diff options
| -rw-r--r-- | drivers/infiniband/hw/hfi1/aspm.h | 15 | ||||
| -rw-r--r-- | drivers/infiniband/hw/hfi1/chip.c | 291 | ||||
| -rw-r--r-- | drivers/infiniband/hw/hfi1/chip.h | 2 | ||||
| -rw-r--r-- | drivers/infiniband/hw/hfi1/debugfs.c | 10 | ||||
| -rw-r--r-- | drivers/infiniband/hw/hfi1/driver.c | 52 | ||||
| -rw-r--r-- | drivers/infiniband/hw/hfi1/file_ops.c | 27 | ||||
| -rw-r--r-- | drivers/infiniband/hw/hfi1/hfi.h | 29 | ||||
| -rw-r--r-- | drivers/infiniband/hw/hfi1/init.c | 29 | ||||
| -rw-r--r-- | drivers/infiniband/hw/hfi1/mad.c | 10 | ||||
| -rw-r--r-- | drivers/infiniband/hw/hfi1/pio.c | 19 | ||||
| -rw-r--r-- | drivers/infiniband/hw/hfi1/pio.h | 8 | ||||
| -rw-r--r-- | drivers/infiniband/hw/hfi1/sysfs.c | 4 | ||||
| -rw-r--r-- | drivers/infiniband/hw/hfi1/user_exp_rcv.c | 8 | ||||
| -rw-r--r-- | drivers/infiniband/hw/hfi1/user_pages.c | 5 | ||||
| -rw-r--r-- | drivers/infiniband/hw/hfi1/verbs.c | 6 | ||||
| -rw-r--r-- | drivers/infiniband/hw/hfi1/vnic.h | 3 | ||||
| -rw-r--r-- | drivers/infiniband/hw/hfi1/vnic_main.c | 245 | ||||
| -rw-r--r-- | include/rdma/opa_port_info.h | 3 |
18 files changed, 661 insertions, 105 deletions
diff --git a/drivers/infiniband/hw/hfi1/aspm.h b/drivers/infiniband/hw/hfi1/aspm.h index 0d58fe3b49b5..794e6814a531 100644 --- a/drivers/infiniband/hw/hfi1/aspm.h +++ b/drivers/infiniband/hw/hfi1/aspm.h @@ -1,5 +1,5 @@ /* - * Copyright(c) 2015, 2016 Intel Corporation. + * Copyright(c) 2015-2017 Intel Corporation. * * This file is provided under a dual BSD/GPLv2 license. When using or * redistributing this file, you may do so under either license. @@ -229,14 +229,17 @@ static inline void aspm_ctx_timer_function(unsigned long data) spin_unlock_irqrestore(&rcd->aspm_lock, flags); } -/* Disable interrupt processing for verbs contexts when PSM contexts are open */ +/* + * Disable interrupt processing for verbs contexts when PSM or VNIC contexts + * are open. + */ static inline void aspm_disable_all(struct hfi1_devdata *dd) { struct hfi1_ctxtdata *rcd; unsigned long flags; unsigned i; - for (i = 0; i < dd->first_user_ctxt; i++) { + for (i = 0; i < dd->first_dyn_alloc_ctxt; i++) { rcd = dd->rcd[i]; del_timer_sync(&rcd->aspm_timer); spin_lock_irqsave(&rcd->aspm_lock, flags); @@ -260,7 +263,7 @@ static inline void aspm_enable_all(struct hfi1_devdata *dd) if (aspm_mode != ASPM_MODE_DYNAMIC) return; - for (i = 0; i < dd->first_user_ctxt; i++) { + for (i = 0; i < dd->first_dyn_alloc_ctxt; i++) { rcd = dd->rcd[i]; spin_lock_irqsave(&rcd->aspm_lock, flags); rcd->aspm_intr_enable = true; @@ -276,7 +279,7 @@ static inline void aspm_ctx_init(struct hfi1_ctxtdata *rcd) (unsigned long)rcd); rcd->aspm_intr_supported = rcd->dd->aspm_supported && aspm_mode == ASPM_MODE_DYNAMIC && - rcd->ctxt < rcd->dd->first_user_ctxt; + rcd->ctxt < rcd->dd->first_dyn_alloc_ctxt; } static inline void aspm_init(struct hfi1_devdata *dd) @@ -286,7 +289,7 @@ static inline void aspm_init(struct hfi1_devdata *dd) spin_lock_init(&dd->aspm_lock); dd->aspm_supported = aspm_hw_l1_supported(dd); - for (i = 0; i < dd->first_user_ctxt; i++) + for (i = 0; i < dd->first_dyn_alloc_ctxt; i++) aspm_ctx_init(dd->rcd[i]); /* Start with ASPM disabled */ diff --git a/drivers/infiniband/hw/hfi1/chip.c b/drivers/infiniband/hw/hfi1/chip.c index 79a316acb8f4..e520929ac501 100644 --- a/drivers/infiniband/hw/hfi1/chip.c +++ b/drivers/infiniband/hw/hfi1/chip.c @@ -126,9 +126,16 @@ struct flag_table { #define DEFAULT_KRCVQS 2 #define MIN_KERNEL_KCTXTS 2 #define FIRST_KERNEL_KCTXT 1 -/* sizes for both the QP and RSM map tables */ -#define NUM_MAP_ENTRIES 256 -#define NUM_MAP_REGS 32 + +/* + * RSM instance allocation + * 0 - Verbs + * 1 - User Fecn Handling + * 2 - Vnic + */ +#define RSM_INS_VERBS 0 +#define RSM_INS_FECN 1 +#define RSM_INS_VNIC 2 /* Bit offset into the GUID which carries HFI id information */ #define GUID_HFI_INDEX_SHIFT 39 @@ -139,8 +146,7 @@ struct flag_table { #define is_emulator_p(dd) ((((dd)->irev) & 0xf) == 3) #define is_emulator_s(dd) ((((dd)->irev) & 0xf) == 4) -/* RSM fields */ - +/* RSM fields for Verbs */ /* packet type */ #define IB_PACKET_TYPE 2ull #define QW_SHIFT 6ull @@ -170,6 +176,28 @@ struct flag_table { /* QPN[m+n:1] QW 1, OFFSET 1 */ #define QPN_SELECT_OFFSET ((1ull << QW_SHIFT) | (1ull)) +/* RSM fields for Vnic */ +/* L2_TYPE: QW 0, OFFSET 61 - for match */ +#define L2_TYPE_QW 0ull +#define L2_TYPE_BIT_OFFSET 61ull +#define L2_TYPE_OFFSET(off) ((L2_TYPE_QW << QW_SHIFT) | (off)) +#define L2_TYPE_MATCH_OFFSET L2_TYPE_OFFSET(L2_TYPE_BIT_OFFSET) +#define L2_TYPE_MASK 3ull +#define L2_16B_VALUE 2ull + +/* L4_TYPE QW 1, OFFSET 0 - for match */ +#define L4_TYPE_QW 1ull +#define L4_TYPE_BIT_OFFSET 0ull +#define L4_TYPE_OFFSET(off) ((L4_TYPE_QW << QW_SHIFT) | (off)) +#define L4_TYPE_MATCH_OFFSET L4_TYPE_OFFSET(L4_TYPE_BIT_OFFSET) +#define L4_16B_TYPE_MASK 0xFFull +#define L4_16B_ETH_VALUE 0x78ull + +/* 16B VESWID - for select */ +#define L4_16B_HDR_VESWID_OFFSET ((2 << QW_SHIFT) | (16ull)) +/* 16B ENTROPY - for select */ +#define L2_16B_ENTROPY_OFFSET ((1 << QW_SHIFT) | (32ull)) + /* defines to build power on SC2VL table */ #define SC2VL_VAL( \ num, \ @@ -1047,6 +1075,7 @@ static int qos_rmt_entries(struct hfi1_devdata *dd, unsigned int *mp, unsigned int *np); static void clear_full_mgmt_pkey(struct hfi1_pportdata *ppd); static int wait_link_transfer_active(struct hfi1_devdata *dd, int wait_ms); +static void clear_rsm_rule(struct hfi1_devdata *dd, u8 rule_index); /* * Error interrupt table entry. This is used as input to the interrupt @@ -6703,7 +6732,13 @@ static void rxe_kernel_unfreeze(struct hfi1_devdata *dd) int i; /* enable all kernel contexts */ - for (i = 0; i < dd->n_krcv_queues; i++) { + for (i = 0; i < dd->num_rcv_contexts; i++) { + struct hfi1_ctxtdata *rcd = dd->rcd[i]; + + /* Ensure all non-user contexts(including vnic) are enabled */ + if (!rcd || !rcd->sc || (rcd->sc->type == SC_USER)) + continue; + rcvmask = HFI1_RCVCTRL_CTXT_ENB; /* HFI1_RCVCTRL_TAILUPD_[ENB|DIS] needs to be set explicitly */ rcvmask |= HFI1_CAP_KGET_MASK(dd->rcd[i]->flags, DMA_RTAIL) ? @@ -8000,7 +8035,9 @@ static void is_rcv_avail_int(struct hfi1_devdata *dd, unsigned int source) if (likely(source < dd->num_rcv_contexts)) { rcd = dd->rcd[source]; if (rcd) { - if (source < dd->first_user_ctxt) + /* Check for non-user contexts, including vnic */ + if ((source < dd->first_dyn_alloc_ctxt) || + (rcd->sc && (rcd->sc->type == SC_KERNEL))) rcd->do_interrupt(rcd, 0); else handle_user_interrupt(rcd); @@ -8028,7 +8065,8 @@ static void is_rcv_urgent_int(struct hfi1_devdata *dd, unsigned int source) rcd = dd->rcd[source]; if (rcd) { /* only pay attention to user urgent interrupts */ - if (source >= dd->first_user_ctxt) + if ((source >= dd->first_dyn_alloc_ctxt) && + (!rcd->sc || (rcd->sc->type == SC_USER))) handle_user_interrupt(rcd); return; /* OK */ } @@ -12842,7 +12880,10 @@ static int request_msix_irqs(struct hfi1_devdata *dd) first_sdma = last_general; last_sdma = first_sdma + dd->num_sdma; first_rx = last_sdma; - last_rx = first_rx + dd->n_krcv_queues; + last_rx = first_rx + dd->n_krcv_queues + HFI1_NUM_VNIC_CTXT; + + /* VNIC MSIx interrupts get mapped when VNIC contexts are created */ + dd->first_dyn_msix_idx = first_rx + dd->n_krcv_queues; /* * Sanity check - the code expects all SDMA chip source @@ -12856,7 +12897,7 @@ static int request_msix_irqs(struct hfi1_devdata *dd) const char *err_info; irq_handler_t handler; irq_handler_t thread = NULL; - void *arg; + void *arg = NULL; int idx; struct hfi1_ctxtdata *rcd = NULL; struct sdma_engine *sde = NULL; @@ -12883,24 +12924,25 @@ static int request_msix_irqs(struct hfi1_devdata *dd) } else if (first_rx <= i && i < last_rx) { idx = i - first_rx; rcd = dd->rcd[idx]; - /* no interrupt if no rcd */ - if (!rcd) - continue; - /* - * Set the interrupt register and mask for this - * context's interrupt. - */ - rcd->ireg = (IS_RCVAVAIL_START + idx) / 64; - rcd->imask = ((u64)1) << - ((IS_RCVAVAIL_START + idx) % 64); - handler = receive_context_interrupt; - thread = receive_context_thread; - arg = rcd; - snprintf(me->name, sizeof(me->name), - DRIVER_NAME "_%d kctxt%d", dd->unit, idx); - err_info = "receive context"; - remap_intr(dd, IS_RCVAVAIL_START + idx, i); - me->type = IRQ_RCVCTXT; + if (rcd) { + /* + * Set the interrupt register and mask for this + * context's interrupt. + */ + rcd->ireg = (IS_RCVAVAIL_START + idx) / 64; + rcd->imask = ((u64)1) << + ((IS_RCVAVAIL_START + idx) % 64); + handler = receive_context_interrupt; + thread = receive_context_thread; + arg = rcd; + snprintf(me->name, sizeof(me->name), + DRIVER_NAME "_%d kctxt%d", + dd->unit, idx); + err_info = "receive context"; + remap_intr(dd, IS_RCVAVAIL_START + idx, i); + me->type = IRQ_RCVCTXT; + rcd->msix_intr = i; + } } else { /* not in our expected range - complain, then * ignore it @@ -12938,6 +12980,84 @@ static int request_msix_irqs(struct hfi1_devdata *dd) return ret; } +void hfi1_vnic_synchronize_irq(struct hfi1_devdata *dd) +{ + int i; + + if (!dd->num_msix_entries) { + synchronize_irq(dd->pcidev->irq); + return; + } + + for (i = 0; i < dd->vnic.num_ctxt; i++) { + struct hfi1_ctxtdata *rcd = dd->vnic.ctxt[i]; + struct hfi1_msix_entry *me = &dd->msix_entries[rcd->msix_intr]; + + synchronize_irq(me->msix.vector); + } +} + +void hfi1_reset_vnic_msix_info(struct hfi1_ctxtdata *rcd) +{ + struct hfi1_devdata *dd = rcd->dd; + struct hfi1_msix_entry *me = &dd->msix_entries[rcd->msix_intr]; + + if (!me->arg) /* => no irq, no affinity */ + return; + + hfi1_put_irq_affinity(dd, me); + free_irq(me->msix.vector, me->arg); + + me->arg = NULL; +} + +void hfi1_set_vnic_msix_info(struct hfi1_ctxtdata *rcd) +{ + struct hfi1_devdata *dd = rcd->dd; + struct hfi1_msix_entry *me; + int idx = rcd->ctxt; + void *arg = rcd; + int ret; + + rcd->msix_intr = dd->vnic.msix_idx++; + me = &dd->msix_entries[rcd->msix_intr]; + + /* + * Set the interrupt register and mask for this + * context's interrupt. + */ + rcd->ireg = (IS_RCVAVAIL_START + idx) / 64; + rcd->imask = ((u64)1) << + ((IS_RCVAVAIL_START + idx) % 64); + + snprintf(me->name, sizeof(me->name), + DRIVER_NAME "_%d kctxt%d", dd->unit, idx); + me->name[sizeof(me->name) - 1] = 0; + me->type = IRQ_RCVCTXT; + + remap_intr(dd, IS_RCVAVAIL_START + idx, rcd->msix_intr); + + ret = request_threaded_irq(me->msix.vector, receive_context_interrupt, + receive_context_thread, 0, me->name, arg); + if (ret) { + dd_dev_err(dd, "vnic irq request (vector %d, idx %d) fail %d\n", + me->msix.vector, idx, ret); + return; + } + /* + * assign arg after request_irq call, so it will be + * cleaned up + */ + me->arg = arg; + + ret = hfi1_get_irq_affinity(dd, me); + if (ret) { + dd_dev_err(dd, + "unable to pin IRQ %d\n", ret); + free_irq(me->msix.vector, me->arg); + } +} + /* * Set the general handler to accept all interrupts, remap all * chip interrupts back to MSI-X 0. @@ -12969,7 +13089,7 @@ static int set_up_interrupts(struct hfi1_devdata *dd) * N interrupts - one per used SDMA engine * M interrupt - one per kernel receive context */ - total = 1 + dd->num_sdma + dd->n_krcv_queues; + total = 1 + dd->num_sdma + dd->n_krcv_queues + HFI1_NUM_VNIC_CTXT; entries = kcalloc(total, sizeof(*entries), GFP_KERNEL); if (!entries) { @@ -13034,7 +13154,8 @@ fail: * * num_rcv_contexts - number of contexts being used * n_krcv_queues - number of kernel contexts - * first_user_ctxt - first non-kernel context in array of contexts + * first_dyn_alloc_ctxt - first dynamically allocated context + * in array of contexts * freectxts - number of free user contexts * num_send_contexts - number of PIO send contexts being used */ @@ -13111,10 +13232,14 @@ static int set_up_context_variables(struct hfi1_devdata *dd) total_contexts = num_kernel_contexts + num_user_contexts; } - /* the first N are kernel contexts, the rest are user contexts */ + /* Accommodate VNIC contexts */ + if ((total_contexts + HFI1_NUM_VNIC_CTXT) <= dd->chip_rcv_contexts) + total_contexts += HFI1_NUM_VNIC_CTXT; + + /* the first N are kernel contexts, the rest are user/vnic contexts */ dd->num_rcv_contexts = total_contexts; dd->n_krcv_queues = num_kernel_contexts; - dd->first_user_ctxt = num_kernel_contexts; + dd->first_dyn_alloc_ctxt = num_kernel_contexts; dd->num_user_contexts = num_user_contexts; dd->freectxts = num_user_contexts; dd_dev_info(dd, @@ -13570,11 +13695,8 @@ static void reset_rxe_csrs(struct hfi1_devdata *dd) write_csr(dd, RCV_COUNTER_ARRAY32 + (8 * i), 0); for (i = 0; i < RXE_NUM_64_BIT_COUNTERS; i++) write_csr(dd, RCV_COUNTER_ARRAY64 + (8 * i), 0); - for (i = 0; i < RXE_NUM_RSM_INSTANCES; i++) { - write_csr(dd, RCV_RSM_CFG + (8 * i), 0); - write_csr(dd, RCV_RSM_SELECT + (8 * i), 0); - write_csr(dd, RCV_RSM_MATCH + (8 * i), 0); - } + for (i = 0; i < RXE_NUM_RSM_INSTANCES; i++) + clear_rsm_rule(dd, i); for (i = 0; i < 32; i++) write_csr(dd, RCV_RSM_MAP_TABLE + (8 * i), 0); @@ -13933,6 +14055,16 @@ static void add_rsm_rule(struct hfi1_devdata *dd, u8 rule_index, (u64)rrd->value2 << RCV_RSM_MATCH_VALUE2_SHIFT); } +/* + * Clear a receive side mapping rule. + */ +static void clear_rsm_rule(struct hfi1_devdata *dd, u8 rule_index) +{ + write_csr(dd, RCV_RSM_CFG + (8 * rule_index), 0); + write_csr(dd, RCV_RSM_SELECT + (8 * rule_index), 0); + write_csr(dd, RCV_RSM_MATCH + (8 * rule_index), 0); +} + /* return the number of RSM map table entries that will be used for QOS */ static int qos_rmt_entries(struct hfi1_devdata *dd, unsigned int *mp, unsigned int *np) @@ -14048,7 +14180,7 @@ static void init_qos(struct hfi1_devdata *dd, struct rsm_map_table *rmt) rrd.value2 = LRH_SC_VALUE; /* add rule 0 */ - add_rsm_rule(dd, 0, &rrd); + add_rsm_rule(dd, RSM_INS_VERBS, &rrd); /* mark RSM map entries as used */ rmt->used += rmt_entries; @@ -14078,7 +14210,7 @@ static void init_user_fecn_handling(struct hfi1_devdata *dd, /* * RSM will extract the destination context as an index into the * map table. The destination contexts are a sequential block - * in the range first_user_ctxt...num_rcv_contexts-1 (inclusive). + * in the range first_dyn_alloc_ctxt...num_rcv_contexts-1 (inclusive). * Map entries are accessed as offset + extracted value. Adjust * the added offset so this sequence can be placed anywhere in * the table - as long as the entries themselves do not wrap. @@ -14086,9 +14218,9 @@ static void init_user_fecn_handling(struct hfi1_devdata *dd, * start with that to allow for a "negative" offset. */ offset = (u8)(NUM_MAP_ENTRIES + (int)rmt->used - - (int)dd->first_user_ctxt); + (int)dd->first_dyn_alloc_ctxt); - for (i = dd->first_user_ctxt, idx = rmt->used; + for (i = dd->first_dyn_alloc_ctxt, idx = rmt->used; i < dd->num_rcv_contexts; i++, idx++) { /* replace with identity mapping */ regoff = (idx % 8) * 8; @@ -14122,11 +14254,84 @@ static void init_user_fecn_handling(struct hfi1_devdata *dd, rrd.value2 = 1; /* add rule 1 */ - add_rsm_rule(dd, 1, &rrd); + add_rsm_rule(dd, RSM_INS_FECN, &rrd); rmt->used += dd->num_user_contexts; } +/* Initialize RSM for VNIC */ +void hfi1_init_vnic_rsm(struct hfi1_devdata *dd) +{ + u8 i, j; + u8 ctx_id = 0; + u64 reg; + u32 regoff; + struct rsm_rule_data rrd; + + if (hfi1_vnic_is_rsm_full(dd, NUM_VNIC_MAP_ENTRIES)) { + dd_dev_err(dd, "Vnic RSM disabled, rmt entries used = %d\n", + dd->vnic.rmt_start); + return; + } + + dev_dbg(&(dd)->pcidev->dev, "Vnic rsm start = %d, end %d\n", + dd->vnic.rmt_start, + dd->vnic.rmt_start + NUM_VNIC_MAP_ENTRIES); + + /* Update RSM mapping table, 32 regs, 256 entries - 1 ctx per byte */ + regoff = RCV_RSM_MAP_TABLE + (dd->vnic.rmt_start / 8) * 8; + reg = read_csr(dd, regoff); + for (i = 0; i < NUM_VNIC_MAP_ENTRIES; i++) { + /* Update map register with vnic context */ + j = (dd->vnic.rmt_start + i) % 8; + reg &= ~(0xffllu << (j * 8)); + reg |= (u64)dd->vnic.ctxt[ctx_id++]->ctxt << (j * 8); + /* Wrap up vnic ctx index */ + ctx_id %= dd->vnic.num_ctxt; + /* Write back map register */ + if (j == 7 || ((i + 1) == NUM_VNIC_MAP_ENTRIES)) { + dev_dbg(&(dd)->pcidev->dev, + "Vnic rsm map reg[%d] =0x%llx\n", + regoff - RCV_RSM_MAP_TABLE, reg); + + write_csr(dd, regoff, reg); + regoff += 8; + if (i < (NUM_VNIC_MAP_ENTRIES - 1)) + reg = read_csr(dd, regoff); + } + } + + /* Add rule for vnic */ + rrd.offset = dd->vnic.rmt_start; + rrd.pkt_type = 4; + /* Match 16B packets */ + rrd.field1_off = L2_TYPE_MATCH_OFFSET; + rrd.mask1 = L2_TYPE_MASK; + rrd.value1 = L2_16B_VALUE; + /* Match ETH L4 packets */ + rrd.field2_off = L4_TYPE_MATCH_OFFSET; + rrd.mask2 = L4_16B_TYPE_MASK; + rrd.value2 = L4_16B_ETH_VALUE; + /* Calc context from veswid and entropy */ + rrd.index1_off = L4_16B_HDR_VESWID_OFFSET; + rrd.index1_width = ilog2(NUM_VNIC_MAP_ENTRIES); + rrd.index2_off = L2_16B_ENTROPY_OFFSET; + rrd.index2_width = ilog2(NUM_VNIC_MAP_ENTRIES); + add_rsm_rule(dd, RSM_INS_VNIC, &rrd); + + /* Enable RSM if not already enabled */ + add_rcvctrl(dd, RCV_CTRL_RCV_RSM_ENABLE_SMASK); +} + +void hfi1_deinit_vnic_rsm(struct hfi1_devdata *dd) +{ + clear_rsm_rule(dd, RSM_INS_VNIC); + + /* Disable RSM if used only by vnic */ + if (dd->vnic.rmt_start == 0) + clear_rcvctrl(dd, RCV_CTRL_RCV_RSM_ENABLE_SMASK); +} + static void init_rxe(struct hfi1_devdata *dd) { struct rsm_map_table *rmt; @@ -14139,6 +14344,8 @@ static void init_rxe(struct hfi1_devdata *dd) init_qos(dd, rmt); init_user_fecn_handling(dd, rmt); complete_rsm_map_table(dd, rmt); + /* record number of used rsm map entries for vnic */ + dd->vnic.rmt_start = rmt->used; kfree(rmt); /* diff --git a/drivers/infiniband/hw/hfi1/chip.h b/drivers/infiniband/hw/hfi1/chip.h index 24df45fc8722..b9dbf16d7703 100644 --- a/drivers/infiniband/hw/hfi1/chip.h +++ b/drivers/infiniband/hw/hfi1/chip.h @@ -1362,6 +1362,8 @@ int hfi1_clear_ctxt_jkey(struct hfi1_devdata *dd, unsigned ctxt); int hfi1_set_ctxt_pkey(struct hfi1_devdata *dd, unsigned ctxt, u16 pkey); int hfi1_clear_ctxt_pkey(struct hfi1_devdata *dd, unsigned ctxt); void hfi1_read_link_quality(struct hfi1_devdata *dd, u8 *link_quality); +void hfi1_init_vnic_rsm(struct hfi1_devdata *dd); +void hfi1_deinit_vnic_rsm(struct hfi1_devdata *dd); /* * Interrupt source table. diff --git a/drivers/infiniband/hw/hfi1/debugfs.c b/drivers/infiniband/hw/hfi1/debugfs.c index dc2c1c993f04..e9fa3c293e42 100644 --- a/drivers/infiniband/hw/hfi1/debugfs.c +++ b/drivers/infiniband/hw/hfi1/debugfs.c @@ -1,6 +1,6 @@ #ifdef CONFIG_DEBUG_FS /* - * Copyright(c) 2015, 2016 Intel Corporation. + * Copyright(c) 2015-2017 Intel Corporation. * * This file is provided under a dual BSD/GPLv2 license. When using or * redistributing this file, you may do so under either license. @@ -174,7 +174,7 @@ static int _opcode_stats_seq_show(struct seq_file *s, void *v) struct hfi1_ibdev *ibd = (struct hfi1_ibdev *)s->private; struct hfi1_devdata *dd = dd_from_dev(ibd); - for (j = 0; j < dd->first_user_ctxt; j++) { + for (j = 0; j < dd->first_dyn_alloc_ctxt; j++) { if (!dd->rcd[j]) continue; n_packets += dd->rcd[j]->opstats->stats[i].n_packets; @@ -200,7 +200,7 @@ static void *_ctx_stats_seq_start(struct seq_file *s, loff_t *pos) if (!