^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1) // SPDX-License-Identifier: GPL-2.0-or-later
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3) * RDMA Transport Layer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5) * Copyright (c) 2014 - 2018 ProfitBricks GmbH. All rights reserved.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 6) * Copyright (c) 2018 - 2019 1&1 IONOS Cloud GmbH. All rights reserved.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 7) * Copyright (c) 2019 - 2020 1&1 IONOS SE. All rights reserved.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 8) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 9)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 10) #undef pr_fmt
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 11) #define pr_fmt(fmt) KBUILD_MODNAME " L" __stringify(__LINE__) ": " fmt
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 12)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 13) #include <linux/module.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 14) #include <linux/rculist.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 15) #include <linux/random.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 16)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 17) #include "rtrs-clt.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 18) #include "rtrs-log.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 19)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 20) #define RTRS_CONNECT_TIMEOUT_MS 30000
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 21) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 22) * Wait a bit before trying to reconnect after a failure
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 23) * in order to give server time to finish clean up which
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 24) * leads to "false positives" failed reconnect attempts
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 25) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 26) #define RTRS_RECONNECT_BACKOFF 1000
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 27) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 28) * Wait for additional random time between 0 and 8 seconds
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 29) * before starting to reconnect to avoid clients reconnecting
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 30) * all at once in case of a major network outage
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 31) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 32) #define RTRS_RECONNECT_SEED 8
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 33)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 34) #define FIRST_CONN 0x01
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 35)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 36) MODULE_DESCRIPTION("RDMA Transport Client");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 37) MODULE_LICENSE("GPL");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 38)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 39) static const struct rtrs_rdma_dev_pd_ops dev_pd_ops;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 40) static struct rtrs_rdma_dev_pd dev_pd = {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 41) .ops = &dev_pd_ops
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 42) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 43)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 44) static struct workqueue_struct *rtrs_wq;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 45) static struct class *rtrs_clt_dev_class;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 46)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 47) static inline bool rtrs_clt_is_connected(const struct rtrs_clt *clt)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 48) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 49) struct rtrs_clt_sess *sess;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 50) bool connected = false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 51)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 52) rcu_read_lock();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 53) list_for_each_entry_rcu(sess, &clt->paths_list, s.entry)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 54) connected |= READ_ONCE(sess->state) == RTRS_CLT_CONNECTED;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 55) rcu_read_unlock();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 56)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 57) return connected;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 58) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 59)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 60) static struct rtrs_permit *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 61) __rtrs_get_permit(struct rtrs_clt *clt, enum rtrs_clt_con_type con_type)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 62) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 63) size_t max_depth = clt->queue_depth;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 64) struct rtrs_permit *permit;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 65) int bit;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 66)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 67) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 68) * Adapted from null_blk get_tag(). Callers from different cpus may
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 69) * grab the same bit, since find_first_zero_bit is not atomic.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 70) * But then the test_and_set_bit_lock will fail for all the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 71) * callers but one, so that they will loop again.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 72) * This way an explicit spinlock is not required.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 73) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 74) do {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 75) bit = find_first_zero_bit(clt->permits_map, max_depth);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 76) if (unlikely(bit >= max_depth))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 77) return NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 78) } while (unlikely(test_and_set_bit_lock(bit, clt->permits_map)));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 79)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 80) permit = get_permit(clt, bit);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 81) WARN_ON(permit->mem_id != bit);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 82) permit->cpu_id = raw_smp_processor_id();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 83) permit->con_type = con_type;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 84)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 85) return permit;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 86) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 87)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 88) static inline void __rtrs_put_permit(struct rtrs_clt *clt,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 89) struct rtrs_permit *permit)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 90) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 91) clear_bit_unlock(permit->mem_id, clt->permits_map);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 92) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 93)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 94) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 95) * rtrs_clt_get_permit() - allocates permit for future RDMA operation
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 96) * @clt: Current session
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 97) * @con_type: Type of connection to use with the permit
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 98) * @can_wait: Wait type
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 99) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 100) * Description:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 101) * Allocates permit for the following RDMA operation. Permit is used
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 102) * to preallocate all resources and to propagate memory pressure
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 103) * up earlier.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 104) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 105) * Context:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 106) * Can sleep if @wait == RTRS_TAG_WAIT
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 107) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 108) struct rtrs_permit *rtrs_clt_get_permit(struct rtrs_clt *clt,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 109) enum rtrs_clt_con_type con_type,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 110) int can_wait)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 111) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 112) struct rtrs_permit *permit;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 113) DEFINE_WAIT(wait);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 114)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 115) permit = __rtrs_get_permit(clt, con_type);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 116) if (likely(permit) || !can_wait)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 117) return permit;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 118)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 119) do {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 120) prepare_to_wait(&clt->permits_wait, &wait,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 121) TASK_UNINTERRUPTIBLE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 122) permit = __rtrs_get_permit(clt, con_type);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 123) if (likely(permit))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 124) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 125)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 126) io_schedule();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 127) } while (1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 128)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 129) finish_wait(&clt->permits_wait, &wait);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 130)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 131) return permit;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 132) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 133) EXPORT_SYMBOL(rtrs_clt_get_permit);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 134)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 135) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 136) * rtrs_clt_put_permit() - puts allocated permit
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 137) * @clt: Current session
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 138) * @permit: Permit to be freed
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 139) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 140) * Context:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 141) * Does not matter
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 142) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 143) void rtrs_clt_put_permit(struct rtrs_clt *clt, struct rtrs_permit *permit)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 144) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 145) if (WARN_ON(!test_bit(permit->mem_id, clt->permits_map)))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 146) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 147)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 148) __rtrs_put_permit(clt, permit);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 149)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 150) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 151) * rtrs_clt_get_permit() adds itself to the &clt->permits_wait list
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 152) * before calling schedule(). So if rtrs_clt_get_permit() is sleeping
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 153) * it must have added itself to &clt->permits_wait before
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 154) * __rtrs_put_permit() finished.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 155) * Hence it is safe to guard wake_up() with a waitqueue_active() test.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 156) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 157) if (waitqueue_active(&clt->permits_wait))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 158) wake_up(&clt->permits_wait);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 159) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 160) EXPORT_SYMBOL(rtrs_clt_put_permit);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 161)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 162) void *rtrs_permit_to_pdu(struct rtrs_permit *permit)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 163) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 164) return permit + 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 165) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 166) EXPORT_SYMBOL(rtrs_permit_to_pdu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 167)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 168) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 169) * rtrs_permit_to_clt_con() - returns RDMA connection pointer by the permit
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 170) * @sess: client session pointer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 171) * @permit: permit for the allocation of the RDMA buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 172) * Note:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 173) * IO connection starts from 1.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 174) * 0 connection is for user messages.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 175) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 176) static
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 177) struct rtrs_clt_con *rtrs_permit_to_clt_con(struct rtrs_clt_sess *sess,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 178) struct rtrs_permit *permit)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 179) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 180) int id = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 181)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 182) if (likely(permit->con_type == RTRS_IO_CON))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 183) id = (permit->cpu_id % (sess->s.con_num - 1)) + 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 184)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 185) return to_clt_con(sess->s.con[id]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 186) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 187)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 188) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 189) * __rtrs_clt_change_state() - change the session state through session state
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 190) * machine.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 191) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 192) * @sess: client session to change the state of.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 193) * @new_state: state to change to.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 194) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 195) * returns true if successful, false if the requested state can not be set.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 196) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 197) * Locks:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 198) * state_wq lock must be hold.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 199) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 200) static bool __rtrs_clt_change_state(struct rtrs_clt_sess *sess,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 201) enum rtrs_clt_state new_state)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 202) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 203) enum rtrs_clt_state old_state;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 204) bool changed = false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 205)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 206) lockdep_assert_held(&sess->state_wq.lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 207)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 208) old_state = sess->state;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 209) switch (new_state) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 210) case RTRS_CLT_CONNECTING:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 211) switch (old_state) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 212) case RTRS_CLT_RECONNECTING:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 213) changed = true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 214) fallthrough;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 215) default:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 216) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 217) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 218) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 219) case RTRS_CLT_RECONNECTING:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 220) switch (old_state) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 221) case RTRS_CLT_CONNECTED:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 222) case RTRS_CLT_CONNECTING_ERR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 223) case RTRS_CLT_CLOSED:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 224) changed = true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 225) fallthrough;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 226) default:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 227) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 228) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 229) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 230) case RTRS_CLT_CONNECTED:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 231) switch (old_state) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 232) case RTRS_CLT_CONNECTING:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 233) changed = true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 234) fallthrough;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 235) default:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 236) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 237) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 238) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 239) case RTRS_CLT_CONNECTING_ERR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 240) switch (old_state) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 241) case RTRS_CLT_CONNECTING:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 242) changed = true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 243) fallthrough;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 244) default:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 245) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 246) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 247) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 248) case RTRS_CLT_CLOSING:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 249) switch (old_state) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 250) case RTRS_CLT_CONNECTING:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 251) case RTRS_CLT_CONNECTING_ERR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 252) case RTRS_CLT_RECONNECTING:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 253) case RTRS_CLT_CONNECTED:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 254) changed = true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 255) fallthrough;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 256) default:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 257) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 258) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 259) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 260) case RTRS_CLT_CLOSED:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 261) switch (old_state) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 262) case RTRS_CLT_CLOSING:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 263) changed = true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 264) fallthrough;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 265) default:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 266) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 267) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 268) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 269) case RTRS_CLT_DEAD:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 270) switch (old_state) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 271) case RTRS_CLT_CLOSED:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 272) changed = true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 273) fallthrough;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 274) default:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 275) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 276) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 277) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 278) default:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 279) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 280) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 281) if (changed) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 282) sess->state = new_state;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 283) wake_up_locked(&sess->state_wq);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 284) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 285)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 286) return changed;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 287) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 288)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 289) static bool rtrs_clt_change_state_from_to(struct rtrs_clt_sess *sess,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 290) enum rtrs_clt_state old_state,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 291) enum rtrs_clt_state new_state)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 292) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 293) bool changed = false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 294)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 295) spin_lock_irq(&sess->state_wq.lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 296) if (sess->state == old_state)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 297) changed = __rtrs_clt_change_state(sess, new_state);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 298) spin_unlock_irq(&sess->state_wq.lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 299)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 300) return changed;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 301) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 302)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 303) static void rtrs_rdma_error_recovery(struct rtrs_clt_con *con)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 304) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 305) struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 306)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 307) if (rtrs_clt_change_state_from_to(sess,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 308) RTRS_CLT_CONNECTED,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 309) RTRS_CLT_RECONNECTING)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 310) struct rtrs_clt *clt = sess->clt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 311) unsigned int delay_ms;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 312)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 313) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 314) * Normal scenario, reconnect if we were successfully connected
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 315) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 316) delay_ms = clt->reconnect_delay_sec * 1000;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 317) queue_delayed_work(rtrs_wq, &sess->reconnect_dwork,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 318) msecs_to_jiffies(delay_ms +
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 319) prandom_u32() % RTRS_RECONNECT_SEED));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 320) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 321) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 322) * Error can happen just on establishing new connection,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 323) * so notify waiter with error state, waiter is responsible
