diff options
author | Linus Torvalds <torvalds@linux-foundation.org> | 2015-06-24 16:49:49 -0700 |
---|---|---|
committer | Linus Torvalds <torvalds@linux-foundation.org> | 2015-06-24 16:49:49 -0700 |
commit | e0456717e483bb8a9431b80a5bdc99a928b9b003 (patch) | |
tree | 5eb5add2bafd1f20326d70f5cb3b711d00a40b10 /drivers/infiniband | |
parent | 98ec21a01896751b673b6c731ca8881daa8b2c6d (diff) | |
parent | 1ea2d020ba477cb7011a7174e8501a9e04a325d4 (diff) | |
download | blackbird-op-linux-e0456717e483bb8a9431b80a5bdc99a928b9b003.tar.gz blackbird-op-linux-e0456717e483bb8a9431b80a5bdc99a928b9b003.zip |
Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next
Pull networking updates from David Miller:
1) Add TX fast path in mac80211, from Johannes Berg.
2) Add TSO/GRO support to ibmveth, from Thomas Falcon
3) Move away from cached routes in ipv6, just like ipv4, from Martin
KaFai Lau.
4) Lots of new rhashtable tests, from Thomas Graf.
5) Run ingress qdisc lockless, from Alexei Starovoitov.
6) Allow servers to fetch TCP packet headers for SYN packets of new
connections, for fingerprinting. From Eric Dumazet.
7) Add mode parameter to pktgen, for testing receive. From Alexei
Starovoitov.
8) Cache access optimizations via simplifications of build_skb(), from
Alexander Duyck.
9) Move page frag allocator under mm/, also from Alexander.
10) Add xmit_more support to hv_netvsc, from KY Srinivasan.
11) Add a counter guard in case we try to perform endless reclassify
loops in the packet scheduler.
12) Extern flow dissector to be programmable and use it in new "Flower"
classifier. From Jiri Pirko.
13) AF_PACKET fanout rollover fixes, performance improvements, and new
statistics. From Willem de Bruijn.
14) Add netdev driver for GENEVE tunnels, from John W Linville.
15) Add ingress netfilter hooks and filtering, from Pablo Neira Ayuso.
16) Fix handling of epoll edge triggers in TCP, from Eric Dumazet.
17) Add an ECN retry fallback for the initial TCP handshake, from Daniel
Borkmann.
18) Add tail call support to BPF, from Alexei Starovoitov.
19) Add several pktgen helper scripts, from Jesper Dangaard Brouer.
20) Add zerocopy support to AF_UNIX, from Hannes Frederic Sowa.
21) Favor even port numbers for allocation to connect() requests, and
odd port numbers for bind(0), in an effort to help avoid
ip_local_port_range exhaustion. From Eric Dumazet.
22) Add Cavium ThunderX driver, from Sunil Goutham.
23) Allow bpf programs to access skb_iif and dev->ifindex SKB metadata,
from Alexei Starovoitov.
24) Add support for T6 chips in cxgb4vf driver, from Hariprasad Shenai.
25) Double TCP Small Queues default to 256K to accomodate situations
like the XEN driver and wireless aggregation. From Wei Liu.
26) Add more entropy inputs to flow dissector, from Tom Herbert.
27) Add CDG congestion control algorithm to TCP, from Kenneth Klette
Jonassen.
28) Convert ipset over to RCU locking, from Jozsef Kadlecsik.
29) Track and act upon link status of ipv4 route nexthops, from Andy
Gospodarek.
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next: (1670 commits)
bridge: vlan: flush the dynamically learned entries on port vlan delete
bridge: multicast: add a comment to br_port_state_selection about blocking state
net: inet_diag: export IPV6_V6ONLY sockopt
stmmac: troubleshoot unexpected bits in des0 & des1
net: ipv4 sysctl option to ignore routes when nexthop link is down
net: track link-status of ipv4 nexthops
net: switchdev: ignore unsupported bridge flags
net: Cavium: Fix MAC address setting in shutdown state
drivers: net: xgene: fix for ACPI support without ACPI
ip: report the original address of ICMP messages
net/mlx5e: Prefetch skb data on RX
net/mlx5e: Pop cq outside mlx5e_get_cqe
net/mlx5e: Remove mlx5e_cq.sqrq back-pointer
net/mlx5e: Remove extra spaces
net/mlx5e: Avoid TX CQE generation if more xmit packets expected
net/mlx5e: Avoid redundant dev_kfree_skb() upon NOP completion
net/mlx5e: Remove re-assignment of wq type in mlx5e_enable_rq()
net/mlx5e: Use skb_shinfo(skb)->gso_segs rather than counting them
net/mlx5e: Static mapping of netdev priv resources to/from netdev TX queues
net/mlx4_en: Use HW counters for rx/tx bytes/packets in PF device
...
Diffstat (limited to 'drivers/infiniband')
-rw-r--r-- | drivers/infiniband/hw/cxgb4/provider.c | 8 | ||||
-rw-r--r-- | drivers/infiniband/hw/mlx4/alias_GUID.c | 7 | ||||
-rw-r--r-- | drivers/infiniband/hw/mlx4/mad.c | 49 | ||||
-rw-r--r-- | drivers/infiniband/hw/mlx4/main.c | 118 | ||||
-rw-r--r-- | drivers/infiniband/hw/mlx4/mlx4_ib.h | 8 | ||||
-rw-r--r-- | drivers/infiniband/hw/mlx4/qp.c | 7 | ||||
-rw-r--r-- | drivers/infiniband/hw/mlx5/Kconfig | 4 | ||||
-rw-r--r-- | drivers/infiniband/hw/mlx5/cq.c | 11 | ||||
-rw-r--r-- | drivers/infiniband/hw/mlx5/mad.c | 299 | ||||
-rw-r--r-- | drivers/infiniband/hw/mlx5/main.c | 646 | ||||
-rw-r--r-- | drivers/infiniband/hw/mlx5/mlx5_ib.h | 23 | ||||
-rw-r--r-- | drivers/infiniband/hw/mlx5/mr.c | 3 | ||||
-rw-r--r-- | drivers/infiniband/hw/mlx5/odp.c | 47 | ||||
-rw-r--r-- | drivers/infiniband/hw/mlx5/qp.c | 89 | ||||
-rw-r--r-- | drivers/infiniband/hw/mlx5/srq.c | 11 |
15 files changed, 902 insertions, 428 deletions
diff --git a/drivers/infiniband/hw/cxgb4/provider.c b/drivers/infiniband/hw/cxgb4/provider.c index 62c816af46e4..6eee3d385541 100644 --- a/drivers/infiniband/hw/cxgb4/provider.c +++ b/drivers/infiniband/hw/cxgb4/provider.c @@ -453,10 +453,10 @@ static int c4iw_get_mib(struct ib_device *ibdev, cxgb4_get_tcp_stats(c4iw_dev->rdev.lldi.pdev, &v4, &v6); memset(stats, 0, sizeof *stats); - stats->iw.tcpInSegs = v4.tcpInSegs + v6.tcpInSegs; - stats->iw.tcpOutSegs = v4.tcpOutSegs + v6.tcpOutSegs; - stats->iw.tcpRetransSegs = v4.tcpRetransSegs + v6.tcpRetransSegs; - stats->iw.tcpOutRsts = v4.tcpOutRsts + v6.tcpOutSegs; + stats->iw.tcpInSegs = v4.tcp_in_segs + v6.tcp_in_segs; + stats->iw.tcpOutSegs = v4.tcp_out_segs + v6.tcp_out_segs; + stats->iw.tcpRetransSegs = v4.tcp_retrans_segs + v6.tcp_retrans_segs; + stats->iw.tcpOutRsts = v4.tcp_out_rsts + v6.tcp_out_rsts; return 0; } diff --git a/drivers/infiniband/hw/mlx4/alias_GUID.c b/drivers/infiniband/hw/mlx4/alias_GUID.c index 0f00204d2ece..21cb41a60fe8 100644 --- a/drivers/infiniband/hw/mlx4/alias_GUID.c +++ b/drivers/infiniband/hw/mlx4/alias_GUID.c @@ -189,7 +189,7 @@ void mlx4_ib_notify_slaves_on_guid_change(struct mlx4_ib_dev *dev, { int i; u64 guid_indexes; - int slave_id; + int slave_id, slave_port; enum slave_port_state new_state; enum slave_port_state prev_state; __be64 tmp_cur_ag, form_cache_ag; @@ -217,6 +217,11 @@ void mlx4_ib_notify_slaves_on_guid_change(struct mlx4_ib_dev *dev, slave_id = (block_num * NUM_ALIAS_GUID_IN_REC) + i ; if (slave_id >= dev->dev->persist->num_vfs + 