From 737edea328d701e33cde6ad9601e546ded6dd08f Mon Sep 17 00:00:00 2001 From: Ting Xu Date: Thu, 15 Dec 2022 02:10:59 +0000 Subject: idpf: add native idpf driver plugin Add a new native idpf driver. This patch enables the device initialization. Add some necessary functions and definations for input and output. A new version of virtchnl is introduced. Type: feature Signed-off-by: Ting Xu Change-Id: Ibbd9cd645e64469f1c4c8b33346c1301be3f6927 --- src/plugins/idpf/device.c | 2265 +++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 2265 insertions(+) create mode 100644 src/plugins/idpf/device.c (limited to 'src/plugins/idpf/device.c') diff --git a/src/plugins/idpf/device.c b/src/plugins/idpf/device.c new file mode 100644 index 00000000000..44b8116d996 --- /dev/null +++ b/src/plugins/idpf/device.c @@ -0,0 +1,2265 @@ +/* + *------------------------------------------------------------------ + * Copyright (c) 2023 Intel and/or its affiliates. + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at: + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + *------------------------------------------------------------------ + */ + +#include +#include +#include + +#define IDPF_RXQ_SZ 512 +#define IDPF_TXQ_SZ 512 + +#define PCI_VENDOR_ID_INTEL 0x8086 +#define PCI_DEVICE_ID_INTEL_IDPF_PF 0x1452 +#define PCI_DEVICE_ID_INTEL_IDPF_VF 0x1889 + +VLIB_REGISTER_LOG_CLASS (idpf_log) = { + .class_name = "idpf", +}; + +VLIB_REGISTER_LOG_CLASS (idpf_stats_log) = { + .class_name = "idpf", + .subclass_name = "stats", +}; + +idpf_main_t idpf_main; +void idpf_delete_if (vlib_main_t *vm, idpf_device_t *id, int with_barrier); + +static pci_device_id_t idpf_pci_device_ids[] = { + { .vendor_id = PCI_VENDOR_ID_INTEL, + .device_id = PCI_DEVICE_ID_INTEL_IDPF_PF }, + { .vendor_id = PCI_VENDOR_ID_INTEL, + .device_id = PCI_DEVICE_ID_INTEL_IDPF_VF }, + { 0 }, +}; + +static int +idpf_vc_clean (vlib_main_t *vm, idpf_device_t *id) +{ + idpf_ctlq_msg_t *q_msg[IDPF_CTLQ_LEN]; + uint16_t num_q_msg = IDPF_CTLQ_LEN; + idpf_dma_mem_t *dma_mem; + uint32_t i; + int err; + + for (i = 0; i < 10; i++) + { + err = idpf_ctlq_clean_sq (id->asq, &num_q_msg, q_msg); + vlib_process_suspend (vm, 0.02); + if (num_q_msg > 0) + break; + } + if (err != 0) + return err; + + /* Empty queue is not an error */ + for (i = 0; i < num_q_msg; i++) + { + dma_mem = q_msg[i]->ctx.indirect.payload; + if (dma_mem != NULL) + idpf_free_dma_mem (id, dma_mem); + clib_mem_free (q_msg[i]); + } + + return 0; +} + +static idpf_vc_result_t +idpf_read_msg_from_cp (idpf_device_t *id, u16 buf_len, u8 *buf) +{ + idpf_ctlq_msg_t ctlq_msg; + idpf_dma_mem_t *dma_mem = NULL; + idpf_vc_result_t result = IDPF_MSG_NON; + u32 opcode; + u16 pending = 1; + int ret; + + ret = idpf_ctlq_recv (id->arq, &pending, &ctlq_msg); + if (ret != 0) + { + idpf_log_debug (id, "Can't read msg from AQ"); + if (ret != -ENOMSG) + result = IDPF_MSG_ERR; + return result; + } + + clib_memcpy_fast (buf, ctlq_msg.ctx.indirect.payload->va, buf_len); + + opcode = ctlq_msg.cookie.mbx.chnl_opcode; + id->cmd_retval = ctlq_msg.cookie.mbx.chnl_retval; + + idpf_log_debug (id, "CQ from CP carries opcode %u, retval %d", opcode, + id->cmd_retval); + + if (opcode == VIRTCHNL2_OP_EVENT) + { + virtchnl2_event_t *ve = + (virtchnl2_event_t *) ctlq_msg.ctx.indirect.payload->va; + + result = IDPF_MSG_SYS; + switch (ve->event) + { + case VIRTCHNL2_EVENT_LINK_CHANGE: + break; + default: + idpf_log_err (id, "%s: Unknown event %d from CP", __func__, + ve->event); + break; + } + } + else + { + /* async reply msg on command issued by pf previously */ + result = IDPF_MSG_CMD; + if (opcode != id->pend_cmd) + { + idpf_log_warn (id, "command mismatch, expect %u, get %u", + id->pend_cmd, opcode); + result = IDPF_MSG_ERR; + } + } + + if (ctlq_msg.data_len != 0) + dma_mem = ctlq_msg.ctx.indirect.payload; + else + pending = 0; + + ret = idpf_ctlq_post_rx_buffs (id, id->arq, &pending, &dma_mem); + if (ret != 0 && dma_mem != NULL) + idpf_free_dma_mem (id, dma_mem); + + return result; +} + +clib_error_t * +idpf_send_vc_msg (vlib_main_t *vm, idpf_device_t *id, virtchnl2_op_t op, + u8 *in, u16 in_len) +{ + idpf_ctlq_msg_t *ctlq_msg; + idpf_dma_mem_t *dma_mem; + int error = 0; + + error = idpf_vc_clean (vm, id); + if (error) + goto err; + + ctlq_msg = clib_mem_alloc (sizeof (idpf_ctlq_msg_t)); + if (ctlq_msg == NULL) + goto err; + clib_memset (ctlq_msg, 0, sizeof (idpf_ctlq_msg_t)); + + dma_mem = clib_mem_alloc (sizeof (idpf_dma_mem_t)); + if (dma_mem == NULL) + goto dma_mem_error; + clib_memset (dma_mem, 0, sizeof (idpf_dma_mem_t)); + + dma_mem->va = idpf_alloc_dma_mem (vm, id, dma_mem, IDPF_DFLT_MBX_BUF_SIZE); + if (dma_mem->va == NULL) + { + clib_mem_free (dma_mem); + goto err; + } + + clib_memcpy (dma_mem->va, in, in_len); + + ctlq_msg->opcode = idpf_mbq_opc_send_msg_to_pf; + ctlq_msg->func_id = 0; + ctlq_msg->data_len = in_len; + ctlq_msg->cookie.mbx.chnl_opcode = op; + ctlq_msg->cookie.mbx.chnl_retval = VIRTCHNL2_STATUS_SUCCESS; + ctlq_msg->ctx.indirect.payload = dma_mem; + + error = idpf_ctlq_send (id, id->asq, 1, ctlq_msg); + if (error) + goto send_error; + + return 0; + +send_error: + idpf_free_dma_mem (id, dma_mem); +dma_mem_error: + clib_mem_free (ctlq_msg); +err: + return clib_error_return (0, "idpf send vc msg to PF failed"); +} + +clib_error_t * +idpf_read_one_msg (vlib_main_t *vm, idpf_device_t *id, u32 ops, u8 *buf, + u16 buf_len) +{ + int i = 0, ret; + f64 suspend_time = IDPF_SEND_TO_PF_SUSPEND_TIME; + + do + { + ret = idpf_read_msg_from_cp (id, buf_len, buf); + if (ret == IDPF_MSG_CMD) + break; + vlib_process_suspend (vm, suspend_time); + } + while (i++ < IDPF_SEND_TO_PF_MAX_TRY_TIMES); + if (i >= IDPF_SEND_TO_PF_MAX_TRY_TIMES || + id->cmd_retval != VIRTCHNL2_STATUS_SUCCESS) + return clib_error_return (0, "idpf read one msg failed"); + + return 0; +} + +clib_error_t * +idpf_execute_vc_cmd (vlib_main_t *vm, idpf_device_t *id, idpf_cmd_info_t *args) +{ + clib_error_t *error = 0; + f64 suspend_time = IDPF_SEND_TO_PF_SUSPEND_TIME; + int i = 0; + + if (id->pend_cmd == VIRTCHNL2_OP_UNKNOWN) + id->pend_cmd = args->ops; + else + return clib_error_return (0, "There is incomplete cmd %d", id->pend_cmd); + + if ((error = idpf_send_vc_msg (vm, id, args->ops, args->in_args, + args->in_args_size))) + return error; + + switch (args->ops) + { + case VIRTCHNL2_OP_VERSION: + case VIRTCHNL2_OP_GET_CAPS: + case VIRTCHNL2_OP_CREATE_VPORT: + case VIRTCHNL2_OP_DESTROY_VPORT: + case VIRTCHNL2_OP_SET_RSS_KEY: + case VIRTCHNL2_OP_SET_RSS_LUT: + case VIRTCHNL2_OP_SET_RSS_HASH: + case VIRTCHNL2_OP_CONFIG_RX_QUEUES: + case VIRTCHNL2_OP_CONFIG_TX_QUEUES: + case VIRTCHNL2_OP_ENABLE_QUEUES: + case VIRTCHNL2_OP_DISABLE_QUEUES: + case VIRTCHNL2_OP_ENABLE_VPORT: + case VIRTCHNL2_OP_DISABLE_VPORT: + case VIRTCHNL2_OP_MAP_QUEUE_VECTOR: + case VIRTCHNL2_OP_UNMAP_QUEUE_VECTOR: + case VIRTCHNL2_OP_ALLOC_VECTORS: + case VIRTCHNL2_OP_DEALLOC_VECTORS: + case VIRTCHNL2_OP_GET_STATS: + /* for init virtchnl ops, need to poll the response */ + error = idpf_read_one_msg (vm, id, args->ops, args->out_buffer, + args->out_size); + if (error) + return clib_error_return (0, "idpf read vc message from PF failed"); + clear_cmd (id); + break; + case VIRTCHNL2_OP_GET_PTYPE_INFO: + break; + default: + do + { + if (id->pend_cmd == VIRTCHNL2_OP_UNKNOWN) + break; + vlib_process_suspend (vm, suspend_time); + /* If don't read msg or read sys event, continue */ + } + while (i++ < IDPF_SEND_TO_PF_MAX_TRY_TIMES); + /* If there's no response is received, clear command */ + if (i >= IDPF_SEND_TO_PF_MAX_TRY_TIMES || + id->cmd_retval != VIRTCHNL2_STATUS_SUCCESS) + return clib_error_return ( + 0, "No response or return failure (%d) for cmd %d", id->cmd_retval, + args->ops); + break; + } + + return error; +} + +static inline uword +idpf_dma_addr (vlib_main_t *vm, idpf_device_t *id, void *p) +{ + return (id->flags & IDPF_DEVICE_F_VA_DMA) ? pointer_to_uword (p) : + vlib_physmem_get_pa (vm, p); +} + +clib_error_t * +idpf_vc_config_irq_map_unmap (vlib_main_t *vm, idpf_device_t *id, + idpf_vport_t *vport, bool map) +{ + virtchnl2_queue_vector_maps_t *map_info; + virtchnl2_queue_vector_t *vecmap; + u16 nb_rxq = vport->id->n_rx_queues; + idpf_cmd_info_t args; + clib_error_t *error; + int len, i; + + len = sizeof (virtchnl2_queue_vector_maps_t) + + (nb_rxq - 1) * sizeof (virtchnl2_queue_vector_t); + + map_info = clib_mem_alloc_aligned (len, CLIB_CACHE_LINE_BYTES); + clib_memset (map_info, 0, len); + + map_info->vport_id = vport->vport_id; + map_info->num_qv_maps = nb_rxq; + for (i = 0; i < nb_rxq; i++) + { + vecmap = &map_info->qv_maps[i]; + vecmap->queue_id = vport->qv_map[i].queue_id; + vecmap->vector_id = vport->qv_map[i].vector_id; + vecmap->itr_idx = VIRTCHNL2_ITR_IDX_0; + vecmap->queue_type = VIRTCHNL2_QUEUE_TYPE_RX; + } + + args.ops = + map ? VIRTCHNL2_OP_MAP_QUEUE_VECTOR : VIRTCHNL2_OP_UNMAP_QUEUE_VECTOR; + args.in_args = (u8 *) map_info; + args.in_args_size = len; + args.out_buffer = id->mbx_resp; + args.out_size = IDPF_DFLT_MBX_BUF_SIZE; + error = idpf_execute_vc_cmd (vm, id, &args); + if (error != 0) + return clib_error_return ( + 0, "Failed to execute command of VIRTCHNL2_OP_%s_QUEUE_VECTOR", + map ? "MAP" : "UNMAP"); + + clib_mem_free (map_info); + return error; +} + +clib_error_t * +idpf_config_rx_queues_irqs (vlib_main_t *vm, idpf_device_t *id, + idpf_vport_t *vport) +{ + virtchnl2_queue_vector_t *qv_map; + clib_error_t *error = 0; + u32 dynctl_reg_start; + u32 itrn_reg_start; + u32 dynctl_val, itrn_val; + int i; + + qv_map = clib_mem_alloc_aligned (id->n_rx_queues * + sizeof (virtchnl2_queue_vector_t), + CLIB_CACHE_LINE_BYTES); + clib_memset (qv_map, 0, id->n_rx_queues * sizeof (virtchnl2_queue_vector_t)); + + dynctl_reg_start = vport->recv_vectors->vchunks.vchunks->dynctl_reg_start; + itrn_reg_start = vport->recv_vectors->vchunks.vchunks->itrn_reg_start; + dynctl_val = idpf_reg_read (id, dynctl_reg_start); + idpf_log_debug (id, "Value of dynctl_reg_start is 0x%x", dynctl_val); + itrn_val = idpf_reg_read (id, itrn_reg_start); + idpf_log_debug (id, "Value of itrn_reg_start is 0x%x", itrn_val); + + if (itrn_val != 0) + idpf_reg_write (id, dynctl_reg_start, + VIRTCHNL2_ITR_IDX_0 << PF_GLINT_DYN_CTL_ITR_INDX_S | + PF_GLINT_DYN_CTL_WB_ON_ITR_M | + itrn_val << PF_GLINT_DYN_CTL_INTERVAL_S); + else + idpf_reg_write (id, dynctl_reg_start, + VIRTCHNL2_ITR_IDX_0 << PF_GLINT_DYN_CTL_ITR_INDX_S | + PF_GLINT_DYN_CTL_WB_ON_ITR_M | + IDPF_DFLT_INTERVAL << PF_GLINT_DYN_CTL_INTERVAL_S); + + for (i = 0; i < id->n_rx_queues; i++) + { + /* map all queues to the same vector */ + qv_map[i].queue_id = vport->chunks_info.rx_start_qid + i; + qv_map[i].vector_id = + vport->recv_vectors->vchunks.vchunks->start_vector_id; + } + vport->qv_map = qv_map; + + if ((error = idpf_vc_config_irq_map_unmap (vm, id, vport, true))) + { + idpf_log_err (id, "config interrupt mapping failed"); + goto config_irq_map_err; + } + + return error; + +config_irq_map_err: + clib_mem_free (vport->qv_map); + vport->qv_map = NULL; + + return error; +} + +clib_error_t * +idpf_rx_split_bufq_setup (vlib_main_t *vm, idpf_device_t *id, + idpf_vport_t *vport, idpf_rxq_t *bufq, u16 qid, + u16 rxq_size) +{ + clib_error_t *err; + u32 n_alloc, i; + + bufq->size = rxq_size; + bufq->next = 0; + bufq->descs = vlib_physmem_alloc_aligned_on_numa ( + vm, bufq->size * sizeof (virtchnl2_rx_desc_t), 2 * CLIB_CACHE_LINE_BYTES, + id->numa_node); + + bufq->buffer_pool_index = + vlib_buffer_pool_get_default_for_numa (vm, id->numa_node); + + if ((err = vlib_pci_map_dma (vm, id->pci_dev_handle, (void *) bufq->descs))) + return err; + + clib_memset ((void *) bufq->descs, 0, + bufq->size * sizeof (virtchnl2_rx_desc_t)); + vec_validate_aligned (bufq->bufs, bufq->size, CLIB_CACHE_LINE_BYTES); + bufq->qrx_tail = id->bar0 + (vport->chunks_info.rx_buf_qtail_start + + qid * vport->chunks_info.rx_buf_qtail_spacing); + + n_alloc = vlib_buffer_alloc_from_pool (vm, bufq->bufs, bufq->size - 8, + bufq->buffer_pool_index); + if (n_alloc == 0) + return clib_error_return (0, "buffer allocation error"); + + bufq->n_enqueued = n_alloc; + virtchnl2_rx_desc_t *d = bufq->descs; + for (i = 0; i < n_alloc; i++) + { + vlib_buffer_t *b = vlib_get_buffer (vm, bufq->bufs[i]); + if (id->flags & IDPF_DEVICE_F_VA_DMA) + d->qword[0] = vlib_buffer_get_va (b); + else + d->qword[0] = vlib_buffer_get_pa (vm, b); + d++; + } + + return 0; +} + +clib_error_t * +idpf_split_rxq_init (vlib_main_t *vm, idpf_device_t *id, idpf_vport_t *vport, + u16 qid, u16 rxq_size) +{ + clib_error_t *err; + idpf_rxq_t *rxq; + u32 n_alloc, i; + + vec_validate_aligned (vport->rxqs, qid, CLIB_CACHE_LINE_BYTES); + rxq = vec_elt_at_index (vport->rxqs, qid); + rxq->size = rxq_size; + rxq->next = 0; + rxq->descs = vlib_physmem_alloc_aligned_on_numa ( + vm, rxq->size * sizeof (virtchnl2_rx_desc_t), 2 * CLIB_CACHE_LINE_BYTES, + id->numa_node); + + rxq->buffer_pool_index = + vlib_buffer_pool_get_default_for_numa (vm, id->numa_node); + + if (rxq->descs == 0) + return vlib_physmem_last_error (vm); + + if ((err = vlib_pci_map_dma (vm, id->pci_dev_handle, (void *) rxq->descs))) + return err; + + clib_memset ((void *) rxq->descs, 0, + rxq->size * sizeof (virtchnl2_rx_desc_t)); + vec_validate_aligned (rxq->bufs, rxq->size, CLIB_CACHE_LINE_BYTES); + rxq->qrx_tail = id->bar0 + (vport->chunks_info.rx_qtail_start + + qid * vport->chunks_info.rx_qtail_spacing); + + n_alloc = vlib_buffer_alloc_from_pool (vm, rxq->bufs, rxq->size - 8, + rxq->buffer_pool_index); + + if (n_alloc == 0) + return clib_error_return (0, "buffer allocation error"); + + rxq->n_enqueued = n_alloc; + virtchnl2_rx_desc_t *d = rxq->descs; + for (i = 0; i < n_alloc; i++) + { + vlib_buffer_t *b = vlib_get_buffer (vm, rxq->bufs[i]); + if (id->flags & IDPF_DEVICE_F_VA_DMA) + d->qword[0] = vlib_buffer_get_va (b); + else + d->qword[0] = vlib_buffer_get_pa (vm, b); + d++; + } + + err = + idpf_rx_split_bufq_setup (vm, id, vport, rxq->bufq1, 2 * qid, rxq_size); + if (err) + return err; + err = + idpf_rx_split_bufq_setup (vm, id, vport, rxq->bufq2, 2 * qid, rxq_size); + if (err) + return err; + + return 0; +} + +clib_error_t * +idpf_single_rxq_init (vlib_main_t *vm, idpf_device_t *id, idpf_vport_t *vport, + u16 qid, u16 rxq_size) +{ + clib_error_t *err; + idpf_rxq_t *rxq; + u32 n_alloc, i; + + vec_validate_aligned (vport->rxqs, qid, CLIB_CACHE_LINE_BYTES); + rxq = vec_elt_at_index (vport->rxqs, qid); + rxq->queue_index = vport->chunks_info.rx_start_qid + qid; + rxq->size = rxq_size; + rxq->next = 0; + rxq->descs = vlib_physmem_alloc_aligned_on_numa ( + vm, rxq->size * sizeof (virtchnl2_rx_desc_t), 2 * CLIB_CACHE_LINE_BYTES, + id->numa_node); + + rxq->buffer_pool_index = + vlib_buffer_pool_get_default_for_numa (vm, id->numa_node); + + if (rxq->descs == 0) + return vlib_physmem_last_error (vm); + + err = vlib_pci_map_dma (vm, id->pci_dev_handle, (void *) rxq->descs); + if (err) + return err; + + clib_memset ((void *) rxq->descs, 0, + rxq->size * sizeof (virtchnl2_rx_desc_t)); + vec_validate_aligned (rxq->bufs, rxq->size, CLIB_CACHE_LINE_BYTES); + rxq->qrx_tail = id->bar0 + (vport->chunks_info.rx_qtail_start + + qid * vport->chunks_info.rx_qtail_spacing); + + n_alloc = vlib_buffer_alloc_from_pool (vm, rxq->bufs, rxq->size - 8, + rxq->buffer_pool_index); + + if (n_alloc == 0) + return clib_error_return (0, "buffer allocation error"); + + rxq->n_enqueued = n_alloc; + virtchnl2_rx_desc_t *d = rxq->descs; + for (i = 0; i < n_alloc; i++) + { + vlib_buffer_t *b = vlib_get_buffer (vm, rxq->bufs[i]); + if (id->flags & IDPF_DEVICE_F_VA_DMA) + d->qword[0] = vlib_buffer_get_va (b); + else + d->qword[0] = vlib_buffer_get_pa (vm, b); + d++; + } + + return 0; +} + +clib_error_t * +idpf_rx_queue_setup (vlib_main_t *vm, idpf_device_t *id, idpf_vport_t *vport, + u16 qid, u16 rxq_size) +{ + if (vport->rxq_model == VIRTCHNL2_QUEUE_MODEL_SINGLE) + return idpf_single_rxq_init (vm, id, vport, qid, rxq_size); + else + return idpf_split_rxq_init (vm, id, vport, qid, rxq_size); +} + +clib_error_t * +idpf_tx_split_complq_setup (vlib_main_t *vm, idpf_device_t *id, + idpf_vport_t *vport, idpf_txq_t *complq, u16 qid, + u16 txq_size) +{ + clib_error_t *err; + u16 n; + u8 bpi = vlib_buffer_pool_get_default_for_numa (vm, id->numa_node); + + complq->size = txq_size; + complq->next = 0; + clib_spinlock_init (&complq->lock); + + n = (complq->size / 510) + 1; + vec_validate_aligned (complq->ph_bufs, n, CLIB_CACHE_LINE_BYTES); + + if (!vlib_buffer_alloc_from_pool (vm, complq->ph_bufs, n, bpi)) + return clib_error_return (0, "buffer allocation error"); + + complq->descs = vlib_physmem_alloc_aligned_on_numa ( + vm, complq->size * sizeof (idpf_tx_desc_t), 2 * CLIB_CACHE_LINE_BYTES, + id->numa_node); + if (complq->descs == 0) + return vlib_physmem_last_error (vm); + + if ((err = + vlib_pci_map_dma (vm, id->pci_dev_handle, (void *) complq->descs))) + return err; + + vec_validate_aligned (complq->bufs, complq->size, CLIB_CACHE_LINE_BYTES); + complq->qtx_tail = + id->bar0 + (vport->chunks_info.tx_compl_qtail_start + + qid * vport->chunks_info.tx_compl_qtail_spacing); + + /* initialize ring of pending RS slots */ + clib_ring_new_aligned (complq->rs_slots, 32, CLIB_CACHE_LINE_BYTES); + + vec_validate_aligned (complq->tmp_descs, complq->size, + CLIB_CACHE_LINE_BYTES); + vec_validate_aligned (complq->tmp_bufs, complq->size, CLIB_CACHE_LINE_BYTES); + + return 0; +} + +clib_error_t * +idpf_split_txq_init (vlib_main_t *vm, idpf_device_t *id, idpf_vport_t *vport, + u16 qid, u16 txq_size) +{ + clib_error_t *err; + idpf_txq_t *txq; + u16 n, complq_qid; + u8 bpi = vlib_buffer_pool_get_default_for_numa (vm, id->numa_node); + + vec_validate_aligned (vport->txqs, qid, CLIB_CACHE_LINE_BYTES); + txq = vec_elt_at_index (vport->txqs, qid); + txq->size = txq_size; + txq->next = 0; + clib_spinlock_init (&txq->lock); + + n = (txq->size / 510) + 1; + vec_validate_aligned (txq->ph_bufs, n, CLIB_CACHE_LINE_BYTES); + + if (!vlib_buffer_alloc_from_pool (vm, txq->ph_bufs, n, bpi)) + return clib_error_return (0, "buffer allocation error"); + + txq->descs = vlib_physmem_alloc_aligned_on_numa ( + vm, txq->size * sizeof (idpf_tx_desc_t), 2 * CLIB_CACHE_LINE_BYTES, + id->numa_node); + if (txq->descs == 0) + return vlib_physmem_last_error (vm); + + err = vlib_pci_map_dma (vm, id->pci_dev_handle, (void *) txq->descs); + if (err) + return err; + + vec_validate_aligned (txq->bufs, txq->size, CLIB_CACHE_LINE_BYTES); + txq->qtx_tail = id->bar0 + (vport->chunks_info.tx_qtail_start + + qid * vport->chunks_info.tx_qtail_spacing); + + /* initialize ring of pending RS slots */ + clib_ring_new_aligned (txq->rs_slots, 32, CLIB_CACHE_LINE_BYTES); + + vec_validate_aligned (txq->tmp_descs, txq->size, CLIB_CACHE_LINE_BYTES); + vec_validate_aligned (txq->tmp_bufs, txq->size, CLIB_CACHE_LINE_BYTES); + + complq_qid = vport->chunks_info.tx_compl_start_qid + qid; + err = idpf_tx_split_complq_setup (vm, id, vport, txq->complq, complq_qid, + 2 * txq_size); + if (err) + return err; + + return 0; +} + +clib_error_t * +idpf_single_txq_init (vlib_main_t *vm, idpf_device_t *id, idpf_vport_t *vport, + u16 qid, u16 txq_size) +{ + clib_error_t *err; + idpf_txq_t *txq; + u16 n; + u8 bpi = vlib_buffer_pool_get_default_for_numa (vm, id->numa_node); + + vec_validate_aligned (vport->txqs, qid, CLIB_CACHE_LINE_BYTES); + txq = vec_elt_at_index (vport->txqs, qid); + txq->queue_index = vport->chunks_info.tx_start_qid + qid; + txq->size = txq_size; + txq->next = 0; + clib_spinlock_init (&txq->lock); + + n = (txq->size / 510) + 1; + vec_validate_aligned (txq->ph_bufs, n, CLIB_CACHE_LINE_BYTES); + + if (!vlib_buffer_alloc_from_pool (vm, txq->ph_bufs, n, bpi)) + return clib_error_return (0, "buffer allocation error"); + + txq->descs = vlib_physmem_alloc_aligned_on_numa ( + vm, txq->size * sizeof (idpf_tx_desc_t), 2 * CLIB_CACHE_LINE_BYTES, + id->numa_node); + if (txq->descs == 0) + return vlib_physmem_last_error (vm); + + err = vlib_pci_map_dma (vm, id->pci_dev_handle, (void *) txq->descs); + if (err) + return err; + + vec_validate_aligned (txq->bufs, txq->size, CLIB_CACHE_LINE_BYTES); + txq->qtx_tail = id->bar0 + (vport->chunks_info.tx_qtail_start + + qid * vport->chunks_info.tx_qtail_spacing); + + /* initialize ring of pending RS slots */ + clib_ring_new_aligned (txq->rs_slots, 32, CLIB_CACHE_LINE_BYTES); + + vec_validate_aligned (txq->tmp_descs, txq->size, CLIB_CACHE_LINE_BYTES); + vec_validate_aligned (txq->tmp_bufs, txq->size, CLIB_CACHE_LINE_BYTES); + + return 0; +} + +clib_error_t * +idpf_tx_queue_setup (vlib_main_t *vm, idpf_device_t *id, idpf_vport_t *vport, + u16 qid, u16 txq_size) +{ + if (vport->txq_model == VIRTCHNL2_QUEUE_MODEL_SINGLE) + return idpf_single_txq_init (vm, id, vport, qid, txq_size); + else + return idpf_split_txq_init (vm, id, vport, qid, txq_size); +} + +clib_error_t * +idpf_vc_config_txq (vlib_main_t *vm, idpf_device_t *id, idpf_vport_t *vport, + u16 qid) +{ + idpf_txq_t *txq; + virtchnl2_config_tx_queues_t *vc_txqs = NULL; + virtchnl2_txq_info_t *txq_info; + idpf_cmd_info_t args; + clib_error_t *error; + u16 num_qs; + int size; + + vec_validate_aligned (vport->txqs, qid, CLIB_CACHE_LINE_BYTES); + txq = vec_elt_at_index (vport->txqs, qid); + + if (vport->txq_model == VIRTCHNL2_QUEUE_MODEL_SINGLE) + num_qs = IDPF_TXQ_PER_GRP; + else + num_qs = IDPF_TXQ_PER_GRP + IDPF_TX_COMPLQ_PER_GRP; + + size = sizeof (*vc_txqs) + (num_qs - 1) * sizeof (virtchnl2_txq_info_t); + vc_txqs = clib_mem_alloc_aligned (size, CLIB_CACHE_LINE_BYTES); + clib_memset (vc_txqs, 0, size); + + vc_txqs->vport_id = vport->vport_id; + vc_txqs->num_qinfo = num_qs; + + if (vport->txq_model == VIRTCHNL2_QUEUE_MODEL_SINGLE) + { + txq_info = &vc_txqs->qinfo[0]; + txq_info->dma_ring_addr = idpf_dma_addr (vm, id, (void *) txq->descs); + txq_info->type = VIRTCHNL2_QUEUE_TYPE_TX; + txq_info->queue_id = txq->queue_index; + txq_info->model = VIRTCHNL2_QUEUE_MODEL_SINGLE; + txq_info->sched_mode = VIRTCHNL2_TXQ_SCHED_MODE_QUEUE; + txq_info->ring_len = txq->size; + } + else + { + /* txq info */ + txq_info = &vc_txqs->qinfo[0]; + txq_info->dma_ring_addr = idpf_dma_addr (vm, id, (void *) txq->descs); + txq_info->type = VIRTCHNL2_QUEUE_TYPE_TX; + txq_info->queue_id = txq->queue_index; + txq_info->model = VIRTCHNL2_QUEUE_MODEL_SPLIT; + txq_info->sched_mode = VIRTCHNL2_TXQ_SCHED_MODE_FLOW; + txq_info->ring_len = txq->size; + txq_info->tx_compl_queue_id = txq->complq->queue_index; + txq_info->relative_queue_id = txq_info->queue_id; + + /* tx completion queue info */ + idpf_txq_t *complq = txq->complq; + txq_info = &vc_txqs->qinfo[1]; + txq_info->dma_ring_addr = idpf_dma_addr (vm, id, (void *) complq->descs); + txq_info->type = VIRTCHNL2_QUEUE_TYPE_TX_COMPLETION; + txq_info->queue_id = complq->queue_index; + txq_info->model = VIRTCHNL2_QUEUE_MODEL_SPLIT; + txq_info->sched_mode = VIRTCHNL2_TXQ_SCHED_MODE_FLOW; + txq_info->ring_len = complq->size; + } + + clib_memset (&args, 0, sizeof (args)); + args.ops = VIRTCHNL2_OP_CONFIG_TX_QUEUES; + args.in_args = (u8 *) vc_txqs; + args.in_args_size = size; + args.out_buffer = id->mbx_resp; + args.out_size = IDPF_DFLT_MBX_BUF_SIZE; + + error = idpf_execute_vc_cmd (vm, id, &args); + clib_mem_free (vc_txqs); + if (error != 0) + return clib_error_return ( + 0, "Failed to execute command VIRTCHNL2_OP_CONFIG_TX_QUEUES"); + + return error; +} + +clib_error_t * +idpf_vc_config_rxq (vlib_main_t *vm, idpf_device_t *id, idpf_vport_t *vport, + u16 qid) +{ + idpf_rxq_t *rxq; + virtchnl2_config_rx_queues_t *vc_rxqs = NULL; + virtchnl2_rxq_info_t *rxq_info; + idpf_cmd_info_t args; + clib_error_t *error; + u16 num_qs; + int size, i; + + vec_validate_aligned (vport->rxqs, qid, CLIB_CACHE_LINE_BYTES); + rxq = vec_elt_at_index (vport->rxqs, qid); + + if (vport->rxq_model == VIRTCHNL2_QUEUE_MODEL_SINGLE) + num_qs = IDPF_RXQ_PER_GRP; + else + num_qs = IDPF_RXQ_PER_GRP + IDPF_RX_BUFQ_PER_GRP; + + size = sizeof (*vc_rxqs) + (num_qs - 1) * sizeof (virtchnl2_rxq_info_t); + vc_rxqs = clib_mem_alloc_aligned (size, CLIB_CACHE_LINE_BYTES); + clib_memset (vc_rxqs, 0, size); + + vc_rxqs->vport_id = vport->vport_id; + vc_rxqs->num_qinfo = num_qs; + + if (vport->rxq_model == VIRTCHNL2_QUEUE_MODEL_SINGLE) + { + rxq_info = &vc_rxqs->qinfo[0]; + rxq_info->dma_ring_addr = idpf_dma_addr (vm, id, (void *) rxq->descs); + rxq_info->type = VIRTCHNL2_QUEUE_TYPE_RX; + rxq_info->queue_id = rxq->queue_index; + rxq_info->model = VIRTCHNL2_QUEUE_MODEL_SINGLE; + rxq_info->data_buffer_size = vlib_buffer_get_default_data_size (vm); + rxq_info->max_pkt_size = ETHERNET_MAX_PACKET_BYTES; + + rxq_info->desc_ids = VIRTCHNL2_RXDID_2_FLEX_SQ_NIC_M; + rxq_info->qflags |= VIRTCHNL2_RX_DESC_SIZE_32BYTE; + + rxq_info->ring_len = rxq->size; + } + else + { + /* Rx queue */ + rxq_info = &vc_rxqs->qinfo[0]; + rxq_info->dma_ring_addr = idpf_dma_addr (vm, id, (void *) rxq->descs); + rxq_info->type = VIRTCHNL2_QUEUE_TYPE_RX; + rxq_info->queue_id = rxq->queue_index; + rxq_info->model = VIRTCHNL2_QUEUE_MODEL_SINGLE; + rxq_info->data_buffer_size = vlib_buffer_get_default_data_size (vm); + rxq_info->max_pkt_size = ETHERNET_MAX_PACKET_BYTES; + + rxq_info->desc_ids = VIRTCHNL2_RXDID_2_FLEX_SPLITQ_M; + rxq_info->qflags |= VIRTCHNL2_RX_DESC_SIZE_32BYTE; + + rxq_info->ring_len = rxq->size; + rxq_info->rx_bufq1_id = rxq->bufq1->queue_index; + rxq_info->rx_bufq2_id = rxq->bufq2->queue_index; + rxq_info->rx_buffer_low_watermark = 64; + + /* Buffer queue */ + for (i = 1; i <= IDPF_RX_BUFQ_PER_GRP; i++) + { + idpf_rxq_t *bufq = (i == 1 ? rxq->bufq1 : rxq->bufq2); + rxq_info = &vc_rxqs->qinfo[i]; + rxq_info->dma_ring_addr = + idpf_dma_addr (vm, id, (void *) bufq->descs); + rxq_info->type = VIRTCHNL2_QUEUE_TYPE_RX_BUFFER; + rxq_info->queue_id = bufq->queue_index; + rxq_info->model = VIRTCHNL2_QUEUE_MODEL_SPLIT; + rxq_info->data_buffer_size = vlib_buffer_get_default_data_size (vm); + rxq_info->desc_ids = VIRTCHNL2_RXDID_2_FLEX_SPLITQ_M; + rxq_info->ring_len = bufq->size; + + rxq_info->buffer_notif_stride = IDPF_RX_BUF_STRIDE; + rxq_info->rx_buffer_low_watermark = 64; + } + } + + clib_memset (&args, 0, sizeof (args)); + args.ops = VIRTCHNL2_OP_CONFIG_RX_QUEUES; + args.in_args = (u8 *) vc_rxqs; + args.in_args_size = size; + args.out_buffer = id->mbx_resp; + args.out_size = IDPF_DFLT_MBX_BUF_SIZE; + + error = idpf_execute_vc_cmd (vm, id, &args); + clib_mem_free (vc_rxqs); + if (error != 0) + return clib_error_return ( + 0, "Failed to execute command VIRTCHNL2_OP_CONFIG_RX_QUEUES"); + + return error; +} + +clib_error_t * +idpf_alloc_vectors (vlib_main_t *vm, idpf_device_t *id, idpf_vport_t *vport, + uint16_t num_vectors) +{ + virtchnl2_alloc_vectors_t *alloc_vec; + idpf_cmd_info_t args; + clib_error_t *error; + int len; + + len = sizeof (virtchnl2_alloc_vectors_t) + + (num_vectors - 1) * sizeof (virtchnl2_vector_chunk_t); + alloc_vec = clib_mem_alloc_aligned (len, CLIB_CACHE_LINE_BYTES); + clib_memset (alloc_vec, 0, len); + + alloc_vec->num_vectors = num_vectors; + + args.ops = VIRTCHNL2_OP_ALLOC_VECTORS; + args.in_args = (u8 *) alloc_vec; + args.in_args_size = sizeof (virtchnl2_alloc_vectors_t); + args.out_buffer = id->mbx_resp; + args.out_size = IDPF_DFLT_MBX_BUF_SIZE; + error = idpf_execute_vc_cmd (vm, id, &args); + if (error != 0) + return clib_error_return ( + 0, "Failed to execute command VIRTCHNL2_OP_ALLOC_VECTORS"); + + if (vport->recv_vectors == NULL) + { + vport->recv_vectors = + clib_mem_alloc_aligned (len, CLIB_CACHE_LINE_BYTES); + clib_memset (vport->recv_vectors, 0, len); + } + + clib_memcpy (vport->recv_vectors, args.out_buffer, len); + clib_mem_free (alloc_vec); + return error; +} + +clib_error_t * +idpf_vc_ena_dis_one_queue (vlib_main_t *vm, idpf_device_t *id, + idpf_vport_t *vport, u16 qid, u32 type, bool on) +{ + virtchnl2_del_ena_dis_queues_t *queue_select; + virtchnl2_queue_chunk_t *queue_chunk; + idpf_cmd_info_t args; + clib_error_t *error = 0; + int len; + + len = sizeof (virtchnl2_del_ena_dis_queues_t); + queue_select = clib_mem_alloc_aligned (len, CLIB_CACHE_LINE_BYTES); + clib_memset (queue_select, 0, len); + + queue_chunk = queue_select->chunks.chunks; + queue_select->chunks.num_chunks = 1; + queue_select->vport_id = vport->vport_id; + + queue_chunk->type = type; + queue_chunk->start_queue_id = qid; + queue_chunk->num_queues = 1; + + args.ops = on ? VIRTCHNL2_OP_ENABLE_QUEUES : VIRTCHNL2_OP_DISABLE_QUEUES; + args.in_args = (u8 *) queue_select; + args.in_args_size = len; + args.out_buffer = id->mbx_resp; + args.out_size = IDPF_DFLT_MBX_BUF_SIZE; + error = idpf_execute_vc_cmd (vm, id, &args); + if (error != 0) + return clib_error_return ( + 0, "Failed to execute command of VIRTCHNL2_OP_%s_QUEUES", + on ? "ENABLE" : "DISABLE"); + + clib_mem_free (queue_select); + return error; +} + +clib_error_t * +idpf_op_enable_queues (vlib_main_t *vm, idpf_device_t *id, idpf_vport_t *vport, + u16 qid, bool rx, bool on) +{ + clib_error_t *error; + u16 queue_index; + u32 type; + + /* switch txq/rxq */ + type = rx ? VIRTCHNL2_QUEUE_TYPE_RX : VIRTCHNL2_QUEUE_TYPE_TX; + + if (type == VIRTCHNL2_QUEUE_TYPE_RX) + { + queue_index = vport->chunks_info.rx_start_qid + qid; + error = idpf_vc_ena_dis_one_queue (vm, id, vport, queue_index, type, on); + } + else + { + queue_index = vport->chunks_info.tx_start_qid + qid; + error = idpf_vc_ena_dis_one_queue (vm, id, vport, queue_index, type, on); + } + if (error != 0) + return error; + + /* switch tx completion queue */ + if (!rx && vport->txq_model == VIRTCHNL2_QUEUE_MODEL_SPLIT) + { + type = VIRTCHNL2_QUEUE_TYPE_TX_COMPLETION; + queue_index = vport->chunks_info.tx_compl_start_qid + qid; + error = idpf_vc_ena_dis_one_queue (vm, id, vport, queue_index, type, on); + if (error != 0) + return error; + } + + /* switch rx buffer queue */ + if (rx && vport->rxq_model == VIRTCHNL2_QUEUE_MODEL_SPLIT) + { + type = VIRTCHNL2_QUEUE_TYPE_RX_BUFFER; + queue_index = vport->chunks_info.rx_buf_start_qid + 2 * qid; + error = idpf_vc_ena_dis_one_queue (vm, id, vport, queue_index, type, on); + if (error != 0) + return error; + queue_index++; + error = idpf_vc_ena_dis_one_queue (vm, id, vport, queue_index, type, on); + if (error != 0) + return error; + } + + return error; +} + +clib_error_t * +idpf_queue_init (vlib_main_t *vm, idpf_device_t *id, idpf_vport_t *vport, + idpf_create_if_args_t *args) +{ + clib_error_t *error = 0; + int i; + + for (i = 0; i < id->n_rx_queues; i++) + { + if ((error = idpf_rx_queue_setup (vm, id, vport, i, args->rxq_size))) + return error; + if ((error = idpf_vc_config_rxq (vm, id, vport, i))) + return error; + if ((error = idpf_op_enable_queues (vm, id, vport, i, true, true))) + return error; + } + + for (i = 0; i < id->n_tx_queues; i++) + { + if ((error = idpf_tx_queue_setup (vm, id, vport, i, args->txq_size))) + return error; + if ((error = idpf_vc_config_txq (vm, id, vport, i))) + return error; + if ((error = idpf_op_enable_queues (vm, id, vport, i, false, true))) + return error; + } + + if ((error = idpf_alloc_vectors (vm, id, vport, IDPF_DFLT_Q_VEC_NUM))) + return error; + + if ((error = idpf_config_rx_queues_irqs (vm, id, vport))) + return error; + + return error; +} + +clib_error_t * +idpf_op_version (vlib_main_t *vm, idpf_device_t *id) +{ + clib_error_t *error = 0; + idpf_cmd_info_t args; + virtchnl2_version_info_t myver = { + .major = VIRTCHNL2_VERSION_MAJOR_2, + .minor = VIRTCHNL2_VERSION_MINOR_0, + }; + virtchnl2_version_info_t ver = { 0 }; + + idpf_log_debug (id, "version: major %u minor %u", myver.major, myver.minor); + + args.ops = VIRTCHNL2_OP_VERSION; + args.in_args = (u8 *) &myver; + args.in_args_size = sizeof (myver); + args.out_buffer = id->mbx_resp; + args.out_size = IDPF_DFLT_MBX_BUF_SIZE; + error = idpf_execute_vc_cmd (vm, id, &args); + if (error != 0) + return clib_error_return (0, + "Failed to execute command VIRTCHNL_OP_VERSION"); + + clib_memcpy (&ver, args.out_buffer, sizeof (ver)); + + if (ver.major != VIRTCHNL2_VERSION_MAJOR_2 || + ver.minor != VIRTCHNL2_VERSION_MINOR_0) + return clib_error_return (0, + "incompatible virtchnl version " + "(remote %d.