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 324) * for cleaning the rest and reconnect if needed.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 325) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 326) rtrs_clt_change_state_from_to(sess,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 327) RTRS_CLT_CONNECTING,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 328) RTRS_CLT_CONNECTING_ERR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 329) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 330) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 331)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 332) static void rtrs_clt_fast_reg_done(struct ib_cq *cq, struct ib_wc *wc)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 333) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 334) struct rtrs_clt_con *con = cq->cq_context;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 335)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 336) if (unlikely(wc->status != IB_WC_SUCCESS)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 337) rtrs_err(con->c.sess, "Failed IB_WR_REG_MR: %s\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 338) ib_wc_status_msg(wc->status));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 339) rtrs_rdma_error_recovery(con);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 340) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 341) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 342)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 343) static struct ib_cqe fast_reg_cqe = {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 344) .done = rtrs_clt_fast_reg_done
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 345) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 346)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 347) static void complete_rdma_req(struct rtrs_clt_io_req *req, int errno,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 348) bool notify, bool can_wait);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 349)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 350) static void rtrs_clt_inv_rkey_done(struct ib_cq *cq, struct ib_wc *wc)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 351) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 352) struct rtrs_clt_io_req *req =
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 353) container_of(wc->wr_cqe, typeof(*req), inv_cqe);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 354) struct rtrs_clt_con *con = cq->cq_context;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 355)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 356) if (unlikely(wc->status != IB_WC_SUCCESS)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 357) rtrs_err(con->c.sess, "Failed IB_WR_LOCAL_INV: %s\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 358) ib_wc_status_msg(wc->status));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 359) rtrs_rdma_error_recovery(con);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 360) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 361) req->need_inv = false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 362) if (likely(req->need_inv_comp))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 363) complete(&req->inv_comp);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 364) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 365) /* Complete request from INV callback */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 366) complete_rdma_req(req, req->inv_errno, true, false);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 367) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 368)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 369) static int rtrs_inv_rkey(struct rtrs_clt_io_req *req)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 370) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 371) struct rtrs_clt_con *con = req->con;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 372) struct ib_send_wr wr = {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 373) .opcode = IB_WR_LOCAL_INV,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 374) .wr_cqe = &req->inv_cqe,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 375) .send_flags = IB_SEND_SIGNALED,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 376) .ex.invalidate_rkey = req->mr->rkey,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 377) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 378) req->inv_cqe.done = rtrs_clt_inv_rkey_done;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 379)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 380) return ib_post_send(con->c.qp, &wr, NULL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 381) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 382)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 383) static void complete_rdma_req(struct rtrs_clt_io_req *req, int errno,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 384) bool notify, bool can_wait)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 385) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 386) struct rtrs_clt_con *con = req->con;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 387) struct rtrs_clt_sess *sess;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 388) int err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 389)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 390) if (WARN_ON(!req->in_use))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 391) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 392) if (WARN_ON(!req->con))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 393) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 394) sess = to_clt_sess(con->c.sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 395)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 396) if (req->sg_cnt) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 397) if (unlikely(req->dir == DMA_FROM_DEVICE && req->need_inv)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 398) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 399) * We are here to invalidate read requests
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 400) * ourselves. In normal scenario server should
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 401) * send INV for all read requests, but
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 402) * we are here, thus two things could happen:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 403) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 404) * 1. this is failover, when errno != 0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 405) * and can_wait == 1,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 406) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 407) * 2. something totally bad happened and
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 408) * server forgot to send INV, so we
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 409) * should do that ourselves.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 410) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 411)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 412) if (likely(can_wait)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 413) req->need_inv_comp = true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 414) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 415) /* This should be IO path, so always notify */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 416) WARN_ON(!notify);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 417) /* Save errno for INV callback */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 418) req->inv_errno = errno;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 419) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 420)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 421) err = rtrs_inv_rkey(req);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 422) if (unlikely(err)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 423) rtrs_err(con->c.sess, "Send INV WR key=%#x: %d\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 424) req->mr->rkey, err);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 425) } else if (likely(can_wait)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 426) wait_for_completion(&req->inv_comp);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 427) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 428) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 429) * Something went wrong, so request will be
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 430) * completed from INV callback.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 431) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 432) WARN_ON_ONCE(1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 433)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 434) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 435) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 436) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 437) ib_dma_unmap_sg(sess->s.dev->ib_dev, req->sglist,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 438) req->sg_cnt, req->dir);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 439) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 440) if (sess->clt->mp_policy == MP_POLICY_MIN_INFLIGHT)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 441) atomic_dec(&sess->stats->inflight);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 442)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 443) req->in_use = false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 444) req->con = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 445)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 446) if (notify)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 447) req->conf(req->priv, errno);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 448) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 449)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 450) static int rtrs_post_send_rdma(struct rtrs_clt_con *con,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 451) struct rtrs_clt_io_req *req,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 452) struct rtrs_rbuf *rbuf, u32 off,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 453) u32 imm, struct ib_send_wr *wr)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 454) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 455) struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 456) enum ib_send_flags flags;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 457) struct ib_sge sge;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 458)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 459) if (unlikely(!req->sg_size)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 460) rtrs_wrn(con->c.sess,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 461) "Doing RDMA Write failed, no data supplied\n");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 462) return -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 463) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 464)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 465) /* user data and user message in the first list element */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 466) sge.addr = req->iu->dma_addr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 467) sge.length = req->sg_size;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 468) sge.lkey = sess->s.dev->ib_pd->local_dma_lkey;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 469)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 470) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 471) * From time to time we have to post signalled sends,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 472) * or send queue will fill up and only QP reset can help.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 473) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 474) flags = atomic_inc_return(&con->io_cnt) % sess->queue_depth ?
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 475) 0 : IB_SEND_SIGNALED;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 476)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 477) ib_dma_sync_single_for_device(sess->s.dev->ib_dev, req->iu->dma_addr,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 478) req->sg_size, DMA_TO_DEVICE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 479)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 480) return rtrs_iu_post_rdma_write_imm(&con->c, req->iu, &sge, 1,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 481) rbuf->rkey, rbuf->addr + off,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 482) imm, flags, wr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 483) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 484)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 485) static void process_io_rsp(struct rtrs_clt_sess *sess, u32 msg_id,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 486) s16 errno, bool w_inval)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 487) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 488) struct rtrs_clt_io_req *req;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 489)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 490) if (WARN_ON(msg_id >= sess->queue_depth))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 491) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 492)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 493) req = &sess->reqs[msg_id];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 494) /* Drop need_inv if server responded with send with invalidation */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 495) req->need_inv &= !w_inval;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 496) complete_rdma_req(req, errno, true, false);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 497) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 498)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 499) static void rtrs_clt_recv_done(struct rtrs_clt_con *con, struct ib_wc *wc)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 500) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 501) struct rtrs_iu *iu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 502) int err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 503) struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 504)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 505) WARN_ON((sess->flags & RTRS_MSG_NEW_RKEY_F) == 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 506) iu = container_of(wc->wr_cqe, struct rtrs_iu,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 507) cqe);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 508) err = rtrs_iu_post_recv(&con->c, iu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 509) if (unlikely(err)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 510) rtrs_err(con->c.sess, "post iu failed %d\n", err);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 511) rtrs_rdma_error_recovery(con);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 512) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 513) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 514)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 515) static void rtrs_clt_rkey_rsp_done(struct rtrs_clt_con *con, struct ib_wc *wc)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 516) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 517) struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 518) struct rtrs_msg_rkey_rsp *msg;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 519) u32 imm_type, imm_payload;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 520) bool w_inval = false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 521) struct rtrs_iu *iu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 522) u32 buf_id;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 523) int err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 524)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 525) WARN_ON((sess->flags & RTRS_MSG_NEW_RKEY_F) == 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 526)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 527) iu = container_of(wc->wr_cqe, struct rtrs_iu, cqe);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 528)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 529) if (unlikely(wc->byte_len < sizeof(*msg))) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 530) rtrs_err(con->c.sess, "rkey response is malformed: size %d\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 531) wc->byte_len);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 532) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 533) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 534) ib_dma_sync_single_for_cpu(sess->s.dev->ib_dev, iu->dma_addr,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 535) iu->size, DMA_FROM_DEVICE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 536) msg = iu->buf;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 537) if (unlikely(le16_to_cpu(msg->type) != RTRS_MSG_RKEY_RSP)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 538) rtrs_err(sess->clt, "rkey response is malformed: type %d\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 539) le16_to_cpu(msg->type));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 540) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 541) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 542) buf_id = le16_to_cpu(msg->buf_id);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 543) if (WARN_ON(buf_id >= sess->queue_depth))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 544) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 545)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 546) rtrs_from_imm(be32_to_cpu(wc->ex.imm_data), &imm_type, &imm_payload);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 547) if (likely(imm_type == RTRS_IO_RSP_IMM ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 548) imm_type == RTRS_IO_RSP_W_INV_IMM)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 549) u32 msg_id;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 550)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 551) w_inval = (imm_type == RTRS_IO_RSP_W_INV_IMM);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 552) rtrs_from_io_rsp_imm(imm_payload, &msg_id, &err);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 553)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 554) if (WARN_ON(buf_id != msg_id))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 555) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 556) sess->rbufs[buf_id].rkey = le32_to_cpu(msg->rkey);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 557) process_io_rsp(sess, msg_id, err, w_inval);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 558) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 559) ib_dma_sync_single_for_device(sess->s.dev->ib_dev, iu->dma_addr,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 560) iu->size, DMA_FROM_DEVICE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 561) return rtrs_clt_recv_done(con, wc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 562) out:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 563) rtrs_rdma_error_recovery(con);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 564) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 565)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 566) static void rtrs_clt_rdma_done(struct ib_cq *cq, struct ib_wc *wc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 567)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 568) static struct ib_cqe io_comp_cqe = {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 569) .done = rtrs_clt_rdma_done
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 570) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 571)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 572) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 573) * Post x2 empty WRs: first is for this RDMA with IMM,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 574) * second is for RECV with INV, which happened earlier.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 575) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 576) static int rtrs_post_recv_empty_x2(struct rtrs_con *con, struct ib_cqe *cqe)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 577) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 578) struct ib_recv_wr wr_arr[2], *wr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 579) int i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 580)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 581) memset(wr_arr, 0, sizeof(wr_arr));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 582) for (i = 0; i < ARRAY_SIZE(wr_arr); i++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 583) wr = &wr_arr[i];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 584) wr->wr_cqe = cqe;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 585) if (i)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 586) /* Chain backwards */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 587) wr->next = &wr_arr[i - 1];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 588) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 589)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 590) return ib_post_recv(con->qp, wr, NULL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 591) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 592)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 593) static void rtrs_clt_rdma_done(struct ib_cq *cq, struct ib_wc *wc)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 594) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 595) struct rtrs_clt_con *con = cq->cq_context;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 596) struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 597) u32 imm_type, imm_payload;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 598) bool w_inval = false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 599) int err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 600)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 601) if (unlikely(wc->status != IB_WC_SUCCESS)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 602) if (wc->status != IB_WC_WR_FLUSH_ERR) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 603) rtrs_err(sess->clt, "RDMA failed: %s\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 604) ib_wc_status_msg(wc->status));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 605) rtrs_rdma_error_recovery(con);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 606) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 607) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 608) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 609) rtrs_clt_update_wc_stats(con);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 610)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 611) switch (wc->opcode) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 612) case IB_WC_RECV_RDMA_WITH_IMM:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 613) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 614) * post_recv() RDMA write completions of IO reqs (read/write)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 615) * and hb
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 616) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 617) if (WARN_ON(wc->wr_cqe->done != rtrs_clt_rdma_done))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 618) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 619) rtrs_from_imm(be32_to_cpu(wc->ex.imm_data),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 620) &imm_type, &imm_payload);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 621) if (likely(imm_type == RTRS_IO_RSP_IMM ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 622) imm_type == RTRS_IO_RSP_W_INV_IMM)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 623) u32 msg_id;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 624)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 625) w_inval = (imm_type == RTRS_IO_RSP_W_INV_IMM);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 626) rtrs_from_io_rsp_imm(imm_payload, &msg_id, &err);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 627)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 628) process_io_rsp(sess, msg_id, err, w_inval);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 629) } else if (imm_type == RTRS_HB_MSG_IMM) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 630) WARN_ON(con->c.cid);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 631) rtrs_send_hb_ack(&sess->s);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 632) if (sess->flags & RTRS_MSG_NEW_RKEY_F)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 633) return rtrs_clt_recv_done(con, wc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 634) } else if (imm_type == RTRS_HB_ACK_IMM) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 635) WARN_ON(con->c.cid);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 636) sess->s.hb_missed_cnt = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 637) if (sess->flags & RTRS_MSG_NEW_RKEY_F)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 638) return rtrs_clt_recv_done(con, wc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 639) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 640) rtrs_wrn(con->c.sess, "Unknown IMM type %u\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 641) imm_type);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 642) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 643) if (w_inval)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 644) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 645) * Post x2 empty WRs: first is for this RDMA with IMM,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 646) * second is for RECV with INV, which happened earlier.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 647) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 648) err = rtrs_post_recv_empty_x2(&con->c, &io_comp_cqe);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 649) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 650) err = rtrs_post_recv_empty(&con->c, &io_comp_cqe);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 651) if (unlikely(err)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 652) rtrs_err(con->c.sess, "rtrs_post_recv_empty(): %d\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 653) err);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 654) rtrs_rdma_error_recovery(con);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 655) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 656) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 657) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 658) case IB_WC_RECV:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 659) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 660) * Key invalidations from server side
