drop_qpn, 0LL}; rte_compiler_barrier(); - p2 = (unsigned int)((vector unsigned short)p)[2]; - cqes[2] = (vector unsigned char)(vector unsigned long){ + p2 = (unsigned int)((__vector unsigned short)p)[2]; + cqes[2] = (__vector unsigned char)(__vector unsigned long){ *(__rte_aligned(8) unsigned long *) &cq[pos + p2].sop_drop_qpn, 0LL}; rte_compiler_barrier(); /* B.1 load mbuf pointers. */ - mbp1 = (vector unsigned char)vec_vsx_ld(0, + mbp1 = (__vector unsigned char)vec_vsx_ld(0, (signed int const *)&elts[pos]); - mbp2 = (vector unsigned char)vec_vsx_ld(0, + mbp2 = (__vector unsigned char)vec_vsx_ld(0, (signed int const *)&elts[pos + 2]); /* A.1 load a block having op_own. */ - p1 = (unsigned int)((vector unsigned short)p)[1]; - cqes[1] = (vector unsigned char)(vector unsigned long){ + p1 = (unsigned int)((__vector unsigned short)p)[1]; + cqes[1] = (__vector unsigned char)(__vector unsigned long){ *(__rte_aligned(8) unsigned long *) &cq[pos + p1].sop_drop_qpn, 0LL}; rte_compiler_barrier(); - cqes[0] = (vector unsigned char)(vector unsigned long){ + cqes[0] = (__vector unsigned char)(__vector unsigned long){ *(__rte_aligned(8) unsigned long *) &cq[pos].sop_drop_qpn, 0LL}; rte_compiler_barrier(); /* B.2 copy mbuf pointers. */ - *(vector unsigned char *)&pkts[pos] = mbp1; - *(vector unsigned char *)&pkts[pos + 2] = mbp2; + *(__vector unsigned char *)&pkts[pos] = mbp1; + *(__vector unsigned char *)&pkts[pos + 2] = mbp2; rte_io_rmb(); /* C.1 load remaining CQE data and extract necessary fields. */ - cqe_tmp2 = *(vector unsigned char *) + cqe_tmp2 = *(__vector unsigned char *) &cq[pos + p3].pkt_info; - cqe_tmp1 = *(vector unsigned char *) + cqe_tmp1 = *(__vector unsigned char *) &cq[pos + p2].pkt_info; cqes[3] = vec_sel(cqes[3], cqe_tmp2, blend_mask); cqes[2] = vec_sel(cqes[2], cqe_tmp1, blend_mask); - cqe_tmp2 = (vector unsigned char)vec_vsx_ld(0, + cqe_tmp2 = (__vector unsigned char)vec_vsx_ld(0, (signed int const *)&cq[pos + p3].csum); - cqe_tmp1 = (vector unsigned char)vec_vsx_ld(0, + cqe_tmp1 = (__vector unsigned char)vec_vsx_ld(0, (signed int const *)&cq[pos + p2].csum); - cqes[3] = (vector unsigned char) - vec_sel((vector unsigned short)cqes[3], - (vector unsigned short)cqe_tmp2, cqe_sel_mask1); - cqes[2] = (vector unsigned char) - vec_sel((vector unsigned short)cqes[2], - (vector unsigned short)cqe_tmp1, cqe_sel_mask1); - cqe_tmp2 = (vector unsigned char)(vector unsigned long){ + cqes[3] = (__vector unsigned char) + vec_sel((__vector unsigned short)cqes[3], + (__vector unsigned short)cqe_tmp2, cqe_sel_mask1); + cqes[2] = (__vector unsigned char) + vec_sel((__vector unsigned short)cqes[2], + (__vector unsigned short)cqe_tmp1, cqe_sel_mask1); + cqe_tmp2 = (__vector unsigned char)(__vector unsigned long){ *(__rte_aligned(8) unsigned long *) &cq[pos + p3].rsvd4[2], 0LL}; - cqe_tmp1 = (vector unsigned char)(vector unsigned long){ + cqe_tmp1 = (__vector unsigned char)(__vector unsigned long){ *(__rte_aligned(8) unsigned long *) &cq[pos + p2].rsvd4[2], 0LL}; - cqes[3] = (vector unsigned char) - vec_sel((vector unsigned short)cqes[3], - (vector unsigned short)cqe_tmp2, - (vector unsigned short)cqe_sel_mask2); - cqes[2] = (vector unsigned char) - vec_sel((vector unsigned short)cqes[2], - (vector unsigned short)cqe_tmp1, - (vector unsigned