1) return; + + slave_port = mlx4_phys_to_slave_port(dev->dev, slave_id, port_num); + if (slave_port < 0) /* this port isn't available for the VF */ + continue; + tmp_cur_ag = *(__be64 *)&p_data[i * GUID_REC_SIZE]; form_cache_ag = get_cached_alias_guid(dev, port_num, (NUM_ALIAS_GUID_IN_REC * block_num) + i); diff --git a/drivers/infiniband/hw/mlx4/mad.c b/drivers/infiniband/hw/mlx4/mad.c index 3e2dee46caa2..85a50df2f203 100644 --- a/drivers/infiniband/hw/mlx4/mad.c +++ b/drivers/infiniband/hw/mlx4/mad.c @@ -64,14 +64,6 @@ enum { #define GUID_TBL_BLK_NUM_ENTRIES 8 #define GUID_TBL_BLK_SIZE (GUID_TBL_ENTRY_SIZE * GUID_TBL_BLK_NUM_ENTRIES) -/* Counters should be saturate once they reach their maximum value */ -#define ASSIGN_32BIT_COUNTER(counter, value) do {\ - if ((value) > U32_MAX) \ - counter = cpu_to_be32(U32_MAX); \ - else \ - counter = cpu_to_be32(value); \ -} while (0) - struct mlx4_mad_rcv_buf { struct ib_grh grh; u8 payload[256]; @@ -830,31 +822,25 @@ static int iboe_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) { - struct mlx4_cmd_mailbox *mailbox; + struct mlx4_counter counter_stats; struct mlx4_ib_dev *dev = to_mdev(ibdev); int err; - u32 inmod = dev->counters[port_num - 1] & 0xffff; - u8 mode; if (in_mad->mad_hdr.mgmt_class != IB_MGMT_CLASS_PERF_MGMT) return -EINVAL; - mailbox = mlx4_alloc_cmd_mailbox(dev->dev); - if (IS_ERR(mailbox)) - return IB_MAD_RESULT_FAILURE; - - err = mlx4_cmd_box(dev->dev, 0, mailbox->dma, inmod, 0, - MLX4_CMD_QUERY_IF_STAT, MLX4_CMD_TIME_CLASS_C, - MLX4_CMD_WRAPPED); + memset(&counter_stats, 0, sizeof(counter_stats)); + err = mlx4_get_counter_stats(dev->dev, + dev->counters[port_num - 1].index, + &counter_stats, 0); if (err) err = IB_MAD_RESULT_FAILURE; else { memset(out_mad->data, 0, sizeof out_mad->data); - mode = ((struct mlx4_counter *)mailbox->buf)->counter_mode; - switch (mode & 0xf) { + switch (counter_stats.counter_mode & 0xf) { case 0: - edit_counter(mailbox->buf, - (void *)(out_mad->data + 40)); + edit_counter(&counter_stats, + (void *)(out_mad->data + 40)); err = IB_MAD_RESULT_SUCCESS | IB_MAD_RESULT_REPLY; break; default: @@ -862,8 +848,6 @@ static int iboe_process_mad(struct ib_device *ibdev, int mad_flags, u8 port_num, } } - mlx4_free_cmd_mailbox(dev->dev, mailbox); - return err; } @@ -873,6 +857,7 @@ int mlx4_ib_process_mad(struct ib_device *ibdev, int mad_flags, u8 port_num, struct ib_mad_hdr *out, size_t *out_mad_size, u16 *out_mad_pkey_index) { + struct mlx4_ib_dev *dev = to_mdev(ibdev); const struct ib_mad *in_mad = (const struct ib_mad *)in; struct ib_mad *out_mad = (struct ib_mad *)out; @@ -881,8 +866,9 @@ int mlx4_ib_process_mad(struct ib_device *ibdev, int mad_flags, u8 port_num, switch (rdma_port_get_link_layer(ibdev, port_num)) { case IB_LINK_LAYER_INFINIBAND: - return ib_process_mad(ibdev, mad_flags, port_num, in_wc, - in_grh, in_mad, out_mad); + if (!mlx4_is_slave(dev->dev)) + return ib_process_mad(ibdev, mad_flags, port_num, in_wc, + in_grh, in_mad, out_mad); case IB_LINK_LAYER_ETHERNET: return iboe_process_mad(ibdev, mad_flags, port_num, in_wc, in_grh, in_mad, out_mad); @@ -1375,14 +1361,17 @@ static void mlx4_ib_multiplex_mad(struct mlx4_ib_demux_pv_ctx *ctx, struct ib_wc * stadard address handle by decoding the tunnelled mlx4_ah fields */ memcpy(&ah.av, &tunnel->hdr.av, sizeof (struct mlx4_av)); ah.ibah.device = ctx->ib_dev; + + port = be32_to_cpu(ah.av.ib.port_pd) >> 24; + port = mlx4_slave_convert_port(dev->dev, slave, port); + if (port < 0) + return; + ah.av.ib.port_pd = cpu_to_be32(port << 24 | (be32_to_cpu(ah.av.ib.port_pd) & 0xffffff)); + mlx4_ib_query_ah(&ah.ibah, &ah_attr); if (ah_attr.ah_flags & IB_AH_GRH) fill_in_real_sgid_index(dev, slave, ctx->port, &ah_attr); - port = mlx4_slave_convert_port(dev->dev, slave, ah_attr.port_num); - if (port < 0) - return; - ah_attr.port_num = port; memcpy(ah_attr.dmac, tunnel->hdr.mac, 6); ah_attr.vlan_id = be16_to_cpu(tunnel->hdr.vlan); /* if slave have default vlan use it */ diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c index 166da787780c..067a691ecbed 100644 --- a/drivers/infiniband/hw/mlx4/main.c +++ b/drivers/infiniband/hw/mlx4/main.c @@ -1146,7 +1146,7 @@ static int __mlx4_ib_create_flow(struct ib_qp *qp, struct ib_flow_attr *flow_att ret = mlx4_cmd_imm(mdev->dev, mailbox->dma, reg_id, size >> 2, 0, MLX4_QP_FLOW_STEERING_ATTACH, MLX4_CMD_TIME_CLASS_A, - MLX4_CMD_NATIVE); + MLX4_CMD_WRAPPED); if (ret == -ENOMEM) pr_err("mcg table is full. Fail to register network rule.\n"); else if (ret == -ENXIO) @@ -1163,7 +1163,7 @@ static int __mlx4_ib_destroy_flow(struct mlx4_dev *dev, u64 reg_id) int err; err = mlx4_cmd(dev, reg_id, 0, 0, MLX4_QP_FLOW_STEERING_DETACH, MLX4_CMD_TIME_CLASS_A, - MLX4_CMD_NATIVE); + MLX4_CMD_WRAPPED); if (err) pr_err("Fail to detach network rule. registration id = 0x%llx\n", reg_id); @@ -2098,77 +2098,52 @@ static void init_pkeys(struct mlx4_ib_dev *ibdev) static void mlx4_ib_alloc_eqs(struct mlx4_dev *dev, struct mlx4_ib_dev *ibdev) { - char name[80]; - int eq_per_port = 0; - int added_eqs = 0; - int total_eqs = 0; - int i, j, eq; - - /* Legacy mode or comp_pool is not large enough */ - if (dev->caps.comp_pool == 0 || - dev->caps.num_ports > dev->caps.comp_pool) - return; - - eq_per_port = dev->caps.comp_pool / dev->caps.num_ports; + int i, j, eq = 0, total_eqs = 0; - /* Init eq table */ - added_eqs = 0; - mlx4_foreach_port(i, dev, MLX4_PORT_TYPE_IB) - added_eqs += eq_per_port; - - total_eqs = dev->caps.num_comp_vectors + added_eqs; - - ibdev->eq_table = kzalloc(total_eqs * sizeof(int), GFP_KERNEL); + ibdev->eq_table = kcalloc(dev->caps.num_comp_vectors, + sizeof(ibdev->eq_table[0]), GFP_KERNEL); if (!ibdev->eq_table) return; - ibdev->eq_added = added_eqs; - - eq = 0; - mlx4_foreach_port(i, dev, MLX4_PORT_TYPE_IB) { - for (j = 0; j < eq_per_port; j++) { - snprintf(name, sizeof(name), "mlx4-ib-%d-%d@%s", - i, j, dev->persist->pdev->bus->name); - /* Set IRQ for specific name (per ring) */ - if (mlx4_assign_eq(dev, name, NULL, - &ibdev->eq_table[eq])) { - /* Use legacy (same as mlx4_en driver) */ - pr_warn("Can't allocate EQ %d; reverting to legacy\n", eq); - ibdev->eq_table[eq] = - (eq % dev->caps.num_comp_vectors); - } - eq++; + for (i = 1; i <= dev->caps.num_ports; i++) { + for (j = 0; j < mlx4_get_eqs_per_port(dev, i); + j++, total_eqs++) { + if (i > 1 && mlx4_is_eq_shared(dev, total_eqs)) + continue; + ibdev->eq_table[eq] = total_eqs; + if (!mlx4_assign_eq(dev, i, + &ibdev->eq_table[eq])) + eq++; + else + ibdev->eq_table[eq] = -1; } } - /* Fill the reset of the vector with legacy EQ */ - for (i = 0, eq = added_eqs; i < dev->caps.num_comp_vectors; i++) - ibdev->eq_table[eq++] = i; + for (i = eq; i < dev->caps.num_comp_vectors; + ibdev->eq_table[i++] = -1) + ; /* Advertise