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 661) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 662) WARN_ON(!(wc->wc_flags & IB_WC_WITH_INVALIDATE ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 663) wc->wc_flags & IB_WC_WITH_IMM));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 664) WARN_ON(wc->wr_cqe->done != rtrs_clt_rdma_done);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 665) if (sess->flags & RTRS_MSG_NEW_RKEY_F) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 666) if (wc->wc_flags & IB_WC_WITH_INVALIDATE)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 667) return rtrs_clt_recv_done(con, wc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 668)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 669) return rtrs_clt_rkey_rsp_done(con, wc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 670) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 671) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 672) case IB_WC_RDMA_WRITE:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 673) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 674) * post_send() RDMA write completions of IO reqs (read/write)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 675) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 676) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 677)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 678) default:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 679) rtrs_wrn(sess->clt, "Unexpected WC type: %d\n", wc->opcode);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 680) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 681) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 682) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 683)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 684) static int post_recv_io(struct rtrs_clt_con *con, size_t q_size)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 685) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 686) int err, i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 687) struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 688)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 689) for (i = 0; i < q_size; i++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 690) if (sess->flags & RTRS_MSG_NEW_RKEY_F) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 691) struct rtrs_iu *iu = &con->rsp_ius[i];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 692)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 693) err = rtrs_iu_post_recv(&con->c, iu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 694) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 695) err = rtrs_post_recv_empty(&con->c, &io_comp_cqe);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 696) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 697) if (unlikely(err))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 698) return err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 699) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 700)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 701) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 702) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 703)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 704) static int post_recv_sess(struct rtrs_clt_sess *sess)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 705) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 706) size_t q_size = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 707) int err, cid;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 708)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 709) for (cid = 0; cid < sess->s.con_num; cid++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 710) if (cid == 0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 711) q_size = SERVICE_CON_QUEUE_DEPTH;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 712) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 713) q_size = sess->queue_depth;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 714)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 715) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 716) * x2 for RDMA read responses + FR key invalidations,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 717) * RDMA writes do not require any FR registrations.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 718) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 719) q_size *= 2;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 720)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 721) err = post_recv_io(to_clt_con(sess->s.con[cid]), q_size);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 722) if (unlikely(err)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 723) rtrs_err(sess->clt, "post_recv_io(), err: %d\n", err);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 724) return err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 725) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 726) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 727)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 728) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 729) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 730)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 731) struct path_it {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 732) int i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 733) struct list_head skip_list;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 734) struct rtrs_clt *clt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 735) struct rtrs_clt_sess *(*next_path)(struct path_it *it);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 736) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 737)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 738) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 739) * list_next_or_null_rr_rcu - get next list element in round-robin fashion.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 740) * @head: the head for the list.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 741) * @ptr: the list head to take the next element from.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 742) * @type: the type of the struct this is embedded in.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 743) * @memb: the name of the list_head within the struct.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 744) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 745) * Next element returned in round-robin fashion, i.e. head will be skipped,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 746) * but if list is observed as empty, NULL will be returned.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 747) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 748) * This primitive may safely run concurrently with the _rcu list-mutation
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 749) * primitives such as list_add_rcu() as long as it's guarded by rcu_read_lock().
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 750) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 751) #define list_next_or_null_rr_rcu(head, ptr, type, memb) \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 752) ({ \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 753) list_next_or_null_rcu(head, ptr, type, memb) ?: \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 754) list_next_or_null_rcu(head, READ_ONCE((ptr)->next), \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 755) type, memb); \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 756) })
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 757)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 758) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 759) * get_next_path_rr() - Returns path in round-robin fashion.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 760) * @it: the path pointer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 761) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 762) * Related to @MP_POLICY_RR
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 763) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 764) * Locks:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 765) * rcu_read_lock() must be hold.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 766) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 767) static struct rtrs_clt_sess *get_next_path_rr(struct path_it *it)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 768) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 769) struct rtrs_clt_sess __rcu **ppcpu_path;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 770) struct rtrs_clt_sess *path;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 771) struct rtrs_clt *clt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 772)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 773) clt = it->clt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 774)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 775) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 776) * Here we use two RCU objects: @paths_list and @pcpu_path
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 777) * pointer. See rtrs_clt_remove_path_from_arr() for details
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 778) * how that is handled.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 779) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 780)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 781) ppcpu_path = this_cpu_ptr(clt->pcpu_path);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 782) path = rcu_dereference(*ppcpu_path);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 783) if (unlikely(!path))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 784) path = list_first_or_null_rcu(&clt->paths_list,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 785) typeof(*path), s.entry);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 786) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 787) path = list_next_or_null_rr_rcu(&clt->paths_list,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 788) &path->s.entry,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 789) typeof(*path),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 790) s.entry);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 791) rcu_assign_pointer(*ppcpu_path, path);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 792)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 793) return path;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 794) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 795)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 796) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 797) * get_next_path_min_inflight() - Returns path with minimal inflight count.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 798) * @it: the path pointer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 799) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 800) * Related to @MP_POLICY_MIN_INFLIGHT
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 801) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 802) * Locks:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 803) * rcu_read_lock() must be hold.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 804) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 805) static struct rtrs_clt_sess *get_next_path_min_inflight(struct path_it *it)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 806) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 807) struct rtrs_clt_sess *min_path = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 808) struct rtrs_clt *clt = it->clt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 809) struct rtrs_clt_sess *sess;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 810) int min_inflight = INT_MAX;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 811) int inflight;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 812)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 813) list_for_each_entry_rcu(sess, &clt->paths_list, s.entry) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 814) if (unlikely(READ_ONCE(sess->state) != RTRS_CLT_CONNECTED))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 815) continue;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 816)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 817) if (unlikely(!list_empty(raw_cpu_ptr(sess->mp_skip_entry))))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 818) continue;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 819)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 820) inflight = atomic_read(&sess->stats->inflight);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 821)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 822) if (inflight < min_inflight) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 823) min_inflight = inflight;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 824) min_path = sess;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 825) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 826) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 827)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 828) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 829) * add the path to the skip list, so that next time we can get
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 830) * a different one
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 831) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 832) if (min_path)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 833) list_add(raw_cpu_ptr(min_path->mp_skip_entry), &it->skip_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 834)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 835) return min_path;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 836) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 837)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 838) static inline void path_it_init(struct path_it *it, struct rtrs_clt *clt)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 839) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 840) INIT_LIST_HEAD(&it->skip_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 841) it->clt = clt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 842) it->i = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 843)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 844) if (clt->mp_policy == MP_POLICY_RR)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 845) it->next_path = get_next_path_rr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 846) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 847) it->next_path = get_next_path_min_inflight;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 848) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 849)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 850) static inline void path_it_deinit(struct path_it *it)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 851) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 852) struct list_head *skip, *tmp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 853) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 854) * The skip_list is used only for the MIN_INFLIGHT policy.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 855) * We need to remove paths from it, so that next IO can insert
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 856) * paths (->mp_skip_entry) into a skip_list again.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 857) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 858) list_for_each_safe(skip, tmp, &it->skip_list)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 859) list_del_init(skip);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 860) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 861)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 862) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 863) * rtrs_clt_init_req() Initialize an rtrs_clt_io_req holding information
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 864) * about an inflight IO.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 865) * The user buffer holding user control message (not data) is copied into
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 866) * the corresponding buffer of rtrs_iu (req->iu->buf), which later on will
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 867) * also hold the control message of rtrs.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 868) * @req: an io request holding information about IO.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 869) * @sess: client session
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 870) * @conf: conformation callback function to notify upper layer.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 871) * @permit: permit for allocation of RDMA remote buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 872) * @priv: private pointer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 873) * @vec: kernel vector containing control message
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 874) * @usr_len: length of the user message
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 875) * @sg: scater list for IO data
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 876) * @sg_cnt: number of scater list entries
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 877) * @data_len: length of the IO data
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 878) * @dir: direction of the IO.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 879) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 880) static void rtrs_clt_init_req(struct rtrs_clt_io_req *req,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 881) struct rtrs_clt_sess *sess,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 882) void (*conf)(void *priv, int errno),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 883) struct rtrs_permit *permit, void *priv,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 884) const struct kvec *vec, size_t usr_len,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 885) struct scatterlist *sg, size_t sg_cnt,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 886) size_t data_len, int dir)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 887) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 888) struct iov_iter iter;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 889) size_t len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 890)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 891) req->permit = permit;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 892) req->in_use = true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 893) req->usr_len = usr_len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 894) req->data_len = data_len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 895) req->sglist = sg;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 896) req->sg_cnt = sg_cnt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 897) req->priv = priv;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 898) req->dir = dir;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 899) req->con = rtrs_permit_to_clt_con(sess, permit);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 900) req->conf = conf;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 901) req->need_inv = false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 902) req->need_inv_comp = false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 903) req->inv_errno = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 904)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 905) iov_iter_kvec(&iter, READ, vec, 1, usr_len);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 906) len = _copy_from_iter(req->iu->buf, usr_len, &iter);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 907) WARN_ON(len != usr_len);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 908)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 909) reinit_completion(&req->inv_comp);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 910) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 911)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 912) static struct rtrs_clt_io_req *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 913) rtrs_clt_get_req(struct rtrs_clt_sess *sess,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 914) void (*conf)(void *priv, int errno),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 915) struct rtrs_permit *permit, void *priv,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 916) const struct kvec *vec, size_t usr_len,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 917) struct scatterlist *sg, size_t sg_cnt,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 918) size_t data_len, int dir)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 919) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 920) struct rtrs_clt_io_req *req;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 921)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 922) req = &sess->reqs[permit->mem_id];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 923) rtrs_clt_init_req(req, sess, conf, permit, priv, vec, usr_len,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 924) sg, sg_cnt, data_len, dir);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 925) return req;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 926) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 927)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 928) static struct rtrs_clt_io_req *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 929) rtrs_clt_get_copy_req(struct rtrs_clt_sess *alive_sess,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 930) struct rtrs_clt_io_req *fail_req)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 931) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 932) struct rtrs_clt_io_req *req;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 933) struct kvec vec = {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 934) .iov_base = fail_req->iu->buf,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 935) .iov_len = fail_req->usr_len
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 936) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 937)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 938) req = &alive_sess->reqs[fail_req->permit->mem_id];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 939) rtrs_clt_init_req(req, alive_sess, fail_req->conf, fail_req->permit,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 940) fail_req->priv, &vec, fail_req->usr_len,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 941) fail_req->sglist, fail_req->sg_cnt,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 942) fail_req->data_len, fail_req->dir);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 943) return req;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 944) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 945)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 946) static int rtrs_post_rdma_write_sg(struct rtrs_clt_con *con,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 947) struct rtrs_clt_io_req *req,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 948) struct rtrs_rbuf *rbuf,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 949) u32 size, u32 imm)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 950) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 951) struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 952) struct ib_sge *sge = req->sge;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 953) enum ib_send_flags flags;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 954) struct scatterlist *sg;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 955) size_t num_sge;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 956) int i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 957)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 958) for_each_sg(req->sglist, sg, req->sg_cnt, i) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 959) sge[i].addr = sg_dma_address(sg);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 960) sge[i].length = sg_dma_len(sg);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 961) sge[i].lkey = sess->s.dev->ib_pd->local_dma_lkey;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 962) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 963) sge[i].addr = req->iu->dma_addr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 964) sge[i].length = size;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 965) sge[i].lkey = sess->s.dev->ib_pd->local_dma_lkey;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 966)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 967) num_sge = 1 + req->sg_cnt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 968)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 969) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 970) * From time to time we have to post signalled sends,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 971) * or send queue will fill up and only QP reset can help.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 972) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 973) flags = atomic_inc_return(&con->io_cnt) % sess->queue_depth ?