*pos) return SEQ_START_TOKEN; - if (*pos >= dd->first_user_ctxt) + if (*pos >= dd->first_dyn_alloc_ctxt) return NULL; return pos; } @@ -214,7 +214,7 @@ static void *_ctx_stats_seq_next(struct seq_file *s, void *v, loff_t *pos) return pos; ++*pos; - if (*pos >= dd->first_user_ctxt) + if (*pos >= dd->first_dyn_alloc_ctxt) return NULL; return pos; } @@ -1099,7 +1099,7 @@ static int _fault_stats_seq_show(struct seq_file *s, void *v) struct hfi1_ibdev *ibd = (struct hfi1_ibdev *)s->private; struct hfi1_devdata *dd = dd_from_dev(ibd); - for (j = 0; j < dd->first_user_ctxt; j++) { + for (j = 0; j < dd->first_dyn_alloc_ctxt; j++) { if (!dd->rcd[j]) continue; n_packets += dd->rcd[j]->opstats->stats[i].n_packets; diff --git a/drivers/infiniband/hw/hfi1/driver.c b/drivers/infiniband/hw/hfi1/driver.c index e4dc6a5997e5..6b3869529d5e 100644 --- a/drivers/infiniband/hw/hfi1/driver.c +++ b/drivers/infiniband/hw/hfi1/driver.c @@ -874,20 +874,42 @@ bail: return last; } -static inline void set_all_nodma_rtail(struct hfi1_devdata *dd) +static inline void set_nodma_rtail(struct hfi1_devdata *dd, u8 ctxt) { int i; - for (i = HFI1_CTRL_CTXT + 1; i < dd->first_user_ctxt; i++) + /* + * For dynamically allocated kernel contexts (like vnic) switch + * interrupt handler only for that context. Otherwise, switch + * interrupt handler for all statically allocated kernel contexts. + */ + if (ctxt >= dd->first_dyn_alloc_ctxt) { + dd->rcd[ctxt]->do_interrupt = + &handle_receive_interrupt_nodma_rtail; + return; + } + + for (i = HFI1_CTRL_CTXT + 1; i < dd->first_dyn_alloc_ctxt; i++) dd->rcd[i]->do_interrupt = &handle_receive_interrupt_nodma_rtail; } -static inline void set_all_dma_rtail(struct hfi1_devdata *dd) +static inline void set_dma_rtail(struct hfi1_devdata *dd, u8 ctxt) { int i; - for (i = HFI1_CTRL_CTXT + 1; i < dd->first_user_ctxt; i++) + /* + * For dynamically allocated kernel contexts (like vnic) switch + * interrupt handler only for that context. Otherwise, switch + * interrupt handler for all statically allocated kernel contexts. + */ + if (ctxt >= dd->first_dyn_alloc_ctxt) { + dd->rcd[ctxt]->do_interrupt = + &handle_receive_interrupt_dma_rtail; + return; + } + + for (i = HFI1_CTRL_CTXT + 1; i < dd->first_dyn_alloc_ctxt; i++) dd->rcd[i]->do_interrupt = &handle_receive_interrupt_dma_rtail; } @@ -897,8 +919,13 @@ void set_all_slowpath(struct hfi1_devdata *dd) int i; /* HFI1_CTRL_CTXT must always use the slow path interrupt handler */ - for (i = HFI1_CTRL_CTXT + 1; i < dd->first_user_ctxt; i++) - dd->rcd[i]->do_interrupt = &handle_receive_interrupt; + for (i = HFI1_CTRL_CTXT + 1; i < dd->num_rcv_contexts; i++) { + struct hfi1_ctxtdata *rcd = dd->rcd[i]; + + if ((i < dd->first_dyn_alloc_ctxt) || + (rcd && rcd->sc && (rcd->sc->type == SC_KERNEL))) + rcd->do_interrupt = &handle_receive_interrupt; + } } static inline int set_armed_to_active(struct hfi1_ctxtdata *rcd, @@ -1008,7 +1035,7 @@ int handle_receive_interrupt(struct hfi1_ctxtdata *rcd, int thread) last = RCV_PKT_DONE; if (needset) { dd_dev_info(dd, "Switching to NO_DMA_RTAIL\n"); - set_all_nodma_rtail(dd); + set_nodma_rtail(dd, rcd->ctxt); needset = 0; } } else { @@ -1030,7 +1057,7 @@ int handle_receive_interrupt(struct hfi1_ctxtdata *rcd, int thread) if (needset) { dd_dev_info(dd, "Switching to DMA_RTAIL\n"); - set_all_dma_rtail(dd); + set_dma_rtail(dd, rcd->ctxt); needset = 0; } } @@ -1079,10 +1106,10 @@ void receive_interrupt_work(struct work_struct *work) set_link_state(ppd, HLS_UP_ACTIVE); /* - * Interrupt all kernel contexts that could have had an - * interrupt during auto activation. + * Interrupt all statically allocated kernel contexts that could + * have had an interrupt during auto activation. */ - for (i = HFI1_CTRL_CTXT; i < dd->first_user_ctxt; i++) + for (i = HFI1_CTRL_CTXT; i < dd->first_dyn_alloc_ctxt; i++) force_recv_intr(dd->rcd[i]); } @@ -1296,7 +1323,8 @@ int hfi1_reset_device(int unit) spin_lock_irqsave(&dd->uctxt_lock, flags); if (dd->rcd) - for (i = dd->first_user_ctxt; i < dd->num_rcv_contexts; i++) { + for (i = dd->first_dyn_alloc_ctxt; + i < dd->num_rcv_contexts; i++) { if (!dd->rcd[i] || !dd->rcd[i]->cnt) continue; spin_unlock_irqrestore(&dd->uctxt_lock, flags); diff --git a/drivers/infiniband/hw/hfi1/file_ops.c b/drivers/infiniband/hw/hfi1/file_ops.c index f78c739b330a..60598867f948 100644 --- a/drivers/infiniband/hw/hfi1/file_ops.c +++ b/drivers/infiniband/hw/hfi1/file_ops.c @@ -1,5 +1,5 @@ /* - * Copyright(c) 2015, 2016 Intel Corporation. + * Copyright(c) 2015-2017 Intel Corporation. * * This file is provided under a dual BSD/GPLv2 license. When using or * redistributing this file, you may do so under either license. @@ -586,8 +586,8 @@ static int hfi1_file_mmap(struct file *fp, struct vm_area_struct *vma) * knows where it's own bitmap is within the page. */ memaddr = (unsigned long)(dd->events + - ((uctxt->ctxt - dd->first_user_ctxt) * - HFI1_MAX_SHARED_CTXTS)) & PAGE_MASK; + ((uctxt->ctxt - dd->first_dyn_alloc_ctxt) * + HFI1_MAX_SHARED_CTXTS)) & PAGE_MASK; memlen = PAGE_SIZE; /* * v3.7 removes VM_RESERVED but the effect is kept by @@ -756,7 +756,7 @@ static int hfi1_file_close(struct inode *inode, struct file *fp) * Clear any left over, unhandled events so the next process that * gets this context doesn't get confused. */ - ev = dd->events + ((uctxt->ctxt - dd->first_user_ctxt) * + ev = dd->events + ((uctxt->ctxt - dd->first_dyn_alloc_ctxt) * HFI1_MAX_SHARED_CTXTS) + fdata->subctxt; *ev = 0; @@ -909,12 +909,18 @@ static int find_shared_ctxt(struct file *fp, if (!