the new number of EQs to clients */ - ibdev->ib_dev.num_comp_vectors = total_eqs; + ibdev->ib_dev.num_comp_vectors = eq; } static void mlx4_ib_free_eqs(struct mlx4_dev *dev, struct mlx4_ib_dev *ibdev) { int i; + int total_eqs = ibdev->ib_dev.num_comp_vectors; - /* no additional eqs were added */ + /* no eqs were allocated */ if (!ibdev->eq_table) return; /* Reset the advertised EQ number */ - ibdev->ib_dev.num_comp_vectors = dev->caps.num_comp_vectors; + ibdev->ib_dev.num_comp_vectors = 0; - /* Free only the added eqs */ - for (i = 0; i < ibdev->eq_added; i++) { - /* Don't free legacy eqs if used */ - if (ibdev->eq_table[i] <= dev->caps.num_comp_vectors) - continue; + for (i = 0; i < total_eqs; i++) mlx4_release_eq(dev, ibdev->eq_table[i]); - } kfree(ibdev->eq_table); + ibdev->eq_table = NULL; } static int mlx4_port_immutable(struct ib_device *ibdev, u8 port_num, @@ -2203,6 +2178,8 @@ static void *mlx4_ib_add(struct mlx4_dev *dev) struct mlx4_ib_iboe *iboe; int ib_num_ports = 0; int num_req_counters; + int allocated; + u32 counter_index; pr_info_once("%s", mlx4_ib_version); @@ -2373,19 +2350,31 @@ static void *mlx4_ib_add(struct mlx4_dev *dev) num_req_counters = mlx4_is_bonded(dev) ? 1 : ibdev->num_ports; for (i = 0; i < num_req_counters; ++i) { mutex_init(&ibdev->qp1_proxy_lock[i]); + allocated = 0; if (mlx4_ib_port_link_layer(&ibdev->ib_dev, i + 1) == IB_LINK_LAYER_ETHERNET) { - err = mlx4_counter_alloc(ibdev->dev, &ibdev->counters[i]); + err = mlx4_counter_alloc(ibdev->dev, &counter_index); + /* if failed to allocate a new counter, use default */ if (err) - ibdev->counters[i] = -1; - } else { - ibdev->counters[i] = -1; + counter_index = + mlx4_get_default_counter_index(dev, + i + 1); + else + allocated = 1; + } else { /* IB_LINK_LAYER_INFINIBAND use the default counter */ + counter_index = mlx4_get_default_counter_index(dev, + i + 1); } + ibdev->counters[i].index = counter_index; + ibdev->counters[i].allocated = allocated; + pr_info("counter index %d for port %d allocated %d\n", + counter_index, i + 1, allocated); } if (mlx4_is_bonded(dev)) - for (i = 1; i < ibdev->num_ports ; ++i) - ibdev->counters[i] = ibdev->counters[0]; - + for (i = 1; i < ibdev->num_ports ; ++i) { + ibdev->counters[i].index = ibdev->counters[0].index; + ibdev->counters[i].allocated = 0; + } mlx4_foreach_port(i, dev, MLX4_PORT_TYPE_IB) ib_num_ports++; @@ -2525,10 +2514,12 @@ err_steer_qp_release: mlx4_qp_release_range(dev, ibdev->steer_qpn_base, ibdev->steer_qpn_count); err_counter: - for (; i; --i) - if (ibdev->counters[i - 1] != -1) - mlx4_counter_free(ibdev->dev, ibdev->counters[i - 1]); - + for (i = 0; i < ibdev->num_ports; ++i) { + if (ibdev->counters[i].index != -1 && + ibdev->counters[i].allocated) + mlx4_counter_free(ibdev->dev, + ibdev->counters[i].index); + } err_map: iounmap(ibdev->uar_map); @@ -2645,8 +2636,9 @@ static void mlx4_ib_remove(struct mlx4_dev *dev, void *ibdev_ptr) iounmap(ibdev->uar_map); for (p = 0; p < ibdev->num_ports; ++p) - if (ibdev->counters[p] != -1) - mlx4_counter_free(ibdev->dev, ibdev->counters[p]); + if (ibdev->counters[p].index != -1 && + ibdev->counters[p].allocated) + mlx4_counter_free(ibdev->dev, ibdev->counters[p].index); mlx4_foreach_port(p, dev, MLX4_PORT_TYPE_IB) mlx4_CLOSE_PORT(dev, p); diff --git a/drivers/infiniband/hw/mlx4/mlx4_ib.h b/drivers/infiniband/hw/mlx4/mlx4_ib.h index 7933adfff662..334387f63358 100644 --- a/drivers/infiniband/hw/mlx4/mlx4_ib.h +++ b/drivers/infiniband/hw/mlx4/mlx4_ib.h @@ -504,6 +504,11 @@ struct mlx4_ib_iov_port { struct mlx4_ib_iov_sysfs_attr mcg_dentry; }; +struct counter_index { + u32 index; + u8 allocated; +}; + struct mlx4_ib_dev { struct ib_device ib_dev; struct mlx4_dev *dev; @@ -522,9 +527,8 @@ struct mlx4_ib_dev { struct mutex cap_mask_mutex; bool ib_active; struct mlx4_ib_iboe iboe; - int counters[MLX4_MAX_PORTS]; + struct counter_index counters[MLX4_MAX_PORTS]; int *eq_table; - int eq_added; struct kobject *iov_parent; struct kobject *ports_parent; struct kobject *dev_ports_parent[MLX4_MFUNC_MAX]; diff --git a/drivers/infiniband/hw/mlx4/qp.c b/drivers/infiniband/hw/mlx4/qp.c index 02fc91c68027..c5a3a5f0de41 100644 --- a/drivers/infiniband/hw/mlx4/qp.c +++ b/drivers/infiniband/hw/mlx4/qp.c @@ -1539,12 +1539,13 @@ static int __mlx4_ib_modify_qp(struct ib_qp *ibqp, } if (cur_state == IB_QPS_INIT && new_state == IB_QPS_RTR) { - if (dev->counters[qp->port - 1] != -1) { + if (dev->counters[qp->port - 1].index != -1) { context->pri_path.counter_index = - dev->counters[qp->port - 1]; + dev->counters[qp->port - 1].index; optpar |= MLX4_QP_OPTPAR_COUNTER_INDEX; } else - context->pri_path.counter_index = 0xff; + context->pri_path.counter_index = + MLX4_SINK_COUNTER_INDEX(dev->dev); if (qp->flags & MLX4_IB_QP_NETIF) { mlx4_ib_steer_qp_reg(dev, qp, 1); diff --git a/drivers/infiniband/hw/mlx5/Kconfig b/drivers/infiniband/hw/mlx5/Kconfig index 10df386c6344..bce263b92821 100644 --- a/drivers/infiniband/hw/mlx5/Kconfig +++ b/drivers/infiniband/hw/mlx5/Kconfig @@ -1,8 +1,6 @@ config MLX5_INFINIBAND tristate "Mellanox Connect-IB HCA support" - depends on NETDEVICES && ETHERNET && PCI - select NET_VENDOR_MELLANOX - select MLX5_CORE + depends on NETDEVICES && ETHERNET && PCI && MLX5_CORE ---help--- This driver provides low-level InfiniBand support for Mellanox Connect-IB PCI Express host channel adapters (HCAs). diff --git a/drivers/infiniband/hw/mlx5/cq.c b/drivers/infiniband/hw/mlx5/cq.c index 09fbae618d35..5c9eeea62805 100644 --- a/drivers/infiniband/hw/mlx5/cq.c +++ b/drivers/infiniband/hw/mlx5/cq.c @@ -590,8 +590,7 @@ static int alloc_cq_buf(struct mlx5_ib_dev *dev, struct mlx5_ib_cq_buf *buf, { int err; - err = mlx5_buf_alloc(dev->mdev, nent * cqe_size, - PAGE_SIZE * 2, &buf->buf); + err = mlx5_buf_alloc(dev->mdev, nent * cqe_size, &buf->buf); if (err) return err; @@ -760,7 +759,7 @@ struct ib_cq *mlx5_ib_create_cq(struct ib_device *ibdev, return ERR_PTR(-EINVAL); entries = roundup_pow_of_two(entries + 1); - if (entries > dev->mdev->caps.gen.max_cqes) + if (entries > (1 << MLX5_CAP_GEN(dev->mdev, log_max_cq_sz))) return ERR_PTR(-EINVAL); cq = kzalloc(sizeof(*cq), GFP_KERNEL); @@ -927,7 +926,7 @@ int mlx5_ib_modify_cq(struct ib_cq *cq, u16 cq_count, u16 cq_period) int err; u32 fsel; - if (!(dev->mdev->caps.gen.flags & MLX5_DEV_CAP_FLAG_CQ_MODER)) + if (!MLX5_CAP_GEN(dev->mdev, cq_moderation)) return -ENOSYS; in = kzalloc(sizeof(*in), GFP_KERNEL); @@ -1082,7 +1081,7 @@ int mlx5_ib_resize_cq(struct ib_cq *ibcq, int entries, struct ib_udata *udata) int uninitialized_var(cqe_size); unsigned long flags; - if (!