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 974) 0 : IB_SEND_SIGNALED;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 975)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 976) ib_dma_sync_single_for_device(sess->s.dev->ib_dev, req->iu->dma_addr,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 977) size, DMA_TO_DEVICE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 978)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 979) return rtrs_iu_post_rdma_write_imm(&con->c, req->iu, sge, num_sge,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 980) rbuf->rkey, rbuf->addr, imm,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 981) flags, NULL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 982) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 983)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 984) static int rtrs_clt_write_req(struct rtrs_clt_io_req *req)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 985) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 986) struct rtrs_clt_con *con = req->con;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 987) struct rtrs_sess *s = con->c.sess;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 988) struct rtrs_clt_sess *sess = to_clt_sess(s);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 989) struct rtrs_msg_rdma_write *msg;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 990)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 991) struct rtrs_rbuf *rbuf;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 992) int ret, count = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 993) u32 imm, buf_id;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 994)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 995) const size_t tsize = sizeof(*msg) + req->data_len + req->usr_len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 996)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 997) if (unlikely(tsize > sess->chunk_size)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 998) rtrs_wrn(s, "Write request failed, size too big %zu > %d\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 999) tsize, sess->chunk_size);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1000) return -EMSGSIZE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1001) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1002) if (req->sg_cnt) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1003) count = ib_dma_map_sg(sess->s.dev->ib_dev, req->sglist,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1004) req->sg_cnt, req->dir);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1005) if (unlikely(!count)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1006) rtrs_wrn(s, "Write request failed, map failed\n");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1007) return -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1008) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1009) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1010) /* put rtrs msg after sg and user message */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1011) msg = req->iu->buf + req->usr_len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1012) msg->type = cpu_to_le16(RTRS_MSG_WRITE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1013) msg->usr_len = cpu_to_le16(req->usr_len);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1014)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1015) /* rtrs message on server side will be after user data and message */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1016) imm = req->permit->mem_off + req->data_len + req->usr_len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1017) imm = rtrs_to_io_req_imm(imm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1018) buf_id = req->permit->mem_id;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1019) req->sg_size = tsize;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1020) rbuf = &sess->rbufs[buf_id];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1021)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1022) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1023) * Update stats now, after request is successfully sent it is not
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1024) * safe anymore to touch it.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1025) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1026) rtrs_clt_update_all_stats(req, WRITE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1027)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1028) ret = rtrs_post_rdma_write_sg(req->con, req, rbuf,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1029) req->usr_len + sizeof(*msg),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1030) imm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1031) if (unlikely(ret)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1032) rtrs_err(s, "Write request failed: %d\n", ret);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1033) if (sess->clt->mp_policy == MP_POLICY_MIN_INFLIGHT)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1034) atomic_dec(&sess->stats->inflight);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1035) if (req->sg_cnt)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1036) ib_dma_unmap_sg(sess->s.dev->ib_dev, req->sglist,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1037) req->sg_cnt, req->dir);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1038) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1039)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1040) return ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1041) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1042)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1043) static int rtrs_map_sg_fr(struct rtrs_clt_io_req *req, size_t count)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1044) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1045) int nr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1046)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1047) /* Align the MR to a 4K page size to match the block virt boundary */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1048) nr = ib_map_mr_sg(req->mr, req->sglist, count, NULL, SZ_4K);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1049) if (nr < 0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1050) return nr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1051) if (unlikely(nr < req->sg_cnt))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1052) return -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1053) ib_update_fast_reg_key(req->mr, ib_inc_rkey(req->mr->rkey));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1054)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1055) return nr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1056) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1057)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1058) static int rtrs_clt_read_req(struct rtrs_clt_io_req *req)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1059) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1060) struct rtrs_clt_con *con = req->con;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1061) struct rtrs_sess *s = con->c.sess;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1062) struct rtrs_clt_sess *sess = to_clt_sess(s);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1063) struct rtrs_msg_rdma_read *msg;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1064) struct rtrs_ib_dev *dev;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1065)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1066) struct ib_reg_wr rwr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1067) struct ib_send_wr *wr = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1068)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1069) int ret, count = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1070) u32 imm, buf_id;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1071)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1072) const size_t tsize = sizeof(*msg) + req->data_len + req->usr_len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1073)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1074) s = &sess->s;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1075) dev = sess->s.dev;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1076)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1077) if (unlikely(tsize > sess->chunk_size)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1078) rtrs_wrn(s,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1079) "Read request failed, message size is %zu, bigger than CHUNK_SIZE %d\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1080) tsize, sess->chunk_size);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1081) return -EMSGSIZE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1082) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1083)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1084) if (req->sg_cnt) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1085) count = ib_dma_map_sg(dev->ib_dev, req->sglist, req->sg_cnt,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1086) req->dir);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1087) if (unlikely(!count)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1088) rtrs_wrn(s,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1089) "Read request failed, dma map failed\n");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1090) return -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1091) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1092) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1093) /* put our message into req->buf after user message*/
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1094) msg = req->iu->buf + req->usr_len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1095) msg->type = cpu_to_le16(RTRS_MSG_READ);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1096) msg->usr_len = cpu_to_le16(req->usr_len);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1097)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1098) if (count) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1099) ret = rtrs_map_sg_fr(req, count);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1100) if (ret < 0) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1101) rtrs_err_rl(s,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1102) "Read request failed, failed to map fast reg. data, err: %d\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1103) ret);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1104) ib_dma_unmap_sg(dev->ib_dev, req->sglist, req->sg_cnt,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1105) req->dir);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1106) return ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1107) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1108) rwr = (struct ib_reg_wr) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1109) .wr.opcode = IB_WR_REG_MR,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1110) .wr.wr_cqe = &fast_reg_cqe,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1111) .mr = req->mr,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1112) .key = req->mr->rkey,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1113) .access = (IB_ACCESS_LOCAL_WRITE |
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1114) IB_ACCESS_REMOTE_WRITE),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1115) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1116) wr = &rwr.wr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1117)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1118) msg->sg_cnt = cpu_to_le16(1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1119) msg->flags = cpu_to_le16(RTRS_MSG_NEED_INVAL_F);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1120)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1121) msg->desc[0].addr = cpu_to_le64(req->mr->iova);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1122) msg->desc[0].key = cpu_to_le32(req->mr->rkey);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1123) msg->desc[0].len = cpu_to_le32(req->mr->length);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1124)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1125) /* Further invalidation is required */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1126) req->need_inv = !!RTRS_MSG_NEED_INVAL_F;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1127)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1128) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1129) msg->sg_cnt = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1130) msg->flags = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1131) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1132) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1133) * rtrs message will be after the space reserved for disk data and
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1134) * user message
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1135) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1136) imm = req->permit->mem_off + req->data_len + req->usr_len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1137) imm = rtrs_to_io_req_imm(imm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1138) buf_id = req->permit->mem_id;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1139)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1140) req->sg_size = sizeof(*msg);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1141) req->sg_size += le16_to_cpu(msg->sg_cnt) * sizeof(struct rtrs_sg_desc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1142) req->sg_size += req->usr_len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1143)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1144) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1145) * Update stats now, after request is successfully sent it is not
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1146) * safe anymore to touch it.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1147) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1148) rtrs_clt_update_all_stats(req, READ);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1149)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1150) ret = rtrs_post_send_rdma(req->con, req, &sess->rbufs[buf_id],
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1151) req->data_len, imm, wr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1152) if (unlikely(ret)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1153) rtrs_err(s, "Read request failed: %d\n", ret);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1154) if (sess->clt->mp_policy == MP_POLICY_MIN_INFLIGHT)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1155) atomic_dec(&sess->stats->inflight);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1156) req->need_inv = false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1157) if (req->sg_cnt)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1158) ib_dma_unmap_sg(dev->ib_dev, req->sglist,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1159) req->sg_cnt, req->dir);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1160) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1161)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1162) return ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1163) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1164)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1165) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1166) * rtrs_clt_failover_req() Try to find an active path for a failed request
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1167) * @clt: clt context
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1168) * @fail_req: a failed io request.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1169) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1170) static int rtrs_clt_failover_req(struct rtrs_clt *clt,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1171) struct rtrs_clt_io_req *fail_req)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1172) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1173) struct rtrs_clt_sess *alive_sess;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1174) struct rtrs_clt_io_req *req;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1175) int err = -ECONNABORTED;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1176) struct path_it it;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1177)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1178) rcu_read_lock();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1179) for (path_it_init(&it, clt);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1180) (alive_sess = it.next_path(&it)) && it.i < it.clt->paths_num;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1181) it.i++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1182) if (unlikely(READ_ONCE(alive_sess->state) !=
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1183) RTRS_CLT_CONNECTED))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1184) continue;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1185) req = rtrs_clt_get_copy_req(alive_sess, fail_req);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1186) if (req->dir == DMA_TO_DEVICE)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1187) err = rtrs_clt_write_req(req);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1188) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1189) err = rtrs_clt_read_req(req);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1190) if (unlikely(err)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1191) req->in_use = false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1192) continue;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1193) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1194) /* Success path */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1195) rtrs_clt_inc_failover_cnt(alive_sess->stats);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1196) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1197) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1198) path_it_deinit(&it);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1199) rcu_read_unlock();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1200)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1201) return err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1202) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1203)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1204) static void fail_all_outstanding_reqs(struct rtrs_clt_sess *sess)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1205) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1206) struct rtrs_clt *clt = sess->clt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1207) struct rtrs_clt_io_req *req;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1208) int i, err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1209)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1210) if (!sess->reqs)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1211) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1212) for (i = 0; i < sess->queue_depth; ++i) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1213) req = &sess->reqs[i];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1214) if (!req->in_use)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1215) continue;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1216)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1217) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1218) * Safely (without notification) complete failed request.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1219) * After completion this request is still useble and can
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1220) * be failovered to another path.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1221) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1222) complete_rdma_req(req, -ECONNABORTED, false, true);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1223)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1224) err = rtrs_clt_failover_req(clt, req);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1225) if (unlikely(err))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1226) /* Failover failed, notify anyway */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1227) req->conf(req->priv, err);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1228) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1229) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1230)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1231) static void free_sess_reqs(struct rtrs_clt_sess *sess)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1232) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1233) struct rtrs_clt_io_req *req;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1234) int i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1235)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1236) if (!sess->reqs)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1237) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1238) for (i = 0; i < sess->queue_depth; ++i) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1239) req = &sess->reqs[i];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1240) if (req->mr)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1241) ib_dereg_mr(req->mr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1242) kfree(req->sge);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1243) rtrs_iu_free(req->iu, sess->s.dev->ib_dev, 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1244) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1245) kfree(sess->reqs);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1246) sess->reqs = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1247) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1248)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1249) static int alloc_sess_reqs(struct rtrs_clt_sess *sess)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1250) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1251) struct rtrs_clt_io_req *req;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1252) struct rtrs_clt *clt = sess->clt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1253) int i, err = -ENOMEM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1254)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1255) sess->reqs = kcalloc(sess->queue_depth, sizeof(*sess->reqs),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1256) GFP_KERNEL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1257) if (!sess->reqs)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1258) return -ENOMEM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1259)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1260) for (i = 0; i < sess->queue_depth; ++i) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1261) req = &sess->reqs[i];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1262) req->iu = rtrs_iu_alloc(1, sess->max_hdr_size, GFP_KERNEL,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1263) sess->s.dev->ib_dev,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1264) DMA_TO_DEVICE,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1265) rtrs_clt_rdma_done);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1266) if (!req->iu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1267) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1268)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1269) req->sge = kmalloc_array(clt->max_segments + 1,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1270) sizeof(*req->sge), GFP_KERNEL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1271) if (!req->sge)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1272) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1273)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1274) req->mr = ib_alloc_mr(sess->s.dev->ib_pd, IB_MR_TYPE_MEM_REG,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1275) sess->max_pages_per_mr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1276) if (IS_ERR(req->mr)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1277) err = PTR_ERR(req->mr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1278) req->mr = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1279) pr_err("Failed to alloc sess->max_pages_per_mr %d\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1280) sess->max_pages_per_mr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1281) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1282) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1283)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1284) init_completion(&req->inv_comp);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1285) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1286)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1287) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1288)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1289) out:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1290) free_sess_reqs(sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1291)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1292) return