diff --git a/sys/dev/mlx5/mlx5_ib/mlx5_ib_ah.c b/sys/dev/mlx5/mlx5_ib/mlx5_ib_ah.c index ce3ef156147a..8371f2b808b2 100644 --- a/sys/dev/mlx5/mlx5_ib/mlx5_ib_ah.c +++ b/sys/dev/mlx5/mlx5_ib/mlx5_ib_ah.c @@ -1,133 +1,133 @@ /*- * Copyright (c) 2013-2015, Mellanox Technologies, Ltd. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY AUTHOR AND CONTRIBUTORS `AS IS' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $FreeBSD$ */ #include "opt_rss.h" #include "opt_ratelimit.h" -#include "mlx5_ib.h" +#include static struct ib_ah *create_ib_ah(struct mlx5_ib_dev *dev, struct mlx5_ib_ah *ah, struct ib_ah_attr *ah_attr, enum rdma_link_layer ll) { if (ah_attr->ah_flags & IB_AH_GRH) { memcpy(ah->av.rgid, &ah_attr->grh.dgid, 16); ah->av.grh_gid_fl = cpu_to_be32(ah_attr->grh.flow_label | (1 << 30) | ah_attr->grh.sgid_index << 20); ah->av.hop_limit = ah_attr->grh.hop_limit; ah->av.tclass = ah_attr->grh.traffic_class; } ah->av.stat_rate_sl = (ah_attr->static_rate << 4); if (ll == IB_LINK_LAYER_ETHERNET) { memcpy(ah->av.rmac, ah_attr->dmac, sizeof(ah_attr->dmac)); ah->av.udp_sport = mlx5_get_roce_udp_sport(dev, ah_attr->port_num, ah_attr->grh.sgid_index); ah->av.stat_rate_sl |= (ah_attr->sl & 0x7) << 1; } else { ah->av.rlid = cpu_to_be16(ah_attr->dlid); ah->av.fl_mlid = ah_attr->src_path_bits & 0x7f; ah->av.stat_rate_sl |= (ah_attr->sl & 0xf); } return &ah->ibah; } struct ib_ah *mlx5_ib_create_ah(struct ib_pd *pd, struct ib_ah_attr *ah_attr, struct ib_udata *udata) { struct mlx5_ib_ah *ah; struct mlx5_ib_dev *dev = to_mdev(pd->device); enum rdma_link_layer ll; ll = pd->device->get_link_layer(pd->device, ah_attr->port_num); if (ll == IB_LINK_LAYER_ETHERNET && !(ah_attr->ah_flags & IB_AH_GRH)) return ERR_PTR(-EINVAL); if (ll == IB_LINK_LAYER_ETHERNET && udata) { int err; struct mlx5_ib_create_ah_resp resp = {}; u32 min_resp_len = offsetof(typeof(resp), dmac) + sizeof(resp.dmac); if (udata->outlen < min_resp_len) return ERR_PTR(-EINVAL); resp.response_length = min_resp_len; err = ib_resolve_eth_dmac(pd->device, ah_attr); if (err) return ERR_PTR(err); memcpy(resp.dmac, ah_attr->dmac, ETH_ALEN); err = ib_copy_to_udata(udata, &resp, resp.response_length); if (err) return ERR_PTR(err); } ah = kzalloc(sizeof(*ah), GFP_ATOMIC); if (!ah) return ERR_PTR(-ENOMEM); return create_ib_ah(dev, ah, ah_attr, ll); /* never fails */ } int mlx5_ib_query_ah(struct ib_ah *ibah, struct ib_ah_attr *ah_attr) { struct mlx5_ib_ah *ah = to_mah(ibah); u32 tmp; memset(ah_attr, 0, sizeof(*ah_attr)); tmp = be32_to_cpu(ah->av.grh_gid_fl); if (tmp & (1 << 30)) { ah_attr->ah_flags = IB_AH_GRH; ah_attr->grh.sgid_index = (tmp >> 20) & 0xff; ah_attr->grh.flow_label = tmp & 0xfffff; memcpy(&ah_attr->grh.dgid, ah->av.rgid, 16); ah_attr->grh.hop_limit = ah->av.hop_limit; ah_attr->grh.traffic_class = ah->av.tclass; } ah_attr->dlid = be16_to_cpu(ah->av.rlid); ah_attr->static_rate = ah->av.stat_rate_sl >> 4; ah_attr->sl = ah->av.stat_rate_sl & 0xf; return 0; } int mlx5_ib_destroy_ah(struct ib_ah *ah) { kfree(to_mah(ah)); return 0; } diff --git a/sys/dev/mlx5/mlx5_ib/mlx5_ib_cong.c b/sys/dev/mlx5/mlx5_ib/mlx5_ib_cong.c index afc2609919fb..da2fe114a35e 100644 --- a/sys/dev/mlx5/mlx5_ib/mlx5_ib_cong.c +++ b/sys/dev/mlx5/mlx5_ib/mlx5_ib_cong.c @@ -1,553 +1,552 @@ /*- * Copyright (c) 2013-2020, Mellanox Technologies, Ltd. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY AUTHOR AND CONTRIBUTORS `AS IS' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $FreeBSD$ */ #include "opt_rss.h" #include "opt_ratelimit.h" -#include "mlx5_ib.h" - +#include #include static const char *mlx5_ib_cong_params_desc[] = { MLX5_IB_CONG_PARAMS(MLX5_IB_STATS_DESC) }; static const char *mlx5_ib_cong_status_desc[] = { MLX5_IB_CONG_STATUS(MLX5_IB_STATS_DESC) }; static const char *mlx5_ib_cong_stats_desc[] = { MLX5_IB_CONG_STATS(MLX5_IB_STATS_DESC) }; #define MLX5_IB_INDEX(field) ( \ (__offsetof(struct mlx5_ib_congestion, field) - \ __offsetof(struct mlx5_ib_congestion, arg[0])) / sizeof(u64)) #define MLX5_IB_FLD_MAX(type, field) ((1ULL << __mlx5_bit_sz(type, field)) - 1ULL) #define MLX5_IB_SET_CLIPPED(type, ptr, field, var) do { \ /* rangecheck */ \ if ((var) > MLX5_IB_FLD_MAX(type, field)) \ (var) = MLX5_IB_FLD_MAX(type, field); \ /* set value */ \ MLX5_SET(type, ptr, field, var); \ } while (0) #define CONG_LOCK(dev) sx_xlock(&(dev)->congestion.lock) #define CONG_UNLOCK(dev) sx_xunlock(&(dev)->congestion.lock) #define CONG_LOCKED(dev) sx_xlocked(&(dev)->congestion.lock) #define MLX5_IB_RP_CLAMP_TGT_RATE_ATTR BIT(1) #define MLX5_IB_RP_CLAMP_TGT_RATE_ATI_ATTR BIT(2) #define MLX5_IB_RP_TIME_RESET_ATTR BIT(3) #define MLX5_IB_RP_BYTE_RESET_ATTR BIT(4) #define MLX5_IB_RP_THRESHOLD_ATTR BIT(5) #define MLX5_IB_RP_AI_RATE_ATTR BIT(7) #define MLX5_IB_RP_HAI_RATE_ATTR BIT(8) #define MLX5_IB_RP_MIN_DEC_FAC_ATTR BIT(9) #define MLX5_IB_RP_MIN_RATE_ATTR BIT(10) #define MLX5_IB_RP_RATE_TO_SET_ON_FIRST_CNP_ATTR BIT(11) #define MLX5_IB_RP_DCE_TCP_G_ATTR BIT(12) #define MLX5_IB_RP_DCE_TCP_RTT_ATTR BIT(13) #define MLX5_IB_RP_RATE_REDUCE_MONITOR_PERIOD_ATTR BIT(14) #define MLX5_IB_RP_INITIAL_ALPHA_VALUE_ATTR BIT(15) #define MLX5_IB_RP_GD_ATTR BIT(16) #define MLX5_IB_NP_CNP_DSCP_ATTR BIT(3) #define MLX5_IB_NP_CNP_PRIO_MODE_ATTR BIT(4) enum mlx5_ib_cong_node_type { MLX5_IB_RROCE_ECN_RP = 1, MLX5_IB_RROCE_ECN_NP = 2, }; static enum mlx5_ib_cong_node_type mlx5_ib_param_to_node(u32 index) { if (index >= MLX5_IB_INDEX(rp_clamp_tgt_rate) && index <= MLX5_IB_INDEX(rp_gd)) return MLX5_IB_RROCE_ECN_RP; else return MLX5_IB_RROCE_ECN_NP; } static u64 mlx5_get_cc_param_val(void *field, u32 index) { switch (index) { case MLX5_IB_INDEX(rp_clamp_tgt_rate): return MLX5_GET(cong_control_r_roce_ecn_rp, field, clamp_tgt_rate); case MLX5_IB_INDEX(rp_clamp_tgt_rate_ati): return MLX5_GET(cong_control_r_roce_ecn_rp, field, clamp_tgt_rate_after_time_inc); case MLX5_IB_INDEX(rp_time_reset): return MLX5_GET(cong_control_r_roce_ecn_rp, field, rpg_time_reset); case MLX5_IB_INDEX(rp_byte_reset): return MLX5_GET(cong_control_r_roce_ecn_rp, field, rpg_byte_reset); case MLX5_IB_INDEX(rp_threshold): return MLX5_GET(cong_control_r_roce_ecn_rp, field, rpg_threshold); case MLX5_IB_INDEX(rp_ai_rate): return MLX5_GET(cong_control_r_roce_ecn_rp, field, rpg_ai_rate); case MLX5_IB_INDEX(rp_hai_rate): return MLX5_GET(cong_control_r_roce_ecn_rp, field, rpg_hai_rate); case MLX5_IB_INDEX(rp_min_dec_fac): return MLX5_GET(cong_control_r_roce_ecn_rp, field, rpg_min_dec_fac); case MLX5_IB_INDEX(rp_min_rate): return MLX5_GET(cong_control_r_roce_ecn_rp, field, rpg_min_rate); case MLX5_IB_INDEX(rp_rate_to_set_on_first_cnp): return MLX5_GET(cong_control_r_roce_ecn_rp, field, rate_to_set_on_first_cnp); case MLX5_IB_INDEX(rp_dce_tcp_g): return MLX5_GET(cong_control_r_roce_ecn_rp, field, dce_tcp_g); case MLX5_IB_INDEX(rp_dce_tcp_rtt): return MLX5_GET(cong_control_r_roce_ecn_rp, field, dce_tcp_rtt); case MLX5_IB_INDEX(rp_rate_reduce_monitor_period): return MLX5_GET(cong_control_r_roce_ecn_rp, field, rate_reduce_monitor_period); case MLX5_IB_INDEX(rp_initial_alpha_value): return MLX5_GET(cong_control_r_roce_ecn_rp, field, initial_alpha_value); case MLX5_IB_INDEX(rp_gd): return MLX5_GET(cong_control_r_roce_ecn_rp, field, rpg_gd); case MLX5_IB_INDEX(np_cnp_dscp): return MLX5_GET(cong_control_r_roce_ecn_np, field, cnp_dscp); case MLX5_IB_INDEX(np_cnp_prio_mode): return MLX5_GET(cong_control_r_roce_ecn_np, field, cnp_prio_mode); case MLX5_IB_INDEX(np_cnp_prio): return MLX5_GET(cong_control_r_roce_ecn_np, field, cnp_802p_prio); default: return 0; } } static void mlx5_ib_set_cc_param_mask_val(void *field, u32 index, u64 var, u32 *attr_mask) { switch (index) { case MLX5_IB_INDEX(rp_clamp_tgt_rate): *attr_mask |= MLX5_IB_RP_CLAMP_TGT_RATE_ATTR; MLX5_IB_SET_CLIPPED(cong_control_r_roce_ecn_rp, field, clamp_tgt_rate, var); break; case MLX5_IB_INDEX(rp_clamp_tgt_rate_ati): *attr_mask |= MLX5_IB_RP_CLAMP_TGT_RATE_ATI_ATTR; MLX5_IB_SET_CLIPPED(cong_control_r_roce_ecn_rp, field, clamp_tgt_rate_after_time_inc, var); break; case MLX5_IB_INDEX(rp_time_reset): *attr_mask |= MLX5_IB_RP_TIME_RESET_ATTR; MLX5_IB_SET_CLIPPED(cong_control_r_roce_ecn_rp, field, rpg_time_reset, var); break; case MLX5_IB_INDEX(rp_byte_reset): *attr_mask |= MLX5_IB_RP_BYTE_RESET_ATTR; MLX5_IB_SET_CLIPPED(cong_control_r_roce_ecn_rp, field, rpg_byte_reset, var); break; case MLX5_IB_INDEX(rp_threshold): *attr_mask |= MLX5_IB_RP_THRESHOLD_ATTR; MLX5_IB_SET_CLIPPED(cong_control_r_roce_ecn_rp, field, rpg_threshold, var); break; case MLX5_IB_INDEX(rp_ai_rate): *attr_mask |= MLX5_IB_RP_AI_RATE_ATTR; MLX5_IB_SET_CLIPPED(cong_control_r_roce_ecn_rp, field, rpg_ai_rate, var); break; case MLX5_IB_INDEX(rp_hai_rate): *attr_mask |= MLX5_IB_RP_HAI_RATE_ATTR; MLX5_IB_SET_CLIPPED(cong_control_r_roce_ecn_rp, field, rpg_hai_rate, var); break; case MLX5_IB_INDEX(rp_min_dec_fac): *attr_mask |= MLX5_IB_RP_MIN_DEC_FAC_ATTR; MLX5_IB_SET_CLIPPED(cong_control_r_roce_ecn_rp, field, rpg_min_dec_fac, var); break; case MLX5_IB_INDEX(rp_min_rate): *attr_mask |= MLX5_IB_RP_MIN_RATE_ATTR; MLX5_IB_SET_CLIPPED(cong_control_r_roce_ecn_rp, field, rpg_min_rate, var); break; case MLX5_IB_INDEX(rp_rate_to_set_on_first_cnp): *attr_mask |= MLX5_IB_RP_RATE_TO_SET_ON_FIRST_CNP_ATTR; MLX5_IB_SET_CLIPPED(cong_control_r_roce_ecn_rp, field, rate_to_set_on_first_cnp, var); break; case MLX5_IB_INDEX(rp_dce_tcp_g): *attr_mask |= MLX5_IB_RP_DCE_TCP_G_ATTR; MLX5_IB_SET_CLIPPED(cong_control_r_roce_ecn_rp, field, dce_tcp_g, var); break; case MLX5_IB_INDEX(rp_dce_tcp_rtt): *attr_mask |= MLX5_IB_RP_DCE_TCP_RTT_ATTR; MLX5_IB_SET_CLIPPED(cong_control_r_roce_ecn_rp, field, dce_tcp_rtt, var); break; case MLX5_IB_INDEX(rp_rate_reduce_monitor_period): *attr_mask |= MLX5_IB_RP_RATE_REDUCE_MONITOR_PERIOD_ATTR; MLX5_IB_SET_CLIPPED(cong_control_r_roce_ecn_rp, field, rate_reduce_monitor_period, var); break; case MLX5_IB_INDEX(rp_initial_alpha_value): *attr_mask |= MLX5_IB_RP_INITIAL_ALPHA_VALUE_ATTR; MLX5_IB_SET_CLIPPED(cong_control_r_roce_ecn_rp, field, initial_alpha_value, var); break; case MLX5_IB_INDEX(rp_gd): *attr_mask |= MLX5_IB_RP_GD_ATTR; MLX5_IB_SET_CLIPPED(cong_control_r_roce_ecn_rp, field, rpg_gd, var); break; case MLX5_IB_INDEX(np_cnp_dscp): *attr_mask |= MLX5_IB_NP_CNP_DSCP_ATTR; MLX5_IB_SET_CLIPPED(cong_control_r_roce_ecn_np, field, cnp_dscp, var); break; case MLX5_IB_INDEX(np_cnp_prio_mode): *attr_mask |= MLX5_IB_NP_CNP_PRIO_MODE_ATTR; MLX5_IB_SET_CLIPPED(cong_control_r_roce_ecn_np, field, cnp_prio_mode, var); break; case MLX5_IB_INDEX(np_cnp_prio): *attr_mask |= MLX5_IB_NP_CNP_PRIO_MODE_ATTR; MLX5_SET(cong_control_r_roce_ecn_np, field, cnp_prio_mode, 0); MLX5_IB_SET_CLIPPED(cong_control_r_roce_ecn_np, field, cnp_802p_prio, var); break; default: break; } } static int mlx5_ib_get_all_cc_params(struct mlx5_ib_dev *dev) { int outlen = MLX5_ST_SZ_BYTES(query_cong_params_out); enum mlx5_ib_cong_node_type node = 0; void *out; void *field; u32 x; int err = 0; out = kzalloc(outlen, GFP_KERNEL); if (!out) return -ENOMEM; /* get the current values */ for (x = 0; x != MLX5_IB_CONG_PARAMS_NUM; x++) { if (node != mlx5_ib_param_to_node(x)) { node = mlx5_ib_param_to_node(x); err = mlx5_cmd_query_cong_params(dev->mdev, node, out, outlen); if (err) break; } field = MLX5_ADDR_OF(query_cong_params_out, out, congestion_parameters); dev->congestion.arg[x] = mlx5_get_cc_param_val(field, x); } kfree(out); return err; } static int mlx5_ib_set_cc_params(struct mlx5_ib_dev *dev, u32 index, u64 var) { int inlen = MLX5_ST_SZ_BYTES(modify_cong_params_in); enum mlx5_ib_cong_node_type node; u32 attr_mask = 0; void *field; void *in; int err; in = kzalloc(inlen, GFP_KERNEL); if (!in) return -ENOMEM; MLX5_SET(modify_cong_params_in, in, opcode, MLX5_CMD_OP_MODIFY_CONG_PARAMS); node = mlx5_ib_param_to_node(index); MLX5_SET(modify_cong_params_in, in, cong_protocol, node); field = MLX5_ADDR_OF(modify_cong_params_in, in, congestion_parameters); mlx5_ib_set_cc_param_mask_val(field, index, var, &attr_mask); field = MLX5_ADDR_OF(modify_cong_params_in, in, field_select); MLX5_SET(field_select_r_roce_rp, field, field_select_r_roce_rp, attr_mask); err = mlx5_cmd_modify_cong_params(dev->mdev, in, inlen); kfree(in); return err; } static int mlx5_ib_cong_params_handler(SYSCTL_HANDLER_ARGS) { struct mlx5_ib_dev *dev = arg1; u64 value; int error; CONG_LOCK(dev); value = dev->congestion.arg[arg2]; if (req != NULL) { error = sysctl_handle_64(oidp, &value, 0, req); if (error || req->newptr == NULL || value == dev->congestion.arg[arg2]) goto done; /* assign new value */ dev->congestion.arg[arg2] = value; } else { error = 0; } if (!MLX5_CAP_GEN(dev->mdev, cc_modify_allowed)) error = EPERM; else { error = -mlx5_ib_set_cc_params(dev, MLX5_IB_INDEX(arg[arg2]), dev->congestion.arg[arg2]); } done: CONG_UNLOCK(dev); return (error); } static int mlx5_ib_get_all_cc_status(struct mlx5_ib_dev *dev) { const int outlen = MLX5_ST_SZ_BYTES(query_cong_status_out); uint32_t out[MLX5_ST_SZ_DW(query_cong_status_out)] = {}; int error; #define MLX5_IB_CONG_STATUS_READ(a,b,c,d,e,node,prio,field) do { \ error = mlx5_cmd_query_cong_status(dev->mdev, node, prio, out, outlen); \ if (error) \ goto done; \ dev->congestion.c = MLX5_GET(query_cong_status_out, out, field); \ } while (0); MLX5_IB_CONG_STATUS(MLX5_IB_CONG_STATUS_READ); done: return (error); } static int mlx5_ib_cong_status_handler(SYSCTL_HANDLER_ARGS) { const int inlen = MLX5_ST_SZ_BYTES(modify_cong_status_in); uint32_t in[MLX5_ST_SZ_DW(modify_cong_status_in)] = {}; struct mlx5_ib_dev *dev = arg1; u64 value; int error; CONG_LOCK(dev); value = dev->congestion.arg[arg2]; if (req != NULL) { error = sysctl_handle_64(oidp, &value, 0, req); /* convert value into a boolean */ value = value ? 1 : 0; if (error || req->newptr == NULL || value == dev->congestion.arg[arg2]) goto done; /* assign new binary value */ dev->congestion.arg[arg2] = value; } else { error = 0; } if (!MLX5_CAP_GEN(dev->mdev, cc_modify_allowed)) error = EPERM; else switch (arg2) { #define MLX5_IB_CONG_STATUS_WRITE(a,b,c,d,e,node,prio,field) \ case MLX5_IB_INDEX(c): \ MLX5_SET(modify_cong_status_in, in, opcode, \ MLX5_CMD_OP_MODIFY_CONG_STATUS); \ MLX5_SET(modify_cong_status_in, in, priority, prio); \ MLX5_SET(modify_cong_status_in, in, cong_protocol, node); \ MLX5_SET(modify_cong_status_in, in, field, value); \ error = -mlx5_cmd_modify_cong_status(dev->mdev, in, inlen); \ break; MLX5_IB_CONG_STATUS(MLX5_IB_CONG_STATUS_WRITE) default: error = EINVAL; break; } done: CONG_UNLOCK(dev); return (error); } #define MLX5_GET_UNALIGNED_64(t,p,f) \ (((u64)MLX5_GET(t,p,f##_high) << 32) | MLX5_GET(t,p,f##_low)) static void mlx5_ib_read_cong_stats(struct work_struct *work) { struct mlx5_ib_dev *dev = container_of(work, struct mlx5_ib_dev, congestion.dwork.work); const int outlen = MLX5_ST_SZ_BYTES(query_cong_statistics_out); void *out; out = kzalloc(outlen, GFP_KERNEL); if (!out) goto done; CONG_LOCK(dev); if (mlx5_cmd_query_cong_counter(dev->mdev, 0, out, outlen)) memset(out, 0, outlen); dev->congestion.syndrome = MLX5_GET(query_cong_statistics_out, out, syndrome); dev->congestion.rp_cur_flows = MLX5_GET(query_cong_statistics_out, out, rp_cur_flows); dev->congestion.sum_flows = MLX5_GET(query_cong_statistics_out, out, sum_flows); dev->congestion.rp_cnp_ignored = MLX5_GET_UNALIGNED_64(query_cong_statistics_out, out, rp_cnp_ignored); dev->congestion.rp_cnp_handled = MLX5_GET_UNALIGNED_64(query_cong_statistics_out, out, rp_cnp_handled); dev->congestion.time_stamp = MLX5_GET_UNALIGNED_64(query_cong_statistics_out, out, time_stamp); dev->congestion.accumulators_period = MLX5_GET(query_cong_statistics_out, out, accumulators_period); dev->congestion.np_ecn_marked_roce_packets = MLX5_GET_UNALIGNED_64(query_cong_statistics_out, out, np_ecn_marked_roce_packets); dev->congestion.np_cnp_sent = MLX5_GET_UNALIGNED_64(query_cong_statistics_out, out, np_cnp_sent); CONG_UNLOCK(dev); kfree(out); done: schedule_delayed_work(&dev->congestion.dwork, hz); } void mlx5_ib_cleanup_congestion(struct mlx5_ib_dev *dev) { while (cancel_delayed_work_sync(&dev->congestion.dwork)) ; sysctl_ctx_free(&dev->congestion.ctx); sx_destroy(&dev->congestion.lock); } int mlx5_ib_init_congestion(struct mlx5_ib_dev *dev) { struct sysctl_ctx_list *ctx; struct sysctl_oid *parent; struct sysctl_oid *node; int err; u32 x; ctx = &dev->congestion.ctx; sysctl_ctx_init(ctx); sx_init(&dev->congestion.lock, "mlx5ibcong"); INIT_DELAYED_WORK(&dev->congestion.dwork, mlx5_ib_read_cong_stats); if (!MLX5_CAP_GEN(dev->mdev, cc_query_allowed)) return (0); err = mlx5_ib_get_all_cc_params(dev); if (err) return (err); err = mlx5_ib_get_all_cc_status(dev); if (err) return (err); parent = SYSCTL_ADD_NODE(ctx, SYSCTL_CHILDREN(dev->ib_dev.dev.kobj.oidp), OID_AUTO, "cong", CTLFLAG_RW | CTLFLAG_MPSAFE, NULL, "Congestion control"); if (parent == NULL) return (-ENOMEM); node = SYSCTL_ADD_NODE(ctx, SYSCTL_CHILDREN(parent), OID_AUTO, "conf", CTLFLAG_RW | CTLFLAG_MPSAFE, NULL, "Configuration"); if (node == NULL) { sysctl_ctx_free(&dev->congestion.ctx); return (-ENOMEM); } for (x = 0; x != MLX5_IB_CONG_PARAMS_NUM; x++) { SYSCTL_ADD_PROC(ctx, SYSCTL_CHILDREN(node), OID_AUTO, mlx5_ib_cong_params_desc[2 * x], CTLTYPE_U64 | CTLFLAG_RWTUN | CTLFLAG_MPSAFE, dev, x, &mlx5_ib_cong_params_handler, "QU", mlx5_ib_cong_params_desc[2 * x + 1]); } for (x = 0; x != MLX5_IB_CONG_STATUS_NUM; x++) { SYSCTL_ADD_PROC(ctx, SYSCTL_CHILDREN(node), OID_AUTO, mlx5_ib_cong_status_desc[2 * x], CTLTYPE_U64 | CTLFLAG_RWTUN | CTLFLAG_MPSAFE, dev, x + MLX5_IB_CONG_PARAMS_NUM + MLX5_IB_CONG_STATS_NUM, &mlx5_ib_cong_status_handler, "QU", mlx5_ib_cong_status_desc[2 * x + 1]); } node = SYSCTL_ADD_NODE(ctx, SYSCTL_CHILDREN(parent), OID_AUTO, "stats", CTLFLAG_RD | CTLFLAG_MPSAFE, NULL, "Statistics"); if (node == NULL) { sysctl_ctx_free(&dev->congestion.ctx); return (-ENOMEM); } for (x = 0; x != MLX5_IB_CONG_STATS_NUM; x++) { /* read-only SYSCTLs */ SYSCTL_ADD_U64(ctx, SYSCTL_CHILDREN(node), OID_AUTO, mlx5_ib_cong_stats_desc[2 * x], CTLFLAG_RD | CTLFLAG_MPSAFE, &dev->congestion.arg[x + MLX5_IB_CONG_PARAMS_NUM], 0, mlx5_ib_cong_stats_desc[2 * x + 1]); } schedule_delayed_work(&dev->congestion.dwork, hz); return (0); } diff --git a/sys/dev/mlx5/mlx5_ib/mlx5_ib_cq.c b/sys/dev/mlx5/mlx5_ib/mlx5_ib_cq.c index 26bc8f70779f..9ce8941bbc6f 100644 --- a/sys/dev/mlx5/mlx5_ib/mlx5_ib_cq.c +++ b/sys/dev/mlx5/mlx5_ib/mlx5_ib_cq.c @@ -1,1415 +1,1415 @@ /*- * Copyright (c) 2013-2015, Mellanox Technologies, Ltd. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY AUTHOR AND CONTRIBUTORS `AS IS' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $FreeBSD$ */ #include "opt_rss.h" #include "opt_ratelimit.h" #include #include #include #include -#include "mlx5_ib.h" +#include static void mlx5_ib_cq_comp(struct mlx5_core_cq *cq, struct mlx5_eqe *eqe __unused) { struct ib_cq *ibcq = &to_mibcq(cq)->ibcq; ibcq->comp_handler(ibcq, ibcq->cq_context); } static void mlx5_ib_cq_event(struct mlx5_core_cq *mcq, int type) { struct mlx5_ib_cq *cq = container_of(mcq, struct mlx5_ib_cq, mcq); struct mlx5_ib_dev *dev = to_mdev(cq->ibcq.device); struct ib_cq *ibcq = &cq->ibcq; struct ib_event event; if (type != MLX5_EVENT_TYPE_CQ_ERROR) { mlx5_ib_warn(dev, "Unexpected event type %d on CQ %06x\n", type, mcq->cqn); return; } if (ibcq->event_handler) { event.device = &dev->ib_dev; event.event = IB_EVENT_CQ_ERR; event.element.cq = ibcq; ibcq->event_handler(&event, ibcq->cq_context); } } static void *get_cqe_from_buf(struct mlx5_ib_cq_buf *buf, int n, int size) { return mlx5_buf_offset(&buf->buf, n * size); } static void *get_cqe(struct mlx5_ib_cq *cq, int n) { return get_cqe_from_buf(&cq->buf, n, cq->mcq.cqe_sz); } static u8 sw_ownership_bit(int n, int nent) { return (n & nent) ? 1 : 0; } static void *get_sw_cqe(struct mlx5_ib_cq *cq, int n) { void *cqe = get_cqe(cq, n & cq->ibcq.cqe); struct mlx5_cqe64 *cqe64; cqe64 = (cq->mcq.cqe_sz == 64) ? cqe : cqe + 64; if (likely((cqe64->op_own) >> 4 != MLX5_CQE_INVALID) && !((cqe64->op_own & MLX5_CQE_OWNER_MASK) ^ !!(n & (cq->ibcq.cqe + 1)))) { return cqe; } else { return NULL; } } static void *next_cqe_sw(struct mlx5_ib_cq *cq) { return get_sw_cqe(cq, cq->mcq.cons_index); } static enum ib_wc_opcode get_umr_comp(struct mlx5_ib_wq *wq, int idx) { switch (wq->wr_data[idx]) { case MLX5_IB_WR_UMR: return 0; case IB_WR_LOCAL_INV: return IB_WC_LOCAL_INV; case IB_WR_REG_MR: return IB_WC_REG_MR; default: pr_warn("unknown completion status\n"); return 0; } } static void handle_good_req(struct ib_wc *wc, struct mlx5_cqe64 *cqe, struct mlx5_ib_wq *wq, int idx) { wc->wc_flags = 0; switch (be32_to_cpu(cqe->sop_drop_qpn) >> 24) { case MLX5_OPCODE_RDMA_WRITE_IMM: wc->wc_flags |= IB_WC_WITH_IMM; case MLX5_OPCODE_RDMA_WRITE: wc->opcode = IB_WC_RDMA_WRITE; break; case MLX5_OPCODE_SEND_IMM: wc->wc_flags |= IB_WC_WITH_IMM; case MLX5_OPCODE_SEND: case MLX5_OPCODE_SEND_INVAL: wc->opcode = IB_WC_SEND; break; case MLX5_OPCODE_RDMA_READ: wc->opcode = IB_WC_RDMA_READ; wc->byte_len = be32_to_cpu(cqe->byte_cnt); break; case MLX5_OPCODE_ATOMIC_CS: wc->opcode = IB_WC_COMP_SWAP; wc->byte_len = 8; break; case MLX5_OPCODE_ATOMIC_FA: wc->opcode = IB_WC_FETCH_ADD; wc->byte_len = 8; break; case MLX5_OPCODE_ATOMIC_MASKED_CS: wc->opcode = IB_WC_MASKED_COMP_SWAP; wc->byte_len = 8; break; case MLX5_OPCODE_ATOMIC_MASKED_FA: wc->opcode = IB_WC_MASKED_FETCH_ADD; wc->byte_len = 8; break; case MLX5_OPCODE_UMR: wc->opcode = get_umr_comp(wq, idx); break; } } enum { MLX5_GRH_IN_BUFFER = 1, MLX5_GRH_IN_CQE = 2, }; static void handle_responder(struct ib_wc *wc, struct mlx5_cqe64 *cqe, struct mlx5_ib_qp *qp) { enum rdma_link_layer ll = rdma_port_get_link_layer(qp->ibqp.device, 1); struct mlx5_ib_dev *dev = to_mdev(qp->ibqp.device); struct mlx5_ib_srq *srq; struct mlx5_ib_wq *wq; u16 wqe_ctr; u8 roce_packet_type; bool vlan_present; u8 g; if (qp->ibqp.srq || qp->ibqp.xrcd) { struct mlx5_core_srq *msrq = NULL; if (qp->ibqp.xrcd) { msrq = mlx5_core_get_srq(dev->mdev, be32_to_cpu(cqe->srqn)); srq = to_mibsrq(msrq); } else { srq = to_msrq(qp->ibqp.srq); } if (srq) { wqe_ctr = be16_to_cpu(cqe->wqe_counter); wc->wr_id = srq->wrid[wqe_ctr]; mlx5_ib_free_srq_wqe(srq, wqe_ctr); if (msrq && atomic_dec_and_test(&msrq->refcount)) complete(&msrq->free); } } else { wq = &qp->rq; wc->wr_id = wq->wrid[wq->tail & (wq->wqe_cnt - 1)]; ++wq->tail; } wc->byte_len = be32_to_cpu(cqe->byte_cnt); switch (cqe->op_own >> 4) { case MLX5_CQE_RESP_WR_IMM: wc->opcode = IB_WC_RECV_RDMA_WITH_IMM; wc->wc_flags = IB_WC_WITH_IMM; wc->ex.imm_data = cqe->imm_inval_pkey; break; case MLX5_CQE_RESP_SEND: wc->opcode = IB_WC_RECV; wc->wc_flags = IB_WC_IP_CSUM_OK; if (unlikely(!((cqe->hds_ip_ext & CQE_L3_OK) && (cqe->hds_ip_ext & CQE_L4_OK)))) wc->wc_flags = 0; break; case MLX5_CQE_RESP_SEND_IMM: wc->opcode = IB_WC_RECV; wc->wc_flags = IB_WC_WITH_IMM; wc->ex.imm_data = cqe->imm_inval_pkey; break; case MLX5_CQE_RESP_SEND_INV: wc->opcode = IB_WC_RECV; wc->wc_flags = IB_WC_WITH_INVALIDATE; wc->ex.invalidate_rkey = be32_to_cpu(cqe->imm_inval_pkey); break; } wc->src_qp = be32_to_cpu(cqe->flags_rqpn) & 0xffffff; wc->dlid_path_bits = cqe->ml_path; g = (be32_to_cpu(cqe->flags_rqpn) >> 28) & 3; wc->wc_flags |= g ? IB_WC_GRH : 0; if (unlikely(is_qp1(qp->ibqp.qp_type))) { u16 pkey = be32_to_cpu(cqe->imm_inval_pkey) & 0xffff; ib_find_cached_pkey(&dev->ib_dev, qp->port, pkey, &wc->pkey_index); } else { wc->pkey_index = 0; } if (ll != IB_LINK_LAYER_ETHERNET) { wc->slid = be16_to_cpu(cqe->slid); wc->sl = (be32_to_cpu(cqe->flags_rqpn) >> 24) & 0xf; return; } wc->slid = 0; vlan_present = cqe_has_vlan(cqe); roce_packet_type = (be32_to_cpu(cqe->flags_rqpn) >> 24) & 0x3; if (vlan_present) { wc->vlan_id = (be16_to_cpu(cqe->vlan_info)) & 0xfff; wc->sl = (be16_to_cpu(cqe->vlan_info) >> 13) & 0x7; wc->wc_flags |= IB_WC_WITH_VLAN; } else { wc->sl = 0; } switch (roce_packet_type) { case MLX5_CQE_ROCE_L3_HEADER_TYPE_GRH: wc->network_hdr_type = RDMA_NETWORK_IB; break; case MLX5_CQE_ROCE_L3_HEADER_TYPE_IPV6: wc->network_hdr_type = RDMA_NETWORK_IPV6; break; case MLX5_CQE_ROCE_L3_HEADER_TYPE_IPV4: wc->network_hdr_type = RDMA_NETWORK_IPV4; break; } wc->wc_flags |= IB_WC_WITH_NETWORK_HDR_TYPE; } static void dump_cqe(struct mlx5_ib_dev *dev, struct mlx5_err_cqe *cqe) { __be32 *p = (__be32 *)cqe; int i; mlx5_ib_warn(dev, "dump error cqe\n"); for (i = 0; i < sizeof(*cqe) / 16; i++, p += 4) pr_info("%08x %08x %08x %08x\n", be32_to_cpu(p[0]), be32_to_cpu(p[1]), be32_to_cpu(p[2]), be32_to_cpu(p[3])); } static void mlx5_handle_error_cqe(struct mlx5_ib_dev *dev, struct mlx5_err_cqe *cqe, struct ib_wc *wc) { int dump = 1; switch (cqe->syndrome) { case MLX5_CQE_SYNDROME_LOCAL_LENGTH_ERR: wc->status = IB_WC_LOC_LEN_ERR; break; case MLX5_CQE_SYNDROME_LOCAL_QP_OP_ERR: wc->status = IB_WC_LOC_QP_OP_ERR; break; case MLX5_CQE_SYNDROME_LOCAL_PROT_ERR: wc->status = IB_WC_LOC_PROT_ERR; break; case MLX5_CQE_SYNDROME_WR_FLUSH_ERR: dump = 0; wc->status = IB_WC_WR_FLUSH_ERR; break; case MLX5_CQE_SYNDROME_MW_BIND_ERR: wc->status = IB_WC_MW_BIND_ERR; break; case MLX5_CQE_SYNDROME_BAD_RESP_ERR: wc->status = IB_WC_BAD_RESP_ERR; break; case MLX5_CQE_SYNDROME_LOCAL_ACCESS_ERR: wc->status = IB_WC_LOC_ACCESS_ERR; break; case MLX5_CQE_SYNDROME_REMOTE_INVAL_REQ_ERR: wc->status = IB_WC_REM_INV_REQ_ERR; break; case MLX5_CQE_SYNDROME_REMOTE_ACCESS_ERR: wc->status = IB_WC_REM_ACCESS_ERR; break; case MLX5_CQE_SYNDROME_REMOTE_OP_ERR: wc->status = IB_WC_REM_OP_ERR; break; case MLX5_CQE_SYNDROME_TRANSPORT_RETRY_EXC_ERR: wc->status = IB_WC_RETRY_EXC_ERR; dump = 0; break; case MLX5_CQE_SYNDROME_RNR_RETRY_EXC_ERR: wc->status = IB_WC_RNR_RETRY_EXC_ERR; dump = 0; break; case MLX5_CQE_SYNDROME_REMOTE_ABORTED_ERR: wc->status = IB_WC_REM_ABORT_ERR; break; default: wc->status = IB_WC_GENERAL_ERR; break; } wc->vendor_err = cqe->vendor_err_synd; if (dump) dump_cqe(dev, cqe); } static int is_atomic_response(struct mlx5_ib_qp *qp, uint16_t idx) { /* TBD: waiting decision */ return 0; } static void *mlx5_get_atomic_laddr(struct mlx5_ib_qp *qp, uint16_t idx) { struct mlx5_wqe_data_seg *dpseg; void *addr; dpseg = mlx5_get_send_wqe(qp, idx) + sizeof(struct mlx5_wqe_ctrl_seg) + sizeof(struct mlx5_wqe_raddr_seg) + sizeof(struct mlx5_wqe_atomic_seg); addr = (void *)(unsigned long)be64_to_cpu(dpseg->addr); return addr; } static void handle_atomic(struct mlx5_ib_qp *qp, struct mlx5_cqe64 *cqe64, uint16_t idx) { void *addr; int byte_count; int i; if (!is_atomic_response(qp, idx)) return; byte_count = be32_to_cpu(cqe64->byte_cnt); addr = mlx5_get_atomic_laddr(qp, idx); if (byte_count == 4) { *(uint32_t *)addr = be32_to_cpu(*((__be32 *)addr)); } else { for (i = 0; i < byte_count; i += 8) { *(uint64_t *)addr = be64_to_cpu(*((__be64 *)addr)); addr += 8; } } return; } static void handle_atomics(struct mlx5_ib_qp *qp, struct mlx5_cqe64 *cqe64, u16 tail, u16 head) { u16 idx; do { idx = tail & (qp->sq.wqe_cnt - 1); handle_atomic(qp, cqe64, idx); if (idx == head) break; tail = qp->sq.w_list[idx].next; } while (1); tail = qp->sq.w_list[idx].next; qp->sq.last_poll = tail; } static void free_cq_buf(struct mlx5_ib_dev *dev, struct mlx5_ib_cq_buf *buf) { mlx5_buf_free(dev->mdev, &buf->buf); } static void get_sig_err_item(struct mlx5_sig_err_cqe *cqe, struct ib_sig_err *item) { u16 syndrome = be16_to_cpu(cqe->syndrome); #define GUARD_ERR (1 << 13) #define APPTAG_ERR (1 << 12) #define REFTAG_ERR (1 << 11) if (syndrome & GUARD_ERR) { item->err_type = IB_SIG_BAD_GUARD; item->expected = be32_to_cpu(cqe->expected_trans_sig) >> 16; item->actual = be32_to_cpu(cqe->actual_trans_sig) >> 16; } else if (syndrome & REFTAG_ERR) { item->err_type = IB_SIG_BAD_REFTAG; item->expected = be32_to_cpu(cqe->expected_reftag); item->actual = be32_to_cpu(cqe->actual_reftag); } else if (syndrome & APPTAG_ERR) { item->err_type = IB_SIG_BAD_APPTAG; item->expected = be32_to_cpu(cqe->expected_trans_sig) & 0xffff; item->actual = be32_to_cpu(cqe->actual_trans_sig) & 0xffff; } else { pr_err("Got signature completion error with bad syndrome %04x\n", syndrome); } item->sig_err_offset = be64_to_cpu(cqe->err_offset); item->key = be32_to_cpu(cqe->mkey); } static void sw_send_comp(struct mlx5_ib_qp *qp, int num_entries, struct ib_wc *wc, int *npolled) { struct mlx5_ib_wq *wq; unsigned int cur; unsigned int idx; int np; int i; wq = &qp->sq; cur = wq->head - wq->tail; np = *npolled; if (cur == 0) return; for (i = 0; i < cur && np < num_entries; i++) { idx = wq->last_poll & (wq->wqe_cnt - 1); wc->wr_id = wq->wrid[idx]; wc->status = IB_WC_WR_FLUSH_ERR; wc->vendor_err = MLX5_CQE_SYNDROME_WR_FLUSH_ERR; wq->tail++; np++; wc->qp = &qp->ibqp; wc++; wq->last_poll = wq->w_list[idx].next; } *npolled = np; } static void sw_recv_comp(struct mlx5_ib_qp *qp, int num_entries, struct ib_wc *wc, int *npolled) { struct mlx5_ib_wq *wq; unsigned int cur; int np; int i; wq = &qp->rq; cur = wq->head - wq->tail; np = *npolled; if (cur == 0) return; for (i = 0; i < cur && np < num_entries; i++) { wc->wr_id = wq->wrid[wq->tail & (wq->wqe_cnt - 1)]; wc->status = IB_WC_WR_FLUSH_ERR; wc->vendor_err = MLX5_CQE_SYNDROME_WR_FLUSH_ERR; wq->tail++; np++; wc->qp = &qp->ibqp; wc++; } *npolled = np; } static void mlx5_ib_poll_sw_comp(struct mlx5_ib_cq *cq, int num_entries, struct ib_wc *wc, int *npolled) { struct mlx5_ib_qp *qp; *npolled = 0; /* Find uncompleted WQEs belonging to that cq and retrun mmics ones */ list_for_each_entry(qp, &cq->list_send_qp, cq_send_list) { sw_send_comp(qp, num_entries, wc + *npolled, npolled); if (*npolled >= num_entries) return; } list_for_each_entry(qp, &cq->list_recv_qp, cq_recv_list) { sw_recv_comp(qp, num_entries, wc + *npolled, npolled); if (*npolled >= num_entries) return; } } static int mlx5_poll_one(struct mlx5_ib_cq *cq, struct mlx5_ib_qp **cur_qp, struct ib_wc *wc) { struct mlx5_ib_dev *dev = to_mdev(cq->ibcq.device); struct mlx5_err_cqe *err_cqe; struct mlx5_cqe64 *cqe64; struct mlx5_core_qp *mqp; struct mlx5_ib_wq *wq; struct mlx5_sig_err_cqe *sig_err_cqe; struct mlx5_core_mr *mmkey; struct mlx5_ib_mr *mr; unsigned long flags; uint8_t opcode; uint32_t qpn; u16 wqe_ctr; void *cqe; int idx; repoll: cqe = next_cqe_sw(cq); if (!cqe) return -EAGAIN; cqe64 = (cq->mcq.cqe_sz == 64) ? cqe : cqe + 64; ++cq->mcq.cons_index; /* Make sure we read CQ entry contents after we've checked the * ownership bit. */ rmb(); opcode = cqe64->op_own >> 4; if (unlikely(opcode == MLX5_CQE_RESIZE_CQ)) { if (likely(cq->resize_buf)) { free_cq_buf(dev, &cq->buf); cq->buf = *cq->resize_buf; kfree(cq->resize_buf); cq->resize_buf = NULL; goto repoll; } else { mlx5_ib_warn(dev, "unexpected resize cqe\n"); } } qpn = ntohl(cqe64->sop_drop_qpn) & 0xffffff; if (!*cur_qp || (qpn != (*cur_qp)->ibqp.qp_num)) { /* We do not have to take the QP table lock here, * because CQs will be locked while QPs are removed * from the table. */ mqp = __mlx5_qp_lookup(dev->mdev, qpn); *cur_qp = to_mibqp(mqp); } wc->qp = &(*cur_qp)->ibqp; switch (opcode) { case MLX5_CQE_REQ: wq = &(*cur_qp)->sq; wqe_ctr = be16_to_cpu(cqe64->wqe_counter); idx = wqe_ctr & (wq->wqe_cnt - 1); handle_good_req(wc, cqe64, wq, idx); handle_atomics(*cur_qp, cqe64, wq->last_poll, idx); wc->wr_id = wq->wrid[idx]; wq->tail = wq->wqe_head[idx] + 1; wc->status = IB_WC_SUCCESS; break; case MLX5_CQE_RESP_WR_IMM: case MLX5_CQE_RESP_SEND: case MLX5_CQE_RESP_SEND_IMM: case MLX5_CQE_RESP_SEND_INV: handle_responder(wc, cqe64, *cur_qp); wc->status = IB_WC_SUCCESS; break; case MLX5_CQE_RESIZE_CQ: break; case MLX5_CQE_REQ_ERR: case MLX5_CQE_RESP_ERR: err_cqe = (struct mlx5_err_cqe *)cqe64; mlx5_handle_error_cqe(dev, err_cqe, wc); mlx5_ib_dbg(dev, "%s error cqe on cqn 0x%x:\n", opcode == MLX5_CQE_REQ_ERR ? "Requestor" : "Responder", cq->mcq.cqn); mlx5_ib_dbg(dev, "syndrome 0x%x, vendor syndrome 0x%x\n", err_cqe->syndrome, err_cqe->vendor_err_synd); if (opcode == MLX5_CQE_REQ_ERR) { wq = &(*cur_qp)->sq; wqe_ctr = be16_to_cpu(cqe64->wqe_counter); idx = wqe_ctr & (wq->wqe_cnt - 1); wc->wr_id = wq->wrid[idx]; wq->tail = wq->wqe_head[idx] + 1; } else { struct mlx5_ib_srq *srq; if ((*cur_qp)->ibqp.srq) { srq = to_msrq((*cur_qp)->ibqp.srq); wqe_ctr = be16_to_cpu(cqe64->wqe_counter); wc->wr_id = srq->wrid[wqe_ctr]; mlx5_ib_free_srq_wqe(srq, wqe_ctr); } else { wq = &(*cur_qp)->rq; wc->wr_id = wq->wrid[wq->tail & (wq->wqe_cnt - 1)]; ++wq->tail; } } break; case MLX5_CQE_SIG_ERR: sig_err_cqe = (struct mlx5_sig_err_cqe *)cqe64; spin_lock_irqsave(&dev->mdev->priv.mr_table.lock, flags); mmkey = __mlx5_mr_lookup(dev->mdev, mlx5_base_mkey(be32_to_cpu(sig_err_cqe->mkey))); mr = to_mibmr(mmkey); get_sig_err_item(sig_err_cqe, &mr->sig->err_item); mr->sig->sig_err_exists = true; mr->sig->sigerr_count++; mlx5_ib_warn(dev, "CQN: 0x%x Got SIGERR on key: 0x%x err_type %x err_offset %llx expected %x actual %x\n", cq->mcq.cqn, mr->sig->err_item.key, mr->sig->err_item.err_type, (long long)mr->sig->err_item.sig_err_offset, mr->sig->err_item.expected, mr->sig->err_item.actual); spin_unlock_irqrestore(&dev->mdev->priv.mr_table.lock, flags); goto repoll; } return 0; } static int poll_soft_wc(struct mlx5_ib_cq *cq, int num_entries, struct ib_wc *wc) { struct mlx5_ib_dev *dev = to_mdev(cq->ibcq.device); struct mlx5_ib_wc *soft_wc, *next; int npolled = 0; list_for_each_entry_safe(soft_wc, next, &cq->wc_list, list) { if (npolled >= num_entries) break; mlx5_ib_dbg(dev, "polled software generated completion on CQ 0x%x\n", cq->mcq.cqn); wc[npolled++] = soft_wc->wc; list_del(&soft_wc->list); kfree(soft_wc); } return npolled; } int mlx5_ib_poll_cq(struct ib_cq *ibcq, int num_entries, struct ib_wc *wc) { struct mlx5_ib_cq *cq = to_mcq(ibcq); struct mlx5_ib_qp *cur_qp = NULL; struct mlx5_ib_dev *dev = to_mdev(cq->ibcq.device); struct mlx5_core_dev *mdev = dev->mdev; unsigned long flags; int soft_polled = 0; int npolled; spin_lock_irqsave(&cq->lock, flags); if (unlikely(mdev->state == MLX5_DEVICE_STATE_INTERNAL_ERROR)) { mlx5_ib_poll_sw_comp(cq, num_entries, wc, &npolled); goto out; } if (unlikely(!list_empty(&cq->wc_list))) soft_polled = poll_soft_wc(cq, num_entries, wc); for (npolled = 0; npolled < num_entries - soft_polled; npolled++) { if (mlx5_poll_one(cq, &cur_qp, wc + soft_polled + npolled)) break; } if (npolled) mlx5_cq_set_ci(&cq->mcq); out: spin_unlock_irqrestore(&cq->lock, flags); return soft_polled + npolled; } int mlx5_ib_arm_cq(struct ib_cq *ibcq, enum ib_cq_notify_flags flags) { struct mlx5_core_dev *mdev = to_mdev(ibcq->device)->mdev; struct mlx5_ib_cq *cq = to_mcq(ibcq); void __iomem *uar_page = mdev->priv.uar->map; unsigned long irq_flags; int ret = 0; if (unlikely(mdev->state == MLX5_DEVICE_STATE_INTERNAL_ERROR)) return -1; spin_lock_irqsave(&cq->lock, irq_flags); if (cq->notify_flags != IB_CQ_NEXT_COMP) cq->notify_flags = flags & IB_CQ_SOLICITED_MASK; if ((flags & IB_CQ_REPORT_MISSED_EVENTS) && !list_empty(&cq->wc_list)) ret = 1; spin_unlock_irqrestore(&cq->lock, irq_flags); mlx5_cq_arm(&cq->mcq, (flags & IB_CQ_SOLICITED_MASK) == IB_CQ_SOLICITED ? MLX5_CQ_DB_REQ_NOT_SOL : MLX5_CQ_DB_REQ_NOT, uar_page, MLX5_GET_DOORBELL_LOCK(&mdev->priv.cq_uar_lock), cq->mcq.cons_index); return ret; } static int alloc_cq_buf(struct mlx5_ib_dev *dev, struct mlx5_ib_cq_buf *buf, int nent, int cqe_size) { int err; err = mlx5_buf_alloc(dev->mdev, nent * cqe_size, 2 * PAGE_SIZE, &buf->buf); if (err) return err; buf->cqe_size = cqe_size; buf->nent = nent; return 0; } static int create_cq_user(struct mlx5_ib_dev *dev, struct ib_udata *udata, struct ib_ucontext *context, struct mlx5_ib_cq *cq, int entries, u32 **cqb, int *cqe_size, int *index, int *inlen) { struct mlx5_ib_create_cq ucmd; size_t ucmdlen; int page_shift; __be64 *pas; int npages; int ncont; void *cqc; int err; ucmdlen = min(udata->inlen, sizeof(ucmd)); if (ucmdlen < offsetof(struct mlx5_ib_create_cq, flags)) return -EINVAL; if (ib_copy_from_udata(&ucmd, udata, ucmdlen)) return -EFAULT; if ((ucmd.flags & ~(MLX5_IB_CREATE_CQ_FLAGS_UAR_PAGE_INDEX))) return -EINVAL; if (ucmd.cqe_size != 64 && ucmd.cqe_size != 128) return -EINVAL; *cqe_size = ucmd.cqe_size; cq->buf.umem = ib_umem_get(context, ucmd.buf_addr, entries * ucmd.cqe_size, IB_ACCESS_LOCAL_WRITE, 1); if (IS_ERR(cq->buf.umem)) { err = PTR_ERR(cq->buf.umem); return err; } err = mlx5_ib_db_map_user(to_mucontext(context), ucmd.db_addr, &cq->db); if (err) goto err_umem; mlx5_ib_cont_pages(cq->buf.umem, ucmd.buf_addr, 0, &npages, &page_shift, &ncont, NULL); mlx5_ib_dbg(dev, "addr 0x%llx, size %u, npages %d, page_shift %d, ncont %d\n", (long long)ucmd.buf_addr, entries * ucmd.cqe_size, npages, page_shift, ncont); *inlen = MLX5_ST_SZ_BYTES(create_cq_in) + MLX5_FLD_SZ_BYTES(create_cq_in, pas[0]) * ncont; *cqb = mlx5_vzalloc(*inlen); if (!*cqb) { err = -ENOMEM; goto err_db; } pas = (__be64 *)MLX5_ADDR_OF(create_cq_in, *cqb, pas); mlx5_ib_populate_pas(dev, cq->buf.umem, page_shift, pas, 0); cqc = MLX5_ADDR_OF(create_cq_in, *cqb, cq_context); MLX5_SET(cqc, cqc, log_page_size, page_shift - MLX5_ADAPTER_PAGE_SHIFT); if (ucmd.flags & MLX5_IB_CREATE_CQ_FLAGS_UAR_PAGE_INDEX) { *index = ucmd.uar_page_index; } else if (to_mucontext(context)->bfregi.lib_uar_dyn) { err = -EINVAL; goto err_cqb; } else { *index = to_mucontext(context)->bfregi.sys_pages[0]; } return 0; err_cqb: kvfree(*cqb); err_db: mlx5_ib_db_unmap_user(to_mucontext(context), &cq->db); err_umem: ib_umem_release(cq->buf.umem); return err; } static void destroy_cq_user(struct mlx5_ib_cq *cq, struct ib_ucontext *context) { mlx5_ib_db_unmap_user(to_mucontext(context), &cq->db); ib_umem_release(cq->buf.umem); } static void init_cq_buf(struct mlx5_ib_cq *cq, struct mlx5_ib_cq_buf *buf) { int i; void *cqe; struct mlx5_cqe64 *cqe64; for (i = 0; i < buf->nent; i++) { cqe = get_cqe_from_buf(buf, i, buf->cqe_size); cqe64 = buf->cqe_size == 64 ? cqe : cqe + 64; cqe64->op_own = MLX5_CQE_INVALID << 4; } } static int create_cq_kernel(struct mlx5_ib_dev *dev, struct mlx5_ib_cq *cq, int entries, int cqe_size, u32 **cqb, int *index, int *inlen) { __be64 *pas; void *cqc; int err; err = mlx5_db_alloc(dev->mdev, &cq->db); if (err) return err; cq->mcq.set_ci_db = cq->db.db; cq->mcq.arm_db = cq->db.db + 1; cq->mcq.cqe_sz = cqe_size; err = alloc_cq_buf(dev, &cq->buf, entries, cqe_size); if (err) goto err_db; init_cq_buf(cq, &cq->buf); *inlen = MLX5_ST_SZ_BYTES(create_cq_in) + MLX5_FLD_SZ_BYTES(create_cq_in, pas[0]) * cq->buf.buf.npages; *cqb = mlx5_vzalloc(*inlen); if (!*cqb) { err = -ENOMEM; goto err_buf; } pas = (__be64 *)MLX5_ADDR_OF(create_cq_in, *cqb, pas); mlx5_fill_page_array(&cq->buf.buf, pas); cqc = MLX5_ADDR_OF(create_cq_in, *cqb, cq_context); MLX5_SET(cqc, cqc, log_page_size, cq->buf.buf.page_shift - MLX5_ADAPTER_PAGE_SHIFT); *index = dev->mdev->priv.uar->index; return 0; err_buf: free_cq_buf(dev, &cq->buf); err_db: mlx5_db_free(dev->mdev, &cq->db); return err; } static void destroy_cq_kernel(struct mlx5_ib_dev *dev, struct mlx5_ib_cq *cq) { free_cq_buf(dev, &cq->buf); mlx5_db_free(dev->mdev, &cq->db); } static void notify_soft_wc_handler(struct work_struct *work) { struct mlx5_ib_cq *cq = container_of(work, struct mlx5_ib_cq, notify_work); cq->ibcq.comp_handler(&cq->ibcq, cq->ibcq.cq_context); } struct ib_cq *mlx5_ib_create_cq(struct ib_device *ibdev, const struct ib_cq_init_attr *attr, struct ib_ucontext *context, struct ib_udata *udata) { int entries = attr->cqe; int vector = attr->comp_vector; struct mlx5_ib_dev *dev = to_mdev(ibdev); u32 out[MLX5_ST_SZ_DW(create_cq_out)]; struct mlx5_ib_cq *cq; int uninitialized_var(index); int uninitialized_var(inlen); u32 *cqb = NULL; void *cqc; int cqe_size; unsigned int irqn; int eqn; int err; if (entries < 0 || (entries > (1 << MLX5_CAP_GEN(dev->mdev, log_max_cq_sz)))) return ERR_PTR(-EINVAL); if (check_cq_create_flags(attr->flags)) return ERR_PTR(-EOPNOTSUPP); entries = roundup_pow_of_two(entries + 1); if (entries > (1 << MLX5_CAP_GEN(dev->mdev, log_max_cq_sz))) return ERR_PTR(-EINVAL); cq = kzalloc(sizeof(*cq), GFP_KERNEL); if (!cq) return ERR_PTR(-ENOMEM); cq->ibcq.cqe = entries - 1; mutex_init(&cq->resize_mutex); spin_lock_init(&cq->lock); cq->resize_buf = NULL; cq->resize_umem = NULL; cq->create_flags = attr->flags; INIT_LIST_HEAD(&cq->list_send_qp); INIT_LIST_HEAD(&cq->list_recv_qp); if (context) { err = create_cq_user(dev, udata, context, cq, entries, &cqb, &cqe_size, &index, &inlen); if (err) goto err_create; } else { cqe_size = cache_line_size() == 128 ? 