short)cqe_sel_mask2); + cqes[3] = (__vector unsigned char) + vec_sel((__vector unsigned short)cqes[3], + (__vector unsigned short)cqe_tmp2, + (__vector unsigned short)cqe_sel_mask2); + cqes[2] = (__vector unsigned char) + vec_sel((__vector unsigned short)cqes[2], + (__vector unsigned short)cqe_tmp1, + (__vector unsigned short)cqe_sel_mask2); /* C.2 generate final structure for mbuf with swapping bytes. */ pkt_mb3 = vec_perm(cqes[3], zero, shuf_mask); pkt_mb2 = vec_perm(cqes[2], zero, shuf_mask); /* C.3 adjust CRC length. */ - pkt_mb3 = (vector unsigned char) - ((vector unsigned short)pkt_mb3 - - (vector unsigned short)crc_adj); - pkt_mb2 = (vector unsigned char) - ((vector unsigned short)pkt_mb2 - - (vector unsigned short)crc_adj); + pkt_mb3 = (__vector unsigned char) + ((__vector unsigned short)pkt_mb3 - + (__vector unsigned short)crc_adj); + pkt_mb2 = (__vector unsigned char) + ((__vector unsigned short)pkt_mb2 - + (__vector unsigned short)crc_adj); /* C.4 adjust flow mark. */ - pkt_mb3 = (vector unsigned char) - ((vector unsigned int)pkt_mb3 + - (vector unsigned int)flow_mark_adj); - pkt_mb2 = (vector unsigned char) - ((vector unsigned int)pkt_mb2 + - (vector unsigned int)flow_mark_adj); + pkt_mb3 = (__vector unsigned char) + ((__vector unsigned int)pkt_mb3 + + (__vector unsigned int)flow_mark_adj); + pkt_mb2 = (__vector unsigned char) + ((__vector unsigned int)pkt_mb2 + + (__vector unsigned int)flow_mark_adj); /* D.1 fill in mbuf - rx_descriptor_fields1. */ - *(vector unsigned char *) + *(__vector unsigned char *) &pkts[pos + 3]->pkt_len = pkt_mb3; - *(vector unsigned char *) + *(__vector unsigned char *) &pkts[pos + 2]->pkt_len = pkt_mb2; /* E.1 extract op_own field. */ - op_own_tmp2 = (vector unsigned char) - vec_mergeh((vector unsigned int)cqes[2], - (vector unsigned int)cqes[3]); + op_own_tmp2 = (__vector unsigned char) + vec_mergeh((__vector unsigned int)cqes[2], + (__vector unsigned int)cqes[3]); /* C.1 load remaining CQE data and extract necessary fields. */ - cqe_tmp2 = *(vector unsigned char *) + cqe_tmp2 = *(__vector unsigned char *) &cq[pos + p1].pkt_info; - cqe_tmp1 = *(vector unsigned char *) + cqe_tmp1 = *(__vector unsigned char *) &cq[pos].pkt_info; cqes[1] = vec_sel(cqes[1], cqe_tmp2, blend_mask); cqes[0] = vec_sel(cqes[0], cqe_tmp2, blend_mask); - cqe_tmp2 = (vector unsigned char)vec_vsx_ld(0, + cqe_tmp2 = (__vector unsigned char)vec_vsx_ld(0, (signed int const *)&cq[pos + p1].csum); - cqe_tmp1 = (vector unsigned char)vec_vsx_ld(0, + cqe_tmp1 = (__vector unsigned char)vec_vsx_ld(0, (signed int const *)&cq[pos].csum); - cqes[1] = (vector unsigned char) - vec_sel((vector unsigned short)cqes[1], - (vector unsigned short)cqe_tmp2, cqe_sel_mask1); - cqes[0] = (vector unsigned char) - vec_sel((vector unsigned short)cqes[0], - (vector unsigned short)cqe_tmp1, cqe_sel_mask1); - cqe_tmp2 = (vector unsigned char)(vector unsigned long){ + cqes[1] = (__vector unsigned char) + vec_sel((__vector unsigned short)cqes[1], + (__vector unsigned short)cqe_tmp2, cqe_sel_mask1); + cqes[0] = (__vector unsigned char) + vec_sel((__vector unsigned short)cqes[0], + (__vector unsigned short)cqe_tmp1, cqe_sel_mask1); + cqe_tmp2 = (__vector unsigned char)(__vector unsigned long){ *(__rte_aligned(8) unsigned long *) &cq[pos + p1].rsvd4[2], 0LL}; - cqe_tmp1 = (vector unsigned char)(vector unsigned long){ + cqe_tmp1 = (__vector unsigned char)(__vector unsigned long){ *(__rte_aligned(8) unsigned long *) &cq[pos].rsvd4[2], 0LL}; - cqes[1] = (vector unsigned char) - vec_sel((vector unsigned short)cqes[1], - (vector unsigned short)cqe_tmp2, cqe_sel_mask2); - cqes[0] = (vector unsigned char) - vec_sel((vector unsigned short)cqes[0], - (vector unsigned short)cqe_tmp1, cqe_sel_mask2); + cqes[1] = (__vector unsigned char) + vec_sel((__vector unsigned short)cqes[1], + (__vector unsigned short)cqe_tmp2, cqe_sel_mask2); + cqes[0] = (__vector unsigned char) + vec_sel((__vector unsigned short)cqes[0], + (__vector unsigned short)cqe_tmp1, cqe_sel_mask2); /* C.2 generate final structure for mbuf with swapping bytes. */ pkt_mb1 = vec_perm(cqes[1], zero, shuf_mask); pkt_mb0 = vec_perm(cqes[0], zero, shuf_mask); /* C.3 adjust CRC length. */ - pkt_mb1 = (vector unsigned char) - ((vector unsigned short)pkt_mb1 - - (vector unsigned short)crc_adj); - pkt_mb0 = (vector unsigned char) - ((vector unsigned short)pkt_mb0 - - (vector unsigned short)crc_adj); + pkt_mb1 = (__vector unsigned char) + ((__vector unsigned short)pkt_mb1 - + (__vector unsigned short)crc_adj); + pkt_mb0 = (__vector unsigned char) + ((__vector unsigned short)pkt_mb0 - + (__vector unsigned short)crc_adj); /* C.4 adjust flow mark. */ - pkt_mb1 = (vector unsigned char) - ((vector unsigned int)pkt_mb1 + - (vector unsigned int)flow_mark_adj); - pkt_mb0 = (vector unsigned char) - ((vector unsigned int)pkt_mb0 + - (vector unsigned int)flow_mark_adj); + pkt_mb1 = (__vector unsigned char) + ((__vector unsigned int)pkt_mb1 + + (__vector unsigned int)flow_mark_adj); + pkt_mb0 = (__vector unsigned char) + ((__vector unsigned int)pkt_mb0 + + (__vector unsigned int)flow_mark_adj); /* E.1 extract op_own byte. */ - op_own_tmp1 = (vector unsigned char) - vec_mergeh((vector unsigned int)cqes[0], - (vector unsigned int)cqes[1]); - op_own = (vector unsigned char) - vec_mergel((vector unsigned long)op_own_tmp1, - (vector unsigned long)op_own_tmp2); + op_own_tmp1 = (__vector unsigned char) + vec_mergeh((__vector unsigned int)cqes[0], + (__vector unsigned int)cqes[1]); + op_own = (__vector unsigned char) + vec_mergel((__vector unsigned long)op_own_tmp1, + (__vector unsigned long)op_own_tmp2); /* D.1 fill in mbuf - rx_descriptor_fields1. */ - *(vector unsigned char *) + *(__vector unsigned char *) &pkts[pos + 1]->pkt_len = pkt_mb1; - *(vector unsigned char *) + *(__vector unsigned char *) &pkts[pos]->pkt_len = pkt_mb0; /* E.2 flip owner bit to mark CQEs from last round. */ - owner_mask = (vector unsigned char) - vec_and((vector unsigned long)op_own, - (vector unsigned long)owner_check); + owner_mask = (__vector unsigned char) + vec_and((__vector unsigned long)op_own, + (__vector unsigned long)owner_check); if (ownership) - owner_mask = (vector unsigned char) - vec_xor((vector unsigned long)owner_mask, - (vector unsigned long)owner_check); - owner_mask = (vector unsigned char) - vec_cmpeq((vector unsigned int)owner_mask, - (vector unsigned int)owner_check); - owner_mask = (vector unsigned char) - vec_packs((vector unsigned int)owner_mask, - (vector unsigned int)zero); + owner_mask = (__vector unsigned char) + vec_xor((__vector unsigned long)owner_mask, + (__vector