err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1293) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1294)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1295) static int alloc_permits(struct rtrs_clt *clt)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1296) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1297) unsigned int chunk_bits;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1298) int err, i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1299)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1300) clt->permits_map = kcalloc(BITS_TO_LONGS(clt->queue_depth),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1301) sizeof(long), GFP_KERNEL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1302) if (!clt->permits_map) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1303) err = -ENOMEM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1304) goto out_err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1305) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1306) clt->permits = kcalloc(clt->queue_depth, permit_size(clt), GFP_KERNEL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1307) if (!clt->permits) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1308) err = -ENOMEM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1309) goto err_map;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1310) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1311) chunk_bits = ilog2(clt->queue_depth - 1) + 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1312) for (i = 0; i < clt->queue_depth; i++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1313) struct rtrs_permit *permit;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1314)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1315) permit = get_permit(clt, i);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1316) permit->mem_id = i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1317) permit->mem_off = i << (MAX_IMM_PAYL_BITS - chunk_bits);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1318) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1319)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1320) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1321)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1322) err_map:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1323) kfree(clt->permits_map);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1324) clt->permits_map = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1325) out_err:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1326) return err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1327) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1328)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1329) static void free_permits(struct rtrs_clt *clt)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1330) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1331) if (clt->permits_map) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1332) size_t sz = clt->queue_depth;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1333)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1334) wait_event(clt->permits_wait,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1335) find_first_bit(clt->permits_map, sz) >= sz);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1336) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1337) kfree(clt->permits_map);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1338) clt->permits_map = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1339) kfree(clt->permits);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1340) clt->permits = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1341) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1342)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1343) static void query_fast_reg_mode(struct rtrs_clt_sess *sess)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1344) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1345) struct ib_device *ib_dev;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1346) u64 max_pages_per_mr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1347) int mr_page_shift;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1348)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1349) ib_dev = sess->s.dev->ib_dev;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1350)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1351) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1352) * Use the smallest page size supported by the HCA, down to a
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1353) * minimum of 4096 bytes. We're unlikely to build large sglists
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1354) * out of smaller entries.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1355) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1356) mr_page_shift = max(12, ffs(ib_dev->attrs.page_size_cap) - 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1357) max_pages_per_mr = ib_dev->attrs.max_mr_size;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1358) do_div(max_pages_per_mr, (1ull << mr_page_shift));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1359) sess->max_pages_per_mr =
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1360) min3(sess->max_pages_per_mr, (u32)max_pages_per_mr,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1361) ib_dev->attrs.max_fast_reg_page_list_len);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1362) sess->max_send_sge = ib_dev->attrs.max_send_sge;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1363) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1364)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1365) static bool rtrs_clt_change_state_get_old(struct rtrs_clt_sess *sess,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1366) enum rtrs_clt_state new_state,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1367) enum rtrs_clt_state *old_state)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1368) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1369) bool changed;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1370)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1371) spin_lock_irq(&sess->state_wq.lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1372) *old_state = sess->state;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1373) changed = __rtrs_clt_change_state(sess, new_state);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1374) spin_unlock_irq(&sess->state_wq.lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1375)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1376) return changed;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1377) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1378)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1379) static bool rtrs_clt_change_state(struct rtrs_clt_sess *sess,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1380) enum rtrs_clt_state new_state)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1381) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1382) enum rtrs_clt_state old_state;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1383)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1384) return rtrs_clt_change_state_get_old(sess, new_state, &old_state);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1385) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1386)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1387) static void rtrs_clt_hb_err_handler(struct rtrs_con *c)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1388) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1389) struct rtrs_clt_con *con = container_of(c, typeof(*con), c);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1390)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1391) rtrs_rdma_error_recovery(con);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1392) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1393)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1394) static void rtrs_clt_init_hb(struct rtrs_clt_sess *sess)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1395) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1396) rtrs_init_hb(&sess->s, &io_comp_cqe,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1397) RTRS_HB_INTERVAL_MS,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1398) RTRS_HB_MISSED_MAX,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1399) rtrs_clt_hb_err_handler,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1400) rtrs_wq);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1401) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1402)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1403) static void rtrs_clt_start_hb(struct rtrs_clt_sess *sess)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1404) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1405) rtrs_start_hb(&sess->s);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1406) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1407)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1408) static void rtrs_clt_stop_hb(struct rtrs_clt_sess *sess)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1409) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1410) rtrs_stop_hb(&sess->s);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1411) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1412)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1413) static void rtrs_clt_reconnect_work(struct work_struct *work);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1414) static void rtrs_clt_close_work(struct work_struct *work);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1415)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1416) static struct rtrs_clt_sess *alloc_sess(struct rtrs_clt *clt,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1417) const struct rtrs_addr *path,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1418) size_t con_num, u16 max_segments,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1419) size_t max_segment_size)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1420) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1421) struct rtrs_clt_sess *sess;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1422) int err = -ENOMEM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1423) int cpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1424)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1425) sess = kzalloc(sizeof(*sess), GFP_KERNEL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1426) if (!sess)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1427) goto err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1428)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1429) /* Extra connection for user messages */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1430) con_num += 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1431)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1432) sess->s.con = kcalloc(con_num, sizeof(*sess->s.con), GFP_KERNEL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1433) if (!sess->s.con)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1434) goto err_free_sess;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1435)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1436) sess->stats = kzalloc(sizeof(*sess->stats), GFP_KERNEL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1437) if (!sess->stats)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1438) goto err_free_con;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1439)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1440) mutex_init(&sess->init_mutex);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1441) uuid_gen(&sess->s.uuid);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1442) memcpy(&sess->s.dst_addr, path->dst,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1443) rdma_addr_size((struct sockaddr *)path->dst));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1444)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1445) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1446) * rdma_resolve_addr() passes src_addr to cma_bind_addr, which
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1447) * checks the sa_family to be non-zero. If user passed src_addr=NULL
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1448) * the sess->src_addr will contain only zeros, which is then fine.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1449) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1450) if (path->src)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1451) memcpy(&sess->s.src_addr, path->src,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1452) rdma_addr_size((struct sockaddr *)path->src));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1453) strlcpy(sess->s.sessname, clt->sessname, sizeof(sess->s.sessname));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1454) sess->s.con_num = con_num;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1455) sess->clt = clt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1456) sess->max_pages_per_mr = max_segments * max_segment_size >> 12;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1457) init_waitqueue_head(&sess->state_wq);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1458) sess->state = RTRS_CLT_CONNECTING;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1459) atomic_set(&sess->connected_cnt, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1460) INIT_WORK(&sess->close_work, rtrs_clt_close_work);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1461) INIT_DELAYED_WORK(&sess->reconnect_dwork, rtrs_clt_reconnect_work);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1462) rtrs_clt_init_hb(sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1463)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1464) sess->mp_skip_entry = alloc_percpu(typeof(*sess->mp_skip_entry));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1465) if (!sess->mp_skip_entry)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1466) goto err_free_stats;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1467)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1468) for_each_possible_cpu(cpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1469) INIT_LIST_HEAD(per_cpu_ptr(sess->mp_skip_entry, cpu));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1470)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1471) err = rtrs_clt_init_stats(sess->stats);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1472) if (err)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1473) goto err_free_percpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1474)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1475) return sess;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1476)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1477) err_free_percpu:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1478) free_percpu(sess->mp_skip_entry);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1479) err_free_stats:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1480) kfree(sess->stats);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1481) err_free_con:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1482) kfree(sess->s.con);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1483) err_free_sess:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1484) kfree(sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1485) err:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1486) return ERR_PTR(err);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1487) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1488)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1489) void free_sess(struct rtrs_clt_sess *sess)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1490) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1491) free_percpu(sess->mp_skip_entry);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1492) mutex_destroy(&sess->init_mutex);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1493) kfree(sess->s.con);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1494) kfree(sess->rbufs);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1495) kfree(sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1496) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1497)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1498) static int create_con(struct rtrs_clt_sess *sess, unsigned int cid)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1499) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1500) struct rtrs_clt_con *con;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1501)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1502) con = kzalloc(sizeof(*con), GFP_KERNEL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1503) if (!con)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1504) return -ENOMEM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1505)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1506) /* Map first two connections to the first CPU */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1507) con->cpu = (cid ? cid - 1 : 0) % nr_cpu_ids;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1508) con->c.cid = cid;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1509) con->c.sess = &sess->s;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1510) atomic_set(&con->io_cnt, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1511)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1512) sess->s.con[cid] = &con->c;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1513)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1514) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1515) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1516)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1517) static void destroy_con(struct rtrs_clt_con *con)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1518) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1519) struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1520)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1521) sess->s.con[con->c.cid] = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1522) kfree(con);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1523) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1524)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1525) static int create_con_cq_qp(struct rtrs_clt_con *con)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1526) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1527) struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1528) u32 max_send_wr, max_recv_wr, cq_size;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1529) int err, cq_vector;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1530) struct rtrs_msg_rkey_rsp *rsp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1531)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1532) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1533) * This function can fail, but still destroy_con_cq_qp() should
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1534) * be called, this is because create_con_cq_qp() is called on cm
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1535) * event path, thus caller/waiter never knows: have we failed before
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1536) * create_con_cq_qp() or after. To solve this dilemma without
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1537) * creating any additional flags just allow destroy_con_cq_qp() be
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1538) * called many times.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1539) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1540)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1541) if (con->c.cid == 0) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1542) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1543) * One completion for each receive and two for each send
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1544) * (send request + registration)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1545) * + 2 for drain and heartbeat
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1546) * in case qp gets into error state
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1547) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1548) max_send_wr = SERVICE_CON_QUEUE_DEPTH * 2 + 2;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1549) max_recv_wr = SERVICE_CON_QUEUE_DEPTH * 2 + 2;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1550) /* We must be the first here */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1551) if (WARN_ON(sess->s.dev))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1552) return -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1553)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1554) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1555) * The whole session uses device from user connection.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1556) * Be careful not to close user connection before ib dev
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1557) * is gracefully put.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1558) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1559) sess->s.dev = rtrs_ib_dev_find_or_add(con->c.cm_id->device,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1560) &dev_pd);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1561) if (!sess->s.dev) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1562) rtrs_wrn(sess->clt,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1563) "rtrs_ib_dev_find_get_or_add(): no memory\n");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1564) return -ENOMEM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1565) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1566) sess->s.dev_ref = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1567) query_fast_reg_mode(sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1568) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1569) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1570) * Here we assume that session members are correctly set.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1571) * This is always true if user connection (cid == 0) is
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1572) * established first.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1573) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1574) if (WARN_ON(!sess->s.dev))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1575) return -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1576) if (WARN_ON(!sess->queue_depth))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1577) return -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1578)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1579) /* Shared between connections */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1580) sess->s.dev_ref++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1581) max_send_wr =
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1582) min_t(int, sess->s.dev->ib_dev->attrs.max_qp_wr,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1583) /* QD * (REQ + RSP + FR REGS or INVS) + drain */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1584) sess->queue_depth * 3 + 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1585) max_recv_wr =
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1586) min_t(int, sess->s.dev->ib_dev->attrs.max_qp_wr,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1587) sess->queue_depth * 3 + 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1588) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1589) /* alloc iu to recv new rkey reply when server reports flags set */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1590) if (sess->flags & RTRS_MSG_NEW_RKEY_F || con->c.cid == 0) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1591) con->rsp_ius = rtrs_iu_alloc(max_recv_wr, sizeof(*rsp),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1592) GFP_KERNEL, sess->s.dev->ib_dev,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1593) DMA_FROM_DEVICE,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1594) rtrs_clt_rdma_done);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1595) if (!con->rsp_ius)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1596) return -ENOMEM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1597) con->queue_size = max_recv_wr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1598) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1599) cq_size = max_send_wr + max_recv_wr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1600) cq_vector = con->cpu % sess->s.dev->ib_dev->num_comp_vectors;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1601) err = rtrs_cq_qp_create(&sess->s, &con->c, sess->max_send_sge,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1602) cq_vector, cq_size, max_send_wr,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1603) max_recv_wr, IB_POLL_SOFTIRQ);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1604) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1605) * In case of error we do not bother to clean previous allocations,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1606) * since destroy_con_cq_qp() must be called.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1607) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1608) return err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1609) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1610)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1611) static void destroy_con_cq_qp(struct rtrs_clt_con *con)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1612) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1613) struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1614)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1615) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1616) * Be careful here: destroy_con_cq_qp() can be called even