(dev->mdev->caps.gen.flags & MLX5_DEV_CAP_FLAG_RESIZE_CQ)) { + if (!MLX5_CAP_GEN(dev->mdev, cq_resize)) { pr_info("Firmware does not support resize CQ\n"); return -ENOSYS; } @@ -1091,7 +1090,7 @@ int mlx5_ib_resize_cq(struct ib_cq *ibcq, int entries, struct ib_udata *udata) return -EINVAL; entries = roundup_pow_of_two(entries + 1); - if (entries > dev->mdev->caps.gen.max_cqes + 1) + if (entries > (1 << MLX5_CAP_GEN(dev->mdev, log_max_cq_sz)) + 1) return -EINVAL; if (entries == ibcq->cqe + 1) diff --git a/drivers/infiniband/hw/mlx5/mad.c b/drivers/infiniband/hw/mlx5/mad.c index 8e45714fa369..01fc97db45d6 100644 --- a/drivers/infiniband/hw/mlx5/mad.c +++ b/drivers/infiniband/hw/mlx5/mad.c @@ -136,7 +136,7 @@ int mlx5_query_ext_port_caps(struct mlx5_ib_dev *dev, u8 port) packet_error = be16_to_cpu(out_mad->status); - dev->mdev->caps.gen.ext_port_cap[port - 1] = (!err && !packet_error) ? + dev->mdev->port_caps[port - 1].ext_port_cap = (!err && !packet_error) ? MLX_EXT_PORT_CAP_FLAG_EXTENDED_PORT_INFO : 0; out: @@ -144,3 +144,300 @@ out: 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, 64); +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/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c index c6cb26e0c866..085c24b4b603 100644 --- a/drivers/infiniband/hw/mlx5/main.c +++ b/drivers/infiniband/hw/mlx5/main.c @@ -40,6 +40,7 @@ #include <linux/io-mapping.h> #include <linux/sched.h> #include <rdma/ib_user_verbs.h> +#include <linux/mlx5/vport.h> #include <rdma/ib_smi.h> #include <rdma/ib_umem.h> #include "user.h" @@ -62,36 +63,172 @@ static char mlx5_version[] = DRIVER_NAME ": Mellanox Connect-IB Infiniband driver v" DRIVER_VERSION " (" DRIVER_RELDATE ")\n"; +static enum rdma_link_layer +mlx5_ib_port_link_layer(struct ib_device *device) +{ + struct mlx5_ib_dev *dev = to_mdev(device); + + switch (MLX5_CAP_GEN(dev->mdev, port_type)) { + 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 int mlx5_use_mad_ifc(struct mlx5_ib_dev *dev) +{ + return !dev->mdev->issi; +} + +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) == + IB_LINK_LAYER_ETHERNET) + return MLX5_VPORT_ACCESS_METHOD_NIC; + + return MLX5_VPORT_ACCESS_METHOD_HCA; +} + +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); + if (!err) + *sys_image_guid = cpu_to_be64(tmp); + return err; + + default: + return -EINVAL; + } +} + +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); + if (!err) + *node_guid = cpu_to_be64(tmp); + return err; + + default: + return -EINVAL; + } +} + +struct mlx5_reg_node_desc { + u8 desc[64]; +}; + +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 ib_smp *in_mad = NULL; - struct ib_smp *out_mad = NULL; - struct mlx5_general_caps *gen; + struct mlx5_core_dev *mdev = dev->mdev; int err = -ENOMEM; int max_rq_sg; int max_sq_sg; - u64 flags; if (uhw->inlen || uhw->outlen) return -EINVAL; - gen = &dev->mdev->caps.gen; - 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; + memset(props, 0, sizeof(*props)); + err = mlx5_query_system_image_guid(ibdev, + &props->sys_image_guid); + if (err) + return err; - err = mlx5_MAD_IFC(to_mdev(ibdev), 1, 1, 1, NULL, NULL, in_mad, out_mad); + err = mlx5_query_max_pkeys(ibdev, &props->max_pkeys); if (err) - goto out; + return err; - memset(props, 0, sizeof(*props)); + err = mlx5_query_vendor_id(ibdev, &props->vendor_id); + if (err) + return err; props->fw_ver = ((u64)fw_rev_maj(dev->mdev) << 32) | (fw_rev_min(dev->mdev) << 16) | @@ -100,18 +237,18 @@ static int mlx5_ib_query_device(struct ib_device *ibdev, IB_DEVICE_PORT_ACTIVE_EVENT | IB_DEVICE_SYS_IMAGE_GUID | IB_DEVICE_RC_RNR_NAK_GEN; - flags = gen->flags; - if (flags & MLX5_DEV_CAP_FLAG_BAD_PKEY_CNTR) + + if (MLX5_CAP_GEN(mdev, pkv)) props->device_cap_flags |= IB_DEVICE_BAD_PKEY_CNTR; - if (flags & MLX5_DEV_CAP_FLAG_BAD_QKEY_CNTR) + if (MLX5_CAP_GEN(mdev, qkv)) props->device_cap_flags |= IB_DEVICE_BAD_QKEY_CNTR; - if (flags & MLX5_DEV_CAP_FLAG_APM) + if (MLX5_CAP_GEN(mdev, apm)) props->device_cap_flags |= IB_DEVICE_AUTO_PATH_MIG; props->device_cap_flags |= IB_DEVICE_LOCAL_DMA_LKEY; - if (flags & MLX5_DEV_CAP_FLAG_XRC) + if (MLX5_CAP_GEN(mdev, xrc)) props->device_cap_flags |= IB_DEVICE_XRC; props->device_cap_flags |= IB_DEVICE_MEM_MGT_EXTENSIONS; - if (flags & MLX5_DEV_CAP_FLAG_SIG_HAND_OVER) { + 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 | @@ -120,221 +257,270 @@ static int mlx5_ib_query_device(struct ib_device *ibdev, props->sig_guard_cap = IB_GUARD_T10DIF_CRC | IB_GUARD_T10DIF_CSUM; } - if (flags & MLX5_DEV_CAP_FLAG_BLOCK_MCAST) + if (MLX5_CAP_GEN(mdev, block_lb_mc)) props->device_cap_flags |= IB_DEVICE_BLOCK_MULTICAST_LOOPBACK; - props->vendor_id = be32_to_cpup((__be32 *)(out_mad->data + 36)) & - 0xffffff; - props->vendor_part_id = be16_to_cpup((__be16 *)(out_mad->data + 30)); - props->hw_ver = be32_to_cpup((__be32 *)(out_mad->data + 32)); - memcpy(&props->sys_image_guid, out_mad->data + 4, 8); + props->vendor_part_id = mdev->pdev->device; + props->hw_ver = mdev->pdev->revision; props->max_mr_size = ~0ull; - props->page_size_cap = gen->min_page_sz; - props->max_qp = 1 << gen->log_max_qp; - props->max_qp_wr = gen->max_wqes; - max_rq_sg = gen->max_rq_desc_sz / sizeof(struct mlx5_wqe_data_seg); - max_sq_sg = (gen->max_sq_desc_sz - sizeof(struct mlx5_wqe_ctrl_seg)) / - sizeof(struct mlx5_wqe_data_seg); + props->page_size_cap = 1ull << MLX5_CAP_GEN(mdev, log_pg_sz); + 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_sg = (MLX5_CAP_GEN(mdev, max_wqe_sz_sq) - + sizeof(struct mlx5_wqe_ctrl_seg)) / + sizeof(struct mlx5_wqe_data_seg); props->max_sge = min(max_rq_sg, max_sq_sg); - props->max_cq = 1 << gen->log_max_cq; - props->max_cqe = gen->max_cqes - 1; - props->max_mr = 1 << gen->log_max_mkey; - props->max_pd = 1 << gen->log_max_pd; - props->max_qp_rd_atom = 1 << gen->log_max_ra_req_qp; - props->max_qp_init_rd_atom = 1 << gen->log_max_ra_res_qp; - props->max_srq = 1 << gen->log_max_srq; - props->max_srq_wr = gen->max_srq_wqes - 1; - props->local_ca_ack_delay = gen->local_ca_ack_delay; + props->max_cq = 1 << MLX5_CAP_GEN(mdev, log_max_cq); + props->max_cqe = (1 << MLX5_CAP_GEN(mdev, log_max_eq_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 = (unsigned int)-1; - props->local_ca_ack_delay = gen->local_ca_ack_delay; props->atomic_cap = IB_ATOMIC_NONE; props->masked_atomic_cap = IB_ATOMIC_NONE; - props->max_pkeys = be16_to_cpup((__be16 *)(out_mad->data + 28)); - props->max_mcast_grp = 1 << gen->log_max_mcg; - props->max_mcast_qp_attach = gen->max_qp_mcg; + 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 */ #ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING - if (dev->mdev->caps.gen.flags & MLX5_DEV_CAP_FLAG_ON_DMND_PG) + if (MLX5_CAP_GEN(mdev, pg)) props->device_cap_flags |= IB_DEVICE_ON_DEMAND_PAGING; props->odp_caps = dev->odp_caps; #endif -out: - kfree(in_mad); - kfree(out_mad); - - return err; + return 0; } -int