%d)", + ver.major, ver.minor); + + return 0; +} + +clib_error_t * +idpf_op_get_caps (vlib_main_t *vm, idpf_device_t *id, + virtchnl2_get_capabilities_t *caps) +{ + virtchnl2_get_capabilities_t caps_msg = { 0 }; + idpf_cmd_info_t args; + clib_error_t *error = 0; + + caps_msg.csum_caps = + VIRTCHNL2_CAP_TX_CSUM_L3_IPV4 | VIRTCHNL2_CAP_TX_CSUM_L4_IPV4_TCP | + VIRTCHNL2_CAP_TX_CSUM_L4_IPV4_UDP | VIRTCHNL2_CAP_TX_CSUM_L4_IPV4_SCTP | + VIRTCHNL2_CAP_TX_CSUM_L4_IPV6_TCP | VIRTCHNL2_CAP_TX_CSUM_L4_IPV6_UDP | + VIRTCHNL2_CAP_TX_CSUM_L4_IPV6_SCTP | VIRTCHNL2_CAP_TX_CSUM_GENERIC | + VIRTCHNL2_CAP_RX_CSUM_L3_IPV4 | VIRTCHNL2_CAP_RX_CSUM_L4_IPV4_TCP | + VIRTCHNL2_CAP_RX_CSUM_L4_IPV4_UDP | VIRTCHNL2_CAP_RX_CSUM_L4_IPV4_SCTP | + VIRTCHNL2_CAP_RX_CSUM_L4_IPV6_TCP | VIRTCHNL2_CAP_RX_CSUM_L4_IPV6_UDP | + VIRTCHNL2_CAP_RX_CSUM_L4_IPV6_SCTP | VIRTCHNL2_CAP_RX_CSUM_GENERIC; + + caps_msg.other_caps = VIRTCHNL2_CAP_WB_ON_ITR; + + args.ops = VIRTCHNL2_OP_GET_CAPS; + args.in_args = (u8 *) &caps_msg; + args.in_args_size = sizeof (caps_msg); + args.out_buffer = id->mbx_resp; + args.out_size = IDPF_DFLT_MBX_BUF_SIZE; + error = idpf_execute_vc_cmd (vm, id, &args); + if (error != 0) + return clib_error_return ( + 0, "Failed to execute command VIRTCHNL2_OP_GET_CAPS"); + + clib_memcpy (caps, args.out_buffer, sizeof (*caps)); + return error; +} + +#define CTLQ_NUM 2 +clib_error_t * +idpf_mbx_init (vlib_main_t *vm, idpf_device_t *id) +{ + idpf_ctlq_create_info_t ctlq_info[CTLQ_NUM] = { + { + .type = IDPF_CTLQ_TYPE_MAILBOX_TX, + .id = IDPF_CTLQ_ID, + .len = IDPF_CTLQ_LEN, + .buf_size = IDPF_DFLT_MBX_BUF_SIZE, + .reg = { + .head = PF_FW_ATQH, + .tail = PF_FW_ATQT, + .len = PF_FW_ATQLEN, + .bah = PF_FW_ATQBAH, + .bal = PF_FW_ATQBAL, + .len_mask = PF_FW_ATQLEN_ATQLEN_M, + .len_ena_mask = PF_FW_ATQLEN_ATQENABLE_M, + .head_mask = PF_FW_ATQH_ATQH_M, + } + }, + { + .type = IDPF_CTLQ_TYPE_MAILBOX_RX, + .id = IDPF_CTLQ_ID, + .len = IDPF_CTLQ_LEN, + .buf_size = IDPF_DFLT_MBX_BUF_SIZE, + .reg = { + .head = PF_FW_ARQH, + .tail = PF_FW_ARQT, + .len = PF_FW_ARQLEN, + .bah = PF_FW_ARQBAH, + .bal = PF_FW_ARQBAL, + .len_mask = PF_FW_ARQLEN_ARQLEN_M, + .len_ena_mask = PF_FW_ARQLEN_ARQENABLE_M, + .head_mask = PF_FW_ARQH_ARQH_M, + } + } + }; + struct idpf_ctlq_info *ctlq; + + if (idpf_ctlq_init (vm, id, CTLQ_NUM, ctlq_info)) + return clib_error_return (0, "ctlq init failed"); + + LIST_FOR_EACH_ENTRY_SAFE (ctlq, NULL, &id->cq_list_head, + struct idpf_ctlq_info, cq_list) + { + if (ctlq->q_id == IDPF_CTLQ_ID && + ctlq->cq_type == IDPF_CTLQ_TYPE_MAILBOX_TX) + id->asq = ctlq; + if (ctlq->q_id == IDPF_CTLQ_ID && + ctlq->cq_type == IDPF_CTLQ_TYPE_MAILBOX_RX) + id->arq = ctlq; + } + + if (!id->asq || !id->arq) + { + idpf_ctlq_deinit (id); + return clib_error_return (0, "ctlq deinit"); + } + + return 0; +} + +clib_error_t * +idpf_vc_query_ptype_info (vlib_main_t *vm, idpf_device_t *id) +{ + virtchnl2_get_ptype_info_t ptype_info; + idpf_cmd_info_t args; + clib_error_t *error; + + ptype_info.start_ptype_id = 0; + ptype_info.num_ptypes = IDPF_MAX_PKT_TYPE; + args.ops = VIRTCHNL2_OP_GET_PTYPE_INFO; + args.in_args = (u8 *) &ptype_info; + args.in_args_size = sizeof (virtchnl2_get_ptype_info_t); + args.out_buffer = NULL; + args.out_size = 0; + + error = idpf_execute_vc_cmd (vm, id, &args); + if (error != 0) + return clib_error_return ( + 0, "Failed to execute command VIRTCHNL2_OP_GET_PTYPE_INFO"); + + return error; +} + +clib_error_t * +idpf_get_pkt_type (vlib_main_t *vm, idpf_device_t *id) +{ + virtchnl2_get_ptype_info_t *ptype_info; + u16 ptype_recvd = 0, ptype_offset, i, j; + clib_error_t *error; + + error = idpf_vc_query_ptype_info (vm, id); + if (error != 0) + return clib_error_return (0, "Fail to query packet type information"); + + ptype_info = + clib_mem_alloc_aligned (IDPF_DFLT_MBX_BUF_SIZE, CLIB_CACHE_LINE_BYTES); + + while (ptype_recvd < IDPF_MAX_PKT_TYPE) + { + error = idpf_read_one_msg (vm, id, VIRTCHNL2_OP_GET_PTYPE_INFO, + (u8 *) ptype_info, IDPF_DFLT_MBX_BUF_SIZE); + if (error != 0) + { + error = clib_error_return (0, "Fail to get packet type information"); + goto free_ptype_info; + } + + ptype_recvd += ptype_info->num_ptypes; + ptype_offset = + sizeof (virtchnl2_get_ptype_info_t) - sizeof (virtchnl2_ptype_t); + + for (i = 0; i < ptype_info->num_ptypes; i++) + { + bool is_inner = false, is_ip = false; + virtchnl2_ptype_t *ptype; + u32 proto_hdr = 0; + + ptype = (virtchnl2_ptype_t *) ((u8 *) ptype_info + ptype_offset); + ptype_offset += IDPF_GET_PTYPE_SIZE (ptype); + if (ptype_offset > IDPF_DFLT_MBX_BUF_SIZE) + { + error = + clib_error_return (0, "Ptype offset exceeds mbx buffer size"); + goto free_ptype_info; + } + + if (ptype->ptype_id_10 == 0xFFFF) + goto free_ptype_info; + + for (j = 0; j < ptype->proto_id_count; j++) + { + switch (ptype->proto_id[j]) + { + case VIRTCHNL2_PROTO_HDR_GRE: + case VIRTCHNL2_PROTO_HDR_VXLAN: + proto_hdr &= ~IDPF_PTYPE_L4_MASK; + proto_hdr |= IDPF_PTYPE_TUNNEL_GRENAT; + is_inner = true; + break; + case VIRTCHNL2_PROTO_HDR_MAC: + if (is_inner) + { + proto_hdr &= ~IDPF_PTYPE_INNER_L2_MASK; + proto_hdr |= IDPF_PTYPE_INNER_L2_ETHER; + } + else + { + proto_hdr &= ~IDPF_PTYPE_L2_MASK; + proto_hdr |= IDPF_PTYPE_L2_ETHER; + } + break; + case VIRTCHNL2_PROTO_HDR_VLAN: + if (is_inner) + { + proto_hdr &= ~IDPF_PTYPE_INNER_L2_MASK; + proto_hdr |= IDPF_PTYPE_INNER_L2_ETHER_VLAN; + } + break; + case VIRTCHNL2_PROTO_HDR_PTP: + proto_hdr &= ~IDPF_PTYPE_L2_MASK; + proto_hdr |= IDPF_PTYPE_L2_ETHER_TIMESYNC; + break; + case VIRTCHNL2_PROTO_HDR_LLDP: + proto_hdr &= ~IDPF_PTYPE_L2_MASK; + proto_hdr |= IDPF_PTYPE_L2_ETHER_LLDP; + break; + case VIRTCHNL2_PROTO_HDR_ARP: + proto_hdr &= ~IDPF_PTYPE_L2_MASK; + proto_hdr |= IDPF_PTYPE_L2_ETHER_ARP; + break; + case VIRTCHNL2_PROTO_HDR_PPPOE: + proto_hdr &= ~IDPF_PTYPE_L2_MASK; + proto_hdr |= IDPF_PTYPE_L2_ETHER_PPPOE; + break; + case VIRTCHNL2_PROTO_HDR_IPV4: + if (!is_ip) + { + proto_hdr |= IDPF_PTYPE_L3_IPV4_EXT_UNKNOWN; + is_ip = true; + } + else + { + proto_hdr |= IDPF_PTYPE_INNER_L3_IPV4_EXT_UNKNOWN | + IDPF_PTYPE_TUNNEL_IP; + is_inner = true; + } + break; + case VIRTCHNL2_PROTO_HDR_IPV6: + if (!is_ip) + { + proto_hdr |= IDPF_PTYPE_L3_IPV6_EXT_UNKNOWN; + is_ip = true; + } + else + { + proto_hdr |= IDPF_PTYPE_INNER_L3_IPV6_EXT_UNKNOWN | + IDPF_PTYPE_TUNNEL_IP; + is_inner = true; + } + break; + case VIRTCHNL2_PROTO_HDR_IPV4_FRAG: + case VIRTCHNL2_PROTO_HDR_IPV6_FRAG: + if (is_inner) + proto_hdr |= IDPF_PTYPE_INNER_L4_FRAG; + else + proto_hdr |= IDPF_PTYPE_L4_FRAG; + break; + case VIRTCHNL2_PROTO_HDR_UDP: + if (is_inner) + proto_hdr |= IDPF_PTYPE_INNER_L4_UDP; + else + proto_hdr |= IDPF_PTYPE_L4_UDP; + break; + case VIRTCHNL2_PROTO_HDR_TCP: + if (is_inner) + proto_hdr |= IDPF_PTYPE_INNER_L4_TCP; + else + proto_hdr |= IDPF_PTYPE_L4_TCP; + break; + case VIRTCHNL2_PROTO_HDR_SCTP: + if (is_inner) + proto_hdr |= IDPF_PTYPE_INNER_L4_SCTP; + else + proto_hdr |= IDPF_PTYPE_L4_SCTP; + break; + case VIRTCHNL2_PROTO_HDR_ICMP: + if (is_inner) + proto_hdr |= IDPF_PTYPE_INNER_L4_ICMP; + else + proto_hdr |= IDPF_PTYPE_L4_ICMP; + break; + case VIRTCHNL2_PROTO_HDR_ICMPV6: + if (is_inner) + proto_hdr |= IDPF_PTYPE_INNER_L4_ICMP; + else + proto_hdr |= IDPF_PTYPE_L4_ICMP; + break; + case VIRTCHNL2_PROTO_HDR_L2TPV2: + case VIRTCHNL2_PROTO_HDR_L2TPV2_CONTROL: + case VIRTCHNL2_PROTO_HDR_L2TPV3: + is_inner = true; + proto_hdr |= IDPF_PTYPE_TUNNEL_L2TP; + break; + case VIRTCHNL2_PROTO_HDR_NVGRE: + is_inner = true; + proto_hdr |= IDPF_PTYPE_TUNNEL_NVGRE; + break; + case VIRTCHNL2_PROTO_HDR_GTPC_TEID: + is_inner = true; + proto_hdr |= IDPF_PTYPE_TUNNEL_GTPC; + break; + case VIRTCHNL2_PROTO_HDR_GTPU: + case VIRTCHNL2_PROTO_HDR_GTPU_UL: + case VIRTCHNL2_PROTO_HDR_GTPU_DL: + is_inner = true; + proto_hdr |= IDPF_PTYPE_TUNNEL_GTPU; + break; + case VIRTCHNL2_PROTO_HDR_PAY: + case VIRTCHNL2_PROTO_HDR_IPV6_EH: + case VIRTCHNL2_PROTO_HDR_PRE_MAC: + case VIRTCHNL2_PROTO_HDR_POST_MAC: + case VIRTCHNL2_PROTO_HDR_ETHERTYPE: + case VIRTCHNL2_PROTO_HDR_SVLAN: + case VIRTCHNL2_PROTO_HDR_CVLAN: + case VIRTCHNL2_PROTO_HDR_MPLS: + case VIRTCHNL2_PROTO_HDR_MMPLS: + case VIRTCHNL2_PROTO_HDR_CTRL: + case VIRTCHNL2_PROTO_HDR_ECP: + case VIRTCHNL2_PROTO_HDR_EAPOL: + case VIRTCHNL2_PROTO_HDR_PPPOD: + case VIRTCHNL2_PROTO_HDR_IGMP: + case VIRTCHNL2_PROTO_HDR_AH: + case VIRTCHNL2_PROTO_HDR_ESP: + case VIRTCHNL2_PROTO_HDR_IKE: + case VIRTCHNL2_PROTO_HDR_NATT_KEEP: + case VIRTCHNL2_PROTO_HDR_GTP: + case VIRTCHNL2_PROTO_HDR_GTP_EH: + case VIRTCHNL2_PROTO_HDR_GTPCV2: + case VIRTCHNL2_PROTO_HDR_ECPRI: + case VIRTCHNL2_PROTO_HDR_VRRP: + case VIRTCHNL2_PROTO_HDR_OSPF: + case VIRTCHNL2_PROTO_HDR_TUN: + case VIRTCHNL2_PROTO_HDR_VXLAN_GPE: + case VIRTCHNL2_PROTO_HDR_GENEVE: + case VIRTCHNL2_PROTO_HDR_NSH: + case VIRTCHNL2_PROTO_HDR_QUIC: + case VIRTCHNL2_PROTO_HDR_PFCP: + case VIRTCHNL2_PROTO_HDR_PFCP_NODE: + case VIRTCHNL2_PROTO_HDR_PFCP_SESSION: + case VIRTCHNL2_PROTO_HDR_RTP: + case VIRTCHNL2_PROTO_HDR_NO_PROTO: + default: + continue; + } + id->ptype_tbl[ptype->ptype_id_10] = proto_hdr; + } + } + } + +free_ptype_info: + clib_mem_free (ptype_info); + clear_cmd (id); + return error; +} + +static void +idpf_reset_pf (idpf_device_t *id) +{ + u32 reg; + + reg = idpf_reg_read (id, PFGEN_CTRL); + idpf_reg_write (id, PFGEN_CTRL, (reg | PFGEN_CTRL_PFSWR)); +} + +#define IDPF_RESET_WAIT_CNT 100 +clib_error_t * +idpf_check_pf_reset_done (vlib_main_t *vm, idpf_device_t *id) +{ + u32 reg; + int i; + + for (i = 0; i < IDPF_RESET_WAIT_CNT; i++) + { + reg = idpf_reg_read (id, PFGEN_RSTAT); + if (reg != 0xFFFFFFFF && (reg & PFGEN_RSTAT_PFR_STATE_M)) + return 0; + vlib_process_suspend (vm, 1.0); + } + + return clib_error_return (0, "pf reset time out"); +} + +void +idpf_init_vport_req_info (idpf_device_t *id, + virtchnl2_create_vport_t *vport_info) +{ + vport_info->vport_type = VIRTCHNL2_VPORT_TYPE_DEFAULT; + if (id->txq_model == 1) + { + vport_info->txq_model = VIRTCHNL2_QUEUE_MODEL_SPLIT; + vport_info->num_tx_q = IDPF_DEFAULT_TXQ_NUM; + vport_info->num_tx_complq = + IDPF_DEFAULT_TXQ_NUM * IDPF_TX_COMPLQ_PER_GRP; + } + else + { + vport_info->txq_model = VIRTCHNL2_QUEUE_MODEL_SINGLE; + vport_info->num_tx_q = IDPF_DEFAULT_TXQ_NUM; + vport_info->num_tx_complq = 0; + } + if (id->rxq_model == 1) + { + vport_info->rxq_model = VIRTCHNL2_QUEUE_MODEL_SPLIT; + vport_info->num_rx_q = IDPF_DEFAULT_RXQ_NUM; + vport_info->num_rx_bufq = IDPF_DEFAULT_RXQ_NUM * IDPF_RX_BUFQ_PER_GRP; + } + else + { + vport_info->rxq_model = VIRTCHNL2_QUEUE_MODEL_SINGLE; + vport_info->num_rx_q = IDPF_DEFAULT_RXQ_NUM; + vport_info->num_rx_bufq = 0; + } + + return; +} + +clib_error_t * +idpf_vc_create_vport (vlib_main_t *vm, idpf_device_t *id, idpf_vport_t *vport, + virtchnl2_create_vport_t *vport_req_info) +{ + virtchnl2_create_vport_t vport_msg = { 0 }; + idpf_cmd_info_t args; + clib_error_t *error; + + vport_msg.vport_type = vport_req_info->vport_type; + vport_msg.txq_model = vport_req_info->txq_model; + vport_msg.rxq_model = vport_req_info->rxq_model; + vport_msg.num_tx_q = vport_req_info->num_tx_q; + vport_msg.num_tx_complq = vport_req_info->num_tx_complq; + vport_msg.num_rx_q = vport_req_info->num_rx_q; + vport_msg.num_rx_bufq = vport_req_info->num_rx_bufq; + + clib_memset (&args, 0, sizeof (args)); + args.ops = VIRTCHNL2_OP_CREATE_VPORT; + args.in_args = (u8 *) &vport_msg; + args.in_args_size = sizeof (vport_msg); + args.out_buffer = id->mbx_resp; + args.out_size = IDPF_DFLT_MBX_BUF_SIZE; + error = idpf_execute_vc_cmd (vm, id, &args); + if (error != 0) + return clib_error_return ( + 0, "Failed to execute command of VIRTCHNL2_OP_CREATE_VPORT"); + + clib_memcpy (vport->vport_info, args.out_buffer, IDPF_DFLT_MBX_BUF_SIZE); + return error; +} + +clib_error_t * +idpf_vc_destroy_vport (vlib_main_t *vm, idpf_device_t *id, idpf_vport_t *vport) +{ + virtchnl2_vport_t vc_vport; + idpf_cmd_info_t args; + clib_error_t *error = 0; + + vc_vport.vport_id = vport->vport_id; + + clib_memset (&args, 0, sizeof (args)); + args.ops = VIRTCHNL2_OP_DESTROY_VPORT; + args.in_args = (u8 *) &vc_vport; + args.in_args_size = sizeof (vc_vport); + args.out_buffer = id->mbx_resp; + args.out_size = IDPF_DFLT_MBX_BUF_SIZE; + error = idpf_execute_vc_cmd (vm, id, &args); + if (error != 0) + return clib_error_return ( + 0, "Failed to execute command of VIRTCHNL2_OP_DESTROY_VPORT"); + + return error; +} + +clib_error_t * +idpf_init_vport (idpf_device_t *id, idpf_vport_t *vport) +{ + virtchnl2_create_vport_t *vport_info = vport->vport_info; + int i, type; + + vport->vport_id = vport_info->vport_id; + vport->txq_model = vport_info->txq_model; + vport->rxq_model = vport_info->rxq_model; + vport->num_tx_q = vport_info->num_tx_q; + vport->num_tx_complq = vport_info->num_tx_complq; + vport->num_rx_q = vport_info->num_rx_q; + vport->num_rx_bufq = vport_info->num_rx_bufq; + vport->max_mtu = vport_info->max_mtu; + clib_memcpy (vport->default_mac_addr, vport_info->default_mac_addr, + IDPF_ETH_ALEN); + + for (i = 0; i < vport_info->chunks.num_chunks; i++) + { + type = vport_info->chunks.chunks[i].type; + switch (type) + { + case VIRTCHNL2_QUEUE_TYPE_TX: + vport->chunks_info.tx_start_qid = + vport_info->chunks.chunks[i].start_queue_id; + vport->chunks_info.tx_qtail_start = + vport_info->chunks.chunks[i].qtail_reg_start; + vport->chunks_info.tx_qtail_spacing = + vport_info->chunks.chunks[i].qtail_reg_spacing; + break; + case VIRTCHNL2_QUEUE_TYPE_RX: + vport->chunks_info.rx_start_qid = + vport_info->chunks.chunks[i].start_queue_id; + vport->chunks_info.rx_qtail_start = + vport_info->chunks.chunks[i].qtail_reg_start; + vport->chunks_info.rx_qtail_spacing = + vport_info->chunks.chunks[i].qtail_reg_spacing; + break; + case VIRTCHNL2_QUEUE_TYPE_TX_COMPLETION: + vport->chunks_info.tx_compl_start_qid = + vport_info->chunks.chunks[i].start_queue_id; + vport->chunks_info.tx_compl_qtail_start = + vport_info->chunks.chunks[i].qtail_reg_start; + vport->chunks_info.tx_compl_qtail_spacing = + vport_info->chunks.chunks[i].qtail_reg_spacing; + break; + case VIRTCHNL2_QUEUE_TYPE_RX_BUFFER: + vport->chunks_info.rx_buf_start_qid = + vport_info->chunks.chunks[i].start_queue_id; + vport->chunks_info.rx_buf_qtail_start = + vport_info->chunks.chunks[i].qtail_reg_start; + vport->chunks_info.rx_buf_qtail_spacing = + vport_info->chunks.chunks[i].qtail_reg_spacing; + break; + default: + return clib_error_return (0, "Unsupported queue type"); + } + } + + return 0; +} + +clib_error_t * +idpf_ena_dis_vport (vlib_main_t *vm, idpf_device_t *id, idpf_vport_t *vport, + bool enable) +{ + virtchnl2_vport_t vc_vport; + idpf_cmd_info_t args; + clib_error_t *error; + + vc_vport.vport_id = vport->vport_id; + args.ops = enable ? VIRTCHNL2_OP_ENABLE_VPORT : VIRTCHNL2_OP_DISABLE_VPORT; + args.in_args = (u8 *) &vc_vport; + args.in_args_size = sizeof (vc_vport); + args.out_buffer = id->mbx_resp; + args.out_size = IDPF_DFLT_MBX_BUF_SIZE; + + error = idpf_execute_vc_cmd (vm, id, &args); + if (error != 0) + { + return clib_error_return ( + 0, "Failed to execute command of VIRTCHNL2_OP_%s_VPORT", + enable ? "ENABLE" : "DISABLE"); + } + + return error; +} + +clib_error_t * +idpf_dealloc_vectors (vlib_main_t *vm, idpf_device_t *id, idpf_vport_t *vport) +{ + virtchnl2_alloc_vectors_t *alloc_vec; + virtchnl2_vector_chunks_t *vcs; + idpf_cmd_info_t args; + clib_error_t *error; + int len; + + alloc_vec = vport->recv_vectors; + vcs = &alloc_vec->vchunks; + + len = sizeof (virtchnl2_vector_chunks_t) + + (vcs->num_vchunks - 1) * sizeof (virtchnl2_vector_chunk_t); + + args.ops = VIRTCHNL2_OP_DEALLOC_VECTORS; + args.in_args = (u8 *) vcs; + args.in_args_size = len; + args.out_buffer = id->mbx_resp; + args.out_size = IDPF_DFLT_MBX_BUF_SIZE; + error = idpf_execute_vc_cmd (vm, id, &args); + if (error != 0) + return clib_error_return ( + 0, "Failed to execute command VIRTCHNL2_OP_DEALLOC_VECTORS"); + + return error; +} + +clib_error_t * +idpf_dev_vport_init (vlib_main_t *vm, idpf_device_t *id, + idpf_vport_param_t *param) +{ + idpf_vport_t *vport; + virtchnl2_create_vport_t vport_req_info = { 0 }; + clib_error_t *error = 0; + + vport = clib_mem_alloc (sizeof (idpf_vport_t)); + clib_memset (vport, 0, sizeof (idpf_vport_t)); + + vport->vport_info = clib_mem_alloc (IDPF_DFLT_MBX_BUF_SIZE); + clib_memset (vport->vport_info, 0, IDPF_DFLT_MBX_BUF_SIZE); + + id->vports[param->idx] = vport; + vport->id = id; + vport->idx = param->idx; + + idpf_init_vport_req_info (id, &vport_req_info); + + error = idpf_vc_create_vport (vm, id, vport, &vport_req_info); + if (error != 0) + { + idpf_log_err (id, "Failed to create vport."); + goto err_create_vport; + } + + error = idpf_init_vport (id, vport); + if (error != 0) + { + idpf_log_err (id, "Failed to init vports."); + goto err_init_vport; + } + + id->vports[param->idx] = vport; + + clib_memcpy (id->hwaddr, vport->default_mac_addr, IDPF_ETH_ALEN); + + return error; + +err_init_vport: + id->vports[param->idx] = NULL; /* reset */ + idpf_vc_destroy_vport (vm, id, vport); +err_create_vport: + clib_mem_free (vport->vport_info); + clib_mem_free (vport); + return error; +} + +/* dev configure */ +clib_error_t * +idpf_device_init (vlib_main_t *vm, idpf_main_t *im, idpf_device_t *id, + idpf_create_if_args_t *args) +{ + idpf_vport_t *vport; + idpf_vport_param_t vport_param = { 0 }; + virtchnl2_get_capabilities_t caps = { 0 }; + clib_error_t *error; + u16 rxq_num, txq_num; + int i; + + idpf_reset_pf (id); + error = idpf_check_pf_reset_done (vm, id); + if (error) + return error; + + /* + * Init mailbox configuration + */ + if ((error = idpf_mbx_init (vm, id))) + return error; + + /* + * Check API version + */ + error = idpf_op_version (vm, id); + if (error) + return error; + + /* + * Get pkt type table + */ + error = idpf_get_pkt_type (vm, id); + if (error) + return error; + + /* Get idpf capability */ + error = idpf_op_get_caps (vm, id, &caps); + if (error) + return error; + + rxq_num = args->rxq_num ? args->rxq_num : 1; + txq_num = args->txq_num ? args->txq_num : vlib_get_n_threads (); + + /* Sync capabilities */ + id->n_rx_queues = rxq_num; + id->n_tx_queues = txq_num; + id->csum_caps = caps.csum_caps; + id->seg_caps = caps.seg_caps; + id->hsplit_caps = caps.hsplit_caps; + id->rsc_caps = caps.rsc_caps; + id->rss_caps = caps.rss_caps; + id->other_caps = caps.other_caps; + id->max_rx_q = caps.max_rx_q; + id->max_tx_q = caps.max_tx_q; + id->max_rx_bufq = caps.max_rx_bufq; + id->max_tx_complq = caps.max_tx_complq; + id->max_sriov_vfs = caps.max_sriov_vfs; + id->max_vports = caps.max_vports; + id->default_num_vports = caps.default_num_vports; + + id->vports = clib_mem_alloc (id->max_vports * sizeof (*id->vports)); + id->max_rxq_per_msg = + (IDPF_DFLT_MBX_BUF_SIZE - sizeof (virtchnl2_config_rx_queues_t)) / + sizeof (virtchnl2_rxq_info_t); + id->max_txq_per_msg = + (IDPF_DFLT_MBX_BUF_SIZE - sizeof (virtchnl2_config_tx_queues_t)) / + sizeof (virtchnl2_txq_info_t); + + id->cur_vport_idx = 0; + id->cur_vports = 0; + id->cur_vport_nb = 0; + + if (!args->rxq_single) + id->rxq_model = 1; + if (!args->txq_single) + id->txq_model = 1; + + /* Init and enable vports */ + if (args->req_vport_nb == 1) + { + vport_param.id = id; + vport_param.idx = 0; + error = idpf_dev_vport_init (vm, id, &vport_param); + if (error) + return error; + vport = id->vports[vport_param.idx]; + error = idpf_ena_dis_vport (vm, id, vport, true); + if (error) + return error; + id->cur_vports |= 1ULL << vport_param.idx; + id->cur_vport_nb++; + id->cur_vport_idx++; + error = idpf_queue_init (vm, id, vport, args); + if (error) + return error; + } + else + { + for (i = 0; i < args->req_vport_nb; i++) + { + vport_param.id = id; + vport_param.idx = i; + if ((error = idpf_dev_vport_init (vm, id, &vport_param))) + return error; + vport = id->vports[vport_param.idx]; + error = idpf_ena_dis_vport (vm, id, vport, true); + if (error) + return error; + id->cur_vports |= 1ULL << vport_param.idx; + id->cur_vport_nb++; + id->cur_vport_idx++; + error = idpf_queue_init (vm, id, vport, args); + if (error) + return error; + } + } + + id->flags |= IDPF_DEVICE_F_INITIALIZED; + return error; +} + +static u32 +idpf_flag_change (vnet_main_t *vnm, vnet_hw_interface_t *hw, u32 flags) +{ + idpf_device_t *id = idpf_get_device (hw->dev_instance); + + switch (flags) + { + case ETHERNET_INTERFACE_FLAG_DEFAULT_L3: + id->flags &= ~IDPF_DEVICE_F_PROMISC; + break; + case ETHERNET_INTERFACE_FLAG_ACCEPT_ALL: + id->flags |= IDPF_DEVICE_F_PROMISC; + break; + default: + return ~0; + } + + return 0; +} + +void +idpf_delete_if (vlib_main_t *vm, idpf_device_t *id, int with_barrier) +{ + vnet_main_t *vnm = vnet_get_main (); + idpf_main_t *im = &idpf_main; + idpf_vport_t *vport; + int i; + u32 dev_instance; + + id->flags &= ~IDPF_DEVICE_F_ADMIN_UP; + + if (id->hw_if_index) + { + if (with_barrier) + vlib_worker_thread_barrier_sync (vm); + vnet_hw_interface_set_flags (vnm, id->hw_if_index, 0); + ethernet_delete_interface (vnm, id->hw_if_index); + if (with_barrier) + vlib_worker_thread_barrier_release (vm); + } + + for (i = 0; i < id->cur_vport_nb; i++) + { + vport = id->vports[i]; + if (vport->recv_vectors != NULL) + idpf_dealloc_vectors (vm, id, vport); + } + + vlib_pci_device_close (vm, id->pci_dev_handle); + + vlib_physmem_free (vm, id->asq); + vlib_physmem_free (vm, id->arq); + + for (i = 0; i < id->cur_vport_nb; i++) + { + vport = id->vports[i]; + vec_foreach_index (i, vport->rxqs) + { + idpf_rxq_t *rxq = vec_elt_at_index (vport->rxqs, i); + vlib_physmem_free (vm, (void *) rxq->descs); + if (rxq->n_enqueued) + vlib_buffer_free_from_ring (vm, rxq->bufs, rxq->next, rxq->size, + rxq->n_enqueued); + vec_free (rxq->bufs); + } + + vec_free (vport->rxqs); + + vec_foreach_index (i, vport->txqs) + { + idpf_txq_t *txq = vec_elt_at_index (vport->txqs, i); + vlib_physmem_free (vm, (void *) txq->descs); + if (txq->n_enqueued) + { + u16 first = (txq->next - txq->n_enqueued) & (txq->size - 1); + vlib_buffer_free_from_ring (vm, txq->bufs, first, txq->size, + txq->n_enqueued); + } + vec_free (txq->ph_bufs); + vec_free (txq->bufs); + clib_ring_free (txq->rs_slots); + vec_free (txq->tmp_bufs); + vec_free (txq->tmp_descs); + clib_spinlock_free (&txq->lock); + } + vec_free (vport->txqs); + } + + vec_free (id->name); + + clib_error_free (id->error); + dev_instance = id->dev_instance; + clib_mem_free (id->mbx_resp); + clib_memset (id, 0, sizeof (*id)); + pool_put_index (im->devices, dev_instance); + clib_mem_free (id); +} + +static u8 +idpf_validate_queue_size (idpf_create_if_args_t *args) +{ + clib_error_t *error = 0; + + args->rxq_size = (args->rxq_size == 0) ? IDPF_RXQ_SZ : args->rxq_size; + args->txq_size = (args->txq_size == 0) ? IDPF_TXQ_SZ : args->txq_size; + + if ((args->rxq_size > IDPF_QUEUE_SZ_MAX) || + (args->txq_size > IDPF_QUEUE_SZ_MAX)) + { + args->rv = VNET_API_ERROR_INVALID_VALUE; + args->error = clib_error_return ( + error, "queue size must not be greater than %u", IDPF_QUEUE_SZ_MAX); + return 