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1617) * create_con_cq_qp() failed, see comments there.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1618) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1619)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1620) rtrs_cq_qp_destroy(&con->c);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1621) if (con->rsp_ius) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1622) rtrs_iu_free(con->rsp_ius, sess->s.dev->ib_dev, con->queue_size);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1623) con->rsp_ius = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1624) con->queue_size = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1625) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1626) if (sess->s.dev_ref && !--sess->s.dev_ref) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1627) rtrs_ib_dev_put(sess->s.dev);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1628) sess->s.dev = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1629) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1630) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1631)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1632) static void stop_cm(struct rtrs_clt_con *con)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1633) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1634) rdma_disconnect(con->c.cm_id);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1635) if (con->c.qp)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1636) ib_drain_qp(con->c.qp);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1637) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1638)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1639) static void destroy_cm(struct rtrs_clt_con *con)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1640) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1641) rdma_destroy_id(con->c.cm_id);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1642) con->c.cm_id = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1643) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1644)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1645) static int rtrs_rdma_addr_resolved(struct rtrs_clt_con *con)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1646) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1647) struct rtrs_sess *s = con->c.sess;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1648) int err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1649)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1650) err = create_con_cq_qp(con);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1651) if (err) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1652) rtrs_err(s, "create_con_cq_qp(), err: %d\n", err);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1653) return err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1654) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1655) err = rdma_resolve_route(con->c.cm_id, RTRS_CONNECT_TIMEOUT_MS);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1656) if (err)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1657) rtrs_err(s, "Resolving route failed, err: %d\n", err);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1658)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1659) return err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1660) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1661)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1662) static int rtrs_rdma_route_resolved(struct rtrs_clt_con *con)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1663) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1664) struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1665) struct rtrs_clt *clt = sess->clt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1666) struct rtrs_msg_conn_req msg;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1667) struct rdma_conn_param param;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1668)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1669) int err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1670)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1671) param = (struct rdma_conn_param) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1672) .retry_count = 7,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1673) .rnr_retry_count = 7,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1674) .private_data = &msg,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1675) .private_data_len = sizeof(msg),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1676) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1677)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1678) msg = (struct rtrs_msg_conn_req) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1679) .magic = cpu_to_le16(RTRS_MAGIC),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1680) .version = cpu_to_le16(RTRS_PROTO_VER),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1681) .cid = cpu_to_le16(con->c.cid),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1682) .cid_num = cpu_to_le16(sess->s.con_num),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1683) .recon_cnt = cpu_to_le16(sess->s.recon_cnt),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1684) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1685) msg.first_conn = sess->for_new_clt ? FIRST_CONN : 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1686) uuid_copy(&msg.sess_uuid, &sess->s.uuid);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1687) uuid_copy(&msg.paths_uuid, &clt->paths_uuid);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1688)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1689) err = rdma_connect_locked(con->c.cm_id, ¶m);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1690) if (err)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1691) rtrs_err(clt, "rdma_connect_locked(): %d\n", err);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1692)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1693) return err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1694) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1695)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1696) static int rtrs_rdma_conn_established(struct rtrs_clt_con *con,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1697) struct rdma_cm_event *ev)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1698) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1699) struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1700) struct rtrs_clt *clt = sess->clt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1701) const struct rtrs_msg_conn_rsp *msg;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1702) u16 version, queue_depth;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1703) int errno;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1704) u8 len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1705)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1706) msg = ev->param.conn.private_data;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1707) len = ev->param.conn.private_data_len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1708) if (len < sizeof(*msg)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1709) rtrs_err(clt, "Invalid RTRS connection response\n");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1710) return -ECONNRESET;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1711) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1712) if (le16_to_cpu(msg->magic) != RTRS_MAGIC) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1713) rtrs_err(clt, "Invalid RTRS magic\n");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1714) return -ECONNRESET;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1715) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1716) version = le16_to_cpu(msg->version);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1717) if (version >> 8 != RTRS_PROTO_VER_MAJOR) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1718) rtrs_err(clt, "Unsupported major RTRS version: %d, expected %d\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1719) version >> 8, RTRS_PROTO_VER_MAJOR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1720) return -ECONNRESET;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1721) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1722) errno = le16_to_cpu(msg->errno);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1723) if (errno) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1724) rtrs_err(clt, "Invalid RTRS message: errno %d\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1725) errno);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1726) return -ECONNRESET;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1727) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1728) if (con->c.cid == 0) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1729) queue_depth = le16_to_cpu(msg->queue_depth);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1730)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1731) if (queue_depth > MAX_SESS_QUEUE_DEPTH) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1732) rtrs_err(clt, "Invalid RTRS message: queue=%d\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1733) queue_depth);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1734) return -ECONNRESET;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1735) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1736) if (sess->queue_depth > 0 && queue_depth != sess->queue_depth) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1737) rtrs_err(clt, "Error: queue depth changed\n");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1738)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1739) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1740) * Stop any more reconnection attempts
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1741) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1742) sess->reconnect_attempts = -1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1743) rtrs_err(clt,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1744) "Disabling auto-reconnect. Trigger a manual reconnect after issue is resolved\n");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1745) return -ECONNRESET;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1746) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1747)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1748) if (!sess->rbufs) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1749) kfree(sess->rbufs);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1750) sess->rbufs = kcalloc(queue_depth, sizeof(*sess->rbufs),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1751) GFP_KERNEL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1752) if (!sess->rbufs)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1753) return -ENOMEM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1754) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1755) sess->queue_depth = queue_depth;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1756) sess->max_hdr_size = le32_to_cpu(msg->max_hdr_size);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1757) sess->max_io_size = le32_to_cpu(msg->max_io_size);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1758) sess->flags = le32_to_cpu(msg->flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1759) sess->chunk_size = sess->max_io_size + sess->max_hdr_size;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1760)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1761) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1762) * Global IO size is always a minimum.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1763) * If while a reconnection server sends us a value a bit
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1764) * higher - client does not care and uses cached minimum.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1765) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1766) * Since we can have several sessions (paths) restablishing
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1767) * connections in parallel, use lock.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1768) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1769) mutex_lock(&clt->paths_mutex);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1770) clt->queue_depth = sess->queue_depth;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1771) clt->max_io_size = min_not_zero(sess->max_io_size,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1772) clt->max_io_size);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1773) mutex_unlock(&clt->paths_mutex);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1774)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1775) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1776) * Cache the hca_port and hca_name for sysfs
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1777) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1778) sess->hca_port = con->c.cm_id->port_num;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1779) scnprintf(sess->hca_name, sizeof(sess->hca_name),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1780) sess->s.dev->ib_dev->name);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1781) sess->s.src_addr = con->c.cm_id->route.addr.src_addr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1782) /* set for_new_clt, to allow future reconnect on any path */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1783) sess->for_new_clt = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1784) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1785)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1786) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1787) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1788)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1789) static inline void flag_success_on_conn(struct rtrs_clt_con *con)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1790) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1791) struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1792)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1793) atomic_inc(&sess->connected_cnt);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1794) con->cm_err = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1795) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1796)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1797) static int rtrs_rdma_conn_rejected(struct rtrs_clt_con *con,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1798) struct rdma_cm_event *ev)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1799) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1800) struct rtrs_sess *s = con->c.sess;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1801) const struct rtrs_msg_conn_rsp *msg;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1802) const char *rej_msg;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1803) int status, errno;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1804) u8 data_len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1805)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1806) status = ev->status;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1807) rej_msg = rdma_reject_msg(con->c.cm_id, status);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1808) msg = rdma_consumer_reject_data(con->c.cm_id, ev, &data_len);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1809)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1810) if (msg && data_len >= sizeof(*msg)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1811) errno = (int16_t)le16_to_cpu(msg->errno);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1812) if (errno == -EBUSY)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1813) rtrs_err(s,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1814) "Previous session is still exists on the server, please reconnect later\n");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1815) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1816) rtrs_err(s,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1817) "Connect rejected: status %d (%s), rtrs errno %d\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1818) status, rej_msg, errno);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1819) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1820) rtrs_err(s,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1821) "Connect rejected but with malformed message: status %d (%s)\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1822) status, rej_msg);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1823) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1824)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1825) return -ECONNRESET;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1826) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1827)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1828) static void rtrs_clt_close_conns(struct rtrs_clt_sess *sess, bool wait)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1829) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1830) if (rtrs_clt_change_state(sess, RTRS_CLT_CLOSING))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1831) queue_work(rtrs_wq, &sess->close_work);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1832) if (wait)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1833) flush_work(&sess->close_work);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1834) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1835)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1836) static inline void flag_error_on_conn(struct rtrs_clt_con *con, int cm_err)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1837) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1838) if (con->cm_err == 1) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1839) struct rtrs_clt_sess *sess;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1840)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1841) sess = to_clt_sess(con->c.sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1842) if (atomic_dec_and_test(&sess->connected_cnt))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1843)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1844) wake_up(&sess->state_wq);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1845) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1846) con->cm_err = cm_err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1847) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1848)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1849) static int rtrs_clt_rdma_cm_handler(struct rdma_cm_id *cm_id,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1850) struct rdma_cm_event *ev)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1851) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1852) struct rtrs_clt_con *con = cm_id->context;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1853) struct rtrs_sess *s = con->c.sess;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1854) struct rtrs_clt_sess *sess = to_clt_sess(s);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1855) int cm_err = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1856)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1857) switch (ev->event) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1858) case RDMA_CM_EVENT_ADDR_RESOLVED:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1859) cm_err = rtrs_rdma_addr_resolved(con);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1860) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1861) case RDMA_CM_EVENT_ROUTE_RESOLVED:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1862) cm_err = rtrs_rdma_route_resolved(con);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1863) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1864) case RDMA_CM_EVENT_ESTABLISHED:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1865) cm_err = rtrs_rdma_conn_established(con, ev);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1866) if (likely(!cm_err)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1867) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1868) * Report success and wake up. Here we abuse state_wq,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1869) * i.e. wake up without state change, but we set cm_err.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1870) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1871) flag_success_on_conn(con);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1872) wake_up(&sess->state_wq);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1873) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1874) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1875) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1876) case RDMA_CM_EVENT_REJECTED:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1877) cm_err = rtrs_rdma_conn_rejected(con, ev);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1878) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1879) case RDMA_CM_EVENT_CONNECT_ERROR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1880) case RDMA_CM_EVENT_UNREACHABLE:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1881) rtrs_wrn(s, "CM error event %d\n", ev->event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1882) cm_err = -ECONNRESET;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1883) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1884) case RDMA_CM_EVENT_ADDR_ERROR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1885) case RDMA_CM_EVENT_ROUTE_ERROR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1886) cm_err = -EHOSTUNREACH;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1887) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1888) case RDMA_CM_EVENT_DISCONNECTED:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1889) case RDMA_CM_EVENT_ADDR_CHANGE:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1890) case RDMA_CM_EVENT_TIMEWAIT_EXIT:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1891) cm_err = -ECONNRESET;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1892) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1893) case RDMA_CM_EVENT_DEVICE_REMOVAL:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1894) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1895) * Device removal is a special case. Queue close and return 0.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1896) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1897) rtrs_clt_close_conns(sess, false);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1898) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1899) default:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1900) rtrs_err(s, "Unexpected RDMA CM event (%d)\n", ev->event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1901) cm_err = -ECONNRESET;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1902) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1903) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1904)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1905) if (cm_err) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1906) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1907) * cm error makes sense only on connection establishing,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1908) * in other cases we rely on normal procedure of reconnecting.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1909) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1910) flag_error_on_conn(con, cm_err);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1911) rtrs_rdma_error_recovery(con);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1912) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1913)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1914) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1915) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1916)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1917) static int create_cm(struct rtrs_clt_con *con)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1918) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1919) struct rtrs_sess *s = con->c.sess;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1920) struct rtrs_clt_sess *sess = to_clt_sess(s);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1921) struct rdma_cm_id *cm_id;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1922) int err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1923)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1924) cm_id = rdma_create_id(&init_net, rtrs_clt_rdma_cm_handler, con,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1925) sess->s.dst_addr.ss_family == AF_IB ?