(dd && (dd->flags & HFI1_PRESENT) && dd->kregbase)) continue; - for (i = dd->first_user_ctxt; i < dd->num_rcv_contexts; i++) { + for (i = dd->first_dyn_alloc_ctxt; + i < dd->num_rcv_contexts; i++) { struct hfi1_ctxtdata *uctxt = dd->rcd[i]; /* Skip ctxts which are not yet open */ if (!uctxt || !uctxt->cnt) continue; + + /* Skip dynamically allocted kernel contexts */ + if (uctxt->sc && (uctxt->sc->type == SC_KERNEL)) + continue; + /* Skip ctxt if it doesn't match the requested one */ if (memcmp(uctxt->uuid, uinfo->uuid, sizeof(uctxt->uuid)) || @@ -960,7 +966,8 @@ static int allocate_ctxt(struct file *fp, struct hfi1_devdata *dd, return -EIO; } - for (ctxt = dd->first_user_ctxt; ctxt < dd->num_rcv_contexts; ctxt++) + for (ctxt = dd->first_dyn_alloc_ctxt; + ctxt < dd->num_rcv_contexts; ctxt++) if (!dd->rcd[ctxt]) break; @@ -1306,7 +1313,7 @@ static int get_base_info(struct file *fp, void __user *ubase, __u32 len) */ binfo.user_regbase = HFI1_MMAP_TOKEN(UREGS, uctxt->ctxt, fd->subctxt, 0); - offset = offset_in_page((((uctxt->ctxt - dd->first_user_ctxt) * + offset = offset_in_page((((uctxt->ctxt - dd->first_dyn_alloc_ctxt) * HFI1_MAX_SHARED_CTXTS) + fd->subctxt) * sizeof(*dd->events)); binfo.events_bufbase = HFI1_MMAP_TOKEN(EVENTS, uctxt->ctxt, @@ -1400,12 +1407,12 @@ int hfi1_set_uevent_bits(struct hfi1_pportdata *ppd, const int evtbit) } spin_lock_irqsave(&dd->uctxt_lock, flags); - for (ctxt = dd->first_user_ctxt; ctxt < dd->num_rcv_contexts; + for (ctxt = dd->first_dyn_alloc_ctxt; ctxt < dd->num_rcv_contexts; ctxt++) { uctxt = dd->rcd[ctxt]; if (uctxt) { unsigned long *evs = dd->events + - (uctxt->ctxt - dd->first_user_ctxt) * + (uctxt->ctxt - dd->first_dyn_alloc_ctxt) * HFI1_MAX_SHARED_CTXTS; int i; /* @@ -1477,7 +1484,7 @@ static int user_event_ack(struct hfi1_ctxtdata *uctxt, int subctxt, if (!dd->events) return 0; - evs = dd->events + ((uctxt->ctxt - dd->first_user_ctxt) * + evs = dd->events + ((uctxt->ctxt - dd->first_dyn_alloc_ctxt) * HFI1_MAX_SHARED_CTXTS) + subctxt; for (i = 0; i <= _HFI1_MAX_EVENT_BIT; i++) { diff --git a/drivers/infiniband/hw/hfi1/hfi.h b/drivers/infiniband/hw/hfi1/hfi.h index f85e8f4eae69..a12bb462d83f 100644 --- a/drivers/infiniband/hw/hfi1/hfi.h +++ b/drivers/infiniband/hw/hfi1/hfi.h @@ -54,6 +54,7 @@ #include <linux/list.h> #include <linux/scatterlist.h> #include <linux/slab.h> +#include <linux/idr.h> #include <linux/io.h> #include <linux/fs.h> #include <linux/completion.h> @@ -66,6 +67,7 @@ #include <linux/i2c-algo-bit.h> #include <rdma/ib_hdrs.h> #include <linux/rhashtable.h> +#include <linux/netdevice.h> #include <rdma/rdma_vt.h> #include "chip_registers.h" @@ -278,6 +280,8 @@ struct hfi1_ctxtdata { struct hfi1_devdata *dd; /* so functions that need physical port can get it easily */ struct hfi1_pportdata *ppd; + /* associated msix interrupt */ + u32 msix_intr; /* A page of memory for rcvhdrhead, rcvegrhead, rcvegrtail * N */ void *subctxt_uregbase; /* An array of pages for the eager receive buffers * N */ @@ -814,15 +818,27 @@ struct hfi1_asic_data { struct hfi1_i2c_bus *i2c_bus1; }; +/* sizes for both the QP and RSM map tables */ +#define NUM_MAP_ENTRIES 256 +#define NUM_MAP_REGS 32 + /* * Number of VNIC contexts used. Ensure it is less than or equal to * max queues supported by VNIC (HFI1_VNIC_MAX_QUEUE). */ #define HFI1_NUM_VNIC_CTXT 8 +/* Number of VNIC RSM entries */ +#define NUM_VNIC_MAP_ENTRIES 8 + /* Virtual NIC information */ struct hfi1_vnic_data { + struct hfi1_ctxtdata *ctxt[HFI1_NUM_VNIC_CTXT]; + u8 num_vports; struct idr vesw_idr; + u8 rmt_start; + u8 num_ctxt; + u32 msix_idx; }; struct hfi1_vnic_vport_info; @@ -1050,6 +1066,7 @@ struct hfi1_devdata { /* MSI-X information */ struct hfi1_msix_entry *msix_entries; u32 num_msix_entries; + u32 first_dyn_msix_idx; /* INTx information */ u32 requested_intx_irq; /* did we request one? */ @@ -1148,8 +1165,8 @@ struct hfi1_devdata { u16 flags; /* Number of physical ports available */ u8 num_pports; - /* Lowest context number which can be used by user processes */ - u8 first_user_ctxt; + /* Lowest context number which can be used by user processes or VNIC */ + u8 first_dyn_alloc_ctxt; /* adding a new field here would make it part of this cacheline */ /* seqlock for sc2vl */ @@ -1197,6 +1214,11 @@ struct hfi1_devdata { struct hfi1_vnic_data vnic; }; +static inline bool hfi1_vnic_is_rsm_full(struct hfi1_devdata *dd, int spare) +{ + return (dd->vnic.rmt_start + spare) > NUM_MAP_ENTRIES; +} + /* 8051 firmware version helper */ #define dc8051_ver(a, b, c) ((a) << 16 | (b) << 8 | (c)) #define dc8051_ver_maj(a) (((a) & 0xff0000) >> 16) @@ -1261,6 +1283,9 @@ int handle_receive_interrupt(struct hfi1_ctxtdata *, int); int handle_receive_interrupt_nodma_rtail(struct hfi1_ctxtdata *, int); int handle_receive_interrupt_dma_rtail(struct hfi1_ctxtdata *, int); void set_all_slowpath(struct hfi1_devdata *dd); +void hfi1_vnic_synchronize_irq(struct hfi1_devdata *dd); +void hfi1_set_vnic_msix_info(struct hfi1_ctxtdata *rcd); +void hfi1_reset_vnic_msix_info(struct hfi1_ctxtdata *rcd); extern const struct pci_device_id hfi1_pci_tbl[]; diff --git a/drivers/infiniband/hw/hfi1/init.c b/drivers/infiniband/hw/hfi1/init.c index e84f95d50e79..de2eec40f2a0 100644 --- a/drivers/infiniband/hw/hfi1/init.c +++ b/drivers/infiniband/hw/hfi1/init.c @@ -140,7 +140,7 @@ int hfi1_create_ctxts(struct hfi1_devdata *dd) goto nomem; /* create one or more kernel contexts */ - for (i = 0; i < dd->first_user_ctxt; ++i) { + for (i = 0; i < dd->first_dyn_alloc_ctxt; ++i) { struct hfi1_pportdata *ppd; struct hfi1_ctxtdata *rcd; @@ -215,9 +215,9 @@ struct hfi1_ctxtdata *hfi1_create_ctxtdata(struct hfi1_pportdata *ppd, u32 ctxt, u32 base; if (dd->rcv_entries.nctxt_extra > - dd->num_rcv_contexts - dd->first_user_ctxt) + dd->num_rcv_contexts - dd->first_dyn_alloc_ctxt) kctxt_ngroups = (dd->rcv_entries.nctxt_extra - - (dd->num_rcv_contexts - dd->first_user_ctxt)); + (dd->num_rcv_contexts - dd->first_dyn_alloc_ctxt)); rcd = kzalloc_node(sizeof(*rcd), GFP_KERNEL, numa); if (rcd) { u32 rcvtids, max_entries; @@ -239,10 +239,10 @@ struct hfi1_ctxtdata *hfi1_create_ctxtdata(struct hfi1_pportdata *ppd, u32 ctxt, * Calculate the context's RcvArray entry starting point. * We do this here because we have to take into account all * the RcvArray entries that previous context would have - * taken and we have to account for any extra groups - * assigned to the kernel or user contexts. + * taken and we have to account for any extra groups assigned + * to the static (kernel) or dynamic (vnic/user) contexts. */ - if (ctxt < dd->first_user_ctxt) { + if (ctxt < dd->first_dyn_alloc_ctxt) { if (ctxt < kctxt_ngroups) { base = ctxt * (dd->rcv_entries.ngroups + 1); rcd->rcv_array_groups++; @@ -250,7 +250,7 @@ struct hfi1_ctxtdata *hfi1_create_ctxtdata(struct hfi1_pportdata *ppd, u32 ctxt, base = kctxt_ngroups + (ctxt * dd->rcv_entries.ngroups); } else { - u16 ct = ctxt - dd->first_user_ctxt; + u16 ct = ctxt - dd->first_dyn_alloc_ctxt; base = ((dd->n_krcv_queues * dd->rcv_entries.ngroups) + kctxt_ngroups); @@ -323,7 +323,8 @@ struct hfi1_ctxtdata *hfi1_create_ctxtdata(struct hfi1_pportdata *ppd, u32 ctxt, } rcd->egrbufs.rcvtid_size = HFI1_MAX_EAGER_BUFFER_SIZE; - if (ctxt < dd->first_user_ctxt) { /* N/A for PSM contexts */ + /* Applicable only for statically created kernel contexts */ + if (ctxt < dd->first_dyn_alloc_ctxt) { rcd->opstats = kzalloc_node(sizeof(*rcd->opstats), GFP_KERNEL, numa); if (!rcd->opstats) @@ -586,7 +587,7 @@ static void enable_chip(struct hfi1_devdata *dd) * Enable kernel ctxts' receive and receive interrupt. * Other ctxts done as user opens and initializes them. */ - for (i = 0; i < dd->first_user_ctxt; ++i) { + for (i = 0; i < dd->first_dyn_alloc_ctxt; ++i) { rcvmask = HFI1_RCVCTRL_CTXT_ENB | HFI1_RCVCTRL_INTRAVAIL_ENB; rcvmask |= HFI1_CAP_KGET_MASK(dd->rcd[i]->flags, DMA_RTAIL) ? HFI1_RCVCTRL_TAILUPD_ENB : HFI1_RCVCTRL_TAILUPD_DIS; @@ -715,7 +716,7 @@ int hfi1_init(struct hfi1_devdata *dd, int reinit) } /* dd->rcd can be NULL if early initialization failed */ - for (i = 0; dd->rcd && i < dd->first_user_ctxt; ++i) { + for (i = 0; dd->rcd && i < dd->first_dyn_alloc_ctxt; ++i) { /* * Set up the (kernel) rcvhdr queue and egr TIDs. If doing * re-init, the simplest way to handle this is to free @@ -1535,6 +1536,7 @@ static int init_one(struct pci_dev *pdev, const struct pci_device_id *ent) hfi1_device_remove(dd); if (!ret) hfi1_unregister_ib_device(dd); + hfi1_vnic_cleanup(dd); postinit_cleanup(dd); if (initfail) ret = initfail; @@ -1621,8 +1623,11 @@ int hfi1_create_rcvhdrq(struct hfi1_devdata *dd, struct hfi1_ctxtdata *rcd) amt = PAGE_ALIGN(rcd->rcvhdrq_cnt * rcd->rcvhdrqentsize * sizeof(u32)); - gfp_flags = (rcd->ctxt >= dd->first_user_ctxt) ? - GFP_USER : GFP_KERNEL; + if ((rcd->ctxt < dd->first_dyn_alloc_ctxt) || + (rcd->sc && (rcd->sc->type == SC_KERNEL))) + gfp_flags = GFP_KERNEL; + else + gfp_flags = GFP_USER; rcd->rcvhdrq = dma_zalloc_coherent( &dd->pcidev->dev, amt, &rcd->rcvhdrq_dma, gfp_flags | __GFP_COMP); diff --git a/drivers/infiniband/hw/hfi1/mad.c b/drivers/infiniband/hw/hfi1/mad.c index 09cda3c35e82..955e5fce6573 100644 --- a/drivers/infiniband/hw/hfi1/mad.c +++ b/drivers/infiniband/hw/hfi1/mad.c @@ -53,6 +53,7 @@ #include "mad.h" #include "trace.h" #include "qp.h" +#include "vnic.h" /* the reset value from the FM is supposed to be 0xffff, handle both */ #define OPA_LINK_WIDTH_RESET_OLD 0x0fff @@ -650,9 +651,11 @@ static int __subn_get_opa_portinfo(struct opa_smp *smp, u32 am, u8 *data, OPA_PI_MASK_PORT_ACTIVE_OPTOMIZE : 0); pi->port_packet_format.supported = - cpu_to_be16(OPA_PORT_PACKET_FORMAT_9B); + cpu_to_be16(OPA_PORT_PACKET_FORMAT_9B | + OPA_PORT_PACKET_FORMAT_16B); pi->port_packet_format.enabled = - cpu_to_be16(OPA_PORT_PACKET_FORMAT_9B); + cpu_to_be16(OPA_PORT_PACKET_FORMAT_9B | + OPA_PORT_PACKET_FORMAT_16B); /* flit_control.interleave is (OPA V1, version .76): * bits use @@ -701,7 +704,8 @@ static int __subn_get_opa_portinfo(struct opa_smp *smp, u32 am, u8 *data, buffer_units |= (dd->vl15_init << 11) & OPA_PI_MASK_BUF_UNIT_VL15_INIT; pi->buffer_units = cpu_to_be32(buffer_units); - pi->opa_cap_mask = cpu_to_be16(OPA_CAP_MASK3_IsSharedSpaceSupported); + pi->opa_cap_mask = cpu_to_be16(OPA_CAP_MASK3_IsSharedSpaceSupported | + OPA_CAP_MASK3_IsEthOnFabricSupported); /* HFI supports a replay buffer 128 LTPs in size */ pi->replay_depth.buffer = 0x80; diff --git a/drivers/infiniband/hw/hfi1/pio.c b/drivers/infiniband/hw/hfi1/pio.c index 615be68e40b3..ed72b5aca139 100644 --- a/drivers/infiniband/hw/hfi1/pio.c +++ b/drivers/infiniband/hw/hfi1/pio.c @@ -1,5 +1,5 @@ /* - * Copyright(c) 2015, 2016 Intel Corporation. + * Copyright(c) 2015-2017 Intel Corporation. * * This file is provided under a dual BSD/GPLv2 license. When using or * redistributing this file, you may do so under either license. @@ -703,6 +703,7 @@ struct send_context *sc_alloc(struct hfi1_devdata *dd, int type, { struct send_context_info *sci; struct send_context *sc = NULL; + int req_type = type; dma_addr_t dma; unsigned long flags; u64 reg; @@ -729,6 +730,13 @@ struct send_context *sc_alloc(struct hfi1_devdata *dd, int type, return NULL; } + /* + * VNIC contexts are dynamically allocated. + * Hence, pick a user context for VNIC. + */ + if (type == SC_VNIC) + type = SC_USER; + spin_lock_irqsave(&dd->sc_lock, flags); ret = sc_hw_alloc(dd, type, &sw_index, &hw_context); if (ret) { @@ -738,6 +746,15 @@ struct send_context *sc_alloc(struct hfi1_devdata *dd, int type, return NULL; } + /* + * VNIC contexts are used by kernel driver. + * Hence, mark them as kernel contexts. + */ + if (req_type == SC_VNIC) { + dd->send_contexts[sw_index].type = SC_KERNEL; + type = SC_KERNEL; + } + sci = &dd->send_contexts[sw_index]; sci->sc = sc; diff --git a/drivers/infiniband/hw/hfi1/pio.h b/drivers/infiniband/hw/hfi1/pio.h index 867e5ffc3595..a6fb70093a70 100644 --- a/drivers/infiniband/hw/hfi1/pio.h +++ b/drivers/infiniband/hw/hfi1/pio.h @@ -1,7 +1,7 @@ #ifndef _PIO_H #define _PIO_H /* - * Copyright(c) 2015, 2016 Intel Corporation. + * Copyright(c) 2015-2017 Intel Corporation. * * This file is provided under a dual BSD/GPLv2 license. When using or * redistributing this file, you may do so under either license. @@ -54,6 +54,12 @@ #define SC_USER 3 /* must be the last one: it may take all left */ #define SC_MAX 4 /* count of send context types */ +/* + * SC_VNIC types are allocated (dynamically) from the user context pool, + * (SC_USER) and used by kernel driver as kernel contexts (SC_KERNEL). + */ +#define SC_VNIC SC_MAX + /* invalid send context index */ #define INVALID_SCI 0xff diff --git a/drivers/infiniband/hw/hfi1/sysfs.c b/drivers/infiniband/hw/hfi1/sysfs.c index 919a5474e651..50d140d25e38 100644 --- a/drivers/infiniband/hw/hfi1/sysfs.c +++ b/drivers/infiniband/hw/hfi1/sysfs.c @@ -1,5 +1,5 @@ /* - * Copyright(c) 2015, 2016 Intel Corporation. + * Copyright(c) 2015-2017 Intel Corporation. * * This file is provided under a dual BSD/GPLv2 license. When