mlx5_ib_query_port(struct ib_device *ibdev, u8 port, - struct ib_port_attr *props) +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); - struct ib_smp *in_mad = NULL; - struct ib_smp *out_mad = NULL; - struct mlx5_general_caps *gen; - int ext_active_speed; - int err = -ENOMEM; - - gen = &dev->mdev->caps.gen; - if (port < 1 || port > gen->num_ports) { - mlx5_ib_warn(dev, "invalid port number %d\n", port); - return -EINVAL; + int err = 0; + + if (active_width & MLX5_IB_WIDTH_1X) { + *ib_width = IB_WIDTH_1X; + } else if (active_width & MLX5_IB_WIDTH_2X) { + mlx5_ib_dbg(dev, "active_width %d is not supported by IB spec\n", + (int)active_width); + err = -EINVAL; + } 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; } - 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); + return err; +} - 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; +static int mlx5_mtu_to_ib_mtu(int mtu) +{ + switch (mtu) { + case 256: return 1; + case 512: return 2; + case 1024: return 3; + case 2048: return 4; + case 4096: return 5; + default: + pr_warn("invalid mtu\n"); + return -1; } +} +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, +}; - 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 << gen->log_max_msg; - props->pkey_tbl_len = gen->port[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; - } - } +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 +}; - /* If reported active speed is QDR, check if is FDR-10 */ - if (props->active_speed == 4) { - if (gen->ext_port_cap[port - 1] & - 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; - } - } +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; -out: - kfree(in_mad); - kfree(out_mad); + default: + return -EINVAL; + } - return err; + return 0; } -static int mlx5_ib_query_gid(struct ib_device *ibdev, u8 port, int index, - union ib_gid *gid) +static int mlx5_query_hca_port(struct ib_device *ibdev, u8 port, + struct ib_port_attr *props) { - struct ib_smp *in_mad = NULL; - struct ib_smp *out_mad = NULL; - int err = -ENOMEM; + struct mlx5_ib_dev *dev = to_mdev(ibdev); + struct mlx5_core_dev *mdev = dev->mdev; + struct mlx5_hca_vport_context *rep; + int max_mtu; + int oper_mtu; + int err; + u8 ib_link_width_oper; + u8 vl_hw_cap; - in_mad = kzalloc(sizeof(*in_mad), GFP_KERNEL); - out_mad = kmalloc(sizeof(*out_mad), GFP_KERNEL); - if (!in_mad || !out_mad) + rep = kzalloc(sizeof(*rep), GFP_KERNEL); + if (!rep) { + err = -ENOMEM; goto out; + } - init_query_mad(in_mad); - in_mad->attr_id = IB_SMP_ATTR_PORT_INFO; - in_mad->attr_mod = cpu_to_be32(port); + memset(props, 0, sizeof(*props)); - err = mlx5_MAD_IFC(to_mdev(ibdev), 1, 1, port, NULL, NULL, in_mad, out_mad); + err = mlx5_query_hca_vport_context(mdev, 0, port, 0, rep); 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); + props->lid = rep->lid; + props->lmc = rep->lmc; + props->sm_lid = rep->sm_lid; + props->sm_sl = rep->sm_sl; + props->state = rep->vport_state; + props->phys_state = rep->port_physical_state; + props->port_cap_flags = rep->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 = rep->pkey_violation_counter; + props->qkey_viol_cntr = rep->qkey_violation_counter; + props->subnet_timeout = rep->subnet_timeout; + props->init_type_reply = rep->init_type_reply; + + err = mlx5_query_port_link_width_oper(mdev, &ib_link_width_oper, port); + if (err) + goto out; - err = mlx5_MAD_IFC(to_mdev(ibdev), 1, 1, port, NULL, NULL, in_mad, out_mad); + err = translate_active_width(ibdev, ib_link_width_oper, + &props->active_width); + if (err) + goto out; + err = mlx5_query_port_proto_oper(mdev, &props->active_speed, MLX5_PTYS_IB, + port); if (err) goto out; - memcpy(gid->raw + 8, out_mad->data + (index % 8) * 8, 8); + mlx5_query_port_max_mtu(mdev, &max_mtu, port); + + props->max_mtu = mlx5_mtu_to_ib_mtu(max_mtu); + + mlx5_query_port_oper_mtu(mdev, &oper_mtu, port); + + props->active_mtu = mlx5_mtu_to_ib_mtu(oper_mtu); + err = mlx5_query_port_vl_hw_cap(mdev, &vl_hw_cap, port); + if (err) + goto out; + + err = translate_max_vl_num(ibdev, vl_hw_cap, + &props->max_vl_num); out: - kfree(in_mad); - kfree(out_mad); + kfree(rep); return err; } -static int mlx5_ib_query_pkey(struct ib_device *ibdev, u8 port, u16 index, - u16 *pkey) +int mlx5_ib_query_port(struct ib_device *ibdev, u8 port, + struct ib_port_attr *props) { - struct ib_smp *in_mad = NULL; - struct ib_smp *out_mad = NULL; - int err = -ENOMEM; + switch (mlx5_get_vport_access_method(ibdev)) { + case MLX5_VPORT_ACCESS_METHOD_MAD: + return mlx5_query_mad_ifc_port(ibdev, port, props); - in_mad = kzalloc(sizeof(*in_mad), GFP_KERNEL); - out_mad = kmalloc(sizeof(*out_mad), GFP_KERNEL); - if (!in_mad || !out_mad) - goto out; + case MLX5_VPORT_ACCESS_METHOD_HCA: + return mlx5_query_hca_port(ibdev, port, props); + + default: + return -EINVAL; + } +} - init_query_mad(in_mad); - in_mad->attr_id = IB_SMP_ATTR_PKEY_TABLE; - in_mad->attr_mod = cpu_to_be32(index / 32); +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; - err = mlx5_MAD_IFC(to_mdev(ibdev), 1, 1, port, NULL, NULL, in_mad, out_mad); - if (err) - goto out; + switch (mlx5_get_vport_access_method(ibdev)) { + case MLX5_VPORT_ACCESS_METHOD_MAD: + return mlx5_query_mad_ifc_gids(ibdev, port, index, gid); - *pkey = be16_to_cpu(((__be16 *)out_mad->data)[index % 32]); + case MLX5_VPORT_ACCESS_METHOD_HCA: + return mlx5_query_hca_vport_gid(mdev, 0, port, 0, index, gid); + + default: + return -EINVAL; + } -out: - kfree(in_mad); - kfree(out_mad); - return err; } -struct mlx5_reg_node_desc { - u8 desc[64]; -}; +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) @@ -396,7 +582,6 @@ static struct ib_ucontext *mlx5_ib_alloc_ucontext(struct ib_device *ibdev, struct mlx5_ib_alloc_ucontext_req_v2 req; struct mlx5_ib_alloc_ucontext_resp resp; struct mlx5_ib_ucontext *context; - struct mlx5_general_caps *gen; struct mlx5_uuar_info *uuari; struct mlx5_uar *uars; int gross_uuars; @@ -407,7 +592,6 @@ static struct ib_ucontext *mlx5_ib_alloc_ucontext(struct ib_device *ibdev, int i; size_t reqlen; - gen = &dev->mdev->caps.gen; if (!dev->ib_active) return ERR_PTR(-EAGAIN); @@ -440,14 +624,14 @@ static struct ib_ucontext *mlx5_ib_alloc_ucontext(struct ib_device *ibdev, num_uars = req.total_num_uuars / MLX5_NON_FP_BF_REGS_PER_PAGE; gross_uuars = num_uars * MLX5_BF_REGS_PER_PAGE; - resp.qp_tab_size = 1 << gen->log_max_qp; - resp.bf_reg_size = gen->bf_reg_size; - resp.cache_line_size = L1_CACHE_BYTES; - resp.max_sq_desc_sz = gen->max_sq_desc_sz; - resp.max_rq_desc_sz = gen->max_rq_desc_sz; - resp.max_send_wqebb = gen->max_wqes; - resp.max_recv_wr = gen->max_wqes; - resp.max_srq_recv_wr = gen->max_srq_wqes; + resp.qp_tab_size = 1 << MLX5_CAP_GEN(dev->mdev, log_max_qp); + resp.bf_reg_size = 1 << MLX5_CAP_GEN(dev->mdev, log_bf_reg_size); + resp.cache_line_size = L1_CACHE_BYTES; + 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); context = kzalloc(sizeof(*context), GFP_KERNEL); if (!context) @@ -497,7 +681,7 @@ static struct ib_ucontext *mlx5_ib_alloc_ucontext(struct ib_device *ibdev, mutex_init(&context->db_page_mutex); resp.tot_uuars = req.total_num_uuars; - resp.num_ports = gen->num_ports; + resp.num_ports = MLX5_CAP_GEN(dev->mdev, num_ports); err = ib_copy_to_udata(udata, &resp, sizeof(resp) - sizeof(resp.reserved)); if (err) @@ -735,37 +919,15 @@ static int mlx5_ib_mcg_detach(struct ib_qp *ibqp, union ib_gid *gid, u16 lid) static int init_node_data(struct mlx5_ib_dev *dev) { - 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(dev->ib_dev.node_desc, out_mad->data, 64); - - in_mad->attr_id = IB_SMP_ATTR_NODE_INFO; + int err; - err = mlx5_MAD_IFC(dev, 1, 1, 1, NULL, NULL, in_mad, out_mad); + err = mlx5_query_node_desc(dev, dev->ib_dev.node_desc); if (err) - goto out; + return err; - dev->mdev->rev_id = be32_to_cpup((__be32 *)(out_mad->data + 32)); - memcpy(&dev->ib_dev.node_guid, out_mad->data + 12, 8); + dev->mdev->rev_id = dev->mdev->pdev->revision; -out: - kfree(in_mad); - kfree(out_mad); - 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, @@ -899,11 +1061,9 @@ static void mlx5_ib_event(struct mlx5_core_dev *dev, void *context, static void get_ext_port_caps(struct mlx5_ib_dev *dev) { - struct mlx5_general_caps *gen; int port; - gen = &dev->mdev->caps.gen; - for (port = 1; port <= gen->num_ports; port++) + for (port = 1; port <= MLX5_CAP_GEN(dev->mdev, num_ports); port++) mlx5_query_ext_port_caps(dev, port); } @@ -911,12 +1071,10 @@ static int get_port_caps(struct mlx5_ib_dev *dev) { struct ib_device_attr *dprops = NULL; struct ib_port_attr *pprops = NULL; - struct mlx5_general_caps *gen; int err = -ENOMEM; int port; struct ib_udata uhw = {.inlen = 0, .outlen = 0}; - gen = &dev->mdev->caps.gen; pprops = kmalloc(sizeof(*pprops), GFP_KERNEL); if (!pprops) goto out; @@ -931,14 +1089,17 @@ static int get_port_caps(struct mlx5_ib_dev *dev) goto out; } - for (port = 1; port <= gen->num_ports; port++) { + 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); + mlx5_ib_warn(dev, "query_port %d failed %d\n", + port, err); break; } - gen->port[port - 1].pkey_table_len = dprops->max_pkeys; - gen->port[port - 1].gid_table_len = pprops->gid_tbl_len; + 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); } @@ -1167,8 +1328,29 @@ static int create_dev_resources(struct mlx5_ib_resources *devr) 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); + return 0; +error5: + mlx5_ib_destroy_srq(devr->s0); error4: mlx5_ib_dealloc_xrcd(devr->x1); error3: @@ -1183,6 +1365,7 @@ error0: static void destroy_dev_resources(struct mlx5_ib_resources *devr) { + mlx5_ib_destroy_srq(devr->s1); mlx5_ib_destroy_srq(devr->s0); mlx5_ib_dealloc_xrcd(devr->x0); mlx5_ib_dealloc_xrcd(devr->x1); @@ -1214,6 +1397,10 @@ static void *mlx5_ib_add(struct mlx5_core_dev *mdev) int err; int i; + /* don't create IB instance over Eth ports, no RoCE yet! */ + if (MLX5_CAP_GEN(mdev, port_type) == MLX5_CAP_PORT_TYPE_ETH) + return NULL; + printk_once(KERN_INFO "%s", mlx5_version); dev = (struct mlx5_ib_dev *)ib_alloc_device(sizeof(*dev)); @@ -1226,15 +1413,16 @@ static void *mlx5_ib_add(struct mlx5_core_dev *mdev) if (err) goto err_dealloc; - get_ext_port_caps(dev); + if (mlx5_use_mad_ifc(dev)) + get_ext_port_caps(dev); MLX5_INIT_DOORBELL_LOCK(&dev->uar_lock); strlcpy(dev->ib_dev.name, "mlx5_%d", IB_DEVICE_NAME_MAX); dev->ib_dev.owner = THIS_MODULE; dev->ib_dev.node_type = RDMA_NODE_IB_CA; - dev->ib_dev.local_dma_lkey = mdev->caps.gen.reserved_lkey; - dev->num_ports = mdev->caps.gen.num_ports; + 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; @@ -1313,9 +1501,9 @@ static void *mlx5_ib_add(struct mlx5_core_dev *mdev) dev->ib_dev.check_mr_status = mlx5_ib_check_mr_status; dev->ib_dev.get_port_immutable = mlx5_port_immutable; - mlx5_ib_internal_query_odp_caps(dev); + mlx5_ib_internal_fill_odp_caps(dev); - if (mdev->caps.gen.flags & MLX5_DEV_CAP_FLAG_XRC) { + 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 |= diff --git a/drivers/infiniband/hw/mlx5/mlx5_ib.h b/drivers/infiniband/hw/mlx5/mlx5_ib.h index 178314e764da..7cae09836481 100644 --- a/drivers/infiniband/hw/mlx5/mlx5_ib.h +++ b/drivers/infiniband/hw/mlx5/mlx5_ib.h @@ -415,6 +415,7 @@ struct mlx5_ib_resources { struct ib_xrcd *x1; struct ib_pd *p0; struct ib_srq *s0; + struct ib_srq *s1; }; struct mlx5_ib_dev { @@ -597,6 +598,22 @@ struct ib_xrcd *mlx5_ib_alloc_xrcd(struct ib_device *ibdev, int mlx5_ib_dealloc_xrcd(struct ib_xrcd *xrcd); int mlx5_ib_get_buf_offset(u64 addr, int page_shift, u32 *offset); int mlx5_query_ext_port_caps(struct mlx5_ib_dev *dev, u8 port); +int mlx5_query_mad_ifc_smp_attr_node_info(struct ib_device *ibdev, + struct ib_smp *out_mad); +int mlx5_query_mad_ifc_system_image_guid(struct ib_device *ibdev, + __be64 *sys_image_guid); +int mlx5_query_mad_ifc_max_pkeys(struct ib_device *ibdev, + u16 *max_pkeys); +int mlx5_query_mad_ifc_vendor_id(struct ib_device *ibdev, + u32 *vendor_id); +int mlx5_query_mad_ifc_node_desc(struct mlx5_ib_dev *dev, char *node_desc); +int mlx5_query_mad_ifc_node_guid(struct mlx5_ib_dev *dev, __be64 *node_guid); +int mlx5_query_mad_ifc_pkey(struct ib_device *ibdev, u8 port, u16 index, + u16 *pkey); +int mlx5_query_mad_ifc_gids(struct ib_device *ibdev, u8 port, int index, + union ib_gid *gid); +int mlx5_query_mad_ifc_port(struct ib_device *ibdev, u8 port, + struct ib_port_attr *props); int mlx5_ib_query_port(struct ib_device *ibdev, u8 port, struct ib_port_attr *props); int mlx5_ib_init_fmr(struct mlx5_ib_dev *dev); @@ -620,7 +637,7 @@ int mlx5_ib_check_mr_status(struct ib_mr *ibmr, u32 check_mask, #ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING extern struct workqueue_struct *mlx5_ib_page_fault_wq; -int mlx5_ib_internal_query_odp_caps(struct mlx5_ib_dev *dev); +void mlx5_ib_internal_fill_odp_caps(struct mlx5_ib_dev *dev); void mlx5_ib_mr_pfault_handler(struct mlx5_ib_qp *qp, struct mlx5_ib_pfault *pfault); void mlx5_ib_odp_create_qp(struct mlx5_ib_qp *qp); @@ -634,9 +651,9 @@ void mlx5_ib_invalidate_range(struct ib_umem *umem, unsigned long start, unsigned long end); #else /* CONFIG_INFINIBAND_ON_DEMAND_PAGING */ -static inline int mlx5_ib_internal_query_odp_caps(struct mlx5_ib_dev *dev) +static inline void mlx5_ib_internal_fill_odp_caps(struct mlx5_ib_dev *dev) { - return 0; + return; } static inline void mlx5_ib_odp_create_qp(struct mlx5_ib_qp *qp) {} diff --git a/drivers/infiniband/hw/mlx5/mr.c b/drivers/infiniband/hw/mlx5/mr.c index 71c593583864..bc9a0de897cb 100644 --- a/drivers/infiniband/hw/mlx5/mr.c +++ b/drivers/infiniband/hw/mlx5/mr.c @@ -975,8 +975,7 @@ static struct mlx5_ib_mr *reg_create(struct ib_pd *pd, u64 virt_addr, struct mlx5_ib_mr *mr; int inlen; int err; - bool pg_cap = !!