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1926) RDMA_PS_IB : RDMA_PS_TCP, IB_QPT_RC);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1927) if (IS_ERR(cm_id)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1928) err = PTR_ERR(cm_id);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1929) rtrs_err(s, "Failed to create CM ID, err: %d\n", err);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1930)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1931) return err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1932) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1933) con->c.cm_id = cm_id;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1934) con->cm_err = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1935) /* allow the port to be reused */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1936) err = rdma_set_reuseaddr(cm_id, 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1937) if (err != 0) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1938) rtrs_err(s, "Set address reuse failed, err: %d\n", err);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1939) goto destroy_cm;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1940) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1941) err = rdma_resolve_addr(cm_id, (struct sockaddr *)&sess->s.src_addr,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1942) (struct sockaddr *)&sess->s.dst_addr,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1943) RTRS_CONNECT_TIMEOUT_MS);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1944) if (err) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1945) rtrs_err(s, "Failed to resolve address, err: %d\n", err);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1946) goto destroy_cm;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1947) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1948) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1949) * Combine connection status and session events. This is needed
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1950) * for waiting two possible cases: cm_err has something meaningful
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1951) * or session state was really changed to error by device removal.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1952) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1953) err = wait_event_interruptible_timeout(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1954) sess->state_wq,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1955) con->cm_err || sess->state != RTRS_CLT_CONNECTING,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1956) msecs_to_jiffies(RTRS_CONNECT_TIMEOUT_MS));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1957) if (err == 0 || err == -ERESTARTSYS) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1958) if (err == 0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1959) err = -ETIMEDOUT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1960) /* Timedout or interrupted */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1961) goto errr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1962) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1963) if (con->cm_err < 0) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1964) err = con->cm_err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1965) goto errr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1966) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1967) if (READ_ONCE(sess->state) != RTRS_CLT_CONNECTING) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1968) /* Device removal */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1969) err = -ECONNABORTED;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1970) goto errr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1971) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1972)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1973) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1974)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1975) errr:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1976) stop_cm(con);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1977) /* Is safe to call destroy if cq_qp is not inited */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1978) destroy_con_cq_qp(con);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1979) destroy_cm:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1980) destroy_cm(con);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1981)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1982) return err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1983) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1984)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1985) static void rtrs_clt_sess_up(struct rtrs_clt_sess *sess)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1986) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1987) struct rtrs_clt *clt = sess->clt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1988) int up;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1989)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1990) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1991) * We can fire RECONNECTED event only when all paths were
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1992) * connected on rtrs_clt_open(), then each was disconnected
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1993) * and the first one connected again. That's why this nasty
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1994) * game with counter value.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1995) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1996)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1997) mutex_lock(&clt->paths_ev_mutex);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1998) up = ++clt->paths_up;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1999) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2000) * Here it is safe to access paths num directly since up counter
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2001) * is greater than MAX_PATHS_NUM only while rtrs_clt_open() is
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2002) * in progress, thus paths removals are impossible.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2003) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2004) if (up > MAX_PATHS_NUM && up == MAX_PATHS_NUM + clt->paths_num)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2005) clt->paths_up = clt->paths_num;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2006) else if (up == 1)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2007) clt->link_ev(clt->priv, RTRS_CLT_LINK_EV_RECONNECTED);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2008) mutex_unlock(&clt->paths_ev_mutex);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2009)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2010) /* Mark session as established */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2011) sess->established = true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2012) sess->reconnect_attempts = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2013) sess->stats->reconnects.successful_cnt++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2014) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2015)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2016) static void rtrs_clt_sess_down(struct rtrs_clt_sess *sess)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2017) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2018) struct rtrs_clt *clt = sess->clt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2019)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2020) if (!sess->established)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2021) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2022)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2023) sess->established = false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2024) mutex_lock(&clt->paths_ev_mutex);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2025) WARN_ON(!clt->paths_up);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2026) if (--clt->paths_up == 0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2027) clt->link_ev(clt->priv, RTRS_CLT_LINK_EV_DISCONNECTED);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2028) mutex_unlock(&clt->paths_ev_mutex);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2029) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2030)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2031) static void rtrs_clt_stop_and_destroy_conns(struct rtrs_clt_sess *sess)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2032) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2033) struct rtrs_clt_con *con;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2034) unsigned int cid;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2035)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2036) WARN_ON(READ_ONCE(sess->state) == RTRS_CLT_CONNECTED);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2037)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2038) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2039) * Possible race with rtrs_clt_open(), when DEVICE_REMOVAL comes
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2040) * exactly in between. Start destroying after it finishes.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2041) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2042) mutex_lock(&sess->init_mutex);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2043) mutex_unlock(&sess->init_mutex);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2044)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2045) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2046) * All IO paths must observe !CONNECTED state before we
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2047) * free everything.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2048) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2049) synchronize_rcu();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2050)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2051) rtrs_clt_stop_hb(sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2052)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2053) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2054) * The order it utterly crucial: firstly disconnect and complete all
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2055) * rdma requests with error (thus set in_use=false for requests),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2056) * then fail outstanding requests checking in_use for each, and
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2057) * eventually notify upper layer about session disconnection.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2058) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2059)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2060) for (cid = 0; cid < sess->s.con_num; cid++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2061) if (!sess->s.con[cid])
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2062) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2063) con = to_clt_con(sess->s.con[cid]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2064) stop_cm(con);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2065) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2066) fail_all_outstanding_reqs(sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2067) free_sess_reqs(sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2068) rtrs_clt_sess_down(sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2069)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2070) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2071) * Wait for graceful shutdown, namely when peer side invokes
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2072) * rdma_disconnect(). 'connected_cnt' is decremented only on
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2073) * CM events, thus if other side had crashed and hb has detected
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2074) * something is wrong, here we will stuck for exactly timeout ms,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2075) * since CM does not fire anything. That is fine, we are not in
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2076) * hurry.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2077) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2078) wait_event_timeout(sess->state_wq, !atomic_read(&sess->connected_cnt),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2079) msecs_to_jiffies(RTRS_CONNECT_TIMEOUT_MS));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2080)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2081) for (cid = 0; cid < sess->s.con_num; cid++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2082) if (!sess->s.con[cid])
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2083) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2084) con = to_clt_con(sess->s.con[cid]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2085) destroy_con_cq_qp(con);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2086) destroy_cm(con);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2087) destroy_con(con);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2088) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2089) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2090)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2091) static inline bool xchg_sessions(struct rtrs_clt_sess __rcu **rcu_ppcpu_path,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2092) struct rtrs_clt_sess *sess,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2093) struct rtrs_clt_sess *next)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2094) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2095) struct rtrs_clt_sess **ppcpu_path;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2096)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2097) /* Call cmpxchg() without sparse warnings */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2098) ppcpu_path = (typeof(ppcpu_path))rcu_ppcpu_path;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2099) return sess == cmpxchg(ppcpu_path, sess, next);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2100) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2101)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2102) static void rtrs_clt_remove_path_from_arr(struct rtrs_clt_sess *sess)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2103) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2104) struct rtrs_clt *clt = sess->clt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2105) struct rtrs_clt_sess *next;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2106) bool wait_for_grace = false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2107) int cpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2108)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2109) mutex_lock(&clt->paths_mutex);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2110) list_del_rcu(&sess->s.entry);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2111)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2112) /* Make sure everybody observes path removal. */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2113) synchronize_rcu();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2114)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2115) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2116) * At this point nobody sees @sess in the list, but still we have
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2117) * dangling pointer @pcpu_path which _can_ point to @sess. Since
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2118) * nobody can observe @sess in the list, we guarantee that IO path
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2119) * will not assign @sess to @pcpu_path, i.e. @pcpu_path can be equal
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2120) * to @sess, but can never again become @sess.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2121) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2122)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2123) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2124) * Decrement paths number only after grace period, because
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2125) * caller of do_each_path() must firstly observe list without
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2126) * path and only then decremented paths number.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2127) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2128) * Otherwise there can be the following situation:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2129) * o Two paths exist and IO is coming.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2130) * o One path is removed:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2131) * CPU#0 CPU#1
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2132) * do_each_path(): rtrs_clt_remove_path_from_arr():
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2133) * path = get_next_path()
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2134) * ^^^ list_del_rcu(path)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2135) * [!CONNECTED path] clt->paths_num--
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2136) * ^^^^^^^^^
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2137) * load clt->paths_num from 2 to 1
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2138) * ^^^^^^^^^
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2139) * sees 1
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2140) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2141) * path is observed as !CONNECTED, but do_each_path() loop
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2142) * ends, because expression i < clt->paths_num is false.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2143) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2144) clt->paths_num--;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2145)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2146) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2147) * Get @next connection from current @sess which is going to be
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2148) * removed. If @sess is the last element, then @next is NULL.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2149) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2150) rcu_read_lock();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2151) next = list_next_or_null_rr_rcu(&clt->paths_list, &sess->s.entry,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2152) typeof(*next), s.entry);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2153) rcu_read_unlock();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2154)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2155) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2156) * @pcpu paths can still point to the path which is going to be
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2157) * removed, so change the pointer manually.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2158) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2159) for_each_possible_cpu(cpu) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2160) struct rtrs_clt_sess __rcu **ppcpu_path;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2161)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2162) ppcpu_path = per_cpu_ptr(clt->pcpu_path, cpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2163) if (rcu_dereference_protected(*ppcpu_path,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2164) lockdep_is_held(&clt->paths_mutex)) != sess)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2165) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2166) * synchronize_rcu() was called just after deleting
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2167) * entry from the list, thus IO code path cannot
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2168) * change pointer back to the pointer which is going
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2169) * to be removed, we are safe here.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2170) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2171) continue;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2172)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2173) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2174) * We race with IO code path, which also changes pointer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2175) * thus we have to be careful not to overwrite it.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2176) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2177) if (xchg_sessions(ppcpu_path, sess, next))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2178) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2179) * @ppcpu_path was successfully replaced with @next,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2180) * that means that someone could also pick up the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2181) * @sess and dereferencing it right now, so wait for