128 : 64; err = create_cq_kernel(dev, cq, entries, cqe_size, &cqb, &index, &inlen); if (err) goto err_create; INIT_WORK(&cq->notify_work, notify_soft_wc_handler); } err = mlx5_vector2eqn(dev->mdev, vector, &eqn, &irqn); if (err) goto err_cqb; cq->cqe_size = cqe_size; cqc = MLX5_ADDR_OF(create_cq_in, cqb, cq_context); MLX5_SET(cqc, cqc, cqe_sz, cqe_sz_to_mlx_sz(cqe_size)); MLX5_SET(cqc, cqc, log_cq_size, ilog2(entries)); MLX5_SET(cqc, cqc, uar_page, index); MLX5_SET(cqc, cqc, c_eqn, eqn); MLX5_SET64(cqc, cqc, dbr_addr, cq->db.dma); if (cq->create_flags & IB_CQ_FLAGS_IGNORE_OVERRUN) MLX5_SET(cqc, cqc, oi, 1); err = mlx5_core_create_cq(dev->mdev, &cq->mcq, cqb, inlen, out, sizeof(out)); if (err) goto err_cqb; mlx5_ib_dbg(dev, "cqn 0x%x\n", cq->mcq.cqn); cq->mcq.irqn = irqn; cq->mcq.comp = mlx5_ib_cq_comp; cq->mcq.event = mlx5_ib_cq_event; INIT_LIST_HEAD(&cq->wc_list); if (context) if (ib_copy_to_udata(udata, &cq->mcq.cqn, sizeof(__u32))) { err = -EFAULT; goto err_cmd; } kvfree(cqb); return &cq->ibcq; err_cmd: mlx5_core_destroy_cq(dev->mdev, &cq->mcq); err_cqb: kvfree(cqb); if (context) destroy_cq_user(cq, context); else destroy_cq_kernel(dev, cq); err_create: kfree(cq); return ERR_PTR(err); } int mlx5_ib_destroy_cq(struct ib_cq *cq) { struct mlx5_ib_dev *dev = to_mdev(cq->device); struct mlx5_ib_cq *mcq = to_mcq(cq); struct ib_ucontext *context = NULL; if (cq->uobject) context = cq->uobject->context; mlx5_core_destroy_cq(dev->mdev, &mcq->mcq); if (context) destroy_cq_user(mcq, context); else destroy_cq_kernel(dev, mcq); kfree(mcq); return 0; } static int is_equal_rsn(struct mlx5_cqe64 *cqe64, u32 rsn) { return rsn == (ntohl(cqe64->sop_drop_qpn) & 0xffffff); } void __mlx5_ib_cq_clean(struct mlx5_ib_cq *cq, u32 rsn, struct mlx5_ib_srq *srq) { struct mlx5_cqe64 *cqe64, *dest64; void *cqe, *dest; u32 prod_index; int nfreed = 0; u8 owner_bit; if (!cq) return; /* First we need to find the current producer index, so we * know where to start cleaning from. It doesn't matter if HW * adds new entries after this loop -- the QP we're worried * about is already in RESET, so the new entries won't come * from our QP and therefore don't need to be checked. */ for (prod_index = cq->mcq.cons_index; get_sw_cqe(cq, prod_index); prod_index++) if (prod_index == cq->mcq.cons_index + cq->ibcq.cqe) break; /* Now sweep backwards through the CQ, removing CQ entries * that match our QP by copying older entries on top of them. */ while ((int) --prod_index - (int) cq->mcq.cons_index >= 0) { cqe = get_cqe(cq, prod_index & cq->ibcq.cqe); cqe64 = (cq->mcq.cqe_sz == 64) ? cqe : cqe + 64; if (is_equal_rsn(cqe64, rsn)) { if (srq && (ntohl(cqe64->srqn) & 0xffffff)) mlx5_ib_free_srq_wqe(srq, be16_to_cpu(cqe64->wqe_counter)); ++nfreed; } else if (nfreed) { dest = get_cqe(cq, (prod_index + nfreed) & cq->ibcq.cqe); dest64 = (cq->mcq.cqe_sz == 64) ? dest : dest + 64; owner_bit = dest64->op_own & MLX5_CQE_OWNER_MASK; memcpy(dest, cqe, cq->mcq.cqe_sz); dest64->op_own = owner_bit | (dest64->op_own & ~MLX5_CQE_OWNER_MASK); } } if (nfreed) { cq->mcq.cons_index += nfreed; /* Make sure update of buffer contents is done before * updating consumer index. */ wmb(); mlx5_cq_set_ci(&cq->mcq); } } void mlx5_ib_cq_clean(struct mlx5_ib_cq *cq, u32 qpn, struct mlx5_ib_srq *srq) { if (!cq) return; spin_lock_irq(&cq->lock); __mlx5_ib_cq_clean(cq, qpn, srq); spin_unlock_irq(&cq->lock); } int mlx5_ib_modify_cq(struct ib_cq *cq, u16 cq_count, u16 cq_period) { struct mlx5_ib_dev *dev = to_mdev(cq->device); struct mlx5_ib_cq *mcq = to_mcq(cq); int err; if (!MLX5_CAP_GEN(dev->mdev, cq_moderation)) return -ENOSYS; err = mlx5_core_modify_cq_moderation(dev->mdev, &mcq->mcq, cq_period, cq_count); if (err) mlx5_ib_warn(dev, "modify cq 0x%x failed\n", mcq->mcq.cqn); return err; } static int resize_user(struct mlx5_ib_dev *dev, struct mlx5_ib_cq *cq, int entries, struct ib_udata *udata, int *npas, int *page_shift, int *cqe_size) { struct mlx5_ib_resize_cq ucmd; struct ib_umem *umem; int err; int npages; struct ib_ucontext *context = cq->buf.umem->context; err = ib_copy_from_udata(&ucmd, udata, sizeof(ucmd)); if (err) return err; if (ucmd.reserved0 || ucmd.reserved1) return -EINVAL; /* check multiplication overflow */ if (ucmd.cqe_size && SIZE_MAX / ucmd.cqe_size <= entries - 1) return -EINVAL; umem = ib_umem_get(context, ucmd.buf_addr, (size_t)ucmd.cqe_size * entries, IB_ACCESS_LOCAL_WRITE, 1); if (IS_ERR(umem)) { err = PTR_ERR(umem); return err; } mlx5_ib_cont_pages(umem, ucmd.buf_addr, 0, &npages, page_shift, npas, NULL); cq->resize_umem = umem; *cqe_size = ucmd.cqe_size; return 0; } static void un_resize_user(struct mlx5_ib_cq *cq) { ib_umem_release(cq->resize_umem); } static int resize_kernel(struct mlx5_ib_dev *dev, struct mlx5_ib_cq *cq, int entries, int cqe_size) { int err; cq->resize_buf = kzalloc(sizeof(*cq->resize_buf), GFP_KERNEL); if (!cq->resize_buf) return -ENOMEM; err = alloc_cq_buf(dev, cq->resize_buf, entries, cqe_size); if (err) goto ex; init_cq_buf(cq, cq->resize_buf); return 0; ex: kfree(cq->resize_buf); return err; } static void un_resize_kernel(struct mlx5_ib_dev *dev, struct mlx5_ib_cq *cq) { free_cq_buf(dev, cq->resize_buf); cq->resize_buf = NULL; } static int copy_resize_cqes(struct mlx5_ib_cq *cq) { struct mlx5_ib_dev *dev = to_mdev(cq->ibcq.device); struct mlx5_cqe64 *scqe64; struct mlx5_cqe64 *dcqe64; void *start_cqe; void *scqe; void *dcqe; int ssize; int dsize; int i; u8 sw_own; ssize = cq->buf.cqe_size; dsize = cq->resize_buf->cqe_size; if (ssize != dsize) { mlx5_ib_warn(dev, "resize from different cqe size is not supported\n"); return -EINVAL; } i = cq->mcq.cons_index; scqe = get_sw_cqe(cq, i); scqe64 = ssize == 64 ? scqe : scqe + 64; start_cqe = scqe; if (!scqe) { mlx5_ib_warn(dev, "expected cqe in sw ownership\n"); return -EINVAL; } while ((scqe64->op_own >> 4) != MLX5_CQE_RESIZE_CQ) { dcqe = get_cqe_from_buf(cq->resize_buf, (i + 1) & (cq->resize_buf->nent), dsize); dcqe64 = dsize == 64 ? dcqe : dcqe + 64; sw_own = sw_ownership_bit(i + 1, cq->resize_buf->nent); memcpy(dcqe, scqe, dsize); dcqe64->op_own = (dcqe64->op_own & ~MLX5_CQE_OWNER_MASK) | sw_own; ++i; scqe = get_sw_cqe(cq, i); scqe64 = ssize == 64 ? scqe : scqe + 64; if (!scqe) { mlx5_ib_warn(dev, "expected cqe in sw ownership\n"); return -EINVAL; } if (scqe == start_cqe) { pr_warn("resize CQ failed to get resize CQE, CQN 0x%x\n", cq->mcq.cqn); return -ENOMEM; } } ++cq->mcq.cons_index; return 0; } int mlx5_ib_resize_cq(struct ib_cq *ibcq, int entries, struct ib_udata *udata) { struct mlx5_ib_dev *dev = to_mdev(ibcq->device); struct mlx5_ib_cq *cq = to_mcq(ibcq); void *cqc; u32 *in; int err; int npas; __be64 *pas; int page_shift; int inlen; int uninitialized_var(cqe_size); unsigned long flags; if (!MLX5_CAP_GEN(dev->mdev, cq_resize)) { pr_info("Firmware does not support resize CQ\n"); return -ENOSYS; } if (entries < 1 || entries > (1 << MLX5_CAP_GEN(dev->mdev, log_max_cq_sz))) { mlx5_ib_warn(dev, "wrong entries number %d, max %d\n", entries, 1 << MLX5_CAP_GEN(dev->mdev, log_max_cq_sz)); return -EINVAL; } entries = roundup_pow_of_two(entries + 1); if (entries > (1 << MLX5_CAP_GEN(dev->mdev, log_max_cq_sz)) + 1) return -EINVAL; if (entries == ibcq->cqe + 1) return 0; mutex_lock(&cq->resize_mutex); if (udata) { err = resize_user(dev, cq, entries, udata, &npas, &page_shift, &cqe_size); } else { cqe_size = 64; err = resize_kernel(dev, cq, entries, cqe_size); if (!err) { npas = cq->resize_buf->buf.npages; page_shift = cq->resize_buf->buf.page_shift; } } if (err) goto ex; inlen = MLX5_ST_SZ_BYTES(modify_cq_in) + MLX5_FLD_SZ_BYTES(modify_cq_in, pas[0]) * npas; in = mlx5_vzalloc(inlen); if (!in) { err = -ENOMEM; goto ex_resize; } pas = (__be64 *)MLX5_ADDR_OF(modify_cq_in, in, pas); if (udata) mlx5_ib_populate_pas(dev, cq->resize_umem, page_shift, pas, 0); else mlx5_fill_page_array(&cq->resize_buf->buf, pas); MLX5_SET(modify_cq_in, in, modify_field_select_resize_field_select.resize_field_select.resize_field_select, MLX5_MODIFY_CQ_MASK_LOG_SIZE | MLX5_MODIFY_CQ_MASK_PG_OFFSET | MLX5_MODIFY_CQ_MASK_PG_SIZE); cqc = MLX5_ADDR_OF(modify_cq_in, in, cq_context); MLX5_SET(cqc, cqc, log_page_size, page_shift - MLX5_ADAPTER_PAGE_SHIFT); MLX5_SET(cqc, cqc, cqe_sz, cqe_sz_to_mlx_sz(cqe_size)); MLX5_SET(cqc, cqc, log_cq_size, ilog2(entries)); MLX5_SET(modify_cq_in, in, op_mod, MLX5_CQ_OPMOD_RESIZE); MLX5_SET(modify_cq_in, in, cqn, cq->mcq.cqn); err = mlx5_core_modify_cq(dev->mdev, &cq->mcq, in, inlen); if (err) goto ex_alloc; if (udata) { cq->ibcq.cqe = entries - 1; ib_umem_release(cq->buf.umem); cq->buf.umem = cq->resize_umem; cq->resize_umem = NULL; } else { struct mlx5_ib_cq_buf tbuf; int resized = 0; spin_lock_irqsave(&cq->lock, flags); if (cq->resize_buf) { err = copy_resize_cqes(cq); if (!err) { tbuf = cq->buf; cq->buf = *cq->resize_buf; kfree(cq->resize_buf); cq->resize_buf = NULL; resized = 1; } } cq->ibcq.cqe = entries - 1; spin_unlock_irqrestore(&cq->lock, flags); if (resized) free_cq_buf(dev, &tbuf); } mutex_unlock(&cq->resize_mutex); kvfree(in); return 0; ex_alloc: kvfree(in); ex_resize: if (udata) un_resize_user(cq); else un_resize_kernel(dev, cq); ex: mutex_unlock(&cq->resize_mutex); return err; } int mlx5_ib_get_cqe_size(struct mlx5_ib_dev *dev, struct ib_cq *ibcq) { struct mlx5_ib_cq *cq; if (!ibcq) return 128; cq = to_mcq(ibcq); return cq->cqe_size; } /* Called from atomic context */ int mlx5_ib_generate_wc(struct ib_cq *ibcq, struct ib_wc *wc) { struct mlx5_ib_wc *soft_wc; struct mlx5_ib_cq *cq = to_mcq(ibcq); unsigned long flags; soft_wc = kmalloc(sizeof(*soft_wc), GFP_ATOMIC); if (!soft_wc) return -ENOMEM; soft_wc->wc = *wc; spin_lock_irqsave(&cq->lock, flags); list_add_tail(&soft_wc->list, &cq->wc_list); if (cq->notify_flags == IB_CQ_NEXT_COMP || wc->status != IB_WC_SUCCESS) { cq->notify_flags = 0; schedule_work(&cq->notify_work); } spin_unlock_irqrestore(&cq->lock, flags); return 0; } diff --git a/sys/dev/mlx5/mlx5_ib/mlx5_ib_doorbell.c b/sys/dev/mlx5/mlx5_ib/mlx5_ib_doorbell.c index 04579fbc2346..5e7c4d1e914b 100644 --- a/sys/dev/mlx5/mlx5_ib/mlx5_ib_doorbell.c +++ b/sys/dev/mlx5/mlx5_ib/mlx5_ib_doorbell.c @@ -1,96 +1,96 @@ /*- * Copyright (c) 2013-2015, Mellanox Technologies, Ltd. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY AUTHOR AND CONTRIBUTORS `AS IS' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $FreeBSD$ */ #include "opt_rss.h" #include "opt_ratelimit.h" #include #include #include -#include "mlx5_ib.h" +#include struct mlx5_ib_user_db_page { struct list_head list; struct ib_umem *umem; unsigned long user_virt; int refcnt; }; int mlx5_ib_db_map_user(struct mlx5_ib_ucontext *context, unsigned long virt, struct mlx5_db *db) { struct mlx5_ib_user_db_page *page; int err = 0; mutex_lock(&context->db_page_mutex); list_for_each_entry(page, &context->db_page_list, list) if (page->user_virt == (virt & PAGE_MASK)) goto found; page = kmalloc(sizeof(*page), GFP_KERNEL); if (!page) { err = -ENOMEM; goto out; } page->user_virt = (virt & PAGE_MASK); page->refcnt = 0; page->umem = ib_umem_get(&context->ibucontext, virt & PAGE_MASK, PAGE_SIZE, 0, 0); if (IS_ERR(page->umem)) { err = PTR_ERR(page->umem); kfree(page); goto out; } list_add(&page->list, &context->db_page_list); found: db->dma = sg_dma_address(page->umem->sg_head.sgl) + (virt & ~PAGE_MASK); db->u.user_page = page; ++page->refcnt; out: mutex_unlock(&context->db_page_mutex); return err; } void mlx5_ib_db_unmap_user(struct mlx5_ib_ucontext *context, struct mlx5_db *db) { mutex_lock(&context->db_page_mutex); if (!--db->u.user_page->refcnt) { list_del(&db->u.user_page->list); ib_umem_release(db->u.user_page->umem); kfree(db->u.user_page); } mutex_unlock(&context->db_page_mutex); } diff --git a/sys/dev/mlx5/mlx5_ib/mlx5_ib_gsi.c b/sys/dev/mlx5/mlx5_ib/mlx5_ib_gsi.c index cb9b5cfabf48..dc273bf5d179 100644 --- a/sys/dev/mlx5/mlx5_ib/mlx5_ib_gsi.c +++ b/sys/dev/mlx5/mlx5_ib/mlx5_ib_gsi.c @@ -1,539 +1,539 @@ /*- * Copyright (c) 2016, Mellanox Technologies, Ltd. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY AUTHOR AND CONTRIBUTORS `AS IS' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $FreeBSD$ */ #include "opt_rss.h" #include "opt_ratelimit.h" -#include "mlx5_ib.h" +#include struct mlx5_ib_gsi_wr { struct ib_cqe cqe; struct ib_wc wc; int send_flags; bool completed:1; }; struct mlx5_ib_gsi_qp { struct ib_qp ibqp; struct ib_qp *rx_qp; u8 port_num; struct ib_qp_cap cap; enum ib_sig_type sq_sig_type; /* Serialize qp state modifications */ struct mutex mutex; struct ib_cq *cq; struct mlx5_ib_gsi_wr *outstanding_wrs; u32 outstanding_pi, outstanding_ci; int num_qps; /* Protects access to the tx_qps. Post send operations synchronize * with tx_qp creation in setup_qp(). Also protects the * outstanding_wrs array and indices. */ spinlock_t lock; struct ib_qp **tx_qps; }; static struct mlx5_ib_gsi_qp *gsi_qp(struct ib_qp *qp) { return container_of(qp, struct mlx5_ib_gsi_qp, ibqp); } static bool mlx5_ib_deth_sqpn_cap(struct mlx5_ib_dev *dev) { return MLX5_CAP_GEN(dev->mdev, set_deth_sqpn); } /* Call with gsi->lock locked */ static void generate_completions(struct mlx5_ib_gsi_qp *gsi) { struct ib_cq *gsi_cq = gsi->ibqp.send_cq; struct mlx5_ib_gsi_wr *wr; u32 index; for (index = gsi->outstanding_ci; index != gsi->outstanding_pi; index++) { wr = &gsi->outstanding_wrs[index % gsi->cap.max_send_wr]; if (!wr->completed) break; if (gsi->sq_sig_type == IB_SIGNAL_ALL_WR || wr->send_flags & IB_SEND_SIGNALED) WARN_ON_ONCE(mlx5_ib_generate_wc(gsi_cq, &wr->wc)); wr->completed = false; } gsi->outstanding_ci = index; } static void handle_single_completion(struct ib_cq *cq, struct ib_wc *wc) { struct mlx5_ib_gsi_qp *gsi = cq->cq_context; struct mlx5_ib_gsi_wr *wr = container_of(wc->wr_cqe, struct mlx5_ib_gsi_wr, cqe); u64 wr_id; unsigned long flags; spin_lock_irqsave(&gsi->lock, flags); wr->completed = true; wr_id = wr->wc.wr_id; wr->wc = *wc; wr->wc.wr_id = wr_id; wr->wc.qp = &gsi->ibqp; generate_completions(gsi); spin_unlock_irqrestore(&gsi->lock, flags); } struct ib_qp *mlx5_ib_gsi_create_qp(struct ib_pd *pd, struct ib_qp_init_attr *init_attr) { struct mlx5_ib_dev *dev = to_mdev(pd->device); struct mlx5_ib_gsi_qp *gsi; struct ib_qp_init_attr hw_init_attr = *init_attr; const u8 port_num = init_attr->port_num; const int num_pkeys = pd->device->attrs.max_pkeys; const int num_qps = mlx5_ib_deth_sqpn_cap(dev) ? num_pkeys : 0; int ret; mlx5_ib_dbg(dev, "creating GSI QP\n"); if (port_num > ARRAY_SIZE(dev->devr.ports) || port_num < 1) { mlx5_ib_warn(dev, "invalid port number %d during GSI QP creation\n", port_num); return ERR_PTR(-EINVAL); } gsi = kzalloc(sizeof(*gsi), GFP_KERNEL); if (!gsi) return ERR_PTR(-ENOMEM); gsi->tx_qps = kcalloc(num_qps, sizeof(*gsi->tx_qps), GFP_KERNEL); if (!gsi->tx_qps) { ret = -ENOMEM; goto err_free; } gsi->outstanding_wrs = kcalloc(init_attr->cap.max_send_wr, sizeof(*gsi->outstanding_wrs), GFP_KERNEL); if (!gsi->outstanding_wrs) { ret = -ENOMEM; goto err_free_tx; } mutex_init(&gsi->mutex); mutex_lock(&dev->devr.mutex); if (dev->devr.ports[port_num - 1].gsi) { mlx5_ib_warn(dev, "GSI QP already exists on port %d\n", port_num); ret = -EBUSY; goto err_free_wrs; } gsi->num_qps = num_qps; spin_lock_init(&gsi->lock); gsi->cap = init_attr->cap; gsi->sq_sig_type = init_attr->sq_sig_type; gsi->ibqp.qp_num = 1; gsi->port_num = port_num; gsi->cq = ib_alloc_cq(pd->device, gsi, init_attr->cap.max_send_wr, 0, IB_POLL_SOFTIRQ); if (IS_ERR(gsi->cq)) { mlx5_ib_warn(dev, "unable to create send CQ for GSI QP. error %ld\n", PTR_ERR(gsi->cq)); ret = PTR_ERR(gsi->cq); goto err_free_wrs; } hw_init_attr.qp_type = MLX5_IB_QPT_HW_GSI; hw_init_attr.send_cq = gsi->cq; if (num_qps) { hw_init_attr.cap.max_send_wr = 0; hw_init_attr.cap.max_send_sge = 0; hw_init_attr.cap.max_inline_data = 0; } gsi->rx_qp = ib_create_qp(pd, &hw_init_attr); if (IS_ERR(gsi->rx_qp)) { mlx5_ib_warn(dev, "unable to create hardware GSI QP. error %ld\n", PTR_ERR(gsi->rx_qp)); ret = PTR_ERR(gsi->rx_qp); goto err_destroy_cq; } dev->devr.ports[init_attr->port_num - 1].gsi = gsi; mutex_unlock(&dev->devr.mutex); return &gsi->ibqp; err_destroy_cq: ib_free_cq(gsi->cq); err_free_wrs: mutex_unlock(&dev->devr.mutex); kfree(gsi->outstanding_wrs); err_free_tx: kfree(gsi->tx_qps); err_free: kfree(gsi); return ERR_PTR(ret); } int mlx5_ib_gsi_destroy_qp(struct ib_qp *qp) { struct mlx5_ib_dev *dev = to_mdev(qp->device); struct mlx5_ib_gsi_qp *gsi = gsi_qp(qp); const int port_num = gsi->port_num; int qp_index; int ret; mlx5_ib_dbg(dev, "destroying GSI QP\n"); mutex_lock(&dev->devr.mutex); ret = ib_destroy_qp(gsi->rx_qp); if (ret) { mlx5_ib_warn(dev, "unable to destroy hardware GSI QP. error %d\n", ret); mutex_unlock(&dev->devr.mutex); return ret; } dev->devr.ports[port_num - 1].gsi = NULL; mutex_unlock(&dev->devr.mutex); gsi->rx_qp = NULL; for (qp_index = 0; qp_index < gsi->num_qps; ++qp_index) { if (!gsi->tx_qps[qp_index]) continue; WARN_ON_ONCE(ib_destroy_qp(gsi->tx_qps[qp_index])); gsi->tx_qps[qp_index] = NULL; } ib_free_cq(gsi->cq); kfree(gsi->outstanding_wrs); kfree(gsi->tx_qps); kfree(gsi); return 0; } static struct ib_qp *create_gsi_ud_qp(struct mlx5_ib_gsi_qp *gsi) { struct ib_pd *pd = gsi->rx_qp->pd; struct ib_qp_init_attr init_attr = { .event_handler = gsi->rx_qp->event_handler, .qp_context = gsi->rx_qp->qp_context, .send_cq = gsi->cq, .recv_cq = gsi->rx_qp->recv_cq, .cap = { .max_send_wr = gsi->cap.max_send_wr, .max_send_sge = gsi->cap.max_send_sge, .max_inline_data = gsi->cap.max_inline_data, }, .sq_sig_type = gsi->sq_sig_type, .qp_type = IB_QPT_UD, .create_flags = MLX5_IB_QP_CREATE_SQPN_QP1, }; return ib_create_qp(pd, &init_attr); } static int modify_to_rts(struct mlx5_ib_gsi_qp *gsi, struct ib_qp *qp, u16 qp_index) { struct mlx5_ib_dev *dev = to_mdev(qp->device); struct ib_qp_attr attr; int mask; int ret; mask = IB_QP_STATE | IB_QP_PKEY_INDEX | IB_QP_QKEY | IB_QP_PORT; attr.qp_state = IB_QPS_INIT; attr.pkey_index = qp_index; attr.qkey = IB_QP1_QKEY; attr.port_num = gsi->port_num; ret = ib_modify_qp(qp, &attr, mask); if (ret) { mlx5_ib_err(dev, "could not change QP%d state to INIT: %d\n", qp->qp_num, ret); return ret; } attr.qp_state = IB_QPS_RTR; ret = ib_modify_qp(qp, &attr, IB_QP_STATE); if (ret) { mlx5_ib_err(dev, "could not change QP%d state to RTR: %d\n", qp->qp_num, ret); return ret; } attr.qp_state = IB_QPS_RTS; attr.sq_psn = 0; ret = ib_modify_qp(qp, &attr, IB_QP_STATE | IB_QP_SQ_PSN); if (ret) { mlx5_ib_err(dev, "could not change QP%d state to RTS: %d\n", qp->qp_num, ret); return ret; } return 0; } static void setup_qp(struct mlx5_ib_gsi_qp *gsi, u16 qp_index) { struct ib_device *device = gsi->rx_qp->device; struct mlx5_ib_dev *dev = to_mdev(device); struct ib_qp *qp; unsigned long flags; u16 pkey; int ret; ret = ib_query_pkey(device, gsi->port_num, qp_index, &pkey); if (ret) { mlx5_ib_warn(dev, "unable to read P_Key at port %d, index %d\n", gsi->port_num, qp_index); return; } if (!pkey) { mlx5_ib_dbg(dev, "invalid P_Key at port %d, index %d. Skipping.\n", gsi->port_num, qp_index); return; } spin_lock_irqsave(&gsi->lock, flags); qp = gsi->tx_qps[qp_index]; spin_unlock_irqrestore(&gsi->lock, flags); if (qp) { mlx5_ib_dbg(dev, "already existing GSI TX QP at port %d, index %d. Skipping\n", gsi->port_num, qp_index); return; } qp = create_gsi_ud_qp(gsi); if (IS_ERR(qp)) { mlx5_ib_warn(dev, "unable to create hardware UD QP for GSI: %ld\n", PTR_ERR(qp)); return; } ret = modify_to_rts(gsi, qp, qp_index); if (ret) goto err_destroy_qp; spin_lock_irqsave(&gsi->lock, flags); WARN_ON_ONCE(gsi->tx_qps[qp_index]); gsi->tx_qps[qp_index] = qp; spin_unlock_irqrestore(&gsi->lock, flags); return; err_destroy_qp: WARN_ON_ONCE(qp); } static void setup_qps(struct mlx5_ib_gsi_qp *gsi) { u16 qp_index; for (qp_index = 0; qp_index < gsi->num_qps; ++qp_index) setup_qp(gsi, qp_index); } int mlx5_ib_gsi_modify_qp(struct ib_qp *qp, struct ib_qp_attr *attr, int attr_mask) { struct mlx5_ib_dev *dev = to_mdev(qp->device); struct mlx5_ib_gsi_qp *gsi = gsi_qp(qp); int ret; mlx5_ib_dbg(dev, "modifying GSI QP to state %d\n", attr->qp_state); mutex_lock(&gsi->mutex); ret = ib_modify_qp(gsi->rx_qp, attr, attr_mask); if (ret) { mlx5_ib_warn(dev, "unable to modify GSI rx QP: %d\n", ret); goto unlock; } if (to_mqp(gsi->rx_qp)->state == IB_QPS_RTS) setup_qps(gsi); unlock: mutex_unlock(&gsi->mutex); return ret; } int mlx5_ib_gsi_query_qp(struct ib_qp *qp, struct ib_qp_attr *qp_attr, int qp_attr_mask, struct ib_qp_init_attr *qp_init_attr) { struct mlx5_ib_gsi_qp *gsi = gsi_qp(qp); int ret; mutex_lock(&gsi->mutex); ret = ib_query_qp(gsi->rx_qp, qp_attr, qp_attr_mask, qp_init_attr); qp_init_attr->cap = gsi->cap; mutex_unlock(&gsi->mutex); return ret; } /* Call with gsi->lock locked */ static int mlx5_ib_add_outstanding_wr(struct mlx5_ib_gsi_qp *gsi, struct ib_ud_wr *wr, struct ib_wc *wc) { struct mlx5_ib_dev *dev = to_mdev(gsi->rx_qp->device); struct mlx5_ib_gsi_wr *gsi_wr; if (gsi->outstanding_pi == gsi->outstanding_ci + gsi->cap.max_send_wr) { mlx5_ib_warn(dev, "no available GSI work request.\n"); return -ENOMEM; } gsi_wr = &gsi->outstanding_wrs[gsi->outstanding_pi % gsi->cap.max_send_wr]; gsi->outstanding_pi++; if (!wc) { memset(&gsi_wr->wc, 0, sizeof(gsi_wr->wc)); gsi_wr->wc.pkey_index = wr->pkey_index; gsi_wr->wc.wr_id = wr->wr.wr_id; } else { gsi_wr->wc = *wc; gsi_wr->completed = true; } gsi_wr->cqe.done = &handle_single_completion; wr->wr.wr_cqe = &gsi_wr->cqe; return 0; } /* Call with gsi->lock locked */ static int mlx5_ib_gsi_silent_drop(struct mlx5_ib_gsi_qp *gsi, struct ib_ud_wr *wr) { struct ib_wc wc = { { .wr_id = wr->wr.wr_id }, .status = IB_WC_SUCCESS, .opcode = IB_WC_SEND, .qp = &gsi->ibqp, }; int ret; ret = mlx5_ib_add_outstanding_wr(gsi, wr, &wc); if (ret) return ret; generate_completions(gsi); return 0; } /* Call with gsi->lock locked */ static struct ib_qp *get_tx_qp(struct mlx5_ib_gsi_qp *gsi, struct ib_ud_wr *wr) { struct mlx5_ib_dev *dev = to_mdev(gsi->rx_qp->device); int qp_index = wr->pkey_index; if (!mlx5_ib_deth_sqpn_cap(dev)) return gsi->rx_qp; if (qp_index >= gsi->num_qps) return NULL; return gsi->tx_qps[qp_index]; } int mlx5_ib_gsi_post_send(struct ib_qp *qp, const struct ib_send_wr *wr, const struct ib_send_wr **bad_wr) { struct mlx5_ib_gsi_qp *gsi = gsi_qp(qp); struct ib_qp *tx_qp; unsigned long flags; int ret; for (; wr; wr = wr->next) { struct ib_ud_wr cur_wr = *ud_wr(wr); cur_wr.wr.next = NULL; spin_lock_irqsave(&gsi->lock, flags); tx_qp = get_tx_qp(gsi, &cur_wr); if (!tx_qp) { ret = mlx5_ib_gsi_silent_drop(gsi, &cur_wr); if (ret) goto err; spin_unlock_irqrestore(&gsi->lock, flags); continue; } ret = mlx5_ib_add_outstanding_wr(gsi, &cur_wr, NULL); if (ret) goto err; ret = ib_post_send(tx_qp, &cur_wr.wr, bad_wr); if (ret) { /* Undo the effect of adding the outstanding wr */ gsi->outstanding_pi = (gsi->outstanding_pi - 1) % gsi->cap.max_send_wr; goto err; } spin_unlock_irqrestore(&gsi->lock, flags); } return 0; err: spin_unlock_irqrestore(&gsi->lock, flags); *bad_wr = wr; return ret; } int mlx5_ib_gsi_post_recv(struct ib_qp *qp, const struct ib_recv_wr *wr, const struct ib_recv_wr **bad_wr) { struct mlx5_ib_gsi_qp *gsi = gsi_qp(qp); return ib_post_recv(gsi->rx_qp, wr, bad_wr); } void mlx5_ib_gsi_pkey_change(struct mlx5_ib_gsi_qp *gsi) { if (!gsi) return; mutex_lock(&gsi->mutex); setup_qps(gsi); mutex_unlock(&gsi->mutex); } diff --git a/sys/dev/mlx5/mlx5_ib/mlx5_ib_mad.c b/sys/dev/mlx5/mlx5_ib/mlx5_ib_mad.c index 6abca62ebe47..155f6b88e675 100644 --- a/sys/dev/mlx5/mlx5_ib/mlx5_ib_mad.c +++ b/sys/dev/mlx5/mlx5_ib/mlx5_ib_mad.c @@ -1,585 +1,585 @@ /*- * Copyright (c) 2013-2015, Mellanox Technologies, Ltd. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY AUTHOR AND CONTRIBUTORS `AS IS' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $FreeBSD$ */ #include "opt_rss.h" #include "opt_ratelimit.h" #include #include #include #include -#include "mlx5_ib.h" +#include enum { MLX5_IB_VENDOR_CLASS1 = 0x9, MLX5_IB_VENDOR_CLASS2 = 0xa }; int mlx5_MAD_IFC(struct mlx5_ib_dev *dev, int ignore_mkey, int ignore_bkey, u8 port, const struct ib_wc *in_wc, const struct ib_grh *in_grh, const void *in_mad, void *response_mad) { u8 op_modifier = 0; /* Key check traps can't be generated unless we have in_wc to * tell us where to send the trap. */ if (ignore_mkey || !in_wc) op_modifier |= 0x1; if (ignore_bkey || !in_wc) op_modifier |= 0x2; return mlx5_core_mad_ifc(dev->mdev, in_mad, response_mad, op_modifier, port); } static int process_mad(struct ib_device *ibdev, int mad_flags, u8 port_num, const struct ib_wc *in_wc, const struct ib_grh *in_grh, const struct ib_mad *in_mad, struct ib_mad *out_mad) { u16 slid; int err; slid = in_wc ? in_wc->slid : be16_to_cpu(IB_LID_PERMISSIVE); if (in_mad->mad_hdr.method == IB_MGMT_METHOD_TRAP && slid == 0) return IB_MAD_RESULT_SUCCESS | IB_MAD_RESULT_CONSUMED; if (in_mad->mad_hdr.mgmt_class == IB_MGMT_CLASS_SUBN_LID_ROUTED || in_mad->mad_hdr.mgmt_class == IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE) { if (in_mad->mad_hdr.method != IB_MGMT_METHOD_GET && in_mad->mad_hdr.method != IB_MGMT_METHOD_SET && in_mad->mad_hdr.method != IB_MGMT_METHOD_TRAP_REPRESS) return IB_MAD_RESULT_SUCCESS; /* Don't process SMInfo queries -- the SMA can't handle them. */ if (in_mad->mad_hdr.attr_id == IB_SMP_ATTR_SM_INFO) return IB_MAD_RESULT_SUCCESS; } else if (in_mad->mad_hdr.mgmt_class == IB_MGMT_CLASS_PERF_MGMT || in_mad->mad_hdr.mgmt_class == MLX5_IB_VENDOR_CLASS1 || in_mad->mad_hdr.mgmt_class == MLX5_IB_VENDOR_CLASS2 || in_mad->mad_hdr.mgmt_class == IB_MGMT_CLASS_CONG_MGMT) { if (in_mad->mad_hdr.method != IB_MGMT_METHOD_GET && in_mad->mad_hdr.method != IB_MGMT_METHOD_SET) return IB_MAD_RESULT_SUCCESS; } else { return IB_MAD_RESULT_SUCCESS; } err = mlx5_MAD_IFC(to_mdev(ibdev), mad_flags & IB_MAD_IGNORE_MKEY, mad_flags & IB_MAD_IGNORE_BKEY, port_num, in_wc, in_grh, in_mad, out_mad); if (err) return IB_MAD_RESULT_FAILURE; /* set return bit in status of directed route responses */ if (in_mad->mad_hdr.mgmt_class == IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE) out_mad->mad_hdr.status |= cpu_to_be16(1 << 15); if (in_mad->mad_hdr.method == IB_MGMT_METHOD_TRAP_REPRESS) /* no response for trap repress */ return IB_MAD_RESULT_SUCCESS | IB_MAD_RESULT_CONSUMED; return IB_MAD_RESULT_SUCCESS | IB_MAD_RESULT_REPLY; } static void pma_cnt_ext_assign(struct ib_pma_portcounters_ext *pma_cnt_ext, void *out) { #define MLX5_SUM_CNT(p, cntr1, cntr2) \ (MLX5_GET64(query_vport_counter_out, p, cntr1) + \ MLX5_GET64(query_vport_counter_out, p, cntr2)) pma_cnt_ext->port_xmit_data = cpu_to_be64(MLX5_SUM_CNT(out, transmitted_ib_unicast.octets, transmitted_ib_multicast.octets) >> 2); pma_cnt_ext->port_rcv_data = cpu_to_be64(MLX5_SUM_CNT(out, received_ib_unicast.octets, received_ib_multicast.octets) >> 2); pma_cnt_ext->port_xmit_packets = cpu_to_be64(MLX5_SUM_CNT(out, transmitted_ib_unicast.packets, transmitted_ib_multicast.packets)); pma_cnt_ext->port_rcv_packets = cpu_to_be64(MLX5_SUM_CNT(out, received_ib_unicast.packets, received_ib_multicast.packets)); pma_cnt_ext->port_unicast_xmit_packets = MLX5_GET64_BE(query_vport_counter_out, out, transmitted_ib_unicast.packets); pma_cnt_ext->port_unicast_rcv_packets = MLX5_GET64_BE(query_vport_counter_out, out, received_ib_unicast.packets); pma_cnt_ext->port_multicast_xmit_packets = MLX5_GET64_BE(query_vport_counter_out, out, transmitted_ib_multicast.packets); pma_cnt_ext->port_multicast_rcv_packets = MLX5_GET64_BE(query_vport_counter_out, out, received_ib_multicast.packets); } static void pma_cnt_assign(struct ib_pma_portcounters *pma_cnt, void *out) { /* Traffic counters will be reported in * their 64bit form via ib_pma_portcounters_ext by default. */ void *out_pma = MLX5_ADDR_OF(ppcnt_reg, out, counter_set); #define MLX5_ASSIGN_PMA_CNTR(counter_var, counter_name) { \ counter_var = MLX5_GET_BE(typeof(counter_var), \ ib_port_cntrs_grp_data_layout, \ out_pma, counter_name); \ } MLX5_ASSIGN_PMA_CNTR(pma_cnt->symbol_error_counter, symbol_error_counter); MLX5_ASSIGN_PMA_CNTR(pma_cnt->link_error_recovery_counter, link_error_recovery_counter); MLX5_ASSIGN_PMA_CNTR(pma_cnt->link_downed_counter, link_downed_counter); MLX5_ASSIGN_PMA_CNTR(pma_cnt->port_rcv_errors, port_rcv_errors); MLX5_ASSIGN_PMA_CNTR(pma_cnt->port_rcv_remphys_errors, port_rcv_remote_physical_errors); MLX5_ASSIGN_PMA_CNTR(pma_cnt->port_rcv_switch_relay_errors, port_rcv_switch_relay_errors); MLX5_ASSIGN_PMA_CNTR(pma_cnt->port_xmit_discards, port_xmit_discards); MLX5_ASSIGN_PMA_CNTR(pma_cnt->port_xmit_constraint_errors, port_xmit_constraint_errors); MLX5_ASSIGN_PMA_CNTR(pma_cnt->port_rcv_constraint_errors, port_rcv_constraint_errors); MLX5_ASSIGN_PMA_CNTR(pma_cnt->link_overrun_errors, link_overrun_errors); MLX5_ASSIGN_PMA_CNTR(pma_cnt->vl15_dropped, vl_15_dropped); } static int process_pma_cmd(struct ib_device *ibdev, u8 port_num, const struct ib_mad *in_mad, struct ib_mad *out_mad) { struct mlx5_ib_dev *dev = to_mdev(ibdev); int err; void *out_cnt; /* Decalring support of extended counters */ if (in_mad->mad_hdr.attr_id == IB_PMA_CLASS_PORT_INFO) { struct ib_class_port_info cpi = {}; cpi.capability_mask = IB_PMA_CLASS_CAP_EXT_WIDTH; memcpy((out_mad->data + 40), &cpi, sizeof(cpi)); return IB_MAD_RESULT_SUCCESS | IB_MAD_RESULT_REPLY; } if (in_mad->mad_hdr.attr_id == IB_PMA_PORT_COUNTERS_EXT) { struct ib_pma_portcounters_ext *pma_cnt_ext = (struct ib_pma_portcounters_ext *)(out_mad->data + 40); int sz = MLX5_ST_SZ_BYTES(query_vport_counter_out); out_cnt = mlx5_vzalloc(sz); if (!out_cnt) return IB_MAD_RESULT_FAILURE; err = mlx5_core_query_vport_counter(dev->mdev, 0, 0, port_num, out_cnt, sz); if (!err) pma_cnt_ext_assign(pma_cnt_ext, out_cnt); } else { struct ib_pma_portcounters *pma_cnt = (struct ib_pma_portcounters *)(out_mad->data + 40); int sz = MLX5_ST_SZ_BYTES(ppcnt_reg); out_cnt = mlx5_vzalloc(sz); if (!out_cnt) return IB_MAD_RESULT_FAILURE; err = mlx5_core_query_ib_ppcnt(dev->mdev, port_num, out_cnt, sz); if (!err) pma_cnt_assign(pma_cnt, out_cnt); } kvfree(out_cnt); if (err) return IB_MAD_RESULT_FAILURE; return IB_MAD_RESULT_SUCCESS | IB_MAD_RESULT_REPLY; } int mlx5_ib_process_mad(struct ib_device *ibdev, int mad_flags, u8 port_num, const struct ib_wc *in_wc, const struct ib_grh *in_grh, const struct ib_mad_hdr *in, size_t in_mad_size, struct ib_mad_hdr *out, size_t *out_mad_size, u16 *out_mad_pkey_index) { struct mlx5_ib_dev *dev = to_mdev(ibdev); struct mlx5_core_dev *mdev = dev->mdev; const struct ib_mad *in_mad = (const struct ib_mad *)in; struct ib_mad *out_mad = (struct ib_mad *)out; if (WARN_ON_ONCE(in_mad_size != sizeof(*in_mad) || *out_mad_size != sizeof(*out_mad))) return IB_MAD_RESULT_FAILURE; memset(out_mad->data, 0, sizeof(out_mad->data)); if (MLX5_CAP_GEN(mdev, vport_counters) && in_mad->mad_hdr.mgmt_class == IB_MGMT_CLASS_PERF_MGMT && in_mad->mad_hdr.method == IB_MGMT_METHOD_GET) { return process_pma_cmd(ibdev, port_num, in_mad, out_mad); } else { return process_mad(ibdev, mad_flags, port_num, in_wc, in_grh, in_mad, out_mad); } } int mlx5_query_ext_port_caps(struct mlx5_ib_dev *dev, u8 port) { struct ib_smp *in_mad = NULL; struct ib_smp *out_mad = NULL; int err = -ENOMEM; u16 packet_error; in_mad = kzalloc(sizeof(*in_mad), GFP_KERNEL); out_mad = kmalloc(sizeof(*out_mad), GFP_KERNEL); if (!in_mad || !out_mad) goto out; init_query_mad(in_mad); in_mad->attr_id = MLX5_ATTR_EXTENDED_PORT_INFO; in_mad->attr_mod = cpu_to_be32(port); err = mlx5_MAD_IFC(dev, 1, 1, 1, NULL, NULL, in_mad, out_mad); packet_error = be16_to_cpu(out_mad->status); dev->mdev->port_caps[port - 1].ext_port_cap = (!err && !packet_error) ? MLX_EXT_PORT_CAP_FLAG_EXTENDED_PORT_INFO : 0; out: kfree(in_mad); kfree(out_mad); return err; } int mlx5_query_mad_ifc_smp_attr_node_info(struct ib_device *ibdev, struct ib_smp *out_mad) { struct ib_smp *in_mad = NULL; int err = -ENOMEM; in_mad = kzalloc(sizeof(*in_mad), GFP_KERNEL); if (!in_mad) return -ENOMEM; init_query_mad(in_mad); in_mad->attr_id = IB_SMP_ATTR_NODE_INFO; err = mlx5_MAD_IFC(to_mdev(ibdev), 1, 1, 1, NULL, NULL, in_mad, out_mad); kfree(in_mad); return err; } int mlx5_query_mad_ifc_system_image_guid(struct ib_device *ibdev, __be64 *sys_image_guid) { struct ib_smp *out_mad = NULL; int err = -ENOMEM; out_mad = kmalloc(sizeof(*out_mad), GFP_KERNEL); if (!out_mad) return -ENOMEM; err = mlx5_query_mad_ifc_smp_attr_node_info(ibdev, out_mad); if (err) goto out; memcpy(sys_image_guid, out_mad->data + 4, 8); out: kfree(out_mad); return err; } int mlx5_query_mad_ifc_max_pkeys(struct ib_device *ibdev, u16 *max_pkeys) { struct ib_smp *out_mad = NULL; int err = -ENOMEM; out_mad = kmalloc(sizeof(*out_mad), GFP_KERNEL); if (!out_mad) return -ENOMEM; err = mlx5_query_mad_ifc_smp_attr_node_info(ibdev, out_mad); if (err) goto out; *max_pkeys = be16_to_cpup((__be16 *)(out_mad->data + 28)); out: kfree(out_mad); return err; } int mlx5_query_mad_ifc_vendor_id(struct ib_device *ibdev, u32 *vendor_id) { struct ib_smp *out_mad = NULL; int err = -ENOMEM; out_mad = kmalloc(sizeof(*out_mad), GFP_KERNEL); if (!out_mad) return -ENOMEM; err = mlx5_query_mad_ifc_smp_attr_node_info(ibdev, out_mad); if (err) goto out; *vendor_id = be32_to_cpup((__be32 *)(out_mad->data + 36)) & 0xffff; out: kfree(out_mad); return err; } int mlx5_query_mad_ifc_node_desc(struct mlx5_ib_dev *dev, char *node_desc) { struct ib_smp *in_mad = NULL; struct ib_smp *out_mad = NULL; int err = -ENOMEM; in_mad = kzalloc(sizeof(*in_mad), GFP_KERNEL); out_mad = kmalloc(sizeof(*out_mad), GFP_KERNEL); if (!in_mad || !out_mad) goto out; init_query_mad(in_mad); in_mad->attr_id = IB_SMP_ATTR_NODE_DESC; err = mlx5_MAD_IFC(dev, 1, 1, 1, NULL, NULL, in_mad, out_mad); if (err) goto out; memcpy(node_desc, out_mad->data, IB_DEVICE_NODE_DESC_MAX); out: kfree(in_mad); kfree(out_mad); return err; } int mlx5_query_mad_ifc_node_guid(struct mlx5_ib_dev *dev, __be64 *node_guid) { struct ib_smp *in_mad = NULL; struct ib_smp *out_mad = NULL; int err = -ENOMEM; in_mad = kzalloc(sizeof(*in_mad), GFP_KERNEL); out_mad = kmalloc(sizeof(*out_mad), GFP_KERNEL); if (!in_mad || !out_mad) goto out; init_query_mad(in_mad); in_mad->attr_id = IB_SMP_ATTR_NODE_INFO; err = mlx5_MAD_IFC(dev, 1, 1, 1, NULL, NULL, in_mad, out_mad); if (err) goto out; memcpy(node_guid, out_mad->data + 12, 8); out: kfree(in_mad); kfree(out_mad); return err; } int mlx5_query_mad_ifc_pkey(struct ib_device *ibdev, u8 port, u16 index, u16 *pkey) { struct ib_smp *in_mad = NULL; struct ib_smp *out_mad = NULL; int err = -ENOMEM; in_mad = kzalloc(sizeof(*in_mad), GFP_KERNEL); out_mad = kmalloc(sizeof(*out_mad), GFP_KERNEL); if (!in_mad || !out_mad) goto out; init_query_mad(in_mad); in_mad->attr_id = IB_SMP_ATTR_PKEY_TABLE; in_mad->attr_mod = cpu_to_be32(index / 32); err = mlx5_MAD_IFC(to_mdev(ibdev), 1, 1, port, NULL, NULL, in_mad, out_mad); if (err) goto out; *pkey = be16_to_cpu(((__be16 *)out_mad->data)[index % 32]); out: kfree(in_mad); kfree(out_mad); return err; } int mlx5_query_mad_ifc_gids(struct ib_device *ibdev, u8 port, int index, union ib_gid *gid) { struct ib_smp *in_mad = NULL; struct ib_smp *out_mad = NULL; int err = -ENOMEM; in_mad = kzalloc(sizeof(*in_mad), GFP_KERNEL); out_mad = kmalloc(sizeof(*out_mad), GFP_KERNEL); if (!in_mad || !out_mad) goto out; init_query_mad(in_mad); in_mad->attr_id = IB_SMP_ATTR_PORT_INFO; in_mad->attr_mod = cpu_to_be32(port); err = mlx5_MAD_IFC(to_mdev(ibdev), 1, 1, port, NULL, NULL, in_mad, out_mad); if (err) goto out; memcpy(gid->raw, out_mad->data + 8, 8); init_query_mad(in_mad); in_mad->attr_id = IB_SMP_ATTR_GUID_INFO; in_mad->attr_mod = cpu_to_be32(index / 8); err = mlx5_MAD_IFC(to_mdev(ibdev), 1, 1, port, NULL, NULL, in_mad, out_mad); if (err) goto out; memcpy(gid->raw + 8, out_mad->data + (index % 8) * 8, 8); out: kfree(in_mad); kfree(out_mad); return err; } int mlx5_query_mad_ifc_port(struct ib_device *ibdev, u8 port, struct ib_port_attr *props) { struct mlx5_ib_dev *dev = to_mdev(ibdev); struct mlx5_core_dev *mdev = dev->mdev; struct ib_smp *in_mad = NULL; struct ib_smp *out_mad = NULL; int ext_active_speed; int err = -ENOMEM; if (port < 1 || port > MLX5_CAP_GEN(mdev, num_ports)) { mlx5_ib_warn(dev, "invalid port number %d\n", port); return -EINVAL; } in_mad = kzalloc(sizeof(*in_mad), GFP_KERNEL); out_mad = kmalloc(sizeof(*out_mad), GFP_KERNEL); if (!in_mad || !out_mad) goto out; memset(props, 0, sizeof(*props)); init_query_mad(in_mad); in_mad->attr_id = IB_SMP_ATTR_PORT_INFO; in_mad->attr_mod = cpu_to_be32(port); err = mlx5_MAD_IFC(dev, 1, 1, port, NULL, NULL, in_mad, out_mad); if (err) { mlx5_ib_warn(dev, "err %d\n", err); goto out; } props->lid = be16_to_cpup((__be16 *)(out_mad->data + 16)); props->lmc = out_mad->data[34] & 0x7; props->sm_lid = be16_to_cpup((__be16 *)(out_mad->data + 18)); props->sm_sl = out_mad->data[36] & 0xf; props->state = out_mad->data[32] & 0xf; props->phys_state = out_mad->data[33] >> 4; props->port_cap_flags = be32_to_cpup((__be32 *)(out_mad->data + 20)); props->gid_tbl_len = out_mad->data[50]; props->max_msg_sz = 1 << MLX5_CAP_GEN(mdev, log_max_msg); props->pkey_tbl_len = mdev->port_caps[port - 1].pkey_table_len; props->bad_pkey_cntr = be16_to_cpup((__be16 *)(out_mad->data + 46)); props->qkey_viol_cntr = be16_to_cpup((__be16 *)(out_mad->data + 48)); props->active_width = out_mad->data[31] & 0xf; props->active_speed = out_mad->data[35] >> 4; props->max_mtu = out_mad->data[41] & 0xf; props->active_mtu = out_mad->data[36] >> 4; props->subnet_timeout = out_mad->data[51] & 0x1f; props->max_vl_num = out_mad->data[37] >> 4; props->init_type_reply = out_mad->data[41] >> 4; /* Check if extended speeds (EDR/FDR/...) are supported */ if (props->port_cap_flags & IB_PORT_EXTENDED_SPEEDS_SUP) { ext_active_speed = out_mad->data[62] >> 4; switch (ext_active_speed) { case 1: props->active_speed = 16; /* FDR */ break; case 2: props->active_speed = 32; /* EDR */ break; } } /* If reported active speed is QDR, check if is FDR-10 */ if (props->active_speed == 4) { if (mdev->port_caps[port - 1].ext_port_cap & MLX_EXT_PORT_CAP_FLAG_EXTENDED_PORT_INFO) { init_query_mad(in_mad); in_mad->attr_id = MLX5_ATTR_EXTENDED_PORT_INFO; in_mad->attr_mod = cpu_to_be32(port); err = mlx5_MAD_IFC(dev, 1, 1, port, NULL, NULL, in_mad, out_mad); if (err) goto out; /* Checking LinkSpeedActive for FDR-10 */ if (out_mad->data[15] & 0x1) props->active_speed = 8; } } out: kfree(in_mad); kfree(out_mad); return err; } diff --git a/sys/dev/mlx5/mlx5_ib/mlx5_ib_main.c b/sys/dev/mlx5/mlx5_ib/mlx5_ib_main.c index 23fc38424d02..f517dfc58fac 100644 --- a/sys/dev/mlx5/mlx5_ib/mlx5_ib_main.c +++ b/sys/dev/mlx5/mlx5_ib/mlx5_ib_main.c @@ -1,3573 +1,3573 @@ /*- * Copyright (c) 2013-2021, Mellanox Technologies, Ltd. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY AUTHOR AND CONTRIBUTORS `AS IS' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $FreeBSD$ */ #include "opt_rss.h" #include "opt_ratelimit.h" #include #include #include #include #include #if defined(CONFIG_X86) #include #endif #include #include #include #undef inode #include #include #include #include #include #include #include #include #include #include #include -#include "mlx5_ib.h" +#include MODULE_DESCRIPTION("Mellanox Connect-IB HCA IB driver"); MODULE_LICENSE("Dual BSD/GPL"); MODULE_DEPEND(mlx5ib, linuxkpi, 1, 1, 1); MODULE_DEPEND(mlx5ib, mlx5, 1, 1, 1); MODULE_DEPEND(mlx5ib, ibcore, 1, 1, 1); MODULE_VERSION(mlx5ib, 1); enum { MLX5_ATOMIC_SIZE_QP_8BYTES = 1 << 3, }; static enum rdma_link_layer mlx5_port_type_cap_to_rdma_ll(int port_type_cap) { switch (port_type_cap) { case MLX5_CAP_PORT_TYPE_IB: return IB_LINK_LAYER_INFINIBAND; case MLX5_CAP_PORT_TYPE_ETH: return IB_LINK_LAYER_ETHERNET; default: return IB_LINK_LAYER_UNSPECIFIED; } } static enum rdma_link_layer mlx5_ib_port_link_layer(struct ib_device *device, u8 port_num) { struct mlx5_ib_dev *dev = to_mdev(device); int port_type_cap = MLX5_CAP_GEN(dev->mdev, port_type); return mlx5_port_type_cap_to_rdma_ll(port_type_cap); } static bool mlx5_netdev_match(struct ifnet *ndev, struct mlx5_core_dev *mdev, const char *dname) { return ndev->if_type == IFT_ETHER && ndev->if_dname != NULL && strcmp(ndev->if_dname, dname) == 0 && ndev->if_softc != NULL && *(struct mlx5_core_dev **)ndev->if_softc == mdev; } static int mlx5_netdev_event(struct notifier_block *this, unsigned long event, void *ptr) { struct ifnet *ndev = netdev_notifier_info_to_ifp(ptr); struct mlx5_ib_dev *ibdev = container_of(this, struct mlx5_ib_dev, roce.nb); switch (event) { case NETDEV_REGISTER: case NETDEV_UNREGISTER: write_lock(&ibdev->roce.netdev_lock); /* check if network interface belongs to mlx5en */ if (mlx5_netdev_match(ndev, ibdev->mdev, "mce")) ibdev->roce.netdev = (event == NETDEV_UNREGISTER) ? NULL : ndev; write_unlock(&ibdev->roce.netdev_lock); break; case NETDEV_UP: case NETDEV_DOWN: { struct ifnet *upper = NULL; if ((upper == ndev || (!upper && ndev == ibdev->roce.netdev)) && ibdev->ib_active) { struct ib_event ibev = {0}; ibev.device = &ibdev->ib_dev; ibev.event = (event == NETDEV_UP) ? IB_EVENT_PORT_ACTIVE : IB_EVENT_PORT_ERR; ibev.element.port_num = 1; ib_dispatch_event(&ibev); } break; } default: break; } return NOTIFY_DONE; } static struct ifnet *mlx5_ib_get_netdev(struct ib_device *device, u8 port_num) { struct mlx5_ib_dev *ibdev = to_mdev(device); struct ifnet *ndev; /* Ensure ndev does not disappear before we invoke if_ref() */ read_lock(&ibdev->roce.netdev_lock); ndev = ibdev->roce.netdev; if (ndev) if_ref(ndev); read_unlock(&ibdev->roce.netdev_lock); return ndev; } static int translate_eth_proto_oper(u32 eth_proto_oper, u8 *active_speed, u8 *active_width) { switch (eth_proto_oper) { case MLX5E_PROT_MASK(MLX5E_1000BASE_CX_SGMII): case MLX5E_PROT_MASK(MLX5E_1000BASE_KX): case MLX5E_PROT_MASK(MLX5E_100BASE_TX): case MLX5E_PROT_MASK(MLX5E_1000BASE_T): *active_width = IB_WIDTH_1X; *active_speed = IB_SPEED_SDR; break; case MLX5E_PROT_MASK(MLX5E_10GBASE_T): case MLX5E_PROT_MASK(MLX5E_10GBASE_CX4): case MLX5E_PROT_MASK(MLX5E_10GBASE_KX4): case MLX5E_PROT_MASK(MLX5E_10GBASE_KR): case MLX5E_PROT_MASK(MLX5E_10GBASE_CR): case MLX5E_PROT_MASK(MLX5E_10GBASE_SR): case MLX5E_PROT_MASK(MLX5E_10GBASE_ER_LR): *active_width = IB_WIDTH_1X; *active_speed = IB_SPEED_QDR; break; case MLX5E_PROT_MASK(MLX5E_25GBASE_CR): case MLX5E_PROT_MASK(MLX5E_25GBASE_KR): case MLX5E_PROT_MASK(MLX5E_25GBASE_SR): *active_width = IB_WIDTH_1X; *active_speed = IB_SPEED_EDR; break; case MLX5E_PROT_MASK(MLX5E_40GBASE_CR4): case MLX5E_PROT_MASK(MLX5E_40GBASE_KR4): case MLX5E_PROT_MASK(MLX5E_40GBASE_SR4): case MLX5E_PROT_MASK(MLX5E_40GBASE_LR4_ER4): *active_width = IB_WIDTH_4X; *active_speed = IB_SPEED_QDR; break; case MLX5E_PROT_MASK(MLX5E_50GBASE_CR2): case MLX5E_PROT_MASK(MLX5E_50GBASE_KR2): case MLX5E_PROT_MASK(MLX5E_50GBASE_KR4): case MLX5E_PROT_MASK(MLX5E_50GBASE_SR2): *active_width = IB_WIDTH_1X; *active_speed = IB_SPEED_HDR; break; case MLX5E_PROT_MASK(MLX5E_56GBASE_R4): *active_width = IB_WIDTH_4X; *active_speed = IB_SPEED_FDR; break; case MLX5E_PROT_MASK(MLX5E_100GBASE_CR4): case MLX5E_PROT_MASK(MLX5E_100GBASE_SR4): case MLX5E_PROT_MASK(MLX5E_100GBASE_KR4): case MLX5E_PROT_MASK(MLX5E_100GBASE_LR4): *active_width = IB_WIDTH_4X; *active_speed = IB_SPEED_EDR; break; default: *active_width = IB_WIDTH_4X; *active_speed = IB_SPEED_QDR; return -EINVAL; } return 0; } static int translate_eth_ext_proto_oper(u32 eth_proto_oper, u8 *active_speed, u8 *active_width) { switch (eth_proto_oper) { case MLX5E_PROT_MASK(MLX5E_SGMII_100M): case MLX5E_PROT_MASK(MLX5E_1000BASE_X_SGMII): *active_width = IB_WIDTH_1X; *active_speed = IB_SPEED_SDR; break; case MLX5E_PROT_MASK(MLX5E_5GBASE_R): *active_width = IB_WIDTH_1X; *active_speed = IB_SPEED_DDR; break; case MLX5E_PROT_MASK(MLX5E_10GBASE_XFI_XAUI_1): *active_width = IB_WIDTH_1X; *active_speed = IB_SPEED_QDR; break; case MLX5E_PROT_MASK(MLX5E_40GBASE_XLAUI_4_XLPPI_4): *active_width = IB_WIDTH_4X; *active_speed = IB_SPEED_QDR; break; case MLX5E_PROT_MASK(MLX5E_25GAUI_1_25GBASE_CR_KR): *active_width = IB_WIDTH_1X; *active_speed = IB_SPEED_EDR; break; case MLX5E_PROT_MASK(MLX5E_50GAUI_2_LAUI_2_50GBASE_CR2_KR2): *active_width = IB_WIDTH_2X; *active_speed = IB_SPEED_EDR; break; case MLX5E_PROT_MASK(MLX5E_50GAUI_1_LAUI_1_50GBASE_CR_KR): *active_width = IB_WIDTH_1X; *active_speed = IB_SPEED_HDR; break; case MLX5E_PROT_MASK(MLX5E_CAUI_4_100GBASE_CR4_KR4): *active_width = IB_WIDTH_4X; *active_speed = IB_SPEED_EDR; break; case MLX5E_PROT_MASK(MLX5E_100GAUI_2_100GBASE_CR2_KR2): *active_width = IB_WIDTH_2X; *active_speed = IB_SPEED_HDR; break; case MLX5E_PROT_MASK(MLX5E_200GAUI_4_200GBASE_CR4_KR4): *active_width = IB_WIDTH_4X; *active_speed = IB_SPEED_HDR; break; default: *active_width = IB_WIDTH_4X; *active_speed = IB_SPEED_QDR; return -EINVAL; } return 0; } static int mlx5_query_port_roce(struct ib_device *device, u8 port_num, struct ib_port_attr *props) { struct mlx5_ib_dev *dev = to_mdev(device); u32 out[MLX5_ST_SZ_DW(ptys_reg)] = {}; struct ifnet *ndev; enum ib_mtu ndev_ib_mtu; u16 qkey_viol_cntr; u32 eth_prot_oper; bool ext; int err; memset(props, 0, sizeof(*props)); /* Possible bad flows are checked before filling out props so in case * of an error it will still be zeroed out. */ err = mlx5_query_port_ptys(dev->mdev, out, sizeof(out), MLX5_PTYS_EN, port_num); if (err) return err; ext = MLX5_CAP_PCAM_FEATURE(dev->mdev, ptys_extended_ethernet); eth_prot_oper = MLX5_GET_ETH_PROTO(ptys_reg, out, ext, eth_proto_oper); if (ext) translate_eth_ext_proto_oper(eth_prot_oper, &props->active_speed, &props->active_width); else translate_eth_proto_oper(eth_prot_oper, &props->active_speed, &props->active_width); props->port_cap_flags |= IB_PORT_CM_SUP; props->port_cap_flags |= IB_PORT_IP_BASED_GIDS; props->gid_tbl_len = MLX5_CAP_ROCE(dev->mdev, roce_address_table_size); props->max_mtu = IB_MTU_4096; props->max_msg_sz = 1 << MLX5_CAP_GEN(dev->mdev, log_max_msg); props->pkey_tbl_len = 1; props->state = IB_PORT_DOWN; props->phys_state = IB_PORT_PHYS_STATE_DISABLED; mlx5_query_nic_vport_qkey_viol_cntr(dev->mdev, &qkey_viol_cntr); props->qkey_viol_cntr = qkey_viol_cntr; ndev = mlx5_ib_get_netdev(device, port_num); if (!ndev) return 0; if (ndev->if_drv_flags & IFF_DRV_RUNNING && ndev->if_link_state == LINK_STATE_UP) { props->state = IB_PORT_ACTIVE; props->phys_state = IB_PORT_PHYS_STATE_LINK_UP; } ndev_ib_mtu = iboe_get_mtu(ndev->if_mtu); if_rele(ndev); props->active_mtu = min(props->max_mtu, ndev_ib_mtu); return 0; } static void ib_gid_to_mlx5_roce_addr(const union ib_gid *gid, const struct ib_gid_attr *attr, void *mlx5_addr) { #define MLX5_SET_RA(p, f, v) MLX5_SET(roce_addr_layout, p, f, v) char *mlx5_addr_l3_addr = MLX5_ADDR_OF(roce_addr_layout, mlx5_addr, source_l3_address); void *mlx5_addr_mac = MLX5_ADDR_OF(roce_addr_layout, mlx5_addr, source_mac_47_32); u16 vlan_id; if (!gid) return; ether_addr_copy(mlx5_addr_mac, IF_LLADDR(attr->ndev)); vlan_id = rdma_vlan_dev_vlan_id(attr->ndev); if (vlan_id != 0xffff) { MLX5_SET_RA(mlx5_addr, vlan_valid, 1); MLX5_SET_RA(mlx5_addr, vlan_id, vlan_id); } switch (attr->gid_type) { case IB_GID_TYPE_IB: MLX5_SET_RA(mlx5_addr, roce_version, MLX5_ROCE_VERSION_1); break; case