(dev->mdev->caps.gen.flags & - MLX5_DEV_CAP_FLAG_ON_DMND_PG); + bool pg_cap = !!(MLX5_CAP_GEN(dev->mdev, pg)); mr = kzalloc(sizeof(*mr), GFP_KERNEL); if (!mr) diff --git a/drivers/infiniband/hw/mlx5/odp.c b/drivers/infiniband/hw/mlx5/odp.c index 5099db08afd2..aa8391e75385 100644 --- a/drivers/infiniband/hw/mlx5/odp.c +++ b/drivers/infiniband/hw/mlx5/odp.c @@ -109,40 +109,33 @@ void mlx5_ib_invalidate_range(struct ib_umem *umem, unsigned long start, ib_umem_odp_unmap_dma_pages(umem, start, end); } -#define COPY_ODP_BIT_MLX_TO_IB(reg, ib_caps, field_name, bit_name) do { \ - if (be32_to_cpu(reg.field_name) & MLX5_ODP_SUPPORT_##bit_name) \ - ib_caps->field_name |= IB_ODP_SUPPORT_##bit_name; \ -} while (0) - -int mlx5_ib_internal_query_odp_caps(struct mlx5_ib_dev *dev) +void mlx5_ib_internal_fill_odp_caps(struct mlx5_ib_dev *dev) { - int err; - struct mlx5_odp_caps hw_caps; struct ib_odp_caps *caps = &dev->odp_caps; memset(caps, 0, sizeof(*caps)); - if (!(dev->mdev->caps.gen.flags & MLX5_DEV_CAP_FLAG_ON_DMND_PG)) - return 0; - - err = mlx5_query_odp_caps(dev->mdev, &hw_caps); - if (err) - goto out; + if (!MLX5_CAP_GEN(dev->mdev, pg)) + return; caps->general_caps = IB_ODP_SUPPORT; - COPY_ODP_BIT_MLX_TO_IB(hw_caps, caps, per_transport_caps.ud_odp_caps, - SEND); - COPY_ODP_BIT_MLX_TO_IB(hw_caps, caps, per_transport_caps.rc_odp_caps, - SEND); - COPY_ODP_BIT_MLX_TO_IB(hw_caps, caps, per_transport_caps.rc_odp_caps, - RECV); - COPY_ODP_BIT_MLX_TO_IB(hw_caps, caps, per_transport_caps.rc_odp_caps, - WRITE); - COPY_ODP_BIT_MLX_TO_IB(hw_caps, caps, per_transport_caps.rc_odp_caps, - READ); - -out: - return err; + + if (MLX5_CAP_ODP(dev->mdev, ud_odp_caps.send)) + caps->per_transport_caps.ud_odp_caps |= IB_ODP_SUPPORT_SEND; + + if (MLX5_CAP_ODP(dev->mdev, rc_odp_caps.send)) + caps->per_transport_caps.rc_odp_caps |= IB_ODP_SUPPORT_SEND; + + if (MLX5_CAP_ODP(dev->mdev, rc_odp_caps.receive)) + caps->per_transport_caps.rc_odp_caps |= IB_ODP_SUPPORT_RECV; + + if (MLX5_CAP_ODP(dev->mdev, rc_odp_caps.write)) + caps->per_transport_caps.rc_odp_caps |= IB_ODP_SUPPORT_WRITE; + + if (MLX5_CAP_ODP(dev->mdev, rc_odp_caps.read)) + caps->per_transport_caps.rc_odp_caps |= IB_ODP_SUPPORT_READ; + + return; } static struct mlx5_ib_mr *mlx5_ib_odp_find_mr_lkey(struct mlx5_ib_dev *dev, diff --git a/drivers/infiniband/hw/mlx5/qp.c b/drivers/infiniband/hw/mlx5/qp.c index d35f62d4f4c5..203c8a45e095 100644 --- a/drivers/infiniband/hw/mlx5/qp.c +++ b/drivers/infiniband/hw/mlx5/qp.c @@ -220,13 +220,11 @@ static void mlx5_ib_qp_event(struct mlx5_core_qp *qp, int type) 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) { - struct mlx5_general_caps *gen; int wqe_size; int wq_size; - gen = &dev->mdev->caps.gen; /* Sanity check RQ size before proceeding */ - if (cap->max_recv_wr > gen->max_wqes) + if (cap->max_recv_wr > (1 << MLX5_CAP_GEN(dev->mdev, log_max_qp_sz))) return -EINVAL; if (!has_rq) { @@ -246,10 +244,11 @@ static int set_rq_size(struct mlx5_ib_dev *dev, struct ib_qp_cap *cap, 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 > gen->max_rq_desc_sz) { + if (wqe_size > MLX5_CAP_GEN(dev->mdev, max_wqe_sz_rq)) { mlx5_ib_dbg(dev, "wqe_size %d, max %d\n", wqe_size, - gen->max_rq_desc_sz); + MLX5_CAP_GEN(dev->mdev, + max_wqe_sz_rq)); return -EINVAL; } qp->rq.wqe_shift = ilog2(wqe_size); @@ -330,11 +329,9 @@ static int calc_send_wqe(struct ib_qp_init_attr *attr) static int calc_sq_size(struct mlx5_ib_dev *dev, struct ib_qp_init_attr *attr, struct mlx5_ib_qp *qp) { - struct mlx5_general_caps *gen; int wqe_size; int wq_size; - gen = &dev->mdev->caps.gen; if (!attr->cap.max_send_wr) return 0; @@ -343,9 +340,9 @@ static int calc_sq_size(struct mlx5_ib_dev *dev, struct ib_qp_init_attr *attr, if (wqe_size < 0) return wqe_size; - if (wqe_size > gen->max_sq_desc_sz) { + 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, gen->max_sq_desc_sz); + wqe_size, MLX5_CAP_GEN(dev->mdev, max_wqe_sz_sq)); return -EINVAL; } @@ -358,9 +355,10 @@ static int calc_sq_size(struct mlx5_ib_dev *dev, struct ib_qp_init_attr *attr, 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 > gen->max_wqes) { + 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, gen->max_wqes); + 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); @@ -375,13 +373,11 @@ static int set_user_buf_size(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, struct mlx5_ib_create_qp *ucmd) { - struct mlx5_general_caps *gen; int desc_sz = 1 << qp->sq.wqe_shift; - gen = &dev->mdev->caps.gen; - if (desc_sz > gen->max_sq_desc_sz) { + 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, gen->max_sq_desc_sz); + desc_sz, MLX5_CAP_GEN(dev->mdev, max_wqe_sz_sq)); return -EINVAL; } @@ -393,9 +389,10 @@ static int set_user_buf_size(struct mlx5_ib_dev *dev, qp->sq.wqe_cnt = ucmd->sq_wqe_count; - if (qp->sq.wqe_cnt > gen->max_wqes) { + 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, gen->max_wqes); + qp->sq.wqe_cnt, + 1 << MLX5_CAP_GEN(dev->mdev, log_max_qp_sz)); return -EINVAL; } @@ -768,7 +765,7 @@ static int create_kernel_qp(struct mlx5_ib_dev *dev, qp->sq.offset = qp->rq.wqe_cnt << qp->rq.wqe_shift; qp->buf_size = err + (qp->rq.wqe_cnt << qp->rq.wqe_shift); - err = mlx5_buf_alloc(dev->mdev, qp->buf_size, PAGE_SIZE * 2, &qp->buf); + err = mlx5_buf_alloc(dev->mdev, qp->buf_size, &qp->buf); if (err) { mlx5_ib_dbg(dev, "err %d\n", err); goto err_uuar; @@ -866,22 +863,21 @@ static int create_qp_common(struct mlx5_ib_dev *dev, struct ib_pd *pd, struct ib_udata *udata, struct mlx5_ib_qp *qp) { struct mlx5_ib_resources *devr = &dev->devr; + struct mlx5_core_dev *mdev = dev->mdev; struct mlx5_ib_create_qp_resp resp; struct mlx5_create_qp_mbox_in *in; - struct mlx5_general_caps *gen; struct mlx5_ib_create_qp ucmd; int inlen = sizeof(*in); int err; mlx5_ib_odp_create_qp(qp); - gen = &dev->mdev->caps.gen; mutex_init(&qp->mutex); spin_lock_init(&qp->sq.lock); spin_lock_init(&qp->rq.lock); if (init_attr->create_flags & IB_QP_CREATE_BLOCK_MULTICAST_LOOPBACK) { - if (!