using or * redistributing this file, you may do so under either license. @@ -542,7 +542,7 @@ static ssize_t show_nctxts(struct device *device, * give a more accurate picture of total contexts available. */ return scnprintf(buf, PAGE_SIZE, "%u\n", - min(dd->num_rcv_contexts - dd->first_user_ctxt, + min(dd->num_rcv_contexts - dd->first_dyn_alloc_ctxt, (u32)dd->sc_sizes[SC_USER].count)); } diff --git a/drivers/infiniband/hw/hfi1/user_exp_rcv.c b/drivers/infiniband/hw/hfi1/user_exp_rcv.c index 4a8295399e71..25a8698f7db9 100644 --- a/drivers/infiniband/hw/hfi1/user_exp_rcv.c +++ b/drivers/infiniband/hw/hfi1/user_exp_rcv.c @@ -1,5 +1,5 @@ /* - * Copyright(c) 2015, 2016 Intel Corporation. + * Copyright(c) 2015-2017 Intel Corporation. * * This file is provided under a dual BSD/GPLv2 license. When using or * redistributing this file, you may do so under either license. @@ -607,7 +607,7 @@ int hfi1_user_exp_rcv_invalid(struct file *fp, struct hfi1_tid_info *tinfo) struct hfi1_filedata *fd = fp->private_data; struct hfi1_ctxtdata *uctxt = fd->uctxt; unsigned long *ev = uctxt->dd->events + - (((uctxt->ctxt - uctxt->dd->first_user_ctxt) * + (((uctxt->ctxt - uctxt->dd->first_dyn_alloc_ctxt) * HFI1_MAX_SHARED_CTXTS) + fd->subctxt); u32 *array; int ret = 0; @@ -1011,8 +1011,8 @@ static int tid_rb_invalidate(void *arg, struct mmu_rb_node *mnode) * process in question. */ ev = uctxt->dd->events + - (((uctxt->ctxt - uctxt->dd->first_user_ctxt) * - HFI1_MAX_SHARED_CTXTS) + fdata->subctxt); + (((uctxt->ctxt - uctxt->dd->first_dyn_alloc_ctxt) * + HFI1_MAX_SHARED_CTXTS) + fdata->subctxt); set_bit(_HFI1_EVENT_TID_MMU_NOTIFY_BIT, ev); } fdata->invalid_tid_idx++; diff --git a/drivers/infiniband/hw/hfi1/user_pages.c b/drivers/infiniband/hw/hfi1/user_pages.c index 68295a12b771..e341e6dcc388 100644 --- a/drivers/infiniband/hw/hfi1/user_pages.c +++ b/drivers/infiniband/hw/hfi1/user_pages.c @@ -1,5 +1,5 @@ /* - * Copyright(c) 2015, 2016 Intel Corporation. + * Copyright(c) 2015-2017 Intel Corporation. * * This file is provided under a dual BSD/GPLv2 license. When using or * redistributing this file, you may do so under either license. @@ -73,7 +73,8 @@ bool hfi1_can_pin_pages(struct hfi1_devdata *dd, struct mm_struct *mm, { unsigned long ulimit = rlimit(RLIMIT_MEMLOCK), pinned, cache_limit, size = (cache_size * (1UL << 20)); /* convert to bytes */ - unsigned usr_ctxts = dd->num_rcv_contexts - dd->first_user_ctxt; + unsigned int usr_ctxts = + dd->num_rcv_contexts - dd->first_dyn_alloc_ctxt; bool can_lock = capable(CAP_IPC_LOCK); /* diff --git a/drivers/infiniband/hw/hfi1/verbs.c b/drivers/infiniband/hw/hfi1/verbs.c index 070a349afd78..239fa480555f 100644 --- a/drivers/infiniband/hw/hfi1/verbs.c +++ b/drivers/infiniband/hw/hfi1/verbs.c @@ -61,6 +61,7 @@ #include "qp.h" #include "verbs_txreq.h" #include "debugfs.h" +#include "vnic.h" static unsigned int hfi1_lkey_table_size = 16; module_param_named(lkey_table_size, hfi1_lkey_table_size, uint, @@ -1289,7 +1290,8 @@ static void hfi1_fill_device_attr(struct hfi1_devdata *dd) IB_DEVICE_BAD_QKEY_CNTR | IB_DEVICE_SHUTDOWN_PORT | IB_DEVICE_SYS_IMAGE_GUID | IB_DEVICE_RC_RNR_NAK_GEN | IB_DEVICE_PORT_ACTIVE_EVENT | IB_DEVICE_SRQ_RESIZE | - IB_DEVICE_MEM_MGT_EXTENSIONS; + IB_DEVICE_MEM_MGT_EXTENSIONS | + IB_DEVICE_RDMA_NETDEV_OPA_VNIC; rdi->dparms.props.page_size_cap = PAGE_SIZE; rdi->dparms.props.vendor_id = dd->oui1 << 16 | dd->oui2 << 8 | dd->oui3; rdi->dparms.props.vendor_part_id = dd->pcidev->device; @@ -1772,6 +1774,8 @@ int hfi1_register_ib_device(struct hfi1_devdata *dd) ibdev->modify_device = modify_device; ibdev->alloc_hw_stats = alloc_hw_stats; ibdev->get_hw_stats = get_hw_stats; + ibdev->alloc_rdma_netdev = hfi1_vnic_alloc_rn; + ibdev->free_rdma_netdev = hfi1_vnic_free_rn; /* keep process mad in the driver */ ibdev->process_mad = hfi1_process_mad; diff --git a/drivers/infiniband/hw/hfi1/vnic.h b/drivers/infiniband/hw/hfi1/vnic.h index 04723b1ab246..9bed40d85cff 100644 --- a/drivers/infiniband/hw/hfi1/vnic.h +++ b/drivers/infiniband/hw/hfi1/vnic.h @@ -149,5 +149,8 @@ struct net_device *hfi1_vnic_alloc_rn(struct ib_device *device, unsigned char name_assign_type, void (*setup)(struct net_device *)); void hfi1_vnic_free_rn(struct net_device *netdev); +int hfi1_vnic_send_dma(struct hfi1_devdata *dd, u8 q_idx, + struct hfi1_vnic_vport_info *vinfo, + struct sk_buff *skb, u64 pbc, u8 plen); #endif /* _HFI1_VNIC_H */ diff --git a/drivers/infiniband/hw/hfi1/vnic_main.c b/drivers/infiniband/hw/hfi1/vnic_main.c index fb23f9ff6bc1..32d91b60972b 100644 --- a/drivers/infiniband/hw/hfi1/vnic_main.c +++ b/drivers/infiniband/hw/hfi1/vnic_main.c @@ -62,6 +62,159 @@ static DEFINE_SPINLOCK(vport_cntr_lock); +static int setup_vnic_ctxt(struct hfi1_devdata *dd, struct hfi1_ctxtdata *uctxt) +{ + unsigned int rcvctrl_ops = 0; + int ret; + + ret = hfi1_init_ctxt(uctxt->sc); + if (ret) + goto done; + + uctxt->do_interrupt = &handle_receive_interrupt; + + /* Now allocate the RcvHdr queue and eager buffers. */ + ret = hfi1_create_rcvhdrq(dd, uctxt); + if (ret) + goto done; + + ret = hfi1_setup_eagerbufs(uctxt); + if (ret) + goto done; + + set_bit(HFI1_CTXT_SETUP_DONE, &uctxt->event_flags); + + if (uctxt->rcvhdrtail_kvaddr) + clear_rcvhdrtail(uctxt); + + rcvctrl_ops = HFI1_RCVCTRL_CTXT_ENB; + rcvctrl_ops |= HFI1_RCVCTRL_INTRAVAIL_ENB; + + if (!HFI1_CAP_KGET_MASK(uctxt->flags, MULTI_PKT_EGR)) + rcvctrl_ops |= HFI1_RCVCTRL_ONE_PKT_EGR_ENB; + if (HFI1_CAP_KGET_MASK(uctxt->flags, NODROP_EGR_FULL)) + rcvctrl_ops |= HFI1_RCVCTRL_NO_EGR_DROP_ENB; + if (HFI1_CAP_KGET_MASK(uctxt->flags, NODROP_RHQ_FULL)) + rcvctrl_ops |= HFI1_RCVCTRL_NO_RHQ_DROP_ENB; + if (HFI1_CAP_KGET_MASK(uctxt->flags, DMA_RTAIL)) + rcvctrl_ops |= HFI1_RCVCTRL_TAILUPD_ENB; + + hfi1_rcvctrl(uctxt->dd, rcvctrl_ops, uctxt->ctxt); + + uctxt->is_vnic = true; +done: + return ret; +} + +static int allocate_vnic_ctxt(struct hfi1_devdata *dd, + struct hfi1_ctxtdata **vnic_ctxt) +{ + struct hfi1_ctxtdata *uctxt; + unsigned int ctxt; + int ret; + + if (dd->flags & HFI1_FROZEN) + return -EIO; + + for (ctxt = dd->first_dyn_alloc_ctxt; + ctxt < dd->num_rcv_contexts; ctxt++) + if (!dd->rcd[ctxt]) + break; + + if (ctxt == dd->num_rcv_contexts) + return -EBUSY; + + uctxt = hfi1_create_ctxtdata(dd->pport, ctxt, dd->node); + if (!uctxt) { + dd_dev_err(dd, "Unable to create ctxtdata, failing open\n"); + return -ENOMEM; + } + + uctxt->flags = HFI1_CAP_KGET(MULTI_PKT_EGR) | + HFI1_CAP_KGET(NODROP_RHQ_FULL) | + HFI1_CAP_KGET(NODROP_EGR_FULL) | + HFI1_CAP_KGET(DMA_RTAIL); + uctxt->seq_cnt = 1; + + /* Allocate and enable a PIO send context */ + uctxt->sc = sc_alloc(dd, SC_VNIC, uctxt->rcvhdrqentsize, + uctxt->numa_id); + + ret = uctxt->sc ? 