IB_GID_TYPE_ROCE_UDP_ENCAP: MLX5_SET_RA(mlx5_addr, roce_version, MLX5_ROCE_VERSION_2); break; default: WARN_ON(true); } if (attr->gid_type != IB_GID_TYPE_IB) { if (ipv6_addr_v4mapped((void *)gid)) MLX5_SET_RA(mlx5_addr, roce_l3_type, MLX5_ROCE_L3_TYPE_IPV4); else MLX5_SET_RA(mlx5_addr, roce_l3_type, MLX5_ROCE_L3_TYPE_IPV6); } if ((attr->gid_type == IB_GID_TYPE_IB) || !ipv6_addr_v4mapped((void *)gid)) memcpy(mlx5_addr_l3_addr, gid, sizeof(*gid)); else memcpy(&mlx5_addr_l3_addr[12], &gid->raw[12], 4); } static int set_roce_addr(struct ib_device *device, u8 port_num, unsigned int index, const union ib_gid *gid, const struct ib_gid_attr *attr) { struct mlx5_ib_dev *dev = to_mdev(device); u32 in[MLX5_ST_SZ_DW(set_roce_address_in)] = {0}; u32 out[MLX5_ST_SZ_DW(set_roce_address_out)] = {0}; void *in_addr = MLX5_ADDR_OF(set_roce_address_in, in, roce_address); enum rdma_link_layer ll = mlx5_ib_port_link_layer(device, port_num); if (ll != IB_LINK_LAYER_ETHERNET) return -EINVAL; ib_gid_to_mlx5_roce_addr(gid, attr, in_addr); MLX5_SET(set_roce_address_in, in, roce_address_index, index); MLX5_SET(set_roce_address_in, in, opcode, MLX5_CMD_OP_SET_ROCE_ADDRESS); return mlx5_cmd_exec(dev->mdev, in, sizeof(in), out, sizeof(out)); } static int mlx5_ib_add_gid(struct ib_device *device, u8 port_num, unsigned int index, const union ib_gid *gid, const struct ib_gid_attr *attr, __always_unused void **context) { return set_roce_addr(device, port_num, index, gid, attr); } static int mlx5_ib_del_gid(struct ib_device *device, u8 port_num, unsigned int index, __always_unused void **context) { return set_roce_addr(device, port_num, index, NULL, NULL); } __be16 mlx5_get_roce_udp_sport(struct mlx5_ib_dev *dev, u8 port_num, int index) { struct ib_gid_attr attr; union ib_gid gid; if (ib_get_cached_gid(&dev->ib_dev, port_num, index, &gid, &attr)) return 0; if (!attr.ndev) return 0; if_rele(attr.ndev); if (attr.gid_type != IB_GID_TYPE_ROCE_UDP_ENCAP) return 0; return cpu_to_be16(MLX5_CAP_ROCE(dev->mdev, r_roce_min_src_udp_port)); } int mlx5_get_roce_gid_type(struct mlx5_ib_dev *dev, u8 port_num, int index, enum ib_gid_type *gid_type) { struct ib_gid_attr attr; union ib_gid gid; int ret; ret = ib_get_cached_gid(&dev->ib_dev, port_num, index, &gid, &attr); if (ret) return ret; if (!attr.ndev) return -ENODEV; if_rele(attr.ndev); *gid_type = attr.gid_type; return 0; } static int mlx5_use_mad_ifc(struct mlx5_ib_dev *dev) { if (MLX5_CAP_GEN(dev->mdev, port_type) == MLX5_CAP_PORT_TYPE_IB) return !MLX5_CAP_GEN(dev->mdev, ib_virt); return 0; } enum { MLX5_VPORT_ACCESS_METHOD_MAD, MLX5_VPORT_ACCESS_METHOD_HCA, MLX5_VPORT_ACCESS_METHOD_NIC, }; static int mlx5_get_vport_access_method(struct ib_device *ibdev) { if (mlx5_use_mad_ifc(to_mdev(ibdev))) return MLX5_VPORT_ACCESS_METHOD_MAD; if (mlx5_ib_port_link_layer(ibdev, 1) == IB_LINK_LAYER_ETHERNET) return MLX5_VPORT_ACCESS_METHOD_NIC; return MLX5_VPORT_ACCESS_METHOD_HCA; } static void get_atomic_caps(struct mlx5_ib_dev *dev, struct ib_device_attr *props) { u8 tmp; u8 atomic_operations = MLX5_CAP_ATOMIC(dev->mdev, atomic_operations); u8 atomic_size_qp = MLX5_CAP_ATOMIC(dev->mdev, atomic_size_qp); u8 atomic_req_8B_endianness_mode = MLX5_CAP_ATOMIC(dev->mdev, atomic_req_8B_endianess_mode); /* Check if HW supports 8 bytes standard atomic operations and capable * of host endianness respond */ tmp = MLX5_ATOMIC_OPS_CMP_SWAP | MLX5_ATOMIC_OPS_FETCH_ADD; if (((atomic_operations & tmp) == tmp) && (atomic_size_qp & MLX5_ATOMIC_SIZE_QP_8BYTES) && (atomic_req_8B_endianness_mode)) { props->atomic_cap = IB_ATOMIC_HCA; } else { props->atomic_cap = IB_ATOMIC_NONE; } } static int mlx5_query_system_image_guid(struct ib_device *ibdev, __be64 *sys_image_guid) { struct mlx5_ib_dev *dev = to_mdev(ibdev); struct mlx5_core_dev *mdev = dev->mdev; u64 tmp; int err; switch (mlx5_get_vport_access_method(ibdev)) { case MLX5_VPORT_ACCESS_METHOD_MAD: return mlx5_query_mad_ifc_system_image_guid(ibdev, sys_image_guid); case MLX5_VPORT_ACCESS_METHOD_HCA: err = mlx5_query_hca_vport_system_image_guid(mdev, &tmp); break; case MLX5_VPORT_ACCESS_METHOD_NIC: err = mlx5_query_nic_vport_system_image_guid(mdev, &tmp); break; default: return -EINVAL; } if (!err) *sys_image_guid = cpu_to_be64(tmp); return err; } static int mlx5_query_max_pkeys(struct ib_device *ibdev, u16 *max_pkeys) { struct mlx5_ib_dev *dev = to_mdev(ibdev); struct mlx5_core_dev *mdev = dev->mdev; switch (mlx5_get_vport_access_method(ibdev)) { case MLX5_VPORT_ACCESS_METHOD_MAD: return mlx5_query_mad_ifc_max_pkeys(ibdev, max_pkeys); case MLX5_VPORT_ACCESS_METHOD_HCA: case MLX5_VPORT_ACCESS_METHOD_NIC: *max_pkeys = mlx5_to_sw_pkey_sz(MLX5_CAP_GEN(mdev, pkey_table_size)); return 0; default: return -EINVAL; } } static int mlx5_query_vendor_id(struct ib_device *ibdev, u32 *vendor_id) { struct mlx5_ib_dev *dev = to_mdev(ibdev); switch (mlx5_get_vport_access_method(ibdev)) { case MLX5_VPORT_ACCESS_METHOD_MAD: return mlx5_query_mad_ifc_vendor_id(ibdev, vendor_id); case MLX5_VPORT_ACCESS_METHOD_HCA: case MLX5_VPORT_ACCESS_METHOD_NIC: return mlx5_core_query_vendor_id(dev->mdev, vendor_id); default: return -EINVAL; } } static int mlx5_query_node_guid(struct mlx5_ib_dev *dev, __be64 *node_guid) { u64 tmp; int err; switch (mlx5_get_vport_access_method(&dev->ib_dev)) { case MLX5_VPORT_ACCESS_METHOD_MAD: return mlx5_query_mad_ifc_node_guid(dev, node_guid); case MLX5_VPORT_ACCESS_METHOD_HCA: err = mlx5_query_hca_vport_node_guid(dev->mdev, &tmp); break; case MLX5_VPORT_ACCESS_METHOD_NIC: err = mlx5_query_nic_vport_node_guid(dev->mdev, &tmp); break; default: return -EINVAL; } if (!err) *node_guid = cpu_to_be64(tmp); return err; } struct mlx5_reg_node_desc { u8 desc[IB_DEVICE_NODE_DESC_MAX]; }; static int mlx5_query_node_desc(struct mlx5_ib_dev *dev, char *node_desc) { struct mlx5_reg_node_desc in; if (mlx5_use_mad_ifc(dev)) return mlx5_query_mad_ifc_node_desc(dev, node_desc); memset(&in, 0, sizeof(in)); return mlx5_core_access_reg(dev->mdev, &in, sizeof(in), node_desc, sizeof(struct mlx5_reg_node_desc), MLX5_REG_NODE_DESC, 0, 0); } static int mlx5_ib_query_device(struct ib_device *ibdev, struct ib_device_attr *props, struct ib_udata *uhw) { struct mlx5_ib_dev *dev = to_mdev(ibdev); struct mlx5_core_dev *mdev = dev->mdev; int err = -ENOMEM; int max_sq_desc; int max_rq_sg; int max_sq_sg; u64 min_page_size = 1ull << MLX5_CAP_GEN(mdev, log_pg_sz); struct mlx5_ib_query_device_resp resp = {}; size_t resp_len; u64 max_tso; resp_len = sizeof(resp.comp_mask) + sizeof(resp.response_length); if (uhw->outlen && uhw->outlen < resp_len) return -EINVAL; else resp.response_length = resp_len; if (uhw->inlen && !ib_is_udata_cleared(uhw, 0, uhw->inlen)) return -EINVAL; memset(props, 0, sizeof(*props)); err = mlx5_query_system_image_guid(ibdev, &props->sys_image_guid); if (err) return err; err = mlx5_query_max_pkeys(ibdev, &props->max_pkeys); if (err) return err; err = mlx5_query_vendor_id(ibdev, &props->vendor_id); if (err) return err; props->fw_ver = ((u64)fw_rev_maj(dev->mdev) << 32) | ((u32)fw_rev_min(dev->mdev) << 16) | fw_rev_sub(dev->mdev); props->device_cap_flags = IB_DEVICE_CHANGE_PHY_PORT | IB_DEVICE_PORT_ACTIVE_EVENT | IB_DEVICE_SYS_IMAGE_GUID | IB_DEVICE_RC_RNR_NAK_GEN; if (MLX5_CAP_GEN(mdev, pkv)) props->device_cap_flags |= IB_DEVICE_BAD_PKEY_CNTR; if (MLX5_CAP_GEN(mdev, qkv)) props->device_cap_flags |= IB_DEVICE_BAD_QKEY_CNTR; if (MLX5_CAP_GEN(mdev, apm)) props->device_cap_flags |= IB_DEVICE_AUTO_PATH_MIG; if (MLX5_CAP_GEN(mdev, xrc)) props->device_cap_flags |= IB_DEVICE_XRC; if (MLX5_CAP_GEN(mdev, imaicl)) { props->device_cap_flags |= IB_DEVICE_MEM_WINDOW | IB_DEVICE_MEM_WINDOW_TYPE_2B; props->max_mw = 1 << MLX5_CAP_GEN(mdev, log_max_mkey); /* We support 'Gappy' memory registration too */ props->device_cap_flags |= IB_DEVICE_SG_GAPS_REG; } props->device_cap_flags |= IB_DEVICE_MEM_MGT_EXTENSIONS; if (MLX5_CAP_GEN(mdev, sho)) { props->device_cap_flags |= IB_DEVICE_SIGNATURE_HANDOVER; /* At this stage no support for signature handover */ props->sig_prot_cap = IB_PROT_T10DIF_TYPE_1 | IB_PROT_T10DIF_TYPE_2 | IB_PROT_T10DIF_TYPE_3; props->sig_guard_cap = IB_GUARD_T10DIF_CRC | IB_GUARD_T10DIF_CSUM; } if (MLX5_CAP_GEN(mdev, block_lb_mc)) props->device_cap_flags |= IB_DEVICE_BLOCK_MULTICAST_LOOPBACK; if (MLX5_CAP_GEN(dev->mdev, eth_net_offloads)) { if (MLX5_CAP_ETH(mdev, csum_cap)) props->device_cap_flags |= IB_DEVICE_RAW_IP_CSUM; if (field_avail(typeof(resp), tso_caps, uhw->outlen)) { max_tso = MLX5_CAP_ETH(mdev, max_lso_cap); if (max_tso) { resp.tso_caps.max_tso = 1 << max_tso; resp.tso_caps.supported_qpts |= 1 << IB_QPT_RAW_PACKET; resp.response_length += sizeof(resp.tso_caps); } } if (field_avail(typeof(resp), rss_caps, uhw->outlen)) { resp.rss_caps.rx_hash_function = MLX5_RX_HASH_FUNC_TOEPLITZ; resp.rss_caps.rx_hash_fields_mask = MLX5_RX_HASH_SRC_IPV4 | MLX5_RX_HASH_DST_IPV4 | MLX5_RX_HASH_SRC_IPV6 | MLX5_RX_HASH_DST_IPV6 | MLX5_RX_HASH_SRC_PORT_TCP | MLX5_RX_HASH_DST_PORT_TCP | MLX5_RX_HASH_SRC_PORT_UDP | MLX5_RX_HASH_DST_PORT_UDP; resp.response_length += sizeof(resp.rss_caps); } } else { if (field_avail(typeof(resp), tso_caps, uhw->outlen)) resp.response_length += sizeof(resp.tso_caps); if (field_avail(typeof(resp), rss_caps, uhw->outlen)) resp.response_length += sizeof(resp.rss_caps); } if (MLX5_CAP_GEN(mdev, ipoib_ipoib_offloads)) { props->device_cap_flags |= IB_DEVICE_UD_IP_CSUM; props->device_cap_flags |= IB_DEVICE_UD_TSO; } if (MLX5_CAP_GEN(dev->mdev, eth_net_offloads) && MLX5_CAP_ETH(dev->mdev, scatter_fcs)) props->device_cap_flags |= IB_DEVICE_RAW_SCATTER_FCS; if (mlx5_get_flow_namespace(dev->mdev, MLX5_FLOW_NAMESPACE_BYPASS)) props->device_cap_flags |= IB_DEVICE_MANAGED_FLOW_STEERING; props->vendor_part_id = mdev->pdev->device; props->hw_ver = mdev->pdev->revision; props->max_mr_size = ~0ull; props->page_size_cap = ~(min_page_size - 1); props->max_qp = 1 << MLX5_CAP_GEN(mdev, log_max_qp); props->max_qp_wr = 1 << MLX5_CAP_GEN(mdev, log_max_qp_sz); max_rq_sg = MLX5_CAP_GEN(mdev, max_wqe_sz_rq) / sizeof(struct mlx5_wqe_data_seg); max_sq_desc = min_t(int, MLX5_CAP_GEN(mdev, max_wqe_sz_sq), 512); max_sq_sg = (max_sq_desc - sizeof(struct mlx5_wqe_ctrl_seg) - sizeof(struct mlx5_wqe_raddr_seg)) / sizeof(struct mlx5_wqe_data_seg); props->max_sge = min(max_rq_sg, max_sq_sg); props->max_sge_rd = MLX5_MAX_SGE_RD; props->max_cq = 1 << MLX5_CAP_GEN(mdev, log_max_cq); props->max_cqe = (1 << MLX5_CAP_GEN(mdev, log_max_cq_sz)) - 1; props->max_mr = 1 << MLX5_CAP_GEN(mdev, log_max_mkey); props->max_pd = 1 << MLX5_CAP_GEN(mdev, log_max_pd); props->max_qp_rd_atom = 1 << MLX5_CAP_GEN(mdev, log_max_ra_req_qp); props->max_qp_init_rd_atom = 1 << MLX5_CAP_GEN(mdev, log_max_ra_res_qp); props->max_srq = 1 << MLX5_CAP_GEN(mdev, log_max_srq); props->max_srq_wr = (1 << MLX5_CAP_GEN(mdev, log_max_srq_sz)) - 1; props->local_ca_ack_delay = MLX5_CAP_GEN(mdev, local_ca_ack_delay); props->max_res_rd_atom = props->max_qp_rd_atom * props->max_qp; props->max_srq_sge = max_rq_sg - 1; props->max_fast_reg_page_list_len = 1 << MLX5_CAP_GEN(mdev, log_max_klm_list_size); get_atomic_caps(dev, props); props->masked_atomic_cap = IB_ATOMIC_NONE; props->max_mcast_grp = 1 << MLX5_CAP_GEN(mdev, log_max_mcg); props->max_mcast_qp_attach = MLX5_CAP_GEN(mdev, max_qp_mcg); props->max_total_mcast_qp_attach = props->max_mcast_qp_attach * props->max_mcast_grp; props->max_map_per_fmr = INT_MAX; /* no limit in ConnectIB */ props->hca_core_clock = MLX5_CAP_GEN(mdev, device_frequency_khz); props->timestamp_mask = 0x7FFFFFFFFFFFFFFFULL; #ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING if (MLX5_CAP_GEN(mdev, pg)) props->device_cap_flags |= IB_DEVICE_ON_DEMAND_PAGING; props->odp_caps = dev->odp_caps; #endif if (MLX5_CAP_GEN(mdev, cd)) props->device_cap_flags |= IB_DEVICE_CROSS_CHANNEL; if (!mlx5_core_is_pf(mdev)) props->device_cap_flags |= IB_DEVICE_VIRTUAL_FUNCTION; if (mlx5_ib_port_link_layer(ibdev, 1) == IB_LINK_LAYER_ETHERNET) { props->rss_caps.max_rwq_indirection_tables = 1 << MLX5_CAP_GEN(dev->mdev, log_max_rqt); props->rss_caps.max_rwq_indirection_table_size = 1 << MLX5_CAP_GEN(dev->mdev, log_max_rqt_size); props->rss_caps.supported_qpts = 1 << IB_QPT_RAW_PACKET; props->max_wq_type_rq = 1 << MLX5_CAP_GEN(dev->mdev, log_max_rq); } if (uhw->outlen) { err = ib_copy_to_udata(uhw, &resp, resp.response_length); if (err) return err; } return 0; } enum mlx5_ib_width { MLX5_IB_WIDTH_1X = 1 << 0, MLX5_IB_WIDTH_2X = 1 << 1, MLX5_IB_WIDTH_4X = 1 << 2, MLX5_IB_WIDTH_8X = 1 << 3, MLX5_IB_WIDTH_12X = 1 << 4 }; static int translate_active_width(struct ib_device *ibdev, u8 active_width, u8 *ib_width) { struct mlx5_ib_dev *dev = to_mdev(ibdev); int err = 0; if (active_width & MLX5_IB_WIDTH_1X) { *ib_width = IB_WIDTH_1X; } else if (active_width & MLX5_IB_WIDTH_2X) { *ib_width = IB_WIDTH_2X; } else if (active_width & MLX5_IB_WIDTH_4X) { *ib_width = IB_WIDTH_4X; } else if (active_width & MLX5_IB_WIDTH_8X) { *ib_width = IB_WIDTH_8X; } else if (active_width & MLX5_IB_WIDTH_12X) { *ib_width = IB_WIDTH_12X; } else { mlx5_ib_dbg(dev, "Invalid active_width %d\n", (int)active_width); err = -EINVAL; } return err; } enum ib_max_vl_num { __IB_MAX_VL_0 = 1, __IB_MAX_VL_0_1 = 2, __IB_MAX_VL_0_3 = 3, __IB_MAX_VL_0_7 = 4, __IB_MAX_VL_0_14 = 5, }; enum mlx5_vl_hw_cap { MLX5_VL_HW_0 = 1, MLX5_VL_HW_0_1 = 2, MLX5_VL_HW_0_2 = 3, MLX5_VL_HW_0_3 = 4, MLX5_VL_HW_0_4 = 5, MLX5_VL_HW_0_5 = 6, MLX5_VL_HW_0_6 = 7, MLX5_VL_HW_0_7 = 8, MLX5_VL_HW_0_14 = 15 }; static int translate_max_vl_num(struct ib_device *ibdev, u8 vl_hw_cap, u8 *max_vl_num) { switch (vl_hw_cap) { case MLX5_VL_HW_0: *max_vl_num = __IB_MAX_VL_0; break; case MLX5_VL_HW_0_1: *max_vl_num = __IB_MAX_VL_0_1; break; case MLX5_VL_HW_0_3: *max_vl_num = __IB_MAX_VL_0_3; break; case MLX5_VL_HW_0_7: *max_vl_num = __IB_MAX_VL_0_7; break; case MLX5_VL_HW_0_14: *max_vl_num = __IB_MAX_VL_0_14; break; default: return -EINVAL; } return 0; } static int mlx5_query_hca_port(struct ib_device *ibdev, u8 port, struct ib_port_attr *props) { struct mlx5_ib_dev *dev = to_mdev(ibdev); struct mlx5_core_dev *mdev = dev->mdev; u32 *rep; int replen = MLX5_ST_SZ_BYTES(query_hca_vport_context_out); struct mlx5_ptys_reg *ptys; struct mlx5_pmtu_reg *pmtu; struct mlx5_pvlc_reg pvlc; void *ctx; int err; rep = mlx5_vzalloc(replen); ptys = kzalloc(sizeof(*ptys), GFP_KERNEL); pmtu = kzalloc(sizeof(*pmtu), GFP_KERNEL); if (!rep || !ptys || !pmtu) { err = -ENOMEM; goto out; } memset(props, 0, sizeof(*props)); err = mlx5_query_hca_vport_context(mdev, port, 0, rep, replen); if (err) goto out; ctx = MLX5_ADDR_OF(query_hca_vport_context_out, rep, hca_vport_context); props->lid = MLX5_GET(hca_vport_context, ctx, lid); props->lmc = MLX5_GET(hca_vport_context, ctx, lmc); props->sm_lid = MLX5_GET(hca_vport_context, ctx, sm_lid); props->sm_sl = MLX5_GET(hca_vport_context, ctx, sm_sl); props->state = MLX5_GET(hca_vport_context, ctx, vport_state); props->phys_state = MLX5_GET(hca_vport_context, ctx, port_physical_state); props->port_cap_flags = MLX5_GET(hca_vport_context, ctx, cap_mask1); props->gid_tbl_len = mlx5_get_gid_table_len(MLX5_CAP_GEN(mdev, gid_table_size)); props->max_msg_sz = 1 << MLX5_CAP_GEN(mdev, log_max_msg); props->pkey_tbl_len = mlx5_to_sw_pkey_sz(MLX5_CAP_GEN(mdev, pkey_table_size)); props->bad_pkey_cntr = MLX5_GET(hca_vport_context, ctx, pkey_violation_counter); props->qkey_viol_cntr = MLX5_GET(hca_vport_context, ctx, qkey_violation_counter); props->subnet_timeout = MLX5_GET(hca_vport_context, ctx, subnet_timeout); props->init_type_reply = MLX5_GET(hca_vport_context, ctx, init_type_reply); props->grh_required = MLX5_GET(hca_vport_context, ctx, grh_required); ptys->proto_mask |= MLX5_PTYS_IB; ptys->local_port = port; err = mlx5_core_access_ptys(mdev, ptys, 0); if (err) goto out; err = translate_active_width(ibdev, ptys->ib_link_width_oper, &props->active_width); if (err) goto out; props->active_speed = (u8)ptys->ib_proto_oper; pmtu->local_port = port; err = mlx5_core_access_pmtu(mdev, pmtu, 0); if (err) goto out; props->max_mtu = pmtu->max_mtu; props->active_mtu = pmtu->oper_mtu; memset(&pvlc, 0, sizeof(pvlc)); pvlc.local_port = port; err = mlx5_core_access_pvlc(mdev, &pvlc, 0); if (err) goto out; err = translate_max_vl_num(ibdev, pvlc.vl_hw_cap, &props->max_vl_num); out: kvfree(rep); kfree(ptys); kfree(pmtu); return err; } int mlx5_ib_query_port(struct ib_device *ibdev, u8 port, struct ib_port_attr *props) { switch (mlx5_get_vport_access_method(ibdev)) { case MLX5_VPORT_ACCESS_METHOD_MAD: return mlx5_query_mad_ifc_port(ibdev, port, props); case MLX5_VPORT_ACCESS_METHOD_HCA: return mlx5_query_hca_port(ibdev, port, props); case MLX5_VPORT_ACCESS_METHOD_NIC: return mlx5_query_port_roce(ibdev, port, props); default: return -EINVAL; } } static int mlx5_ib_query_gid(struct ib_device *ibdev, u8 port, int index, union ib_gid *gid) { struct mlx5_ib_dev *dev = to_mdev(ibdev); struct mlx5_core_dev *mdev = dev->mdev; switch (mlx5_get_vport_access_method(ibdev)) { case MLX5_VPORT_ACCESS_METHOD_MAD: return mlx5_query_mad_ifc_gids(ibdev, port, index, gid); case MLX5_VPORT_ACCESS_METHOD_HCA: return mlx5_query_hca_vport_gid(mdev, port, 0, index, gid); default: return -EINVAL; } } static int mlx5_ib_query_pkey(struct ib_device *ibdev, u8 port, u16 index, u16 *pkey) { struct mlx5_ib_dev *dev = to_mdev(ibdev); struct mlx5_core_dev *mdev = dev->mdev; switch (mlx5_get_vport_access_method(ibdev)) { case MLX5_VPORT_ACCESS_METHOD_MAD: return mlx5_query_mad_ifc_pkey(ibdev, port, index, pkey); case MLX5_VPORT_ACCESS_METHOD_HCA: case MLX5_VPORT_ACCESS_METHOD_NIC: return mlx5_query_hca_vport_pkey(mdev, 0, port, 0, index, pkey); default: return -EINVAL; } } static int mlx5_ib_modify_device(struct ib_device *ibdev, int mask, struct ib_device_modify *props) { struct mlx5_ib_dev *dev = to_mdev(ibdev); struct mlx5_reg_node_desc in; struct mlx5_reg_node_desc out; int err; if (mask & ~IB_DEVICE_MODIFY_NODE_DESC) return -EOPNOTSUPP; if (!(mask & IB_DEVICE_MODIFY_NODE_DESC)) return 0; /* * If possible, pass node desc to FW, so it can generate * a 144 trap. If cmd fails, just ignore. */ memcpy(&in, props->node_desc, IB_DEVICE_NODE_DESC_MAX); err = mlx5_core_access_reg(dev->mdev, &in, sizeof(in), &out, sizeof(out), MLX5_REG_NODE_DESC, 0, 1); if (err) return err; memcpy(ibdev->node_desc, props->node_desc, IB_DEVICE_NODE_DESC_MAX); return err; } static int mlx5_ib_modify_port(struct ib_device *ibdev, u8 port, int mask, struct ib_port_modify *props) { struct mlx5_ib_dev *dev = to_mdev(ibdev); struct ib_port_attr attr; u32 tmp; int err; /* * CM layer calls ib_modify_port() regardless of the link * layer. For Ethernet ports, qkey violation and Port * capabilities are meaningless. */ if (mlx5_ib_port_link_layer(ibdev, port) == IB_LINK_LAYER_ETHERNET) return 0; mutex_lock(&dev->cap_mask_mutex); err = mlx5_ib_query_port(ibdev, port, &attr); if (err) goto out; tmp = (attr.port_cap_flags | props->set_port_cap_mask) & ~props->clr_port_cap_mask; err = mlx5_set_port_caps(dev->mdev, port, tmp); out: mutex_unlock(&dev->cap_mask_mutex); return err; } static u16 calc_dynamic_bfregs(int uars_per_sys_page) { /* Large page with non 4k uar support might limit the dynamic size */ if (uars_per_sys_page == 1 && PAGE_SIZE > 4096) return MLX5_MIN_DYN_BFREGS; return MLX5_MAX_DYN_BFREGS; } static int calc_total_bfregs(struct mlx5_ib_dev *dev, bool lib_uar_4k, struct mlx5_ib_alloc_ucontext_req_v2 *req, struct mlx5_bfreg_info *bfregi) { int uars_per_sys_page; int bfregs_per_sys_page; int ref_bfregs = req->total_num_bfregs; if (req->total_num_bfregs == 0) return -EINVAL; BUILD_BUG_ON(MLX5_MAX_BFREGS % MLX5_NON_FP_BFREGS_IN_PAGE); BUILD_BUG_ON(MLX5_MAX_BFREGS < MLX5_NON_FP_BFREGS_IN_PAGE); if (req->total_num_bfregs > MLX5_MAX_BFREGS) return -ENOMEM; uars_per_sys_page = get_uars_per_sys_page(dev, lib_uar_4k); bfregs_per_sys_page = uars_per_sys_page * MLX5_NON_FP_BFREGS_PER_UAR; /* This holds the required static allocation asked by the user */ req->total_num_bfregs = ALIGN(req->total_num_bfregs, bfregs_per_sys_page); if (req->num_low_latency_bfregs > req->total_num_bfregs - 1) return -EINVAL; bfregi->num_static_sys_pages = req->total_num_bfregs / bfregs_per_sys_page; bfregi->num_dyn_bfregs = ALIGN(calc_dynamic_bfregs(uars_per_sys_page), bfregs_per_sys_page); bfregi->total_num_bfregs = req->total_num_bfregs + bfregi->num_dyn_bfregs; bfregi->num_sys_pages = bfregi->total_num_bfregs / bfregs_per_sys_page; mlx5_ib_dbg(dev, "uar_4k: fw support %s, lib support %s, user requested %d bfregs, allocated %d, total bfregs %d, using %d sys pages\n", MLX5_CAP_GEN(dev->mdev, uar_4k) ? "yes" : "no", lib_uar_4k ? "yes" : "no", ref_bfregs, req->total_num_bfregs, bfregi->total_num_bfregs, bfregi->num_sys_pages); return 0; } static int allocate_uars(struct mlx5_ib_dev *dev, struct mlx5_ib_ucontext *context) { struct mlx5_bfreg_info *bfregi; int err; int i; bfregi = &context->bfregi; for (i = 0; i < bfregi->num_static_sys_pages; i++) { err = mlx5_cmd_alloc_uar(dev->mdev, &bfregi->sys_pages[i]); if (err) goto error; mlx5_ib_dbg(dev, "allocated uar %d\n", bfregi->sys_pages[i]); } for (i = bfregi->num_static_sys_pages; i < bfregi->num_sys_pages; i++) bfregi->sys_pages[i] = MLX5_IB_INVALID_UAR_INDEX; return 0; error: for (--i; i >= 0; i--) if (mlx5_cmd_free_uar(dev->mdev, bfregi->sys_pages[i])) mlx5_ib_warn(dev, "failed to free uar %d\n", i); return err; } static void deallocate_uars(struct mlx5_ib_dev *dev, struct mlx5_ib_ucontext *context) { struct mlx5_bfreg_info *bfregi; int i; bfregi = &context->bfregi; for (i = 0; i < bfregi->num_sys_pages; i++) if (i < bfregi->num_static_sys_pages || bfregi->sys_pages[i] != MLX5_IB_INVALID_UAR_INDEX) mlx5_cmd_free_uar(dev->mdev, bfregi->sys_pages[i]); } static struct ib_ucontext *mlx5_ib_alloc_ucontext(struct ib_device *ibdev, struct ib_udata *udata) { struct mlx5_ib_dev *dev = to_mdev(ibdev); struct mlx5_ib_alloc_ucontext_req_v2 req = {}; struct mlx5_ib_alloc_ucontext_resp resp = {}; struct mlx5_ib_ucontext *context; struct mlx5_bfreg_info *bfregi; int ver; int err; size_t reqlen; size_t min_req_v2 = offsetof(struct mlx5_ib_alloc_ucontext_req_v2, max_cqe_version); bool lib_uar_4k; bool lib_uar_dyn; if (!dev->ib_active) return ERR_PTR(-EAGAIN); if (udata->inlen < sizeof(struct ib_uverbs_cmd_hdr)) return ERR_PTR(-EINVAL); reqlen = udata->inlen - sizeof(struct ib_uverbs_cmd_hdr); if (reqlen == sizeof(struct mlx5_ib_alloc_ucontext_req)) ver = 0; else if (reqlen >= min_req_v2) ver = 2; else return ERR_PTR(-EINVAL); err = ib_copy_from_udata(&req, udata, min(reqlen, sizeof(req))); if (err) return ERR_PTR(err); if (req.flags) return ERR_PTR(-EINVAL); if (req.comp_mask || req.reserved0 || req.reserved1 || req.reserved2) return ERR_PTR(-EOPNOTSUPP); req.total_num_bfregs = ALIGN(req.total_num_bfregs, MLX5_NON_FP_BFREGS_PER_UAR); if (req.num_low_latency_bfregs > req.total_num_bfregs - 1) return ERR_PTR(-EINVAL); if (reqlen > sizeof(req) && !ib_is_udata_cleared(udata, sizeof(req), reqlen - sizeof(req))) return ERR_PTR(-EOPNOTSUPP); resp.qp_tab_size = 1 << MLX5_CAP_GEN(dev->mdev, log_max_qp); if (mlx5_core_is_pf(dev->mdev) && MLX5_CAP_GEN(dev->mdev, bf)) resp.bf_reg_size = 1 << MLX5_CAP_GEN(dev->mdev, log_bf_reg_size); resp.cache_line_size = cache_line_size(); resp.max_sq_desc_sz = MLX5_CAP_GEN(dev->mdev, max_wqe_sz_sq); resp.max_rq_desc_sz = MLX5_CAP_GEN(dev->mdev, max_wqe_sz_rq); resp.max_send_wqebb = 1 << MLX5_CAP_GEN(dev->mdev, log_max_qp_sz); resp.max_recv_wr = 1 << MLX5_CAP_GEN(dev->mdev, log_max_qp_sz); resp.max_srq_recv_wr = 1 << MLX5_CAP_GEN(dev->mdev, log_max_srq_sz); resp.cqe_version = min_t(__u8, (__u8)MLX5_CAP_GEN(dev->mdev, cqe_version), req.max_cqe_version); resp.log_uar_size = MLX5_CAP_GEN(dev->mdev, uar_4k) ? MLX5_ADAPTER_PAGE_SHIFT : PAGE_SHIFT; resp.num_uars_per_page = MLX5_CAP_GEN(dev->mdev, uar_4k) ? MLX5_CAP_GEN(dev->mdev, num_of_uars_per_page) : 1; resp.response_length = min(offsetof(typeof(resp), response_length) + sizeof(resp.response_length), udata->outlen); context = kzalloc(sizeof(*context), GFP_KERNEL); if (!context) return ERR_PTR(-ENOMEM); lib_uar_4k = req.lib_caps & MLX5_LIB_CAP_4K_UAR; lib_uar_dyn = req.lib_caps & MLX5_LIB_CAP_DYN_UAR; bfregi = &context->bfregi; if (lib_uar_dyn) { bfregi->lib_uar_dyn = lib_uar_dyn; goto uar_done; } /* updates req->total_num_bfregs */ err = calc_total_bfregs(dev, lib_uar_4k, &req, bfregi); if (err) goto out_ctx; mutex_init(&bfregi->lock); bfregi->lib_uar_4k = lib_uar_4k; bfregi->count = kcalloc(bfregi->total_num_bfregs, sizeof(*bfregi->count), GFP_KERNEL); if (!bfregi->count) { err = -ENOMEM; goto out_ctx; } bfregi->sys_pages = kcalloc(bfregi->num_sys_pages, sizeof(*bfregi->sys_pages), GFP_KERNEL); if (!bfregi->sys_pages) { err = -ENOMEM; goto out_count; } err = allocate_uars(dev, context); if (err) goto out_sys_pages; uar_done: #ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING context->ibucontext.invalidate_range = &mlx5_ib_invalidate_range; #endif if (MLX5_CAP_GEN(dev->mdev, log_max_transport_domain)) { err = mlx5_alloc_transport_domain(dev->mdev, &context->tdn); if (err) goto out_uars; } INIT_LIST_HEAD(&context->vma_private_list); INIT_LIST_HEAD(&context->db_page_list); mutex_init(&context->db_page_mutex); resp.tot_bfregs = lib_uar_dyn ? 0 : req.total_num_bfregs; resp.num_ports = MLX5_CAP_GEN(dev->mdev, num_ports); if (field_avail(typeof(resp), cqe_version, udata->outlen)) resp.response_length += sizeof(resp.cqe_version); if (field_avail(typeof(resp), cmds_supp_uhw, udata->outlen)) { resp.cmds_supp_uhw |= MLX5_USER_CMDS_SUPP_UHW_QUERY_DEVICE | MLX5_USER_CMDS_SUPP_UHW_CREATE_AH; resp.response_length += sizeof(resp.cmds_supp_uhw); } /* * We don't want to expose information from the PCI bar that is located * after 4096 bytes, so if the arch only supports larger pages, let's * pretend we don't support reading the HCA's core clock. This is also * forced by mmap function. */ if (offsetofend(typeof(resp), hca_core_clock_offset) <= udata->outlen) { if (PAGE_SIZE <= 4096) { resp.comp_mask |= MLX5_IB_ALLOC_UCONTEXT_RESP_MASK_CORE_CLOCK_OFFSET; resp.hca_core_clock_offset = offsetof(struct mlx5_init_seg, internal_timer_h) % PAGE_SIZE; } resp.response_length += sizeof(resp.hca_core_clock_offset); } if (offsetofend(typeof(resp), log_uar_size) <= udata->outlen) resp.response_length += sizeof(resp.log_uar_size); if (offsetofend(typeof(resp), num_uars_per_page) <= udata->outlen) resp.response_length += sizeof(resp.num_uars_per_page); if (offsetofend(typeof(resp), num_dyn_bfregs) <= udata->outlen) { resp.num_dyn_bfregs = bfregi->num_dyn_bfregs; resp.response_length += sizeof(resp.num_dyn_bfregs); } err = ib_copy_to_udata(udata, &resp, resp.response_length); if (err) goto out_td; bfregi->ver = ver; bfregi->num_low_latency_bfregs = req.num_low_latency_bfregs; context->cqe_version = resp.cqe_version; return &context->ibucontext; out_td: if (MLX5_CAP_GEN(dev->mdev, log_max_transport_domain)) mlx5_dealloc_transport_domain(dev->mdev, context->tdn); out_uars: deallocate_uars(dev, context); out_sys_pages: kfree(bfregi->sys_pages); out_count: kfree(bfregi->count); out_ctx: kfree(context); return ERR_PTR(err); } static int mlx5_ib_dealloc_ucontext(struct ib_ucontext *ibcontext) { struct mlx5_ib_ucontext *context = to_mucontext(ibcontext); struct mlx5_ib_dev *dev = to_mdev(ibcontext->device); struct mlx5_bfreg_info *bfregi; bfregi = &context->bfregi; if (MLX5_CAP_GEN(dev->mdev, log_max_transport_domain)) mlx5_dealloc_transport_domain(dev->mdev, context->tdn); deallocate_uars(dev, context); kfree(bfregi->sys_pages); kfree(bfregi->count); kfree(context); return 0; } static phys_addr_t uar_index2pfn(struct mlx5_ib_dev *dev, int uar_idx) { int fw_uars_per_page; fw_uars_per_page = MLX5_CAP_GEN(dev->mdev, uar_4k) ? MLX5_UARS_IN_PAGE : 1; return (pci_resource_start(dev->mdev->pdev, 0) >> PAGE_SHIFT) + uar_idx / fw_uars_per_page; } static int get_command(unsigned long offset) { return (offset >> MLX5_IB_MMAP_CMD_SHIFT) & MLX5_IB_MMAP_CMD_MASK; } static int get_arg(unsigned long offset) { return offset & ((1 << MLX5_IB_MMAP_CMD_SHIFT) - 1); } static int get_index(unsigned long offset) { return get_arg(offset); } /* Index resides in an extra byte to enable larger values than 255 */ static int get_extended_index(unsigned long offset) { return get_arg(offset) | ((offset >> 16) & 0xff) << 8; } static void mlx5_ib_vma_open(struct vm_area_struct *area) { /* vma_open is called when a new VMA is created on top of our VMA. This * is done through either mremap flow or split_vma (usually due to * mlock, madvise, munmap, etc.) We do not support a clone of the VMA, * as this VMA is strongly hardware related. Therefore we set the * vm_ops of the newly created/cloned VMA to NULL, to prevent it from * calling us again and trying to do incorrect actions. We assume that * the original VMA size is exactly a single page, and therefore all * "splitting" operation will not happen to it. */ area->vm_ops = NULL; } static void mlx5_ib_vma_close(struct vm_area_struct *area) { struct mlx5_ib_vma_private_data *mlx5_ib_vma_priv_data; /* It's guaranteed that all VMAs opened on a FD are closed before the * file itself is closed, therefore no sync is needed with the regular * closing flow. (e.g. mlx5 ib_dealloc_ucontext) * However need a sync with accessing the vma as part of * mlx5_ib_disassociate_ucontext. * The close operation is usually called under mm->mmap_sem except when * process is exiting. * The exiting case is handled explicitly as part of * mlx5_ib_disassociate_ucontext. */ mlx5_ib_vma_priv_data = (struct mlx5_ib_vma_private_data *)area->vm_private_data; /* setting the vma context pointer to null in the mlx5_ib driver's * private data, to protect a race condition in * mlx5_ib_disassociate_ucontext(). */ mlx5_ib_vma_priv_data->vma = NULL; list_del(&mlx5_ib_vma_priv_data->list); kfree(mlx5_ib_vma_priv_data); } static const struct vm_operations_struct mlx5_ib_vm_ops = { .open = mlx5_ib_vma_open, .close = mlx5_ib_vma_close }; static int mlx5_ib_set_vma_data(struct vm_area_struct *vma, struct mlx5_ib_ucontext *ctx) { struct mlx5_ib_vma_private_data *vma_prv; struct list_head *vma_head = &ctx->vma_private_list; vma_prv = kzalloc(sizeof(*vma_prv), GFP_KERNEL); if (!vma_prv) return -ENOMEM; vma_prv->vma = vma; vma->vm_private_data = vma_prv; vma->vm_ops = &mlx5_ib_vm_ops; list_add(&vma_prv->list, vma_head); return 0; } static void mlx5_ib_disassociate_ucontext(struct ib_ucontext *ibcontext) { int ret; struct vm_area_struct *vma; struct mlx5_ib_vma_private_data *vma_private, *n; struct mlx5_ib_ucontext *context = to_mucontext(ibcontext); struct task_struct *owning_process = NULL; struct mm_struct *owning_mm = NULL; owning_process = get_pid_task(ibcontext->tgid, PIDTYPE_PID); if (!owning_process) return; owning_mm = get_task_mm(owning_process); if (!owning_mm) { pr_info("no mm, disassociate ucontext is pending task termination\n"); while (1) { put_task_struct(owning_process); usleep_range(1000, 2000); owning_process = get_pid_task(ibcontext->tgid, PIDTYPE_PID); if (!owning_process || owning_process->task_thread-> td_proc->p_state == PRS_ZOMBIE) { pr_info("disassociate ucontext done, task was terminated\n"); /* in case task was dead need to release the * task struct. */ if (owning_process) put_task_struct(owning_process); return; } } } /* need to protect from a race on closing the vma as part of * mlx5_ib_vma_close. */ down_write(&owning_mm->mmap_sem); list_for_each_entry_safe(vma_private, n, &context->vma_private_list, list) { vma = vma_private->vma; ret = zap_vma_ptes(vma, vma->vm_start, PAGE_SIZE); if (ret == -ENOTSUP) { if (bootverbose) WARN_ONCE( "%s: zap_vma_ptes not implemented for unmanaged mappings", __func__); } else { WARN(ret, "%s: zap_vma_ptes failed, error %d", __func__, -ret); } /* context going to be destroyed, should * not access ops any more. */ /* XXXKIB vma->vm_flags &= ~(VM_SHARED | VM_MAYSHARE); */ vma->vm_ops = NULL; list_del(&vma_private->list); kfree(vma_private); } up_write(&owning_mm->mmap_sem); mmput(owning_mm); put_task_struct(owning_process); } static inline char *mmap_cmd2str(enum mlx5_ib_mmap_cmd cmd) { switch (cmd) { case MLX5_IB_MMAP_WC_PAGE: return "WC"; case MLX5_IB_MMAP_REGULAR_PAGE: return "best effort WC"; case MLX5_IB_MMAP_NC_PAGE: return "NC"; default: return NULL; } } static int uar_mmap(struct mlx5_ib_dev *dev, enum mlx5_ib_mmap_cmd cmd, struct vm_area_struct *vma, struct mlx5_ib_ucontext *context) { struct mlx5_bfreg_info *bfregi = &context->bfregi; int err; unsigned long idx; phys_addr_t pfn; pgprot_t prot; u32 bfreg_dyn_idx = 0; u32 uar_index; int dyn_uar = (cmd == MLX5_IB_MMAP_WC_PAGE); int max_valid_idx = dyn_uar ? bfregi->num_sys_pages : bfregi->num_static_sys_pages; if (bfregi->lib_uar_dyn) return -EINVAL; if (vma->vm_end - vma->vm_start != PAGE_SIZE) return -EINVAL; if (dyn_uar) idx = get_extended_index(vma->vm_pgoff) + bfregi->num_static_sys_pages; else idx = get_index(vma->vm_pgoff); if (idx >= max_valid_idx) { mlx5_ib_warn(dev, "invalid uar index %lu, max=%d\n", idx, max_valid_idx); return -EINVAL; } switch (cmd) { case MLX5_IB_MMAP_WC_PAGE: case MLX5_IB_MMAP_REGULAR_PAGE: /* For MLX5_IB_MMAP_REGULAR_PAGE do the best effort to get WC */ prot = pgprot_writecombine(vma->vm_page_prot); break; case MLX5_IB_MMAP_NC_PAGE: prot = pgprot_noncached(vma->vm_page_prot); break; default: return -EINVAL; } if (dyn_uar) { int uars_per_page; uars_per_page = get_uars_per_sys_page(dev, bfregi->lib_uar_4k); bfreg_dyn_idx = idx * (uars_per_page * MLX5_NON_FP_BFREGS_PER_UAR); if (bfreg_dyn_idx >= bfregi->total_num_bfregs) { mlx5_ib_warn(dev, "invalid bfreg_dyn_idx %u, max=%u\n", bfreg_dyn_idx, bfregi->total_num_bfregs); return -EINVAL; } mutex_lock(&bfregi->lock); /* Fail if uar already allocated, first bfreg index of each * page holds its count. */ if (bfregi->count[bfreg_dyn_idx]) { mlx5_ib_warn(dev, "wrong offset, idx %lu is busy, bfregn=%u\n", idx, bfreg_dyn_idx); mutex_unlock(&bfregi->lock); return -EINVAL; } bfregi->count[bfreg_dyn_idx]++; mutex_unlock(&bfregi->lock); err = mlx5_cmd_alloc_uar(dev->mdev, &uar_index); if (err) { mlx5_ib_warn(dev, "UAR alloc failed\n"); goto free_bfreg; } } else { uar_index = bfregi->sys_pages[idx]; } pfn = uar_index2pfn(dev, uar_index); mlx5_ib_dbg(dev, "uar idx 0x%lx, pfn %pa\n", idx, &pfn); vma->vm_page_prot = prot; err = io_remap_pfn_range(vma, vma->vm_start, pfn, PAGE_SIZE, vma->vm_page_prot); if (err) { mlx5_ib_err(dev, "io_remap_pfn_range failed with error=%d, vm_start=0x%llx, pfn=%pa, mmap_cmd=%s\n", err, (unsigned long long)vma->vm_start, &pfn, mmap_cmd2str(cmd)); goto err; } if (dyn_uar) bfregi->sys_pages[idx] = uar_index; return mlx5_ib_set_vma_data(vma, context); err: if (!dyn_uar) return err; mlx5_cmd_free_uar(dev->mdev, idx); free_bfreg: mlx5_ib_free_bfreg(dev, bfregi, bfreg_dyn_idx); return err; } static int mlx5_ib_mmap(struct ib_ucontext *ibcontext, struct vm_area_struct *vma) { struct mlx5_ib_ucontext *context = to_mucontext(ibcontext); struct mlx5_ib_dev *dev = to_mdev(ibcontext->device); unsigned long command; phys_addr_t pfn; command = get_command(vma->vm_pgoff); switch (command) { case MLX5_IB_MMAP_WC_PAGE: case MLX5_IB_MMAP_NC_PAGE: case MLX5_IB_MMAP_REGULAR_PAGE: return uar_mmap(dev, command, vma, context); case MLX5_IB_MMAP_GET_CONTIGUOUS_PAGES: return -ENOSYS; case MLX5_IB_MMAP_CORE_CLOCK: if (vma->vm_end - vma->vm_start != PAGE_SIZE) return -EINVAL; if (vma->vm_flags & VM_WRITE) return -EPERM; /* Don't expose to user-space information it shouldn't have */ if (PAGE_SIZE > 4096) return -EOPNOTSUPP; vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot); pfn = (dev->mdev->iseg_base + offsetof(struct mlx5_init_seg, internal_timer_h)) >> PAGE_SHIFT; if (io_remap_pfn_range(vma, vma->vm_start, pfn, PAGE_SIZE, vma->vm_page_prot)) return -EAGAIN; mlx5_ib_dbg(dev, "mapped internal timer at 0x%llx, PA 0x%llx\n", (unsigned long long)vma->vm_start, (unsigned long long)pfn << PAGE_SHIFT); break; default: return -EINVAL; } return 0; } static struct ib_pd *mlx5_ib_alloc_pd(struct ib_device *ibdev, struct ib_ucontext *context, struct ib_udata *udata) { struct mlx5_ib_alloc_pd_resp resp; struct mlx5_ib_pd *pd; int err; pd = kmalloc(sizeof(*pd), GFP_KERNEL); if (!pd) return ERR_PTR(-ENOMEM); err = mlx5_core_alloc_pd(to_mdev(ibdev)->mdev, &pd->pdn); if (err) { kfree(pd); return ERR_PTR(err); } if (context) { resp.pdn = pd->pdn; if (ib_copy_to_udata(udata, &resp, sizeof(resp))) { mlx5_core_dealloc_pd(to_mdev(ibdev)->mdev, pd->pdn); kfree(pd); return ERR_PTR(-EFAULT); } } return &pd->ibpd; } static int mlx5_ib_dealloc_pd(struct ib_pd *pd) { struct mlx5_ib_dev *mdev = to_mdev(pd->device); struct mlx5_ib_pd *mpd = to_mpd(pd); mlx5_core_dealloc_pd(mdev->mdev, mpd->pdn); kfree(mpd); return 0; } enum { MATCH_CRITERIA_ENABLE_OUTER_BIT, MATCH_CRITERIA_ENABLE_MISC_BIT, MATCH_CRITERIA_ENABLE_INNER_BIT }; #define HEADER_IS_ZERO(match_criteria, headers) \ !(memchr_inv(MLX5_ADDR_OF(fte_match_param, match_criteria, headers), \ 0, MLX5_FLD_SZ_BYTES(fte_match_param, headers))) \ static u8 get_match_criteria_enable(u32 *match_criteria) { u8 match_criteria_enable; match_criteria_enable = (!HEADER_IS_ZERO(match_criteria, outer_headers)) << MATCH_CRITERIA_ENABLE_OUTER_BIT; match_criteria_enable |= (!HEADER_IS_ZERO(match_criteria, misc_parameters)) << MATCH_CRITERIA_ENABLE_MISC_BIT; match_criteria_enable |= (!HEADER_IS_ZERO(match_criteria, inner_headers)) << MATCH_CRITERIA_ENABLE_INNER_BIT; return match_criteria_enable; } static void set_proto(void *outer_c, void *outer_v, u8 mask, u8 val) { MLX5_SET(fte_match_set_lyr_2_4, outer_c, ip_protocol, mask); MLX5_SET(fte_match_set_lyr_2_4, outer_v, ip_protocol, val); } static void set_tos(void *outer_c, void *outer_v, u8 mask, u8 val) { MLX5_SET(fte_match_set_lyr_2_4, outer_c, ip_ecn, mask); MLX5_SET(fte_match_set_lyr_2_4, outer_v, ip_ecn, val); MLX5_SET(fte_match_set_lyr_2_4, outer_c, ip_dscp, mask >> 2); MLX5_SET(fte_match_set_lyr_2_4, outer_v, ip_dscp, val >> 2); } #define LAST_ETH_FIELD vlan_tag #define LAST_IB_FIELD sl #define LAST_IPV4_FIELD tos #define LAST_IPV6_FIELD traffic_class #define LAST_TCP_UDP_FIELD src_port /* Field is the last supported field */ #define FIELDS_NOT_SUPPORTED(filter, field)\ memchr_inv((void *)&filter.field +\ sizeof(filter.field), 0,\ sizeof(filter) -\ offsetof(typeof(filter), field) -\ sizeof(filter.field)) static int parse_flow_attr(u32 *match_c, u32 *match_v, const union ib_flow_spec *ib_spec) { void *outer_headers_c = MLX5_ADDR_OF(fte_match_param, match_c, outer_headers); void *outer_headers_v = MLX5_ADDR_OF(fte_match_param, match_v, outer_headers); void *misc_params_c = MLX5_ADDR_OF(fte_match_param, match_c, misc_parameters); void *misc_params_v = MLX5_ADDR_OF(fte_match_param, match_v, misc_parameters); switch (ib_spec->type) { case IB_FLOW_SPEC_ETH: if (FIELDS_NOT_SUPPORTED(ib_spec->eth.mask, LAST_ETH_FIELD)) return -ENOTSUPP; ether_addr_copy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, outer_headers_c, dmac_47_16), ib_spec->eth.mask.dst_mac); ether_addr_copy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, outer_headers_v, dmac_47_16), ib_spec->eth.val.dst_mac); ether_addr_copy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, outer_headers_c, smac_47_16), ib_spec->eth.mask.src_mac); ether_addr_copy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, outer_headers_v, smac_47_16), ib_spec->eth.val.src_mac); if (ib_spec->eth.mask.vlan_tag) { MLX5_SET(fte_match_set_lyr_2_4, outer_headers_c, cvlan_tag, 1); MLX5_SET(fte_match_set_lyr_2_4, outer_headers_v, cvlan_tag, 1); MLX5_SET(fte_match_set_lyr_2_4, outer_headers_c, first_vid, ntohs(ib_spec->eth.mask.vlan_tag)); MLX5_SET(fte_match_set_lyr_2_4, outer_headers_v, first_vid, ntohs(ib_spec->eth.val.vlan_tag)); MLX5_SET(fte_match_set_lyr_2_4, outer_headers_c, first_cfi, ntohs(ib_spec->eth.mask.vlan_tag) >> 12); MLX5_SET(fte_match_set_lyr_2_4, outer_headers_v, first_cfi, ntohs(ib_spec->eth.val.vlan_tag) >> 12); MLX5_SET(fte_match_set_lyr_2_4, outer_headers_c, first_prio, ntohs(ib_spec->eth.mask.vlan_tag) >> 13); MLX5_SET(fte_match_set_lyr_2_4, outer_headers_v, first_prio, ntohs(ib_spec->eth.val.vlan_tag) >> 13); } MLX5_SET(fte_match_set_lyr_2_4, outer_headers_c, ethertype, ntohs(ib_spec->eth.mask.ether_type)); MLX5_SET(fte_match_set_lyr_2_4, outer_headers_v, ethertype, ntohs(ib_spec->eth.val.ether_type)); break; case IB_FLOW_SPEC_IPV4: if (FIELDS_NOT_SUPPORTED(ib_spec->ipv4.mask, LAST_IPV4_FIELD)) return -ENOTSUPP; MLX5_SET(fte_match_set_lyr_2_4, outer_headers_c, ethertype, 0xffff); MLX5_SET(fte_match_set_lyr_2_4, outer_headers_v, ethertype, ETH_P_IP); memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, outer_headers_c, src_ipv4_src_ipv6.ipv4_layout.ipv4), &ib_spec->ipv4.mask.src_ip, sizeof(ib_spec->ipv4.mask.src_ip)); memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, outer_headers_v, src_ipv4_src_ipv6.ipv4_layout.ipv4), &ib_spec->ipv4.val.src_ip, sizeof(ib_spec->ipv4.val.src_ip)); memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, outer_headers_c, dst_ipv4_dst_ipv6.ipv4_layout.ipv4), &ib_spec->ipv4.mask.dst_ip, sizeof(ib_spec->ipv4.mask.dst_ip)); memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, outer_headers_v, dst_ipv4_dst_ipv6.ipv4_layout.ipv4), &ib_spec->ipv4.val.dst_ip, sizeof(ib_spec->ipv4.val.dst_ip)); set_tos(outer_headers_c, outer_headers_v, ib_spec->ipv4.mask.tos, ib_spec->ipv4.val.tos); set_proto(outer_headers_c, outer_headers_v, ib_spec->ipv4.mask.proto, ib_spec->ipv4.val.proto); break; case IB_FLOW_SPEC_IPV6: if (FIELDS_NOT_SUPPORTED(ib_spec->ipv6.mask, LAST_IPV6_FIELD)) return -ENOTSUPP; MLX5_SET(fte_match_set_lyr_2_4, outer_headers_c, ethertype, 0xffff); MLX5_SET(fte_match_set_lyr_2_4, outer_headers_v, ethertype, IPPROTO_IPV6); memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, outer_headers_c, src_ipv4_src_ipv6.ipv6_layout.ipv6), &ib_spec->ipv6.mask.src_ip, sizeof(ib_spec->ipv6.mask.src_ip)); memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, outer_headers_v, src_ipv4_src_ipv6.ipv6_layout.ipv6), &ib_spec->ipv6.val.src_ip, sizeof(ib_spec->ipv6.val.src_ip)); memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, outer_headers_c, dst_ipv4_dst_ipv6.ipv6_layout.ipv6), &ib_spec->ipv6.mask.dst_ip, sizeof(ib_spec->ipv6.mask.dst_ip)); memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, outer_headers_v, dst_ipv4_dst_ipv6.ipv6_layout.ipv6), &ib_spec->ipv6.val.dst_ip, sizeof(ib_spec->ipv6.val.dst_ip)); set_tos(outer_headers_c, outer_headers_v, ib_spec->ipv6.mask.traffic_class, ib_spec->ipv6.val.traffic_class); set_proto(outer_headers_c, outer_headers_v, ib_spec->ipv6.mask.next_hdr, ib_spec->ipv6.val.next_hdr); MLX5_SET(fte_match_set_misc, misc_params_c, outer_ipv6_flow_label, ntohl(ib_spec->ipv6.mask.flow_label)); MLX5_SET(fte_match_set_misc, misc_params_v, outer_ipv6_flow_label, ntohl(ib_spec->ipv6.val.flow_label)); break; case IB_FLOW_SPEC_TCP: if (FIELDS_NOT_SUPPORTED(ib_spec->tcp_udp.mask, LAST_TCP_UDP_FIELD)) return -ENOTSUPP; MLX5_SET(fte_match_set_lyr_2_4, outer_headers_c, ip_protocol, 0xff); MLX5_SET(fte_match_set_lyr_2_4, outer_headers_v, ip_protocol, IPPROTO_TCP); MLX5_SET(fte_match_set_lyr_2_4, outer_headers_c, tcp_sport, ntohs(ib_spec->tcp_udp.mask.src_port)); MLX5_SET(fte_match_set_lyr_2_4, outer_headers_v, tcp_sport, ntohs(ib_spec->tcp_udp.val.src_port)); MLX5_SET(fte_match_set_lyr_2_4, outer_headers_c, tcp_dport, ntohs(ib_spec->tcp_udp.mask.dst_port)); MLX5_SET(fte_match_set_lyr_2_4, outer_headers_v, tcp_dport, ntohs(ib_spec->tcp_udp.val.dst_port)); break; case IB_FLOW_SPEC_UDP: if (FIELDS_NOT_SUPPORTED(ib_spec->tcp_udp.mask, LAST_TCP_UDP_FIELD)) return -ENOTSUPP; MLX5_SET(fte_match_set_lyr_2_4, outer_headers_c, ip_protocol, 0xff); MLX5_SET(fte_match_set_lyr_2_4, outer_headers_v, ip_protocol, IPPROTO_UDP); MLX5_SET(fte_match_set_lyr_2_4, outer_headers_c, udp_sport, ntohs(ib_spec->tcp_udp.mask.src_port)); MLX5_SET(fte_match_set_lyr_2_4, outer_headers_v, udp_sport, ntohs(ib_spec->tcp_udp.val.src_port)); MLX5_SET(fte_match_set_lyr_2_4, outer_headers_c, udp_dport, ntohs(ib_spec->tcp_udp.mask.dst_port)); MLX5_SET(fte_match_set_lyr_2_4, outer_headers_v, udp_dport, ntohs(ib_spec->tcp_udp.val.dst_port)); break; default: return -EINVAL; } return 0; } /* If a flow could catch both multicast and unicast packets, * it won't fall into the multicast flow steering table and this rule * could steal other multicast packets. */ static bool flow_is_multicast_only(struct ib_flow_attr *ib_attr) { struct ib_flow_spec_eth *eth_spec; if (ib_attr->type != IB_FLOW_ATTR_NORMAL || ib_attr->size < sizeof(struct ib_flow_attr) + sizeof(struct ib_flow_spec_eth) || ib_attr->num_of_specs < 1) return false; eth_spec = (struct ib_flow_spec_eth *)(ib_attr + 1); if (eth_spec->type != IB_FLOW_SPEC_ETH || eth_spec->size != sizeof(*eth_spec)) return false; return is_multicast_ether_addr(eth_spec->mask.dst_mac) && is_multicast_ether_addr(eth_spec->val.dst_mac); } static bool is_valid_attr(const struct ib_flow_attr *flow_attr) { union ib_flow_spec *ib_spec = (union ib_flow_spec *)(flow_attr + 1); bool has_ipv4_spec = false; bool eth_type_ipv4 = true; unsigned int spec_index; /* Validate that ethertype is correct */ for (spec_index = 0; spec_index < flow_attr->num_of_specs; spec_index++) { if (ib_spec->type == IB_FLOW_SPEC_ETH && ib_spec->eth.mask.ether_type) { if (!