(gen->flags & MLX5_DEV_CAP_FLAG_BLOCK_MCAST)) { + if (!MLX5_CAP_GEN(mdev, block_lb_mc)) { mlx5_ib_dbg(dev, "block multicast loopback isn't supported\n"); return -EINVAL; } else { @@ -914,15 +910,17 @@ static int create_qp_common(struct mlx5_ib_dev *dev, struct ib_pd *pd, 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 > gen->max_wqes) { + if (ucmd.sq_wqe_count > max_wqes) { mlx5_ib_dbg(dev, "requested sq_wqe_count (%d) > max allowed (%d)\n", - ucmd.sq_wqe_count, gen->max_wqes); + ucmd.sq_wqe_count, max_wqes); return -EINVAL; } err = create_user_qp(dev, pd, qp, udata, &in, &resp, &inlen); @@ -1014,7 +1012,8 @@ static int create_qp_common(struct mlx5_ib_dev *dev, struct ib_pd *pd, in->ctx.rq_type_srqn |= cpu_to_be32(to_msrq(init_attr->srq)->msrq.srqn); } else { in->ctx.xrcd = cpu_to_be32(to_mxrcd(devr->x1)->xrcdn); - in->ctx.rq_type_srqn |= cpu_to_be32(to_msrq(devr->s0)->msrq.srqn); + in->ctx.rq_type_srqn |= + cpu_to_be32(to_msrq(devr->s1)->msrq.srqn); } } @@ -1226,7 +1225,6 @@ struct ib_qp *mlx5_ib_create_qp(struct ib_pd *pd, struct ib_qp_init_attr *init_attr, struct ib_udata *udata) { - struct mlx5_general_caps *gen; struct mlx5_ib_dev *dev; struct mlx5_ib_qp *qp; u16 xrcdn = 0; @@ -1244,12 +1242,11 @@ struct ib_qp *mlx5_ib_create_qp(struct ib_pd *pd, } dev = to_mdev(to_mxrcd(init_attr->xrcd)->ibxrcd.device); } - gen = &dev->mdev->caps.gen; switch (init_attr->qp_type) { case IB_QPT_XRC_TGT: case IB_QPT_XRC_INI: - if (!(gen->flags & MLX5_DEV_CAP_FLAG_XRC)) { + if (!MLX5_CAP_GEN(dev->mdev, xrc)) { mlx5_ib_dbg(dev, "XRC not supported\n"); return ERR_PTR(-ENOSYS); } @@ -1356,9 +1353,6 @@ enum { static int ib_rate_to_mlx5(struct mlx5_ib_dev *dev, u8 rate) { - struct mlx5_general_caps *gen; - - gen = &dev->mdev->caps.gen; if (rate == IB_RATE_PORT_CURRENT) { return 0; } else if (rate < IB_RATE_2_5_GBPS || rate > IB_RATE_300_GBPS) { @@ -1366,7 +1360,7 @@ static int ib_rate_to_mlx5(struct mlx5_ib_dev *dev, u8 rate) } else { while (rate != IB_RATE_2_5_GBPS && !(1 << (rate + MLX5_STAT_RATE_OFFSET) & - gen->stat_rate_support)) + MLX5_CAP_GEN(dev->mdev, stat_rate_support))) --rate; } @@ -1377,10 +1371,8 @@ static int mlx5_set_path(struct mlx5_ib_dev *dev, 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) { - struct mlx5_general_caps *gen; int err; - gen = &dev->mdev->caps.gen; path->fl = (path_flags & MLX5_PATH_FLAG_FL) ? 0x80 : 0; path->free_ar = (path_flags & MLX5_PATH_FLAG_FREE_AR) ? 0x80 : 0; @@ -1391,9 +1383,11 @@ static int mlx5_set_path(struct mlx5_ib_dev *dev, const struct ib_ah_attr *ah, path->rlid = cpu_to_be16(ah->dlid); if (ah->ah_flags & IB_AH_GRH) { - if (ah->grh.sgid_index >= gen->port[port - 1].gid_table_len) { + 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, gen->port[port - 1].gid_table_len); + ah->grh.sgid_index, + dev->mdev->port_caps[port - 1].gid_table_len); return -EINVAL; } path->grh_mlid |= 1 << 7; @@ -1570,7 +1564,6 @@ static int __mlx5_ib_modify_qp(struct ib_qp *ibqp, struct mlx5_ib_qp *qp = to_mqp(ibqp); struct mlx5_ib_cq *send_cq, *recv_cq; struct mlx5_qp_context *context; - struct mlx5_general_caps *gen; struct mlx5_modify_qp_mbox_in *in; struct mlx5_ib_pd *pd; enum mlx5_qp_state mlx5_cur, mlx5_new; @@ -1579,7 +1572,6 @@ static int __mlx5_ib_modify_qp(struct ib_qp *ibqp, int mlx5_st; int err; - gen = &dev->mdev->caps.gen; in = kzalloc(sizeof(*in), GFP_KERNEL); if (!in) return -ENOMEM; @@ -1619,7 +1611,8 @@ static int __mlx5_ib_modify_qp(struct ib_qp *ibqp, err = -EINVAL; goto out; } - context->mtu_msgmax = (attr->path_mtu << 5) | gen->log_max_msg; + context->mtu_msgmax = (attr->path_mtu << 5) | + (u8)MLX5_CAP_GEN(dev->mdev, log_max_msg); } if (attr_mask & IB_QP_DEST_QPN) @@ -1777,11 +1770,9 @@ int mlx5_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr, struct mlx5_ib_dev *dev = to_mdev(ibqp->device); struct mlx5_ib_qp *qp = to_mqp(ibqp); enum ib_qp_state cur_state, new_state; - struct mlx5_general_caps *gen; int err = -EINVAL; int port; - gen = &dev->mdev->caps.gen; mutex_lock(&qp->mutex); cur_state = attr_mask & IB_QP_CUR_STATE ? attr->cur_qp_state : qp->state; @@ -1793,21 +1784,25 @@ int mlx5_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr, goto out; if ((attr_mask & IB_QP_PORT) && - (attr->port_num == 0 || attr->port_num > gen->num_ports)) + (attr->port_num == 0 || + attr->port_num > MLX5_CAP_GEN(dev->mdev, 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 >= gen->port[port - 1].pkey_table_len) + if (attr->pkey_index >= + dev->mdev->port_caps[port - 1].pkey_table_len) goto out; } if (attr_mask & IB_QP_MAX_QP_RD_ATOMIC && - attr->max_rd_atomic > (1 << gen->log_max_ra_res_qp)) + attr->max_rd_atomic > + (1 << MLX5_CAP_GEN(dev->mdev, log_max_ra_res_qp))) goto out; if (attr_mask & IB_QP_MAX_DEST_RD_ATOMIC && - attr->max_dest_rd_atomic > (1 << gen->log_max_ra_req_qp)) + attr->max_dest_rd_atomic > + (1 << MLX5_CAP_GEN(dev->mdev, log_max_ra_req_qp))) goto out; if (cur_state == new_state && cur_state == IB_QPS_RESET) { @@ -3009,7 +3004,7 @@ static void to_ib_ah_attr(struct mlx5_ib_dev *ibdev, struct ib_ah_attr *ib_ah_at ib_ah_attr->port_num = path->port; if (ib_ah_attr->port_num == 0 || - ib_ah_attr->port_num > dev->caps.gen.num_ports) + ib_ah_attr->port_num > MLX5_CAP_GEN(dev, num_ports)) return; ib_ah_attr->sl = path->sl & 0xf; @@ -3135,12 +3130,10 @@ struct ib_xrcd *mlx5_ib_alloc_xrcd(struct ib_device *ibdev, struct ib_udata *udata) { struct mlx5_ib_dev *dev = to_mdev(ibdev); - struct mlx5_general_caps *gen; struct mlx5_ib_xrcd *xrcd; int err; - gen = &dev->mdev->caps.gen; - if (!(gen->flags & MLX5_DEV_CAP_FLAG_XRC)) + if (!MLX5_CAP_GEN(dev->mdev, xrc)) return ERR_PTR(-ENOSYS); xrcd = kmalloc(sizeof(*xrcd), GFP_KERNEL); diff --git a/drivers/infiniband/hw/mlx5/srq.c b/drivers/infiniband/hw/mlx5/srq.c index 02d77a29764d..e008505e96e9 100644 --- a/drivers/infiniband/hw/mlx5/srq.c +++ b/drivers/infiniband/hw/mlx5/srq.c @@ -165,7 +165,7 @@ static int create_srq_kernel(struct mlx5_ib_dev *dev, struct mlx5_ib_srq *srq, return err; } - if (mlx5_buf_alloc(dev->mdev, buf_size, PAGE_SIZE * 2, &srq->buf)) { + if (mlx5_buf_alloc(dev->mdev, buf_size, &srq->buf)) { mlx5_ib_dbg(dev, "buf alloc failed\n"); err = -ENOMEM; goto err_db; @@ -236,7 +236,6 @@ struct ib_srq *mlx5_ib_create_srq(struct ib_pd *pd, struct ib_udata *udata) { struct mlx5_ib_dev *dev = to_mdev(pd->device); - struct mlx5_general_caps *gen; struct mlx5_ib_srq *srq; int desc_size; int buf_size; @@ -245,13 +244,13 @@ struct ib_srq *mlx5_ib_create_srq(struct ib_pd *pd, int uninitialized_var(inlen); int is_xrc; u32 flgs, xrcdn; + __u32 max_srq_wqes = 1 << MLX5_CAP_GEN(dev->mdev, log_max_srq_sz); - gen = &dev->mdev->caps.gen; /* Sanity check SRQ size before proceeding */ - if (init_attr->attr.max_wr >= gen->max_srq_wqes) { + if (init_attr->attr.max_wr >= max_srq_wqes) { mlx5_ib_dbg(dev, "max_wr %d, cap %d\n", init_attr->attr.max_wr, - gen->max_srq_wqes); + max_srq_wqes); return ERR_PTR(-EINVAL); } @@ -303,7 +302,7 @@ struct ib_srq *mlx5_ib_create_srq(struct ib_pd *pd, in->ctx.pd = cpu_to_be32(to_mpd(pd)->pdn); in->ctx.db_record = cpu_to_be64(srq->db.dma); - err = mlx5_core_create_srq(dev->mdev, &srq->msrq, in, inlen); + err = mlx5_core_create_srq(dev->mdev, &srq->msrq, in, inlen, is_xrc); kvfree(in); if (err) { mlx5_ib_dbg(dev, "create SRQ failed, err %d\n", err); |