unsigned long)owner_check); + owner_mask = (__vector unsigned char) + vec_cmpeq((__vector unsigned int)owner_mask, + (__vector unsigned int)owner_check); + owner_mask = (__vector unsigned char) + vec_packs((__vector unsigned int)owner_mask, + (__vector unsigned int)zero); /* E.3 get mask for invalidated CQEs. */ - opcode = (vector unsigned char) - vec_and((vector unsigned long)op_own, - (vector unsigned long)opcode_check); - invalid_mask = (vector unsigned char) - vec_cmpeq((vector unsigned int)opcode_check, - (vector unsigned int)opcode); - invalid_mask = (vector unsigned char) - vec_packs((vector unsigned int)invalid_mask, - (vector unsigned int)zero); + opcode = (__vector unsigned char) + vec_and((__vector unsigned long)op_own, + (__vector unsigned long)opcode_check); + invalid_mask = (__vector unsigned char) + vec_cmpeq((__vector unsigned int)opcode_check, + (__vector unsigned int)opcode); + invalid_mask = (__vector unsigned char) + vec_packs((__vector unsigned int)invalid_mask, + (__vector unsigned int)zero); /* E.4 mask out beyond boundary. */ - invalid_mask = (vector unsigned char) - vec_or((vector unsigned long)invalid_mask, - (vector unsigned long)mask); + invalid_mask = (__vector unsigned char) + vec_or((__vector unsigned long)invalid_mask, + (__vector unsigned long)mask); /* E.5 merge invalid_mask with invalid owner. */ - invalid_mask = (vector unsigned char) - vec_or((vector unsigned long)invalid_mask, - (vector unsigned long)owner_mask); + invalid_mask = (__vector unsigned char) + vec_or((__vector unsigned long)invalid_mask, + (__vector unsigned long)owner_mask); /* F.1 find compressed CQE format. */ - comp_mask = (vector unsigned char) - vec_and((vector unsigned long)op_own, - (vector unsigned long)format_check); - comp_mask = (vector unsigned char) - vec_cmpeq((vector unsigned int)comp_mask, - (vector unsigned int)format_check); - comp_mask = (vector unsigned char) - vec_packs((vector unsigned int)comp_mask, - (vector unsigned int)zero); + comp_mask = (__vector unsigned char) + vec_and((__vector unsigned long)op_own, + (__vector unsigned long)format_check); + comp_mask = (__vector unsigned char) + vec_cmpeq((__vector unsigned int)comp_mask, + (__vector unsigned int)format_check); + comp_mask = (__vector unsigned char) + vec_packs((__vector unsigned int)comp_mask, + (__vector unsigned int)zero); /* F.2 mask out invalid entries. */ - comp_mask = (vector unsigned char) - vec_andc((vector unsigned long)comp_mask, - (vector unsigned long)invalid_mask); - comp_idx = ((vector unsigned long)comp_mask)[0]; + comp_mask = (__vector unsigned char) + vec_andc((__vector unsigned long)comp_mask, + (__vector unsigned long)invalid_mask); + comp_idx = ((__vector unsigned long)comp_mask)[0]; /* F.3 get the first compressed CQE. */ comp_idx = comp_idx ? __builtin_ctzll(comp_idx) / (sizeof(uint16_t) * 8) : MLX5_VPMD_DESCS_PER_LOOP; /* E.6 mask out entries after the compressed CQE. */ - mask = (vector unsigned char)(vector unsigned long){ + mask = (__vector unsigned char)(__vector unsigned long){ (comp_idx * sizeof(uint16_t) * 8), 0}; - lshift = vec_splat((vector unsigned long)mask, 0); + lshift = vec_splat((__vector unsigned long)mask, 0); shmask = vec_cmpgt(shmax, lshift); - mask = (vector unsigned char) - vec_sl((vector unsigned long)ones, lshift); - mask = (vector unsigned char) - vec_sel((vector unsigned long)shmask, - (vector unsigned long)mask, shmask); - invalid_mask = (vector unsigned char) - vec_or((vector unsigned long)invalid_mask, - (vector unsigned long)mask); + mask = (__vector unsigned char) + vec_sl((__vector unsigned long)ones, lshift); + mask = (__vector unsigned char) + vec_sel((__vector unsigned long)shmask, + (__vector unsigned long)mask, shmask); + invalid_mask = (__vector unsigned char) + vec_or((__vector unsigned long)invalid_mask, + (__vector unsigned long)mask); /* E.7 count non-compressed valid CQEs. */ - n = ((vector unsigned long)invalid_mask)[0]; + n = ((__vector unsigned long)invalid_mask)[0]; n = n ? __builtin_ctzll(n) / (sizeof(uint16_t) * 8) : MLX5_VPMD_DESCS_PER_LOOP; nocmp_n += n; /* D.2 get the final invalid mask. */ - mask = (vector unsigned char)(vector unsigned long){ + mask = (__vector unsigned char)(__vector unsigned long){ (n * sizeof(uint16_t) * 8), 0}; - lshift = vec_splat((vector unsigned long)mask, 0); + lshift = vec_splat((__vector unsigned long)mask, 0); shmask = vec_cmpgt(shmax, lshift); - mask = (vector unsigned char) - vec_sl((vector unsigned long)ones, lshift); - mask = (vector unsigned char) - vec_sel((vector unsigned long)shmask, - (vector unsigned long)mask, shmask); - invalid_mask = (vector unsigned char) - vec_or((vector unsigned long)invalid_mask, - (vector unsigned long)mask); + mask = (__vector unsigned char) + vec_sl((__vector unsigned long)ones, lshift); + mask = (__vector unsigned char) + vec_sel((__vector unsigned long)shmask, + (__vector unsigned long)mask, shmask); + invalid_mask = (__vector unsigned char) + vec_or((__vector unsigned long)invalid_mask, + (__vector unsigned long)mask); /* D.3 check error in opcode. */ - opcode = (vector unsigned char) - vec_cmpeq((vector unsigned int)resp_err_check, - (vector unsigned int)opcode); - opcode = (vector unsigned char) - vec_packs((vector unsigned int)opcode, - (vector unsigned int)zero); - opcode = (vector unsigned char) - vec_andc((vector unsigned long)opcode, - (vector unsigned long)invalid_mask); + opcode = (__vector unsigned char) + vec_cmpeq((__vector unsigned int)resp_err_check, + (__vector unsigned int)opcode); + opcode = (__vector unsigned char) + vec_packs((__vector unsigned int)opcode, + (__vector unsigned int)zero); + opcode = (__vector unsigned char) + vec_andc((__vector unsigned long)opcode, + (__vector unsigned long)invalid_mask); /* D.4 mark if any error is set */ - *err |= ((vector unsigned long)opcode)[0]; + *err |= ((__vector unsigned long)opcode)[0]; /* D.5 fill in mbuf - rearm_data and packet_type. */ rxq_cq_to_ptype_oflags_v(rxq, cqes, opcode, &pkts[pos]); @@ -1255,20 +1255,20 @@ rxq_cq_process_v(struct mlx5_rxq_data *rxq, volatile struct mlx5_cqe *cq, #ifdef MLX5_PMD_SOFT_COUNTERS /* Add up received bytes count. */ byte_cnt = vec_perm(op_own, zero, len_shuf_mask); - byte_cnt = (vector unsigned char) - vec_andc((vector unsigned long)byte_cnt, - (vector unsigned long)invalid_mask); - left = vec_perm((vector unsigned short)byte_cnt, - (vector unsigned short)zero, lower_half); - right = vec_perm((vector unsigned short)byte_cnt, - (vector unsigned short)zero, upper_half); - byte_cnt = (vector unsigned char)vec_add(left, right); - left = vec_perm((vector