((ib_spec->eth.mask.ether_type == htons(0xffff)) && ib_spec->eth.val.ether_type == htons(ETH_P_IP))) eth_type_ipv4 = false; } else if (ib_spec->type == IB_FLOW_SPEC_IPV4) { has_ipv4_spec = true; } ib_spec = (void *)ib_spec + ib_spec->size; } return !has_ipv4_spec || eth_type_ipv4; } static void put_flow_table(struct mlx5_ib_dev *dev, struct mlx5_ib_flow_prio *prio, bool ft_added) { prio->refcount -= !!ft_added; if (!prio->refcount) { mlx5_destroy_flow_table(prio->flow_table); prio->flow_table = NULL; } } static int mlx5_ib_destroy_flow(struct ib_flow *flow_id) { struct mlx5_ib_dev *dev = to_mdev(flow_id->qp->device); struct mlx5_ib_flow_handler *handler = container_of(flow_id, struct mlx5_ib_flow_handler, ibflow); struct mlx5_ib_flow_handler *iter, *tmp; mutex_lock(&dev->flow_db.lock); list_for_each_entry_safe(iter, tmp, &handler->list, list) { mlx5_del_flow_rule(iter->rule); put_flow_table(dev, iter->prio, true); list_del(&iter->list); kfree(iter); } mlx5_del_flow_rule(handler->rule); put_flow_table(dev, handler->prio, true); mutex_unlock(&dev->flow_db.lock); kfree(handler); return 0; } static int ib_prio_to_core_prio(unsigned int priority, bool dont_trap) { priority *= 2; if (!dont_trap) priority++; return priority; } enum flow_table_type { MLX5_IB_FT_RX, MLX5_IB_FT_TX }; #define MLX5_FS_MAX_TYPES 10 #define MLX5_FS_MAX_ENTRIES 32000UL static struct mlx5_ib_flow_prio *get_flow_table(struct mlx5_ib_dev *dev, struct ib_flow_attr *flow_attr, enum flow_table_type ft_type) { bool dont_trap = flow_attr->flags & IB_FLOW_ATTR_FLAGS_DONT_TRAP; struct mlx5_flow_namespace *ns = NULL; struct mlx5_ib_flow_prio *prio; struct mlx5_flow_table *ft; int num_entries; int num_groups; int priority; int err = 0; if (flow_attr->type == IB_FLOW_ATTR_NORMAL) { if (flow_is_multicast_only(flow_attr) && !dont_trap) priority = MLX5_IB_FLOW_MCAST_PRIO; else priority = ib_prio_to_core_prio(flow_attr->priority, dont_trap); ns = mlx5_get_flow_namespace(dev->mdev, MLX5_FLOW_NAMESPACE_BYPASS); num_entries = MLX5_FS_MAX_ENTRIES; num_groups = MLX5_FS_MAX_TYPES; prio = &dev->flow_db.prios[priority]; } else if (flow_attr->type == IB_FLOW_ATTR_ALL_DEFAULT || flow_attr->type == IB_FLOW_ATTR_MC_DEFAULT) { ns = mlx5_get_flow_namespace(dev->mdev, MLX5_FLOW_NAMESPACE_LEFTOVERS); build_leftovers_ft_param("bypass", &priority, &num_entries, &num_groups); prio = &dev->flow_db.prios[MLX5_IB_FLOW_LEFTOVERS_PRIO]; } else if (flow_attr->type == IB_FLOW_ATTR_SNIFFER) { if (!MLX5_CAP_FLOWTABLE(dev->mdev, allow_sniffer_and_nic_rx_shared_tir)) return ERR_PTR(-ENOTSUPP); ns = mlx5_get_flow_namespace(dev->mdev, ft_type == MLX5_IB_FT_RX ? MLX5_FLOW_NAMESPACE_SNIFFER_RX : MLX5_FLOW_NAMESPACE_SNIFFER_TX); prio = &dev->flow_db.sniffer[ft_type]; priority = 0; num_entries = 1; num_groups = 1; } if (!ns) return ERR_PTR(-ENOTSUPP); ft = prio->flow_table; if (!ft) { ft = mlx5_create_auto_grouped_flow_table(ns, priority, "bypass", num_entries, num_groups); if (!IS_ERR(ft)) { prio->refcount = 0; prio->flow_table = ft; } else { err = PTR_ERR(ft); } } return err ? ERR_PTR(err) : prio; } static struct mlx5_ib_flow_handler *create_flow_rule(struct mlx5_ib_dev *dev, struct mlx5_ib_flow_prio *ft_prio, const struct ib_flow_attr *flow_attr, struct mlx5_flow_destination *dst) { struct mlx5_flow_table *ft = ft_prio->flow_table; struct mlx5_ib_flow_handler *handler; struct mlx5_flow_spec *spec; const void *ib_flow = (const void *)flow_attr + sizeof(*flow_attr); unsigned int spec_index; u32 action; int err = 0; if (!is_valid_attr(flow_attr)) return ERR_PTR(-EINVAL); spec = mlx5_vzalloc(sizeof(*spec)); handler = kzalloc(sizeof(*handler), GFP_KERNEL); if (!handler || !spec) { err = -ENOMEM; goto free; } INIT_LIST_HEAD(&handler->list); for (spec_index = 0; spec_index < flow_attr->num_of_specs; spec_index++) { err = parse_flow_attr(spec->match_criteria, spec->match_value, ib_flow); if (err < 0) goto free; ib_flow += ((union ib_flow_spec *)ib_flow)->size; } spec->match_criteria_enable = get_match_criteria_enable(spec->match_criteria); action = dst ? MLX5_FLOW_CONTEXT_ACTION_FWD_DEST : MLX5_FLOW_CONTEXT_ACTION_FWD_NEXT_PRIO; handler->rule = mlx5_add_flow_rule(ft, spec->match_criteria_enable, spec->match_criteria, spec->match_value, action, MLX5_FS_DEFAULT_FLOW_TAG, dst); if (IS_ERR(handler->rule)) { err = PTR_ERR(handler->rule); goto free; } ft_prio->refcount++; handler->prio = ft_prio; ft_prio->flow_table = ft; free: if (err) kfree(handler); kvfree(spec); return err ? ERR_PTR(err) : handler; } static struct mlx5_ib_flow_handler *create_dont_trap_rule(struct mlx5_ib_dev *dev, struct mlx5_ib_flow_prio *ft_prio, struct ib_flow_attr *flow_attr, struct mlx5_flow_destination *dst) { struct mlx5_ib_flow_handler *handler_dst = NULL; struct mlx5_ib_flow_handler *handler = NULL; handler = create_flow_rule(dev, ft_prio, flow_attr, NULL); if (!IS_ERR(handler)) { handler_dst = create_flow_rule(dev, ft_prio, flow_attr, dst); if (IS_ERR(handler_dst)) { mlx5_del_flow_rule(handler->rule); ft_prio->refcount--; kfree(handler); handler = handler_dst; } else { list_add(&handler_dst->list, &handler->list); } } return handler; } enum { LEFTOVERS_MC, LEFTOVERS_UC, }; static struct mlx5_ib_flow_handler *create_leftovers_rule(struct mlx5_ib_dev *dev, struct mlx5_ib_flow_prio *ft_prio, struct ib_flow_attr *flow_attr, struct mlx5_flow_destination *dst) { struct mlx5_ib_flow_handler *handler_ucast = NULL; struct mlx5_ib_flow_handler *handler = NULL; static struct { struct ib_flow_attr flow_attr; struct ib_flow_spec_eth eth_flow; } leftovers_specs[] = { [LEFTOVERS_MC] = { .flow_attr = { .num_of_specs = 1, .size = sizeof(leftovers_specs[0]) }, .eth_flow = { .type = IB_FLOW_SPEC_ETH, .size = sizeof(struct ib_flow_spec_eth), .mask = {.dst_mac = {0x1} }, .val = {.dst_mac = {0x1} } } }, [LEFTOVERS_UC] = { .flow_attr = { .num_of_specs = 1, .size = sizeof(leftovers_specs[0]) }, .eth_flow = { .type = IB_FLOW_SPEC_ETH, .size = sizeof(struct ib_flow_spec_eth), .mask = {.dst_mac = {0x1} }, .val = {.dst_mac = {} } } } }; handler = create_flow_rule(dev, ft_prio, &leftovers_specs[LEFTOVERS_MC].flow_attr, dst); if (!IS_ERR(handler) && flow_attr->type == IB_FLOW_ATTR_ALL_DEFAULT) { handler_ucast = create_flow_rule(dev, ft_prio, &leftovers_specs[LEFTOVERS_UC].flow_attr, dst); if (IS_ERR(handler_ucast)) { mlx5_del_flow_rule(handler->rule); ft_prio->refcount--; kfree(handler); handler = handler_ucast; } else { list_add(&handler_ucast->list, &handler->list); } } return handler; } static struct mlx5_ib_flow_handler *create_sniffer_rule(struct mlx5_ib_dev *dev, struct mlx5_ib_flow_prio *ft_rx, struct mlx5_ib_flow_prio *ft_tx, struct mlx5_flow_destination *dst) { struct mlx5_ib_flow_handler *handler_rx; struct mlx5_ib_flow_handler *handler_tx; int err; static const struct ib_flow_attr flow_attr = { .num_of_specs = 0, .size = sizeof(flow_attr) }; handler_rx = create_flow_rule(dev, ft_rx, &flow_attr, dst); if (IS_ERR(handler_rx)) { err = PTR_ERR(handler_rx); goto err; } handler_tx = create_flow_rule(dev, ft_tx, &flow_attr, dst); if (IS_ERR(handler_tx)) { err = PTR_ERR(handler_tx); goto err_tx; } list_add(&handler_tx->list, &handler_rx->list); return handler_rx; err_tx: mlx5_del_flow_rule(handler_rx->rule); ft_rx->refcount--; kfree(handler_rx); err: return ERR_PTR(err); } static struct ib_flow *mlx5_ib_create_flow(struct ib_qp *qp, struct ib_flow_attr *flow_attr, int domain) { struct mlx5_ib_dev *dev = to_mdev(qp->device); struct mlx5_ib_qp *mqp = to_mqp(qp); struct mlx5_ib_flow_handler *handler = NULL; struct mlx5_flow_destination *dst = NULL; struct mlx5_ib_flow_prio *ft_prio_tx = NULL; struct mlx5_ib_flow_prio *ft_prio; int err; if (flow_attr->priority > MLX5_IB_FLOW_LAST_PRIO) return ERR_PTR(-ENOSPC); if (domain != IB_FLOW_DOMAIN_USER || flow_attr->port > MLX5_CAP_GEN(dev->mdev, num_ports) || (flow_attr->flags & ~IB_FLOW_ATTR_FLAGS_DONT_TRAP)) return ERR_PTR(-EINVAL); dst = kzalloc(sizeof(*dst), GFP_KERNEL); if (!dst) return ERR_PTR(-ENOMEM); mutex_lock(&dev->flow_db.lock); ft_prio = get_flow_table(dev, flow_attr, MLX5_IB_FT_RX); if (IS_ERR(ft_prio)) { err = PTR_ERR(ft_prio); goto unlock; } if (flow_attr->type == IB_FLOW_ATTR_SNIFFER) { ft_prio_tx = get_flow_table(dev, flow_attr, MLX5_IB_FT_TX); if (IS_ERR(ft_prio_tx)) { err = PTR_ERR(ft_prio_tx); ft_prio_tx = NULL; goto destroy_ft; } } dst->type = MLX5_FLOW_DESTINATION_TYPE_TIR; if (mqp->flags & MLX5_IB_QP_RSS) dst->tir_num = mqp->rss_qp.tirn; else dst->tir_num = mqp->raw_packet_qp.rq.tirn; if (flow_attr->type == IB_FLOW_ATTR_NORMAL) { if (flow_attr->flags & IB_FLOW_ATTR_FLAGS_DONT_TRAP) { handler = create_dont_trap_rule(dev, ft_prio, flow_attr, dst); } else { handler = create_flow_rule(dev, ft_prio, flow_attr, dst); } } else if (flow_attr->type == IB_FLOW_ATTR_ALL_DEFAULT || flow_attr->type == IB_FLOW_ATTR_MC_DEFAULT) { handler = create_leftovers_rule(dev, ft_prio, flow_attr, dst); } else if (flow_attr->type == IB_FLOW_ATTR_SNIFFER) { handler = create_sniffer_rule(dev, ft_prio, ft_prio_tx, dst); } else { err = -EINVAL; goto destroy_ft; } if (IS_ERR(handler)) { err = PTR_ERR(handler); handler = NULL; goto destroy_ft; } mutex_unlock(&dev->flow_db.lock); kfree(dst); return &handler->ibflow; destroy_ft: put_flow_table(dev, ft_prio, false); if (ft_prio_tx) put_flow_table(dev, ft_prio_tx, false); unlock: mutex_unlock(&dev->flow_db.lock); kfree(dst); kfree(handler); return ERR_PTR(err); } static int mlx5_ib_mcg_attach(struct ib_qp *ibqp, union ib_gid *gid, u16 lid) { struct mlx5_ib_dev *dev = to_mdev(ibqp->device); int err; err = mlx5_core_attach_mcg(dev->mdev, gid, ibqp->qp_num); if (err) mlx5_ib_warn(dev, "failed attaching QPN 0x%x, MGID %pI6\n", ibqp->qp_num, gid->raw); return err; } static int mlx5_ib_mcg_detach(struct ib_qp *ibqp, union ib_gid *gid, u16 lid) { struct mlx5_ib_dev *dev = to_mdev(ibqp->device); int err; err = mlx5_core_detach_mcg(dev->mdev, gid, ibqp->qp_num); if (err) mlx5_ib_warn(dev, "failed detaching QPN 0x%x, MGID %pI6\n", ibqp->qp_num, gid->raw); return err; } static int init_node_data(struct mlx5_ib_dev *dev) { int err; err = mlx5_query_node_desc(dev, dev->ib_dev.node_desc); if (err) return err; return mlx5_query_node_guid(dev, &dev->ib_dev.node_guid); } static ssize_t show_fw_pages(struct device *device, struct device_attribute *attr, char *buf) { struct mlx5_ib_dev *dev = container_of(device, struct mlx5_ib_dev, ib_dev.dev); return sprintf(buf, "%lld\n", (long long)dev->mdev->priv.fw_pages); } static ssize_t show_reg_pages(struct device *device, struct device_attribute *attr, char *buf) { struct mlx5_ib_dev *dev = container_of(device, struct mlx5_ib_dev, ib_dev.dev); return sprintf(buf, "%d\n", atomic_read(&dev->mdev->priv.reg_pages)); } static ssize_t show_hca(struct device *device, struct device_attribute *attr, char *buf) { struct mlx5_ib_dev *dev = container_of(device, struct mlx5_ib_dev, ib_dev.dev); return sprintf(buf, "MT%d\n", dev->mdev->pdev->device); } static ssize_t show_rev(struct device *device, struct device_attribute *attr, char *buf) { struct mlx5_ib_dev *dev = container_of(device, struct mlx5_ib_dev, ib_dev.dev); return sprintf(buf, "%x\n", dev->mdev->pdev->revision); } static ssize_t show_board(struct device *device, struct device_attribute *attr, char *buf) { struct mlx5_ib_dev *dev = container_of(device, struct mlx5_ib_dev, ib_dev.dev); return sprintf(buf, "%.*s\n", MLX5_BOARD_ID_LEN, dev->mdev->board_id); } static DEVICE_ATTR(hw_rev, S_IRUGO, show_rev, NULL); static DEVICE_ATTR(hca_type, S_IRUGO, show_hca, NULL); static DEVICE_ATTR(board_id, S_IRUGO, show_board, NULL); static DEVICE_ATTR(fw_pages, S_IRUGO, show_fw_pages, NULL); static DEVICE_ATTR(reg_pages, S_IRUGO, show_reg_pages, NULL); static struct device_attribute *mlx5_class_attributes[] = { &dev_attr_hw_rev, &dev_attr_hca_type, &dev_attr_board_id, &dev_attr_fw_pages, &dev_attr_reg_pages, }; static void pkey_change_handler(struct work_struct *work) { struct mlx5_ib_port_resources *ports = container_of(work, struct mlx5_ib_port_resources, pkey_change_work); mutex_lock(&ports->devr->mutex); mlx5_ib_gsi_pkey_change(ports->gsi); mutex_unlock(&ports->devr->mutex); } static void mlx5_ib_handle_internal_error(struct mlx5_ib_dev *ibdev) { struct mlx5_ib_qp *mqp; struct mlx5_ib_cq *send_mcq, *recv_mcq; struct mlx5_core_cq *mcq; struct list_head cq_armed_list; unsigned long flags_qp; unsigned long flags_cq; unsigned long flags; INIT_LIST_HEAD(&cq_armed_list); /* Go over qp list reside on that ibdev, sync with create/destroy qp.*/ spin_lock_irqsave(&ibdev->reset_flow_resource_lock, flags); list_for_each_entry(mqp, &ibdev->qp_list, qps_list) { spin_lock_irqsave(&mqp->sq.lock, flags_qp); if (mqp->sq.tail != mqp->sq.head) { send_mcq = to_mcq(mqp->ibqp.send_cq); spin_lock_irqsave(&send_mcq->lock, flags_cq); if (send_mcq->mcq.comp && mqp->ibqp.send_cq->comp_handler) { if (!send_mcq->mcq.reset_notify_added) { send_mcq->mcq.reset_notify_added = 1; list_add_tail(&send_mcq->mcq.reset_notify, &cq_armed_list); } } spin_unlock_irqrestore(&send_mcq->lock, flags_cq); } spin_unlock_irqrestore(&mqp->sq.lock, flags_qp); spin_lock_irqsave(&mqp->rq.lock, flags_qp); /* no handling is needed for SRQ */ if (!mqp->ibqp.srq) { if (mqp->rq.tail != mqp->rq.head) { recv_mcq = to_mcq(mqp->ibqp.recv_cq); spin_lock_irqsave(&recv_mcq->lock, flags_cq); if (recv_mcq->mcq.comp && mqp->ibqp.recv_cq->comp_handler) { if (!recv_mcq->mcq.reset_notify_added) { recv_mcq->mcq.reset_notify_added = 1; list_add_tail(&recv_mcq->mcq.reset_notify, &cq_armed_list); } } spin_unlock_irqrestore(&recv_mcq->lock, flags_cq); } } spin_unlock_irqrestore(&mqp->rq.lock, flags_qp); } /*At that point all inflight post send were put to be executed as of we * lock/unlock above locks Now need to arm all involved CQs. */ list_for_each_entry(mcq, &cq_armed_list, reset_notify) { mcq->comp(mcq, NULL); } spin_unlock_irqrestore(&ibdev->reset_flow_resource_lock, flags); } static void mlx5_ib_event(struct mlx5_core_dev *dev, void *context, enum mlx5_dev_event event, unsigned long param) { struct mlx5_ib_dev *ibdev = (struct mlx5_ib_dev *)context; struct ib_event ibev; bool fatal = false; u8 port = (u8)param; switch (event) { case MLX5_DEV_EVENT_SYS_ERROR: ibev.event = IB_EVENT_DEVICE_FATAL; mlx5_ib_handle_internal_error(ibdev); fatal = true; break; case MLX5_DEV_EVENT_PORT_UP: case MLX5_DEV_EVENT_PORT_DOWN: case MLX5_DEV_EVENT_PORT_INITIALIZED: /* In RoCE, port up/down events are handled in * mlx5_netdev_event(). */ if (mlx5_ib_port_link_layer(&ibdev->ib_dev, port) == IB_LINK_LAYER_ETHERNET) return; ibev.event = (event == MLX5_DEV_EVENT_PORT_UP) ? IB_EVENT_PORT_ACTIVE : IB_EVENT_PORT_ERR; break; case MLX5_DEV_EVENT_LID_CHANGE: ibev.event = IB_EVENT_LID_CHANGE; break; case MLX5_DEV_EVENT_PKEY_CHANGE: ibev.event = IB_EVENT_PKEY_CHANGE; schedule_work(&ibdev->devr.ports[port - 1].pkey_change_work); break; case MLX5_DEV_EVENT_GUID_CHANGE: ibev.event = IB_EVENT_GID_CHANGE; break; case MLX5_DEV_EVENT_CLIENT_REREG: ibev.event = IB_EVENT_CLIENT_REREGISTER; break; default: /* unsupported event */ return; } ibev.device = &ibdev->ib_dev; ibev.element.port_num = port; if (!rdma_is_port_valid(&ibdev->ib_dev, port)) { mlx5_ib_warn(ibdev, "warning: event(%d) on port %d\n", event, port); return; } if (ibdev->ib_active) ib_dispatch_event(&ibev); if (fatal) ibdev->ib_active = false; } static void get_ext_port_caps(struct mlx5_ib_dev *dev) { int port; for (port = 1; port <= MLX5_CAP_GEN(dev->mdev, num_ports); port++) mlx5_query_ext_port_caps(dev, port); } static int get_port_caps(struct mlx5_ib_dev *dev) { struct ib_device_attr *dprops = NULL; struct ib_port_attr *pprops = NULL; int err = -ENOMEM; int port; struct ib_udata uhw = {.inlen = 0, .outlen = 0}; pprops = kmalloc(sizeof(*pprops), GFP_KERNEL); if (!pprops) goto out; dprops = kmalloc(sizeof(*dprops), GFP_KERNEL); if (!dprops) goto out; err = mlx5_ib_query_device(&dev->ib_dev, dprops, &uhw); if (err) { mlx5_ib_warn(dev, "query_device failed %d\n", err); goto out; } for (port = 1; port <= MLX5_CAP_GEN(dev->mdev, num_ports); port++) { err = mlx5_ib_query_port(&dev->ib_dev, port, pprops); if (err) { mlx5_ib_warn(dev, "query_port %d failed %d\n", port, err); break; } dev->mdev->port_caps[port - 1].pkey_table_len = dprops->max_pkeys; dev->mdev->port_caps[port - 1].gid_table_len = pprops->gid_tbl_len; mlx5_ib_dbg(dev, "pkey_table_len %d, gid_table_len %d\n", dprops->max_pkeys, pprops->gid_tbl_len); } out: kfree(pprops); kfree(dprops); return err; } static void destroy_umrc_res(struct mlx5_ib_dev *dev) { int err; err = mlx5_mr_cache_cleanup(dev); if (err) mlx5_ib_warn(dev, "mr cache cleanup failed\n"); mlx5_ib_destroy_qp(dev->umrc.qp); ib_free_cq(dev->umrc.cq); ib_dealloc_pd(dev->umrc.pd); } enum { MAX_UMR_WR = 128, }; static int create_umr_res(struct mlx5_ib_dev *dev) { struct ib_qp_init_attr *init_attr = NULL; struct ib_qp_attr *attr = NULL; struct ib_pd *pd; struct ib_cq *cq; struct ib_qp *qp; int ret; attr = kzalloc(sizeof(*attr), GFP_KERNEL); init_attr = kzalloc(sizeof(*init_attr), GFP_KERNEL); if (!attr || !init_attr) { ret = -ENOMEM; goto error_0; } pd = ib_alloc_pd(&dev->ib_dev, 0); if (IS_ERR(pd)) { mlx5_ib_dbg(dev, "Couldn't create PD for sync UMR QP\n"); ret = PTR_ERR(pd); goto error_0; } cq = ib_alloc_cq(&dev->ib_dev, NULL, 128, 0, IB_POLL_SOFTIRQ); if (IS_ERR(cq)) { mlx5_ib_dbg(dev, "Couldn't create CQ for sync UMR QP\n"); ret = PTR_ERR(cq); goto error_2; } init_attr->send_cq = cq; init_attr->recv_cq = cq; init_attr->sq_sig_type = IB_SIGNAL_ALL_WR; init_attr->cap.max_send_wr = MAX_UMR_WR; init_attr->cap.max_send_sge = 1; init_attr->qp_type = MLX5_IB_QPT_REG_UMR; init_attr->port_num = 1; qp = mlx5_ib_create_qp(pd, init_attr, NULL); if (IS_ERR(qp)) { mlx5_ib_dbg(dev, "Couldn't create sync UMR QP\n"); ret = PTR_ERR(qp); goto error_3; } qp->device = &dev->ib_dev; qp->real_qp = qp; qp->uobject = NULL; qp->qp_type = MLX5_IB_QPT_REG_UMR; attr->qp_state = IB_QPS_INIT; attr->port_num = 1; ret = mlx5_ib_modify_qp(qp, attr, IB_QP_STATE | IB_QP_PKEY_INDEX | IB_QP_PORT, NULL); if (ret) { mlx5_ib_dbg(dev, "Couldn't modify UMR QP\n"); goto error_4; } memset(attr, 0, sizeof(*attr)); attr->qp_state = IB_QPS_RTR; attr->path_mtu = IB_MTU_256; ret = mlx5_ib_modify_qp(qp, attr, IB_QP_STATE, NULL); if (ret) { mlx5_ib_dbg(dev, "Couldn't modify umr QP to rtr\n"); goto error_4; } memset(attr, 0, sizeof(*attr)); attr->qp_state = IB_QPS_RTS; ret = mlx5_ib_modify_qp(qp, attr, IB_QP_STATE, NULL); if (ret) { mlx5_ib_dbg(dev, "Couldn't modify umr QP to rts\n"); goto error_4; } dev->umrc.qp = qp; dev->umrc.cq = cq; dev->umrc.pd = pd; sema_init(&dev->umrc.sem, MAX_UMR_WR); ret = mlx5_mr_cache_init(dev); if (ret) { mlx5_ib_warn(dev, "mr cache init failed %d\n", ret); goto error_4; } kfree(attr); kfree(init_attr); return 0; error_4: mlx5_ib_destroy_qp(qp); error_3: ib_free_cq(cq); error_2: ib_dealloc_pd(pd); error_0: kfree(attr); kfree(init_attr); return ret; } static int create_dev_resources(struct mlx5_ib_resources *devr) { struct ib_srq_init_attr attr; struct mlx5_ib_dev *dev; struct ib_cq_init_attr cq_attr = {.cqe = 1}; int port; int ret = 0; dev = container_of(devr, struct mlx5_ib_dev, devr); mutex_init(&devr->mutex); devr->p0 = mlx5_ib_alloc_pd(&dev->ib_dev, NULL, NULL); if (IS_ERR(devr->p0)) { ret = PTR_ERR(devr->p0); goto error0; } devr->p0->device = &dev->ib_dev; devr->p0->uobject = NULL; atomic_set(&devr->p0->usecnt, 0); devr->c0 = mlx5_ib_create_cq(&dev->ib_dev, &cq_attr, NULL, NULL); if (IS_ERR(devr->c0)) { ret = PTR_ERR(devr->c0); goto error1; } devr->c0->device = &dev->ib_dev; devr->c0->uobject = NULL; devr->c0->comp_handler = NULL; devr->c0->event_handler = NULL; devr->c0->cq_context = NULL; atomic_set(&devr->c0->usecnt, 0); devr->x0 = mlx5_ib_alloc_xrcd(&dev->ib_dev, NULL, NULL); if (IS_ERR(devr->x0)) { ret = PTR_ERR(devr->x0); goto error2; } devr->x0->device = &dev->ib_dev; devr->x0->inode = NULL; atomic_set(&devr->x0->usecnt, 0); mutex_init(&devr->x0->tgt_qp_mutex); INIT_LIST_HEAD(&devr->x0->tgt_qp_list); devr->x1 = mlx5_ib_alloc_xrcd(&dev->ib_dev, NULL, NULL); if (IS_ERR(devr->x1)) { ret = PTR_ERR(devr->x1); goto error3; } devr->x1->device = &dev->ib_dev; devr->x1->inode = NULL; atomic_set(&devr->x1->usecnt, 0); mutex_init(&devr->x1->tgt_qp_mutex); INIT_LIST_HEAD(&devr->x1->tgt_qp_list); memset(&attr, 0, sizeof(attr)); attr.attr.max_sge = 1; attr.attr.max_wr = 1; attr.srq_type = IB_SRQT_XRC; attr.ext.xrc.cq = devr->c0; attr.ext.xrc.xrcd = devr->x0; devr->s0 = mlx5_ib_create_srq(devr->p0, &attr, NULL); if (IS_ERR(devr->s0)) { ret = PTR_ERR(devr->s0); goto error4; } devr->s0->device = &dev->ib_dev; devr->s0->pd = devr->p0; devr->s0->uobject = NULL; devr->s0->event_handler = NULL; devr->s0->srq_context = NULL; devr->s0->srq_type = IB_SRQT_XRC; devr->s0->ext.xrc.xrcd = devr->x0; devr->s0->ext.xrc.cq = devr->c0; atomic_inc(&devr->s0->ext.xrc.xrcd->usecnt); atomic_inc(&devr->s0->ext.xrc.cq->usecnt); atomic_inc(&devr->p0->usecnt); atomic_set(&devr->s0->usecnt, 0); memset(&attr, 0, sizeof(attr)); attr.attr.max_sge = 1; attr.attr.max_wr = 1; attr.srq_type = IB_SRQT_BASIC; devr->s1 = mlx5_ib_create_srq(devr->p0, &attr, NULL); if (IS_ERR(devr->s1)) { ret = PTR_ERR(devr->s1); goto error5; } devr->s1->device = &dev->ib_dev; devr->s1->pd = devr->p0; devr->s1->uobject = NULL; devr->s1->event_handler = NULL; devr->s1->srq_context = NULL; devr->s1->srq_type = IB_SRQT_BASIC; devr->s1->ext.xrc.cq = devr->c0; atomic_inc(&devr->p0->usecnt); atomic_set(&devr->s0->usecnt, 0); for (port = 0; port < ARRAY_SIZE(devr->ports); ++port) { INIT_WORK(&devr->ports[port].pkey_change_work, pkey_change_handler); devr->ports[port].devr = devr; } return 0; error5: mlx5_ib_destroy_srq(devr->s0); error4: mlx5_ib_dealloc_xrcd(devr->x1); error3: mlx5_ib_dealloc_xrcd(devr->x0); error2: mlx5_ib_destroy_cq(devr->c0); error1: mlx5_ib_dealloc_pd(devr->p0); error0: return ret; } static void destroy_dev_resources(struct mlx5_ib_resources *devr) { struct mlx5_ib_dev *dev = container_of(devr, struct mlx5_ib_dev, devr); int port; mlx5_ib_destroy_srq(devr->s1); mlx5_ib_destroy_srq(devr->s0); mlx5_ib_dealloc_xrcd(devr->x0); mlx5_ib_dealloc_xrcd(devr->x1); mlx5_ib_destroy_cq(devr->c0); mlx5_ib_dealloc_pd(devr->p0); /* Make sure no change P_Key work items are still executing */ for (port = 0; port < dev->num_ports; ++port) cancel_work_sync(&devr->ports[port].pkey_change_work); } static u32 get_core_cap_flags(struct ib_device *ibdev) { struct mlx5_ib_dev *dev = to_mdev(ibdev); enum rdma_link_layer ll = mlx5_ib_port_link_layer(ibdev, 1); u8 l3_type_cap = MLX5_CAP_ROCE(dev->mdev, l3_type); u8 roce_version_cap = MLX5_CAP_ROCE(dev->mdev, roce_version); u32 ret = 0; if (ll == IB_LINK_LAYER_INFINIBAND) return RDMA_CORE_PORT_IBA_IB; if (!(l3_type_cap & MLX5_ROCE_L3_TYPE_IPV4_CAP)) return 0; if (!(l3_type_cap & MLX5_ROCE_L3_TYPE_IPV6_CAP)) return 0; if (roce_version_cap & MLX5_ROCE_VERSION_1_CAP) ret |= RDMA_CORE_PORT_IBA_ROCE; if (roce_version_cap & MLX5_ROCE_VERSION_2_CAP) ret |= RDMA_CORE_PORT_IBA_ROCE_UDP_ENCAP; return ret; } static int mlx5_port_immutable(struct ib_device *ibdev, u8 port_num, struct ib_port_immutable *immutable) { struct ib_port_attr attr; struct mlx5_ib_dev *dev = to_mdev(ibdev); enum rdma_link_layer ll = mlx5_ib_port_link_layer(ibdev, port_num); int err; err = mlx5_ib_query_port(ibdev, port_num, &attr); if (err) return err; immutable->pkey_tbl_len = attr.pkey_tbl_len; immutable->gid_tbl_len = attr.gid_tbl_len; immutable->core_cap_flags = get_core_cap_flags(ibdev); if ((ll == IB_LINK_LAYER_INFINIBAND) || MLX5_CAP_GEN(dev->mdev, roce)) immutable->max_mad_size = IB_MGMT_MAD_SIZE; return 0; } static void get_dev_fw_str(struct ib_device *ibdev, char *str, size_t str_len) { struct mlx5_ib_dev *dev = container_of(ibdev, struct mlx5_ib_dev, ib_dev); snprintf(str, str_len, "%d.%d.%04d", fw_rev_maj(dev->mdev), fw_rev_min(dev->mdev), fw_rev_sub(dev->mdev)); } static int mlx5_roce_lag_init(struct mlx5_ib_dev *dev) { return 0; } static void mlx5_roce_lag_cleanup(struct mlx5_ib_dev *dev) { } static void mlx5_remove_roce_notifier(struct mlx5_ib_dev *dev) { if (dev->roce.nb.notifier_call) { unregister_netdevice_notifier(&dev->roce.nb); dev->roce.nb.notifier_call = NULL; } } static int mlx5_enable_roce(struct mlx5_ib_dev *dev) { VNET_ITERATOR_DECL(vnet_iter); struct ifnet *idev; int err; /* Check if mlx5en net device already exists */ VNET_LIST_RLOCK(); VNET_FOREACH(vnet_iter) { IFNET_RLOCK(); CURVNET_SET_QUIET(vnet_iter); CK_STAILQ_FOREACH(idev, &V_ifnet, if_link) { /* check if network interface belongs to mlx5en */ if (!mlx5_netdev_match(idev, dev->mdev, "mce")) continue; write_lock(&dev->roce.netdev_lock); dev->roce.netdev = idev; write_unlock(&dev->roce.netdev_lock); } CURVNET_RESTORE(); IFNET_RUNLOCK(); } VNET_LIST_RUNLOCK(); dev->roce.nb.notifier_call = mlx5_netdev_event; err = register_netdevice_notifier(&dev->roce.nb); if (err) { dev->roce.nb.notifier_call = NULL; return err; } if (MLX5_CAP_GEN(dev->mdev, roce)) { err = mlx5_nic_vport_enable_roce(dev->mdev); if (err) goto err_unregister_netdevice_notifier; } err = mlx5_roce_lag_init(dev); if (err) goto err_disable_roce; return 0; err_disable_roce: if (MLX5_CAP_GEN(dev->mdev, roce)) mlx5_nic_vport_disable_roce(dev->mdev); err_unregister_netdevice_notifier: mlx5_remove_roce_notifier(dev); return err; } static void mlx5_disable_roce(struct mlx5_ib_dev *dev) { mlx5_roce_lag_cleanup(dev); if (MLX5_CAP_GEN(dev->mdev, roce)) mlx5_nic_vport_disable_roce(dev->mdev); } static void mlx5_ib_dealloc_q_port_counter(struct mlx5_ib_dev *dev, u8 port_num) { mlx5_vport_dealloc_q_counter(dev->mdev, MLX5_INTERFACE_PROTOCOL_IB, dev->port[port_num].q_cnt_id); dev->port[port_num].q_cnt_id = 0; } static void mlx5_ib_dealloc_q_counters(struct mlx5_ib_dev *dev) { unsigned int i; for (i = 0; i < dev->num_ports; i++) mlx5_ib_dealloc_q_port_counter(dev, i); } static int mlx5_ib_alloc_q_counters(struct mlx5_ib_dev *dev) { int i; int ret; for (i = 0; i < dev->num_ports; i++) { ret = mlx5_vport_alloc_q_counter(dev->mdev, MLX5_INTERFACE_PROTOCOL_IB, &dev->port[i].q_cnt_id); if (ret) { mlx5_ib_warn(dev, "couldn't allocate queue counter for port %d, err %d\n", i + 1, ret); goto dealloc_counters; } } return 0; dealloc_counters: while (--i >= 0) mlx5_ib_dealloc_q_port_counter(dev, i); return ret; } static const char * const names[] = { "rx_write_requests", "rx_read_requests", "rx_atomic_requests", "out_of_buffer", "out_of_sequence", "duplicate_request", "rnr_nak_retry_err", "packet_seq_err", "implied_nak_seq_err", "local_ack_timeout_err", }; static const size_t stats_offsets[] = { MLX5_BYTE_OFF(query_q_counter_out, rx_write_requests), MLX5_BYTE_OFF(query_q_counter_out, rx_read_requests), MLX5_BYTE_OFF(query_q_counter_out, rx_atomic_requests), MLX5_BYTE_OFF(query_q_counter_out, out_of_buffer), MLX5_BYTE_OFF(query_q_counter_out, out_of_sequence), MLX5_BYTE_OFF(query_q_counter_out, duplicate_request), MLX5_BYTE_OFF(query_q_counter_out, rnr_nak_retry_err), MLX5_BYTE_OFF(query_q_counter_out, packet_seq_err), MLX5_BYTE_OFF(query_q_counter_out, implied_nak_seq_err), MLX5_BYTE_OFF(query_q_counter_out, local_ack_timeout_err), }; static struct rdma_hw_stats *mlx5_ib_alloc_hw_stats(struct ib_device *ibdev, u8 port_num) { BUILD_BUG_ON(ARRAY_SIZE(names) != ARRAY_SIZE(stats_offsets)); /* We support only per port stats */ if (port_num == 0) return NULL; return rdma_alloc_hw_stats_struct(names, ARRAY_SIZE(names), RDMA_HW_STATS_DEFAULT_LIFESPAN); } static int mlx5_ib_get_hw_stats(struct ib_device *ibdev, struct rdma_hw_stats *stats, u8 port, int index) { struct mlx5_ib_dev *dev = to_mdev(ibdev); int outlen = MLX5_ST_SZ_BYTES(query_q_counter_out); void *out; __be32 val; int ret; int i; if (!port || !stats) return -ENOSYS; out = mlx5_vzalloc(outlen); if (!out) return -ENOMEM; ret = mlx5_vport_query_q_counter(dev->mdev, dev->port[port - 1].q_cnt_id, 0, out, outlen); if (ret) goto free; for (i = 0; i < ARRAY_SIZE(names); i++) { val = *(__be32 *)(out + stats_offsets[i]); stats->value[i] = (u64)be32_to_cpu(val); } free: kvfree(out); return ARRAY_SIZE(names); } static int mlx5_ib_stage_bfreg_init(struct mlx5_ib_dev *dev) { int err; err = mlx5_alloc_bfreg(dev->mdev, &dev->bfreg, false, false); if (err) return err; err = mlx5_alloc_bfreg(dev->mdev, &dev->fp_bfreg, false, true); if (err) { mlx5_free_bfreg(dev->mdev, &dev->bfreg); return err; } err = mlx5_alloc_bfreg(dev->mdev, &dev->wc_bfreg, true, false); if (err) { mlx5_free_bfreg(dev->mdev, &dev->fp_bfreg); mlx5_free_bfreg(dev->mdev, &dev->bfreg); } return err; } static void mlx5_ib_stage_bfreg_cleanup(struct mlx5_ib_dev *dev) { mlx5_free_bfreg(dev->mdev, &dev->wc_bfreg); mlx5_free_bfreg(dev->mdev, &dev->fp_bfreg); mlx5_free_bfreg(dev->mdev, &dev->bfreg); } static void *mlx5_ib_add(struct mlx5_core_dev *mdev) { struct mlx5_ib_dev *dev; enum rdma_link_layer ll; int port_type_cap; int err; int i; port_type_cap = MLX5_CAP_GEN(mdev, port_type); ll = mlx5_port_type_cap_to_rdma_ll(port_type_cap); dev = (struct mlx5_ib_dev *)ib_alloc_device(sizeof(*dev)); if (!dev) return NULL; dev->mdev = mdev; dev->port = kcalloc(MLX5_CAP_GEN(mdev, num_ports), sizeof(*dev->port), GFP_KERNEL); if (!dev->port) goto err_dealloc; rwlock_init(&dev->roce.netdev_lock); err = get_port_caps(dev); if (err) goto err_free_port; if (mlx5_use_mad_ifc(dev)) get_ext_port_caps(dev); MLX5_INIT_DOORBELL_LOCK(&dev->uar_lock); snprintf(dev->ib_dev.name, IB_DEVICE_NAME_MAX, "mlx5_%d", device_get_unit(mdev->pdev->dev.bsddev)); dev->ib_dev.owner = THIS_MODULE; dev->ib_dev.node_type = RDMA_NODE_IB_CA; dev->ib_dev.local_dma_lkey = 0 /* not supported for now */; dev->num_ports = MLX5_CAP_GEN(mdev, num_ports); dev->ib_dev.phys_port_cnt = dev->num_ports; dev->ib_dev.num_comp_vectors = dev->mdev->priv.eq_table.num_comp_vectors; dev->ib_dev.dma_device = &mdev->pdev->dev; dev->ib_dev.uverbs_abi_ver = MLX5_IB_UVERBS_ABI_VERSION; dev->ib_dev.uverbs_cmd_mask = (1ull << IB_USER_VERBS_CMD_GET_CONTEXT) | (1ull << IB_USER_VERBS_CMD_QUERY_DEVICE) | (1ull << IB_USER_VERBS_CMD_QUERY_PORT) | (1ull << IB_USER_VERBS_CMD_ALLOC_PD) | (1ull << IB_USER_VERBS_CMD_DEALLOC_PD) | (1ull << IB_USER_VERBS_CMD_CREATE_AH) | (1ull << IB_USER_VERBS_CMD_DESTROY_AH) | (1ull << IB_USER_VERBS_CMD_REG_MR) | (1ull << IB_USER_VERBS_CMD_REREG_MR) | (1ull << IB_USER_VERBS_CMD_DEREG_MR) | (1ull << IB_USER_VERBS_CMD_CREATE_COMP_CHANNEL) | (1ull << IB_USER_VERBS_CMD_CREATE_CQ) | (1ull << IB_USER_VERBS_CMD_RESIZE_CQ) | (1ull << IB_USER_VERBS_CMD_DESTROY_CQ) | (1ull << IB_USER_VERBS_CMD_CREATE_QP) | (1ull << IB_USER_VERBS_CMD_MODIFY_QP) | (1ull << IB_USER_VERBS_CMD_QUERY_QP) | (1ull << IB_USER_VERBS_CMD_DESTROY_QP) | (1ull << IB_USER_VERBS_CMD_ATTACH_MCAST) | (1ull << IB_USER_VERBS_CMD_DETACH_MCAST) | (1ull << IB_USER_VERBS_CMD_CREATE_SRQ) | (1ull << IB_USER_VERBS_CMD_MODIFY_SRQ) | (1ull << IB_USER_VERBS_CMD_QUERY_SRQ) | (1ull << IB_USER_VERBS_CMD_DESTROY_SRQ) | (1ull << IB_USER_VERBS_CMD_CREATE_XSRQ) | (1ull << IB_USER_VERBS_CMD_OPEN_QP); dev->ib_dev.uverbs_ex_cmd_mask = (1ull << IB_USER_VERBS_EX_CMD_QUERY_DEVICE) | (1ull << IB_USER_VERBS_EX_CMD_CREATE_CQ) | (1ull << IB_USER_VERBS_EX_CMD_CREATE_QP); dev->ib_dev.query_device = mlx5_ib_query_device; dev->ib_dev.query_port = mlx5_ib_query_port; dev->ib_dev.get_link_layer = mlx5_ib_port_link_layer; if (ll == IB_LINK_LAYER_ETHERNET) dev->ib_dev.get_netdev = mlx5_ib_get_netdev; dev->ib_dev.query_gid = mlx5_ib_query_gid; dev->ib_dev.add_gid = mlx5_ib_add_gid; dev->ib_dev.del_gid = mlx5_ib_del_gid; dev->ib_dev.query_pkey = mlx5_ib_query_pkey; dev->ib_dev.modify_device = mlx5_ib_modify_device; dev->ib_dev.modify_port = mlx5_ib_modify_port; dev->ib_dev.alloc_ucontext = mlx5_ib_alloc_ucontext; dev->ib_dev.dealloc_ucontext = mlx5_ib_dealloc_ucontext; dev->ib_dev.mmap = mlx5_ib_mmap; dev->ib_dev.alloc_pd = mlx5_ib_alloc_pd; dev->ib_dev.dealloc_pd = mlx5_ib_dealloc_pd; dev->ib_dev.create_ah = mlx5_ib_create_ah; dev->ib_dev.query_ah = mlx5_ib_query_ah; dev->ib_dev.destroy_ah = mlx5_ib_destroy_ah; dev->ib_dev.create_srq = mlx5_ib_create_srq; dev->ib_dev.modify_srq = mlx5_ib_modify_srq; dev->ib_dev.query_srq = mlx5_ib_query_srq; dev->ib_dev.destroy_srq = mlx5_ib_destroy_srq; dev->ib_dev.post_srq_recv = mlx5_ib_post_srq_recv; dev->ib_dev.create_qp = mlx5_ib_create_qp; dev->ib_dev.modify_qp = mlx5_ib_modify_qp; dev->ib_dev.query_qp = mlx5_ib_query_qp; dev->ib_dev.destroy_qp = mlx5_ib_destroy_qp; dev->ib_dev.post_send = mlx5_ib_post_send; dev->ib_dev.post_recv = mlx5_ib_post_recv; dev->ib_dev.create_cq = mlx5_ib_create_cq; dev->ib_dev.modify_cq = mlx5_ib_modify_cq; dev->ib_dev.resize_cq = mlx5_ib_resize_cq; dev->ib_dev.destroy_cq = mlx5_ib_destroy_cq; dev->ib_dev.poll_cq = mlx5_ib_poll_cq; dev->ib_dev.req_notify_cq = mlx5_ib_arm_cq; dev->ib_dev.get_dma_mr = mlx5_ib_get_dma_mr; dev->ib_dev.reg_user_mr = mlx5_ib_reg_user_mr; dev->ib_dev.rereg_user_mr = mlx5_ib_rereg_user_mr; dev->ib_dev.dereg_mr = mlx5_ib_dereg_mr; dev->ib_dev.attach_mcast = mlx5_ib_mcg_attach; dev->ib_dev.detach_mcast = mlx5_ib_mcg_detach; dev->ib_dev.process_mad = mlx5_ib_process_mad; dev->ib_dev.alloc_mr = mlx5_ib_alloc_mr; dev->ib_dev.map_mr_sg = mlx5_ib_map_mr_sg; dev->ib_dev.check_mr_status = mlx5_ib_check_mr_status; dev->ib_dev.get_port_immutable = mlx5_port_immutable; dev->ib_dev.get_dev_fw_str = get_dev_fw_str; if (mlx5_core_is_pf(mdev)) { dev->ib_dev.get_vf_config = mlx5_ib_get_vf_config; dev->ib_dev.set_vf_link_state = mlx5_ib_set_vf_link_state; dev->ib_dev.get_vf_stats = mlx5_ib_get_vf_stats; dev->ib_dev.set_vf_guid = mlx5_ib_set_vf_guid; } dev->ib_dev.disassociate_ucontext = mlx5_ib_disassociate_ucontext; mlx5_ib_internal_fill_odp_caps(dev); if (MLX5_CAP_GEN(mdev, imaicl)) { dev->ib_dev.alloc_mw = mlx5_ib_alloc_mw; dev->ib_dev.dealloc_mw = mlx5_ib_dealloc_mw; dev->ib_dev.uverbs_cmd_mask |= (1ull << IB_USER_VERBS_CMD_ALLOC_MW) | (1ull << IB_USER_VERBS_CMD_DEALLOC_MW); } if (MLX5_CAP_GEN(dev->mdev, out_of_seq_cnt) && MLX5_CAP_GEN(dev->mdev, retransmission_q_counters)) { dev->ib_dev.get_hw_stats = mlx5_ib_get_hw_stats; dev->ib_dev.alloc_hw_stats = mlx5_ib_alloc_hw_stats; } if (MLX5_CAP_GEN(mdev, xrc)) { dev->ib_dev.alloc_xrcd = mlx5_ib_alloc_xrcd; dev->ib_dev.dealloc_xrcd = mlx5_ib_dealloc_xrcd; dev->ib_dev.uverbs_cmd_mask |= (1ull << IB_USER_VERBS_CMD_OPEN_XRCD) | (1ull << IB_USER_VERBS_CMD_CLOSE_XRCD); } if (mlx5_ib_port_link_layer(&dev->ib_dev, 1) == IB_LINK_LAYER_ETHERNET) { dev->ib_dev.create_flow = mlx5_ib_create_flow; dev->ib_dev.destroy_flow = mlx5_ib_destroy_flow; dev->ib_dev.create_wq = mlx5_ib_create_wq; dev->ib_dev.modify_wq = mlx5_ib_modify_wq; dev->ib_dev.destroy_wq = mlx5_ib_destroy_wq; dev->ib_dev.create_rwq_ind_table = mlx5_ib_create_rwq_ind_table; dev->ib_dev.destroy_rwq_ind_table = mlx5_ib_destroy_rwq_ind_table; dev->ib_dev.uverbs_ex_cmd_mask |= (1ull << IB_USER_VERBS_EX_CMD_CREATE_FLOW) | (1ull << IB_USER_VERBS_EX_CMD_DESTROY_FLOW) | (1ull << IB_USER_VERBS_EX_CMD_CREATE_WQ) | (1ull << IB_USER_VERBS_EX_CMD_MODIFY_WQ) | (1ull << IB_USER_VERBS_EX_CMD_DESTROY_WQ) | (1ull << IB_USER_VERBS_EX_CMD_CREATE_RWQ_IND_TBL) | (1ull << IB_USER_VERBS_EX_CMD_DESTROY_RWQ_IND_TBL); } err = init_node_data(dev); if (err) goto err_free_port; mutex_init(&dev->flow_db.lock); mutex_init(&dev->cap_mask_mutex); INIT_LIST_HEAD(&dev->qp_list); spin_lock_init(&dev->reset_flow_resource_lock); if (ll == IB_LINK_LAYER_ETHERNET) { err = mlx5_enable_roce(dev); if (err) goto err_free_port; } err = create_dev_resources(&dev->devr); if (err) goto err_disable_roce; err = mlx5_ib_odp_init_one(dev); if (err) goto err_rsrc; err = mlx5_ib_alloc_q_counters(dev); if (err) goto err_odp; err = mlx5_ib_stage_bfreg_init(dev); if (err) goto err_q_cnt; err = ib_register_device(&dev->ib_dev, NULL); if (err) goto err_bfreg; err = create_umr_res(dev); if (err) goto err_dev; for (i = 0; i < ARRAY_SIZE(mlx5_class_attributes); i++) { err = device_create_file(&dev->ib_dev.dev, mlx5_class_attributes[i]); if (err) goto err_umrc; } err = mlx5_ib_init_congestion(dev); if (err) goto err_umrc; dev->ib_active = true; return dev; err_umrc: destroy_umrc_res(dev); err_dev: ib_unregister_device(&dev->ib_dev); err_bfreg: mlx5_ib_stage_bfreg_cleanup(dev); err_q_cnt: mlx5_ib_dealloc_q_counters(dev); err_odp: mlx5_ib_odp_remove_one(dev); err_rsrc: destroy_dev_resources(&dev->devr); err_disable_roce: if (ll == IB_LINK_LAYER_ETHERNET) { mlx5_disable_roce(dev); mlx5_remove_roce_notifier(dev); } err_free_port: kfree(dev->port); err_dealloc: ib_dealloc_device((struct ib_device *)dev); return NULL; } static void mlx5_ib_remove(struct mlx5_core_dev *mdev, void *context) { struct mlx5_ib_dev *dev = context; enum rdma_link_layer ll = mlx5_ib_port_link_layer(&dev->ib_dev, 1); mlx5_ib_cleanup_congestion(dev); mlx5_remove_roce_notifier(dev); ib_unregister_device(&dev->ib_dev); mlx5_ib_stage_bfreg_cleanup(dev); mlx5_ib_dealloc_q_counters(dev); destroy_umrc_res(dev); mlx5_ib_odp_remove_one(dev); destroy_dev_resources(&dev->devr); if (ll == IB_LINK_LAYER_ETHERNET) mlx5_disable_roce(dev); kfree(dev->port); ib_dealloc_device(&dev->ib_dev); } static struct mlx5_interface mlx5_ib_interface = { .add = mlx5_ib_add, .remove = mlx5_ib_remove, .event = mlx5_ib_event, .protocol = MLX5_INTERFACE_PROTOCOL_IB, }; static int __init mlx5_ib_init(void) { int err; err = mlx5_ib_odp_init(); if (err) return err; err = mlx5_register_interface(&mlx5_ib_interface); if (err) goto clean_odp; return err; clean_odp: mlx5_ib_odp_cleanup(); return err; } static void __exit mlx5_ib_cleanup(void) { mlx5_unregister_interface(&mlx5_ib_interface); mlx5_ib_odp_cleanup(); } module_init_order(mlx5_ib_init, SI_ORDER_SEVENTH); module_exit_order(mlx5_ib_cleanup, SI_ORDER_SEVENTH); diff --git a/sys/dev/mlx5/mlx5_ib/mlx5_ib_mem.c b/sys/dev/mlx5/mlx5_ib/mlx5_ib_mem.c index 91e48bce00f0..8316884c258f 100644 --- a/sys/dev/mlx5/mlx5_ib/mlx5_ib_mem.c +++ b/sys/dev/mlx5/mlx5_ib/mlx5_ib_mem.c @@ -1,203 +1,203 @@ /*- * Copyright (c) 2013-2020, Mellanox Technologies, Ltd. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY AUTHOR AND CONTRIBUTORS `AS IS' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $FreeBSD$ */ #include "opt_rss.h" #include "opt_ratelimit.h" #include #include #include -#include "mlx5_ib.h" +#include /* @umem: umem object to scan * @addr: ib virtual address requested by the user * @max_page_shift: high limit for page_shift - 0 means no limit * @count: number of PAGE_SIZE pages covered by umem * @shift: page shift for the compound pages found in the region * @ncont: number of compund pages * @order: log2 of the number of compound pages */ void mlx5_ib_cont_pages(struct ib_umem *umem, u64 addr, unsigned long max_page_shift, int *count, int *shift, int *ncont, int *order) { unsigned long tmp; unsigned long m; u64 base = ~0, p = 0; u64 len, pfn; int i = 0; struct scatterlist *sg; int entry; addr = addr >> PAGE_SHIFT; tmp = (unsigned long)addr; m = find_first_bit(&tmp, BITS_PER_LONG); if (max_page_shift) m = min_t(unsigned long, max_page_shift - PAGE_SHIFT, m); for_each_sg(umem->sg_head.sgl, sg, umem->nmap, entry) { len = sg_dma_len(sg) >> PAGE_SHIFT; pfn = sg_dma_address(sg) >> PAGE_SHIFT; if (base + p != pfn) { /* If either the offset or the new * base are unaligned update m */ tmp = (unsigned long)(pfn | p); if (!IS_ALIGNED(tmp, 1 << m)) m = find_first_bit(&tmp, BITS_PER_LONG); base = pfn; p = 0; } p += len; i += len; } if (i) { m = min_t(unsigned long, ilog2(roundup_pow_of_two(i)), m); if (order) *order = ilog2(roundup_pow_of_two(i) >> m); *ncont = DIV_ROUND_UP(i, (1 << m)); } else { m = 0; if (order) *order = 0; *ncont = 0; } *shift = PAGE_SHIFT + m; *count = i; } #ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING static u64 umem_dma_to_mtt(dma_addr_t umem_dma) { u64 mtt_entry = umem_dma & ODP_DMA_ADDR_MASK; if (umem_dma & ODP_READ_ALLOWED_BIT) mtt_entry |= MLX5_IB_MTT_READ; if (umem_dma & ODP_WRITE_ALLOWED_BIT) mtt_entry |= MLX5_IB_MTT_WRITE; return mtt_entry; } #endif /* * Populate the given array with bus addresses from the umem. * * dev - mlx5_ib device * umem - umem to use to fill the pages * page_shift - determines the page size used in the resulting array * offset - offset into the umem to start from, * only implemented for ODP umems * num_pages - total number of pages to fill * pas - bus addresses array to fill * access_flags - access flags to set on all present pages. use enum mlx5_ib_mtt_access_flags for this. */ void __mlx5_ib_populate_pas(struct mlx5_ib_dev *dev, struct ib_umem *umem, int page_shift, size_t offset, size_t num_pages, __be64 *pas, int access_flags) { unsigned long umem_page_shift = ilog2(umem->page_size); int shift = page_shift - umem_page_shift; int mask = (1 << shift) - 1; int i, k; u64 cur = 0; u64 base; int len; struct scatterlist *sg; int entry; #ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING const bool odp = umem->odp_data != NULL; if (odp) { WARN_ON(shift != 0); WARN_ON(access_flags != (MLX5_IB_MTT_READ | MLX5_IB_MTT_WRITE)); for (i = 0; i < num_pages; ++i) { dma_addr_t pa = umem->odp_data->dma_list[offset + i]; pas[i] = cpu_to_be64(umem_dma_to_mtt(pa)); } return; } #endif i = 0; for_each_sg(umem->sg_head.sgl, sg, umem->nmap, entry) { len = sg_dma_len(sg) >> umem_page_shift; base = sg_dma_address(sg); for (k = 0; k < len; k++) { if (!(i & mask)) { cur = base + (k << umem_page_shift); cur |= access_flags; pas[i >> shift] = cpu_to_be64(cur); mlx5_ib_dbg(dev, "pas[%d] 0x%llx\n", i >> shift, (long long)be64_to_cpu(pas[i >> shift])); } else mlx5_ib_dbg(dev, "=====> 0x%llx\n", (long long)(base + (k << umem_page_shift))); i++; } } } void mlx5_ib_populate_pas(struct mlx5_ib_dev *dev, struct ib_umem *umem, int page_shift, __be64 *pas, int access_flags) { return __mlx5_ib_populate_pas(dev, umem, page_shift, 0, ib_umem_num_pages(umem), pas, access_flags); } int mlx5_ib_get_buf_offset(u64 addr, int page_shift, u32 *offset) { u64 page_size; u64 page_mask; u64 off_size; u64 off_mask; u64 buf_off; page_size = (u64)1 << page_shift; page_mask = page_size - 1; buf_off = addr & page_mask; off_size = page_size >> 6; off_mask = off_size - 1; if (buf_off & off_mask) return -EINVAL; *offset = buf_off >> ilog2(off_size); return 0; } diff --git a/sys/dev/mlx5/mlx5_ib/mlx5_ib_mr.c b/sys/dev/mlx5/mlx5_ib/mlx5_ib_mr.c index 0884ba3f4506..6e4122f57f4b 100644 --- a/sys/dev/mlx5/mlx5_ib/mlx5_ib_mr.c +++ b/sys/dev/mlx5/mlx5_ib/mlx5_ib_mr.c @@ -1,1657 +1,1657 @@ /*- * Copyright (c) 2013-2015, Mellanox Technologies, Ltd. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY AUTHOR AND CONTRIBUTORS `AS IS' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $FreeBSD$ */ #include "opt_rss.h" #include "opt_ratelimit.h" #include #include #include #include #include #include #include -#include "mlx5_ib.h" +#include enum { MAX_PENDING_REG_MR = 8, }; #define MLX5_UMR_ALIGN 2048 #ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING static __be64 mlx5_ib_update_mtt_emergency_buffer[ MLX5_UMR_MTT_MIN_CHUNK_SIZE/sizeof(__be64)] __aligned(MLX5_UMR_ALIGN); static DEFINE_MUTEX(mlx5_ib_update_mtt_emergency_buffer_mutex); #endif static int clean_mr(struct mlx5_ib_mr *mr); static int destroy_mkey(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr) { int err = mlx5_core_destroy_mkey(dev->mdev, &mr->mmkey); #ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING /* Wait until all page fault handlers using the mr complete. */ synchronize_srcu(&dev->mr_srcu); #endif return err; } static int order2idx(struct mlx5_ib_dev *dev, int order) { struct mlx5_mr_cache *cache = &dev->cache; if (order < cache->ent[0].order) return 0; else return order - cache->ent[0].order; } static bool use_umr_mtt_update(struct mlx5_ib_mr *mr, u64 start, u64 length) { return ((u64)1 << mr->order) * MLX5_ADAPTER_PAGE_SIZE >= length + (start & (MLX5_ADAPTER_PAGE_SIZE - 1)); } #ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING static void update_odp_mr(struct mlx5_ib_mr *mr) { if (mr->umem->odp_data) { /* * This barrier prevents the compiler from moving the * setting of umem->odp_data->private to point to our * MR, before reg_umr finished, to ensure that the MR * initialization have finished before starting to * handle invalidations. */ smp_wmb(); mr->umem->odp_data->private = mr; /* * Make sure we will see the new * umem->odp_data->private value in the invalidation * routines, before we can get page faults on the * MR. Page faults can happen once we put the MR in * the tree, below this line. Without the barrier, * there can be a fault handling and an invalidation * before umem->odp_data->private == mr is visible to * the invalidation handler. */ smp_wmb(); } } #endif static void reg_mr_callback(int status, struct mlx5_async_work *context) { struct mlx5_ib_mr *mr = container_of(context, struct mlx5_ib_mr, cb_work); struct mlx5_ib_dev *dev = mr->dev; struct mlx5_mr_cache *cache = &dev->cache; int c = order2idx(dev, mr->order); struct mlx5_cache_ent *ent = &cache->ent[c]; u8 key; unsigned long flags; struct mlx5_mr_table *table = &dev->mdev->priv.mr_table; int err; spin_lock_irqsave(&ent->lock, flags); ent->pending--; spin_unlock_irqrestore(&ent->lock, flags); if (status) { mlx5_ib_warn(dev, "async reg mr failed. status %d\n", status); kfree(mr); dev->fill_delay = 1; mod_timer(&dev->delay_timer, jiffies + HZ); return; } spin_lock_irqsave(&dev->mdev->priv.mkey_lock, flags); key = dev->mdev->priv.mkey_key++; spin_unlock_irqrestore(&dev->mdev->priv.mkey_lock, flags); mr->mmkey.key = mlx5_idx_to_mkey(MLX5_GET(create_mkey_out, mr->out, mkey_index)) | key; cache->last_add = jiffies; spin_lock_irqsave(&ent->lock, flags); list_add_tail(&mr->list, &ent->head); ent->cur++; ent->size++; spin_unlock_irqrestore(&ent->lock, flags); spin_lock_irqsave(&table->lock, flags); err = radix_tree_insert(&table->tree, mlx5_mkey_to_idx(mr->mmkey.key), &mr->mmkey); if (err) pr_err("Error inserting to mkey tree. 0x%x\n", -err); spin_unlock_irqrestore(&table->lock, flags); } static int add_keys(struct mlx5_ib_dev *dev, int c, int num) { struct mlx5_mr_cache *cache = &dev->cache; struct mlx5_cache_ent *ent = &cache->ent[c]; int inlen = MLX5_ST_SZ_BYTES(create_mkey_in); struct mlx5_ib_mr *mr; int npages = 1 << ent->order; void *mkc; u32 *in; int err = 0; int i; in = kzalloc(inlen, GFP_KERNEL); if (!in) return -ENOMEM; mkc = MLX5_ADDR_OF(create_mkey_in, in, memory_key_mkey_entry); for (i = 0; i < num; i++) { if (ent->pending >= MAX_PENDING_REG_MR) { err = -EAGAIN; break; } mr = kzalloc(sizeof(*mr), GFP_KERNEL); if (!mr) { err = -ENOMEM; break; } mr->order = ent->order; mr->umred = 1; mr->dev = dev; MLX5_SET(mkc, mkc, free, 1); MLX5_SET(mkc, mkc, umr_en, 1); MLX5_SET(mkc, mkc, access_mode, MLX5_ACCESS_MODE_MTT); MLX5_SET(mkc, mkc, qpn, 0xffffff); MLX5_SET(mkc, mkc, translations_octword_size, (npages + 1) / 2); MLX5_SET(mkc, mkc, log_page_size, 12); spin_lock_irq(&ent->lock); ent->pending++; spin_unlock_irq(&ent->lock); err = mlx5_core_create_mkey_cb(dev->mdev, &mr->mmkey, &dev->async_ctx, in, inlen, mr->out, sizeof(mr->out), reg_mr_callback, &mr->cb_work); if (err) { spin_lock_irq(&ent->lock); ent->pending--; spin_unlock_irq(&ent->lock); mlx5_ib_warn(dev, "create mkey failed %d\n", err); kfree(mr); break; } } kfree(in); return err; } static void remove_keys(struct mlx5_ib_dev *dev, int c, int num) { struct mlx5_mr_cache *cache = &dev->cache; struct mlx5_cache_ent *ent = &cache->ent[c]; struct mlx5_ib_mr *mr; int err; int i; for (i = 0; i < num; i++) { spin_lock_irq(&ent->lock); if (list_empty(&ent->head)) { spin_unlock_irq(&ent->lock); return; } mr = list_first_entry(&ent->head, struct mlx5_ib_mr, list); list_del(&mr->list); ent->cur--; ent->size--; spin_unlock_irq(&ent->lock); err = destroy_mkey(dev, mr); if (err) mlx5_ib_warn(dev, "failed destroy mkey\n"); else kfree(mr); } } static int someone_adding(struct mlx5_mr_cache *cache) { int i; for (i = 0; i < MAX_MR_CACHE_ENTRIES; i++) { if (cache->ent[i].cur < cache->ent[i].limit) return 1; } return 0; } static void __cache_work_func(struct mlx5_cache_ent *ent) { struct mlx5_ib_dev *dev = ent->dev; struct mlx5_mr_cache *cache = &dev->cache; int i = order2idx(dev, ent->order); int err; if (cache->stopped) return; ent = &dev->cache.ent[i]; if (ent->cur < 2 * ent->limit && !dev->fill_delay) { err = add_keys(dev, i, 1); if (ent->cur < 2 * ent->limit) { if (err == -EAGAIN) { mlx5_ib_dbg(dev, "returned eagain, order %d\n", i + 2); queue_delayed_work(cache->wq, &ent->dwork, msecs_to_jiffies(3)); } else if (err) { mlx5_ib_warn(dev, "command failed order %d, err %d\n", i + 2, err); queue_delayed_work(cache->wq, &ent->dwork, msecs_to_jiffies(1000)); } else { queue_work(cache->wq, &ent->work); } } } else if (ent->cur > 2 * ent->limit) { /* * The remove_keys() logic is performed as garbage collection * task. Such task is intended to be run when no other active * processes are running. * * The need_resched() will return TRUE if there are user tasks * to be activated in near future. * * In such case, we don't execute remove_keys() and postpone * the garbage collection work to try to run in next cycle, * in order to free CPU resources to other tasks. */ if (!need_resched() && !someone_adding(cache) && time_after(jiffies, cache->last_add + 300 * HZ)) { remove_keys(dev, i, 1); if (ent->cur > ent->limit) queue_work(cache->wq, &ent->work); } else { queue_delayed_work(cache->wq, &ent->dwork, 300 * HZ); } } } static void delayed_cache_work_func(struct work_struct *work) { struct mlx5_cache_ent *ent; ent = container_of(work, struct mlx5_cache_ent, dwork.work); __cache_work_func(ent); } static void cache_work_func(struct work_struct *work) { struct mlx5_cache_ent *ent; ent = container_of(work, struct mlx5_cache_ent, work); __cache_work_func(ent); } static struct mlx5_ib_mr *alloc_cached_mr(struct mlx5_ib_dev *dev, int order) { struct mlx5_mr_cache *cache = &dev->cache; struct mlx5_ib_mr *mr = NULL; struct mlx5_cache_ent *ent; int c; int i; c = order2idx(dev, order); if (c < 0 || c >= MAX_MR_CACHE_ENTRIES) { mlx5_ib_warn(dev, "order %d, cache index %d\n", order, c); return NULL; } for (i = c; i < MAX_MR_CACHE_ENTRIES; i++) { ent = &cache->ent[i]; mlx5_ib_dbg(dev, "order %d, cache index %d\n", ent->order, i); spin_lock_irq(&ent->lock); if (!list_empty(&ent->head)) { mr = list_first_entry(&ent->head, struct mlx5_ib_mr, list); list_del(&mr->list); ent->cur--; spin_unlock_irq(&ent->lock); if (ent->cur < ent->limit) queue_work(cache->wq, &ent->work); break; } spin_unlock_irq(&ent->lock); queue_work(cache->wq, &ent->work); } if (!mr) cache->ent[c].miss++; return mr; } static void free_cached_mr(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr) { struct mlx5_mr_cache *cache = &dev->cache; struct mlx5_cache_ent *ent; int shrink = 0; int c; c = order2idx(dev, mr->order); if (c < 0 || c >= MAX_MR_CACHE_ENTRIES) { mlx5_ib_warn(dev, "order %d, cache index %d\n", mr->order, c); return; } ent = &cache->ent[c]; spin_lock_irq(&ent->lock); list_add_tail(&mr->list, &ent->head); ent->cur++; if (ent->cur > 2 * ent->limit) shrink = 1; spin_unlock_irq(&ent->lock); if (shrink) queue_work(cache->wq, &ent->work); } static void clean_keys(struct mlx5_ib_dev *dev, int c) { struct mlx5_mr_cache *cache = &dev->cache; struct mlx5_cache_ent *ent = &cache->ent[c]; struct mlx5_ib_mr *mr; int err; cancel_delayed_work(&ent->dwork); while (1) { spin_lock_irq(&ent->lock); if (list_empty(&ent->head)) { spin_unlock_irq(&ent->lock); return; } mr = list_first_entry(&ent->head, struct mlx5_ib_mr, list); list_del(&mr->list); ent->cur--; ent->size--; spin_unlock_irq(&ent->lock); err = destroy_mkey(dev, mr); if (err) mlx5_ib_warn(dev, "failed destroy mkey\n"); else kfree(mr); } } static void delay_time_func(unsigned long ctx) { struct mlx5_ib_dev *dev = (struct mlx5_ib_dev *)ctx; dev->fill_delay = 0; } int mlx5_mr_cache_init(struct mlx5_ib_dev *dev) { struct mlx5_mr_cache *cache = &dev->cache; struct mlx5_cache_ent *ent; int limit; int i; mutex_init(&dev->slow_path_mutex); cache->wq = alloc_ordered_workqueue("mkey_cache", WQ_MEM_RECLAIM); if (!cache->wq) { mlx5_ib_warn(dev, "failed to create work queue\n"); return -ENOMEM; } mlx5_cmd_init_async_ctx(dev->mdev, &dev->async_ctx); setup_timer(&dev->delay_timer, delay_time_func, (unsigned long)dev); for (i = 0; i < MAX_MR_CACHE_ENTRIES; i++) { INIT_LIST_HEAD(&cache->ent[i].head); spin_lock_init(&cache->ent[i].lock); ent = &cache->ent[i]; INIT_LIST_HEAD(&ent->head); spin_lock_init(&ent->lock); ent->order = i + 2; ent->dev = dev; if (dev->mdev->profile->mask & MLX5_PROF_MASK_MR_CACHE) limit = dev->mdev->profile->mr_cache[i].limit; else limit = 0; INIT_WORK(&ent->work, cache_work_func); INIT_DELAYED_WORK(&ent->dwork, delayed_cache_work_func); ent->limit = limit; queue_work(cache->wq, &ent->work); } return 0; } int mlx5_mr_cache_cleanup(struct mlx5_ib_dev *dev) { int i; dev->cache.stopped = 1; flush_workqueue(dev->cache.wq); mlx5_cmd_cleanup_async_ctx(&dev->async_ctx); for (i = 0; i < MAX_MR_CACHE_ENTRIES; i++) clean_keys(dev, i); destroy_workqueue(dev->cache.wq); del_timer_sync(&dev->delay_timer); return 0; } struct ib_mr *mlx5_ib_get_dma_mr(struct ib_pd *pd, int acc) { struct mlx5_ib_dev *dev = to_mdev(pd->device); int inlen = MLX5_ST_SZ_BYTES(create_mkey_in); struct mlx5_core_dev *mdev = dev->mdev; struct mlx5_ib_mr *mr; void *mkc; u32 *in; int err; mr = kzalloc(sizeof(*mr), GFP_KERNEL); if (!mr) return ERR_PTR(-ENOMEM); in = kzalloc(inlen, GFP_KERNEL); if (!in) { err = -ENOMEM; goto err_free; } mkc = MLX5_ADDR_OF(create_mkey_in, in, memory_key_mkey_entry); MLX5_SET(mkc, mkc, access_mode, MLX5_ACCESS_MODE_PA); MLX5_SET(mkc, mkc, a, !!(acc & IB_ACCESS_REMOTE_ATOMIC)); MLX5_SET(mkc, mkc, rw, !!(acc & IB_ACCESS_REMOTE_WRITE)); MLX5_SET(mkc, mkc, rr, !!(acc & IB_ACCESS_REMOTE_READ)); MLX5_SET(mkc, mkc, lw, !!(acc & IB_ACCESS_LOCAL_WRITE)); MLX5_SET(mkc, mkc, lr, 1); MLX5_SET(mkc, mkc, length64, 1); MLX5_SET(mkc, mkc, pd, to_mpd(pd)->pdn); MLX5_SET(mkc, mkc, qpn, 0xffffff); MLX5_SET64(mkc, mkc, start_addr, 0); err = mlx5_core_create_mkey(mdev, &mr->mmkey, in, inlen); if (err) goto err_in; kfree(in); mr->ibmr.lkey = mr->mmkey.key; mr->ibmr.rkey = mr->mmkey.key; mr->umem = NULL; return &mr->ibmr; err_in: kfree(in); err_free: kfree(mr); return ERR_PTR(err); } static int get_octo_len(u64 addr, u64 len, int page_size) { u64 offset; int npages; offset = addr & (page_size - 1); npages = ALIGN(len + offset, page_size) >> ilog2(page_size); return (npages + 1) / 2; } static int use_umr(int order) { return order <= MLX5_MAX_UMR_SHIFT; } static int dma_map_mr_pas(struct mlx5_ib_dev *dev, struct ib_umem *umem, int npages, int page_shift, int *size, __be64 **mr_pas, dma_addr_t *dma) { __be64 *pas; struct device *ddev = dev->ib_dev.dma_device; /* * UMR copies MTTs in units of MLX5_UMR_MTT_ALIGNMENT bytes. * To avoid copying garbage after the pas array, we allocate * a little more. */ *size = ALIGN(sizeof(u64) * npages, MLX5_UMR_MTT_ALIGNMENT); *mr_pas = kmalloc(*size + MLX5_UMR_ALIGN - 1, GFP_KERNEL); if (!(*mr_pas)) return -ENOMEM; pas = PTR_ALIGN(*mr_pas, MLX5_UMR_ALIGN); mlx5_ib_populate_pas(dev, umem, page_shift, pas, MLX5_IB_MTT_PRESENT); /* Clear padding after the actual pages. */ memset(pas + npages, 0, *size - npages * sizeof(u64)); *dma = dma_map_single(ddev, pas, *size, DMA_TO_DEVICE); if (dma_mapping_error(ddev, *dma)) { kfree(*mr_pas); return -ENOMEM; } return 0; } static void prep_umr_wqe_common(struct ib_pd *pd, struct mlx5_umr_wr *umrwr, struct ib_sge *sg, u64 dma, int n, u32 key, int page_shift) { struct mlx5_ib_dev *dev = to_mdev(pd->device); sg->addr = dma; sg->length = ALIGN(sizeof(u64) * n, 64); sg->lkey = dev->umrc.pd->local_dma_lkey; umrwr->wr.next = NULL; umrwr->wr.sg_list = sg; if (n) umrwr->wr.num_sge = 1; else umrwr->wr.num_sge = 0; umrwr->wr.opcode = MLX5_IB_WR_UMR; umrwr->npages = n; umrwr->page_shift = page_shift; umrwr->mkey = key; } static void prep_umr_reg_wqe(struct ib_pd *pd, struct mlx5_umr_wr *umrwr, struct ib_sge *sg, u64 dma, int n, u32 key, int page_shift, u64 virt_addr, u64 len, int access_flags) { prep_umr_wqe_common(pd, umrwr, sg, dma, n, key, page_shift); umrwr->wr.send_flags = 0; umrwr->target.virt_addr = virt_addr; umrwr->length = len; umrwr->access_flags = access_flags; umrwr->pd = pd; } static void prep_umr_unreg_wqe(struct mlx5_ib_dev *dev, struct mlx5_umr_wr *umrwr, u32 key) { umrwr->wr.send_flags = MLX5_IB_SEND_UMR_UNREG | MLX5_IB_SEND_UMR_FAIL_IF_FREE; umrwr->wr.opcode = MLX5_IB_WR_UMR; umrwr->mkey = key; } static struct ib_umem *mr_umem_get(struct ib_pd *pd, u64 start, u64 length, int access_flags, int *npages, int *page_shift, int *ncont, int *order) { struct mlx5_ib_dev *dev = to_mdev(pd->device); struct ib_umem *umem = ib_umem_get(pd->uobject->context, start, length, access_flags, 0); if (IS_ERR(umem)) { mlx5_ib_err(dev, "umem get failed (%ld)\n", PTR_ERR(umem)); return (void *)umem; } mlx5_ib_cont_pages(umem, start, MLX5_MKEY_PAGE_SHIFT_MASK, npages, page_shift, ncont, order); if (!*npages) { mlx5_ib_warn(dev, "avoid zero region\n"); ib_umem_release(umem); return ERR_PTR(-EINVAL); } mlx5_ib_dbg(dev, "npages %d, ncont %d, order %d, page_shift %d\n", *npages, *ncont, *order, *page_shift); return umem; } static void mlx5_ib_umr_done(struct ib_cq *cq, struct ib_wc *wc) { struct mlx5_ib_umr_context *context = container_of(wc->wr_cqe, struct mlx5_ib_umr_context, cqe); context->status = wc->status; complete(&context->done); } static inline void mlx5_ib_init_umr_context(struct mlx5_ib_umr_context *context) { context->cqe.done = mlx5_ib_umr_done; context->status = -1; init_completion(&context->done); } static struct mlx5_ib_mr *reg_umr(struct ib_pd *pd, struct ib_umem *umem, u64 virt_addr, u64 len, int npages, int page_shift, int order, int access_flags) { struct mlx5_ib_dev *dev = to_mdev(pd->device); struct device *ddev = dev->ib_dev.dma_device; struct umr_common *umrc = &dev->umrc; struct mlx5_ib_umr_context umr_context; struct mlx5_umr_wr umrwr = {}; const struct ib_send_wr *bad; struct mlx5_ib_mr *mr; struct ib_sge sg; int size; __be64 *mr_pas; dma_addr_t dma; int err = 0; int i; for (i = 0; i < 1; i++) { mr = alloc_cached_mr(dev, order); if (mr) break; err = add_keys(dev, order2idx(dev, order), 1); if (err && err != -EAGAIN) { mlx5_ib_warn(dev, "add_keys failed, err %d\n", err); break; } } if (!mr) return ERR_PTR(-EAGAIN); err = dma_map_mr_pas(dev, umem, npages, page_shift, &size, &mr_pas, &dma); if (err) goto free_mr; mlx5_ib_init_umr_context(&umr_context); umrwr.wr.wr_cqe = &umr_context.cqe; prep_umr_reg_wqe(pd, &umrwr, &sg, dma, npages, mr->mmkey.key, page_shift, virt_addr, len, access_flags); down(&umrc->sem); err = ib_post_send(umrc->qp, &umrwr.wr, &bad); if (err) { mlx5_ib_warn(dev, "post send failed, err %d\n", err); goto unmap_dma; } else { wait_for_completion(&umr_context.done); if (umr_context.status != IB_WC_SUCCESS) { mlx5_ib_warn(dev, "reg umr failed\n"); err = -EFAULT; } } mr->mmkey.iova = virt_addr; mr->mmkey.size = len; mr->mmkey.pd = to_mpd(pd)->pdn; mr->live = 1; unmap_dma: up(&umrc->sem); dma_unmap_single(ddev, dma, size, DMA_TO_DEVICE); kfree(mr_pas); free_mr: if (err) { free_cached_mr(dev, mr); return ERR_PTR(err); } return mr; } #ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING int mlx5_ib_update_mtt(struct mlx5_ib_mr *mr, u64 start_page_index, int npages, int zap) { struct mlx5_ib_dev *dev = mr->dev; struct device *ddev = dev->ib_dev.dma_device; struct umr_common *umrc = &dev->umrc; struct mlx5_ib_umr_context umr_context; struct ib_umem *umem = mr->umem; int size; __be64 *pas; dma_addr_t dma; const struct ib_send_wr *bad; struct mlx5_umr_wr wr; struct ib_sge sg; int err = 0; const int page_index_alignment = MLX5_UMR_MTT_ALIGNMENT / sizeof(u64); const int page_index_mask = page_index_alignment - 1; size_t pages_mapped = 0; size_t pages_to_map = 0; size_t pages_iter = 0; int use_emergency_buf = 0; /* UMR copies MTTs in units of MLX5_UMR_MTT_ALIGNMENT bytes, * so we need to align the offset and length accordingly */ if (start_page_index & page_index_mask) { npages += start_page_index & page_index_mask; start_page_index &= ~page_index_mask; } pages_to_map = ALIGN(npages, page_index_alignment); if (start_page_index + pages_to_map > MLX5_MAX_UMR_PAGES) return -EINVAL; size = sizeof(u64) * pages_to_map; size = min_t(int, PAGE_SIZE, size); /* We allocate with GFP_ATOMIC to avoid recursion into page-reclaim * code, when we are called from an invalidation. The pas buffer must * be 2k-aligned for Connect-IB. */ pas = (__be64 *)get_zeroed_page(GFP_ATOMIC); if (!pas) { mlx5_ib_warn(dev, "unable to allocate memory during MTT update, falling back to slower chunked mechanism.\n"); pas = mlx5_ib_update_mtt_emergency_buffer; size = MLX5_UMR_MTT_MIN_CHUNK_SIZE; use_emergency_buf = 1; mutex_lock(&mlx5_ib_update_mtt_emergency_buffer_mutex); memset(pas, 0, size); } pages_iter = size / sizeof(u64); dma = dma_map_single(ddev, pas, size, DMA_TO_DEVICE); if (dma_mapping_error(ddev, dma)) { mlx5_ib_err(dev, "unable to map DMA during MTT update.\n"); err = -ENOMEM; goto free_pas; } for (pages_mapped = 0; pages_mapped < pages_to_map && !err; pages_mapped += pages_iter, start_page_index += pages_iter) { dma_sync_single_for_cpu(ddev, dma, size, DMA_TO_DEVICE); npages = min_t(size_t, pages_iter, ib_umem_num_pages(umem) - start_page_index); if (!zap) { __mlx5_ib_populate_pas(dev, umem, PAGE_SHIFT, start_page_index, npages, pas, MLX5_IB_MTT_PRESENT); /* Clear padding after the pages brought from the * umem. */ memset(pas + npages, 0, size - npages * sizeof(u64)); } dma_sync_single_for_device(ddev, dma, size, DMA_TO_DEVICE); mlx5_ib_init_umr_context(&umr_context); memset(&wr, 0, sizeof(wr)); wr.wr.wr_cqe = &umr_context.cqe; sg.addr = dma; sg.length = ALIGN(npages * sizeof(u64), MLX5_UMR_MTT_ALIGNMENT); sg.lkey = dev->umrc.pd->local_dma_lkey; wr.wr.send_flags = MLX5_IB_SEND_UMR_FAIL_IF_FREE | MLX5_IB_SEND_UMR_UPDATE_MTT; wr.wr.sg_list = &sg; wr.wr.num_sge = 1; wr.wr.opcode = MLX5_IB_WR_UMR; wr.npages = sg.length / sizeof(u64); wr.page_shift = PAGE_SHIFT; wr.mkey = mr->mmkey.key; wr.target.offset = start_page_index; down(&umrc->sem); err = ib_post_send(umrc->qp, &wr.wr, &bad); if (err) { mlx5_ib_err(dev, "UMR post send failed, err %d\n", err); } else { wait_for_completion(&umr_context.done); if (umr_context.status != IB_WC_SUCCESS) { mlx5_ib_err(dev, "UMR completion failed, code %d\n", umr_context.status); err = -EFAULT; } } up(&umrc->sem); } dma_unmap_single(ddev, dma, size, DMA_TO_DEVICE); free_pas: if (!use_emergency_buf) free_page((unsigned long)pas); else mutex_unlock(&mlx5_ib_update_mtt_emergency_buffer_mutex); return err; } #endif /* * If ibmr is NULL it will be allocated by reg_create. * Else, the given ibmr will be used. */ static struct mlx5_ib_mr *reg_create(struct ib_mr *ibmr, struct ib_pd *pd, u64 virt_addr, u64 length, struct ib_umem *umem, int npages, int page_shift, int access_flags) { struct mlx5_ib_dev *dev = to_mdev(pd->device); struct mlx5_ib_mr *mr; __be64 *pas; void *mkc; int inlen; u32 *in; int err; bool pg_cap = !!(MLX5_CAP_GEN(dev->mdev, pg)); mr = ibmr ? to_mmr(ibmr) : kzalloc(sizeof(*mr), GFP_KERNEL); if (!mr) return ERR_PTR(-ENOMEM); inlen = MLX5_ST_SZ_BYTES(create_mkey_in) + sizeof(*pas) * ((npages + 1) / 2) * 2; in = mlx5_vzalloc(inlen); if (!in) { err = -ENOMEM; goto err_1; } pas = (__be64 *)MLX5_ADDR_OF(create_mkey_in, in, klm_pas_mtt); mlx5_ib_populate_pas(dev, umem, page_shift, pas, pg_cap ? MLX5_IB_MTT_PRESENT : 0); /* The pg_access bit allows setting the access flags * in the page list submitted with the command. */ MLX5_SET(create_mkey_in, in, pg_access, !!(pg_cap)); mkc = MLX5_ADDR_OF(create_mkey_in, in, memory_key_mkey_entry); MLX5_SET(mkc, mkc, access_mode, MLX5_ACCESS_MODE_MTT); MLX5_SET(mkc, mkc, a, !!(access_flags & IB_ACCESS_REMOTE_ATOMIC)); MLX5_SET(mkc, mkc, rw, !!(access_flags & IB_ACCESS_REMOTE_WRITE)); MLX5_SET(mkc, mkc, rr, !!(access_flags & IB_ACCESS_REMOTE_READ)); MLX5_SET(mkc, mkc, lw, !!(access_flags & IB_ACCESS_LOCAL_WRITE)); MLX5_SET(mkc, mkc, lr, 1); MLX5_SET64(mkc, mkc, start_addr, virt_addr); MLX5_SET64(mkc, mkc, len, length); MLX5_SET(mkc, mkc, pd, to_mpd(pd)->pdn); MLX5_SET(mkc, mkc, bsf_octword_size, 0); MLX5_SET(mkc, mkc, translations_octword_size, get_octo_len(virt_addr, length, 1 << page_shift)); MLX5_SET(mkc, mkc, log_page_size, page_shift); MLX5_SET(mkc, mkc, qpn, 0xffffff); MLX5_SET(create_mkey_in, in, translations_octword_actual_size, get_octo_len(virt_addr, length, 1 << page_shift)); err = mlx5_core_create_mkey(dev->mdev, &mr->mmkey, in, inlen); if (err) { mlx5_ib_warn(dev, "create mkey failed\n"); goto err_2; } mr->umem = umem; mr->dev = dev; mr->live = 1; kvfree(in); mlx5_ib_dbg(dev, "mkey = 0x%x\n", mr->mmkey.key); return mr; err_2: kvfree(in); err_1: if (!ibmr) kfree(mr); return ERR_PTR(err); } static void set_mr_fileds(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr, int npages, u64 length, int access_flags) { mr->npages = npages; atomic_add(npages, &dev->mdev->priv.reg_pages); mr->ibmr.lkey = mr->mmkey.key; mr->ibmr.rkey = mr->mmkey.key; mr->ibmr.length = length; mr->access_flags = access_flags; } struct ib_mr *mlx5_ib_reg_user_mr(struct ib_pd *pd, u64 start, u64 length, u64 virt_addr, int access_flags, struct ib_udata *udata) { struct mlx5_ib_dev *dev = to_mdev(pd->device); struct mlx5_ib_mr *mr = NULL; struct ib_umem *umem; int page_shift; int npages; int ncont; int order; int err; mlx5_ib_dbg(dev, "start 0x%llx, virt_addr 0x%llx, length 0x%llx, access_flags 0x%x\n", (long long)start, (long long)virt_addr, (long long)length, access_flags); umem = mr_umem_get(pd, start, length, access_flags, &npages, &page_shift, &ncont, &order); if (IS_ERR(umem)) return (void *)umem; if (use_umr(order)) { mr = reg_umr(pd, umem, virt_addr, length, ncont, page_shift, order, access_flags); if (PTR_ERR(mr) == -EAGAIN) { mlx5_ib_dbg(dev, "cache empty for order %d", order); mr = NULL; } } else if (access_flags & IB_ACCESS_ON_DEMAND) { err = -EINVAL; pr_err("Got MR registration for ODP MR > 512MB, not supported for Connect-IB"); goto error; } if (!mr) { mutex_lock(&dev->slow_path_mutex); mr = reg_create(NULL, pd, virt_addr, length, umem, ncont, page_shift, access_flags); mutex_unlock(&dev->slow_path_mutex); } if (IS_ERR(mr)) { err = PTR_ERR(mr); goto error; } mlx5_ib_dbg(dev, "mkey 0x%x\n", mr->mmkey.key); mr->umem = umem; set_mr_fileds(dev, mr, npages, length, access_flags); #ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING update_odp_mr(mr); #endif return &mr->ibmr; error: ib_umem_release(umem); return ERR_PTR(err); } static int unreg_umr(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr) { struct mlx5_core_dev *mdev = dev->mdev; struct umr_common *umrc = &dev->umrc; struct mlx5_ib_umr_context umr_context; struct mlx5_umr_wr umrwr = {}; const struct ib_send_wr *bad; int err; if (mdev->state == MLX5_DEVICE_STATE_INTERNAL_ERROR) return 0; mlx5_ib_init_umr_context(&umr_context); umrwr.wr.wr_cqe = &umr_context.cqe; prep_umr_unreg_wqe(dev, &umrwr, mr->mmkey.key); down(&umrc->sem); err = ib_post_send(umrc->qp, &umrwr.wr, &bad); if (err) { up(&umrc->sem); mlx5_ib_dbg(dev, "err %d\n", err); goto error; } else { wait_for_completion(&umr_context.done); up(&umrc->sem); } if (umr_context.status != IB_WC_SUCCESS) { mlx5_ib_warn(dev, "unreg umr failed\n"); err = -EFAULT; goto error; } return 0; error: return err; } static int rereg_umr(struct ib_pd *pd, struct mlx5_ib_mr *mr, u64 virt_addr, u64 length, int npages, int page_shift, int order, int access_flags, int flags) { struct mlx5_ib_dev *dev = to_mdev(pd->device); struct device *ddev = dev->ib_dev.dma_device; struct mlx5_ib_umr_context umr_context; const struct ib_send_wr *bad; struct mlx5_umr_wr umrwr = {}; struct ib_sge sg; struct umr_common *umrc = &dev->umrc; dma_addr_t dma = 0; __be64 *mr_pas = NULL; int size; int err; mlx5_ib_init_umr_context(&umr_context); umrwr.wr.wr_cqe = &umr_context.cqe; umrwr.wr.send_flags = MLX5_IB_SEND_UMR_FAIL_IF_FREE; if (flags & IB_MR_REREG_TRANS) { err = dma_map_mr_pas(dev, mr->umem, npages, page_shift, &size, &mr_pas, &dma); if (err) return err; umrwr.target.virt_addr = virt_addr; umrwr.length = length; umrwr.wr.send_flags |= MLX5_IB_SEND_UMR_UPDATE_TRANSLATION; } prep_umr_wqe_common(pd, &umrwr, &sg, dma, npages, mr->mmkey.key, page_shift); if (flags & IB_MR_REREG_PD) { umrwr.pd = pd; umrwr.wr.send_flags |= MLX5_IB_SEND_UMR_UPDATE_PD; } if (flags & IB_MR_REREG_ACCESS) { umrwr.access_flags = access_flags; umrwr.wr.send_flags |= MLX5_IB_SEND_UMR_UPDATE_ACCESS; } /* post send request to UMR QP */ down(&umrc->sem); err = ib_post_send(umrc->qp, &umrwr.wr, &bad); if (err) { mlx5_ib_warn(dev, "post send failed, err %d\n", err); } else { wait_for_completion(&umr_context.done); if (umr_context.status != IB_WC_SUCCESS) { mlx5_ib_warn(dev, "reg umr failed (%u)\n", umr_context.status); err = -EFAULT; } } up(&umrc->sem); if (flags & IB_MR_REREG_TRANS) { dma_unmap_single(ddev, dma, size, DMA_TO_DEVICE); kfree(mr_pas); } return err; } int mlx5_ib_rereg_user_mr(struct ib_mr *ib_mr, int flags, u64 start, u64 length, u64 virt_addr, int new_access_flags, struct ib_pd *new_pd, struct ib_udata *udata) { struct mlx5_ib_dev *dev = to_mdev(ib_mr->device); struct mlx5_ib_mr *mr = to_mmr(ib_mr); struct ib_pd *pd = (flags & IB_MR_REREG_PD) ? new_pd : ib_mr->pd; int access_flags = flags & IB_MR_REREG_ACCESS ? new_access_flags : mr->access_flags; u64 addr = (flags & IB_MR_REREG_TRANS) ? virt_addr : mr->umem->address; u64 len = (flags & IB_MR_REREG_TRANS) ? length : mr->umem->length; int page_shift = 0; int npages = 0; int ncont = 0; int order = 0; int err; mlx5_ib_dbg(dev, "start 0x%llx, virt_addr 0x%llx, length 0x%llx, access_flags 0x%x\n", (long long)start, (long long)virt_addr, (long long)length, access_flags); if (flags != IB_MR_REREG_PD) { /* * Replace umem. This needs to be done whether or not UMR is * used. */ flags |= IB_MR_REREG_TRANS; ib_umem_release(mr->umem); mr->umem = mr_umem_get(pd, addr, len, access_flags, &npages, &page_shift, &ncont, &order); if (IS_ERR(mr->umem)) { err = PTR_ERR(mr->umem); mr->umem = NULL; return err; } } if (flags & IB_MR_REREG_TRANS && !use_umr_mtt_update(mr, addr, len)) { /* * UMR can't be used - MKey needs to be replaced. */ if (mr->umred) { err = unreg_umr(dev, mr); if (err) mlx5_ib_warn(dev, "Failed to unregister MR\n"); } else { err = destroy_mkey(dev, mr); if (err) mlx5_ib_warn(dev, "Failed to destroy MKey\n"); } if (err) return err; mr = reg_create(ib_mr, pd, addr, len, mr->umem, ncont, page_shift, access_flags); if (IS_ERR(mr)) return PTR_ERR(mr); mr->umred = 0; } else { /* * Send a UMR WQE */ err = rereg_umr(pd, mr, addr, len, npages, page_shift, order, access_flags, flags); if (err) { mlx5_ib_warn(dev, "Failed to rereg UMR\n"); return err; } } if (flags & IB_MR_REREG_PD) { ib_mr->pd = pd; mr->mmkey.pd = to_mpd(pd)->pdn; } if (flags & IB_MR_REREG_ACCESS) mr->access_flags = access_flags; if (flags & IB_MR_REREG_TRANS) { atomic_sub(mr->npages, &dev->mdev->priv.reg_pages); set_mr_fileds(dev, mr, npages, len, access_flags); mr->mmkey.iova = addr; mr->mmkey.size = len; } #ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING update_odp_mr(mr); #endif return 0; } static int mlx5_alloc_priv_descs(struct ib_device *device, struct mlx5_ib_mr *mr, int ndescs, int desc_size) { int size = ndescs * desc_size; int add_size; int ret; add_size = max_t(int, MLX5_UMR_ALIGN - 1, 0); mr->descs_alloc = kzalloc(size + add_size, GFP_KERNEL); if (!mr->descs_alloc) return -ENOMEM; mr->descs = PTR_ALIGN(mr->descs_alloc, MLX5_UMR_ALIGN); mr->desc_map = dma_map_single(device->dma_device, mr->descs, size, DMA_TO_DEVICE); if (dma_mapping_error(device->dma_device, mr->desc_map)) { ret = -ENOMEM; goto err; } return 0; err: kfree(mr->descs_alloc); return ret; } static void mlx5_free_priv_descs(struct mlx5_ib_mr *mr) { if (mr->descs) { struct ib_device *device = mr->ibmr.device; int size = mr->max_descs * mr->desc_size; dma_unmap_single(device->dma_device, mr->desc_map, size, DMA_TO_DEVICE); kfree(mr->descs_alloc); mr->descs = NULL; } } static int clean_mr(struct mlx5_ib_mr *mr) { struct mlx5_ib_dev *dev = to_mdev(mr->ibmr.device); int umred = mr->umred; int err; if (mr->sig) { if (mlx5_core_destroy_psv(dev->mdev, mr->sig->psv_memory.psv_idx)) mlx5_ib_warn(dev, "failed to destroy mem psv %d\n", mr->sig->psv_memory.psv_idx); if (mlx5_core_destroy_psv(dev->mdev, mr->sig->psv_wire.psv_idx)) mlx5_ib_warn(dev, "failed to destroy wire psv %d\n", mr->sig->psv_wire.psv_idx); kfree(mr->sig); mr->sig = NULL; } mlx5_free_priv_descs(mr); if (!umred) { err = destroy_mkey(dev, mr); if (err) { mlx5_ib_warn(dev, "failed to destroy mkey 0x%x (%d)\n", mr->mmkey.key, err); return err; } } else { err = unreg_umr(dev, mr); if (err) { mlx5_ib_warn(dev, "failed unregister\n"); return err; } free_cached_mr(dev, mr); } if (!umred) kfree(mr); return 0; } int mlx5_ib_dereg_mr(struct ib_mr *ibmr) { struct mlx5_ib_dev *dev = to_mdev(ibmr->device); struct mlx5_ib_mr *mr = to_mmr(ibmr); int npages = mr->npages; struct ib_umem *umem = mr->umem; #ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING if (umem && umem->odp_data) { /* Prevent new page faults from succeeding */ mr->live = 0; /* Wait for all running page-fault handlers to finish. */ synchronize_srcu(&dev->mr_srcu); /* Destroy all page mappings */ mlx5_ib_invalidate_range(umem, ib_umem_start(umem), ib_umem_end(umem)); /* * We kill the umem before the MR for ODP, * so that there will not be any invalidations in * flight, looking at the *mr struct. */ ib_umem_release(umem); atomic_sub(npages, &dev->mdev->priv.reg_pages); /* Avoid double-freeing the umem. */ umem = NULL; } #endif clean_mr(mr); if (umem) { ib_umem_release(umem); atomic_sub(npages, &dev->mdev->priv.reg_pages); } return 0; } struct ib_mr *mlx5_ib_alloc_mr(struct ib_pd *pd, enum ib_mr_type mr_type, u32 max_num_sg) { struct mlx5_ib_dev *dev = to_mdev(pd->device); int inlen = MLX5_ST_SZ_BYTES(create_mkey_in); int ndescs = ALIGN(max_num_sg, 4); struct mlx5_ib_mr *mr; void *mkc; u32 *in; int err; mr = kzalloc(sizeof(*mr), GFP_KERNEL); if (!mr) return ERR_PTR(-ENOMEM); in = kzalloc(inlen, GFP_KERNEL); if (!in) { err = -ENOMEM; goto err_free; } mkc = MLX5_ADDR_OF(create_mkey_in, in, memory_key_mkey_entry); MLX5_SET(mkc, mkc, free, 1); MLX5_SET(mkc, mkc, translations_octword_size, ndescs); MLX5_SET(mkc, mkc, qpn, 0xffffff); MLX5_SET(mkc, mkc, pd, to_mpd(pd)->pdn); if (mr_type == IB_MR_TYPE_MEM_REG) { mr->access_mode = MLX5_ACCESS_MODE_MTT; MLX5_SET(mkc, mkc, log_page_size, PAGE_SHIFT); err = mlx5_alloc_priv_descs(pd->device, mr, ndescs, sizeof(u64)); if (err) goto err_free_in; mr->desc_size = sizeof(u64); mr->max_descs = ndescs; } else if (mr_type == IB_MR_TYPE_SG_GAPS) { mr->access_mode = MLX5_ACCESS_MODE_KLM; err = mlx5_alloc_priv_descs(pd->device, mr, ndescs, sizeof(struct mlx5_klm)); if (err) goto err_free_in; mr->desc_size = sizeof(struct mlx5_klm); mr->max_descs = ndescs; } else if (mr_type == IB_MR_TYPE_SIGNATURE) { u32 psv_index[2]; MLX5_SET(mkc, mkc, bsf_en, 1); MLX5_SET(mkc, mkc, bsf_octword_size, MLX5_MKEY_BSF_OCTO_SIZE); mr->sig = kzalloc(sizeof(*mr->sig), GFP_KERNEL); if (!mr->sig) { err = -ENOMEM; goto err_free_in; } /* create mem & wire PSVs */ err = mlx5_core_create_psv(dev->mdev, to_mpd(pd)->pdn, 2, psv_index); if (err) goto err_free_sig; mr->access_mode = MLX5_ACCESS_MODE_KLM; mr->sig->psv_memory.psv_idx = psv_index[0]; mr->sig->psv_wire.psv_idx = psv_index[1]; mr->sig->sig_status_checked = true; mr->sig->sig_err_exists = false; /* Next UMR, Arm SIGERR */ ++mr->sig->sigerr_count; } else { mlx5_ib_warn(dev, "Invalid mr type %d\n", mr_type); err = -EINVAL; goto err_free_in; } MLX5_SET(mkc, mkc, access_mode, mr->access_mode); MLX5_SET(mkc, mkc, umr_en, 1); err = mlx5_core_create_mkey(dev->mdev, &mr->mmkey, in, inlen); if (err) goto err_destroy_psv; mr->ibmr.lkey = mr->mmkey.key; mr->ibmr.rkey = mr->mmkey.key; mr->umem = NULL; kfree(in); return &mr->ibmr; err_destroy_psv: if (mr->sig) { if (mlx5_core_destroy_psv(dev->mdev, mr->sig->psv_memory.psv_idx)) mlx5_ib_warn(dev, "failed to destroy mem psv %d\n", mr->sig->psv_memory.psv_idx); if (mlx5_core_destroy_psv(dev->mdev, mr->sig->psv_wire.psv_idx)) mlx5_ib_warn(dev, "failed to destroy wire psv %d\n", mr->sig->psv_wire.psv_idx); } mlx5_free_priv_descs(mr); err_free_sig: kfree(mr->sig); err_free_in: kfree(in); err_free: kfree(mr); return ERR_PTR(err); } struct ib_mw *mlx5_ib_alloc_mw(struct ib_pd *pd, enum ib_mw_type type, struct ib_udata *udata) { struct mlx5_ib_dev *dev = to_mdev(pd->device); int inlen = MLX5_ST_SZ_BYTES(create_mkey_in); struct mlx5_ib_mw *mw = NULL; u32 *in = NULL; void *mkc; int ndescs; int err; struct mlx5_ib_alloc_mw req = {}; struct { __u32 comp_mask; __u32 response_length; } resp = {}; err = ib_copy_from_udata(&req, udata, min(udata->inlen, sizeof(req))); if (err) return ERR_PTR(err); if (req.comp_mask || req.reserved1 || req.reserved2) return ERR_PTR(-EOPNOTSUPP); if (udata->inlen > sizeof(req) && !ib_is_udata_cleared(udata, sizeof(req), udata->inlen - sizeof(req))) return ERR_PTR(-EOPNOTSUPP); ndescs = req.num_klms ? roundup(req.num_klms, 4) : roundup(1, 4); mw = kzalloc(sizeof(*mw), GFP_KERNEL); in = kzalloc(inlen, GFP_KERNEL); if (!mw || !in) { err = -ENOMEM; goto free; } mkc = MLX5_ADDR_OF(create_mkey_in, in, memory_key_mkey_entry); MLX5_SET(mkc, mkc, free, 1); MLX5_SET(mkc, mkc, translations_octword_size, ndescs); MLX5_SET(mkc, mkc, pd, to_mpd(pd)->pdn); MLX5_SET(mkc, mkc, umr_en, 1); MLX5_SET(mkc, mkc, lr, 1); MLX5_SET(mkc, mkc, access_mode, MLX5_ACCESS_MODE_KLM); MLX5_SET(mkc, mkc, en_rinval, !!((type == IB_MW_TYPE_2))); MLX5_SET(mkc, mkc, qpn, 0xffffff); err = mlx5_core_create_mkey(dev->mdev, &mw->mmkey, in, inlen); if (err) goto free; mw->ibmw.rkey = mw->mmkey.key; resp.response_length = min(offsetof(typeof(resp), response_length) + sizeof(resp.response_length), udata->outlen); if (resp.response_length) { err = ib_copy_to_udata(udata, &resp, resp.response_length); if (err) { mlx5_core_destroy_mkey(dev->mdev, &mw->mmkey); goto free; } } kfree(in); return &mw->ibmw; free: kfree(mw); kfree(in); return ERR_PTR(err); } int mlx5_ib_dealloc_mw(struct ib_mw *mw) { struct mlx5_ib_mw *mmw = to_mmw(mw); int err; err = mlx5_core_destroy_mkey((to_mdev(mw->device))->mdev, &mmw->mmkey); if (!err) kfree(mmw); return err; } int mlx5_ib_check_mr_status(struct ib_mr *ibmr, u32 check_mask, struct ib_mr_status *mr_status) { struct mlx5_ib_mr *mmr = to_mmr(ibmr); int ret = 0; if (check_mask & ~IB_MR_CHECK_SIG_STATUS) { pr_err("Invalid status check mask\n"); ret = -EINVAL; goto done; } mr_status->fail_status = 0; if (check_mask & IB_MR_CHECK_SIG_STATUS) { if (!mmr->sig) { ret = -EINVAL; pr_err("signature status check requested on a non-signature enabled MR\n"); goto done; } mmr->sig->sig_status_checked = true; if (!mmr->sig->sig_err_exists) goto done; if (ibmr->lkey == mmr->sig->err_item.key) memcpy(&mr_status->sig_err, &mmr->sig->err_item, sizeof(mr_status->sig_err)); else { mr_status->sig_err.err_type = IB_SIG_BAD_GUARD; mr_status->sig_err.sig_err_offset = 0; mr_status->sig_err.key = mmr->sig->err_item.key; } mmr->sig->sig_err_exists = false; mr_status->fail_status |= IB_MR_CHECK_SIG_STATUS; } done: return ret; } static int mlx5_ib_sg_to_klms(struct mlx5_ib_mr *mr, struct scatterlist *sgl, unsigned short sg_nents, unsigned int *sg_offset_p) { struct scatterlist *sg = sgl; struct mlx5_klm *klms = mr->descs; unsigned int sg_offset = sg_offset_p ? *sg_offset_p : 0; u32 lkey = mr->ibmr.pd->local_dma_lkey; int i; mr->ibmr.iova = sg_dma_address(sg) + sg_offset; mr->ibmr.length = 0; mr->ndescs = sg_nents; for_each_sg(sgl, sg, sg_nents, i) { if (unlikely(i > mr->max_descs)) break; klms[i].va = cpu_to_be64(sg_dma_address(sg) + sg_offset); klms[i].bcount = cpu_to_be32(sg_dma_len(sg) - sg_offset); klms[i].key = cpu_to_be32(lkey); mr->ibmr.length += sg_dma_len(sg); sg_offset = 0; } if (sg_offset_p) *sg_offset_p = sg_offset; return i; } static int mlx5_set_page(struct ib_mr *ibmr, u64 addr) { struct mlx5_ib_mr *mr = to_mmr(ibmr); __be64 *descs; if (unlikely(mr->ndescs == mr->max_descs)) return -ENOMEM; descs = mr->descs; descs[mr->ndescs++] = cpu_to_be64(addr | MLX5_EN_RD | MLX5_EN_WR); return 0; } int mlx5_ib_map_mr_sg(struct ib_mr *ibmr, struct scatterlist *sg, int sg_nents, unsigned int *sg_offset) { struct mlx5_ib_mr *mr = to_mmr(ibmr); int n; mr->ndescs = 0; ib_dma_sync_single_for_cpu(ibmr->device, mr->desc_map, mr->desc_size * mr->max_descs, DMA_TO_DEVICE); if (mr->access_mode == MLX5_ACCESS_MODE_KLM) n = mlx5_ib_sg_to_klms(mr, sg, sg_nents, sg_offset); else n = ib_sg_to_pages(ibmr, sg, sg_nents, sg_offset, mlx5_set_page); ib_dma_sync_single_for_device(ibmr->device, mr->desc_map, mr->desc_size * mr->max_descs, DMA_TO_DEVICE); return n; } diff --git a/sys/dev/mlx5/mlx5_ib/mlx5_ib_qp.c b/sys/dev/mlx5/mlx5_ib/mlx5_ib_qp.c index 0ee66192a191..d930d08208c8 100644 --- a/sys/dev/mlx5/mlx5_ib/mlx5_ib_qp.c +++ b/sys/dev/mlx5/mlx5_ib/mlx5_ib_qp.c @@ -1,5006 +1,5006 @@ /*- * Copyright (c) 2013-2021, Mellanox Technologies. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY AUTHOR AND CONTRIBUTORS `AS IS' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $FreeBSD$ */ #include "opt_rss.h" #include "opt_ratelimit.h" #include #include #include #include -#include "mlx5_ib.h" +#include /* not supported currently */ static int wq_signature; enum { MLX5_IB_ACK_REQ_FREQ = 8, }; enum { MLX5_IB_DEFAULT_SCHED_QUEUE = 0x83, MLX5_IB_DEFAULT_QP0_SCHED_QUEUE = 0x3f, MLX5_IB_LINK_TYPE_IB = 0, MLX5_IB_LINK_TYPE_ETH = 1 }; enum { MLX5_IB_SQ_STRIDE = 6, }; static const u32 mlx5_ib_opcode[] = { [IB_WR_SEND] = MLX5_OPCODE_SEND, [IB_WR_LSO] = MLX5_OPCODE_LSO, [IB_WR_SEND_WITH_IMM] = MLX5_OPCODE_SEND_IMM, [IB_WR_RDMA_WRITE] = MLX5_OPCODE_RDMA_WRITE, [IB_WR_RDMA_WRITE_WITH_IMM] = MLX5_OPCODE_RDMA_WRITE_IMM, [IB_WR_RDMA_READ] = MLX5_OPCODE_RDMA_READ, [IB_WR_ATOMIC_CMP_AND_SWP] = MLX5_OPCODE_ATOMIC_CS, [IB_WR_ATOMIC_FETCH_AND_ADD] = MLX5_OPCODE_ATOMIC_FA, [IB_WR_SEND_WITH_INV] = MLX5_OPCODE_SEND_INVAL, [IB_WR_LOCAL_INV] = MLX5_OPCODE_UMR, [IB_WR_REG_MR] = MLX5_OPCODE_UMR, [IB_WR_MASKED_ATOMIC_CMP_AND_SWP] = MLX5_OPCODE_ATOMIC_MASKED_CS, [IB_WR_MASKED_ATOMIC_FETCH_AND_ADD] = MLX5_OPCODE_ATOMIC_MASKED_FA, [MLX5_IB_WR_UMR] = MLX5_OPCODE_UMR, }; struct mlx5_wqe_eth_pad { u8 rsvd0[16]; }; enum raw_qp_set_mask_map { MLX5_RAW_QP_MOD_SET_RQ_Q_CTR_ID = 1UL << 0, }; struct mlx5_modify_raw_qp_param { u16 operation; u32 set_mask; /* raw_qp_set_mask_map */ u8 rq_q_ctr_id; }; static void get_cqs(enum ib_qp_type qp_type, struct ib_cq *ib_send_cq, struct ib_cq *ib_recv_cq, struct mlx5_ib_cq **send_cq, struct mlx5_ib_cq **recv_cq); static int is_qp0(enum ib_qp_type qp_type) { return qp_type == IB_QPT_SMI; } static int is_sqp(enum ib_qp_type qp_type) { return is_qp0(qp_type) || is_qp1(qp_type); } static void *get_wqe(struct mlx5_ib_qp *qp, int offset) { return mlx5_buf_offset(&qp->buf, offset); } static void *get_recv_wqe(struct mlx5_ib_qp *qp, int n) { return get_wqe(qp, qp->rq.offset + (n << qp->rq.wqe_shift)); } void *mlx5_get_send_wqe(struct mlx5_ib_qp *qp, int n) { return get_wqe(qp, qp->sq.offset + (n << MLX5_IB_SQ_STRIDE)); } /** * mlx5_ib_read_user_wqe() - Copy a user-space WQE to kernel space. * * @qp: QP to copy from. * @send: copy from the send queue when non-zero, use the receive queue * otherwise. * @wqe_index: index to start copying from. For send work queues, the * wqe_index is in units of MLX5_SEND_WQE_BB. * For receive work queue, it is the number of work queue * element in the queue. * @buffer: destination buffer. * @length: maximum number of bytes to copy. * * Copies at least a single WQE, but may copy more data. * * Return: the number of bytes copied, or an error code. */ int mlx5_ib_read_user_wqe(struct mlx5_ib_qp *qp, int send, int wqe_index, void *buffer, u32 length, struct mlx5_ib_qp_base *base) { struct ib_device *ibdev = qp->ibqp.device; struct mlx5_ib_dev *dev = to_mdev(ibdev); struct mlx5_ib_wq *wq = send ? &qp->sq : &qp->rq; size_t offset; size_t wq_end; struct ib_umem *umem = base->ubuffer.umem; u32 first_copy_length; int wqe_length; int ret; if (wq->wqe_cnt == 0) { mlx5_ib_dbg(dev, "mlx5_ib_read_user_wqe for a QP with wqe_cnt == 0. qp_type: 0x%x\n", qp->ibqp.qp_type); return -EINVAL; } offset = wq->offset + ((wqe_index % wq->wqe_cnt) << wq->wqe_shift); wq_end = wq->offset + (wq->wqe_cnt << wq->wqe_shift); if (send && length < sizeof(struct mlx5_wqe_ctrl_seg)) return -EINVAL; if (offset > umem->length || (send && offset + sizeof(struct mlx5_wqe_ctrl_seg) > umem->length)) return -EINVAL; first_copy_length = min_t(u32, offset + length, wq_end) - offset; ret = ib_umem_copy_from(buffer, umem, offset, first_copy_length); if (ret) return ret; if (send) { struct mlx5_wqe_ctrl_seg *ctrl = buffer; int ds = be32_to_cpu(ctrl->qpn_ds) & MLX5_WQE_CTRL_DS_MASK; wqe_length = ds * MLX5_WQE_DS_UNITS; } else { wqe_length = 1 << wq->wqe_shift; } if (wqe_length <= first_copy_length) return first_copy_length; ret = ib_umem_copy_from(buffer + first_copy_length, umem, wq->offset, wqe_length - first_copy_length); if (ret) return ret; return wqe_length; } static void mlx5_ib_qp_event(struct mlx5_core_qp *qp, int type) { struct ib_qp *ibqp = &to_mibqp(qp)->ibqp; struct ib_event event; if (type == MLX5_EVENT_TYPE_PATH_MIG) { /* This event is only valid for trans_qps */ to_mibqp(qp)->port = to_mibqp(qp)->trans_qp.alt_port; } if (ibqp->event_handler) { event.device = ibqp->device; event.element.qp = ibqp; switch (type) { case MLX5_EVENT_TYPE_PATH_MIG: event.event = IB_EVENT_PATH_MIG; break; case MLX5_EVENT_TYPE_COMM_EST: event.event = IB_EVENT_COMM_EST; break; case MLX5_EVENT_TYPE_SQ_DRAINED: event.event = IB_EVENT_SQ_DRAINED; break; case MLX5_EVENT_TYPE_SRQ_LAST_WQE: event.event = IB_EVENT_QP_LAST_WQE_REACHED; break; case MLX5_EVENT_TYPE_WQ_CATAS_ERROR: event.event = IB_EVENT_QP_FATAL; break; case MLX5_EVENT_TYPE_PATH_MIG_FAILED: event.event = IB_EVENT_PATH_MIG_ERR; break; case MLX5_EVENT_TYPE_WQ_INVAL_REQ_ERROR: event.event = IB_EVENT_QP_REQ_ERR; break; case MLX5_EVENT_TYPE_WQ_ACCESS_ERROR: event.event = IB_EVENT_QP_ACCESS_ERR; break; default: pr_warn("mlx5_ib: Unexpected event type %d on QP %06x\n", type, qp->qpn); return; } ibqp->event_handler(&event, ibqp->qp_context); } } static int set_rq_size(struct mlx5_ib_dev *dev, struct ib_qp_cap *cap, int has_rq, struct mlx5_ib_qp *qp, struct mlx5_ib_create_qp *ucmd) { int wqe_size; int wq_size; /* Sanity check RQ size before proceeding */ if (cap->max_recv_wr > (1 << MLX5_CAP_GEN(dev->mdev, log_max_qp_sz))) return -EINVAL; if (!has_rq) { qp->rq.max_gs = 0; qp->rq.wqe_cnt = 0; qp->rq.wqe_shift = 0; cap->max_recv_wr = 0; cap->max_recv_sge = 0; } else { if (ucmd) { qp->rq.wqe_cnt = ucmd->rq_wqe_count; qp->rq.wqe_shift = ucmd->rq_wqe_shift; qp->rq.max_gs = (1 << qp->rq.wqe_shift) / sizeof(struct mlx5_wqe_data_seg) - qp->wq_sig; qp->rq.max_post = qp->rq.wqe_cnt; } else { wqe_size = qp->wq_sig ? sizeof(struct mlx5_wqe_signature_seg) : 0; wqe_size += cap->max_recv_sge * sizeof(struct mlx5_wqe_data_seg); wqe_size = roundup_pow_of_two(wqe_size); wq_size = roundup_pow_of_two(cap->max_recv_wr) * wqe_size; wq_size = max_t(int, wq_size, MLX5_SEND_WQE_BB); qp->rq.wqe_cnt = wq_size / wqe_size; if (wqe_size > MLX5_CAP_GEN(dev->mdev, max_wqe_sz_rq)) { mlx5_ib_dbg(dev, "wqe_size %d, max %d\n", wqe_size, MLX5_CAP_GEN(dev->mdev, max_wqe_sz_rq)); return -EINVAL; } qp->rq.wqe_shift = ilog2(wqe_size); qp->rq.max_gs = (1 << qp->rq.wqe_shift) / sizeof(struct mlx5_wqe_data_seg) - qp->wq_sig; qp->rq.max_post = qp->rq.wqe_cnt; } } return 0; } static int sq_overhead(struct ib_qp_init_attr *attr) { int size = 0; switch (attr->qp_type) { case IB_QPT_XRC_INI: size += sizeof(struct mlx5_wqe_xrc_seg); /* fall through */ case IB_QPT_RC: size += sizeof(struct mlx5_wqe_ctrl_seg) + max(sizeof(struct mlx5_wqe_atomic_seg) + sizeof(struct mlx5_wqe_raddr_seg), sizeof(struct mlx5_wqe_umr_ctrl_seg) + sizeof(struct mlx5_mkey_seg)); break; case IB_QPT_XRC_TGT: return 0; case IB_QPT_UC: size += sizeof(struct mlx5_wqe_ctrl_seg) + max(sizeof(struct mlx5_wqe_raddr_seg), sizeof(struct mlx5_wqe_umr_ctrl_seg) + sizeof(struct mlx5_mkey_seg)); break; case IB_QPT_UD: if (attr->create_flags & IB_QP_CREATE_IPOIB_UD_LSO) size += sizeof(struct mlx5_wqe_eth_pad) + sizeof(struct mlx5_wqe_eth_seg); /* fall through */ case IB_QPT_SMI: case MLX5_IB_QPT_HW_GSI: size += sizeof(struct mlx5_wqe_ctrl_seg) + sizeof(struct mlx5_wqe_datagram_seg); break; case MLX5_IB_QPT_REG_UMR: size += sizeof(struct mlx5_wqe_ctrl_seg) + sizeof(struct mlx5_wqe_umr_ctrl_seg) + sizeof(struct mlx5_mkey_seg); break; default: return -EINVAL; } return size; } static int calc_send_wqe(struct ib_qp_init_attr *attr) { int inl_size = 0; int size; size = sq_overhead(attr); if (size < 0) return size; if (attr->cap.max_inline_data) { inl_size = size + sizeof(struct mlx5_wqe_inline_seg) + attr->cap.max_inline_data; } size += attr->cap.max_send_sge * sizeof(struct mlx5_wqe_data_seg); if (attr->create_flags & IB_QP_CREATE_SIGNATURE_EN && ALIGN(max_t(int, inl_size, size), MLX5_SEND_WQE_BB) < MLX5_SIG_WQE_SIZE) return MLX5_SIG_WQE_SIZE; else return ALIGN(max_t(int, inl_size, size), MLX5_SEND_WQE_BB); } static int get_send_sge(struct ib_qp_init_attr *attr, int wqe_size) { int max_sge; if (attr->qp_type == IB_QPT_RC) max_sge = (min_t(int, wqe_size, 512) - sizeof(struct mlx5_wqe_ctrl_seg) - sizeof(struct mlx5_wqe_raddr_seg)) / sizeof(struct mlx5_wqe_data_seg); else if (attr->qp_type == IB_QPT_XRC_INI) max_sge = (min_t(int, wqe_size, 512) - sizeof(struct mlx5_wqe_ctrl_seg) - sizeof(struct mlx5_wqe_xrc_seg) - sizeof(struct mlx5_wqe_raddr_seg)) / sizeof(struct mlx5_wqe_data_seg); else max_sge = (wqe_size - sq_overhead(attr)) / sizeof(struct mlx5_wqe_data_seg); return min_t(int, max_sge, wqe_size - sq_overhead(attr) / sizeof(struct mlx5_wqe_data_seg)); } static int calc_sq_size(struct mlx5_ib_dev *dev, struct ib_qp_init_attr *attr, struct mlx5_ib_qp *qp) { int wqe_size; int wq_size; if (!attr->cap.max_send_wr) return 0; wqe_size = calc_send_wqe(attr); mlx5_ib_dbg(dev, "wqe_size %d\n", wqe_size); if (wqe_size < 0) return wqe_size; if (wqe_size > MLX5_CAP_GEN(dev->mdev, max_wqe_sz_sq)) { mlx5_ib_dbg(dev, "wqe_size(%d) > max_sq_desc_sz(%d)\n", wqe_size, MLX5_CAP_GEN(dev->mdev, max_wqe_sz_sq)); return -EINVAL; } qp->max_inline_data = wqe_size - sq_overhead(attr) - sizeof(struct mlx5_wqe_inline_seg); attr->cap.max_inline_data = qp->max_inline_data; if (attr->create_flags & IB_QP_CREATE_SIGNATURE_EN) qp->signature_en = true; wq_size = roundup_pow_of_two(attr->cap.max_send_wr * wqe_size); qp->sq.wqe_cnt = wq_size / MLX5_SEND_WQE_BB; if (qp->sq.wqe_cnt > (1 << MLX5_CAP_GEN(dev->mdev, log_max_qp_sz))) { mlx5_ib_dbg(dev, "wqe count(%d) exceeds limits(%d)\n", qp->sq.wqe_cnt, 1 << MLX5_CAP_GEN(dev->mdev, log_max_qp_sz)); return -ENOMEM; } qp->sq.wqe_shift = ilog2(MLX5_SEND_WQE_BB); qp->sq.max_gs = get_send_sge(attr, wqe_size); if (qp->sq.max_gs < attr->cap.max_send_sge) return -ENOMEM; attr->cap.max_send_sge = qp->sq.max_gs; qp->sq.max_post = wq_size / wqe_size; attr->cap.max_send_wr = qp->sq.max_post; return wq_size; } static int set_user_buf_size(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, struct mlx5_ib_create_qp *ucmd, struct mlx5_ib_qp_base *base, struct ib_qp_init_attr *attr) { int desc_sz = 1 << qp->sq.wqe_shift; if (desc_sz > MLX5_CAP_GEN(dev->mdev, max_wqe_sz_sq)) { mlx5_ib_warn(dev, "desc_sz %d, max_sq_desc_sz %d\n", desc_sz, MLX5_CAP_GEN(dev->mdev, max_wqe_sz_sq)); return -EINVAL; } if (ucmd->sq_wqe_count && ((1 << ilog2(ucmd->sq_wqe_count)) != ucmd->sq_wqe_count)) { mlx5_ib_warn(dev, "sq_wqe_count %d, sq_wqe_count %d\n", ucmd->sq_wqe_count, ucmd->sq_wqe_count); return -EINVAL; } qp->sq.wqe_cnt = ucmd->sq_wqe_count; if (qp->sq.wqe_cnt > (1 << MLX5_CAP_GEN(dev->mdev, log_max_qp_sz))) { mlx5_ib_warn(dev, "wqe_cnt %d, max_wqes %d\n", qp->sq.wqe_cnt, 1 << MLX5_CAP_GEN(dev->mdev, log_max_qp_sz)); return -EINVAL; } if (attr->qp_type == IB_QPT_RAW_PACKET) { base->ubuffer.buf_size = qp->rq.wqe_cnt << qp->rq.wqe_shift; qp->raw_packet_qp.sq.ubuffer.buf_size = qp->sq.wqe_cnt << 6; } else { base->ubuffer.buf_size = (qp->rq.wqe_cnt << qp->rq.wqe_shift) + (qp->sq.wqe_cnt << 6); } return 0; } static int qp_has_rq(struct ib_qp_init_attr *attr) { if (attr->qp_type == IB_QPT_XRC_INI || attr->qp_type == IB_QPT_XRC_TGT || attr->srq || attr->qp_type == MLX5_IB_QPT_REG_UMR || !attr->cap.max_recv_wr) return 0; return 1; } enum { /* this is the first blue flame register in the array of bfregs assigned * to a processes. Since we do not use it for blue flame but rather * regular 64 bit doorbells, we do not need a lock for maintaiing * "odd/even" order */ NUM_NON_BLUE_FLAME_BFREGS = 1, }; static int max_bfregs(struct mlx5_ib_dev *dev, struct mlx5_bfreg_info *bfregi) { return get_num_static_uars(dev, bfregi) * MLX5_NON_FP_BFREGS_PER_UAR; } static int num_med_bfreg(struct mlx5_ib_dev *dev, struct mlx5_bfreg_info *bfregi) { int n; n = max_bfregs(dev, bfregi) - bfregi->num_low_latency_bfregs - NUM_NON_BLUE_FLAME_BFREGS; return n >= 0 ? n : 0; } static int first_med_bfreg(struct mlx5_ib_dev *dev, struct mlx5_bfreg_info *bfregi) { return num_med_bfreg(dev, bfregi) ? 1 : -ENOMEM; } static int first_hi_bfreg(struct mlx5_ib_dev *dev, struct mlx5_bfreg_info *bfregi) { int med; med = num_med_bfreg(dev, bfregi); return ++med; } static int alloc_high_class_bfreg(struct mlx5_ib_dev *dev, struct mlx5_bfreg_info *bfregi) { int i; for (i = first_hi_bfreg(dev, bfregi); i < max_bfregs(dev, bfregi); i++) { if (!bfregi->count[i]) { bfregi->count[i]++; return i; } } return -ENOMEM; } static int alloc_med_class_bfreg(struct mlx5_ib_dev *dev, struct mlx5_bfreg_info *bfregi) { int minidx = first_med_bfreg(dev, bfregi); int i; if (minidx < 0) return minidx; for (i = minidx; i < first_hi_bfreg(dev, bfregi); i++) { if (bfregi->count[i] < bfregi->count[minidx]) minidx = i; if (!bfregi->count[minidx]) break; } bfregi->count[minidx]++; return minidx; } static int alloc_bfreg(struct mlx5_ib_dev *dev, struct mlx5_bfreg_info *bfregi) { int bfregn = -ENOMEM; if (bfregi->lib_uar_dyn) return -EINVAL; mutex_lock(&bfregi->lock); if (bfregi->ver >= 2) { bfregn = alloc_high_class_bfreg(dev, bfregi); if (bfregn < 0) bfregn = alloc_med_class_bfreg(dev, bfregi); } if (bfregn < 0) { BUILD_BUG_ON(NUM_NON_BLUE_FLAME_BFREGS != 1); bfregn = 0; bfregi->count[bfregn]++; } mutex_unlock(&bfregi->lock); return bfregn; } void mlx5_ib_free_bfreg(struct mlx5_ib_dev *dev, struct mlx5_bfreg_info *bfregi, int bfregn) { mutex_lock(&bfregi->lock); bfregi->count[bfregn]--; mutex_unlock(&bfregi->lock); } static enum mlx5_qp_state to_mlx5_state(enum ib_qp_state state) { switch (state) { case IB_QPS_RESET: return MLX5_QP_STATE_RST; case IB_QPS_INIT: return MLX5_QP_STATE_INIT; case IB_QPS_RTR: return MLX5_QP_STATE_RTR; case IB_QPS_RTS: return MLX5_QP_STATE_RTS; case IB_QPS_SQD: return MLX5_QP_STATE_SQD; case IB_QPS_SQE: return MLX5_QP_STATE_SQER; case IB_QPS_ERR: return MLX5_QP_STATE_ERR; default: return -1; } } static int to_mlx5_st(enum ib_qp_type type) { switch (type) { case IB_QPT_RC: return MLX5_QP_ST_RC; case IB_QPT_UC: return MLX5_QP_ST_UC; case IB_QPT_UD: return MLX5_QP_ST_UD; case MLX5_IB_QPT_REG_UMR: return MLX5_QP_ST_REG_UMR; case IB_QPT_XRC_INI: case IB_QPT_XRC_TGT: return MLX5_QP_ST_XRC; case IB_QPT_SMI: return MLX5_QP_ST_QP0; case MLX5_IB_QPT_HW_GSI: return MLX5_QP_ST_QP1; case IB_QPT_RAW_IPV6: return MLX5_QP_ST_RAW_IPV6; case IB_QPT_RAW_PACKET: case IB_QPT_RAW_ETHERTYPE: return MLX5_QP_ST_RAW_ETHERTYPE; case IB_QPT_MAX: default: return -EINVAL; } } static void mlx5_ib_lock_cqs(struct mlx5_ib_cq *send_cq, struct mlx5_ib_cq *recv_cq); static void mlx5_ib_unlock_cqs(struct mlx5_ib_cq *send_cq, struct mlx5_ib_cq *recv_cq); int bfregn_to_uar_index(struct mlx5_ib_dev *dev, struct mlx5_bfreg_info *bfregi, u32 bfregn, bool dyn_bfreg) { unsigned int bfregs_per_sys_page; u32 index_of_sys_page; u32 offset; if (bfregi->lib_uar_dyn) return -EINVAL; bfregs_per_sys_page = get_uars_per_sys_page(dev, bfregi->lib_uar_4k) * MLX5_NON_FP_BFREGS_PER_UAR; index_of_sys_page = bfregn / bfregs_per_sys_page; if (dyn_bfreg) { index_of_sys_page += bfregi->num_static_sys_pages; if (index_of_sys_page >= bfregi->num_sys_pages) return -EINVAL; if (bfregn > bfregi->num_dyn_bfregs || bfregi->sys_pages[index_of_sys_page] == MLX5_IB_INVALID_UAR_INDEX) { mlx5_ib_dbg(dev, "Invalid dynamic uar index\n"); return -EINVAL; } } offset = bfregn % bfregs_per_sys_page / MLX5_NON_FP_BFREGS_PER_UAR; return bfregi->sys_pages[index_of_sys_page] + offset; } static int mlx5_ib_umem_get(struct mlx5_ib_dev *dev, struct ib_pd *pd, unsigned long addr, size_t size, struct ib_umem **umem, int *npages, int *page_shift, int *ncont, u32 *offset) { int err; *umem = ib_umem_get(pd->uobject->context, addr, size, 0, 0); if (IS_ERR(*umem)) { mlx5_ib_dbg(dev, "umem_get failed\n"); return PTR_ERR(*umem); } mlx5_ib_cont_pages(*umem, addr, 0, npages, page_shift, ncont, NULL); err = mlx5_ib_get_buf_offset(addr, *page_shift, offset); if (err) { mlx5_ib_warn(dev, "bad offset\n"); goto err_umem; } mlx5_ib_dbg(dev, "addr 0x%lx, size %zu, npages %d, page_shift %d, ncont %d, offset %d\n", addr, size, *npages, *page_shift, *ncont, *offset); return 0; err_umem: ib_umem_release(*umem); *umem = NULL; return err; } static void destroy_user_rq(struct ib_pd *pd, struct mlx5_ib_rwq *rwq) { struct mlx5_ib_ucontext *context; context = to_mucontext(pd->uobject->context); mlx5_ib_db_unmap_user(context, &rwq->db); if (rwq->umem) ib_umem_release(rwq->umem); } static int create_user_rq(struct mlx5_ib_dev *dev, struct ib_pd *pd, struct mlx5_ib_rwq *rwq, struct mlx5_ib_create_wq *ucmd) { struct mlx5_ib_ucontext *context; int page_shift = 0; int npages; u32 offset = 0; int ncont = 0; int err; if (!ucmd->buf_addr) return -EINVAL; context = to_mucontext(pd->uobject->context); rwq->umem = ib_umem_get(pd->uobject->context, ucmd->buf_addr, rwq->buf_size, 0, 0); if (IS_ERR(rwq->umem)) { mlx5_ib_dbg(dev, "umem_get failed\n"); err = PTR_ERR(rwq->umem); return err; } mlx5_ib_cont_pages(rwq->umem, ucmd->buf_addr, 0, &npages, &page_shift, &ncont, NULL); err = mlx5_ib_get_buf_offset(ucmd->buf_addr, page_shift, &rwq->rq_page_offset); if (err) { mlx5_ib_warn(dev, "bad offset\n"); goto err_umem; } rwq->rq_num_pas = ncont; rwq->page_shift = page_shift; rwq->log_page_size = page_shift - MLX5_ADAPTER_PAGE_SHIFT; rwq->wq_sig = !!(ucmd->flags & MLX5_WQ_FLAG_SIGNATURE); mlx5_ib_dbg(dev, "addr 0x%llx, size %zd, npages %d, page_shift %d, ncont %d, offset %d\n", (unsigned long long)ucmd->buf_addr, rwq->buf_size, npages, page_shift, ncont, offset); err = mlx5_ib_db_map_user(context, ucmd->db_addr, &rwq->db); if (err) { mlx5_ib_dbg(dev, "map failed\n"); goto err_umem; } rwq->create_type = MLX5_WQ_USER; return 0; err_umem: ib_umem_release(rwq->umem); return err; } static int adjust_bfregn(struct mlx5_ib_dev *dev, struct mlx5_bfreg_info *bfregi, int bfregn) { return bfregn / MLX5_NON_FP_BFREGS_PER_UAR * MLX5_BFREGS_PER_UAR + bfregn % MLX5_NON_FP_BFREGS_PER_UAR; } static int create_user_qp(struct mlx5_ib_dev *dev, struct ib_pd *pd, struct mlx5_ib_qp *qp, struct ib_udata *udata, struct ib_qp_init_attr *attr, u32 **in, struct mlx5_ib_create_qp_resp *resp, int *inlen, struct mlx5_ib_qp_base *base) { struct mlx5_ib_ucontext *context; struct mlx5_ib_create_qp ucmd; struct mlx5_ib_ubuffer *ubuffer = &base->ubuffer; int page_shift = 0; int uar_index = 0; int npages; u32 offset = 0; int bfregn; int ncont = 0; __be64 *pas; void *qpc; int err; u32 uar_flags; err = ib_copy_from_udata(&ucmd, udata, sizeof(ucmd)); if (err) { mlx5_ib_dbg(dev, "copy failed\n"); return err; } context = to_mucontext(pd->uobject->context); uar_flags = ucmd.flags & (MLX5_QP_FLAG_UAR_PAGE_INDEX | MLX5_QP_FLAG_BFREG_INDEX); switch (uar_flags) { case MLX5_QP_FLAG_UAR_PAGE_INDEX: uar_index = ucmd.bfreg_index; bfregn = MLX5_IB_INVALID_BFREG; break; case MLX5_QP_FLAG_BFREG_INDEX: uar_index = bfregn_to_uar_index(dev, &context->bfregi, ucmd.bfreg_index, true); if (uar_index < 0) return uar_index; bfregn = MLX5_IB_INVALID_BFREG; break; case 0: if (qp->flags & MLX5_IB_QP_CROSS_CHANNEL) return -EINVAL; bfregn = alloc_bfreg(dev, &context->bfregi); if (bfregn < 0) return bfregn; break; default: return -EINVAL; } mlx5_ib_dbg(dev, "bfregn 0x%x, uar_index 0x%x\n", bfregn, uar_index); if (bfregn != MLX5_IB_INVALID_BFREG) uar_index = bfregn_to_uar_index(dev, &context->bfregi, bfregn, false); qp->rq.offset = 0; qp->sq.wqe_shift = ilog2(MLX5_SEND_WQE_BB); qp->sq.offset = qp->rq.wqe_cnt << qp->rq.wqe_shift; err = set_user_buf_size(dev, qp, &ucmd, base, attr); if (err) goto err_bfreg; if (ucmd.buf_addr && ubuffer->buf_size) { ubuffer->buf_addr = ucmd.buf_addr; err = mlx5_ib_umem_get(dev, pd, ubuffer->buf_addr, ubuffer->buf_size, &ubuffer->umem, &npages, &page_shift, &ncont, &offset); if (err) goto err_bfreg; } else { ubuffer->umem = NULL; } *inlen = MLX5_ST_SZ_BYTES(create_qp_in) + MLX5_FLD_SZ_BYTES(create_qp_in, pas[0]) * ncont; *in = mlx5_vzalloc(*inlen); if (!*in) { err = -ENOMEM; goto err_umem; } pas = (__be64 *)MLX5_ADDR_OF(create_qp_in, *in, pas); if (ubuffer->umem) mlx5_ib_populate_pas(dev, ubuffer->umem, page_shift, pas, 0); qpc = MLX5_ADDR_OF(create_qp_in, *in, qpc); MLX5_SET(qpc, qpc, log_page_size, page_shift - MLX5_ADAPTER_PAGE_SHIFT); MLX5_SET(qpc, qpc, page_offset, offset); MLX5_SET(qpc, qpc, uar_page, uar_index); if (bfregn != MLX5_IB_INVALID_BFREG) resp->bfreg_index = adjust_bfregn(dev, &context->bfregi, bfregn); else resp->bfreg_index = MLX5_IB_INVALID_BFREG; qp->bfregn = bfregn; err = mlx5_ib_db_map_user(context, ucmd.db_addr, &qp->db); if (err) { mlx5_ib_dbg(dev, "map failed\n"); goto err_free; } err = ib_copy_to_udata(udata, resp, sizeof(*resp)); if (err) { mlx5_ib_dbg(dev, "copy failed\n"); goto err_unmap; } qp->create_type = MLX5_QP_USER; return 0; err_unmap: mlx5_ib_db_unmap_user(context, &qp->db); err_free: kvfree(*in); err_umem: if (ubuffer->umem) ib_umem_release(ubuffer->umem); err_bfreg: if (bfregn != MLX5_IB_INVALID_BFREG) mlx5_ib_free_bfreg(dev, &context->bfregi, bfregn); return err; } static void destroy_qp_user(struct mlx5_ib_dev *dev, struct ib_pd *pd, struct mlx5_ib_qp *qp, struct mlx5_ib_qp_base *base) { struct mlx5_ib_ucontext *context; context = to_mucontext(pd->uobject->context); mlx5_ib_db_unmap_user(context, &qp->db); if (base->ubuffer.umem) ib_umem_release(base->ubuffer.umem); /* * Free only the BFREGs which are handled by the kernel. * BFREGs of UARs allocated dynamically are handled by user. */ if (qp->bfregn != MLX5_IB_INVALID_BFREG) mlx5_ib_free_bfreg(dev, &context->bfregi, qp->bfregn); } static int create_kernel_qp(struct mlx5_ib_dev *dev, struct ib_qp_init_attr *init_attr, struct mlx5_ib_qp *qp, u32 **in, int *inlen, struct mlx5_ib_qp_base *base) { int uar_index; void *qpc; int err; if (init_attr->create_flags & ~(IB_QP_CREATE_SIGNATURE_EN | IB_QP_CREATE_BLOCK_MULTICAST_LOOPBACK | IB_QP_CREATE_IPOIB_UD_LSO | MLX5_IB_QP_CREATE_SQPN_QP1 | MLX5_IB_QP_CREATE_WC_TEST)) return -EINVAL; spin_lock_init(&qp->bf.lock32); if (init_attr->qp_type == MLX5_IB_QPT_REG_UMR) qp->bf.bfreg = &dev->fp_bfreg; else if (init_attr->create_flags & MLX5_IB_QP_CREATE_WC_TEST) qp->bf.bfreg = &dev->wc_bfreg; else qp->bf.bfreg = &dev->bfreg; /* We need to divide by two since each register is comprised of * two buffers of identical size, namely odd and even */ qp->bf.buf_size = (1 << MLX5_CAP_GEN(dev->mdev, log_bf_reg_size)) / 2; uar_index = qp->bf.bfreg->index; err = calc_sq_size(dev, init_attr, qp); if (err < 0) { mlx5_ib_dbg(dev, "err %d\n", err); return err; } qp->rq.offset = 0; qp->sq.offset = qp->rq.wqe_cnt << qp->rq.wqe_shift; base->ubuffer.buf_size = err + (qp->rq.wqe_cnt << qp->rq.wqe_shift); err = mlx5_buf_alloc(dev->mdev, base->ubuffer.buf_size, 2 * PAGE_SIZE, &qp->buf); if (err) { mlx5_ib_dbg(dev, "err %d\n", err); return err; } qp->sq.qend = mlx5_get_send_wqe(qp, qp->sq.wqe_cnt); *inlen = MLX5_ST_SZ_BYTES(create_qp_in) + MLX5_FLD_SZ_BYTES(create_qp_in, pas[0]) * qp->buf.npages; *in = mlx5_vzalloc(*inlen); if (!*in) { err = -ENOMEM; goto err_buf; } qpc = MLX5_ADDR_OF(create_qp_in, *in, qpc); MLX5_SET(qpc, qpc, uar_page, uar_index); MLX5_SET(qpc, qpc, ts_format, mlx5_get_qp_default_ts(dev->mdev)); MLX5_SET(qpc, qpc, log_page_size, qp->buf.page_shift - MLX5_ADAPTER_PAGE_SHIFT); /* Set "fast registration enabled" for all kernel QPs */ MLX5_SET(qpc, qpc, fre, 1); MLX5_SET(qpc, qpc, rlky, 1); if (init_attr->create_flags & MLX5_IB_QP_CREATE_SQPN_QP1) { MLX5_SET(qpc, qpc, deth_sqpn, 1); qp->flags |= MLX5_IB_QP_SQPN_QP1; } mlx5_fill_page_array(&qp->buf, (__be64 *)MLX5_ADDR_OF(create_qp_in, *in, pas)); err = mlx5_db_alloc(dev->mdev, &qp->db); if (err) { mlx5_ib_dbg(dev, "err %d\n", err); goto err_free; } qp->sq.wrid = kmalloc(qp->sq.wqe_cnt * sizeof(*qp->sq.wrid), GFP_KERNEL); qp->sq.wr_data = kmalloc(qp->sq.wqe_cnt * sizeof(*qp->sq.wr_data), GFP_KERNEL); qp->rq.wrid = kmalloc(qp->rq.wqe_cnt * sizeof(*qp->rq.wrid), GFP_KERNEL); qp->sq.w_list = kmalloc(qp->sq.wqe_cnt * sizeof(*qp->sq.w_list), GFP_KERNEL); qp->sq.wqe_head = kmalloc(qp->sq.wqe_cnt * sizeof(*qp->sq.wqe_head), GFP_KERNEL); if (!qp->sq.wrid || !qp->sq.wr_data || !qp->rq.wrid || !qp->sq.w_list || !qp->sq.wqe_head) { err = -ENOMEM; goto err_wrid; } qp->create_type = MLX5_QP_KERNEL; return 0; err_wrid: kfree(qp->sq.wqe_head); kfree(qp->sq.w_list); kfree(qp->sq.wrid); kfree(qp->sq.wr_data); kfree(qp->rq.wrid); mlx5_db_free(dev->mdev, &qp->db); err_free: kvfree(*in); err_buf: mlx5_buf_free(dev->mdev, &qp->buf); return err; } static void destroy_qp_kernel(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp) { kfree(qp->sq.wqe_head); kfree(qp->sq.w_list); kfree(qp->sq.wrid); kfree(qp->sq.wr_data); kfree(qp->rq.wrid); mlx5_db_free(dev->mdev, &qp->db); mlx5_buf_free(dev->mdev, &qp->buf); } static u32 get_rx_type(struct mlx5_ib_qp *qp, struct ib_qp_init_attr *attr) { if (attr->srq || (attr->qp_type == IB_QPT_XRC_TGT) || (attr->qp_type == IB_QPT_XRC_INI)) return MLX5_SRQ_RQ; else if (!qp->has_rq) return MLX5_ZERO_LEN_RQ; else return MLX5_NON_ZERO_RQ; } static int is_connected(enum ib_qp_type qp_type) { if (qp_type == IB_QPT_RC || qp_type == IB_QPT_UC) return 1; return 0; } static int create_raw_packet_qp_tis(struct mlx5_ib_dev *dev, struct mlx5_ib_sq *sq, u32 tdn) { u32 in[MLX5_ST_SZ_DW(create_tis_in)] = {0}; void *tisc = MLX5_ADDR_OF(create_tis_in, in, ctx); MLX5_SET(tisc, tisc, transport_domain, tdn); return mlx5_core_create_tis(dev->mdev, in, sizeof(in), &sq->tisn); } static void destroy_raw_packet_qp_tis(struct mlx5_ib_dev *dev, struct mlx5_ib_sq *sq) { mlx5_core_destroy_tis(dev->mdev, sq->tisn); } static int create_raw_packet_qp_sq(struct mlx5_ib_dev *dev, struct mlx5_ib_sq *sq, void *qpin, struct ib_pd *pd) { struct mlx5_ib_ubuffer *ubuffer = &sq->ubuffer; __be64 *pas; void *in; void *sqc; void *qpc = MLX5_ADDR_OF(create_qp_in, qpin, qpc); void *wq; int inlen; int err; int page_shift = 0; int npages; int ncont = 0; u32 offset = 0; u8 ts_format; ts_format = mlx5_get_sq_default_ts(dev->mdev); err = mlx5_ib_umem_get(dev, pd, ubuffer->buf_addr, ubuffer->buf_size, &sq->ubuffer.umem, &npages, &page_shift, &ncont, &offset); if (err) return err; inlen = MLX5_ST_SZ_BYTES(create_sq_in) + sizeof(u64) * ncont; in = mlx5_vzalloc(inlen); if (!in) { err = -ENOMEM; goto err_umem; } sqc = MLX5_ADDR_OF(create_sq_in, in, ctx); MLX5_SET(sqc, sqc, flush_in_error_en, 1); MLX5_SET(sqc, sqc, state, MLX5_SQC_STATE_RST); MLX5_SET(sqc, sqc, ts_format, ts_format); MLX5_SET(sqc, sqc, user_index, MLX5_GET(qpc, qpc, user_index)); MLX5_SET(sqc, sqc, cqn, MLX5_GET(qpc, qpc, cqn_snd)); MLX5_SET(sqc, sqc, tis_lst_sz, 1); MLX5_SET(sqc, sqc, tis_num_0, sq->tisn); wq = MLX5_ADDR_OF(sqc, sqc, wq); MLX5_SET(wq, wq, wq_type, MLX5_WQ_TYPE_CYCLIC); MLX5_SET(wq, wq, pd, MLX5_GET(qpc, qpc, pd)); MLX5_SET(wq, wq, uar_page, MLX5_GET(qpc, qpc, uar_page)); MLX5_SET64(wq, wq, dbr_addr, MLX5_GET64(qpc, qpc, dbr_addr)); MLX5_SET(wq, wq, log_wq_stride, ilog2(MLX5_SEND_WQE_BB)); MLX5_SET(wq, wq, log_wq_sz, MLX5_GET(qpc, qpc, log_sq_size)); MLX5_SET(wq, wq, log_wq_pg_sz, page_shift - MLX5_ADAPTER_PAGE_SHIFT); MLX5_SET(wq, wq, page_offset, offset); pas = (__be64 *)MLX5_ADDR_OF(wq, wq, pas); mlx5_ib_populate_pas(dev, sq->ubuffer.umem, page_shift, pas, 0); err = mlx5_core_create_sq_tracked(dev->mdev, in, inlen, &sq->base.mqp); kvfree(in); if (err) goto err_umem; return 0; err_umem: ib_umem_release(sq->ubuffer.umem); sq->ubuffer.umem = NULL; return err; } static void destroy_raw_packet_qp_sq(struct mlx5_ib_dev *dev, struct mlx5_ib_sq *sq) { mlx5_core_destroy_sq_tracked(dev->mdev, &sq->base.mqp); ib_umem_release(sq->ubuffer.umem); } static int get_rq_pas_size(void *qpc) { u32 log_page_size = MLX5_GET(qpc, qpc, log_page_size) + 12; u32 log_rq_stride = MLX5_GET(qpc, qpc, log_rq_stride); u32 log_rq_size = MLX5_GET(qpc, qpc, log_rq_size); u32 page_offset = MLX5_GET(qpc, qpc, page_offset); u32 po_quanta = 1 << (log_page_size - 6); u32 rq_sz = 1 << (log_rq_size + 4 + log_rq_stride); u32 page_size = 1 << log_page_size; u32 rq_sz_po = rq_sz + (page_offset * po_quanta); u32 rq_num_pas = (rq_sz_po + page_size - 1) / page_size; return rq_num_pas * sizeof(u64); } static int create_raw_packet_qp_rq(struct mlx5_ib_dev *dev, struct mlx5_ib_rq *rq, void *qpin) { struct mlx5_ib_qp *mqp = rq->base.container_mibqp; __be64 *pas; __be64 *qp_pas; void *in; void *rqc; void *wq; void *qpc = MLX5_ADDR_OF(create_qp_in, qpin, qpc); int inlen; int err; u32 rq_pas_size = get_rq_pas_size(qpc); u8 ts_format; ts_format = mlx5_get_rq_default_ts(dev->mdev); inlen = MLX5_ST_SZ_BYTES(create_rq_in) + rq_pas_size; in = mlx5_vzalloc(inlen); if (!in) return -ENOMEM; rqc = MLX5_ADDR_OF(create_rq_in, in, ctx); MLX5_SET(rqc, rqc, vlan_strip_disable, 1); MLX5_SET(rqc, rqc, mem_rq_type, MLX5_RQC_RQ_TYPE_MEMORY_RQ_INLINE); MLX5_SET(rqc, rqc, state, MLX5_RQC_STATE_RST); MLX5_SET(rqc, rqc, ts_format, ts_format); MLX5_SET(rqc, rqc, flush_in_error_en, 1); MLX5_SET(rqc, rqc, user_index, MLX5_GET(qpc, qpc, user_index)); MLX5_SET(rqc, rqc, cqn, MLX5_GET(qpc, qpc, cqn_rcv)); if (mqp->flags & MLX5_IB_QP_CAP_SCATTER_FCS) MLX5_SET(rqc, rqc, scatter_fcs, 1); wq = MLX5_ADDR_OF(rqc, rqc, wq); MLX5_SET(wq, wq, wq_type, MLX5_WQ_TYPE_CYCLIC); MLX5_SET(wq, wq, end_padding_mode, MLX5_GET(qpc, qpc, end_padding_mode)); MLX5_SET(wq, wq, page_offset, MLX5_GET(qpc, qpc, page_offset)); MLX5_SET(wq, wq, pd, MLX5_GET(qpc, qpc, pd)); MLX5_SET64(wq, wq, dbr_addr, MLX5_GET64(qpc, qpc, dbr_addr)); MLX5_SET(wq, wq, log_wq_stride, MLX5_GET(qpc, qpc, log_rq_stride) + 4); MLX5_SET(wq, wq, log_wq_pg_sz, MLX5_GET(qpc, qpc, log_page_size)); MLX5_SET(wq, wq, log_wq_sz, MLX5_GET(qpc, qpc, log_rq_size)); pas = (__be64 *)MLX5_ADDR_OF(wq, wq, pas); qp_pas = (__be64 *)MLX5_ADDR_OF(create_qp_in, qpin, pas); memcpy(pas, qp_pas, rq_pas_size); err = mlx5_core_create_rq_tracked(dev->mdev, in, inlen, &rq->base.mqp); kvfree(in); return err; } static void destroy_raw_packet_qp_rq(struct mlx5_ib_dev *dev, struct mlx5_ib_rq *rq) { mlx5_core_destroy_rq_tracked(dev->mdev, &rq->base.mqp); } static int create_raw_packet_qp_tir(struct mlx5_ib_dev *dev, struct mlx5_ib_rq *rq, u32 tdn) { u32 *in; void *tirc; int inlen; int err; inlen = MLX5_ST_SZ_BYTES(create_tir_in); in = mlx5_vzalloc(inlen); if (!in) return -ENOMEM; tirc = MLX5_ADDR_OF(create_tir_in, in, tir_context); MLX5_SET(tirc, tirc, disp_type, MLX5_TIRC_DISP_TYPE_DIRECT); MLX5_SET(tirc, tirc, inline_rqn, rq->base.mqp.qpn); MLX5_SET(tirc, tirc, transport_domain, tdn); err = mlx5_core_create_tir(dev->mdev, in, inlen, &rq->tirn); kvfree(in); return err; } static void destroy_raw_packet_qp_tir(struct mlx5_ib_dev *dev, struct mlx5_ib_rq *rq) { mlx5_core_destroy_tir(dev->mdev, rq->tirn); } static int create_raw_packet_qp(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, u32 *in, struct ib_pd *pd) { struct mlx5_ib_raw_packet_qp *raw_packet_qp = &qp->raw_packet_qp; struct mlx5_ib_sq *sq = &raw_packet_qp->sq; struct mlx5_ib_rq *rq = &raw_packet_qp->rq; struct ib_uobject *uobj = pd->uobject; struct ib_ucontext *ucontext = uobj->context; struct mlx5_ib_ucontext *mucontext = to_mucontext(ucontext); int err; u32 tdn = mucontext->tdn; if (qp->sq.wqe_cnt) { err = create_raw_packet_qp_tis(dev, sq, tdn); if (err) return err; err = create_raw_packet_qp_sq(dev, sq, in, pd); if (err) goto err_destroy_tis; sq->base.container_mibqp = qp; } if (qp->rq.wqe_cnt) { rq->base.container_mibqp = qp; err = create_raw_packet_qp_rq(dev, rq, in); if (err) goto err_destroy_sq; err = create_raw_packet_qp_tir(dev, rq, tdn); if (err) goto err_destroy_rq; } qp->trans_qp.base.mqp.qpn = qp->sq.wqe_cnt ? sq->base.mqp.qpn : rq->base.mqp.qpn; return 0; err_destroy_rq: destroy_raw_packet_qp_rq(dev, rq); err_destroy_sq: if (!qp->sq.wqe_cnt) return err; destroy_raw_packet_qp_sq(dev, sq); err_destroy_tis: destroy_raw_packet_qp_tis(dev, sq); return err; } static void destroy_raw_packet_qp(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp) { struct mlx5_ib_raw_packet_qp *raw_packet_qp = &qp->raw_packet_qp; struct mlx5_ib_sq *sq = &raw_packet_qp->sq; struct mlx5_ib_rq *rq = &raw_packet_qp->rq; if (qp->rq.wqe_cnt) { destroy_raw_packet_qp_tir(dev, rq); destroy_raw_packet_qp_rq(dev, rq); } if (qp->sq.wqe_cnt) { destroy_raw_packet_qp_sq(dev, sq); destroy_raw_packet_qp_tis(dev, sq); } } static void raw_packet_qp_copy_info(struct mlx5_ib_qp *qp, struct mlx5_ib_raw_packet_qp *raw_packet_qp) { struct mlx5_ib_sq *sq = &raw_packet_qp->sq; struct mlx5_ib_rq *rq = &raw_packet_qp->rq; sq->sq = &qp->sq; rq->rq = &qp->rq; sq->doorbell = &qp->db; rq->doorbell = &qp->db; } static void destroy_rss_raw_qp_tir(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp) { mlx5_core_destroy_tir(dev->mdev, qp->rss_qp.tirn); } static int create_rss_raw_qp_tir(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, struct ib_pd *pd, struct ib_qp_init_attr *init_attr, struct ib_udata *udata) { struct ib_uobject *uobj = pd->uobject; struct ib_ucontext *ucontext = uobj->context; struct mlx5_ib_ucontext *mucontext = to_mucontext(ucontext); struct mlx5_ib_create_qp_resp resp = {}; int inlen; int err; u32 *in; void *tirc; void *hfso; u32 selected_fields = 0; size_t min_resp_len; u32 tdn = mucontext->tdn; struct mlx5_ib_create_qp_rss ucmd = {}; size_t required_cmd_sz; if (init_attr->qp_type != IB_QPT_RAW_PACKET) return -EOPNOTSUPP; if (init_attr->create_flags || init_attr->send_cq) return -EINVAL; min_resp_len = offsetof(typeof(resp), bfreg_index) + sizeof(resp.bfreg_index); if (udata->outlen < min_resp_len) return -EINVAL; required_cmd_sz = offsetof(typeof(ucmd), reserved1) + sizeof(ucmd.reserved1); if (udata->inlen < required_cmd_sz) { mlx5_ib_dbg(dev, "invalid inlen\n"); return -EINVAL; } if (udata->inlen > sizeof(ucmd) && !ib_is_udata_cleared(udata, sizeof(ucmd), udata->inlen - sizeof(ucmd))) { mlx5_ib_dbg(dev, "inlen is not supported\n"); return -EOPNOTSUPP; } if (ib_copy_from_udata(&ucmd, udata, min(sizeof(ucmd), udata->inlen))) { mlx5_ib_dbg(dev, "copy failed\n"); return -EFAULT; } if (ucmd.comp_mask) { mlx5_ib_dbg(dev, "invalid comp mask\n"); return -EOPNOTSUPP; } if (memchr_inv(ucmd.reserved, 0, sizeof(ucmd.reserved)) || ucmd.reserved1) { mlx5_ib_dbg(dev, "invalid reserved\n"); return -EOPNOTSUPP; } err = ib_copy_to_udata(udata, &resp, min_resp_len); if (err) { mlx5_ib_dbg(dev, "copy failed\n"); return -EINVAL; } inlen = MLX5_ST_SZ_BYTES(create_tir_in); in = mlx5_vzalloc(inlen); if (!in) return -ENOMEM; tirc = MLX5_ADDR_OF(create_tir_in, in, tir_context); MLX5_SET(tirc, tirc, disp_type, MLX5_TIRC_DISP_TYPE_INDIRECT); MLX5_SET(tirc, tirc, indirect_table, init_attr->rwq_ind_tbl->ind_tbl_num); MLX5_SET(tirc, tirc, transport_domain, tdn); hfso = MLX5_ADDR_OF(tirc, tirc, rx_hash_field_selector_outer); switch (ucmd.rx_hash_function) { case MLX5_RX_HASH_FUNC_TOEPLITZ: { void *rss_key = MLX5_ADDR_OF(tirc, tirc, rx_hash_toeplitz_key); size_t len = MLX5_FLD_SZ_BYTES(tirc, rx_hash_toeplitz_key); if (len != ucmd.rx_key_len) { err = -EINVAL; goto err; } MLX5_SET(tirc, tirc, rx_hash_fn, MLX5_RX_HASH_FUNC_TOEPLITZ); MLX5_SET(tirc, tirc, rx_hash_symmetric, 1); memcpy(rss_key, ucmd.rx_hash_key, len); break; } default: err = -EOPNOTSUPP; goto err; } if (!ucmd.rx_hash_fields_mask) { /* special case when this TIR serves as steering entry without hashing */ if (!init_attr->rwq_ind_tbl->log_ind_tbl_size) goto create_tir; err = -EINVAL; goto err; } if (((ucmd.rx_hash_fields_mask & MLX5_RX_HASH_SRC_IPV4) || (ucmd.rx_hash_fields_mask & MLX5_RX_HASH_DST_IPV4)) && ((ucmd.rx_hash_fields_mask & MLX5_RX_HASH_SRC_IPV6) || (ucmd.rx_hash_fields_mask & MLX5_RX_HASH_DST_IPV6))) { err = -EINVAL; goto err; } /* If none of IPV4 & IPV6 SRC/DST was set - this bit field is ignored */ if ((ucmd.rx_hash_fields_mask & MLX5_RX_HASH_SRC_IPV4) || (ucmd.rx_hash_fields_mask & MLX5_RX_HASH_DST_IPV4)) MLX5_SET(rx_hash_field_select, hfso, l3_prot_type, MLX5_L3_PROT_TYPE_IPV4); else if ((ucmd.rx_hash_fields_mask & MLX5_RX_HASH_SRC_IPV6) || (ucmd.rx_hash_fields_mask & MLX5_RX_HASH_DST_IPV6)) MLX5_SET(rx_hash_field_select, hfso, l3_prot_type, MLX5_L3_PROT_TYPE_IPV6); if (((ucmd.rx_hash_fields_mask & MLX5_RX_HASH_SRC_PORT_TCP) || (ucmd.rx_hash_fields_mask & MLX5_RX_HASH_DST_PORT_TCP)) && ((ucmd.rx_hash_fields_mask & MLX5_RX_HASH_SRC_PORT_UDP) || (ucmd.rx_hash_fields_mask & MLX5_RX_HASH_DST_PORT_UDP))) { err = -EINVAL; goto err; } /* If none of TCP & UDP SRC/DST was set - this bit field is ignored */ if ((ucmd.rx_hash_fields_mask & MLX5_RX_HASH_SRC_PORT_TCP) || (ucmd.rx_hash_fields_mask & MLX5_RX_HASH_DST_PORT_TCP)) MLX5_SET(rx_hash_field_select, hfso, l4_prot_type, MLX5_L4_PROT_TYPE_TCP); else if ((ucmd.rx_hash_fields_mask & MLX5_RX_HASH_SRC_PORT_UDP) || (ucmd.rx_hash_fields_mask & MLX5_RX_HASH_DST_PORT_UDP)) MLX5_SET(rx_hash_field_select, hfso, l4_prot_type, MLX5_L4_PROT_TYPE_UDP); if ((ucmd.rx_hash_fields_mask & MLX5_RX_HASH_SRC_IPV4) || (ucmd.rx_hash_fields_mask & MLX5_RX_HASH_SRC_IPV6)) selected_fields |= MLX5_HASH_FIELD_SEL_SRC_IP; if ((ucmd.rx_hash_fields_mask & MLX5_RX_HASH_DST_IPV4) || (ucmd.rx_hash_fields_mask & MLX5_RX_HASH_DST_IPV6)) selected_fields |= MLX5_HASH_FIELD_SEL_DST_IP; if ((ucmd.rx_hash_fields_mask & MLX5_RX_HASH_SRC_PORT_TCP) || (ucmd.rx_hash_fields_mask & MLX5_RX_HASH_SRC_PORT_UDP)) selected_fields |= MLX5_HASH_FIELD_SEL_L4_SPORT; if ((ucmd.rx_hash_fields_mask & MLX5_RX_HASH_DST_PORT_TCP) || (ucmd.rx_hash_fields_mask & MLX5_RX_HASH_DST_PORT_UDP)) selected_fields |= MLX5_HASH_FIELD_SEL_L4_DPORT; MLX5_SET(rx_hash_field_select, hfso, selected_fields, selected_fields); create_tir: err = mlx5_core_create_tir(dev->mdev, in, inlen, &qp->rss_qp.tirn); if (err) goto err; kvfree(in); /* qpn is reserved for that QP */ qp->trans_qp.base.mqp.qpn = 0; qp->flags |= MLX5_IB_QP_RSS; return 0; err: kvfree(in); return err; } static int atomic_size_to_mode(int size_mask) { /* driver does not support atomic_size > 256B * and does not know how to translate bigger sizes */ int supported_size_mask = size_mask & 0x1ff; int log_max_size; if (!supported_size_mask) return -EOPNOTSUPP; log_max_size = __fls(supported_size_mask); if (log_max_size > 3) return log_max_size; return MLX5_ATOMIC_MODE_8B; } static int get_atomic_mode(struct mlx5_ib_dev *dev, enum ib_qp_type qp_type) { u8 atomic_operations = MLX5_CAP_ATOMIC(dev->mdev, atomic_operations); u8 atomic = MLX5_CAP_GEN(dev->mdev, atomic); int atomic_mode = -EOPNOTSUPP; int atomic_size_mask; if (!atomic) return -EOPNOTSUPP; if (qp_type == MLX5_IB_QPT_DCT) atomic_size_mask = MLX5_CAP_ATOMIC(dev->mdev, atomic_size_dc); else atomic_size_mask = MLX5_CAP_ATOMIC(dev->mdev, atomic_size_qp); if ((atomic_operations & MLX5_ATOMIC_OPS_MASKED_CMP_SWAP) || (atomic_operations & MLX5_ATOMIC_OPS_MASKED_FETCH_ADD)) atomic_mode = atomic_size_to_mode(atomic_size_mask); if (atomic_mode <= 0 && (atomic_operations & MLX5_ATOMIC_OPS_CMP_SWAP && atomic_operations & MLX5_ATOMIC_OPS_FETCH_ADD)) atomic_mode = MLX5_ATOMIC_MODE_IB_COMP; return atomic_mode; } static int create_qp_common(struct mlx5_ib_dev *dev, struct ib_pd *pd, struct ib_qp_init_attr *init_attr, struct ib_udata *udata, struct mlx5_ib_qp *qp) { struct mlx5_ib_resources *devr = &dev->devr; int inlen = MLX5_ST_SZ_BYTES(create_qp_in); struct mlx5_core_dev *mdev = dev->mdev; struct mlx5_ib_create_qp_resp resp; struct mlx5_ib_cq *send_cq; struct mlx5_ib_cq *recv_cq; unsigned long flags; u32 uidx = MLX5_IB_DEFAULT_UIDX; struct mlx5_ib_create_qp ucmd; struct mlx5_ib_qp_base *base; void *qpc; u32 *in; int err; base = init_attr->qp_type == IB_QPT_RAW_PACKET ? &qp->raw_packet_qp.rq.base : &qp->trans_qp.base; if (init_attr->qp_type != IB_QPT_RAW_PACKET) mlx5_ib_odp_create_qp(qp); mutex_init(&qp->mutex); spin_lock_init(&qp->sq.lock); spin_lock_init(&qp->rq.lock); if (init_attr->rwq_ind_tbl) { if (!udata) return -ENOSYS; err = create_rss_raw_qp_tir(dev, qp, pd, init_attr, udata); return err; } if (init_attr->create_flags & IB_QP_CREATE_BLOCK_MULTICAST_LOOPBACK) { if (!MLX5_CAP_GEN(mdev, block_lb_mc)) { mlx5_ib_dbg(dev, "block multicast loopback isn't supported\n"); return -EINVAL; } else { qp->flags |= MLX5_IB_QP_BLOCK_MULTICAST_LOOPBACK; } } if (init_attr->create_flags & (IB_QP_CREATE_CROSS_CHANNEL | IB_QP_CREATE_MANAGED_SEND | IB_QP_CREATE_MANAGED_RECV)) { if (!MLX5_CAP_GEN(mdev, cd)) { mlx5_ib_dbg(dev, "cross-channel isn't supported\n"); return -EINVAL; } if (init_attr->create_flags & IB_QP_CREATE_CROSS_CHANNEL) qp->flags |= MLX5_IB_QP_CROSS_CHANNEL; if (init_attr->create_flags & IB_QP_CREATE_MANAGED_SEND) qp->flags |= MLX5_IB_QP_MANAGED_SEND; if (init_attr->create_flags & IB_QP_CREATE_MANAGED_RECV) qp->flags |= MLX5_IB_QP_MANAGED_RECV; } if (init_attr->qp_type == IB_QPT_UD && (init_attr->create_flags & IB_QP_CREATE_IPOIB_UD_LSO)) if (!MLX5_CAP_GEN(mdev, ipoib_ipoib_offloads)) { mlx5_ib_dbg(dev, "ipoib UD lso qp isn't supported\n"); return -EOPNOTSUPP; } if (init_attr->create_flags & IB_QP_CREATE_SCATTER_FCS) { if (init_attr->qp_type != IB_QPT_RAW_PACKET) { mlx5_ib_dbg(dev, "Scatter FCS is supported only for Raw Packet QPs"); return -EOPNOTSUPP; } if (!MLX5_CAP_GEN(dev->mdev, eth_net_offloads) || !MLX5_CAP_ETH(dev->mdev, scatter_fcs)) { mlx5_ib_dbg(dev, "Scatter FCS isn't supported\n"); return -EOPNOTSUPP; } qp->flags |= MLX5_IB_QP_CAP_SCATTER_FCS; } if (init_attr->sq_sig_type == IB_SIGNAL_ALL_WR) qp->sq_signal_bits = MLX5_WQE_CTRL_CQ_UPDATE; if (pd && pd->uobject) { if (ib_copy_from_udata(&ucmd, udata, sizeof(ucmd))) { mlx5_ib_dbg(dev, "copy failed\n"); return -EFAULT; } err = get_qp_user_index(to_mucontext(pd->uobject->context), &ucmd, udata->inlen, &uidx); if (err) return err; qp->wq_sig = !!(ucmd.flags & MLX5_QP_FLAG_SIGNATURE); qp->scat_cqe = !!(ucmd.flags & MLX5_QP_FLAG_SCATTER_CQE); } else { qp->wq_sig = !!wq_signature; } qp->has_rq = qp_has_rq(init_attr); err = set_rq_size(dev, &init_attr->cap, qp->has_rq, qp, (pd && pd->uobject) ? &ucmd : NULL); if (err) { mlx5_ib_dbg(dev, "err %d\n", err); return err; } if (pd) { if (pd->uobject) { __u32 max_wqes = 1 << MLX5_CAP_GEN(mdev, log_max_qp_sz); mlx5_ib_dbg(dev, "requested sq_wqe_count (%d)\n", ucmd.sq_wqe_count); if (ucmd.rq_wqe_shift != qp->rq.wqe_shift || ucmd.rq_wqe_count != qp->rq.wqe_cnt) { mlx5_ib_dbg(dev, "invalid rq params\n"); return -EINVAL; } if (ucmd.sq_wqe_count > max_wqes) { mlx5_ib_dbg(dev, "requested sq_wqe_count (%d) > max allowed (%d)\n", ucmd.sq_wqe_count, max_wqes); return -EINVAL; } if (init_attr->create_flags & MLX5_IB_QP_CREATE_SQPN_QP1) { mlx5_ib_dbg(dev, "user-space is not allowed to create UD QPs spoofing as QP1\n"); return -EINVAL; } err = create_user_qp(dev, pd, qp, udata, init_attr, &in, &resp, &inlen, base); if (err) mlx5_ib_dbg(dev, "err %d\n", err); } else { err = create_kernel_qp(dev, init_attr, qp, &in, &inlen, base); if (err) mlx5_ib_dbg(dev, "err %d\n", err); } if (err) return err; } else { in = mlx5_vzalloc(inlen); if (!in) return -ENOMEM; qp->create_type = MLX5_QP_EMPTY; } if (is_sqp(init_attr->qp_type)) qp->port = init_attr->port_num; qpc = MLX5_ADDR_OF(create_qp_in, in, qpc); MLX5_SET(qpc, qpc, st, to_mlx5_st(init_attr->qp_type)); MLX5_SET(qpc, qpc, pm_state, MLX5_QP_PM_MIGRATED); if (init_attr->qp_type != MLX5_IB_QPT_REG_UMR) MLX5_SET(qpc, qpc, pd, to_mpd(pd ? pd : devr->p0)->pdn); else MLX5_SET(qpc, qpc, latency_sensitive, 1); if (qp->wq_sig) MLX5_SET(qpc, qpc, wq_signature, 1); if (qp->flags & MLX5_IB_QP_BLOCK_MULTICAST_LOOPBACK) MLX5_SET(qpc, qpc, block_lb_mc, 1); if (qp->flags & MLX5_IB_QP_CROSS_CHANNEL) MLX5_SET(qpc, qpc, cd_master, 1); if (qp->flags & MLX5_IB_QP_MANAGED_SEND) MLX5_SET(qpc, qpc, cd_slave_send, 1); if (qp->flags & MLX5_IB_QP_MANAGED_RECV) MLX5_SET(qpc, qpc, cd_slave_receive, 1); if (qp->scat_cqe && is_connected(init_attr->qp_type)) { int rcqe_sz; int scqe_sz; rcqe_sz = mlx5_ib_get_cqe_size(dev, init_attr->recv_cq); scqe_sz = mlx5_ib_get_cqe_size(dev, init_attr->send_cq); if (rcqe_sz == 128) MLX5_SET(qpc, qpc, cs_res, MLX5_RES_SCAT_DATA64_CQE); else MLX5_SET(qpc, qpc, cs_res, MLX5_RES_SCAT_DATA32_CQE); if (init_attr->sq_sig_type == IB_SIGNAL_ALL_WR) { if (scqe_sz == 128) MLX5_SET(qpc, qpc, cs_req, MLX5_REQ_SCAT_DATA64_CQE); else MLX5_SET(qpc, qpc, cs_req, MLX5_REQ_SCAT_DATA32_CQE); } } if (qp->rq.wqe_cnt) { MLX5_SET(qpc, qpc, log_rq_stride, qp->rq.wqe_shift - 4); MLX5_SET(qpc, qpc, log_rq_size, ilog2(qp->rq.wqe_cnt)); } if (init_attr->qp_type != IB_QPT_RAW_PACKET) MLX5_SET(qpc, qpc, ts_format, mlx5_get_qp_default_ts(dev->mdev)); MLX5_SET(qpc, qpc, rq_type, get_rx_type(qp, init_attr)); if (qp->sq.wqe_cnt) MLX5_SET(qpc, qpc, log_sq_size, ilog2(qp->sq.wqe_cnt)); else MLX5_SET(qpc, qpc, no_sq, 1); /* Set default resources */ switch (init_attr->qp_type) { case IB_QPT_XRC_TGT: MLX5_SET(qpc, qpc, cqn_rcv, to_mcq(devr->c0)->mcq.cqn); MLX5_SET(qpc, qpc, cqn_snd, to_mcq(devr->c0)->mcq.cqn); MLX5_SET(qpc, qpc, srqn_rmpn, to_msrq(devr->s0)->msrq.srqn); MLX5_SET(qpc, qpc, xrcd, to_mxrcd(init_attr->xrcd)->xrcdn); break; case IB_QPT_XRC_INI: MLX5_SET(qpc, qpc, cqn_rcv, to_mcq(devr->c0)->mcq.cqn); MLX5_SET(qpc, qpc, xrcd, to_mxrcd(devr->x1)->xrcdn); MLX5_SET(qpc, qpc, srqn_rmpn, to_msrq(devr->s0)->msrq.srqn); break; default: if (init_attr->srq) { MLX5_SET(qpc, qpc, xrcd, to_mxrcd(devr->x0)->xrcdn); MLX5_SET(qpc, qpc, srqn_rmpn, to_msrq(init_attr->srq)->msrq.srqn); } else { MLX5_SET(qpc, qpc, xrcd, to_mxrcd(devr->x1)->xrcdn); MLX5_SET(qpc, qpc, srqn_rmpn, to_msrq(devr->s1)->msrq.srqn); } } if (init_attr->send_cq) MLX5_SET(qpc, qpc, cqn_snd, to_mcq(init_attr->send_cq)->mcq.cqn); if (init_attr->recv_cq) MLX5_SET(qpc, qpc, cqn_rcv, to_mcq(init_attr->recv_cq)->mcq.cqn); MLX5_SET64(qpc, qpc, dbr_addr, qp->db.dma); /* 0xffffff means we ask to work with cqe version 0 */ if (MLX5_CAP_GEN(mdev, cqe_version) == MLX5_CQE_VERSION_V1) MLX5_SET(qpc, qpc, user_index, uidx); /* we use IB_QP_CREATE_IPOIB_UD_LSO to indicates ipoib qp */ if (init_attr->qp_type == IB_QPT_UD && (init_attr->create_flags & IB_QP_CREATE_IPOIB_UD_LSO)) { MLX5_SET(qpc, qpc, ulp_stateless_offload_mode, 1); qp->flags |= MLX5_IB_QP_LSO; } if (init_attr->qp_type == IB_QPT_RAW_PACKET) { qp->raw_packet_qp.sq.ubuffer.buf_addr = ucmd.sq_buf_addr; raw_packet_qp_copy_info(qp, &qp->raw_packet_qp); err = create_raw_packet_qp(dev, qp, in, pd); } else { err = mlx5_core_create_qp(dev->mdev, &base->mqp, in, inlen); } if (err) { mlx5_ib_dbg(dev, "create qp failed\n"); goto err_create; } kvfree(in); base->container_mibqp = qp; base->mqp.event = mlx5_ib_qp_event; get_cqs(init_attr->qp_type, init_attr->send_cq, init_attr->recv_cq, &send_cq, &recv_cq); spin_lock_irqsave(&dev->reset_flow_resource_lock, flags); mlx5_ib_lock_cqs(send_cq, recv_cq); /* Maintain device to QPs access, needed for further handling via reset * flow */ list_add_tail(&qp->qps_list, &dev->qp_list); /* Maintain CQ to QPs access, needed for further handling via reset flow */ if (send_cq) list_add_tail(&qp->cq_send_list, &send_cq->list_send_qp); if (recv_cq) list_add_tail(&qp->cq_recv_list, &recv_cq->list_recv_qp); mlx5_ib_unlock_cqs(send_cq, recv_cq); spin_unlock_irqrestore(&dev->reset_flow_resource_lock, flags); return 0; err_create: if (qp->create_type == MLX5_QP_USER) destroy_qp_user(dev, pd, qp, base); else if (qp->create_type == MLX5_QP_KERNEL) destroy_qp_kernel(dev, qp); kvfree(in); return err; } static void mlx5_ib_lock_cqs(struct mlx5_ib_cq *send_cq, struct mlx5_ib_cq *recv_cq) __acquires(&send_cq->lock) __acquires(&recv_cq->lock) { if (send_cq) { if (recv_cq) { if (send_cq->mcq.cqn < recv_cq->mcq.cqn) { spin_lock(&send_cq->lock); spin_lock_nested(&recv_cq->lock, SINGLE_DEPTH_NESTING); } else if (send_cq->mcq.cqn == recv_cq->mcq.cqn) { spin_lock(&send_cq->lock); __acquire(&recv_cq->lock); } else { spin_lock(&recv_cq->lock); spin_lock_nested(&send_cq->lock, SINGLE_DEPTH_NESTING); } } else { spin_lock(&send_cq->lock); __acquire(&recv_cq->lock); } } else if (recv_cq) { spin_lock(&recv_cq->lock); __acquire(&send_cq->lock); } else { __acquire(&send_cq->lock); __acquire(&recv_cq->lock); } } static void mlx5_ib_unlock_cqs(struct mlx5_ib_cq *send_cq, struct mlx5_ib_cq *recv_cq) __releases(&send_cq->lock) __releases(&recv_cq->lock) { if (send_cq) { if (recv_cq) { if (send_cq->mcq.cqn < recv_cq->mcq.cqn) { spin_unlock(&recv_cq->lock); spin_unlock(&send_cq->lock); } else if (send_cq->mcq.cqn == recv_cq->mcq.cqn) { __release(&recv_cq->lock); spin_unlock(&send_cq->lock); } else { spin_unlock(&send_cq->lock); spin_unlock(&recv_cq->lock); } } else { __release(&recv_cq->lock); spin_unlock(&send_cq->lock); } } else