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 9464DA0508; Mon, 4 Apr 2022 23:17:10 +0200 (CEST) Received: from [217.70.189.124] (localhost [127.0.0.1]) by mails.dpdk.org (Postfix) with ESMTP id 053D142869; Mon, 4 Apr 2022 23:16:32 +0200 (CEST) Received: from mga11.intel.com (mga11.intel.com [192.55.52.93]) by mails.dpdk.org (Postfix) with ESMTP id D351642830 for ; Mon, 4 Apr 2022 23:16:25 +0200 (CEST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1649106986; x=1680642986; h=from:to:cc:subject:date:message-id:in-reply-to: references; bh=Ae64RFyFoodMG2eZ4Ma0OEUZaQyS0J6YAhj0+Z+9/+E=; b=RlGJRUfO8IqdYd/MWBlNUapqExIbLd7jlv9p1W9S0bWu1EdVV5HxuvqQ oBUEzC1TFf65ie1Tia8hftQENJbtv9wy2OSaZ+QgWWXaUhnuRqhb2kp5d BdxVBj6E59lQRKykf0jGFUz2+3IUogq59K+nFlGGGggTyRnpZvQTpFuZF KHFTO4iX7eFArVH0d2er+lP4uetqShlDO2Agmyp9GM0IbVbfaxmglM69F OYw36dM51DrdeT7ilt9PSZ2bnt9tVzMpipOGZvt0BQ38tlImFdSdmd0Mo 8oRyBrUJnh79xES1Ytph0EnEExGtuf/ZiqX0H4pYNL+izOOI7zLLpWhcA A==; X-IronPort-AV: E=McAfee;i="6200,9189,10307"; a="258194700" X-IronPort-AV: E=Sophos;i="5.90,235,1643702400"; d="scan'208";a="258194700" Received: from orsmga008.jf.intel.com ([10.7.209.65]) by fmsmga102.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 04 Apr 2022 14:16:24 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.90,235,1643702400"; d="scan'208";a="569552055" Received: from skx-5gnr-sc12-4.sc.intel.com ([172.25.69.210]) by orsmga008.jf.intel.com with ESMTP; 04 Apr 2022 14:16:23 -0700 From: Nicolas Chautru To: dev@dpdk.org, gakhil@marvell.com Cc: trix@redhat.com, thomas@monjalon.net, ray.kinsella@intel.com, bruce.richardson@intel.com, hemant.agrawal@nxp.com, mingshan.zhang@intel.com, david.marchand@redhat.com, Nicolas Chautru Subject: [PATCH v1 5/9] baseband/acc101: add LDPC processing Date: Mon, 4 Apr 2022 14:13:44 -0700 Message-Id: <1649106828-116338-6-git-send-email-nicolas.chautru@intel.com> X-Mailer: git-send-email 1.8.3.1 In-Reply-To: <1649106828-116338-1-git-send-email-nicolas.chautru@intel.com> References: <1649106828-116338-1-git-send-email-nicolas.chautru@intel.com> 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 Add capabilities and function to do actual LDPC encode and decode processing operations. Signed-off-by: Nicolas Chautru --- doc/guides/bbdevs/acc101.rst | 22 + doc/guides/bbdevs/features/acc101.ini | 8 +- drivers/baseband/acc101/rte_acc101_pmd.c | 1737 ++++++++++++++++++++++++++++++ drivers/baseband/acc101/rte_acc101_pmd.h | 26 + 4 files changed, 1789 insertions(+), 4 deletions(-) diff --git a/doc/guides/bbdevs/acc101.rst b/doc/guides/bbdevs/acc101.rst index fb5f9f1..ae27bf3 100644 --- a/doc/guides/bbdevs/acc101.rst +++ b/doc/guides/bbdevs/acc101.rst @@ -15,12 +15,34 @@ Features ACC101 5G/4G FEC PMD supports the following features: +- LDPC Encode in the DL (5GNR) +- LDPC Decode in the UL (5GNR) - 16 VFs per PF (physical device) - Maximum of 128 queues per VF - PCIe Gen-3 x16 Interface - MSI - SR-IOV +ACC101 5G/4G FEC PMD supports the following BBDEV capabilities: + +* For the LDPC encode operation: + - ``RTE_BBDEV_LDPC_CRC_24B_ATTACH`` : set to attach CRC24B to CB(s) + - ``RTE_BBDEV_LDPC_RATE_MATCH`` : if set then do not do Rate Match bypass + - ``RTE_BBDEV_LDPC_INTERLEAVER_BYPASS`` : if set then bypass interleaver + +* For the LDPC decode operation: + - ``RTE_BBDEV_LDPC_CRC_TYPE_24B_CHECK`` : check CRC24B from CB(s) + - ``RTE_BBDEV_LDPC_ITERATION_STOP_ENABLE`` : disable early termination + - ``RTE_BBDEV_LDPC_CRC_TYPE_24B_DROP`` : drops CRC24B bits appended while decoding + - ``RTE_BBDEV_LDPC_HQ_COMBINE_IN_ENABLE`` : provides an input for HARQ combining + - ``RTE_BBDEV_LDPC_HQ_COMBINE_OUT_ENABLE`` : provides an input for HARQ combining + - ``RTE_BBDEV_LDPC_INTERNAL_HARQ_MEMORY_IN_ENABLE`` : HARQ memory input is internal + - ``RTE_BBDEV_LDPC_INTERNAL_HARQ_MEMORY_OUT_ENABLE`` : HARQ memory output is internal + - ``RTE_BBDEV_LDPC_INTERNAL_HARQ_MEMORY_FILLERS`` : HARQ memory includes the fillers bits + - ``RTE_BBDEV_LDPC_DEC_SCATTER_GATHER`` : supports scatter-gather for input/output data + - ``RTE_BBDEV_LDPC_HARQ_6BIT_COMPRESSION`` : supports compression of the HARQ input/output + - ``RTE_BBDEV_LDPC_LLR_COMPRESSION`` : supports LLR input compression + Installation ------------ diff --git a/doc/guides/bbdevs/features/acc101.ini b/doc/guides/bbdevs/features/acc101.ini index 1a62d13..4a88932 100644 --- a/doc/guides/bbdevs/features/acc101.ini +++ b/doc/guides/bbdevs/features/acc101.ini @@ -6,8 +6,8 @@ [Features] Turbo Decoder (4G) = N Turbo Encoder (4G) = N -LDPC Decoder (5G) = N -LDPC Encoder (5G) = N -LLR/HARQ Compression = N -External DDR Access = N +LDPC Decoder (5G) = Y +LDPC Encoder (5G) = Y +LLR/HARQ Compression = Y +External DDR Access = Y HW Accelerated = Y diff --git a/drivers/baseband/acc101/rte_acc101_pmd.c b/drivers/baseband/acc101/rte_acc101_pmd.c index e490e97..2861907 100644 --- a/drivers/baseband/acc101/rte_acc101_pmd.c +++ b/drivers/baseband/acc101/rte_acc101_pmd.c @@ -762,6 +762,46 @@ { struct acc101_device *d = dev->data->dev_private; static const struct rte_bbdev_op_cap bbdev_capabilities[] = { + { + .type = RTE_BBDEV_OP_LDPC_ENC, + .cap.ldpc_enc = { + .capability_flags = + RTE_BBDEV_LDPC_RATE_MATCH | + RTE_BBDEV_LDPC_CRC_24B_ATTACH | + RTE_BBDEV_LDPC_INTERLEAVER_BYPASS, + .num_buffers_src = + RTE_BBDEV_LDPC_MAX_CODE_BLOCKS, + .num_buffers_dst = + RTE_BBDEV_LDPC_MAX_CODE_BLOCKS, + } + }, + { + .type = RTE_BBDEV_OP_LDPC_DEC, + .cap.ldpc_dec = { + .capability_flags = + RTE_BBDEV_LDPC_CRC_TYPE_24B_CHECK | + RTE_BBDEV_LDPC_CRC_TYPE_24B_DROP | + RTE_BBDEV_LDPC_HQ_COMBINE_IN_ENABLE | + RTE_BBDEV_LDPC_HQ_COMBINE_OUT_ENABLE | +#ifdef ACC101_EXT_MEM + RTE_BBDEV_LDPC_INTERNAL_HARQ_MEMORY_IN_ENABLE | + RTE_BBDEV_LDPC_INTERNAL_HARQ_MEMORY_OUT_ENABLE | +#endif + RTE_BBDEV_LDPC_ITERATION_STOP_ENABLE | + RTE_BBDEV_LDPC_DEINTERLEAVER_BYPASS | + RTE_BBDEV_LDPC_DECODE_BYPASS | + RTE_BBDEV_LDPC_DEC_SCATTER_GATHER | + RTE_BBDEV_LDPC_HARQ_6BIT_COMPRESSION | + RTE_BBDEV_LDPC_LLR_COMPRESSION, + .llr_size = 8, + .llr_decimals = 1, + .num_buffers_src = + RTE_BBDEV_LDPC_MAX_CODE_BLOCKS, + .num_buffers_hard_out = + RTE_BBDEV_LDPC_MAX_CODE_BLOCKS, + .num_buffers_soft_out = 0, + } + }, RTE_BBDEV_END_OF_CAPABILITIES_LIST() }; @@ -796,7 +836,11 @@ dev_info->cpu_flag_reqs = NULL; dev_info->min_alignment = 64; dev_info->capabilities = bbdev_capabilities; +#ifdef ACC101_EXT_MEM + dev_info->harq_buffer_size = d->ddr_size; +#else dev_info->harq_buffer_size = 0; +#endif } static const struct rte_bbdev_ops acc101_bbdev_ops = { @@ -824,6 +868,1695 @@ {.device_id = 0}, }; +/* Read flag value 0/1 from bitmap */ +static inline bool +check_bit(uint32_t bitmap, uint32_t bitmask) +{ + return bitmap & bitmask; +} + +static inline char * +mbuf_append(struct rte_mbuf *m_head, struct rte_mbuf *m, uint16_t len) +{ + if (unlikely(len > rte_pktmbuf_tailroom(m))) + return NULL; + + char *tail = (char *)m->buf_addr + m->data_off + m->data_len; + m->data_len = (uint16_t)(m->data_len + len); + m_head->pkt_len = (m_head->pkt_len + len); + return tail; +} + +/* Compute value of k0. + * Based on 3GPP 38.212 Table 5.4.2.1-2 + * Starting position of different redundancy versions, k0 + */ +static inline uint16_t +get_k0(uint16_t n_cb, uint16_t z_c, uint8_t bg, uint8_t rv_index) +{ + if (rv_index == 0) + return 0; + uint16_t n = (bg == 1 ? ACC101_N_ZC_1 : ACC101_N_ZC_2) * z_c; + if (n_cb == n) { + if (rv_index == 1) + return (bg == 1 ? ACC101_K0_1_1 : ACC101_K0_1_2) * z_c; + else if (rv_index == 2) + return (bg == 1 ? ACC101_K0_2_1 : ACC101_K0_2_2) * z_c; + else + return (bg == 1 ? ACC101_K0_3_1 : ACC101_K0_3_2) * z_c; + } + /* LBRM case - includes a division by N */ + if (unlikely(z_c == 0)) + return 0; + if (rv_index == 1) + return (((bg == 1 ? ACC101_K0_1_1 : ACC101_K0_1_2) * n_cb) + / n) * z_c; + else if (rv_index == 2) + return (((bg == 1 ? ACC101_K0_2_1 : ACC101_K0_2_2) * n_cb) + / n) * z_c; + else + return (((bg == 1 ? ACC101_K0_3_1 : ACC101_K0_3_2) * n_cb) + / n) * z_c; +} + +/* Fill in a frame control word for LDPC encoding. */ +static inline void +acc101_fcw_le_fill(const struct rte_bbdev_enc_op *op, + struct acc101_fcw_le *fcw, int num_cb, uint32_t default_e) +{ + fcw->qm = op->ldpc_enc.q_m; + fcw->nfiller = op->ldpc_enc.n_filler; + fcw->BG = (op->ldpc_enc.basegraph - 1); + fcw->Zc = op->ldpc_enc.z_c; + fcw->ncb = op->ldpc_enc.n_cb; + fcw->k0 = get_k0(fcw->ncb, fcw->Zc, op->ldpc_enc.basegraph, + op->ldpc_enc.rv_index); + fcw->rm_e = (default_e == 0) ? op->ldpc_enc.cb_params.e : default_e; + fcw->crc_select = check_bit(op->ldpc_enc.op_flags, + RTE_BBDEV_LDPC_CRC_24B_ATTACH); + fcw->bypass_intlv = check_bit(op->ldpc_enc.op_flags, + RTE_BBDEV_LDPC_INTERLEAVER_BYPASS); + fcw->mcb_count = num_cb; +} + +/* Convert offset to harq index for harq_layout structure */ +static inline uint32_t hq_index(uint32_t offset) +{ + return (offset >> ACC101_HARQ_OFFSET_SHIFT) & ACC101_HARQ_OFFSET_MASK; +} + +/* Fill in a frame control word for LDPC decoding. */ +static inline void +acc101_fcw_ld_fill(struct rte_bbdev_dec_op *op, struct acc101_fcw_ld *fcw, + union acc101_harq_layout_data *harq_layout) +{ + uint16_t harq_out_length, harq_in_length, ncb_p, k0_p, parity_offset; + uint32_t harq_index; + uint32_t l; + + fcw->qm = op->ldpc_dec.q_m; + fcw->nfiller = op->ldpc_dec.n_filler; + fcw->BG = (op->ldpc_dec.basegraph - 1); + fcw->Zc = op->ldpc_dec.z_c; + fcw->ncb = op->ldpc_dec.n_cb; + fcw->k0 = get_k0(fcw->ncb, fcw->Zc, op->ldpc_dec.basegraph, + op->ldpc_dec.rv_index); + if (op->ldpc_dec.code_block_mode == RTE_BBDEV_CODE_BLOCK) + fcw->rm_e = op->ldpc_dec.cb_params.e; + else + fcw->rm_e = (op->ldpc_dec.tb_params.r < + op->ldpc_dec.tb_params.cab) ? + op->ldpc_dec.tb_params.ea : + op->ldpc_dec.tb_params.eb; + + if (unlikely(check_bit(op->ldpc_dec.op_flags, + RTE_BBDEV_LDPC_HQ_COMBINE_IN_ENABLE) && + (op->ldpc_dec.harq_combined_input.length == 0))) { + rte_bbdev_log(WARNING, "Null HARQ input size provided"); + /* Disable HARQ input in that case to carry forward */ + op->ldpc_dec.op_flags ^= RTE_BBDEV_LDPC_HQ_COMBINE_IN_ENABLE; + } + + fcw->hcin_en = check_bit(op->ldpc_dec.op_flags, + RTE_BBDEV_LDPC_HQ_COMBINE_IN_ENABLE); + fcw->hcout_en = check_bit(op->ldpc_dec.op_flags, + RTE_BBDEV_LDPC_HQ_COMBINE_OUT_ENABLE); + fcw->crc_select = check_bit(op->ldpc_dec.op_flags, + RTE_BBDEV_LDPC_CRC_TYPE_24B_CHECK); + fcw->bypass_dec = check_bit(op->ldpc_dec.op_flags, + RTE_BBDEV_LDPC_DECODE_BYPASS); + fcw->bypass_intlv = check_bit(op->ldpc_dec.op_flags, + RTE_BBDEV_LDPC_DEINTERLEAVER_BYPASS); + if (op->ldpc_dec.q_m == 1) { + fcw->bypass_intlv = 1; + fcw->qm = 2; + } + fcw->hcin_decomp_mode = check_bit(op->ldpc_dec.op_flags, + RTE_BBDEV_LDPC_HARQ_6BIT_COMPRESSION); + fcw->hcout_comp_mode = check_bit(op->ldpc_dec.op_flags, + RTE_BBDEV_LDPC_HARQ_6BIT_COMPRESSION); + fcw->llr_pack_mode = check_bit(op->ldpc_dec.op_flags, + RTE_BBDEV_LDPC_LLR_COMPRESSION); + harq_index = hq_index(op->ldpc_dec.harq_combined_output.offset); + if (fcw->hcin_en > 0) { + harq_in_length = op->ldpc_dec.harq_combined_input.length; + if (fcw->hcin_decomp_mode > 0) + harq_in_length = harq_in_length * 8 / 6; + harq_in_length = RTE_MIN(harq_in_length, op->ldpc_dec.n_cb + - op->ldpc_dec.n_filler); + /* Alignment on next 64B - Already enforced from HC output */ + harq_in_length = RTE_ALIGN_FLOOR(harq_in_length, 64); + fcw->hcin_size0 = harq_in_length; + fcw->hcin_offset = 0; + fcw->hcin_size1 = 0; + } else { + fcw->hcin_size0 = 0; + fcw->hcin_offset = 0; + fcw->hcin_size1 = 0; + } + + fcw->itmax = op->ldpc_dec.iter_max; + fcw->itstop = check_bit(op->ldpc_dec.op_flags, + RTE_BBDEV_LDPC_ITERATION_STOP_ENABLE); + fcw->synd_precoder = fcw->itstop; + /* + * These are all implicitly set + * fcw->synd_post = 0; + * fcw->so_en = 0; + * fcw->so_bypass_rm = 0; + * fcw->so_bypass_intlv = 0; + * fcw->dec_convllr = 0; + * fcw->hcout_convllr = 0; + * fcw->hcout_size1 = 0; + * fcw->so_it = 0; + * fcw->hcout_offset = 0; + * fcw->negstop_th = 0; + * fcw->negstop_it = 0; + * fcw->negstop_en = 0; + * fcw->gain_i = 1; + * fcw->gain_h = 1; + */ + if (fcw->hcout_en > 0) { + parity_offset = (op->ldpc_dec.basegraph == 1 ? 20 : 8) + * op->ldpc_dec.z_c - op->ldpc_dec.n_filler; + k0_p = (fcw->k0 > parity_offset) ? + fcw->k0 - op->ldpc_dec.n_filler : fcw->k0; + ncb_p = fcw->ncb - op->ldpc_dec.n_filler; + l = RTE_MIN(k0_p + fcw->rm_e, INT16_MAX); + harq_out_length = (uint16_t) fcw->hcin_size0; + harq_out_length = RTE_MAX(harq_out_length, l); + /* Cannot exceed the pruned Ncb circular buffer */ + harq_out_length = RTE_MIN(harq_out_length, ncb_p); + /* Alignment on next 64B */ + harq_out_length = RTE_ALIGN_CEIL(harq_out_length, 64); + fcw->hcout_size0 = harq_out_length; + fcw->hcout_size1 = 0; + fcw->hcout_offset = 0; + harq_layout[harq_index].offset = fcw->hcout_offset; + harq_layout[harq_index].size0 = fcw->hcout_size0; + } else { + fcw->hcout_size0 = 0; + fcw->hcout_size1 = 0; + fcw->hcout_offset = 0; + } +} + +/** + * Fills descriptor with data pointers of one block type. + * + * @param desc + * Pointer to DMA descriptor. + * @param input + * Pointer to pointer to input data which will be encoded. It can be changed + * and points to next segment in scatter-gather case. + * @param offset + * Input offset in rte_mbuf structure. It is used for calculating the point + * where data is starting. + * @param cb_len + * Length of currently processed Code Block + * @param seg_total_left + * It indicates how many bytes still left in segment (mbuf) for further + * processing. + * @param op_flags + * Store information about device capabilities + * @param next_triplet + * Index for ACC101 DMA Descriptor triplet + * @param scattergather + * Flag to support scatter-gather for the mbuf + * + * @return + * Returns index of next triplet on success, other value if lengths of + * pkt and processed cb do not match. + * + */ +static inline int +acc101_dma_fill_blk_type_in(struct acc101_dma_req_desc *desc, + struct rte_mbuf **input, uint32_t *offset, uint32_t cb_len, + uint32_t *seg_total_left, int next_triplet, + bool scattergather) +{ + uint32_t part_len; + struct rte_mbuf *m = *input; + + if (scattergather) + part_len = (*seg_total_left < cb_len) ? + *seg_total_left : cb_len; + else + part_len = cb_len; + cb_len -= part_len; + *seg_total_left -= part_len; + + desc->data_ptrs[next_triplet].address = + rte_pktmbuf_iova_offset(m, *offset); + desc->data_ptrs[next_triplet].blen = part_len; + desc->data_ptrs[next_triplet].blkid = ACC101_DMA_BLKID_IN; + desc->data_ptrs[next_triplet].last = 0; + desc->data_ptrs[next_triplet].dma_ext = 0; + *offset += part_len; + next_triplet++; + + while (cb_len > 0) { + if (next_triplet < ACC101_DMA_MAX_NUM_POINTERS_IN && m->next != NULL) { + + m = m->next; + *seg_total_left = rte_pktmbuf_data_len(m); + part_len = (*seg_total_left < cb_len) ? + *seg_total_left : + cb_len; + desc->data_ptrs[next_triplet].address = + rte_pktmbuf_iova_offset(m, 0); + desc->data_ptrs[next_triplet].blen = part_len; + desc->data_ptrs[next_triplet].blkid = + ACC101_DMA_BLKID_IN; + desc->data_ptrs[next_triplet].last = 0; + desc->data_ptrs[next_triplet].dma_ext = 0; + cb_len -= part_len; + *seg_total_left -= part_len; + /* Initializing offset for next segment (mbuf) */ + *offset = part_len; + next_triplet++; + } else { + rte_bbdev_log(ERR, + "Some data still left for processing: " + "data_left: %u, next_triplet: %u, next_mbuf: %p", + cb_len, next_triplet, m->next); + return -EINVAL; + } + } + /* Storing new mbuf as it could be changed in scatter-gather case*/ + *input = m; + + return next_triplet; +} + +/* Fills descriptor with data pointers of one block type. + * Returns index of next triplet on success, other value if lengths of + * output data and processed mbuf do not match. + */ +static inline int +acc101_dma_fill_blk_type_out(struct acc101_dma_req_desc *desc, + struct rte_mbuf *output, uint32_t out_offset, + uint32_t output_len, int next_triplet, int blk_id) +{ + desc->data_ptrs[next_triplet].address = + rte_pktmbuf_iova_offset(output, out_offset); + desc->data_ptrs[next_triplet].blen = output_len; + desc->data_ptrs[next_triplet].blkid = blk_id; + desc->data_ptrs[next_triplet].last = 0; + desc->data_ptrs[next_triplet].dma_ext = 0; + next_triplet++; + + return next_triplet; +} + +static inline void +acc101_header_init(struct acc101_dma_req_desc *desc) +{ + desc->word0 = ACC101_DMA_DESC_TYPE; + desc->word1 = 0; /**< Timestamp could be disabled */ + desc->word2 = 0; + desc->word3 = 0; + desc->numCBs = 1; +} + +#ifdef RTE_LIBRTE_BBDEV_DEBUG +/* Check if any input data is unexpectedly left for processing */ +static inline int +check_mbuf_total_left(uint32_t mbuf_total_left) +{ + if (mbuf_total_left == 0) + return 0; + rte_bbdev_log(ERR, + "Some date still left for processing: mbuf_total_left = %u", + mbuf_total_left); + return -EINVAL; +} +#endif + +static inline int +acc101_dma_desc_le_fill(struct rte_bbdev_enc_op *op, + struct acc101_dma_req_desc *desc, struct rte_mbuf **input, + struct rte_mbuf *output, uint32_t *in_offset, + uint32_t *out_offset, uint32_t *out_length, + uint32_t *mbuf_total_left, uint32_t *seg_total_left) +{ + int next_triplet = 1; /* FCW already done */ + uint16_t K, in_length_in_bits, in_length_in_bytes; + struct rte_bbdev_op_ldpc_enc *enc = &op->ldpc_enc; + + acc101_header_init(desc); + + K = (enc->basegraph == 1 ? 22 : 10) * enc->z_c; + in_length_in_bits = K - enc->n_filler; + if (enc->op_flags & RTE_BBDEV_LDPC_CRC_24B_ATTACH) + in_length_in_bits -= 24; + in_length_in_bytes = in_length_in_bits >> 3; + + if (unlikely((*mbuf_total_left == 0) || + (*mbuf_total_left < in_length_in_bytes))) { + rte_bbdev_log(ERR, + "Mismatch between mbuf length and included CB sizes: mbuf len %u, cb len %u", + *mbuf_total_left, in_length_in_bytes); + return -1; + } + + next_triplet = acc101_dma_fill_blk_type_in(desc, input, in_offset, + in_length_in_bytes, + seg_total_left, next_triplet, + false); + if (unlikely(next_triplet < 0)) { + rte_bbdev_log(ERR, + "Mismatch between data to process and mbuf data length in bbdev_op: %p", + op); + return -1; + } + desc->data_ptrs[next_triplet - 1].last = 1; + desc->m2dlen = next_triplet; + *mbuf_total_left -= in_length_in_bytes; + + /* Set output length */ + /* Integer round up division by 8 */ + *out_length = (enc->cb_params.e + 7) >> 3; + + next_triplet = acc101_dma_fill_blk_type_out(desc, output, *out_offset, + *out_length, next_triplet, ACC101_DMA_BLKID_OUT_ENC); + op->ldpc_enc.output.length += *out_length; + *out_offset += *out_length; + desc->data_ptrs[next_triplet - 1].last = 1; + desc->data_ptrs[next_triplet - 1].dma_ext = 0; + desc->d2mlen = next_triplet - desc->m2dlen; + + desc->op_addr = op; + + return 0; +} + +static inline int +acc101_dma_desc_ld_fill(struct rte_bbdev_dec_op *op, + struct acc101_dma_req_desc *desc, + struct rte_mbuf **input, struct rte_mbuf *h_output, + uint32_t *in_offset, uint32_t *h_out_offset, + uint32_t *h_out_length, uint32_t *mbuf_total_left, + uint32_t *seg_total_left, + struct acc101_fcw_ld *fcw) +{ + struct rte_bbdev_op_ldpc_dec *dec = &op->ldpc_dec; + int next_triplet = 1; /* FCW already done */ + uint32_t input_length; + uint16_t output_length, crc24_overlap = 0; + uint16_t sys_cols, K, h_p_size, h_np_size; + bool h_comp = check_bit(dec->op_flags, + RTE_BBDEV_LDPC_HARQ_6BIT_COMPRESSION); + + acc101_header_init(desc); + + if (check_bit(op->ldpc_dec.op_flags, + RTE_BBDEV_LDPC_CRC_TYPE_24B_DROP)) + crc24_overlap = 24; + + /* Compute some LDPC BG lengths */ + input_length = fcw->rm_e; + if (check_bit(op->ldpc_dec.op_flags, + RTE_BBDEV_LDPC_LLR_COMPRESSION)) + input_length = (input_length * 3 + 3) / 4; + sys_cols = (dec->basegraph == 1) ? 22 : 10; + K = sys_cols * dec->z_c; + output_length = K - dec->n_filler - crc24_overlap; + + if (unlikely((*mbuf_total_left == 0) || + (*mbuf_total_left < input_length))) { + rte_bbdev_log(ERR, + "Mismatch between mbuf length and included CB sizes: mbuf len %u, cb len %u", + *mbuf_total_left, input_length); + return -1; + } + + next_triplet = acc101_dma_fill_blk_type_in(desc, input, + in_offset, input_length, + seg_total_left, next_triplet, + check_bit(op->ldpc_dec.op_flags, + RTE_BBDEV_LDPC_DEC_SCATTER_GATHER)); + + if (unlikely(next_triplet < 0)) { + rte_bbdev_log(ERR, + "Mismatch between data to process and mbuf data length in bbdev_op: %p", + op); + return -1; + } + + if (check_bit(op->ldpc_dec.op_flags, + RTE_BBDEV_LDPC_HQ_COMBINE_IN_ENABLE)) { + h_p_size = fcw->hcin_size0 + fcw->hcin_size1; + if (h_comp) + h_p_size = (h_p_size * 3 + 3) / 4; + desc->data_ptrs[next_triplet].address = + dec->harq_combined_input.offset; + desc->data_ptrs[next_triplet].blen = h_p_size; + desc->data_ptrs[next_triplet].blkid = ACC101_DMA_BLKID_IN_HARQ; + desc->data_ptrs[next_triplet].dma_ext = 1; +#ifndef ACC101_EXT_MEM + acc101_dma_fill_blk_type_out( + desc, + op->ldpc_dec.harq_combined_input.data, + op->ldpc_dec.harq_combined_input.offset, + h_p_size, + next_triplet, + ACC101_DMA_BLKID_IN_HARQ); +#endif + next_triplet++; + } + + desc->data_ptrs[next_triplet - 1].last = 1; + desc->m2dlen = next_triplet; + *mbuf_total_left -= input_length; + + next_triplet = acc101_dma_fill_blk_type_out(desc, h_output, + *h_out_offset, output_length >> 3, next_triplet, + ACC101_DMA_BLKID_OUT_HARD); + + if (check_bit(op->ldpc_dec.op_flags, + RTE_BBDEV_LDPC_HQ_COMBINE_OUT_ENABLE)) { + /* Pruned size of the HARQ */ + h_p_size = fcw->hcout_size0 + fcw->hcout_size1; + /* Non-Pruned size of the HARQ */ + h_np_size = fcw->hcout_offset > 0 ? + fcw->hcout_offset + fcw->hcout_size1 : + h_p_size; + if (h_comp) { + h_np_size = (h_np_size * 3 + 3) / 4; + h_p_size = (h_p_size * 3 + 3) / 4; + } + dec->harq_combined_output.length = h_np_size; + desc->data_ptrs[next_triplet].address = + dec->harq_combined_output.offset; + desc->data_ptrs[next_triplet].blen = h_p_size; + desc->data_ptrs[next_triplet].blkid = ACC101_DMA_BLKID_OUT_HARQ; + desc->data_ptrs[next_triplet].dma_ext = 1; +#ifndef ACC101_EXT_MEM + acc101_dma_fill_blk_type_out( + desc, + dec->harq_combined_output.data, + dec->harq_combined_output.offset, + h_p_size, + next_triplet, + ACC101_DMA_BLKID_OUT_HARQ); +#endif + next_triplet++; + } + + *h_out_length = output_length >> 3; + dec->hard_output.length += *h_out_length; + *h_out_offset += *h_out_length; + desc->data_ptrs[next_triplet - 1].last = 1; + desc->d2mlen = next_triplet - desc->m2dlen; + + desc->op_addr = op; + + return 0; +} + +static inline void +acc101_dma_desc_ld_update(struct rte_bbdev_dec_op *op, + struct acc101_dma_req_desc *desc, + struct rte_mbuf *input, struct rte_mbuf *h_output, + uint32_t *in_offset, uint32_t *h_out_offset, + uint32_t *h_out_length, + union acc101_harq_layout_data *harq_layout) +{ + int next_triplet = 1; /* FCW already done */ + desc->data_ptrs[next_triplet].address = + rte_pktmbuf_iova_offset(input, *in_offset); + next_triplet++; + + if (check_bit(op->ldpc_dec.op_flags, + RTE_BBDEV_LDPC_HQ_COMBINE_IN_ENABLE)) { + struct rte_bbdev_op_data hi = op->ldpc_dec.harq_combined_input; + desc->data_ptrs[next_triplet].address = hi.offset; +#ifndef ACC101_EXT_MEM + desc->data_ptrs[next_triplet].address = + rte_pktmbuf_iova_offset(hi.data, hi.offset); +#endif + next_triplet++; + } + + desc->data_ptrs[next_triplet].address = + rte_pktmbuf_iova_offset(h_output, *h_out_offset); + *h_out_length = desc->data_ptrs[next_triplet].blen; + next_triplet++; + + if (check_bit(op->ldpc_dec.op_flags, + RTE_BBDEV_LDPC_HQ_COMBINE_OUT_ENABLE)) { + desc->data_ptrs[next_triplet].address = + op->ldpc_dec.harq_combined_output.offset; + /* Adjust based on previous operation */ + struct rte_bbdev_dec_op *prev_op = desc->op_addr; + op->ldpc_dec.harq_combined_output.length = + prev_op->ldpc_dec.harq_combined_output.length; + uint32_t harq_idx = hq_index( + op->ldpc_dec.harq_combined_output.offset); + uint32_t prev_harq_idx = hq_index( + prev_op->ldpc_dec.harq_combined_output.offset); + harq_layout[harq_idx].val = harq_layout[prev_harq_idx].val; +#ifndef ACC101_EXT_MEM + struct rte_bbdev_op_data ho = + op->ldpc_dec.harq_combined_output; + desc->data_ptrs[next_triplet].address = + rte_pktmbuf_iova_offset(ho.data, ho.offset); +#endif + next_triplet++; + } + + op->ldpc_dec.hard_output.length += *h_out_length; + desc->op_addr = op; +} + + +/* Enqueue a number of operations to HW and update software rings */ +static inline void +acc101_dma_enqueue(struct acc101_queue *q, uint16_t n, + struct rte_bbdev_stats *queue_stats) +{ + union acc101_enqueue_reg_fmt enq_req; +#ifdef RTE_BBDEV_OFFLOAD_COST + uint64_t start_time = 0; + queue_stats->acc_offload_cycles = 0; +#else + RTE_SET_USED(queue_stats); +#endif + + enq_req.val = 0; + /* Setting offset, 100b for 256 DMA Desc */ + enq_req.addr_offset = ACC101_DESC_OFFSET; + + /* Split ops into batches */ + do { + union acc101_dma_desc *desc; + uint16_t enq_batch_size; + uint64_t offset; + rte_iova_t req_elem_addr; + + enq_batch_size = RTE_MIN(n, MAX_ENQ_BATCH_SIZE); + + /* Set flag on last descriptor in a batch */ + desc = q->ring_addr + ((q->sw_ring_head + enq_batch_size - 1) & + q->sw_ring_wrap_mask); + desc->req.last_desc_in_batch = 1; + + /* Calculate the 1st descriptor's address */ + offset = ((q->sw_ring_head & q->sw_ring_wrap_mask) * + sizeof(union acc101_dma_desc)); + req_elem_addr = q->ring_addr_iova + offset; + + /* Fill enqueue struct */ + enq_req.num_elem = enq_batch_size; + /* low 6 bits are not needed */ + enq_req.req_elem_addr = (uint32_t)(req_elem_addr >> 6); + +#ifdef RTE_LIBRTE_BBDEV_DEBUG + rte_memdump(stderr, "Req sdone", desc, sizeof(*desc)); +#endif + rte_bbdev_log_debug( + "Enqueue %u reqs (phys %#"PRIx64") to reg %p", + enq_batch_size, + req_elem_addr, + (void *)q->mmio_reg_enqueue); + + rte_wmb(); + +#ifdef RTE_BBDEV_OFFLOAD_COST + /* Start time measurement for enqueue function offload. */ + start_time = rte_rdtsc_precise(); +#endif + rte_bbdev_log(DEBUG, "Debug : MMIO Enqueue"); + mmio_write(q->mmio_reg_enqueue, enq_req.val); + +#ifdef RTE_BBDEV_OFFLOAD_COST + queue_stats->acc_offload_cycles += + rte_rdtsc_precise() - start_time; +#endif + + q->aq_enqueued++; + q->sw_ring_head += enq_batch_size; + n -= enq_batch_size; + + } while (n); + + +} + +/* Enqueue one encode operations for ACC101 device in CB mode + * multiplexed on the same descriptor + */ +static inline int +enqueue_ldpc_enc_n_op_cb(struct acc101_queue *q, struct rte_bbdev_enc_op **ops, + uint16_t total_enqueued_descs, int16_t num) +{ + union acc101_dma_desc *desc = NULL; + uint32_t out_length; + struct rte_mbuf *output_head, *output; + int i, next_triplet; + uint16_t in_length_in_bytes; + struct rte_bbdev_op_ldpc_enc *enc = &ops[0]->ldpc_enc; + + uint16_t desc_idx = ((q->sw_ring_head + total_enqueued_descs) + & q->sw_ring_wrap_mask); + desc = q->ring_addr + desc_idx; + acc101_fcw_le_fill(ops[0], &desc->req.fcw_le, num, 0); + + /** This could be done at polling */ + acc101_header_init(&desc->req); + desc->req.numCBs = num; + + in_length_in_bytes = ops[0]->ldpc_enc.input.data->data_len; + out_length = (enc->cb_params.e + 7) >> 3; + desc->req.m2dlen = 1 + num; + desc->req.d2mlen = num; + next_triplet = 1; + + for (i = 0; i < num; i++) { + desc->req.data_ptrs[next_triplet].address = + rte_pktmbuf_iova_offset(ops[i]->ldpc_enc.input.data, 0); + desc->req.data_ptrs[next_triplet].blen = in_length_in_bytes; + next_triplet++; + desc->req.data_ptrs[next_triplet].address = + rte_pktmbuf_iova_offset( + ops[i]->ldpc_enc.output.data, 0); + desc->req.data_ptrs[next_triplet].blen = out_length; + next_triplet++; + ops[i]->ldpc_enc.output.length = out_length; + output_head = output = ops[i]->ldpc_enc.output.data; + mbuf_append(output_head, output, out_length); + output->data_len = out_length; + } + + desc->req.op_addr = ops[0]; + /* Keep track of pointers even when multiplexed in single descriptor */ + struct acc101_ptrs *context_ptrs = q->companion_ring_addr + desc_idx; + for (i = 0; i < num; i++) + context_ptrs->ptr[i].op_addr = ops[i]; + +#ifdef RTE_LIBRTE_BBDEV_DEBUG + rte_memdump(stderr, "FCW", &desc->req.fcw_le, + sizeof(desc->req.fcw_le) - 8); + rte_memdump(stderr, "Req Desc.", desc, sizeof(*desc)); +#endif + + /* One CB (one op) was successfully prepared to enqueue */ + return num; +} + +/* Enqueue one encode operations for ACC101 device for a partial TB + * all codes blocks have same configuration multiplexed on the same descriptor + */ +static inline void +enqueue_ldpc_enc_part_tb(struct acc101_queue *q, struct rte_bbdev_enc_op *op, + uint16_t total_enqueued_descs, int16_t num_cbs, uint32_t e, + uint16_t in_len_B, uint32_t out_len_B, uint32_t *in_offset, + uint32_t *out_offset) +{ + + union acc101_dma_desc *desc = NULL; + struct rte_mbuf *output_head, *output; + int i, next_triplet; + struct rte_bbdev_op_ldpc_enc *enc = &op->ldpc_enc; + + + uint16_t desc_idx = ((q->sw_ring_head + total_enqueued_descs) + & q->sw_ring_wrap_mask); + desc = q->ring_addr + desc_idx; + acc101_fcw_le_fill(op, &desc->req.fcw_le, num_cbs, e); + + /** This could be done at polling */ + acc101_header_init(&desc->req); + desc->req.numCBs = num_cbs; + + desc->req.m2dlen = 1 + num_cbs; + desc->req.d2mlen = num_cbs; + next_triplet = 1; + + for (i = 0; i < num_cbs; i++) { + desc->req.data_ptrs[next_triplet].address = + rte_pktmbuf_iova_offset(enc->input.data, + *in_offset); + *in_offset += in_len_B; + desc->req.data_ptrs[next_triplet].blen = in_len_B; + next_triplet++; + desc->req.data_ptrs[next_triplet].address = + rte_pktmbuf_iova_offset( + enc->output.data, *out_offset); + *out_offset += out_len_B; + desc->req.data_ptrs[next_triplet].blen = out_len_B; + next_triplet++; + enc->output.length += out_len_B; + output_head = output = enc->output.data; + mbuf_append(output_head, output, out_len_B); + } + +#ifdef RTE_LIBRTE_BBDEV_DEBUG + rte_memdump(stderr, "FCW", &desc->req.fcw_le, + sizeof(desc->req.fcw_le) - 8); + rte_memdump(stderr, "Req Desc.", desc, sizeof(*desc)); +#endif + +} + +/* Enqueue one encode operations for ACC101 device in CB mode */ +static inline int +enqueue_ldpc_enc_one_op_cb(struct acc101_queue *q, struct rte_bbdev_enc_op *op, + uint16_t total_enqueued_cbs) +{ + union acc101_dma_desc *desc = NULL; + int ret; + uint32_t in_offset, out_offset, out_length, mbuf_total_left, + seg_total_left; + struct rte_mbuf *input, *output_head, *output; + + uint16_t desc_idx = ((q->sw_ring_head + total_enqueued_cbs) + & q->sw_ring_wrap_mask); + desc = q->ring_addr + desc_idx; + acc101_fcw_le_fill(op, &desc->req.fcw_le, 1, 0); + + input = op->ldpc_enc.input.data; + output_head = output = op->ldpc_enc.output.data; + in_offset = op->ldpc_enc.input.offset; + out_offset = op->ldpc_enc.output.offset; + out_length = 0; + mbuf_total_left = op->ldpc_enc.input.length; + seg_total_left = rte_pktmbuf_data_len(op->ldpc_enc.input.data) + - in_offset; + + ret = acc101_dma_desc_le_fill(op, &desc->req, &input, output, + &in_offset, &out_offset, &out_length, &mbuf_total_left, + &seg_total_left); + + if (unlikely(ret < 0)) + return ret; + + mbuf_append(output_head, output, out_length); + +#ifdef RTE_LIBRTE_BBDEV_DEBUG + rte_memdump(stderr, "FCW", &desc->req.fcw_le, + sizeof(desc->req.fcw_le) - 8); + rte_memdump(stderr, "Req Desc.", desc, sizeof(*desc)); + + if (check_mbuf_total_left(mbuf_total_left) != 0) + return -EINVAL; +#endif + /* One CB (one op) was successfully prepared to enqueue */ + return 1; +} + +/* Enqueue one encode operations for ACC101 device in TB mode. + * returns the number of descs used + */ +static inline int +enqueue_ldpc_enc_one_op_tb(struct acc101_queue *q, struct rte_bbdev_enc_op *op, + uint16_t enq_descs, uint8_t cbs_in_tb) +{ + uint8_t num_a, num_b; + uint16_t desc_idx; + uint8_t r = op->ldpc_enc.tb_params.r; + uint8_t cab = op->ldpc_enc.tb_params.cab; + union acc101_dma_desc *desc; + uint16_t init_enq_descs = enq_descs; + uint16_t input_len_B = ((op->ldpc_enc.basegraph == 1 ? 22 : 10) * + op->ldpc_enc.z_c - op->ldpc_enc.n_filler) >> 3; + if (check_bit(op->ldpc_enc.op_flags, RTE_BBDEV_LDPC_CRC_24B_ATTACH)) + input_len_B -= 3; + + if (r < cab) { + num_a = cab - r; + num_b = cbs_in_tb - cab; + } else { + num_a = 0; + num_b = cbs_in_tb - r; + } + uint32_t in_offset = 0, out_offset = 0; + + while (num_a > 0) { + uint32_t e = op->ldpc_enc.tb_params.ea; + uint32_t out_len_B = (e + 7) >> 3; + uint8_t enq = RTE_MIN(num_a, ACC101_MUX_5GDL_DESC); + num_a -= enq; + enqueue_ldpc_enc_part_tb(q, op, enq_descs, enq, e, input_len_B, + out_len_B, &in_offset, &out_offset); + enq_descs++; + } + while (num_b > 0) { + uint32_t e = op->ldpc_enc.tb_params.eb; + uint32_t out_len_B = (e + 7) >> 3; + uint8_t enq = RTE_MIN(num_b, ACC101_MUX_5GDL_DESC); + num_b -= enq; + enqueue_ldpc_enc_part_tb(q, op, enq_descs, enq, e, input_len_B, + out_len_B, &in_offset, &out_offset); + enq_descs++; + } + + uint16_t return_descs = enq_descs - init_enq_descs; + /* Keep total number of CBs in first TB */ + desc_idx = ((q->sw_ring_head + init_enq_descs) + & q->sw_ring_wrap_mask); + desc = q->ring_addr + desc_idx; + desc->req.cbs_in_tb = return_descs; /** Actual number of descriptors */ + desc->req.op_addr = op; + + /* Set SDone on last CB descriptor for TB mode. */ + desc_idx = ((q->sw_ring_head + enq_descs - 1) + & q->sw_ring_wrap_mask); + desc = q->ring_addr + desc_idx; + desc->req.sdone_enable = 1; + desc->req.irq_enable = q->irq_enable; + desc->req.op_addr = op; + return return_descs; +} + +/** Enqueue one decode operations for ACC101 device in CB mode */ +static inline int +enqueue_ldpc_dec_one_op_cb(struct acc101_queue *q, struct rte_bbdev_dec_op *op, + uint16_t total_enqueued_cbs, bool same_op, + struct rte_bbdev_queue_data *q_data) +{ + RTE_SET_USED(q_data); + int ret; + + union acc101_dma_desc *desc; + uint16_t desc_idx = ((q->sw_ring_head + total_enqueued_cbs) + & q->sw_ring_wrap_mask); + desc = q->ring_addr + desc_idx; + struct rte_mbuf *input, *h_output_head, *h_output; + uint32_t in_offset, h_out_offset, mbuf_total_left, h_out_length = 0; + input = op->ldpc_dec.input.data; + h_output_head = h_output = op->ldpc_dec.hard_output.data; + in_offset = op->ldpc_dec.input.offset; + h_out_offset = op->ldpc_dec.hard_output.offset; + mbuf_total_left = op->ldpc_dec.input.length; +#ifdef RTE_LIBRTE_BBDEV_DEBUG + if (unlikely(input == NULL)) { + rte_bbdev_log(ERR, "Invalid mbuf pointer"); + return -EFAULT; + } +#endif + union acc101_harq_layout_data *harq_layout = q->d->harq_layout; + + if (same_op) { + union acc101_dma_desc *prev_desc; + desc_idx = ((q->sw_ring_head + total_enqueued_cbs - 1) + & q->sw_ring_wrap_mask); + prev_desc = q->ring_addr + desc_idx; + uint8_t *prev_ptr = (uint8_t *) prev_desc; + uint8_t *new_ptr = (uint8_t *) desc; + /* Copy first 4 words and BDESCs */ + rte_memcpy(new_ptr, prev_ptr, ACC101_5GUL_SIZE_0); + rte_memcpy(new_ptr + ACC101_5GUL_OFFSET_0, + prev_ptr + ACC101_5GUL_OFFSET_0, + ACC101_5GUL_SIZE_1); + desc->req.op_addr = prev_desc->req.op_addr; + /* Copy FCW */ + rte_memcpy(new_ptr + ACC101_DESC_FCW_OFFSET, + prev_ptr + ACC101_DESC_FCW_OFFSET, + ACC101_FCW_LD_BLEN); + acc101_dma_desc_ld_update(op, &desc->req, input, h_output, + &in_offset, &h_out_offset, + &h_out_length, harq_layout); + } else { + struct acc101_fcw_ld *fcw; + uint32_t seg_total_left; + fcw = &desc->req.fcw_ld; + acc101_fcw_ld_fill(op, fcw, harq_layout); + + /* Special handling when using mbuf or not */ + if (check_bit(op->ldpc_dec.op_flags, + RTE_BBDEV_LDPC_DEC_SCATTER_GATHER)) + seg_total_left = rte_pktmbuf_data_len(input) + - in_offset; + else + seg_total_left = fcw->rm_e; + + ret = acc101_dma_desc_ld_fill(op, &desc->req, &input, h_output, + &in_offset, &h_out_offset, + &h_out_length, &mbuf_total_left, + &seg_total_left, fcw); + if (unlikely(ret < 0)) + return ret; + } + + /* Hard output */ + mbuf_append(h_output_head, h_output, h_out_length); +#ifndef ACC101_EXT_MEM + if (op->ldpc_dec.harq_combined_output.length > 0) { + /* Push the HARQ output into host memory */ + struct rte_mbuf *hq_output_head, *hq_output; + hq_output_head = op->ldpc_dec.harq_combined_output.data; + hq_output = op->ldpc_dec.harq_combined_output.data; + mbuf_append(hq_output_head, hq_output, + op->ldpc_dec.harq_combined_output.length); + } +#endif + +#ifdef RTE_LIBRTE_BBDEV_DEBUG + rte_memdump(stderr, "FCW", &desc->req.fcw_ld, + sizeof(desc->req.fcw_ld)); + rte_memdump(stderr, "Req Desc.", desc, sizeof(*desc)); +#endif + + /* One CB (one op) was successfully prepared to enqueue */ + return 1; +} + + +/* Enqueue one decode operations for ACC101 device in TB mode */ +static inline int +enqueue_ldpc_dec_one_op_tb(struct acc101_queue *q, struct rte_bbdev_dec_op *op, + uint16_t total_enqueued_cbs, uint8_t cbs_in_tb) +{ + union acc101_dma_desc *desc = NULL; + union acc101_dma_desc *desc_first = NULL; + int ret; + uint8_t r, c; + uint32_t in_offset, h_out_offset, + h_out_length, mbuf_total_left, seg_total_left; + struct rte_mbuf *input, *h_output_head, *h_output; + uint16_t current_enqueued_cbs = 0; + + uint16_t desc_idx = ((q->sw_ring_head + total_enqueued_cbs) + & q->sw_ring_wrap_mask); + desc = q->ring_addr + desc_idx; + desc_first = desc; + uint64_t fcw_offset = (desc_idx << 8) + ACC101_DESC_FCW_OFFSET; + union acc101_harq_layout_data *harq_layout = q->d->harq_layout; + acc101_fcw_ld_fill(op, &desc->req.fcw_ld, harq_layout); + + input = op->ldpc_dec.input.data; + h_output_head = h_output = op->ldpc_dec.hard_output.data; + in_offset = op->ldpc_dec.input.offset; + h_out_offset = op->ldpc_dec.hard_output.offset; + h_out_length = 0; + mbuf_total_left = op->ldpc_dec.input.length; + c = op->ldpc_dec.tb_params.c; + r = op->ldpc_dec.tb_params.r; + + while (mbuf_total_left > 0 && r < c) { + if (check_bit(op->ldpc_dec.op_flags, + RTE_BBDEV_LDPC_DEC_SCATTER_GATHER)) + seg_total_left = rte_pktmbuf_data_len(input) + - in_offset; + else + seg_total_left = op->ldpc_dec.input.length; + /* Set up DMA descriptor */ + desc = q->ring_addr + ((q->sw_ring_head + total_enqueued_cbs) + & q->sw_ring_wrap_mask); + desc->req.data_ptrs[0].address = q->ring_addr_iova + fcw_offset; + desc->req.data_ptrs[0].blen = ACC101_FCW_LD_BLEN; + rte_memcpy(&desc->req.fcw_ld, &desc_first->req.fcw_ld, + ACC101_FCW_LD_BLEN); + ret = acc101_dma_desc_ld_fill(op, &desc->req, &input, + h_output, &in_offset, &h_out_offset, + &h_out_length, + &mbuf_total_left, &seg_total_left, + &desc->req.fcw_ld); + + if (unlikely(ret < 0)) + return ret; + + /* Hard output */ + mbuf_append(h_output_head, h_output, h_out_length); + + /* Set total number of CBs in TB */ + desc->req.cbs_in_tb = cbs_in_tb; +#ifdef RTE_LIBRTE_BBDEV_DEBUG + rte_memdump(stderr, "FCW", &desc->req.fcw_td, + sizeof(desc->req.fcw_td) - 8); + rte_memdump(stderr, "Req Desc.", desc, sizeof(*desc)); +#endif + if (check_bit(op->ldpc_dec.op_flags, + RTE_BBDEV_LDPC_DEC_SCATTER_GATHER) + && (seg_total_left == 0)) { + /* Go to the next mbuf */ + input = input->next; + in_offset = 0; + h_output = h_output->next; + h_out_offset = 0; + } + total_enqueued_cbs++; + current_enqueued_cbs++; + r++; + } +#ifdef RTE_LIBRTE_BBDEV_DEBUG + if (check_mbuf_total_left(mbuf_total_left) != 0) + return -EINVAL; +#endif + /* Set SDone on last CB descriptor for TB mode */ + desc->req.sdone_enable = 1; + desc->req.irq_enable = q->irq_enable; + + return current_enqueued_cbs; +} + +/* Calculates number of CBs in processed encoder TB based on 'r' and input + * length. + */ +static inline uint8_t +get_num_cbs_in_tb_ldpc_enc(struct rte_bbdev_op_ldpc_enc *ldpc_enc) +{ + uint8_t c, r, crc24_bits = 0; + uint16_t k = (ldpc_enc->basegraph == 1 ? 22 : 10) * ldpc_enc->z_c + - ldpc_enc->n_filler; + uint8_t cbs_in_tb = 0; + int32_t length; + + length = ldpc_enc->input.length; + r = ldpc_enc->tb_params.r; + c = ldpc_enc->tb_params.c; + crc24_bits = 0; + if (check_bit(ldpc_enc->op_flags, RTE_BBDEV_LDPC_CRC_24B_ATTACH)) + crc24_bits = 24; + while (length > 0 && r < c) { + length -= (k - crc24_bits) >> 3; + r++; + cbs_in_tb++; + } + return cbs_in_tb; +} + +/* Calculates number of CBs in processed decoder TB based on 'r' and input + * length. + */ +static inline uint16_t +get_num_cbs_in_tb_ldpc_dec(struct rte_bbdev_op_ldpc_dec *ldpc_dec) +{ + uint16_t r, cbs_in_tb = 0; + int32_t length = ldpc_dec->input.length; + r = ldpc_dec->tb_params.r; + while (length > 0 && r < ldpc_dec->tb_params.c) { + length -= (r < ldpc_dec->tb_params.cab) ? + ldpc_dec->tb_params.ea : + ldpc_dec->tb_params.eb; + r++; + cbs_in_tb++; + } + return cbs_in_tb; +} + +/* Number of available descriptor in ring to enqueue */ +static uint32_t +acc101_ring_avail_enq(struct acc101_queue *q) +{ + return (q->sw_ring_depth - 1 + q->sw_ring_tail - q->sw_ring_head) % q->sw_ring_depth; +} + +/* Number of available descriptor in ring to dequeue */ +static uint32_t +acc101_ring_avail_deq(struct acc101_queue *q) +{ + return (q->sw_ring_depth + q->sw_ring_head - q->sw_ring_tail) % q->sw_ring_depth; +} + +/* Check we can mux encode operations with common FCW */ +static inline int16_t +check_mux(struct rte_bbdev_enc_op **ops, uint16_t num) { + uint16_t i; + if (num <= 1) + return 1; + for (i = 1; i < num; ++i) { + /* Only mux compatible code blocks */ + if (memcmp((uint8_t *)(&ops[i]->ldpc_enc) + ACC101_ENC_OFFSET, + (uint8_t *)(&ops[0]->ldpc_enc) + + ACC101_ENC_OFFSET, + ACC101_CMP_ENC_SIZE) != 0) + return i; + } + /* Avoid multiplexing small inbound size frames */ + int Kp = (ops[0]->ldpc_enc.basegraph == 1 ? 22 : 10) * + ops[0]->ldpc_enc.z_c - ops[0]->ldpc_enc.n_filler; + if (Kp <= ACC101_LIMIT_DL_MUX_BITS) + return 1; + return num; +} + +/** Enqueue encode operations for ACC101 device in CB mode. */ +static inline uint16_t +acc101_enqueue_ldpc_enc_cb(struct rte_bbdev_queue_data *q_data, + struct rte_bbdev_enc_op **ops, uint16_t num) +{ + struct acc101_queue *q = q_data->queue_private; + int32_t avail = acc101_ring_avail_enq(q); + uint16_t i = 0; + union acc101_dma_desc *desc; + int ret, desc_idx = 0; + int16_t enq, left = num; + + while (left > 0) { + if (unlikely(avail < 1)) + break; + avail--; + enq = RTE_MIN(left, ACC101_MUX_5GDL_DESC); + enq = check_mux(&ops[i], enq); + if (enq > 1) { + ret = enqueue_ldpc_enc_n_op_cb(q, &ops[i], + desc_idx, enq); + if (ret < 0) + break; + i += enq; + } else { + ret = enqueue_ldpc_enc_one_op_cb(q, ops[i], desc_idx); + if (ret < 0) + break; + i++; + } + desc_idx++; + left = num - i; + } + + if (unlikely(i == 0)) + return 0; /* Nothing to enqueue */ + + /* Set SDone in last CB in enqueued ops for CB mode*/ + desc = q->ring_addr + ((q->sw_ring_head + desc_idx - 1) + & q->sw_ring_wrap_mask); + desc->req.sdone_enable = 1; + desc->req.irq_enable = q->irq_enable; + + acc101_dma_enqueue(q, desc_idx, &q_data->queue_stats); + + /* Update stats */ + q_data->queue_stats.enqueued_count += i; + q_data->queue_stats.enqueue_err_count += num - i; + + return i; +} + +/* Enqueue LDPC encode operations for ACC101 device in TB mode. */ +static uint16_t +acc101_enqueue_ldpc_enc_tb(struct rte_bbdev_queue_data *q_data, + struct rte_bbdev_enc_op **ops, uint16_t num) +{ + struct acc101_queue *q = q_data->queue_private; + int32_t avail = acc101_ring_avail_enq(q); + uint16_t i, enqueued_descs = 0; + uint8_t cbs_in_tb; + int descs_used; + + for (i = 0; i < num; ++i) { + cbs_in_tb = get_num_cbs_in_tb_ldpc_enc(&ops[i]->ldpc_enc); + /* Check if there are available space for further processing */ + if (unlikely(avail - cbs_in_tb < 0)) + break; + + + descs_used = enqueue_ldpc_enc_one_op_tb(q, ops[i], + enqueued_descs, cbs_in_tb); + if (descs_used < 0) + break; + enqueued_descs += descs_used; + avail -= descs_used; + } + if (unlikely(enqueued_descs == 0)) + return 0; /* Nothing to enqueue */ + + acc101_dma_enqueue(q, enqueued_descs, &q_data->queue_stats); + + /* Update stats */ + q_data->queue_stats.enqueued_count += i; + q_data->queue_stats.enqueue_err_count += num - i; + + return i; +} + +/* Check room in AQ for the enqueues batches into Qmgr */ +static int32_t +acc101_aq_avail(struct rte_bbdev_queue_data *q_data, uint16_t num_ops) +{ + struct acc101_queue *q = q_data->queue_private; + int32_t aq_avail = q->aq_depth - + ((q->aq_enqueued - q->aq_dequeued + + ACC101_MAX_QUEUE_DEPTH) % ACC101_MAX_QUEUE_DEPTH) + - (num_ops >> 7); + return aq_avail; +} + +/* Enqueue encode operations for ACC101 device. */ +static uint16_t +acc101_enqueue_ldpc_enc(struct rte_bbdev_queue_data *q_data, + struct rte_bbdev_enc_op **ops, uint16_t num) +{ + uint16_t ret; + int32_t aq_avail = acc101_aq_avail(q_data, num); + if (unlikely((aq_avail <= 0) || (num == 0))) + return 0; + if (ops[0]->ldpc_enc.code_block_mode == RTE_BBDEV_TRANSPORT_BLOCK) + ret = acc101_enqueue_ldpc_enc_tb(q_data, ops, num); + else + ret = acc101_enqueue_ldpc_enc_cb(q_data, ops, num); + return ret; +} + +/* Enqueue decode operations for ACC101 device in TB mode */ +static uint16_t +acc101_enqueue_ldpc_dec_tb(struct rte_bbdev_queue_data *q_data, + struct rte_bbdev_dec_op **ops, uint16_t num) +{ + struct acc101_queue *q = q_data->queue_private; + int32_t avail = acc101_ring_avail_enq(q); + uint16_t i, enqueued_cbs = 0; + uint8_t cbs_in_tb; + int ret; + + for (i = 0; i < num; ++i) { + cbs_in_tb = get_num_cbs_in_tb_ldpc_dec(&ops[i]->ldpc_dec); + /* Check if there are available space for further processing */ + if (unlikely(avail - cbs_in_tb < 0)) + break; + avail -= cbs_in_tb; + + ret = enqueue_ldpc_dec_one_op_tb(q, ops[i], + enqueued_cbs, cbs_in_tb); + if (ret < 0) + break; + enqueued_cbs += ret; + } + if (unlikely(enqueued_cbs == 0)) + return 0; /* Nothing to enqueue */ + + acc101_dma_enqueue(q, enqueued_cbs, &q_data->queue_stats); + + /* Update stats */ + q_data->queue_stats.enqueued_count += i; + q_data->queue_stats.enqueue_err_count += num - i; + return i; +} + +/* Enqueue decode operations for ACC101 device in CB mode */ +static uint16_t +acc101_enqueue_ldpc_dec_cb(struct rte_bbdev_queue_data *q_data, + struct rte_bbdev_dec_op **ops, uint16_t num) +{ + struct acc101_queue *q = q_data->queue_private; + int32_t avail = acc101_ring_avail_enq(q); + uint16_t i; + union acc101_dma_desc *desc; + int ret; + bool same_op = false; + for (i = 0; i < num; ++i) { + /* Check if there are available space for further processing */ + if (unlikely(avail < 1)) + break; + avail -= 1; + rte_bbdev_log(INFO, "Op %d %d %d %d %d %d %d %d %d %d %d %d\n", + i, ops[i]->ldpc_dec.op_flags, ops[i]->ldpc_dec.rv_index, + ops[i]->ldpc_dec.iter_max, ops[i]->ldpc_dec.iter_count, + ops[i]->ldpc_dec.basegraph, ops[i]->ldpc_dec.z_c, + ops[i]->ldpc_dec.n_cb, ops[i]->ldpc_dec.q_m, + ops[i]->ldpc_dec.n_filler, ops[i]->ldpc_dec.cb_params.e, + same_op); + ret = enqueue_ldpc_dec_one_op_cb(q, ops[i], i, same_op, q_data); + if (ret < 0) + break; + } + + if (unlikely(i == 0)) + return 0; /* Nothing to enqueue */ + + /* Set SDone in last CB in enqueued ops for CB mode*/ + desc = q->ring_addr + ((q->sw_ring_head + i - 1) + & q->sw_ring_wrap_mask); + + desc->req.sdone_enable = 1; + desc->req.irq_enable = q->irq_enable; + + acc101_dma_enqueue(q, i, &q_data->queue_stats); + + /* Update stats */ + q_data->queue_stats.enqueued_count += i; + q_data->queue_stats.enqueue_err_count += num - i; + return i; +} + +/* Enqueue decode operations for ACC101 device. */ +static uint16_t +acc101_enqueue_ldpc_dec(struct rte_bbdev_queue_data *q_data, + struct rte_bbdev_dec_op **ops, uint16_t num) +{ + uint16_t ret; + int32_t aq_avail = acc101_aq_avail(q_data, num); + if (unlikely((aq_avail <= 0) || (num == 0))) + return 0; + if (ops[0]->ldpc_dec.code_block_mode == RTE_BBDEV_TRANSPORT_BLOCK) + ret = acc101_enqueue_ldpc_dec_tb(q_data, ops, num); + else + ret = acc101_enqueue_ldpc_dec_cb(q_data, ops, num); + return ret; +} + + +/* Dequeue one encode operations from ACC101 device in CB mode + */ +static inline int +dequeue_enc_one_op_cb(struct acc101_queue *q, struct rte_bbdev_enc_op **ref_op, + uint16_t *dequeued_ops, uint32_t *aq_dequeued, + uint16_t *dequeued_descs) +{ + union acc101_dma_desc *desc, atom_desc; + union acc101_dma_rsp_desc rsp; + struct rte_bbdev_enc_op *op; + int i; + int desc_idx = ((q->sw_ring_tail + *dequeued_descs) + & q->sw_ring_wrap_mask); + desc = q->ring_addr + desc_idx; + atom_desc.atom_hdr = __atomic_load_n((uint64_t *)desc, + __ATOMIC_RELAXED); + + /* Check fdone bit */ + if (!(atom_desc.rsp.val & ACC101_FDONE)) + return -1; + + rsp.val = atom_desc.rsp.val; + rte_bbdev_log_debug("Resp. desc %p: %x num %d\n", + desc, rsp.val, desc->req.numCBs); + + /* Dequeue */ + op = desc->req.op_addr; + + /* Clearing status, it will be set based on response */ + op->status = 0; + op->status |= ((rsp.dma_err) ? (1 << RTE_BBDEV_DRV_ERROR) : 0); + op->status |= ((rsp.fcw_err) ? (1 << RTE_BBDEV_DRV_ERROR) : 0); + + if (desc->req.last_desc_in_batch) { + (*aq_dequeued)++; + desc->req.last_desc_in_batch = 0; + } + desc->rsp.val = ACC101_DMA_DESC_TYPE; + desc->rsp.add_info_0 = 0; /*Reserved bits */ + desc->rsp.add_info_1 = 0; /*Reserved bits */ + + ref_op[0] = op; + struct acc101_ptrs *context_ptrs = q->companion_ring_addr + desc_idx; + for (i = 1 ; i < desc->req.numCBs; i++) + ref_op[i] = context_ptrs->ptr[i].op_addr; + + /* One op was successfully dequeued */ + (*dequeued_descs)++; + *dequeued_ops += desc->req.numCBs; + return desc->req.numCBs; +} + +/* Dequeue one LDPC encode operations from ACC101 device in TB mode + * That operation may cover multiple descriptors + */ +static inline int +dequeue_enc_one_op_tb(struct acc101_queue *q, struct rte_bbdev_enc_op **ref_op, + uint16_t *dequeued_ops, uint32_t *aq_dequeued, + uint16_t *dequeued_descs) +{ + union acc101_dma_desc *desc, *last_desc, atom_desc; + union acc101_dma_rsp_desc rsp; + struct rte_bbdev_enc_op *op; + uint8_t i = 0; + uint16_t current_dequeued_descs = 0, descs_in_tb; + + desc = q->ring_addr + ((q->sw_ring_tail + *dequeued_descs) + & q->sw_ring_wrap_mask); + atom_desc.atom_hdr = __atomic_load_n((uint64_t *)desc, + __ATOMIC_RELAXED); + + /* Check fdone bit */ + if (!(atom_desc.rsp.val & ACC101_FDONE)) + return -1; + + /* Get number of CBs in dequeued TB */ + descs_in_tb = desc->req.cbs_in_tb; + /* Get last CB */ + last_desc = q->ring_addr + ((q->sw_ring_tail + + *dequeued_descs + descs_in_tb - 1) + & q->sw_ring_wrap_mask); + /* Check if last CB in TB is ready to dequeue (and thus + * the whole TB) - checking sdone bit. If not return. + */ + atom_desc.atom_hdr = __atomic_load_n((uint64_t *)last_desc, + __ATOMIC_RELAXED); + if (!(atom_desc.rsp.val & ACC101_SDONE)) + return -1; + + /* Dequeue */ + op = desc->req.op_addr; + + /* Clearing status, it will be set based on response */ + op->status = 0; + + while (i < descs_in_tb) { + desc = q->ring_addr + ((q->sw_ring_tail + + *dequeued_descs) + & q->sw_ring_wrap_mask); + atom_desc.atom_hdr = __atomic_load_n((uint64_t *)desc, + __ATOMIC_RELAXED); + rsp.val = atom_desc.rsp.val; + rte_bbdev_log_debug("Resp. desc %p: %x descs %d cbs %d\n", + desc, + rsp.val, descs_in_tb, + desc->req.numCBs); + + op->status |= ((rsp.dma_err) ? (1 << RTE_BBDEV_DRV_ERROR) : 0); + op->status |= ((rsp.fcw_err) ? (1 << RTE_BBDEV_DRV_ERROR) : 0); + + if (desc->req.last_desc_in_batch) { + (*aq_dequeued)++; + desc->req.last_desc_in_batch = 0; + } + desc->rsp.val = ACC101_DMA_DESC_TYPE; + desc->rsp.add_info_0 = 0; + desc->rsp.add_info_1 = 0; + (*dequeued_descs)++; + current_dequeued_descs++; + i++; + } + + *ref_op = op; + (*dequeued_ops)++; + return current_dequeued_descs; +} + + +/* Dequeue one decode operations from ACC101 device in CB mode */ +static inline int +dequeue_ldpc_dec_one_op_cb(struct rte_bbdev_queue_data *q_data, + struct acc101_queue *q, struct rte_bbdev_dec_op **ref_op, + uint16_t dequeued_cbs, uint32_t *aq_dequeued) +{ + union acc101_dma_desc *desc, atom_desc; + union acc101_dma_rsp_desc rsp; + struct rte_bbdev_dec_op *op; + + desc = q->ring_addr + ((q->sw_ring_tail + dequeued_cbs) + & q->sw_ring_wrap_mask); + atom_desc.atom_hdr = __atomic_load_n((uint64_t *)desc, + __ATOMIC_RELAXED); + + /* Check fdone bit */ + if (!(atom_desc.rsp.val & ACC101_FDONE)) + return -1; + + rsp.val = atom_desc.rsp.val; + rte_bbdev_log_debug("Resp. desc %p: %x\n", desc, rsp.val); + + /* Dequeue */ + op = desc->req.op_addr; + + /* Clearing status, it will be set based on response */ + op->status = 0; + op->status |= rsp.input_err << RTE_BBDEV_DATA_ERROR; + op->status |= rsp.dma_err << RTE_BBDEV_DRV_ERROR; + op->status |= rsp.fcw_err << RTE_BBDEV_DRV_ERROR; + if (op->status != 0) + q_data->queue_stats.dequeue_err_count++; + + op->status |= rsp.crc_status << RTE_BBDEV_CRC_ERROR; + if (op->ldpc_dec.hard_output.length > 0 && !rsp.synd_ok) + op->status |= 1 << RTE_BBDEV_SYNDROME_ERROR; + op->ldpc_dec.iter_count = (uint8_t) rsp.iter_cnt; + + /* Check if this is the last desc in batch (Atomic Queue) */ + if (desc->req.last_desc_in_batch) { + (*aq_dequeued)++; + desc->req.last_desc_in_batch = 0; + } + + desc->rsp.val = ACC101_DMA_DESC_TYPE; + desc->rsp.add_info_0 = 0; + desc->rsp.add_info_1 = 0; + + *ref_op = op; + + /* One CB (op) was successfully dequeued */ + return 1; +} + +/* Dequeue one decode operations from ACC101 device in TB mode. */ +static inline int +dequeue_dec_one_op_tb(struct acc101_queue *q, struct rte_bbdev_dec_op **ref_op, + uint16_t dequeued_cbs, uint32_t *aq_dequeued) +{ + union acc101_dma_desc *desc, *last_desc, atom_desc; + union acc101_dma_rsp_desc rsp; + struct rte_bbdev_dec_op *op; + uint8_t cbs_in_tb = 1, cb_idx = 0; + + desc = q->ring_addr + ((q->sw_ring_tail + dequeued_cbs) + & q->sw_ring_wrap_mask); + atom_desc.atom_hdr = __atomic_load_n((uint64_t *)desc, + __ATOMIC_RELAXED); + + /* Check fdone bit */ + if (!