unsigned short)byte_cnt, - (vector unsigned short)zero, lower_half); - right = vec_perm((vector unsigned short)byte_cnt, - (vector unsigned short)zero, upper_half); - byte_cnt = (vector unsigned char)vec_add(left, right); - rcvd_byte += ((vector unsigned long)byte_cnt)[0]; + byte_cnt = (__vector unsigned char) + vec_andc((__vector unsigned long)byte_cnt, + (__vector unsigned long)invalid_mask); + left = vec_perm((__vector unsigned short)byte_cnt, + (__vector unsigned short)zero, lower_half); + right = vec_perm((__vector unsigned short)byte_cnt, + (__vector unsigned short)zero, upper_half); + byte_cnt = (__vector unsigned char)vec_add(left, right); + left = vec_perm((__vector unsigned short)byte_cnt, + (__vector unsigned short)zero, lower_half); + right = vec_perm((__vector unsigned short)byte_cnt, + (__vector unsigned short)zero, upper_half); + byte_cnt = (__vector unsigned char)vec_add(left, right); + rcvd_byte += ((__vector unsigned long)byte_cnt)[0]; #endif /* diff --git a/drivers/net/virtio/virtio_rxtx_simple_altivec.c b/drivers/net/virtio/virtio_rxtx_simple_altivec.c index 7534974ef4..e7f0ed6068 100644 --- a/drivers/net/virtio/virtio_rxtx_simple_altivec.c +++ b/drivers/net/virtio/virtio_rxtx_simple_altivec.c @@ -50,9 +50,9 @@ virtio_recv_pkts_vec(void *rx_queue, struct rte_mbuf **rx_pkts, struct rte_mbuf **sw_ring_end; struct rte_mbuf **ref_rx_pkts; uint16_t nb_pkts_received = 0; - const vector unsigned char zero = {0}; + const __vector unsigned char zero = {0}; - const vector unsigned char shuf_msk1 = { + const __vector unsigned char shuf_msk1 = { 0xFF, 0xFF, 0xFF, 0xFF, /* packet type */ 4, 5, 0xFF, 0xFF, /* vlan tci */ 4, 5, /* dat len */ @@ -60,7 +60,7 @@ virtio_recv_pkts_vec(void *rx_queue, struct rte_mbuf **rx_pkts, 0xFF, 0xFF, 0xFF, 0xFF }; - const vector unsigned char shuf_msk2 = { + const __vector unsigned char shuf_msk2 = { 0xFF, 0xFF, 0xFF, 0xFF, /* packet type */ 12, 13, 0xFF, 0xFF, /* pkt len */ 12, 13, /* dat len */ @@ -72,7 +72,7 @@ virtio_recv_pkts_vec(void *rx_queue, struct rte_mbuf **rx_pkts, * Subtract the header length. * In which case do we need the header length in used->len ? */ - const vector unsigned short len_adjust = { + const __vector unsigned short len_adjust = { 0, 0, (uint16_t)-vq->hw->vtnet_hdr_size, 0, (uint16_t)-vq->hw->vtnet_hdr_size, 0, @@ -112,68 +112,68 @@ virtio_recv_pkts_vec(void *rx_queue, struct rte_mbuf **rx_pkts, ref_rx_pkts = rx_pkts; for (nb_pkts_received = 0; nb_pkts_received < nb_total;) { - vector unsigned char desc[RTE_VIRTIO_DESC_PER_LOOP / 2]; - vector unsigned char mbp[RTE_VIRTIO_DESC_PER_LOOP / 2]; - vector unsigned char pkt_mb[RTE_VIRTIO_DESC_PER_LOOP]; + __vector unsigned char desc[RTE_VIRTIO_DESC_PER_LOOP / 2]; + __vector unsigned char mbp[RTE_VIRTIO_DESC_PER_LOOP / 2]; + __vector unsigned char pkt_mb[RTE_VIRTIO_DESC_PER_LOOP]; mbp[0] = vec_vsx_ld(0, (unsigned char const *)(sw_ring + 0)); desc[0] = vec_vsx_ld(0, (unsigned char const *)(rused + 0)); - *(vector unsigned char *)&rx_pkts[0] = mbp[0]; + *(__vector unsigned char *)&rx_pkts[0] = mbp[0]; mbp[1] = vec_vsx_ld(0, (unsigned char const *)(sw_ring + 2)); desc[1] = vec_vsx_ld(0, (unsigned char const *)(rused + 2)); - *(vector unsigned char *)&rx_pkts[2] = mbp[1]; + *(__vector unsigned char *)&rx_pkts[2] = mbp[1]; mbp[2] = vec_vsx_ld(0, (unsigned