1; + } + if ((args->rxq_size < IDPF_QUEUE_SZ_MIN) || + (args->txq_size < IDPF_QUEUE_SZ_MIN)) + { + args->rv = VNET_API_ERROR_INVALID_VALUE; + args->error = clib_error_return ( + error, "queue size must not be smaller than %u", IDPF_QUEUE_SZ_MIN); + return 1; + } + if ((args->rxq_size & (args->rxq_size - 1)) || + (args->txq_size & (args->txq_size - 1))) + { + args->rv = VNET_API_ERROR_INVALID_VALUE; + args->error = + clib_error_return (error, "queue size must be a power of two"); + return 1; + } + return 0; +} + +void +idpf_process_one_device (vlib_main_t *vm, idpf_device_t *id, int is_irq) +{ + /* placeholder */ + return; +} + +static uword +idpf_process (vlib_main_t *vm, vlib_node_runtime_t *rt, vlib_frame_t *f) +{ + idpf_main_t *im = &idpf_main; + uword *event_data = 0, event_type; + int enabled = 0, irq; + f64 last_run_duration = 0; + f64 last_periodic_time = 0; + idpf_device_t **dev_pointers = 0; + u32 i; + + while (1) + { + if (enabled) + vlib_process_wait_for_event_or_clock (vm, 5.0 - last_run_duration); + else + vlib_process_wait_for_event (vm); + + event_type = vlib_process_get_events (vm, &event_data); + irq = 0; + + switch (event_type) + { + case ~0: + last_periodic_time = vlib_time_now (vm); + break; + case IDPF_PROCESS_EVENT_START: + enabled = 1; + break; + case IDPF_PROCESS_EVENT_DELETE_IF: + for (int i = 0; i < vec_len (event_data); i++) + { + idpf_device_t *id = idpf_get_device (event_data[i]); + idpf_delete_if (vm, id, /* with_barrier */ 1); + } + if (pool_elts (im->devices) < 1) + enabled = 0; + break; + case IDPF_PROCESS_EVENT_AQ_INT: + irq = 1; + break; + + default: + ASSERT (0); + } + + vec_reset_length (event_data); + + if (enabled == 0) + continue; + + /* create local list of device pointers as device pool may grow + * during suspend */ + vec_reset_length (dev_pointers); + + pool_foreach_index (i, im->devices) + { + vec_add1 (dev_pointers, idpf_get_device (i)); + } + + vec_foreach_index (i, dev_pointers) + { + idpf_process_one_device (vm, dev_pointers[i], irq); + }; + + last_run_duration = vlib_time_now (vm) - last_periodic_time; + } + return 0; +} + +VLIB_REGISTER_NODE (idpf_process_node) = { + .function = idpf_process, + .type = VLIB_NODE_TYPE_PROCESS, + .name = "idpf-process", +}; + +void +idpf_create_if (vlib_main_t *vm, idpf_create_if_args_t *args) +{ + vnet_main_t *vnm = vnet_get_main (); + vnet_eth_interface_registration_t eir = {}; + idpf_main_t *im = &idpf_main; + idpf_device_t *id, **idp; + vlib_pci_dev_handle_t h; + clib_error_t *error = 0; + int i, j, v; + + /* check input args */ + if (idpf_validate_queue_size (args) != 0) + return; + + pool_foreach (idp, im->devices) + { + if ((*idp)->pci_addr.as_u32 == args->addr.as_u32) + { + args->rv = VNET_API_ERROR_ADDRESS_IN_USE; + args->error = + clib_error_return (error, "%U: %s", format_vlib_pci_addr, + &args->addr, "pci address in use"); + return; + } + } + + pool_get (im->devices, idp); + idp[0] = id = + clib_mem_alloc_aligned (sizeof (idpf_device_t), CLIB_CACHE_LINE_BYTES); + clib_memset (id, 0, sizeof (idpf_device_t)); + id->mbx_resp = clib_mem_alloc (IDPF_DFLT_MBX_BUF_SIZE); + id->dev_instance = idp - im->devices; + id->per_interface_next_index = ~0; + id->name = vec_dup (args->name); + + if ((error = + vlib_pci_device_open (vm, &args->addr, idpf_pci_device_ids, &h))) + { + pool_put (im->devices, idp); + clib_mem_free (id); + args->rv = VNET_API_ERROR_INVALID_INTERFACE; + args->error = clib_error_return (error, "pci-addr %U", + format_vlib_pci_addr, &args->addr); + return; + } + id->pci_dev_handle = h; + id->pci_addr = args->addr; + id->numa_node = vlib_pci_get_numa_node (vm, h); + + vlib_pci_set_private_data (vm, h, id->dev_instance); + + if ((error = vlib_pci_bus_master_enable (vm, h))) + goto error; + + if ((error = vlib_pci_map_region (vm, h, 0, &id->bar0))) + goto error; + + if (vlib_pci_supports_virtual_addr_dma (vm, h)) + id->flags |= IDPF_DEVICE_F_VA_DMA; + + if ((error = idpf_device_init (vm, im, id, args))) + goto error; + + /* create interface */ + eir.dev_class_index = idpf_device_class.index; + eir.dev_instance = id->dev_instance; + eir.address = id->hwaddr; + eir.cb.flag_change = idpf_flag_change; + id->hw_if_index = vnet_eth_register_interface (vnm, &eir); + + ethernet_set_flags (vnm, id->hw_if_index, + ETHERNET_INTERFACE_FLAG_DEFAULT_L3); + + vnet_sw_interface_t *sw = vnet_get_hw_sw_interface (vnm, id->hw_if_index); + args->sw_if_index = id->sw_if_index = sw->sw_if_index; + + vnet_hw_if_set_caps (vnm, id->hw_if_index, + VNET_HW_IF_CAP_INT_MODE | VNET_HW_IF_CAP_MAC_FILTER | + VNET_HW_IF_CAP_TX_CKSUM | VNET_HW_IF_CAP_TCP_GSO); + + for (v = 0; v < id->cur_vport_nb; v++) + { + for (j = 0; j < id->n_rx_queues; j++) + { + u32 qi; + i = v * id->n_rx_queues + j; + qi = vnet_hw_if_register_rx_queue (vnm, id->hw_if_index, i, + VNET_HW_IF_RXQ_THREAD_ANY); + id->vports[v]->rxqs[j].queue_index = qi; + } + for (j = 0; j < id->n_tx_queues; j++) + { + u32 qi; + i = v * id->n_tx_queues + j; + qi = vnet_hw_if_register_tx_queue (vnm, id->hw_if_index, i); + id->vports[v]->txqs[j].queue_index = qi; + } + } + + for (v = 0; v < id->cur_vport_nb; v++) + for (i = 0; i < vlib_get_n_threads (); i++) + { + u32 qi = id->vports[v]->txqs[i % id->n_tx_queues].queue_index; + vnet_hw_if_tx_queue_assign_thread (vnm, qi, i); + } + + vnet_hw_if_update_runtime_data (vnm, id->hw_if_index); + + if (pool_elts (im->devices) == 1) + vlib_process_signal_event (vm, idpf_process_node.index, + IDPF_PROCESS_EVENT_START, 0); + + return; + +error: + idpf_delete_if (vm, id, /* with_barrier */ 0); + args->rv = VNET_API_ERROR_INVALID_INTERFACE; + args->error = clib_error_return (error, "pci-addr %U", format_vlib_pci_addr, + &args->addr); + idpf_log_err (id, "error: %U", format_clib_error, args->error); +} + +void * +idpf_alloc_dma_mem (vlib_main_t *vm, idpf_device_t *id, idpf_dma_mem_t *mem, + u64 size) +{ + void *mz = NULL; + vlib_pci_dev_handle_t h = id->pci_dev_handle; + + if (!mem) + return NULL; + + /* Fixme */ + mz = vlib_physmem_alloc_aligned_on_numa (vm, size, CLIB_CACHE_LINE_BYTES, + id->numa_node); + if (!mz) + return NULL; + if (vlib_pci_map_dma (vm, h, mz)) + return NULL; + + mem->size = size; + if (id->flags & IDPF_DEVICE_F_VA_DMA) + { + mem->va = mz; + clib_memset (mem->va, 0, size); + } + else + { + mem->va = NULL; + } + mem->pa = idpf_dma_addr (vm, id, mz); + + return mem->va; +} + +void +idpf_free_dma_mem (idpf_device_t *id, idpf_dma_mem_t *mem) +{ + mem->size = 0; + mem->va = NULL; + mem->pa = 0; + + clib_mem_free (mem); +} + +static clib_error_t * +idpf_interface_admin_up_down (vnet_main_t *vnm, u32 hw_if_index, u32 flags) +{ + vnet_hw_interface_t *hi = vnet_get_hw_interface (vnm, hw_if_index); + idpf_device_t *id = idpf_get_device (hi->dev_instance); + uword is_up = (flags & VNET_SW_INTERFACE_FLAG_ADMIN_UP) != 0; + + if (id->flags & IDPF_DEVICE_F_ERROR) + return clib_error_return (0, "device is in error state"); + + if (is_up) + { + vnet_hw_interface_set_flags (vnm, id->hw_if_index, + VNET_HW_INTERFACE_FLAG_LINK_UP); + id->flags |= IDPF_DEVICE_F_ADMIN_UP; + } + else + { + vnet_hw_interface_set_flags (vnm, id->hw_if_index, 0); + id->flags &= ~IDPF_DEVICE_F_ADMIN_UP; + } + return 0; +} + +VNET_DEVICE_CLASS (idpf_device_class, ) = { + .name = "Infrastructure Data Path Function (IDPF) interface", + .format_device_name = format_idpf_device_name, + .admin_up_down_function = idpf_interface_admin_up_down, +}; + +clib_error_t * +idpf_init (vlib_main_t *vm) +{ + idpf_main_t *im = &idpf_main; + vlib_thread_main_t *tm = vlib_get_thread_main (); + + vec_validate_aligned (im->per_thread_data, tm->n_vlib_mains - 1, + CLIB_CACHE_LINE_BYTES); + + return 0; +} + +VLIB_INIT_FUNCTION (idpf_init) = { + .runs_after = VLIB_INITS ("pci_bus_init"), +}; + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ -- cgit 1.2.3-korg