(atom_desc.rsp.val & ACC101_FDONE)) + return -1; + + /* Dequeue */ + op = desc->req.op_addr; + + /* Get number of CBs in dequeued TB */ + cbs_in_tb = desc->req.cbs_in_tb; + /* Get last CB */ + last_desc = q->ring_addr + ((q->sw_ring_tail + + dequeued_cbs + cbs_in_tb - 1) + & q->sw_ring_wrap_mask); + /* Check if last CB in TB is ready to dequeue (and thus + * the whole TB) - checking sdone bit. If not return. + */ + atom_desc.atom_hdr = __atomic_load_n((uint64_t *)last_desc, + __ATOMIC_RELAXED); + if (!(atom_desc.rsp.val & ACC101_SDONE)) + return -1; + + /* Clearing status, it will be set based on response */ + op->status = 0; + + /* Read remaining CBs if exists */ + while (cb_idx < cbs_in_tb) { + desc = q->ring_addr + ((q->sw_ring_tail + dequeued_cbs) + & q->sw_ring_wrap_mask); + atom_desc.atom_hdr = __atomic_load_n((uint64_t *)desc, + __ATOMIC_RELAXED); + rsp.val = atom_desc.rsp.val; + rte_bbdev_log_debug("Resp. desc %p: %x r %d c %d\n", + desc, rsp.val, + cb_idx, cbs_in_tb); + + op->status |= ((rsp.input_err) + ? (1 << RTE_BBDEV_DATA_ERROR) : 0); + op->status |= ((rsp.dma_err) ? (1 << RTE_BBDEV_DRV_ERROR) : 0); + op->status |= ((rsp.fcw_err) ? (1 << RTE_BBDEV_DRV_ERROR) : 0); + + /* CRC invalid if error exists */ + if (!op->status) + op->status |= rsp.crc_status << RTE_BBDEV_CRC_ERROR; + op->turbo_dec.iter_count = RTE_MAX((uint8_t) rsp.iter_cnt, + op->turbo_dec.iter_count); + + /* Check if this is the last desc in batch (Atomic Queue) */ + if (desc->req.last_desc_in_batch) { + (*aq_dequeued)++; + desc->req.last_desc_in_batch = 0; + } + desc->rsp.val = ACC101_DMA_DESC_TYPE; + desc->rsp.add_info_0 = 0; + desc->rsp.add_info_1 = 0; + dequeued_cbs++; + cb_idx++; + } + + *ref_op = op; + + return cb_idx; +} + +/* Dequeue LDPC encode operations from ACC101 device. */ +static uint16_t +acc101_dequeue_ldpc_enc(struct rte_bbdev_queue_data *q_data, + struct rte_bbdev_enc_op **ops, uint16_t num) +{ + struct acc101_queue *q = q_data->queue_private; + uint32_t avail = acc101_ring_avail_deq(q); + uint32_t aq_dequeued = 0; + uint16_t i, dequeued_ops = 0, dequeued_descs = 0; + int ret; + struct rte_bbdev_enc_op *op; + if (avail == 0) + return 0; + op = (q->ring_addr + (q->sw_ring_tail & + q->sw_ring_wrap_mask))->req.op_addr; +#ifdef RTE_LIBRTE_BBDEV_DEBUG + if (unlikely(ops == NULL || q == NULL || op == NULL)) + return 0; +#endif + int cbm = op->ldpc_enc.code_block_mode; + + for (i = 0; i < num; i++) { + if (cbm == RTE_BBDEV_TRANSPORT_BLOCK) + ret = dequeue_enc_one_op_tb(q, &ops[dequeued_ops], + &dequeued_ops, &aq_dequeued, + &dequeued_descs); + else + ret = dequeue_enc_one_op_cb(q, &ops[dequeued_ops], + &dequeued_ops, &aq_dequeued, + &dequeued_descs); + if (ret < 0) + break; + if (dequeued_ops >= num) + break; + } + + q->aq_dequeued += aq_dequeued; + q->sw_ring_tail += dequeued_descs; + + /* Update enqueue stats */ + q_data->queue_stats.dequeued_count += dequeued_ops; + return dequeued_ops; +} + +/* Dequeue decode operations from ACC101 device. */ +static uint16_t +acc101_dequeue_ldpc_dec(struct rte_bbdev_queue_data *q_data, + struct rte_bbdev_dec_op **ops, uint16_t num) +{ + struct acc101_queue *q = q_data->queue_private; + uint16_t dequeue_num; + uint32_t avail = acc101_ring_avail_deq(q); + uint32_t aq_dequeued = 0; + uint16_t i; + uint16_t dequeued_cbs = 0; + struct rte_bbdev_dec_op *op; + int ret; + +#ifdef RTE_LIBRTE_BBDEV_DEBUG + if (unlikely(ops == 0 && q == NULL)) + return 0; +#endif + + dequeue_num = RTE_MIN(avail, num); + + for (i = 0; i < dequeue_num; ++i) { + op = (q->ring_addr + ((q->sw_ring_tail + dequeued_cbs) + & q->sw_ring_wrap_mask))->req.op_addr; + if (op->ldpc_dec.code_block_mode == RTE_BBDEV_TRANSPORT_BLOCK) + ret = dequeue_dec_one_op_tb(q, &ops[i], dequeued_cbs, + &aq_dequeued); + else + ret = dequeue_ldpc_dec_one_op_cb( + q_data, q, &ops[i], dequeued_cbs, + &aq_dequeued); + + if (ret < 0) + break; + dequeued_cbs += ret; + } + + q->aq_dequeued += aq_dequeued; + q->sw_ring_tail += dequeued_cbs; + + /* Update enqueue stats */ + q_data->queue_stats.dequeued_count += i; + return i; +} + /* Initialization Function */ static void acc101_bbdev_init(struct rte_bbdev *dev, struct rte_pci_driver *drv) @@ -831,6 +2564,10 @@ struct rte_pci_device *pci_dev = RTE_DEV_TO_PCI(dev->device); dev->dev_ops = &acc101_bbdev_ops; + dev->enqueue_ldpc_enc_ops = acc101_enqueue_ldpc_enc; + dev->enqueue_ldpc_dec_ops = acc101_enqueue_ldpc_dec; + dev->dequeue_ldpc_enc_ops = acc101_dequeue_ldpc_enc; + dev->dequeue_ldpc_dec_ops = acc101_dequeue_ldpc_dec; ((struct acc101_device *) dev->data->dev_private)->pf_device = !strcmp(drv->driver.name, diff --git a/drivers/baseband/acc101/rte_acc101_pmd.h b/drivers/baseband/acc101/rte_acc101_pmd.h index 65cab8a..1cbbb1f 100644 --- a/drivers/baseband/acc101/rte_acc101_pmd.h +++ b/drivers/baseband/acc101/rte_acc101_pmd.h @@ -62,6 +62,10 @@ * 128M x 32kB = 4GB addressable memory */ #define ACC101_HARQ_LAYOUT (128 * 1024 * 1024) +/* Assume offset for HARQ in memory */ +#define ACC101_HARQ_OFFSET (32 * 1024) +#define ACC101_HARQ_OFFSET_SHIFT 15 +#define ACC101_HARQ_OFFSET_MASK 0x7ffffff /* Mask used to calculate an index in an Info Ring array (not a byte offset) */ #define ACC101_INFO_RING_MASK (ACC101_INFO_RING_NUM_ENTRIES-1) /* Number of Virtual Functions ACC101 supports */ @@ -76,6 +80,8 @@ #define ACC101_GRP_ID_SHIFT 10 /* Queue Index Hierarchy */ #define ACC101_VF_ID_SHIFT 4 /* Queue Index Hierarchy */ #define ACC101_QUEUE_ENABLE 0x80000000 /* Bit to mark Queue as Enabled */ +#define ACC101_FDONE 0x80000000 +#define ACC101_SDONE 0x40000000 #define ACC101_NUM_ACCS 5 #define ACC101_ACCMAP_0 0 @@ -105,6 +111,26 @@ #define ACC101_COMPANION_PTRS 8 #define ACC101_FCW_VER 2 +#define ACC101_MUX_5GDL_DESC 6 +#define ACC101_CMP_ENC_SIZE 20 +#define ACC101_CMP_DEC_SIZE 24 +#define ACC101_ENC_OFFSET (32) +#define ACC101_DEC_OFFSET (80) +#define ACC101_EXT_MEM /* Default option with memory external to CPU */ +#define ACC101_HARQ_OFFSET_THRESHOLD 1024 +#define ACC101_LIMIT_DL_MUX_BITS 534 + +/* Constants from K0 computation from 3GPP 38.212 Table 5.4.2.1-2 */ +#define ACC101_N_ZC_1 66 /* N = 66 Zc for BG 1 */ +#define ACC101_N_ZC_2 50 /* N = 50 Zc for BG 2 */ +#define ACC101_K_ZC_1 22 /* K = 22 Zc for BG 1 */ +#define ACC101_K_ZC_2 10 /* K = 10 Zc for BG 2 */ +#define ACC101_K0_1_1 17 /* K0 fraction numerator for rv 1 and BG 1 */ +#define ACC101_K0_1_2 13 /* K0 fraction numerator for rv 1 and BG 2 */ +#define ACC101_K0_2_1 33 /* K0 fraction numerator for rv 2 and BG 1 */ +#define ACC101_K0_2_2 25 /* K0 fraction numerator for rv 2 and BG 2 */ +#define ACC101_K0_3_1 56 /* K0 fraction numerator for rv 3 and BG 1 */ +#define ACC101_K0_3_2 43 /* K0 fraction numerator for rv 3 and BG 2 */ #define ACC101_LONG_WAIT 1000 -- 1.8.3.1