if (recv_cq) { __release(&send_cq->lock); spin_unlock(&recv_cq->lock); } else { __release(&recv_cq->lock); __release(&send_cq->lock); } } static struct mlx5_ib_pd *get_pd(struct mlx5_ib_qp *qp) { return to_mpd(qp->ibqp.pd); } static void get_cqs(enum ib_qp_type qp_type, struct ib_cq *ib_send_cq, struct ib_cq *ib_recv_cq, struct mlx5_ib_cq **send_cq, struct mlx5_ib_cq **recv_cq) { switch (qp_type) { case IB_QPT_XRC_TGT: *send_cq = NULL; *recv_cq = NULL; break; case MLX5_IB_QPT_REG_UMR: case IB_QPT_XRC_INI: *send_cq = ib_send_cq ? to_mcq(ib_send_cq) : NULL; *recv_cq = NULL; break; case IB_QPT_SMI: case MLX5_IB_QPT_HW_GSI: case IB_QPT_RC: case IB_QPT_UC: case IB_QPT_UD: case IB_QPT_RAW_IPV6: case IB_QPT_RAW_ETHERTYPE: case IB_QPT_RAW_PACKET: *send_cq = ib_send_cq ? to_mcq(ib_send_cq) : NULL; *recv_cq = ib_recv_cq ? to_mcq(ib_recv_cq) : NULL; break; case IB_QPT_MAX: default: *send_cq = NULL; *recv_cq = NULL; break; } } static int modify_raw_packet_qp(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, const struct mlx5_modify_raw_qp_param *raw_qp_param, u8 lag_tx_affinity); static void destroy_qp_common(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp) { struct mlx5_ib_cq *send_cq, *recv_cq; struct mlx5_ib_qp_base *base = &qp->trans_qp.base; unsigned long flags; int err; if (qp->ibqp.rwq_ind_tbl) { destroy_rss_raw_qp_tir(dev, qp); return; } base = qp->ibqp.qp_type == IB_QPT_RAW_PACKET ? &qp->raw_packet_qp.rq.base : &qp->trans_qp.base; if (qp->state != IB_QPS_RESET) { if (qp->ibqp.qp_type != IB_QPT_RAW_PACKET) { mlx5_ib_qp_disable_pagefaults(qp); err = mlx5_core_qp_modify(dev->mdev, MLX5_CMD_OP_2RST_QP, 0, NULL, &base->mqp); } else { struct mlx5_modify_raw_qp_param raw_qp_param = { .operation = MLX5_CMD_OP_2RST_QP }; err = modify_raw_packet_qp(dev, qp, &raw_qp_param, 0); } if (err) mlx5_ib_warn(dev, "mlx5_ib: modify QP 0x%06x to RESET failed\n", base->mqp.qpn); } get_cqs(qp->ibqp.qp_type, qp->ibqp.send_cq, qp->ibqp.recv_cq, &send_cq, &recv_cq); spin_lock_irqsave(&dev->reset_flow_resource_lock, flags); mlx5_ib_lock_cqs(send_cq, recv_cq); /* del from lists under both locks above to protect reset flow paths */ list_del(&qp->qps_list); if (send_cq) list_del(&qp->cq_send_list); if (recv_cq) list_del(&qp->cq_recv_list); if (qp->create_type == MLX5_QP_KERNEL) { __mlx5_ib_cq_clean(recv_cq, base->mqp.qpn, qp->ibqp.srq ? to_msrq(qp->ibqp.srq) : NULL); if (send_cq != recv_cq) __mlx5_ib_cq_clean(send_cq, base->mqp.qpn, NULL); } mlx5_ib_unlock_cqs(send_cq, recv_cq); spin_unlock_irqrestore(&dev->reset_flow_resource_lock, flags); if (qp->ibqp.qp_type == IB_QPT_RAW_PACKET) { destroy_raw_packet_qp(dev, qp); } else { err = mlx5_core_destroy_qp(dev->mdev, &base->mqp); if (err) mlx5_ib_warn(dev, "failed to destroy QP 0x%x\n", base->mqp.qpn); } if (qp->create_type == MLX5_QP_KERNEL) destroy_qp_kernel(dev, qp); else if (qp->create_type == MLX5_QP_USER) destroy_qp_user(dev, &get_pd(qp)->ibpd, qp, base); } static const char *ib_qp_type_str(enum ib_qp_type type) { switch (type) { case IB_QPT_SMI: return "IB_QPT_SMI"; case IB_QPT_GSI: return "IB_QPT_GSI"; case IB_QPT_RC: return "IB_QPT_RC"; case IB_QPT_UC: return "IB_QPT_UC"; case IB_QPT_UD: return "IB_QPT_UD"; case IB_QPT_RAW_IPV6: return "IB_QPT_RAW_IPV6"; case IB_QPT_RAW_ETHERTYPE: return "IB_QPT_RAW_ETHERTYPE"; case IB_QPT_XRC_INI: return "IB_QPT_XRC_INI"; case IB_QPT_XRC_TGT: return "IB_QPT_XRC_TGT"; case IB_QPT_RAW_PACKET: return "IB_QPT_RAW_PACKET"; case MLX5_IB_QPT_REG_UMR: return "MLX5_IB_QPT_REG_UMR"; case IB_QPT_MAX: default: return "Invalid QP type"; } } struct ib_qp *mlx5_ib_create_qp(struct ib_pd *pd, struct ib_qp_init_attr *init_attr, struct ib_udata *udata) { struct mlx5_ib_dev *dev; struct mlx5_ib_qp *qp; u16 xrcdn = 0; int err; if (pd) { dev = to_mdev(pd->device); if (init_attr->qp_type == IB_QPT_RAW_PACKET) { if (!pd->uobject) { mlx5_ib_dbg(dev, "Raw Packet QP is not supported for kernel consumers\n"); return ERR_PTR(-EINVAL); } else if (!to_mucontext(pd->uobject->context)->cqe_version) { mlx5_ib_dbg(dev, "Raw Packet QP is only supported for CQE version > 0\n"); return ERR_PTR(-EINVAL); } } } else { /* being cautious here */ if (init_attr->qp_type != IB_QPT_XRC_TGT && init_attr->qp_type != MLX5_IB_QPT_REG_UMR) { pr_warn("%s: no PD for transport %s\n", __func__, ib_qp_type_str(init_attr->qp_type)); return ERR_PTR(-EINVAL); } dev = to_mdev(to_mxrcd(init_attr->xrcd)->ibxrcd.device); } switch (init_attr->qp_type) { case IB_QPT_XRC_TGT: case IB_QPT_XRC_INI: if (!MLX5_CAP_GEN(dev->mdev, xrc)) { mlx5_ib_dbg(dev, "XRC not supported\n"); return ERR_PTR(-ENOSYS); } init_attr->recv_cq = NULL; if (init_attr->qp_type == IB_QPT_XRC_TGT) { xrcdn = to_mxrcd(init_attr->xrcd)->xrcdn; init_attr->send_cq = NULL; } /* fall through */ case IB_QPT_RAW_PACKET: case IB_QPT_RC: case IB_QPT_UC: case IB_QPT_UD: case IB_QPT_SMI: case MLX5_IB_QPT_HW_GSI: case MLX5_IB_QPT_REG_UMR: qp = kzalloc(sizeof(*qp), GFP_KERNEL); if (!qp) return ERR_PTR(-ENOMEM); err = create_qp_common(dev, pd, init_attr, udata, qp); if (err) { mlx5_ib_dbg(dev, "create_qp_common failed\n"); kfree(qp); return ERR_PTR(err); } if (is_qp0(init_attr->qp_type)) qp->ibqp.qp_num = 0; else if (is_qp1(init_attr->qp_type)) qp->ibqp.qp_num = 1; else qp->ibqp.qp_num = qp->trans_qp.base.mqp.qpn; mlx5_ib_dbg(dev, "ib qpnum 0x%x, mlx qpn 0x%x, rcqn 0x%x, scqn 0x%x\n", qp->ibqp.qp_num, qp->trans_qp.base.mqp.qpn, init_attr->recv_cq ? to_mcq(init_attr->recv_cq)->mcq.cqn : -1, init_attr->send_cq ? to_mcq(init_attr->send_cq)->mcq.cqn : -1); qp->trans_qp.xrcdn = xrcdn; break; case IB_QPT_GSI: return mlx5_ib_gsi_create_qp(pd, init_attr); case IB_QPT_RAW_IPV6: case IB_QPT_RAW_ETHERTYPE: case IB_QPT_MAX: default: mlx5_ib_dbg(dev, "unsupported qp type %d\n", init_attr->qp_type); /* Don't support raw QPs */ return ERR_PTR(-EINVAL); } return &qp->ibqp; } int mlx5_ib_destroy_qp(struct ib_qp *qp) { struct mlx5_ib_dev *dev = to_mdev(qp->device); struct mlx5_ib_qp *mqp = to_mqp(qp); if (unlikely(qp->qp_type == IB_QPT_GSI)) return mlx5_ib_gsi_destroy_qp(qp); destroy_qp_common(dev, mqp); kfree(mqp); return 0; } static int to_mlx5_access_flags(struct mlx5_ib_qp *qp, const struct ib_qp_attr *attr, int attr_mask, __be32 *hw_access_flags_be) { u8 dest_rd_atomic; u32 access_flags, hw_access_flags = 0; struct mlx5_ib_dev *dev = to_mdev(qp->ibqp.device); if (attr_mask & IB_QP_MAX_DEST_RD_ATOMIC) dest_rd_atomic = attr->max_dest_rd_atomic; else dest_rd_atomic = qp->trans_qp.resp_depth; if (attr_mask & IB_QP_ACCESS_FLAGS) access_flags = attr->qp_access_flags; else access_flags = qp->trans_qp.atomic_rd_en; if (!dest_rd_atomic) access_flags &= IB_ACCESS_REMOTE_WRITE; if (access_flags & IB_ACCESS_REMOTE_READ) hw_access_flags |= MLX5_QP_BIT_RRE; if (access_flags & IB_ACCESS_REMOTE_ATOMIC) { int atomic_mode; atomic_mode = get_atomic_mode(dev, qp->ibqp.qp_type); if (atomic_mode < 0) return -EOPNOTSUPP; hw_access_flags |= MLX5_QP_BIT_RAE; hw_access_flags |= atomic_mode << MLX5_ATOMIC_MODE_OFF; } if (access_flags & IB_ACCESS_REMOTE_WRITE) hw_access_flags |= MLX5_QP_BIT_RWE; *hw_access_flags_be = cpu_to_be32(hw_access_flags); return 0; } enum { MLX5_PATH_FLAG_FL = 1 << 0, MLX5_PATH_FLAG_FREE_AR = 1 << 1, MLX5_PATH_FLAG_COUNTER = 1 << 2, }; static int ib_rate_to_mlx5(struct mlx5_ib_dev *dev, u8 rate) { if (rate == IB_RATE_PORT_CURRENT) { return 0; } else if (rate < IB_RATE_2_5_GBPS || rate > IB_RATE_600_GBPS) { return -EINVAL; } else { while (rate != IB_RATE_2_5_GBPS && !(1 << (rate + MLX5_STAT_RATE_OFFSET) & MLX5_CAP_GEN(dev->mdev, stat_rate_support))) --rate; } return rate + MLX5_STAT_RATE_OFFSET; } static int modify_raw_packet_eth_prio(struct mlx5_core_dev *dev, struct mlx5_ib_sq *sq, u8 sl) { void *in; void *tisc; int inlen; int err; inlen = MLX5_ST_SZ_BYTES(modify_tis_in); in = mlx5_vzalloc(inlen); if (!in) return -ENOMEM; MLX5_SET(modify_tis_in, in, bitmask.prio, 1); tisc = MLX5_ADDR_OF(modify_tis_in, in, ctx); MLX5_SET(tisc, tisc, prio, ((sl & 0x7) << 1)); err = mlx5_core_modify_tis(dev, sq->tisn, in, inlen); kvfree(in); return err; } static int modify_raw_packet_tx_affinity(struct mlx5_core_dev *dev, struct mlx5_ib_sq *sq, u8 tx_affinity) { void *in; void *tisc; int inlen; int err; inlen = MLX5_ST_SZ_BYTES(modify_tis_in); in = mlx5_vzalloc(inlen); if (!in) return -ENOMEM; MLX5_SET(modify_tis_in, in, bitmask.lag_tx_port_affinity, 1); tisc = MLX5_ADDR_OF(modify_tis_in, in, ctx); MLX5_SET(tisc, tisc, lag_tx_port_affinity, tx_affinity); err = mlx5_core_modify_tis(dev, sq->tisn, in, inlen); kvfree(in); return err; } static int mlx5_set_path(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, const struct ib_ah_attr *ah, struct mlx5_qp_path *path, u8 port, int attr_mask, u32 path_flags, const struct ib_qp_attr *attr, bool alt) { enum rdma_link_layer ll = rdma_port_get_link_layer(&dev->ib_dev, port); int err; enum ib_gid_type gid_type; if (attr_mask & IB_QP_PKEY_INDEX) path->pkey_index = cpu_to_be16(alt ? attr->alt_pkey_index : attr->pkey_index); if (ah->ah_flags & IB_AH_GRH) { if (ah->grh.sgid_index >= dev->mdev->port_caps[port - 1].gid_table_len) { pr_err("sgid_index (%u) too large. max is %d\n", ah->grh.sgid_index, dev->mdev->port_caps[port - 1].gid_table_len); return -EINVAL; } } if (ll == IB_LINK_LAYER_ETHERNET) { if (!(ah->ah_flags & IB_AH_GRH)) return -EINVAL; err = mlx5_get_roce_gid_type(dev, port, ah->grh.sgid_index, &gid_type); if (err) return err; memcpy(path->rmac, ah->dmac, sizeof(ah->dmac)); path->udp_sport = mlx5_get_roce_udp_sport(dev, port, ah->grh.sgid_index); path->dci_cfi_prio_sl = (ah->sl & 0x7) << 4; if (gid_type == IB_GID_TYPE_ROCE_UDP_ENCAP) path->ecn_dscp = (ah->grh.traffic_class >> 2) & 0x3f; } else { path->fl_free_ar = (path_flags & MLX5_PATH_FLAG_FL) ? 0x80 : 0; path->fl_free_ar |= (path_flags & MLX5_PATH_FLAG_FREE_AR) ? 0x40 : 0; path->rlid = cpu_to_be16(ah->dlid); path->grh_mlid = ah->src_path_bits & 0x7f; if (ah->ah_flags & IB_AH_GRH) path->grh_mlid |= 1 << 7; path->dci_cfi_prio_sl = ah->sl & 0xf; } if (ah->ah_flags & IB_AH_GRH) { path->mgid_index = ah->grh.sgid_index; path->hop_limit = ah->grh.hop_limit; path->tclass_flowlabel = cpu_to_be32((ah->grh.traffic_class << 20) | (ah->grh.flow_label)); memcpy(path->rgid, ah->grh.dgid.raw, 16); } err = ib_rate_to_mlx5(dev, ah->static_rate); if (err < 0) return err; path->static_rate = err; path->port = port; if (attr_mask & IB_QP_TIMEOUT) path->ackto_lt = (alt ? attr->alt_timeout : attr->timeout) << 3; if ((qp->ibqp.qp_type == IB_QPT_RAW_PACKET) && qp->sq.wqe_cnt) return modify_raw_packet_eth_prio(dev->mdev, &qp->raw_packet_qp.sq, ah->sl & 0xf); return 0; } static enum mlx5_qp_optpar opt_mask[MLX5_QP_NUM_STATE][MLX5_QP_NUM_STATE][MLX5_QP_ST_MAX] = { [MLX5_QP_STATE_INIT] = { [MLX5_QP_STATE_INIT] = { [MLX5_QP_ST_RC] = MLX5_QP_OPTPAR_RRE | MLX5_QP_OPTPAR_RAE | MLX5_QP_OPTPAR_RWE | MLX5_QP_OPTPAR_PKEY_INDEX | MLX5_QP_OPTPAR_PRI_PORT, [MLX5_QP_ST_UC] = MLX5_QP_OPTPAR_RWE | MLX5_QP_OPTPAR_PKEY_INDEX | MLX5_QP_OPTPAR_PRI_PORT, [MLX5_QP_ST_UD] = MLX5_QP_OPTPAR_PKEY_INDEX | MLX5_QP_OPTPAR_Q_KEY | MLX5_QP_OPTPAR_PRI_PORT, }, [MLX5_QP_STATE_RTR] = { [MLX5_QP_ST_RC] = MLX5_QP_OPTPAR_ALT_ADDR_PATH | MLX5_QP_OPTPAR_RRE | MLX5_QP_OPTPAR_RAE | MLX5_QP_OPTPAR_RWE | MLX5_QP_OPTPAR_PKEY_INDEX, [MLX5_QP_ST_UC] = MLX5_QP_OPTPAR_ALT_ADDR_PATH | MLX5_QP_OPTPAR_RWE | MLX5_QP_OPTPAR_PKEY_INDEX, [MLX5_QP_ST_UD] = MLX5_QP_OPTPAR_PKEY_INDEX | MLX5_QP_OPTPAR_Q_KEY, [MLX5_QP_ST_MLX] = MLX5_QP_OPTPAR_PKEY_INDEX | MLX5_QP_OPTPAR_Q_KEY, [MLX5_QP_ST_XRC] = MLX5_QP_OPTPAR_ALT_ADDR_PATH | MLX5_QP_OPTPAR_RRE | MLX5_QP_OPTPAR_RAE | MLX5_QP_OPTPAR_RWE | MLX5_QP_OPTPAR_PKEY_INDEX, }, }, [MLX5_QP_STATE_RTR] = { [MLX5_QP_STATE_RTS] = { [MLX5_QP_ST_RC] = MLX5_QP_OPTPAR_ALT_ADDR_PATH | MLX5_QP_OPTPAR_RRE | MLX5_QP_OPTPAR_RAE | MLX5_QP_OPTPAR_RWE | MLX5_QP_OPTPAR_PM_STATE | MLX5_QP_OPTPAR_RNR_TIMEOUT, [MLX5_QP_ST_UC] = MLX5_QP_OPTPAR_ALT_ADDR_PATH | MLX5_QP_OPTPAR_RWE | MLX5_QP_OPTPAR_PM_STATE, [MLX5_QP_ST_UD] = MLX5_QP_OPTPAR_Q_KEY, }, }, [MLX5_QP_STATE_RTS] = { [MLX5_QP_STATE_RTS] = { [MLX5_QP_ST_RC] = MLX5_QP_OPTPAR_RRE | MLX5_QP_OPTPAR_RAE | MLX5_QP_OPTPAR_RWE | MLX5_QP_OPTPAR_RNR_TIMEOUT | MLX5_QP_OPTPAR_PM_STATE | MLX5_QP_OPTPAR_ALT_ADDR_PATH, [MLX5_QP_ST_UC] = MLX5_QP_OPTPAR_RWE | MLX5_QP_OPTPAR_PM_STATE | MLX5_QP_OPTPAR_ALT_ADDR_PATH, [MLX5_QP_ST_UD] = MLX5_QP_OPTPAR_Q_KEY | MLX5_QP_OPTPAR_SRQN | MLX5_QP_OPTPAR_CQN_RCV, }, }, [MLX5_QP_STATE_SQER] = { [MLX5_QP_STATE_RTS] = { [MLX5_QP_ST_UD] = MLX5_QP_OPTPAR_Q_KEY, [MLX5_QP_ST_MLX] = MLX5_QP_OPTPAR_Q_KEY, [MLX5_QP_ST_UC] = MLX5_QP_OPTPAR_RWE, [MLX5_QP_ST_RC] = MLX5_QP_OPTPAR_RNR_TIMEOUT | MLX5_QP_OPTPAR_RWE | MLX5_QP_OPTPAR_RAE | MLX5_QP_OPTPAR_RRE, }, }, }; static int ib_nr_to_mlx5_nr(int ib_mask) { switch (ib_mask) { case IB_QP_STATE: return 0; case IB_QP_CUR_STATE: return 0; case IB_QP_EN_SQD_ASYNC_NOTIFY: return 0; case IB_QP_ACCESS_FLAGS: return MLX5_QP_OPTPAR_RWE | MLX5_QP_OPTPAR_RRE | MLX5_QP_OPTPAR_RAE; case IB_QP_PKEY_INDEX: return MLX5_QP_OPTPAR_PKEY_INDEX; case IB_QP_PORT: return MLX5_QP_OPTPAR_PRI_PORT; case IB_QP_QKEY: return MLX5_QP_OPTPAR_Q_KEY; case IB_QP_AV: return MLX5_QP_OPTPAR_PRIMARY_ADDR_PATH | MLX5_QP_OPTPAR_PRI_PORT; case IB_QP_PATH_MTU: return 0; case IB_QP_TIMEOUT: return MLX5_QP_OPTPAR_ACK_TIMEOUT; case IB_QP_RETRY_CNT: return MLX5_QP_OPTPAR_RETRY_COUNT; case IB_QP_RNR_RETRY: return MLX5_QP_OPTPAR_RNR_RETRY; case IB_QP_RQ_PSN: return 0; case IB_QP_MAX_QP_RD_ATOMIC: return MLX5_QP_OPTPAR_SRA_MAX; case IB_QP_ALT_PATH: return MLX5_QP_OPTPAR_ALT_ADDR_PATH; case IB_QP_MIN_RNR_TIMER: return MLX5_QP_OPTPAR_RNR_TIMEOUT; case IB_QP_SQ_PSN: return 0; case IB_QP_MAX_DEST_RD_ATOMIC: return MLX5_QP_OPTPAR_RRA_MAX | MLX5_QP_OPTPAR_RWE | MLX5_QP_OPTPAR_RRE | MLX5_QP_OPTPAR_RAE; case IB_QP_PATH_MIG_STATE: return MLX5_QP_OPTPAR_PM_STATE; case IB_QP_CAP: return 0; case IB_QP_DEST_QPN: return 0; } return 0; } static int ib_mask_to_mlx5_opt(int ib_mask) { int result = 0; int i; for (i = 0; i < 8 * sizeof(int); i++) { if ((1 << i) & ib_mask) result |= ib_nr_to_mlx5_nr(1 << i); } return result; } static int modify_raw_packet_qp_rq(struct mlx5_ib_dev *dev, struct mlx5_ib_rq *rq, int new_state, const struct mlx5_modify_raw_qp_param *raw_qp_param) { void *in; void *rqc; int inlen; int err; inlen = MLX5_ST_SZ_BYTES(modify_rq_in); in = mlx5_vzalloc(inlen); if (!in) return -ENOMEM; MLX5_SET(modify_rq_in, in, rqn, rq->base.mqp.qpn); MLX5_SET(modify_rq_in, in, rq_state, rq->state); rqc = MLX5_ADDR_OF(modify_rq_in, in, ctx); MLX5_SET(rqc, rqc, state, new_state); if (raw_qp_param->set_mask & MLX5_RAW_QP_MOD_SET_RQ_Q_CTR_ID) { if (MLX5_CAP_GEN(dev->mdev, modify_rq_counters_set_id)) { MLX5_SET64(modify_rq_in, in, modify_bitmask, MLX5_MODIFY_RQ_IN_MODIFY_BITMASK_MODIFY_RQ_COUNTER_SET_ID); MLX5_SET(rqc, rqc, counter_set_id, raw_qp_param->rq_q_ctr_id); } else pr_info_once("%s: RAW PACKET QP counters are not supported on current FW\n", dev->ib_dev.name); } err = mlx5_core_modify_rq(dev->mdev, in, inlen); if (err) goto out; rq->state = new_state; out: kvfree(in); return err; } static int modify_raw_packet_qp_sq(struct mlx5_core_dev *dev, struct mlx5_ib_sq *sq, int new_state) { void *in; void *sqc; int inlen; int err; inlen = MLX5_ST_SZ_BYTES(modify_sq_in); in = mlx5_vzalloc(inlen); if (!in) return -ENOMEM; MLX5_SET(modify_sq_in, in, sqn, sq->base.mqp.qpn); MLX5_SET(modify_sq_in, in, sq_state, sq->state); sqc = MLX5_ADDR_OF(modify_sq_in, in, ctx); MLX5_SET(sqc, sqc, state, new_state); err = mlx5_core_modify_sq(dev, in, inlen); if (err) goto out; sq->state = new_state; out: kvfree(in); return err; } static int modify_raw_packet_qp(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, const struct mlx5_modify_raw_qp_param *raw_qp_param, u8 tx_affinity) { struct mlx5_ib_raw_packet_qp *raw_packet_qp = &qp->raw_packet_qp; struct mlx5_ib_rq *rq = &raw_packet_qp->rq; struct mlx5_ib_sq *sq = &raw_packet_qp->sq; int rq_state; int sq_state; int err; switch (raw_qp_param->operation) { case MLX5_CMD_OP_RST2INIT_QP: rq_state = MLX5_RQC_STATE_RDY; sq_state = MLX5_SQC_STATE_RDY; break; case MLX5_CMD_OP_2ERR_QP: rq_state = MLX5_RQC_STATE_ERR; sq_state = MLX5_SQC_STATE_ERR; break; case MLX5_CMD_OP_2RST_QP: rq_state = MLX5_RQC_STATE_RST; sq_state = MLX5_SQC_STATE_RST; break; case MLX5_CMD_OP_INIT2INIT_QP: case MLX5_CMD_OP_INIT2RTR_QP: case MLX5_CMD_OP_RTR2RTS_QP: case MLX5_CMD_OP_RTS2RTS_QP: if (raw_qp_param->set_mask) return -EINVAL; else return 0; default: WARN_ON(1); return -EINVAL; } if (qp->rq.wqe_cnt) { err = modify_raw_packet_qp_rq(dev, rq, rq_state, raw_qp_param); if (err) return err; } if (qp->sq.wqe_cnt) { if (tx_affinity) { err = modify_raw_packet_tx_affinity(dev->mdev, sq, tx_affinity); if (err) return err; } return modify_raw_packet_qp_sq(dev->mdev, sq, sq_state); } return 0; } static int __mlx5_ib_modify_qp(struct ib_qp *ibqp, const struct ib_qp_attr *attr, int attr_mask, enum ib_qp_state cur_state, enum ib_qp_state new_state) { static const u16 optab[MLX5_QP_NUM_STATE][MLX5_QP_NUM_STATE] = { [MLX5_QP_STATE_RST] = { [MLX5_QP_STATE_RST] = MLX5_CMD_OP_2RST_QP, [MLX5_QP_STATE_ERR] = MLX5_CMD_OP_2ERR_QP, [MLX5_QP_STATE_INIT] = MLX5_CMD_OP_RST2INIT_QP, }, [MLX5_QP_STATE_INIT] = { [MLX5_QP_STATE_RST] = MLX5_CMD_OP_2RST_QP, [MLX5_QP_STATE_ERR] = MLX5_CMD_OP_2ERR_QP, [MLX5_QP_STATE_INIT] = MLX5_CMD_OP_INIT2INIT_QP, [MLX5_QP_STATE_RTR] = MLX5_CMD_OP_INIT2RTR_QP, }, [MLX5_QP_STATE_RTR] = { [MLX5_QP_STATE_RST] = MLX5_CMD_OP_2RST_QP, [MLX5_QP_STATE_ERR] = MLX5_CMD_OP_2ERR_QP, [MLX5_QP_STATE_RTS] = MLX5_CMD_OP_RTR2RTS_QP, }, [MLX5_QP_STATE_RTS] = { [MLX5_QP_STATE_RST] = MLX5_CMD_OP_2RST_QP, [MLX5_QP_STATE_ERR] = MLX5_CMD_OP_2ERR_QP, [MLX5_QP_STATE_RTS] = MLX5_CMD_OP_RTS2RTS_QP, }, [MLX5_QP_STATE_SQD] = { [MLX5_QP_STATE_RST] = MLX5_CMD_OP_2RST_QP, [MLX5_QP_STATE_ERR] = MLX5_CMD_OP_2ERR_QP, }, [MLX5_QP_STATE_SQER] = { [MLX5_QP_STATE_RST] = MLX5_CMD_OP_2RST_QP, [MLX5_QP_STATE_ERR] = MLX5_CMD_OP_2ERR_QP, [MLX5_QP_STATE_RTS] = MLX5_CMD_OP_SQERR2RTS_QP, }, [MLX5_QP_STATE_ERR] = { [MLX5_QP_STATE_RST] = MLX5_CMD_OP_2RST_QP, [MLX5_QP_STATE_ERR] = MLX5_CMD_OP_2ERR_QP, } }; struct mlx5_ib_dev *dev = to_mdev(ibqp->device); struct mlx5_ib_qp *qp = to_mqp(ibqp); struct mlx5_ib_qp_base *base = &qp->trans_qp.base; struct mlx5_ib_cq *send_cq, *recv_cq; struct mlx5_qp_context *context; struct mlx5_ib_pd *pd; struct mlx5_ib_port *mibport = NULL; enum mlx5_qp_state mlx5_cur, mlx5_new; enum mlx5_qp_optpar optpar; int sqd_event; int mlx5_st; int err; u16 op; context = kzalloc(sizeof(*context), GFP_KERNEL); if (!context) return -ENOMEM; err = to_mlx5_st(ibqp->qp_type); if (err < 0) { mlx5_ib_dbg(dev, "unsupported qp type %d\n", ibqp->qp_type); goto out; } context->flags = cpu_to_be32(err << 16); if (!(attr_mask & IB_QP_PATH_MIG_STATE)) { context->flags |= cpu_to_be32(MLX5_QP_PM_MIGRATED << 11); } else { switch (attr->path_mig_state) { case IB_MIG_MIGRATED: context->flags |= cpu_to_be32(MLX5_QP_PM_MIGRATED << 11); break; case IB_MIG_REARM: context->flags |= cpu_to_be32(MLX5_QP_PM_REARM << 11); break; case IB_MIG_ARMED: context->flags |= cpu_to_be32(MLX5_QP_PM_ARMED << 11); break; } } if (is_sqp(ibqp->qp_type)) { context->mtu_msgmax = (IB_MTU_256 << 5) | 8; } else if (ibqp->qp_type == IB_QPT_UD || ibqp->qp_type == MLX5_IB_QPT_REG_UMR) { context->mtu_msgmax = (IB_MTU_4096 << 5) | 12; } else if (attr_mask & IB_QP_PATH_MTU) { if (attr->path_mtu < IB_MTU_256 || attr->path_mtu > IB_MTU_4096) { mlx5_ib_warn(dev, "invalid mtu %d\n", attr->path_mtu); err = -EINVAL; goto out; } context->mtu_msgmax = (attr->path_mtu << 5) | (u8)MLX5_CAP_GEN(dev->mdev, log_max_msg); } if (attr_mask & IB_QP_DEST_QPN) context->log_pg_sz_remote_qpn = cpu_to_be32(attr->dest_qp_num); if (attr_mask & IB_QP_PKEY_INDEX) context->pri_path.pkey_index = cpu_to_be16(attr->pkey_index); /* todo implement counter_index functionality */ if (is_sqp(ibqp->qp_type)) context->pri_path.port = qp->port; if (attr_mask & IB_QP_PORT) context->pri_path.port = attr->port_num; if (attr_mask & IB_QP_AV) { err = mlx5_set_path(dev, qp, &attr->ah_attr, &context->pri_path, attr_mask & IB_QP_PORT ? attr->port_num : qp->port, attr_mask, 0, attr, false); if (err) goto out; } if (attr_mask & IB_QP_TIMEOUT) context->pri_path.ackto_lt |= attr->timeout << 3; if (attr_mask & IB_QP_ALT_PATH) { err = mlx5_set_path(dev, qp, &attr->alt_ah_attr, &context->alt_path, attr->alt_port_num, attr_mask | IB_QP_PKEY_INDEX | IB_QP_TIMEOUT, 0, attr, true); if (err) goto out; } pd = get_pd(qp); get_cqs(qp->ibqp.qp_type, qp->ibqp.send_cq, qp->ibqp.recv_cq, &send_cq, &recv_cq); context->flags_pd = cpu_to_be32(pd ? pd->pdn : to_mpd(dev->devr.p0)->pdn); context->cqn_send = send_cq ? cpu_to_be32(send_cq->mcq.cqn) : 0; context->cqn_recv = recv_cq ? cpu_to_be32(recv_cq->mcq.cqn) : 0; context->params1 = cpu_to_be32(MLX5_IB_ACK_REQ_FREQ << 28); if (attr_mask & IB_QP_RNR_RETRY) context->params1 |= cpu_to_be32(attr->rnr_retry << 13); if (attr_mask & IB_QP_RETRY_CNT) context->params1 |= cpu_to_be32(attr->retry_cnt << 16); if (attr_mask & IB_QP_MAX_QP_RD_ATOMIC) { if (attr->max_rd_atomic) context->params1 |= cpu_to_be32(fls(attr->max_rd_atomic - 1) << 21); } if (attr_mask & IB_QP_SQ_PSN) context->next_send_psn = cpu_to_be32(attr->sq_psn); if (attr_mask & IB_QP_MAX_DEST_RD_ATOMIC) { if (attr->max_dest_rd_atomic) context->params2 |= cpu_to_be32(fls(attr->max_dest_rd_atomic - 1) << 21); } if (attr_mask & (IB_QP_ACCESS_FLAGS | IB_QP_MAX_DEST_RD_ATOMIC)) { __be32 access_flags; err = to_mlx5_access_flags(qp, attr, attr_mask, &access_flags); if (err) goto out; context->params2 |= access_flags; } if (attr_mask & IB_QP_MIN_RNR_TIMER) context->rnr_nextrecvpsn |= cpu_to_be32(attr->min_rnr_timer << 24); if (attr_mask & IB_QP_RQ_PSN) context->rnr_nextrecvpsn |= cpu_to_be32(attr->rq_psn); if (attr_mask & IB_QP_QKEY) context->qkey = cpu_to_be32(attr->qkey); if (qp->rq.wqe_cnt && cur_state == IB_QPS_RESET && new_state == IB_QPS_INIT) context->db_rec_addr = cpu_to_be64(qp->db.dma); if (cur_state == IB_QPS_RTS && new_state == IB_QPS_SQD && attr_mask & IB_QP_EN_SQD_ASYNC_NOTIFY && attr->en_sqd_async_notify) sqd_event = 1; else sqd_event = 0; if (cur_state == IB_QPS_RESET && new_state == IB_QPS_INIT) { u8 port_num = (attr_mask & IB_QP_PORT ? attr->port_num : qp->port) - 1; mibport = &dev->port[port_num]; context->qp_counter_set_usr_page |= cpu_to_be32((u32)(mibport->q_cnt_id) << 24); } if (!ibqp->uobject && cur_state == IB_QPS_RESET && new_state == IB_QPS_INIT) context->sq_crq_size |= cpu_to_be16(1 << 4); if (qp->flags & MLX5_IB_QP_SQPN_QP1) context->deth_sqpn = cpu_to_be32(1); mlx5_cur = to_mlx5_state(cur_state); mlx5_new = to_mlx5_state(new_state); mlx5_st = to_mlx5_st(ibqp->qp_type); if (mlx5_st < 0) goto out; /* If moving to a reset or error state, we must disable page faults on * this QP and flush all current page faults. Otherwise a stale page * fault may attempt to work on this QP after it is reset and moved * again to RTS, and may cause the driver and the device to get out of * sync. */ if (cur_state != IB_QPS_RESET && cur_state != IB_QPS_ERR && (new_state == IB_QPS_RESET || new_state == IB_QPS_ERR) && (qp->ibqp.qp_type != IB_QPT_RAW_PACKET)) mlx5_ib_qp_disable_pagefaults(qp); if (mlx5_cur >= MLX5_QP_NUM_STATE || mlx5_new >= MLX5_QP_NUM_STATE || !optab[mlx5_cur][mlx5_new]) goto out; op = optab[mlx5_cur][mlx5_new]; optpar = ib_mask_to_mlx5_opt(attr_mask); optpar &= opt_mask[mlx5_cur][mlx5_new][mlx5_st]; if (qp->ibqp.qp_type == IB_QPT_RAW_PACKET) { struct mlx5_modify_raw_qp_param raw_qp_param = {}; raw_qp_param.operation = op; if (cur_state == IB_QPS_RESET && new_state == IB_QPS_INIT) { raw_qp_param.rq_q_ctr_id = mibport->q_cnt_id; raw_qp_param.set_mask |= MLX5_RAW_QP_MOD_SET_RQ_Q_CTR_ID; } err = modify_raw_packet_qp(dev, qp, &raw_qp_param, 0); } else { err = mlx5_core_qp_modify(dev->mdev, op, optpar, context, &base->mqp); } if (err) goto out; if (cur_state == IB_QPS_RESET && new_state == IB_QPS_INIT && (qp->ibqp.qp_type != IB_QPT_RAW_PACKET)) mlx5_ib_qp_enable_pagefaults(qp); qp->state = new_state; if (attr_mask & IB_QP_ACCESS_FLAGS) qp->trans_qp.atomic_rd_en = attr->qp_access_flags; if (attr_mask & IB_QP_MAX_DEST_RD_ATOMIC) qp->trans_qp.resp_depth = attr->max_dest_rd_atomic; if (attr_mask & IB_QP_PORT) qp->port = attr->port_num; if (attr_mask & IB_QP_ALT_PATH) qp->trans_qp.alt_port = attr->alt_port_num; /* * If we moved a kernel QP to RESET, clean up all old CQ * entries and reinitialize the QP. */ if (new_state == IB_QPS_RESET && !ibqp->uobject) { mlx5_ib_cq_clean(recv_cq, base->mqp.qpn, ibqp->srq ? to_msrq(ibqp->srq) : NULL); if (send_cq != recv_cq) mlx5_ib_cq_clean(send_cq, base->mqp.qpn, NULL); qp->rq.head = 0; qp->rq.tail = 0; qp->sq.head = 0; qp->sq.tail = 0; qp->sq.cur_post = 0; qp->sq.last_poll = 0; qp->db.db[MLX5_RCV_DBR] = 0; qp->db.db[MLX5_SND_DBR] = 0; } out: kfree(context); return err; } int mlx5_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr, int attr_mask, struct ib_udata *udata) { struct mlx5_ib_dev *dev = to_mdev(ibqp->device); struct mlx5_ib_qp *qp = to_mqp(ibqp); enum ib_qp_type qp_type; enum ib_qp_state cur_state, new_state; int err = -EINVAL; int port; if (ibqp->rwq_ind_tbl) return -ENOSYS; if (unlikely(ibqp->qp_type == IB_QPT_GSI)) return mlx5_ib_gsi_modify_qp(ibqp, attr, attr_mask); qp_type = (unlikely(ibqp->qp_type == MLX5_IB_QPT_HW_GSI)) ? IB_QPT_GSI : ibqp->qp_type; mutex_lock(&qp->mutex); cur_state = attr_mask & IB_QP_CUR_STATE ? attr->cur_qp_state : qp->state; new_state = attr_mask & IB_QP_STATE ? attr->qp_state : cur_state; if (qp_type != MLX5_IB_QPT_REG_UMR && !ib_modify_qp_is_ok(cur_state, new_state, qp_type, attr_mask)) { mlx5_ib_dbg(dev, "invalid QP state transition from %d to %d, qp_type %d, attr_mask 0x%x\n", cur_state, new_state, ibqp->qp_type, attr_mask); goto out; } if ((attr_mask & IB_QP_PORT) && (attr->port_num == 0 || attr->port_num > MLX5_CAP_GEN(dev->mdev, num_ports))) { mlx5_ib_dbg(dev, "invalid port number %d. number of ports is %d\n", attr->port_num, dev->num_ports); goto out; } if (attr_mask & IB_QP_PKEY_INDEX) { port = attr_mask & IB_QP_PORT ? attr->port_num : qp->port; if (attr->pkey_index >= dev->mdev->port_caps[port - 1].pkey_table_len) { mlx5_ib_dbg(dev, "invalid pkey index %d\n", attr->pkey_index); goto out; } } if (attr_mask & IB_QP_MAX_QP_RD_ATOMIC && attr->max_rd_atomic > (1 << MLX5_CAP_GEN(dev->mdev, log_max_ra_res_qp))) { mlx5_ib_dbg(dev, "invalid max_rd_atomic value %d\n", attr->max_rd_atomic); goto out; } if (attr_mask & IB_QP_MAX_DEST_RD_ATOMIC && attr->max_dest_rd_atomic > (1 << MLX5_CAP_GEN(dev->mdev, log_max_ra_req_qp))) { mlx5_ib_dbg(dev, "invalid max_dest_rd_atomic value %d\n", attr->max_dest_rd_atomic); goto out; } if (cur_state == new_state && cur_state == IB_QPS_RESET) { err = 0; goto out; } err = __mlx5_ib_modify_qp(ibqp, attr, attr_mask, cur_state, new_state); out: mutex_unlock(&qp->mutex); return err; } static int mlx5_wq_overflow(struct mlx5_ib_wq *wq, int nreq, struct ib_cq *ib_cq) { struct mlx5_ib_cq *cq; unsigned cur; cur = wq->head - wq->tail; if (likely(cur + nreq < wq->max_post)) return 0; cq = to_mcq(ib_cq); spin_lock(&cq->lock); cur = wq->head - wq->tail; spin_unlock(&cq->lock); return cur + nreq >= wq->max_post; } static __always_inline void set_raddr_seg(struct mlx5_wqe_raddr_seg *rseg, u64 remote_addr, u32 rkey) { rseg->raddr = cpu_to_be64(remote_addr); rseg->rkey = cpu_to_be32(rkey); rseg->reserved = 0; } static void *set_eth_seg(struct mlx5_wqe_eth_seg *eseg, const struct ib_send_wr *wr, void *qend, struct mlx5_ib_qp *qp, int *size) { void *seg = eseg; memset(eseg, 0, sizeof(struct mlx5_wqe_eth_seg)); if (wr->send_flags & IB_SEND_IP_CSUM) eseg->cs_flags = MLX5_ETH_WQE_L3_CSUM | MLX5_ETH_WQE_L4_CSUM; seg += sizeof(struct mlx5_wqe_eth_seg); *size += sizeof(struct mlx5_wqe_eth_seg) / 16; if (wr->opcode == IB_WR_LSO) { struct ib_ud_wr *ud_wr = container_of(wr, struct ib_ud_wr, wr); int size_of_inl_hdr_start = sizeof(eseg->inline_hdr_start); u64 left, leftlen, copysz; void *pdata = ud_wr->header; left = ud_wr->hlen; eseg->mss = cpu_to_be16(ud_wr->mss); eseg->inline_hdr_sz = cpu_to_be16(left); /* * check if there is space till the end of queue, if yes, * copy all in one shot, otherwise copy till the end of queue, * rollback and than the copy the left */ leftlen = qend - (void *)eseg->inline_hdr_start; copysz = min_t(u64, leftlen, left); memcpy(seg - size_of_inl_hdr_start, pdata, copysz); if (likely(copysz > size_of_inl_hdr_start)) { seg += ALIGN(copysz - size_of_inl_hdr_start, 16); *size += ALIGN(copysz - size_of_inl_hdr_start, 16) / 16; } if (unlikely(copysz < left)) { /* the last wqe in the queue */ seg = mlx5_get_send_wqe(qp, 0); left -= copysz; pdata += copysz; memcpy(seg, pdata, left); seg += ALIGN(left, 16); *size += ALIGN(left, 16) / 16; } } return seg; } static void set_datagram_seg(struct mlx5_wqe_datagram_seg *dseg, const struct ib_send_wr *wr) { memcpy(&dseg->av, &to_mah(ud_wr(wr)->ah)->av, sizeof(struct mlx5_av)); dseg->av.dqp_dct = cpu_to_be32(ud_wr(wr)->remote_qpn | MLX5_EXTENDED_UD_AV); dseg->av.key.qkey.qkey = cpu_to_be32(ud_wr(wr)->remote_qkey); } static void set_data_ptr_seg(struct mlx5_wqe_data_seg *dseg, struct ib_sge *sg) { dseg->byte_count = cpu_to_be32(sg->length); dseg->lkey = cpu_to_be32(sg->lkey); dseg->addr = cpu_to_be64(sg->addr); } static __be16 get_klm_octo(int npages) { return cpu_to_be16(ALIGN(npages, 8) / 2); } static __be64 frwr_mkey_mask(void) { u64 result; result = MLX5_MKEY_MASK_LEN | MLX5_MKEY_MASK_PAGE_SIZE | MLX5_MKEY_MASK_START_ADDR | MLX5_MKEY_MASK_EN_RINVAL | MLX5_MKEY_MASK_KEY | MLX5_MKEY_MASK_LR | MLX5_MKEY_MASK_LW | MLX5_MKEY_MASK_RR | MLX5_MKEY_MASK_RW | MLX5_MKEY_MASK_A | MLX5_MKEY_MASK_SMALL_FENCE | MLX5_MKEY_MASK_FREE; return cpu_to_be64(result); } static __be64 sig_mkey_mask(void) { u64 result; result = MLX5_MKEY_MASK_LEN | MLX5_MKEY_MASK_PAGE_SIZE | MLX5_MKEY_MASK_START_ADDR | MLX5_MKEY_MASK_EN_SIGERR | MLX5_MKEY_MASK_EN_RINVAL | MLX5_MKEY_MASK_KEY | MLX5_MKEY_MASK_LR | MLX5_MKEY_MASK_LW | MLX5_MKEY_MASK_RR | MLX5_MKEY_MASK_RW | MLX5_MKEY_MASK_SMALL_FENCE | MLX5_MKEY_MASK_FREE | MLX5_MKEY_MASK_BSF_EN; return cpu_to_be64(result); } static void set_reg_umr_seg(struct mlx5_wqe_umr_ctrl_seg *umr, struct mlx5_ib_mr *mr) { int ndescs = mr->ndescs; memset(umr, 0, sizeof(*umr)); if (mr->access_mode == MLX5_ACCESS_MODE_KLM) /* KLMs take twice the size of MTTs */ ndescs *= 2; umr->flags = MLX5_UMR_CHECK_NOT_FREE; umr->klm_octowords = get_klm_octo(ndescs); umr->mkey_mask = frwr_mkey_mask(); } static void set_linv_umr_seg(struct mlx5_wqe_umr_ctrl_seg *umr) { memset(umr, 0, sizeof(*umr)); umr->mkey_mask = cpu_to_be64(MLX5_MKEY_MASK_FREE); umr->flags = 1 << 7; } static __be64 get_umr_reg_mr_mask(void) { u64 result; result = MLX5_MKEY_MASK_LEN | MLX5_MKEY_MASK_PAGE_SIZE | MLX5_MKEY_MASK_START_ADDR | MLX5_MKEY_MASK_PD | MLX5_MKEY_MASK_LR | MLX5_MKEY_MASK_LW | MLX5_MKEY_MASK_KEY | MLX5_MKEY_MASK_RR | MLX5_MKEY_MASK_RW | MLX5_MKEY_MASK_A | MLX5_MKEY_MASK_FREE; return cpu_to_be64(result); } static __be64 get_umr_unreg_mr_mask(void) { u64 result; result = MLX5_MKEY_MASK_FREE; return cpu_to_be64(result); } static __be64 get_umr_update_mtt_mask(void) { u64 result; result = MLX5_MKEY_MASK_FREE; return cpu_to_be64(result); } static __be64 get_umr_update_translation_mask(void) { u64 result; result = MLX5_MKEY_MASK_LEN | MLX5_MKEY_MASK_PAGE_SIZE | MLX5_MKEY_MASK_START_ADDR | MLX5_MKEY_MASK_KEY | MLX5_MKEY_MASK_FREE; return cpu_to_be64(result); } static __be64 get_umr_update_access_mask(void) { u64 result; result = MLX5_MKEY_MASK_LW | MLX5_MKEY_MASK_RR | MLX5_MKEY_MASK_RW | MLX5_MKEY_MASK_A | MLX5_MKEY_MASK_KEY | MLX5_MKEY_MASK_FREE; return cpu_to_be64(result); } static __be64 get_umr_update_pd_mask(void) { u64 result; result = MLX5_MKEY_MASK_PD | MLX5_MKEY_MASK_KEY | MLX5_MKEY_MASK_FREE; return cpu_to_be64(result); } static void set_reg_umr_segment(struct mlx5_wqe_umr_ctrl_seg *umr, const struct ib_send_wr *wr) { const struct mlx5_umr_wr *umrwr = umr_wr(wr); memset(umr, 0, sizeof(*umr)); if (wr->send_flags & MLX5_IB_SEND_UMR_FAIL_IF_FREE) umr->flags = MLX5_UMR_CHECK_FREE; /* fail if free */ else umr->flags = MLX5_UMR_CHECK_NOT_FREE; /* fail if not free */ if (!(wr->send_flags & MLX5_IB_SEND_UMR_UNREG)) { umr->klm_octowords = get_klm_octo(umrwr->npages); if (wr->send_flags & MLX5_IB_SEND_UMR_UPDATE_MTT) { umr->mkey_mask = get_umr_update_mtt_mask(); umr->bsf_octowords = get_klm_octo(umrwr->target.offset); umr->flags |= MLX5_UMR_TRANSLATION_OFFSET_EN; } if (wr->send_flags & MLX5_IB_SEND_UMR_UPDATE_TRANSLATION) umr->mkey_mask |= get_umr_update_translation_mask(); if (wr->send_flags & MLX5_IB_SEND_UMR_UPDATE_ACCESS) umr->mkey_mask |= get_umr_update_access_mask(); if (wr->send_flags & MLX5_IB_SEND_UMR_UPDATE_PD) umr->mkey_mask |= get_umr_update_pd_mask(); if (!umr->mkey_mask) umr->mkey_mask = get_umr_reg_mr_mask(); } else { umr->mkey_mask = get_umr_unreg_mr_mask(); } if (!wr->num_sge) umr->flags |= MLX5_UMR_INLINE; } static u8 get_umr_flags(int acc) { return (acc & IB_ACCESS_REMOTE_ATOMIC ? MLX5_PERM_ATOMIC : 0) | (acc & IB_ACCESS_REMOTE_WRITE ? MLX5_PERM_REMOTE_WRITE : 0) | (acc & IB_ACCESS_REMOTE_READ ? MLX5_PERM_REMOTE_READ : 0) | (acc & IB_ACCESS_LOCAL_WRITE ? MLX5_PERM_LOCAL_WRITE : 0) | MLX5_PERM_LOCAL_READ | MLX5_PERM_UMR_EN; } static void set_reg_mkey_seg(struct mlx5_mkey_seg *seg, struct mlx5_ib_mr *mr, u32 key, int access) { int ndescs = ALIGN(mr->ndescs, 8) >> 1; memset(seg, 0, sizeof(*seg)); if (mr->access_mode == MLX5_ACCESS_MODE_MTT) seg->log2_page_size = ilog2(mr->ibmr.page_size); else if (mr->access_mode == MLX5_ACCESS_MODE_KLM) /* KLMs take twice the size of MTTs */ ndescs *= 2; seg->flags = get_umr_flags(access) | mr->access_mode; seg->qpn_mkey7_0 = cpu_to_be32((key & 0xff) | 0xffffff00); seg->flags_pd = cpu_to_be32(MLX5_MKEY_REMOTE_INVAL); seg->start_addr = cpu_to_be64(mr->ibmr.iova); seg->len = cpu_to_be64(mr->ibmr.length); seg->xlt_oct_size = cpu_to_be32(ndescs); } static void set_linv_mkey_seg(struct mlx5_mkey_seg *seg) { memset(seg, 0, sizeof(*seg)); seg->status = MLX5_MKEY_STATUS_FREE; } static void set_reg_mkey_segment(struct mlx5_mkey_seg *seg, const struct ib_send_wr *wr) { const struct mlx5_umr_wr *umrwr = umr_wr(wr); memset(seg, 0, sizeof(*seg)); if (wr->send_flags & MLX5_IB_SEND_UMR_UNREG) { seg->status = MLX5_MKEY_STATUS_FREE; return; } seg->flags = convert_access(umrwr->access_flags); if (!(wr->send_flags & MLX5_IB_SEND_UMR_UPDATE_MTT)) { if (umrwr->pd) seg->flags_pd = cpu_to_be32(to_mpd(umrwr->pd)->pdn); seg->start_addr = cpu_to_be64(umrwr->target.virt_addr); } seg->len = cpu_to_be64(umrwr->length); seg->log2_page_size = umrwr->page_shift; seg->qpn_mkey7_0 = cpu_to_be32(0xffffff00 | mlx5_mkey_variant(umrwr->mkey)); } static void set_reg_data_seg(struct mlx5_wqe_data_seg *dseg, struct mlx5_ib_mr *mr, struct mlx5_ib_pd *pd) { int bcount = mr->desc_size * mr->ndescs; dseg->addr = cpu_to_be64(mr->desc_map); dseg->byte_count = cpu_to_be32(ALIGN(bcount, 64)); dseg->lkey = cpu_to_be32(pd->ibpd.local_dma_lkey); } static __be32 send_ieth(const struct ib_send_wr *wr) { switch (wr->opcode) { case IB_WR_SEND_WITH_IMM: case IB_WR_RDMA_WRITE_WITH_IMM: return wr->ex.imm_data; case IB_WR_SEND_WITH_INV: return cpu_to_be32(wr->ex.invalidate_rkey); default: return 0; } } static u8 calc_sig(void *wqe, int size) { u8 *p = wqe; u8 res = 0; int i; for (i = 0; i < size; i++) res ^= p[i]; return ~res; } static u8 wq_sig(void *wqe) { return calc_sig(wqe, (*((u8 *)wqe + 8) & 0x3f) << 4); } static int set_data_inl_seg(struct mlx5_ib_qp *qp, const struct ib_send_wr *wr, void *wqe, int *sz) { struct mlx5_wqe_inline_seg *seg; void *qend = qp->sq.qend; void *addr; int inl = 0; int copy; int len; int i; seg = wqe; wqe += sizeof(*seg); for (i = 0; i < wr->num_sge; i++) { addr = (void *)(unsigned long)(wr->sg_list[i].addr); len = wr->sg_list[i].length; inl += len; if (unlikely(inl > qp->max_inline_data)) return -ENOMEM; if (unlikely(wqe + len > qend)) { copy = qend - wqe; memcpy(wqe, addr, copy); addr += copy; len -= copy; wqe = mlx5_get_send_wqe(qp, 0); } memcpy(wqe, addr, len); wqe += len; } seg->byte_count = cpu_to_be32(inl | MLX5_INLINE_SEG); *sz = ALIGN(inl + sizeof(seg->byte_count), 16) / 16; return 0; } static u16 prot_field_size(enum ib_signature_type type) { switch (type) { case IB_SIG_TYPE_T10_DIF: return MLX5_DIF_SIZE; default: return 0; } } static u8 bs_selector(int block_size) { switch (block_size) { case 512: return 0x1; case 520: return 0x2; case 4096: return 0x3; case 4160: return 0x4; case 1073741824: return 0x5; default: return 0; } } static void mlx5_fill_inl_bsf(struct ib_sig_domain *domain, struct mlx5_bsf_inl *inl) { /* Valid inline section and allow BSF refresh */ inl->vld_refresh = cpu_to_be16(MLX5_BSF_INL_VALID | MLX5_BSF_REFRESH_DIF); inl->dif_apptag = cpu_to_be16(domain->sig.dif.app_tag); inl->dif_reftag = cpu_to_be32(domain->sig.dif.ref_tag); /* repeating block */ inl->rp_inv_seed = MLX5_BSF_REPEAT_BLOCK; inl->sig_type = domain->sig.dif.bg_type == IB_T10DIF_CRC ? MLX5_DIF_CRC : MLX5_DIF_IPCS; if (domain->sig.dif.ref_remap) inl->dif_inc_ref_guard_check |= MLX5_BSF_INC_REFTAG; if (domain->sig.dif.app_escape) { if (domain->sig.dif.ref_escape) inl->dif_inc_ref_guard_check |= MLX5_BSF_APPREF_ESCAPE; else inl->dif_inc_ref_guard_check |= MLX5_BSF_APPTAG_ESCAPE; } inl->dif_app_bitmask_check = cpu_to_be16(domain->sig.dif.apptag_check_mask); } static int mlx5_set_bsf(struct ib_mr *sig_mr, struct ib_sig_attrs *sig_attrs, struct mlx5_bsf *bsf, u32 data_size) { struct mlx5_core_sig_ctx *msig = to_mmr(sig_mr)->sig; struct mlx5_bsf_basic *basic = &bsf->basic; struct ib_sig_domain *mem = &sig_attrs->mem; struct ib_sig_domain *wire = &sig_attrs->wire; memset(bsf, 0, sizeof(*bsf)); /* Basic + Extended + Inline */ basic->bsf_size_sbs = 1 << 7; /* Input domain check byte mask */ basic->check_byte_mask = sig_attrs->check_mask; basic->raw_data_size = cpu_to_be32(data_size); /* Memory domain */ switch (sig_attrs->mem.sig_type) { case IB_SIG_TYPE_NONE: break; case IB_SIG_TYPE_T10_DIF: basic->mem.bs_selector = bs_selector(mem->sig.dif.pi_interval); basic->m_bfs_psv = cpu_to_be32(msig->psv_memory.psv_idx); mlx5_fill_inl_bsf(mem, &bsf->m_inl); break; default: return -EINVAL; } /* Wire domain */ switch (sig_attrs->wire.sig_type) { case IB_SIG_TYPE_NONE: break; case IB_SIG_TYPE_T10_DIF: if (mem->sig.dif.pi_interval == wire->sig.dif.pi_interval && mem->sig_type == wire->sig_type) { /* Same block structure */ basic->bsf_size_sbs |= 1 << 4; if (mem->sig.dif.bg_type == wire->sig.dif.bg_type) basic->wire.copy_byte_mask |= MLX5_CPY_GRD_MASK; if (mem->sig.dif.app_tag == wire->sig.dif.app_tag) basic->wire.copy_byte_mask |= MLX5_CPY_APP_MASK; if (mem->sig.dif.ref_tag == wire->sig.dif.ref_tag) basic->wire.copy_byte_mask |= MLX5_CPY_REF_MASK; } else basic->wire.bs_selector = bs_selector(wire->sig.dif.pi_interval); basic->w_bfs_psv = cpu_to_be32(msig->psv_wire.psv_idx); mlx5_fill_inl_bsf(wire, &bsf->w_inl); break; default: return -EINVAL; } return 0; } static int set_sig_data_segment(const struct ib_sig_handover_wr *wr, struct mlx5_ib_qp *qp, void **seg, int *size) { struct ib_sig_attrs *sig_attrs = wr->sig_attrs; struct ib_mr *sig_mr = wr->sig_mr; struct mlx5_bsf *bsf; u32 data_len = wr->wr.sg_list->length; u32 data_key = wr->wr.sg_list->lkey; u64 data_va = wr->wr.sg_list->addr; int ret; int wqe_size; if (!wr->prot || (data_key == wr->prot->lkey && data_va == wr->prot->addr && data_len == wr->prot->length)) { /** * Source domain doesn't contain signature information * or data and protection are interleaved in memory. * So need construct: * ------------------ * | data_klm | * ------------------ * | BSF | * ------------------ **/ struct mlx5_klm *data_klm = *seg; data_klm->bcount = cpu_to_be32(data_len); data_klm->key = cpu_to_be32(data_key); data_klm->va = cpu_to_be64(data_va); wqe_size = ALIGN(sizeof(*data_klm), 64); } else { /** * Source domain contains signature information * So need construct a strided block format: * --------------------------- * | stride_block_ctrl | * --------------------------- * | data_klm | * --------------------------- * | prot_klm | * --------------------------- * | BSF | * --------------------------- **/ struct mlx5_stride_block_ctrl_seg *sblock_ctrl; struct mlx5_stride_block_entry *data_sentry; struct mlx5_stride_block_entry *prot_sentry; u32 prot_key = wr->prot->lkey; u64 prot_va = wr->prot->addr; u16 block_size = sig_attrs->mem.sig.dif.pi_interval; int prot_size; sblock_ctrl = *seg; data_sentry = (void *)sblock_ctrl + sizeof(*sblock_ctrl); prot_sentry = (void *)data_sentry + sizeof(*data_sentry); prot_size = prot_field_size(sig_attrs->mem.sig_type); if (!prot_size) { pr_err("Bad block size given: %u\n", block_size); return -EINVAL; } sblock_ctrl->bcount_per_cycle = cpu_to_be32(block_size + prot_size); sblock_ctrl->op = cpu_to_be32(MLX5_STRIDE_BLOCK_OP); sblock_ctrl->repeat_count = cpu_to_be32(data_len / block_size); sblock_ctrl->num_entries = cpu_to_be16(2); data_sentry->bcount = cpu_to_be16(block_size); data_sentry->key = cpu_to_be32(data_key); data_sentry->va = cpu_to_be64(data_va); data_sentry->stride = cpu_to_be16(block_size); prot_sentry->bcount = cpu_to_be16(prot_size); prot_sentry->key = cpu_to_be32(prot_key); prot_sentry->va = cpu_to_be64(prot_va); prot_sentry->stride = cpu_to_be16(prot_size); wqe_size = ALIGN(sizeof(*sblock_ctrl) + sizeof(*data_sentry) + sizeof(*prot_sentry), 64); } *seg += wqe_size; *size += wqe_size / 16; if (unlikely((*seg == qp->sq.qend))) *seg = mlx5_get_send_wqe(qp, 0); bsf = *seg; ret = mlx5_set_bsf(sig_mr, sig_attrs, bsf, data_len); if (ret) return -EINVAL; *seg += sizeof(*bsf); *size += sizeof(*bsf) / 16; if (unlikely((*seg == qp->sq.qend))) *seg = mlx5_get_send_wqe(qp, 0); return 0; } static void set_sig_mkey_segment(struct mlx5_mkey_seg *seg, const struct ib_sig_handover_wr *wr, u32 nelements, u32 length, u32 pdn) { struct ib_mr *sig_mr = wr->sig_mr; u32 sig_key = sig_mr->rkey; u8 sigerr = to_mmr(sig_mr)->sig->sigerr_count & 1; memset(seg, 0, sizeof(*seg)); seg->flags = get_umr_flags(wr->access_flags) | MLX5_ACCESS_MODE_KLM; seg->qpn_mkey7_0 = cpu_to_be32((sig_key & 0xff) | 0xffffff00); seg->flags_pd = cpu_to_be32(MLX5_MKEY_REMOTE_INVAL | sigerr << 26 | MLX5_MKEY_BSF_EN | pdn); seg->len = cpu_to_be64(length); seg->xlt_oct_size = cpu_to_be32(be16_to_cpu(get_klm_octo(nelements))); seg->bsfs_octo_size = cpu_to_be32(MLX5_MKEY_BSF_OCTO_SIZE); } static void set_sig_umr_segment(struct mlx5_wqe_umr_ctrl_seg *umr, u32 nelements) { memset(umr, 0, sizeof(*umr)); umr->flags = MLX5_FLAGS_INLINE | MLX5_FLAGS_CHECK_FREE; umr->klm_octowords = get_klm_octo(nelements); umr->bsf_octowords = cpu_to_be16(MLX5_MKEY_BSF_OCTO_SIZE); umr->mkey_mask = sig_mkey_mask(); } static int set_sig_umr_wr(const struct ib_send_wr *send_wr, struct mlx5_ib_qp *qp, void **seg, int *size) { const struct ib_sig_handover_wr *wr = sig_handover_wr(send_wr); struct mlx5_ib_mr *sig_mr = to_mmr(wr->sig_mr); u32 pdn = get_pd(qp)->pdn; u32 klm_oct_size; int region_len, ret; if (unlikely(wr->wr.num_sge != 1) || unlikely(wr->access_flags & IB_ACCESS_REMOTE_ATOMIC) || unlikely(!sig_mr->sig) || unlikely(!qp->signature_en) || unlikely(!sig_mr->sig->sig_status_checked)) return -EINVAL; /* length of the protected region, data + protection */ region_len = wr->wr.sg_list->length; if (wr->prot && (wr->prot->lkey != wr->wr.sg_list->lkey || wr->prot->addr != wr->wr.sg_list->addr || wr->prot->length != wr->wr.sg_list->length)) region_len += wr->prot->length; /** * KLM octoword size - if protection was provided * then we use strided block format (3 octowords), * else we use single KLM (1 octoword) **/ klm_oct_size = wr->prot ? 