0 : -ENOMEM; + if (ret) + goto bail; + + dd_dev_dbg(dd, "allocated vnic send context %u(%u)\n", + uctxt->sc->sw_index, uctxt->sc->hw_context); + ret = sc_enable(uctxt->sc); + if (ret) + goto bail; + + if (dd->num_msix_entries) + hfi1_set_vnic_msix_info(uctxt); + + hfi1_stats.sps_ctxts++; + dd_dev_dbg(dd, "created vnic context %d\n", uctxt->ctxt); + *vnic_ctxt = uctxt; + + return ret; +bail: + /* + * hfi1_free_ctxtdata() also releases send_context + * structure if uctxt->sc is not null + */ + dd->rcd[uctxt->ctxt] = NULL; + hfi1_free_ctxtdata(dd, uctxt); + dd_dev_dbg(dd, "vnic allocation failed. rc %d\n", ret); + return ret; +} + +static void deallocate_vnic_ctxt(struct hfi1_devdata *dd, + struct hfi1_ctxtdata *uctxt) +{ + unsigned long flags; + + dd_dev_dbg(dd, "closing vnic context %d\n", uctxt->ctxt); + flush_wc(); + + if (dd->num_msix_entries) + hfi1_reset_vnic_msix_info(uctxt); + + spin_lock_irqsave(&dd->uctxt_lock, flags); + /* + * Disable receive context and interrupt available, reset all + * RcvCtxtCtrl bits to default values. + */ + hfi1_rcvctrl(dd, HFI1_RCVCTRL_CTXT_DIS | + HFI1_RCVCTRL_TIDFLOW_DIS | + HFI1_RCVCTRL_INTRAVAIL_DIS | + HFI1_RCVCTRL_ONE_PKT_EGR_DIS | + HFI1_RCVCTRL_NO_RHQ_DROP_DIS | + HFI1_RCVCTRL_NO_EGR_DROP_DIS, uctxt->ctxt); + /* + * VNIC contexts are allocated from user context pool. + * Release them back to user context pool. + * + * Reset context integrity checks to default. + * (writes to CSRs probably belong in chip.c) + */ + write_kctxt_csr(dd, uctxt->sc->hw_context, SEND_CTXT_CHECK_ENABLE, + hfi1_pkt_default_send_ctxt_mask(dd, SC_USER)); + sc_disable(uctxt->sc); + + dd->send_contexts[uctxt->sc->sw_index].type = SC_USER; + spin_unlock_irqrestore(&dd->uctxt_lock, flags); + + dd->rcd[uctxt->ctxt] = NULL; + uctxt->event_flags = 0; + + hfi1_clear_tids(uctxt); + hfi1_clear_ctxt_pkey(dd, uctxt->ctxt); + + hfi1_stats.sps_ctxts--; + hfi1_free_ctxtdata(dd, uctxt); +} + void hfi1_vnic_setup(struct hfi1_devdata *dd) { idr_init(&dd->vnic.vesw_idr); @@ -519,6 +672,9 @@ static void hfi1_vnic_down(struct hfi1_vnic_vport_info *vinfo) netif_tx_disable(vinfo->netdev); idr_remove(&dd->vnic.vesw_idr, vinfo->vesw_id); + /* ensure irqs see the change */ + hfi1_vnic_synchronize_irq(dd); + /* remove unread skbs */ for (i = 0; i < vinfo->num_rx_q; i++) { struct hfi1_vnic_rx_queue *rxq = &vinfo->rxq[i]; @@ -550,6 +706,84 @@ static int hfi1_netdev_close(struct net_device *netdev) return 0; } +static int hfi1_vnic_allot_ctxt(struct hfi1_devdata *dd, + struct hfi1_ctxtdata **vnic_ctxt) +{ + int rc; + + rc = allocate_vnic_ctxt(dd, vnic_ctxt); + if (rc) { + dd_dev_err(dd, "vnic ctxt alloc failed %d\n", rc); + return rc; + } + + rc = setup_vnic_ctxt(dd, *vnic_ctxt); + if (rc) { + dd_dev_err(dd, "vnic ctxt setup failed %d\n", rc); + deallocate_vnic_ctxt(dd, *vnic_ctxt); + *vnic_ctxt = NULL; + } + + return rc; +} + +static int hfi1_vnic_init(struct hfi1_vnic_vport_info *vinfo) +{ + struct hfi1_devdata *dd = vinfo->dd; + int i, rc = 0; + + mutex_lock(&hfi1_mutex); + if (!dd->vnic.num_vports) + dd->vnic.msix_idx = dd->first_dyn_msix_idx; + + for (i = dd->vnic.num_ctxt; i < vinfo->num_rx_q; i++) { + rc = hfi1_vnic_allot_ctxt(dd, &dd->vnic.ctxt[i]); + if (rc) + break; + dd->vnic.ctxt[i]->vnic_q_idx = i; + } + + if (i < vinfo->num_rx_q) { + /* + * If required amount of contexts is not + * allocated successfully then remaining contexts + * are released. + */ + while (i-- > dd->vnic.num_ctxt) { + deallocate_vnic_ctxt(dd, dd->vnic.ctxt[i]); + dd->vnic.ctxt[i] = NULL; + } + goto alloc_fail; + } + + if (dd->vnic.num_ctxt != i) { + dd->vnic.num_ctxt = i; + hfi1_init_vnic_rsm(dd); + } + + dd->vnic.num_vports++; +alloc_fail: + mutex_unlock(&hfi1_mutex); + return rc; +} + +static void hfi1_vnic_deinit(struct hfi1_vnic_vport_info *vinfo) +{ + struct hfi1_devdata *dd = vinfo->dd; + int i; + + mutex_lock(&hfi1_mutex); + if (--dd->vnic.num_vports == 0) { + for (i = 0; i < dd->vnic.num_ctxt; i++) { + deallocate_vnic_ctxt(dd, dd->vnic.ctxt[i]); + dd->vnic.ctxt[i] = NULL; + } + hfi1_deinit_vnic_rsm(dd); + dd->vnic.num_ctxt = 0; + } + mutex_unlock(&hfi1_mutex); +} + static void hfi1_vnic_set_vesw_id(struct net_device *netdev, int id) { struct hfi1_vnic_vport_info *vinfo = opa_vnic_dev_priv(netdev); @@ -594,7 +828,7 @@ struct net_device *hfi1_vnic_alloc_rn(struct ib_device *device, struct hfi1_vnic_vport_info *vinfo; struct net_device *netdev; struct rdma_netdev *rn; - int i, size; + int i, size, rc; if (!port_num || (port_num > dd->num_pports)) return ERR_PTR(-EINVAL); @@ -632,13 +866,22 @@ struct net_device *hfi1_vnic_alloc_rn(struct ib_device *device, netif_napi_add(netdev, &rxq->napi, hfi1_vnic_napi, 64); } + rc = hfi1_vnic_init(vinfo); + if (rc) + goto init_fail; + return netdev; +init_fail: + mutex_destroy(&vinfo->lock); + free_netdev(netdev); + return ERR_PTR(rc); } void hfi1_vnic_free_rn(struct net_device *netdev) { struct hfi1_vnic_vport_info *vinfo = opa_vnic_dev_priv(netdev); + hfi1_vnic_deinit(vinfo); mutex_destroy(&vinfo->lock); free_netdev(netdev); } diff --git a/include/rdma/opa_port_info.h b/include/rdma/opa_port_info.h index 9303e0e4f508..b4f0ac02f283 100644 --- a/include/rdma/opa_port_info.h +++ b/include/rdma/opa_port_info.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2014 Intel Corporation. All rights reserved. + * Copyright (c) 2014-2017 Intel Corporation. All rights reserved. * * This software is available to you under a choice of one of two * licenses. You may choose to be licensed under the terms of the GNU @@ -127,6 +127,7 @@ #define OPA_LINK_WIDTH_3X 0x0004 #define OPA_LINK_WIDTH_4X 0x0008 +#define OPA_CAP_MASK3_IsEthOnFabricSupported (1 << 13) #define OPA_CAP_MASK3_IsSnoopSupported (1 << 7) #define OPA_CAP_MASK3_IsAsyncSC2VLSupported (1 << 6) #define OPA_CAP_MASK3_IsAddrRangeConfigSupported (1 << 5) |