char const *)(sw_ring + 4)); desc[2] = vec_vsx_ld(0, (unsigned char const *)(rused + 4)); - *(vector unsigned char *)&rx_pkts[4] = mbp[2]; + *(__vector unsigned char *)&rx_pkts[4] = mbp[2]; mbp[3] = vec_vsx_ld(0, (unsigned char const *)(sw_ring + 6)); desc[3] = vec_vsx_ld(0, (unsigned char const *)(rused + 6)); - *(vector unsigned char *)&rx_pkts[6] = mbp[3]; + *(__vector unsigned char *)&rx_pkts[6] = mbp[3]; pkt_mb[0] = vec_perm(desc[0], zero, shuf_msk1); pkt_mb[1] = vec_perm(desc[0], zero, shuf_msk2); - pkt_mb[0] = (vector unsigned char) - ((vector unsigned short)pkt_mb[0] + len_adjust); - pkt_mb[1] = (vector unsigned char) - ((vector unsigned short)pkt_mb[1] + len_adjust); - *(vector unsigned char *)&rx_pkts[0]->rx_descriptor_fields1 = + pkt_mb[0] = (__vector unsigned char) + ((__vector unsigned short)pkt_mb[0] + len_adjust); + pkt_mb[1] = (__vector unsigned char) + ((__vector unsigned short)pkt_mb[1] + len_adjust); + *(__vector unsigned char *)&rx_pkts[0]->rx_descriptor_fields1 = pkt_mb[0]; - *(vector unsigned char *)&rx_pkts[1]->rx_descriptor_fields1 = + *(__vector unsigned char *)&rx_pkts[1]->rx_descriptor_fields1 = pkt_mb[1]; pkt_mb[2] = vec_perm(desc[1], zero, shuf_msk1); pkt_mb[3] = vec_perm(desc[1], zero, shuf_msk2); - pkt_mb[2] = (vector unsigned char) - ((vector unsigned short)pkt_mb[2] + len_adjust); - pkt_mb[3] = (vector unsigned char) - ((vector unsigned short)pkt_mb[3] + len_adjust); - *(vector unsigned char *)&rx_pkts[2]->rx_descriptor_fields1 = + pkt_mb[2] = (__vector unsigned char) + ((__vector unsigned short)pkt_mb[2] + len_adjust); + pkt_mb[3] = (__vector unsigned char) + ((__vector unsigned short)pkt_mb[3] + len_adjust); + *(__vector unsigned char *)&rx_pkts[2]->rx_descriptor_fields1 = pkt_mb[2]; - *(vector unsigned char *)&rx_pkts[3]->rx_descriptor_fields1 = + *(__vector unsigned char *)&rx_pkts[3]->rx_descriptor_fields1 = pkt_mb[3]; pkt_mb[4] = vec_perm(desc[2], zero, shuf_msk1); pkt_mb[5] = vec_perm(desc[2], zero, shuf_msk2); - pkt_mb[4] = (vector unsigned char) - ((vector unsigned short)pkt_mb[4] + len_adjust); - pkt_mb[5] = (vector unsigned char) - ((vector unsigned short)pkt_mb[5] + len_adjust); - *(vector unsigned char *)&rx_pkts[4]->rx_descriptor_fields1 = + pkt_mb[4] = (__vector unsigned char) + ((__vector unsigned short)pkt_mb[4] + len_adjust); + pkt_mb[5] = (__vector unsigned char) + ((__vector unsigned short)pkt_mb[5] + len_adjust); + *(__vector unsigned char *)&rx_pkts[4]->rx_descriptor_fields1 = pkt_mb[4]; - *(vector unsigned char *)&rx_pkts[5]->rx_descriptor_fields1 = + *(__vector unsigned char *)&rx_pkts[5]->rx_descriptor_fields1 = pkt_mb[5]; pkt_mb[6] = vec_perm(desc[3], zero, shuf_msk1); pkt_mb[7] = vec_perm(desc[3], zero, shuf_msk2); - pkt_mb[6] = (vector unsigned char) - ((vector unsigned short)pkt_mb[6] + len_adjust); - pkt_mb[7] = (vector unsigned char) - ((vector unsigned short)pkt_mb[7] + len_adjust); - *(vector unsigned char *)&rx_pkts[6]->rx_descriptor_fields1 = + pkt_mb[6] = (__vector unsigned char) + ((__vector unsigned short)pkt_mb[6] + len_adjust); + pkt_mb[7] = (__vector unsigned char) + ((__vector unsigned short)pkt_mb[7] + len_adjust); + *(__vector unsigned char *)&rx_pkts[6]->rx_descriptor_fields1 = pkt_mb[6]; - *(vector unsigned char *)&rx_pkts[7]->rx_descriptor_fields1 = + *(__vector unsigned char *)&rx_pkts[7]->rx_descriptor_fields1 = pkt_mb[7]; if (unlikely(nb_used <= RTE_VIRTIO_DESC_PER_LOOP)) { diff --git a/examples/l3fwd/l3fwd_lpm_altivec.h b/examples/l3fwd/l3fwd_lpm_altivec.h index 7c6814252a..0c6852a7bb 100644 --- a/examples/l3fwd/l3fwd_lpm_altivec.h +++ b/examples/l3fwd/l3fwd_lpm_altivec.h @@ -14,7 +14,7 @@ */ static inline void processx4_step1(struct rte_mbuf *pkt[FWDSTEP], - vector unsigned int *dip, + __vector unsigned int *dip, uint32_t *ipv4_flag) { struct rte_ipv4_hdr *ipv4_hdr; @@ -45,7 +45,7 @@ processx4_step1(struct rte_mbuf *pkt[FWDSTEP], ipv4_flag[0] &= pkt[3]->packet_type; rte_compiler_barrier(); - dip[0] = (vector unsigned int){x0, x1, x2, x3}; + dip[0] = (__vector unsigned int){x0, x1, x2, x3}; } /* @@ -54,22 +54,22 @@ processx4_step1(struct rte_mbuf *pkt[FWDSTEP], */ static inline void processx4_step2(const struct lcore_conf *qconf, - vector unsigned int dip, + __vector unsigned int dip, uint32_t ipv4_flag, uint8_t portid, struct rte_mbuf *pkt[FWDSTEP], uint16_t dprt[FWDSTEP]) { rte_xmm_t dst; - const vector unsigned char bswap_mask = (vector unsigned char){ + const __vector unsigned char bswap_mask = (__vector unsigned char){ 3, 2, 1, 0, 7, 6, 5, 4, 11, 10, 9, 8, 15, 14, 13, 12}; /* Byte swap 4 IPV4 addresses. */ - dip = (vector unsigned int)vec_perm(*(vector unsigned char *)&dip, - (vector unsigned char){}, bswap_mask); + dip = (__vector unsigned int)vec_perm(*(__vector unsigned char *)&dip, + (__vector unsigned char){}, bswap_mask); /* if all 4 packets are IPV4. */ if (likely(ipv4_flag)) { @@ -101,7 +101,7 @@ l3fwd_lpm_send_packets(int nb_rx, struct rte_mbuf **pkts_burst, { int32_t j; uint16_t dst_port[MAX_PKT_BURST]; - vector unsigned int dip[MAX_PKT_BURST / FWDSTEP]; + __vector unsigned int dip[MAX_PKT_BURST / FWDSTEP]; uint32_t ipv4_flag[MAX_PKT_BURST / FWDSTEP]; const int32_t k = RTE_ALIGN_FLOOR(nb_rx, FWDSTEP); diff --git a/lib/eal/ppc/include/rte_vect.h b/lib/eal/ppc/include/rte_vect.h index c1f0b0672c..a5f009b7df 100644 --- a/lib/eal/ppc/include/rte_vect.h +++ b/lib/eal/ppc/include/rte_vect.h @@ -17,7 +17,7 @@ extern "C" { #define RTE_VECT_DEFAULT_SIMD_BITWIDTH RTE_VECT_SIMD_256 -typedef vector signed int xmm_t; +typedef __vector signed int xmm_t; #define XMM_SIZE (sizeof(xmm_t)) #define XMM_MASK (XMM_SIZE - 1) diff --git a/lib/lpm/rte_lpm_altivec.h b/lib/lpm/rte_lpm_altivec.h index 4fbc1b595d..bab8929495 100644 --- a/lib/lpm/rte_lpm_altivec.h +++ b/lib/lpm/rte_lpm_altivec.h @@ -19,14 +19,14 @@ static inline void rte_lpm_lookupx4(const struct rte_lpm *lpm, xmm_t ip, uint32_t hop[4], uint32_t defv) { - vector signed int i24; + xmm_t i24; rte_xmm_t i8; uint32_t tbl[4]; uint64_t idx, pt, pt2; const uint32_t *ptbl; const uint32_t mask = UINT8_MAX; - const vector signed int mask8 = (xmm_t){mask, mask, mask, mask}; + const xmm_t mask8 = (xmm_t){mask, mask, mask, mask}; /* * RTE_LPM_VALID_EXT_ENTRY_BITMASK for 2 LPM entries @@ -46,7 +46,7 @@ rte_lpm_lookupx4(const struct rte_lpm *lpm, xmm_t ip, uint32_t hop[4], /* get 4 indexes for tbl24[]. */ i24 = vec_sr((xmm_t) ip, - (vector unsigned int){CHAR_BIT, CHAR_BIT, CHAR_BIT, CHAR_BIT}); + (__vector unsigned int){CHAR_BIT, CHAR_BIT, CHAR_BIT, CHAR_BIT}); /* extract values from tbl24[] */ idx = (uint32_t)i24[0]; -- 2.35.3