3 : 1; set_sig_umr_segment(*seg, klm_oct_size); *seg += sizeof(struct mlx5_wqe_umr_ctrl_seg); *size += sizeof(struct mlx5_wqe_umr_ctrl_seg) / 16; if (unlikely((*seg == qp->sq.qend))) *seg = mlx5_get_send_wqe(qp, 0); set_sig_mkey_segment(*seg, wr, klm_oct_size, region_len, pdn); *seg += sizeof(struct mlx5_mkey_seg); *size += sizeof(struct mlx5_mkey_seg) / 16; if (unlikely((*seg == qp->sq.qend))) *seg = mlx5_get_send_wqe(qp, 0); ret = set_sig_data_segment(wr, qp, seg, size); if (ret) return ret; sig_mr->sig->sig_status_checked = false; return 0; } static int set_psv_wr(struct ib_sig_domain *domain, u32 psv_idx, void **seg, int *size) { struct mlx5_seg_set_psv *psv_seg = *seg; memset(psv_seg, 0, sizeof(*psv_seg)); psv_seg->psv_num = cpu_to_be32(psv_idx); switch (domain->sig_type) { case IB_SIG_TYPE_NONE: break; case IB_SIG_TYPE_T10_DIF: psv_seg->transient_sig = cpu_to_be32(domain->sig.dif.bg << 16 | domain->sig.dif.app_tag); psv_seg->ref_tag = cpu_to_be32(domain->sig.dif.ref_tag); break; default: pr_err("Bad signature type given.\n"); return 1; } *seg += sizeof(*psv_seg); *size += sizeof(*psv_seg) / 16; return 0; } static int set_reg_wr(struct mlx5_ib_qp *qp, const struct ib_reg_wr *wr, void **seg, int *size) { struct mlx5_ib_mr *mr = to_mmr(wr->mr); struct mlx5_ib_pd *pd = to_mpd(qp->ibqp.pd); if (unlikely(wr->wr.send_flags & IB_SEND_INLINE)) { mlx5_ib_warn(to_mdev(qp->ibqp.device), "Invalid IB_SEND_INLINE send flag\n"); return -EINVAL; } set_reg_umr_seg(*seg, mr); *seg += sizeof(struct mlx5_wqe_umr_ctrl_seg); *size += sizeof(struct mlx5_wqe_umr_ctrl_seg) / 16; if (unlikely((*seg == qp->sq.qend))) *seg = mlx5_get_send_wqe(qp, 0); set_reg_mkey_seg(*seg, mr, wr->key, wr->access); *seg += sizeof(struct mlx5_mkey_seg); *size += sizeof(struct mlx5_mkey_seg) / 16; if (unlikely((*seg == qp->sq.qend))) *seg = mlx5_get_send_wqe(qp, 0); set_reg_data_seg(*seg, mr, pd); *seg += sizeof(struct mlx5_wqe_data_seg); *size += (sizeof(struct mlx5_wqe_data_seg) / 16); return 0; } static void set_linv_wr(struct mlx5_ib_qp *qp, void **seg, int *size) { set_linv_umr_seg(*seg); *seg += sizeof(struct mlx5_wqe_umr_ctrl_seg); *size += sizeof(struct mlx5_wqe_umr_ctrl_seg) / 16; if (unlikely((*seg == qp->sq.qend))) *seg = mlx5_get_send_wqe(qp, 0); set_linv_mkey_seg(*seg); *seg += sizeof(struct mlx5_mkey_seg); *size += sizeof(struct mlx5_mkey_seg) / 16; if (unlikely((*seg == qp->sq.qend))) *seg = mlx5_get_send_wqe(qp, 0); } static void dump_wqe(struct mlx5_ib_qp *qp, int idx, int size_16) { __be32 *p = NULL; int tidx = idx; int i, j; pr_debug("dump wqe at %p\n", mlx5_get_send_wqe(qp, tidx)); for (i = 0, j = 0; i < size_16 * 4; i += 4, j += 4) { if ((i & 0xf) == 0) { void *buf = mlx5_get_send_wqe(qp, tidx); tidx = (tidx + 1) & (qp->sq.wqe_cnt - 1); p = buf; j = 0; } pr_debug("%08x %08x %08x %08x\n", be32_to_cpu(p[j]), be32_to_cpu(p[j + 1]), be32_to_cpu(p[j + 2]), be32_to_cpu(p[j + 3])); } } static u8 get_fence(u8 fence, const struct ib_send_wr *wr) { if (unlikely(wr->opcode == IB_WR_LOCAL_INV && wr->send_flags & IB_SEND_FENCE)) return MLX5_FENCE_MODE_STRONG_ORDERING; if (unlikely(fence)) { if (wr->send_flags & IB_SEND_FENCE) return MLX5_FENCE_MODE_SMALL_AND_FENCE; else return fence; } else if (unlikely(wr->send_flags & IB_SEND_FENCE)) { return MLX5_FENCE_MODE_FENCE; } return 0; } static int __begin_wqe(struct mlx5_ib_qp *qp, void **seg, struct mlx5_wqe_ctrl_seg **ctrl, const struct ib_send_wr *wr, unsigned *idx, int *size, int nreq, bool send_signaled, bool solicited) { if (unlikely(mlx5_wq_overflow(&qp->sq, nreq, qp->ibqp.send_cq))) return -ENOMEM; *idx = qp->sq.cur_post & (qp->sq.wqe_cnt - 1); *seg = mlx5_get_send_wqe(qp, *idx); *ctrl = *seg; *(uint32_t *)(*seg + 8) = 0; (*ctrl)->imm = send_ieth(wr); (*ctrl)->fm_ce_se = qp->sq_signal_bits | (send_signaled ? MLX5_WQE_CTRL_CQ_UPDATE : 0) | (solicited ? MLX5_WQE_CTRL_SOLICITED : 0); *seg += sizeof(**ctrl); *size = sizeof(**ctrl) / 16; return 0; } static int begin_wqe(struct mlx5_ib_qp *qp, void **seg, struct mlx5_wqe_ctrl_seg **ctrl, const struct ib_send_wr *wr, unsigned *idx, int *size, int nreq) { return __begin_wqe(qp, seg, ctrl, wr, idx, size, nreq, wr->send_flags & IB_SEND_SIGNALED, wr->send_flags & IB_SEND_SOLICITED); } static void finish_wqe(struct mlx5_ib_qp *qp, struct mlx5_wqe_ctrl_seg *ctrl, u8 size, unsigned idx, u64 wr_id, int nreq, u8 fence, u8 next_fence, u32 mlx5_opcode) { u8 opmod = 0; ctrl->opmod_idx_opcode = cpu_to_be32(((u32)(qp->sq.cur_post) << 8) | mlx5_opcode | ((u32)opmod << 24)); ctrl->qpn_ds = cpu_to_be32(size | (qp->trans_qp.base.mqp.qpn << 8)); ctrl->fm_ce_se |= fence; qp->fm_cache = next_fence; if (unlikely(qp->wq_sig)) ctrl->signature = wq_sig(ctrl); qp->sq.wrid[idx] = wr_id; qp->sq.w_list[idx].opcode = mlx5_opcode; qp->sq.wqe_head[idx] = qp->sq.head + nreq; qp->sq.cur_post += DIV_ROUND_UP(size * 16, MLX5_SEND_WQE_BB); qp->sq.w_list[idx].next = qp->sq.cur_post; } int mlx5_ib_post_send(struct ib_qp *ibqp, const struct ib_send_wr *wr, const struct ib_send_wr **bad_wr) { struct mlx5_wqe_ctrl_seg *ctrl = NULL; /* compiler warning */ struct mlx5_ib_dev *dev = to_mdev(ibqp->device); struct mlx5_core_dev *mdev = dev->mdev; struct mlx5_ib_qp *qp; struct mlx5_ib_mr *mr; struct mlx5_wqe_data_seg *dpseg; struct mlx5_wqe_xrc_seg *xrc; struct mlx5_bf *bf; int uninitialized_var(size); void *qend; unsigned long flags; unsigned idx; int err = 0; int inl = 0; int num_sge; void *seg; int nreq; int i; u8 next_fence = 0; u8 fence; if (unlikely(ibqp->qp_type == IB_QPT_GSI)) return mlx5_ib_gsi_post_send(ibqp, wr, bad_wr); qp = to_mqp(ibqp); bf = &qp->bf; qend = qp->sq.qend; spin_lock_irqsave(&qp->sq.lock, flags); if (mdev->state == MLX5_DEVICE_STATE_INTERNAL_ERROR) { err = -EIO; *bad_wr = wr; nreq = 0; goto out; } for (nreq = 0; wr; nreq++, wr = wr->next) { if (unlikely(wr->opcode < 0 || wr->opcode >= ARRAY_SIZE(mlx5_ib_opcode))) { mlx5_ib_warn(dev, "\n"); err = -EINVAL; *bad_wr = wr; goto out; } fence = qp->fm_cache; num_sge = wr->num_sge; if (unlikely(num_sge > qp->sq.max_gs)) { mlx5_ib_warn(dev, "\n"); err = -EINVAL; *bad_wr = wr; goto out; } err = begin_wqe(qp, &seg, &ctrl, wr, &idx, &size, nreq); if (err) { mlx5_ib_warn(dev, "\n"); err = -ENOMEM; *bad_wr = wr; goto out; } switch (ibqp->qp_type) { case IB_QPT_XRC_INI: xrc = seg; seg += sizeof(*xrc); size += sizeof(*xrc) / 16; /* fall through */ case IB_QPT_RC: switch (wr->opcode) { case IB_WR_RDMA_READ: case IB_WR_RDMA_WRITE: case IB_WR_RDMA_WRITE_WITH_IMM: set_raddr_seg(seg, rdma_wr(wr)->remote_addr, rdma_wr(wr)->rkey); seg += sizeof(struct mlx5_wqe_raddr_seg); size += sizeof(struct mlx5_wqe_raddr_seg) / 16; break; case IB_WR_ATOMIC_CMP_AND_SWP: case IB_WR_ATOMIC_FETCH_AND_ADD: case IB_WR_MASKED_ATOMIC_CMP_AND_SWP: mlx5_ib_warn(dev, "Atomic operations are not supported yet\n"); err = -ENOSYS; *bad_wr = wr; goto out; case IB_WR_LOCAL_INV: next_fence = MLX5_FENCE_MODE_INITIATOR_SMALL; qp->sq.wr_data[idx] = IB_WR_LOCAL_INV; ctrl->imm = cpu_to_be32(wr->ex.invalidate_rkey); set_linv_wr(qp, &seg, &size); num_sge = 0; break; case IB_WR_REG_MR: next_fence = MLX5_FENCE_MODE_INITIATOR_SMALL; qp->sq.wr_data[idx] = IB_WR_REG_MR; ctrl->imm = cpu_to_be32(reg_wr(wr)->key); err = set_reg_wr(qp, reg_wr(wr), &seg, &size); if (err) { *bad_wr = wr; goto out; } num_sge = 0; break; case IB_WR_REG_SIG_MR: qp->sq.wr_data[idx] = IB_WR_REG_SIG_MR; mr = to_mmr(sig_handover_wr(wr)->sig_mr); ctrl->imm = cpu_to_be32(mr->ibmr.rkey); err = set_sig_umr_wr(wr, qp, &seg, &size); if (err) { mlx5_ib_warn(dev, "\n"); *bad_wr = wr; goto out; } finish_wqe(qp, ctrl, size, idx, wr->wr_id, nreq, get_fence(fence, wr), next_fence, MLX5_OPCODE_UMR); /* * SET_PSV WQEs are not signaled and solicited * on error */ err = __begin_wqe(qp, &seg, &ctrl, wr, &idx, &size, nreq, false, true); if (err) { mlx5_ib_warn(dev, "\n"); err = -ENOMEM; *bad_wr = wr; goto out; } err = set_psv_wr(&sig_handover_wr(wr)->sig_attrs->mem, mr->sig->psv_memory.psv_idx, &seg, &size); if (err) { mlx5_ib_warn(dev, "\n"); *bad_wr = wr; goto out; } finish_wqe(qp, ctrl, size, idx, wr->wr_id, nreq, get_fence(fence, wr), next_fence, MLX5_OPCODE_SET_PSV); err = __begin_wqe(qp, &seg, &ctrl, wr, &idx, &size, nreq, false, true); if (err) { mlx5_ib_warn(dev, "\n"); err = -ENOMEM; *bad_wr = wr; goto out; } next_fence = MLX5_FENCE_MODE_INITIATOR_SMALL; err = set_psv_wr(&sig_handover_wr(wr)->sig_attrs->wire, mr->sig->psv_wire.psv_idx, &seg, &size); if (err) { mlx5_ib_warn(dev, "\n"); *bad_wr = wr; goto out; } finish_wqe(qp, ctrl, size, idx, wr->wr_id, nreq, get_fence(fence, wr), next_fence, MLX5_OPCODE_SET_PSV); num_sge = 0; goto skip_psv; default: break; } break; case IB_QPT_UC: switch (wr->opcode) { case IB_WR_RDMA_WRITE: case IB_WR_RDMA_WRITE_WITH_IMM: set_raddr_seg(seg, rdma_wr(wr)->remote_addr, rdma_wr(wr)->rkey); seg += sizeof(struct mlx5_wqe_raddr_seg); size += sizeof(struct mlx5_wqe_raddr_seg) / 16; break; default: break; } break; case IB_QPT_SMI: case MLX5_IB_QPT_HW_GSI: set_datagram_seg(seg, wr); seg += sizeof(struct mlx5_wqe_datagram_seg); size += sizeof(struct mlx5_wqe_datagram_seg) / 16; if (unlikely((seg == qend))) seg = mlx5_get_send_wqe(qp, 0); break; case IB_QPT_UD: set_datagram_seg(seg, wr); seg += sizeof(struct mlx5_wqe_datagram_seg); size += sizeof(struct mlx5_wqe_datagram_seg) / 16; if (unlikely((seg == qend))) seg = mlx5_get_send_wqe(qp, 0); /* handle qp that supports ud offload */ if (qp->flags & IB_QP_CREATE_IPOIB_UD_LSO) { struct mlx5_wqe_eth_pad *pad; pad = seg; memset(pad, 0, sizeof(struct mlx5_wqe_eth_pad)); seg += sizeof(struct mlx5_wqe_eth_pad); size += sizeof(struct mlx5_wqe_eth_pad) / 16; seg = set_eth_seg(seg, wr, qend, qp, &size); if (unlikely((seg == qend))) seg = mlx5_get_send_wqe(qp, 0); } break; case MLX5_IB_QPT_REG_UMR: if (wr->opcode != MLX5_IB_WR_UMR) { err = -EINVAL; mlx5_ib_warn(dev, "bad opcode\n"); goto out; } qp->sq.wr_data[idx] = MLX5_IB_WR_UMR; ctrl->imm = cpu_to_be32(umr_wr(wr)->mkey); set_reg_umr_segment(seg, wr); seg += sizeof(struct mlx5_wqe_umr_ctrl_seg); size += sizeof(struct mlx5_wqe_umr_ctrl_seg) / 16; if (unlikely((seg == qend))) seg = mlx5_get_send_wqe(qp, 0); set_reg_mkey_segment(seg, wr); seg += sizeof(struct mlx5_mkey_seg); size += sizeof(struct mlx5_mkey_seg) / 16; if (unlikely((seg == qend))) seg = mlx5_get_send_wqe(qp, 0); break; default: break; } if (wr->send_flags & IB_SEND_INLINE && num_sge) { int uninitialized_var(sz); err = set_data_inl_seg(qp, wr, seg, &sz); if (unlikely(err)) { mlx5_ib_warn(dev, "\n"); *bad_wr = wr; goto out; } inl = 1; size += sz; } else { dpseg = seg; for (i = 0; i < num_sge; i++) { if (unlikely(dpseg == qend)) { seg = mlx5_get_send_wqe(qp, 0); dpseg = seg; } if (likely(wr->sg_list[i].length)) { set_data_ptr_seg(dpseg, wr->sg_list + i); size += sizeof(struct mlx5_wqe_data_seg) / 16; dpseg++; } } } finish_wqe(qp, ctrl, size, idx, wr->wr_id, nreq, get_fence(fence, wr), next_fence, mlx5_ib_opcode[wr->opcode]); skip_psv: if (0) dump_wqe(qp, idx, size); } out: if (likely(nreq)) { qp->sq.head += nreq; /* Make sure that descriptors are written before * updating doorbell record and ringing the doorbell */ wmb(); qp->db.db[MLX5_SND_DBR] = cpu_to_be32(qp->sq.cur_post); /* Make sure doorbell record is visible to the HCA before * we hit doorbell */ wmb(); mlx5_write64((__be32 *)ctrl, bf->bfreg->map + bf->offset, MLX5_GET_DOORBELL_LOCK(&bf->lock32)); /* Make sure doorbells don't leak out of SQ spinlock * and reach the HCA out of order. */ bf->offset ^= bf->buf_size; } spin_unlock_irqrestore(&qp->sq.lock, flags); return err; } static void set_sig_seg(struct mlx5_rwqe_sig *sig, int size) { sig->signature = calc_sig(sig, size); } int mlx5_ib_post_recv(struct ib_qp *ibqp, const struct ib_recv_wr *wr, const struct ib_recv_wr **bad_wr) { struct mlx5_ib_qp *qp = to_mqp(ibqp); struct mlx5_wqe_data_seg *scat; struct mlx5_rwqe_sig *sig; struct mlx5_ib_dev *dev = to_mdev(ibqp->device); struct mlx5_core_dev *mdev = dev->mdev; unsigned long flags; int err = 0; int nreq; int ind; int i; if (unlikely(ibqp->qp_type == IB_QPT_GSI)) return mlx5_ib_gsi_post_recv(ibqp, wr, bad_wr); spin_lock_irqsave(&qp->rq.lock, flags); if (mdev->state == MLX5_DEVICE_STATE_INTERNAL_ERROR) { err = -EIO; *bad_wr = wr; nreq = 0; goto out; } ind = qp->rq.head & (qp->rq.wqe_cnt - 1); for (nreq = 0; wr; nreq++, wr = wr->next) { if (mlx5_wq_overflow(&qp->rq, nreq, qp->ibqp.recv_cq)) { err = -ENOMEM; *bad_wr = wr; goto out; } if (unlikely(wr->num_sge > qp->rq.max_gs)) { err = -EINVAL; *bad_wr = wr; goto out; } scat = get_recv_wqe(qp, ind); if (qp->wq_sig) scat++; for (i = 0; i < wr->num_sge; i++) set_data_ptr_seg(scat + i, wr->sg_list + i); if (i < qp->rq.max_gs) { scat[i].byte_count = 0; scat[i].lkey = cpu_to_be32(MLX5_INVALID_LKEY); scat[i].addr = 0; } if (qp->wq_sig) { sig = (struct mlx5_rwqe_sig *)scat; set_sig_seg(sig, (qp->rq.max_gs + 1) << 2); } qp->rq.wrid[ind] = wr->wr_id; ind = (ind + 1) & (qp->rq.wqe_cnt - 1); } out: if (likely(nreq)) { qp->rq.head += nreq; /* Make sure that descriptors are written before * doorbell record. */ wmb(); *qp->db.db = cpu_to_be32(qp->rq.head & 0xffff); } spin_unlock_irqrestore(&qp->rq.lock, flags); return err; } static inline enum ib_qp_state to_ib_qp_state(enum mlx5_qp_state mlx5_state) { switch (mlx5_state) { case MLX5_QP_STATE_RST: return IB_QPS_RESET; case MLX5_QP_STATE_INIT: return IB_QPS_INIT; case MLX5_QP_STATE_RTR: return IB_QPS_RTR; case MLX5_QP_STATE_RTS: return IB_QPS_RTS; case MLX5_QP_STATE_SQ_DRAINING: case MLX5_QP_STATE_SQD: return IB_QPS_SQD; case MLX5_QP_STATE_SQER: return IB_QPS_SQE; case MLX5_QP_STATE_ERR: return IB_QPS_ERR; default: return -1; } } static inline enum ib_mig_state to_ib_mig_state(int mlx5_mig_state) { switch (mlx5_mig_state) { case MLX5_QP_PM_ARMED: return IB_MIG_ARMED; case MLX5_QP_PM_REARM: return IB_MIG_REARM; case MLX5_QP_PM_MIGRATED: return IB_MIG_MIGRATED; default: return -1; } } static int to_ib_qp_access_flags(int mlx5_flags) { int ib_flags = 0; if (mlx5_flags & MLX5_QP_BIT_RRE) ib_flags |= IB_ACCESS_REMOTE_READ; if (mlx5_flags & MLX5_QP_BIT_RWE) ib_flags |= IB_ACCESS_REMOTE_WRITE; if (mlx5_flags & MLX5_QP_BIT_RAE) ib_flags |= IB_ACCESS_REMOTE_ATOMIC; return ib_flags; } static void to_ib_ah_attr(struct mlx5_ib_dev *ibdev, struct ib_ah_attr *ib_ah_attr, struct mlx5_qp_path *path) { struct mlx5_core_dev *dev = ibdev->mdev; memset(ib_ah_attr, 0, sizeof(*ib_ah_attr)); ib_ah_attr->port_num = path->port; if (ib_ah_attr->port_num == 0 || ib_ah_attr->port_num > MLX5_CAP_GEN(dev, num_ports)) return; ib_ah_attr->sl = path->dci_cfi_prio_sl & 0xf; ib_ah_attr->dlid = be16_to_cpu(path->rlid); ib_ah_attr->src_path_bits = path->grh_mlid & 0x7f; ib_ah_attr->static_rate = path->static_rate ? path->static_rate - 5 : 0; ib_ah_attr->ah_flags = (path->grh_mlid & (1 << 7)) ? IB_AH_GRH : 0; if (ib_ah_attr->ah_flags) { ib_ah_attr->grh.sgid_index = path->mgid_index; ib_ah_attr->grh.hop_limit = path->hop_limit; ib_ah_attr->grh.traffic_class = (be32_to_cpu(path->tclass_flowlabel) >> 20) & 0xff; ib_ah_attr->grh.flow_label = be32_to_cpu(path->tclass_flowlabel) & 0xfffff; memcpy(ib_ah_attr->grh.dgid.raw, path->rgid, sizeof(ib_ah_attr->grh.dgid.raw)); } } static int query_raw_packet_qp_sq_state(struct mlx5_ib_dev *dev, struct mlx5_ib_sq *sq, u8 *sq_state) { void *out; void *sqc; int inlen; int err; inlen = MLX5_ST_SZ_BYTES(query_sq_out); out = mlx5_vzalloc(inlen); if (!out) return -ENOMEM; err = mlx5_core_query_sq(dev->mdev, sq->base.mqp.qpn, out); if (err) goto out; sqc = MLX5_ADDR_OF(query_sq_out, out, sq_context); *sq_state = MLX5_GET(sqc, sqc, state); sq->state = *sq_state; out: kvfree(out); return err; } static int query_raw_packet_qp_rq_state(struct mlx5_ib_dev *dev, struct mlx5_ib_rq *rq, u8 *rq_state) { void *out; void *rqc; int inlen; int err; inlen = MLX5_ST_SZ_BYTES(query_rq_out); out = mlx5_vzalloc(inlen); if (!out) return -ENOMEM; err = mlx5_core_query_rq(dev->mdev, rq->base.mqp.qpn, out); if (err) goto out; rqc = MLX5_ADDR_OF(query_rq_out, out, rq_context); *rq_state = MLX5_GET(rqc, rqc, state); rq->state = *rq_state; out: kvfree(out); return err; } static int sqrq_state_to_qp_state(u8 sq_state, u8 rq_state, struct mlx5_ib_qp *qp, u8 *qp_state) { static const u8 sqrq_trans[MLX5_RQ_NUM_STATE][MLX5_SQ_NUM_STATE] = { [MLX5_RQC_STATE_RST] = { [MLX5_SQC_STATE_RST] = IB_QPS_RESET, [MLX5_SQC_STATE_RDY] = MLX5_QP_STATE_BAD, [MLX5_SQC_STATE_ERR] = MLX5_QP_STATE_BAD, [MLX5_SQ_STATE_NA] = IB_QPS_RESET, }, [MLX5_RQC_STATE_RDY] = { [MLX5_SQC_STATE_RST] = MLX5_QP_STATE_BAD, [MLX5_SQC_STATE_RDY] = MLX5_QP_STATE, [MLX5_SQC_STATE_ERR] = IB_QPS_SQE, [MLX5_SQ_STATE_NA] = MLX5_QP_STATE, }, [MLX5_RQC_STATE_ERR] = { [MLX5_SQC_STATE_RST] = MLX5_QP_STATE_BAD, [MLX5_SQC_STATE_RDY] = MLX5_QP_STATE_BAD, [MLX5_SQC_STATE_ERR] = IB_QPS_ERR, [MLX5_SQ_STATE_NA] = IB_QPS_ERR, }, [MLX5_RQ_STATE_NA] = { [MLX5_SQC_STATE_RST] = IB_QPS_RESET, [MLX5_SQC_STATE_RDY] = MLX5_QP_STATE, [MLX5_SQC_STATE_ERR] = MLX5_QP_STATE, [MLX5_SQ_STATE_NA] = MLX5_QP_STATE_BAD, }, }; *qp_state = sqrq_trans[rq_state][sq_state]; if (*qp_state == MLX5_QP_STATE_BAD) { WARN(1, "Buggy Raw Packet QP state, SQ 0x%x state: 0x%x, RQ 0x%x state: 0x%x", qp->raw_packet_qp.sq.base.mqp.qpn, sq_state, qp->raw_packet_qp.rq.base.mqp.qpn, rq_state); return -EINVAL; } if (*qp_state == MLX5_QP_STATE) *qp_state = qp->state; return 0; } static int query_raw_packet_qp_state(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, u8 *raw_packet_qp_state) { struct mlx5_ib_raw_packet_qp *raw_packet_qp = &qp->raw_packet_qp; struct mlx5_ib_sq *sq = &raw_packet_qp->sq; struct mlx5_ib_rq *rq = &raw_packet_qp->rq; int err; u8 sq_state = MLX5_SQ_STATE_NA; u8 rq_state = MLX5_RQ_STATE_NA; if (qp->sq.wqe_cnt) { err = query_raw_packet_qp_sq_state(dev, sq, &sq_state); if (err) return err; } if (qp->rq.wqe_cnt) { err = query_raw_packet_qp_rq_state(dev, rq, &rq_state); if (err) return err; } return sqrq_state_to_qp_state(sq_state, rq_state, qp, raw_packet_qp_state); } static int query_qp_attr(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, struct ib_qp_attr *qp_attr) { int outlen = MLX5_ST_SZ_BYTES(query_qp_out); struct mlx5_qp_context *context; int mlx5_state; u32 *outb; int err = 0; outb = kzalloc(outlen, GFP_KERNEL); if (!outb) return -ENOMEM; err = mlx5_core_qp_query(dev->mdev, &qp->trans_qp.base.mqp, outb, outlen); if (err) goto out; /* FIXME: use MLX5_GET rather than mlx5_qp_context manual struct */ context = (struct mlx5_qp_context *)MLX5_ADDR_OF(query_qp_out, outb, qpc); mlx5_state = be32_to_cpu(context->flags) >> 28; qp->state = to_ib_qp_state(mlx5_state); qp_attr->path_mtu = context->mtu_msgmax >> 5; qp_attr->path_mig_state = to_ib_mig_state((be32_to_cpu(context->flags) >> 11) & 0x3); qp_attr->qkey = be32_to_cpu(context->qkey); qp_attr->rq_psn = be32_to_cpu(context->rnr_nextrecvpsn) & 0xffffff; qp_attr->sq_psn = be32_to_cpu(context->next_send_psn) & 0xffffff; qp_attr->dest_qp_num = be32_to_cpu(context->log_pg_sz_remote_qpn) & 0xffffff; qp_attr->qp_access_flags = to_ib_qp_access_flags(be32_to_cpu(context->params2)); if (qp->ibqp.qp_type == IB_QPT_RC || qp->ibqp.qp_type == IB_QPT_UC) { to_ib_ah_attr(dev, &qp_attr->ah_attr, &context->pri_path); to_ib_ah_attr(dev, &qp_attr->alt_ah_attr, &context->alt_path); qp_attr->alt_pkey_index = be16_to_cpu(context->alt_path.pkey_index); qp_attr->alt_port_num = qp_attr->alt_ah_attr.port_num; } qp_attr->pkey_index = be16_to_cpu(context->pri_path.pkey_index); qp_attr->port_num = context->pri_path.port; /* qp_attr->en_sqd_async_notify is only applicable in modify qp */ qp_attr->sq_draining = mlx5_state == MLX5_QP_STATE_SQ_DRAINING; qp_attr->max_rd_atomic = 1 << ((be32_to_cpu(context->params1) >> 21) & 0x7); qp_attr->max_dest_rd_atomic = 1 << ((be32_to_cpu(context->params2) >> 21) & 0x7); qp_attr->min_rnr_timer = (be32_to_cpu(context->rnr_nextrecvpsn) >> 24) & 0x1f; qp_attr->timeout = context->pri_path.ackto_lt >> 3; qp_attr->retry_cnt = (be32_to_cpu(context->params1) >> 16) & 0x7; qp_attr->rnr_retry = (be32_to_cpu(context->params1) >> 13) & 0x7; qp_attr->alt_timeout = context->alt_path.ackto_lt >> 3; out: kfree(outb); return err; } int mlx5_ib_query_qp(struct ib_qp *ibqp, struct ib_qp_attr *qp_attr, int qp_attr_mask, struct ib_qp_init_attr *qp_init_attr) { struct mlx5_ib_dev *dev = to_mdev(ibqp->device); struct mlx5_ib_qp *qp = to_mqp(ibqp); int err = 0; u8 raw_packet_qp_state; if (ibqp->rwq_ind_tbl) return -ENOSYS; if (unlikely(ibqp->qp_type == IB_QPT_GSI)) return mlx5_ib_gsi_query_qp(ibqp, qp_attr, qp_attr_mask, qp_init_attr); #ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING /* * Wait for any outstanding page faults, in case the user frees memory * based upon this query's result. */ flush_workqueue(mlx5_ib_page_fault_wq); #endif mutex_lock(&qp->mutex); if (qp->ibqp.qp_type == IB_QPT_RAW_PACKET) { err = query_raw_packet_qp_state(dev, qp, &raw_packet_qp_state); if (err) goto out; qp->state = raw_packet_qp_state; qp_attr->port_num = 1; } else { err = query_qp_attr(dev, qp, qp_attr); if (err) goto out; } qp_attr->qp_state = qp->state; qp_attr->cur_qp_state = qp_attr->qp_state; qp_attr->cap.max_recv_wr = qp->rq.wqe_cnt; qp_attr->cap.max_recv_sge = qp->rq.max_gs; if (!ibqp->uobject) { qp_attr->cap.max_send_wr = qp->sq.max_post; qp_attr->cap.max_send_sge = qp->sq.max_gs; qp_init_attr->qp_context = ibqp->qp_context; } else { qp_attr->cap.max_send_wr = 0; qp_attr->cap.max_send_sge = 0; } qp_init_attr->qp_type = ibqp->qp_type; qp_init_attr->recv_cq = ibqp->recv_cq; qp_init_attr->send_cq = ibqp->send_cq; qp_init_attr->srq = ibqp->srq; qp_attr->cap.max_inline_data = qp->max_inline_data; qp_init_attr->cap = qp_attr->cap; qp_init_attr->create_flags = 0; if (qp->flags & MLX5_IB_QP_BLOCK_MULTICAST_LOOPBACK) qp_init_attr->create_flags |= IB_QP_CREATE_BLOCK_MULTICAST_LOOPBACK; if (qp->flags & MLX5_IB_QP_CROSS_CHANNEL) qp_init_attr->create_flags |= IB_QP_CREATE_CROSS_CHANNEL; if (qp->flags & MLX5_IB_QP_MANAGED_SEND) qp_init_attr->create_flags |= IB_QP_CREATE_MANAGED_SEND; if (qp->flags & MLX5_IB_QP_MANAGED_RECV) qp_init_attr->create_flags |= IB_QP_CREATE_MANAGED_RECV; if (qp->flags & MLX5_IB_QP_SQPN_QP1) qp_init_attr->create_flags |= MLX5_IB_QP_CREATE_SQPN_QP1; qp_init_attr->sq_sig_type = qp->sq_signal_bits & MLX5_WQE_CTRL_CQ_UPDATE ? IB_SIGNAL_ALL_WR : IB_SIGNAL_REQ_WR; out: mutex_unlock(&qp->mutex); return err; } struct ib_xrcd *mlx5_ib_alloc_xrcd(struct ib_device *ibdev, struct ib_ucontext *context, struct ib_udata *udata) { struct mlx5_ib_dev *dev = to_mdev(ibdev); struct mlx5_ib_xrcd *xrcd; int err; if (!MLX5_CAP_GEN(dev->mdev, xrc)) return ERR_PTR(-ENOSYS); xrcd = kmalloc(sizeof(*xrcd), GFP_KERNEL); if (!xrcd) return ERR_PTR(-ENOMEM); err = mlx5_core_xrcd_alloc(dev->mdev, &xrcd->xrcdn); if (err) { kfree(xrcd); return ERR_PTR(-ENOMEM); } return &xrcd->ibxrcd; } int mlx5_ib_dealloc_xrcd(struct ib_xrcd *xrcd) { struct mlx5_ib_dev *dev = to_mdev(xrcd->device); u32 xrcdn = to_mxrcd(xrcd)->xrcdn; int err; err = mlx5_core_xrcd_dealloc(dev->mdev, xrcdn); if (err) { mlx5_ib_warn(dev, "failed to dealloc xrcdn 0x%x\n", xrcdn); return err; } kfree(xrcd); return 0; } static void mlx5_ib_wq_event(struct mlx5_core_qp *core_qp, int type) { struct mlx5_ib_rwq *rwq = to_mibrwq(core_qp); struct mlx5_ib_dev *dev = to_mdev(rwq->ibwq.device); struct ib_event event; if (rwq->ibwq.event_handler) { event.device = rwq->ibwq.device; event.element.wq = &rwq->ibwq; switch (type) { case MLX5_EVENT_TYPE_WQ_CATAS_ERROR: event.event = IB_EVENT_WQ_FATAL; break; default: mlx5_ib_warn(dev, "Unexpected event type %d on WQ %06x\n", type, core_qp->qpn); return; } rwq->ibwq.event_handler(&event, rwq->ibwq.wq_context); } } static int create_rq(struct mlx5_ib_rwq *rwq, struct ib_pd *pd, struct ib_wq_init_attr *init_attr) { struct mlx5_ib_dev *dev; __be64 *rq_pas0; void *in; void *rqc; void *wq; int inlen; int err; dev = to_mdev(pd->device); inlen = MLX5_ST_SZ_BYTES(create_rq_in) + sizeof(u64) * rwq->rq_num_pas; in = mlx5_vzalloc(inlen); if (!in) return -ENOMEM; rqc = MLX5_ADDR_OF(create_rq_in, in, ctx); MLX5_SET(rqc, rqc, mem_rq_type, MLX5_RQC_RQ_TYPE_MEMORY_RQ_INLINE); MLX5_SET(rqc, rqc, user_index, rwq->user_index); MLX5_SET(rqc, rqc, cqn, to_mcq(init_attr->cq)->mcq.cqn); MLX5_SET(rqc, rqc, state, MLX5_RQC_STATE_RST); MLX5_SET(rqc, rqc, flush_in_error_en, 1); wq = MLX5_ADDR_OF(rqc, rqc, wq); MLX5_SET(wq, wq, wq_type, MLX5_WQ_TYPE_CYCLIC); MLX5_SET(wq, wq, end_padding_mode, MLX5_WQ_END_PAD_MODE_ALIGN); MLX5_SET(wq, wq, log_wq_stride, rwq->log_rq_stride); MLX5_SET(wq, wq, log_wq_sz, rwq->log_rq_size); MLX5_SET(wq, wq, pd, to_mpd(pd)->pdn); MLX5_SET(wq, wq, page_offset, rwq->rq_page_offset); MLX5_SET(wq, wq, log_wq_pg_sz, rwq->log_page_size); MLX5_SET(wq, wq, wq_signature, rwq->wq_sig); MLX5_SET64(wq, wq, dbr_addr, rwq->db.dma); rq_pas0 = (__be64 *)MLX5_ADDR_OF(wq, wq, pas); mlx5_ib_populate_pas(dev, rwq->umem, rwq->page_shift, rq_pas0, 0); err = mlx5_core_create_rq_tracked(dev->mdev, in, inlen, &rwq->core_qp); kvfree(in); return err; } static int set_user_rq_size(struct mlx5_ib_dev *dev, struct ib_wq_init_attr *wq_init_attr, struct mlx5_ib_create_wq *ucmd, struct mlx5_ib_rwq *rwq) { /* Sanity check RQ size before proceeding */ if (wq_init_attr->max_wr > (1 << MLX5_CAP_GEN(dev->mdev, log_max_wq_sz))) return -EINVAL; if (!ucmd->rq_wqe_count) return -EINVAL; rwq->wqe_count = ucmd->rq_wqe_count; rwq->wqe_shift = ucmd->rq_wqe_shift; rwq->buf_size = (rwq->wqe_count << rwq->wqe_shift); rwq->log_rq_stride = rwq->wqe_shift; rwq->log_rq_size = ilog2(rwq->wqe_count); return 0; } static int prepare_user_rq(struct ib_pd *pd, struct ib_wq_init_attr *init_attr, struct ib_udata *udata, struct mlx5_ib_rwq *rwq) { struct mlx5_ib_dev *dev = to_mdev(pd->device); struct mlx5_ib_create_wq ucmd = {}; int err; size_t required_cmd_sz; required_cmd_sz = offsetof(typeof(ucmd), reserved) + sizeof(ucmd.reserved); if (udata->inlen < required_cmd_sz) { mlx5_ib_dbg(dev, "invalid inlen\n"); return -EINVAL; } if (udata->inlen > sizeof(ucmd) && !ib_is_udata_cleared(udata, sizeof(ucmd), udata->inlen - sizeof(ucmd))) { mlx5_ib_dbg(dev, "inlen is not supported\n"); return -EOPNOTSUPP; } if (ib_copy_from_udata(&ucmd, udata, min(sizeof(ucmd), udata->inlen))) { mlx5_ib_dbg(dev, "copy failed\n"); return -EFAULT; } if (ucmd.comp_mask) { mlx5_ib_dbg(dev, "invalid comp mask\n"); return -EOPNOTSUPP; } if (ucmd.reserved) { mlx5_ib_dbg(dev, "invalid reserved\n"); return -EOPNOTSUPP; } err = set_user_rq_size(dev, init_attr, &ucmd, rwq); if (err) { mlx5_ib_dbg(dev, "err %d\n", err); return err; } err = create_user_rq(dev, pd, rwq, &ucmd); if (err) { mlx5_ib_dbg(dev, "err %d\n", err); if (err) return err; } rwq->user_index = ucmd.user_index; return 0; } struct ib_wq *mlx5_ib_create_wq(struct ib_pd *pd, struct ib_wq_init_attr *init_attr, struct ib_udata *udata) { struct mlx5_ib_dev *dev; struct mlx5_ib_rwq *rwq; struct mlx5_ib_create_wq_resp resp = {}; size_t min_resp_len; int err; if (!udata) return ERR_PTR(-ENOSYS); min_resp_len = offsetof(typeof(resp), reserved) + sizeof(resp.reserved); if (udata->outlen && udata->outlen < min_resp_len) return ERR_PTR(-EINVAL); dev = to_mdev(pd->device); switch (init_attr->wq_type) { case IB_WQT_RQ: rwq = kzalloc(sizeof(*rwq), GFP_KERNEL); if (!rwq) return ERR_PTR(-ENOMEM); err = prepare_user_rq(pd, init_attr, udata, rwq); if (err) goto err; err = create_rq(rwq, pd, init_attr); if (err) goto err_user_rq; break; default: mlx5_ib_dbg(dev, "unsupported wq type %d\n", init_attr->wq_type); return ERR_PTR(-EINVAL); } rwq->ibwq.wq_num = rwq->core_qp.qpn; rwq->ibwq.state = IB_WQS_RESET; if (udata->outlen) { resp.response_length = offsetof(typeof(resp), response_length) + sizeof(resp.response_length); err = ib_copy_to_udata(udata, &resp, resp.response_length); if (err) goto err_copy; } rwq->core_qp.event = mlx5_ib_wq_event; rwq->ibwq.event_handler = init_attr->event_handler; return &rwq->ibwq; err_copy: mlx5_core_destroy_rq_tracked(dev->mdev, &rwq->core_qp); err_user_rq: destroy_user_rq(pd, rwq); err: kfree(rwq); return ERR_PTR(err); } int mlx5_ib_destroy_wq(struct ib_wq *wq) { struct mlx5_ib_dev *dev = to_mdev(wq->device); struct mlx5_ib_rwq *rwq = to_mrwq(wq); mlx5_core_destroy_rq_tracked(dev->mdev, &rwq->core_qp); destroy_user_rq(wq->pd, rwq); kfree(rwq); return 0; } struct ib_rwq_ind_table *mlx5_ib_create_rwq_ind_table(struct ib_device *device, struct ib_rwq_ind_table_init_attr *init_attr, struct ib_udata *udata) { struct mlx5_ib_dev *dev = to_mdev(device); struct mlx5_ib_rwq_ind_table *rwq_ind_tbl; int sz = 1 << init_attr->log_ind_tbl_size; struct mlx5_ib_create_rwq_ind_tbl_resp resp = {}; size_t min_resp_len; int inlen; int err; int i; u32 *in; void *rqtc; if (udata->inlen > 0 && !ib_is_udata_cleared(udata, 0, udata->inlen)) return ERR_PTR(-EOPNOTSUPP); if (init_attr->log_ind_tbl_size > MLX5_CAP_GEN(dev->mdev, log_max_rqt_size)) { mlx5_ib_dbg(dev, "log_ind_tbl_size = %d is bigger than supported = %d\n", init_attr->log_ind_tbl_size, MLX5_CAP_GEN(dev->mdev, log_max_rqt_size)); return ERR_PTR(-EINVAL); } min_resp_len = offsetof(typeof(resp), reserved) + sizeof(resp.reserved); if (udata->outlen && udata->outlen < min_resp_len) return ERR_PTR(-EINVAL); rwq_ind_tbl = kzalloc(sizeof(*rwq_ind_tbl), GFP_KERNEL); if (!rwq_ind_tbl) return ERR_PTR(-ENOMEM); inlen = MLX5_ST_SZ_BYTES(create_rqt_in) + sizeof(u32) * sz; in = mlx5_vzalloc(inlen); if (!in) { err = -ENOMEM; goto err; } rqtc = MLX5_ADDR_OF(create_rqt_in, in, rqt_context); MLX5_SET(rqtc, rqtc, rqt_actual_size, sz); MLX5_SET(rqtc, rqtc, rqt_max_size, sz); for (i = 0; i < sz; i++) MLX5_SET(rqtc, rqtc, rq_num[i], init_attr->ind_tbl[i]->wq_num); err = mlx5_core_create_rqt(dev->mdev, in, inlen, &rwq_ind_tbl->rqtn); kvfree(in); if (err) goto err; rwq_ind_tbl->ib_rwq_ind_tbl.ind_tbl_num = rwq_ind_tbl->rqtn; if (udata->outlen) { resp.response_length = offsetof(typeof(resp), response_length) + sizeof(resp.response_length); err = ib_copy_to_udata(udata, &resp, resp.response_length); if (err) goto err_copy; } return &rwq_ind_tbl->ib_rwq_ind_tbl; err_copy: mlx5_core_destroy_rqt(dev->mdev, rwq_ind_tbl->rqtn); err: kfree(rwq_ind_tbl); return ERR_PTR(err); } int mlx5_ib_destroy_rwq_ind_table(struct ib_rwq_ind_table *ib_rwq_ind_tbl) { struct mlx5_ib_rwq_ind_table *rwq_ind_tbl = to_mrwq_ind_table(ib_rwq_ind_tbl); struct mlx5_ib_dev *dev = to_mdev(ib_rwq_ind_tbl->device); mlx5_core_destroy_rqt(dev->mdev, rwq_ind_tbl->rqtn); kfree(rwq_ind_tbl); return 0; } int mlx5_ib_modify_wq(struct ib_wq *wq, struct ib_wq_attr *wq_attr, u32 wq_attr_mask, struct ib_udata *udata) { struct mlx5_ib_dev *dev = to_mdev(wq->device); struct mlx5_ib_rwq *rwq = to_mrwq(wq); struct mlx5_ib_modify_wq ucmd = {}; size_t required_cmd_sz; int curr_wq_state; int wq_state; int inlen; int err; void *rqc; void *in; required_cmd_sz = offsetof(typeof(ucmd), reserved) + sizeof(ucmd.reserved); if (udata->inlen < required_cmd_sz) return -EINVAL; if (udata->inlen > sizeof(ucmd) && !ib_is_udata_cleared(udata, sizeof(ucmd), udata->inlen - sizeof(ucmd))) return -EOPNOTSUPP; if (ib_copy_from_udata(&ucmd, udata, min(sizeof(ucmd), udata->inlen))) return -EFAULT; if (ucmd.comp_mask || ucmd.reserved) return -EOPNOTSUPP; inlen = MLX5_ST_SZ_BYTES(modify_rq_in); in = mlx5_vzalloc(inlen); if (!in) return -ENOMEM; rqc = MLX5_ADDR_OF(modify_rq_in, in, ctx); MLX5_SET(modify_rq_in, in, rqn, rwq->core_qp.qpn); curr_wq_state = (wq_attr_mask & IB_WQ_CUR_STATE) ? wq_attr->curr_wq_state : wq->state; wq_state = (wq_attr_mask & IB_WQ_STATE) ? wq_attr->wq_state : curr_wq_state; if (curr_wq_state == IB_WQS_ERR) curr_wq_state = MLX5_RQC_STATE_ERR; if (wq_state == IB_WQS_ERR) wq_state = MLX5_RQC_STATE_ERR; MLX5_SET(modify_rq_in, in, rq_state, curr_wq_state); MLX5_SET(rqc, rqc, state, wq_state); err = mlx5_core_modify_rq(dev->mdev, in, inlen); kvfree(in); if (!err) rwq->ibwq.state = (wq_state == MLX5_RQC_STATE_ERR) ? IB_WQS_ERR : wq_state; return err; } diff --git a/sys/dev/mlx5/mlx5_ib/mlx5_ib_srq.c b/sys/dev/mlx5/mlx5_ib/mlx5_ib_srq.c index 3d9165fd71d3..3a70331271dc 100644 --- a/sys/dev/mlx5/mlx5_ib/mlx5_ib_srq.c +++ b/sys/dev/mlx5/mlx5_ib/mlx5_ib_srq.c @@ -1,500 +1,500 @@ /*- * Copyright (c) 2013-2015, Mellanox Technologies, Ltd. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY AUTHOR AND CONTRIBUTORS `AS IS' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $FreeBSD$ */ #include "opt_rss.h" #include "opt_ratelimit.h" #include #include #include #include #include #include -#include "mlx5_ib.h" +#include /* not supported currently */ static int srq_signature; static void *get_wqe(struct mlx5_ib_srq *srq, int n) { return mlx5_buf_offset(&srq->buf, n << srq->msrq.wqe_shift); } static void mlx5_ib_srq_event(struct mlx5_core_srq *srq, int type) { struct ib_event event; struct ib_srq *ibsrq = &to_mibsrq(srq)->ibsrq; if (ibsrq->event_handler) { event.device = ibsrq->device; event.element.srq = ibsrq; switch (type) { case MLX5_EVENT_TYPE_SRQ_RQ_LIMIT: event.event = IB_EVENT_SRQ_LIMIT_REACHED; break; case MLX5_EVENT_TYPE_SRQ_CATAS_ERROR: event.event = IB_EVENT_SRQ_ERR; break; default: pr_warn("mlx5_ib: Unexpected event type %d on SRQ %06x\n", type, srq->srqn); return; } ibsrq->event_handler(&event, ibsrq->srq_context); } } static int create_srq_user(struct ib_pd *pd, struct mlx5_ib_srq *srq, struct mlx5_srq_attr *in, struct ib_udata *udata, int buf_size) { struct mlx5_ib_dev *dev = to_mdev(pd->device); struct mlx5_ib_create_srq ucmd = {}; size_t ucmdlen; int err; int npages; int page_shift; int ncont; u32 offset; u32 uidx = MLX5_IB_DEFAULT_UIDX; ucmdlen = min(udata->inlen, sizeof(ucmd)); if (ib_copy_from_udata(&ucmd, udata, ucmdlen)) { mlx5_ib_dbg(dev, "failed copy udata\n"); return -EFAULT; } if (ucmd.reserved0 || ucmd.reserved1) return -EINVAL; if (udata->inlen > sizeof(ucmd) && !ib_is_udata_cleared(udata, sizeof(ucmd), udata->inlen - sizeof(ucmd))) return -EINVAL; if (in->type == IB_SRQT_XRC) { err = get_srq_user_index(to_mucontext(pd->uobject->context), &ucmd, udata->inlen, &uidx); if (err) return err; } srq->wq_sig = !!(ucmd.flags & MLX5_SRQ_FLAG_SIGNATURE); srq->umem = ib_umem_get(pd->uobject->context, ucmd.buf_addr, buf_size, 0, 0); if (IS_ERR(srq->umem)) { mlx5_ib_dbg(dev, "failed umem get, size %d\n", buf_size); err = PTR_ERR(srq->umem); return err; } mlx5_ib_cont_pages(srq->umem, ucmd.buf_addr, 0, &npages, &page_shift, &ncont, NULL); err = mlx5_ib_get_buf_offset(ucmd.buf_addr, page_shift, &offset); if (err) { mlx5_ib_warn(dev, "bad offset\n"); goto err_umem; } in->pas = mlx5_vzalloc(sizeof(*in->pas) * ncont); if (!in->pas) { err = -ENOMEM; goto err_umem; } mlx5_ib_populate_pas(dev, srq->umem, page_shift, in->pas, 0); err = mlx5_ib_db_map_user(to_mucontext(pd->uobject->context), ucmd.db_addr, &srq->db); if (err) { mlx5_ib_dbg(dev, "map doorbell failed\n"); goto err_in; } in->log_page_size = page_shift - MLX5_ADAPTER_PAGE_SHIFT; in->page_offset = offset; if (MLX5_CAP_GEN(dev->mdev, cqe_version) == MLX5_CQE_VERSION_V1 && in->type == IB_SRQT_XRC) in->user_index = uidx; return 0; err_in: kvfree(in->pas); err_umem: ib_umem_release(srq->umem); return err; } static int create_srq_kernel(struct mlx5_ib_dev *dev, struct mlx5_ib_srq *srq, struct mlx5_srq_attr *in, int buf_size) { int err; int i; struct mlx5_wqe_srq_next_seg *next; err = mlx5_db_alloc(dev->mdev, &srq->db); if (err) { mlx5_ib_warn(dev, "alloc dbell rec failed\n"); return err; } if (mlx5_buf_alloc(dev->mdev, buf_size, 2 * PAGE_SIZE, &srq->buf)) { mlx5_ib_dbg(dev, "buf alloc failed\n"); err = -ENOMEM; goto err_db; } srq->head = 0; srq->tail = srq->msrq.max - 1; srq->wqe_ctr = 0; for (i = 0; i < srq->msrq.max; i++) { next = get_wqe(srq, i); next->next_wqe_index = cpu_to_be16((i + 1) & (srq->msrq.max - 1)); } mlx5_ib_dbg(dev, "srq->buf.page_shift = %d\n", srq->buf.page_shift); in->pas = mlx5_vzalloc(sizeof(*in->pas) * srq->buf.npages); if (!in->pas) { err = -ENOMEM; goto err_buf; } mlx5_fill_page_array(&srq->buf, in->pas); srq->wrid = kmalloc(srq->msrq.max * sizeof(u64), GFP_KERNEL); if (!srq->wrid) { mlx5_ib_dbg(dev, "kmalloc failed %lu\n", (unsigned long)(srq->msrq.max * sizeof(u64))); err = -ENOMEM; goto err_in; } srq->wq_sig = !!srq_signature; in->log_page_size = srq->buf.page_shift - MLX5_ADAPTER_PAGE_SHIFT; if (MLX5_CAP_GEN(dev->mdev, cqe_version) == MLX5_CQE_VERSION_V1 && in->type == IB_SRQT_XRC) in->user_index = MLX5_IB_DEFAULT_UIDX; return 0; err_in: kvfree(in->pas); err_buf: mlx5_buf_free(dev->mdev, &srq->buf); err_db: mlx5_db_free(dev->mdev, &srq->db); return err; } static void destroy_srq_user(struct ib_pd *pd, struct mlx5_ib_srq *srq) { mlx5_ib_db_unmap_user(to_mucontext(pd->uobject->context), &srq->db); ib_umem_release(srq->umem); } static void destroy_srq_kernel(struct mlx5_ib_dev *dev, struct mlx5_ib_srq *srq) { kfree(srq->wrid); mlx5_buf_free(dev->mdev, &srq->buf); mlx5_db_free(dev->mdev, &srq->db); } struct ib_srq *mlx5_ib_create_srq(struct ib_pd *pd, struct ib_srq_init_attr *init_attr, struct ib_udata *udata) { struct mlx5_ib_dev *dev = to_mdev(pd->device); struct mlx5_ib_srq *srq; size_t desc_size; size_t buf_size; int err; struct mlx5_srq_attr in = {0}; __u32 max_srq_wqes = 1 << MLX5_CAP_GEN(dev->mdev, log_max_srq_sz); /* Sanity check SRQ size before proceeding */ if (init_attr->attr.max_wr >= max_srq_wqes) { mlx5_ib_dbg(dev, "max_wr %d, cap %d\n", init_attr->attr.max_wr, max_srq_wqes); return ERR_PTR(-EINVAL); } srq = kmalloc(sizeof(*srq), GFP_KERNEL); if (!srq) return ERR_PTR(-ENOMEM); mutex_init(&srq->mutex); spin_lock_init(&srq->lock); srq->msrq.max = roundup_pow_of_two(init_attr->attr.max_wr + 1); srq->msrq.max_gs = init_attr->attr.max_sge; desc_size = sizeof(struct mlx5_wqe_srq_next_seg) + srq->msrq.max_gs * sizeof(struct mlx5_wqe_data_seg); if (desc_size == 0 || srq->msrq.max_gs > desc_size) { err = -EINVAL; goto err_srq; } desc_size = roundup_pow_of_two(desc_size); desc_size = max_t(size_t, 32, desc_size); if (desc_size < sizeof(struct mlx5_wqe_srq_next_seg)) { err = -EINVAL; goto err_srq; } srq->msrq.max_avail_gather = (desc_size - sizeof(struct mlx5_wqe_srq_next_seg)) / sizeof(struct mlx5_wqe_data_seg); srq->msrq.wqe_shift = ilog2(desc_size); buf_size = srq->msrq.max * desc_size; if (buf_size < desc_size) { err = -EINVAL; goto err_srq; } in.type = init_attr->srq_type; if (pd->uobject) err = create_srq_user(pd, srq, &in, udata, buf_size); else err = create_srq_kernel(dev, srq, &in, buf_size); if (err || !in.pas) { mlx5_ib_warn(dev, "create srq %s failed, err %d\n", pd->uobject ? "user" : "kernel", err); goto err_srq; } in.log_size = ilog2(srq->msrq.max); in.wqe_shift = srq->msrq.wqe_shift - 4; if (srq->wq_sig) in.flags |= MLX5_SRQ_FLAG_WQ_SIG; if (init_attr->srq_type == IB_SRQT_XRC) { in.xrcd = to_mxrcd(init_attr->ext.xrc.xrcd)->xrcdn; in.cqn = to_mcq(init_attr->ext.xrc.cq)->mcq.cqn; } else if (init_attr->srq_type == IB_SRQT_BASIC) { in.xrcd = to_mxrcd(dev->devr.x0)->xrcdn; in.cqn = to_mcq(dev->devr.c0)->mcq.cqn; } in.pd = to_mpd(pd)->pdn; in.db_record = srq->db.dma; err = mlx5_core_create_srq(dev->mdev, &srq->msrq, &in); kvfree(in.pas); if (err) { mlx5_ib_dbg(dev, "create SRQ failed, err %d\n", err); goto err_usr_kern_srq; } mlx5_ib_dbg(dev, "create SRQ with srqn 0x%x\n", srq->msrq.srqn); srq->msrq.event = mlx5_ib_srq_event; srq->ibsrq.ext.xrc.srq_num = srq->msrq.srqn; if (pd->uobject) if (ib_copy_to_udata(udata, &srq->msrq.srqn, sizeof(__u32))) { mlx5_ib_dbg(dev, "copy to user failed\n"); err = -EFAULT; goto err_core; } init_attr->attr.max_wr = srq->msrq.max - 1; return &srq->ibsrq; err_core: mlx5_core_destroy_srq(dev->mdev, &srq->msrq); err_usr_kern_srq: if (pd->uobject) destroy_srq_user(pd, srq); else destroy_srq_kernel(dev, srq); err_srq: kfree(srq); return ERR_PTR(err); } int mlx5_ib_modify_srq(struct ib_srq *ibsrq, struct ib_srq_attr *attr, enum ib_srq_attr_mask attr_mask, struct ib_udata *udata) { struct mlx5_ib_dev *dev = to_mdev(ibsrq->device); struct mlx5_ib_srq *srq = to_msrq(ibsrq); int ret; /* We don't support resizing SRQs yet */ if (attr_mask & IB_SRQ_MAX_WR) return -EINVAL; if (attr_mask & IB_SRQ_LIMIT) { if (attr->srq_limit >= srq->msrq.max) return -EINVAL; mutex_lock(&srq->mutex); ret = mlx5_core_arm_srq(dev->mdev, &srq->msrq, attr->srq_limit, 1); mutex_unlock(&srq->mutex); if (ret) return ret; } return 0; } int mlx5_ib_query_srq(struct ib_srq *ibsrq, struct ib_srq_attr *srq_attr) { struct mlx5_ib_dev *dev = to_mdev(ibsrq->device); struct mlx5_ib_srq *srq = to_msrq(ibsrq); int ret; struct mlx5_srq_attr *out; out = kzalloc(sizeof(*out), GFP_KERNEL); if (!out) return -ENOMEM; ret = mlx5_core_query_srq(dev->mdev, &srq->msrq, out); if (ret) goto out_box; srq_attr->srq_limit = out->lwm; srq_attr->max_wr = srq->msrq.max - 1; srq_attr->max_sge = srq->msrq.max_gs; out_box: kfree(out); return ret; } int mlx5_ib_destroy_srq(struct ib_srq *srq) { struct mlx5_ib_dev *dev = to_mdev(srq->device); struct mlx5_ib_srq *msrq = to_msrq(srq); mlx5_core_destroy_srq(dev->mdev, &msrq->msrq); if (srq->uobject) { mlx5_ib_db_unmap_user(to_mucontext(srq->uobject->context), &msrq->db); ib_umem_release(msrq->umem); } else { destroy_srq_kernel(dev, msrq); } kfree(srq); return 0; } void mlx5_ib_free_srq_wqe(struct mlx5_ib_srq *srq, int wqe_index) { struct mlx5_wqe_srq_next_seg *next; /* always called with interrupts disabled. */ spin_lock(&srq->lock); next = get_wqe(srq, srq->tail); next->next_wqe_index = cpu_to_be16(wqe_index); srq->tail = wqe_index; spin_unlock(&srq->lock); } int mlx5_ib_post_srq_recv(struct ib_srq *ibsrq, const struct ib_recv_wr *wr, const struct ib_recv_wr **bad_wr) { struct mlx5_ib_srq *srq = to_msrq(ibsrq); struct mlx5_wqe_srq_next_seg *next; struct mlx5_wqe_data_seg *scat; struct mlx5_ib_dev *dev = to_mdev(ibsrq->device); struct mlx5_core_dev *mdev = dev->mdev; unsigned long flags; int err = 0; int nreq; int i; spin_lock_irqsave(&srq->lock, flags); if (mdev->state == MLX5_DEVICE_STATE_INTERNAL_ERROR) { err = -EIO; *bad_wr = wr; goto out; } for (nreq = 0; wr; nreq++, wr = wr->next) { if (unlikely(wr->num_sge > srq->msrq.max_gs)) { err = -EINVAL; *bad_wr = wr; break; } if (unlikely(srq->head == srq->tail)) { err = -ENOMEM; *bad_wr = wr; break; } srq->wrid[srq->head] = wr->wr_id; next = get_wqe(srq, srq->head); srq->head = be16_to_cpu(next->next_wqe_index); scat = (struct mlx5_wqe_data_seg *)(next + 1); for (i = 0; i < wr->num_sge; i++) { scat[i].byte_count = cpu_to_be32(wr->sg_list[i].length); scat[i].lkey = cpu_to_be32(wr->sg_list[i].lkey); scat[i].addr = cpu_to_be64(wr->sg_list[i].addr); } if (i < srq->msrq.max_avail_gather) { scat[i].byte_count = 0; scat[i].lkey = cpu_to_be32(MLX5_INVALID_LKEY); scat[i].addr = 0; } } if (likely(nreq)) { srq->wqe_ctr += nreq; /* Make sure that descriptors are written before * doorbell record. */ wmb(); *srq->db.db = cpu_to_be32(srq->wqe_ctr); } out: spin_unlock_irqrestore(&srq->lock, flags); return err; } diff --git a/sys/dev/mlx5/mlx5_ib/mlx5_ib_virt.c b/sys/dev/mlx5/mlx5_ib/mlx5_ib_virt.c index 5bd72e7475fc..d6d15fb2d27e 100644 --- a/sys/dev/mlx5/mlx5_ib/mlx5_ib_virt.c +++ b/sys/dev/mlx5/mlx5_ib/mlx5_ib_virt.c @@ -1,57 +1,57 @@ /*- * Copyright (c) 2016, Mellanox Technologies, Ltd. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY AUTHOR AND CONTRIBUTORS `AS IS' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $FreeBSD$ */ #include "opt_rss.h" #include "opt_ratelimit.h" #include #include -#include "mlx5_ib.h" +#include int mlx5_ib_get_vf_config(struct ib_device *device, int vf, u8 port, struct ifla_vf_info *info) { return -EOPNOTSUPP; } int mlx5_ib_set_vf_link_state(struct ib_device *device, int vf, u8 port, int state) { return -EOPNOTSUPP; } int mlx5_ib_get_vf_stats(struct ib_device *device, int vf, u8 port, struct ifla_vf_stats *stats) { return -EOPNOTSUPP; } int mlx5_ib_set_vf_guid(struct ib_device *device, int vf, u8 port, u64 guid, int type) { return -EOPNOTSUPP; }