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2182) * a grace period is required.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2183) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2184) wait_for_grace = true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2185) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2186) if (wait_for_grace)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2187) synchronize_rcu();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2188)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2189) mutex_unlock(&clt->paths_mutex);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2190) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2191)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2192) static void rtrs_clt_add_path_to_arr(struct rtrs_clt_sess *sess,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2193) struct rtrs_addr *addr)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2194) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2195) struct rtrs_clt *clt = sess->clt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2196)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2197) mutex_lock(&clt->paths_mutex);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2198) clt->paths_num++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2199)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2200) list_add_tail_rcu(&sess->s.entry, &clt->paths_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2201) mutex_unlock(&clt->paths_mutex);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2202) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2203)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2204) static void rtrs_clt_close_work(struct work_struct *work)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2205) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2206) struct rtrs_clt_sess *sess;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2207)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2208) sess = container_of(work, struct rtrs_clt_sess, close_work);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2209)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2210) cancel_delayed_work_sync(&sess->reconnect_dwork);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2211) rtrs_clt_stop_and_destroy_conns(sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2212) rtrs_clt_change_state(sess, RTRS_CLT_CLOSED);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2213) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2214)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2215) static int init_conns(struct rtrs_clt_sess *sess)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2216) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2217) unsigned int cid;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2218) int err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2219)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2220) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2221) * On every new session connections increase reconnect counter
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2222) * to avoid clashes with previous sessions not yet closed
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2223) * sessions on a server side.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2224) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2225) sess->s.recon_cnt++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2226)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2227) /* Establish all RDMA connections */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2228) for (cid = 0; cid < sess->s.con_num; cid++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2229) err = create_con(sess, cid);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2230) if (err)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2231) goto destroy;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2232)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2233) err = create_cm(to_clt_con(sess->s.con[cid]));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2234) if (err) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2235) destroy_con(to_clt_con(sess->s.con[cid]));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2236) goto destroy;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2237) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2238) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2239) err = alloc_sess_reqs(sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2240) if (err)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2241) goto destroy;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2242)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2243) rtrs_clt_start_hb(sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2244)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2245) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2246)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2247) destroy:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2248) while (cid--) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2249) struct rtrs_clt_con *con = to_clt_con(sess->s.con[cid]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2250)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2251) stop_cm(con);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2252) destroy_con_cq_qp(con);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2253) destroy_cm(con);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2254) destroy_con(con);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2255) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2256) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2257) * If we've never taken async path and got an error, say,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2258) * doing rdma_resolve_addr(), switch to CONNECTION_ERR state
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2259) * manually to keep reconnecting.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2260) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2261) rtrs_clt_change_state(sess, RTRS_CLT_CONNECTING_ERR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2262)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2263) return err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2264) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2265)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2266) static void rtrs_clt_info_req_done(struct ib_cq *cq, struct ib_wc *wc)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2267) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2268) struct rtrs_clt_con *con = cq->cq_context;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2269) struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2270) struct rtrs_iu *iu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2271)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2272) iu = container_of(wc->wr_cqe, struct rtrs_iu, cqe);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2273) rtrs_iu_free(iu, sess->s.dev->ib_dev, 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2274)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2275) if (unlikely(wc->status != IB_WC_SUCCESS)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2276) rtrs_err(sess->clt, "Sess info request send failed: %s\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2277) ib_wc_status_msg(wc->status));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2278) rtrs_clt_change_state(sess, RTRS_CLT_CONNECTING_ERR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2279) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2280) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2281)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2282) rtrs_clt_update_wc_stats(con);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2283) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2284)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2285) static int process_info_rsp(struct rtrs_clt_sess *sess,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2286) const struct rtrs_msg_info_rsp *msg)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2287) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2288) unsigned int sg_cnt, total_len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2289) int i, sgi;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2290)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2291) sg_cnt = le16_to_cpu(msg->sg_cnt);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2292) if (unlikely(!sg_cnt))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2293) return -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2294) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2295) * Check if IB immediate data size is enough to hold the mem_id and
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2296) * the offset inside the memory chunk.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2297) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2298) if (unlikely((ilog2(sg_cnt - 1) + 1) +
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2299) (ilog2(sess->chunk_size - 1) + 1) >
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2300) MAX_IMM_PAYL_BITS)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2301) rtrs_err(sess->clt,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2302) "RDMA immediate size (%db) not enough to encode %d buffers of size %dB\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2303) MAX_IMM_PAYL_BITS, sg_cnt, sess->chunk_size);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2304) return -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2305) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2306) if (unlikely(!sg_cnt || (sess->queue_depth % sg_cnt))) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2307) rtrs_err(sess->clt, "Incorrect sg_cnt %d, is not multiple\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2308) sg_cnt);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2309) return -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2310) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2311) total_len = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2312) for (sgi = 0, i = 0; sgi < sg_cnt && i < sess->queue_depth; sgi++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2313) const struct rtrs_sg_desc *desc = &msg->desc[sgi];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2314) u32 len, rkey;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2315) u64 addr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2316)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2317) addr = le64_to_cpu(desc->addr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2318) rkey = le32_to_cpu(desc->key);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2319) len = le32_to_cpu(desc->len);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2320)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2321) total_len += len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2322)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2323) if (unlikely(!len || (len % sess->chunk_size))) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2324) rtrs_err(sess->clt, "Incorrect [%d].len %d\n", sgi,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2325) len);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2326) return -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2327) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2328) for ( ; len && i < sess->queue_depth; i++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2329) sess->rbufs[i].addr = addr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2330) sess->rbufs[i].rkey = rkey;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2331)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2332) len -= sess->chunk_size;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2333) addr += sess->chunk_size;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2334) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2335) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2336) /* Sanity check */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2337) if (unlikely(sgi != sg_cnt || i != sess->queue_depth)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2338) rtrs_err(sess->clt, "Incorrect sg vector, not fully mapped\n");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2339) return -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2340) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2341) if (unlikely(total_len != sess->chunk_size * sess->queue_depth)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2342) rtrs_err(sess->clt, "Incorrect total_len %d\n", total_len);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2343) return -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2344) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2345)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2346) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2347) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2348)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2349) static void rtrs_clt_info_rsp_done(struct ib_cq *cq, struct ib_wc *wc)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2350) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2351) struct rtrs_clt_con *con = cq->cq_context;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2352) struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2353) struct rtrs_msg_info_rsp *msg;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2354) enum rtrs_clt_state state;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2355) struct rtrs_iu *iu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2356) size_t rx_sz;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2357) int err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2358)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2359) state = RTRS_CLT_CONNECTING_ERR;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2360)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2361) WARN_ON(con->c.cid);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2362) iu = container_of(wc->wr_cqe, struct rtrs_iu, cqe);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2363) if (unlikely(wc->status != IB_WC_SUCCESS)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2364) rtrs_err(sess->clt, "Sess info response recv failed: %s\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2365) ib_wc_status_msg(wc->status));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2366) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2367) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2368) WARN_ON(wc->opcode != IB_WC_RECV);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2369)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2370) if (unlikely(wc->byte_len < sizeof(*msg))) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2371) rtrs_err(sess->clt, "Sess info response is malformed: size %d\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2372) wc->byte_len);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2373) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2374) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2375) ib_dma_sync_single_for_cpu(sess->s.dev->ib_dev, iu->dma_addr,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2376) iu->size, DMA_FROM_DEVICE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2377) msg = iu->buf;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2378) if (unlikely(le16_to_cpu(msg->type) != RTRS_MSG_INFO_RSP)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2379) rtrs_err(sess->clt, "Sess info response is malformed: type %d\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2380) le16_to_cpu(msg->type));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2381) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2382) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2383) rx_sz = sizeof(*msg);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2384) rx_sz += sizeof(msg->desc[0]) * le16_to_cpu(msg->sg_cnt);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2385) if (unlikely(wc->byte_len < rx_sz)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2386) rtrs_err(sess->clt, "Sess info response is malformed: size %d\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2387) wc->byte_len);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2388) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2389) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2390) err = process_info_rsp(sess, msg);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2391) if (unlikely(err))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2392) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2393)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2394) err = post_recv_sess(sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2395) if (unlikely(err))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2396) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2397)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2398) state = RTRS_CLT_CONNECTED;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2399)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2400) out:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2401) rtrs_clt_update_wc_stats(con);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2402) rtrs_iu_free(iu, sess->s.dev->ib_dev, 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2403) rtrs_clt_change_state(sess, state);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2404) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2405)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2406) static int rtrs_send_sess_info(struct rtrs_clt_sess *sess)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2407) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2408) struct rtrs_clt_con *usr_con = to_clt_con(sess->s.con[0]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2409) struct rtrs_msg_info_req *msg;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2410) struct rtrs_iu *tx_iu, *rx_iu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2411) size_t rx_sz;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2412) int err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2413)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2414) rx_sz = sizeof(struct rtrs_msg_info_rsp);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2415) rx_sz += sizeof(u64) * MAX_SESS_QUEUE_DEPTH;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2416)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2417) tx_iu = rtrs_iu_alloc(1, sizeof(struct rtrs_msg_info_req), GFP_KERNEL,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2418) sess->s.dev->ib_dev, DMA_TO_DEVICE,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2419) rtrs_clt_info_req_done);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2420) rx_iu = rtrs_iu_alloc(1, rx_sz, GFP_KERNEL, sess->s.dev->ib_dev,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2421) DMA_FROM_DEVICE, rtrs_clt_info_rsp_done);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2422) if (unlikely(!tx_iu || !rx_iu)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2423) err = -ENOMEM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2424) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2425) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2426) /* Prepare for getting info response */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2427) err = rtrs_iu_post_recv(&usr_con->c, rx_iu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2428) if (unlikely(err)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2429) rtrs_err(sess->clt, "rtrs_iu_post_recv(), err: %d\n", err);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2430) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2431) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2432) rx_iu = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2433)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2434) msg = tx_iu->buf;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2435) msg->type = cpu_to_le16(RTRS_MSG_INFO_REQ);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2436) memcpy(msg->sessname, sess->s.sessname, sizeof(msg->sessname));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2437)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2438) ib_dma_sync_single_for_device(sess->s.dev->ib_dev, tx_iu->dma_addr,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2439) tx_iu->size, DMA_TO_DEVICE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2440)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2441) /* Send info request */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2442) err = rtrs_iu_post_send(&usr_con->c, tx_iu, sizeof(*msg), NULL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2443) if (unlikely(err)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2444) rtrs_err(sess->clt, "rtrs_iu_post_send(), err: %d\n", err);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2445) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2446) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2447) tx_iu = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2448)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2449) /* Wait for state change */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2450) wait_event_interruptible_timeout(sess->state_wq,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2451) sess->state != RTRS_CLT_CONNECTING,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2452) msecs_to_jiffies(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2453) RTRS_CONNECT_TIMEOUT_MS));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2454) if (unlikely(READ_ONCE(sess->state) != RTRS_CLT_CONNECTED)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2455) if (READ_ONCE(sess->state) == RTRS_CLT_CONNECTING_ERR)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2456) err = -ECONNRESET;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2457) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2458) err = -ETIMEDOUT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2459) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2460) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2461)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2462) out:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2463) if (tx_iu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2464) rtrs_iu_free(tx_iu, sess->s.dev->ib_dev, 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2465) if (rx_iu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2466) rtrs_iu_free(rx_iu, sess->s.dev->ib_dev, 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2467) if (unlikely(err))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2468) /* If we've never taken async path because of malloc problems */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2469) rtrs_clt_change_state(sess, RTRS_CLT_CONNECTING_ERR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2470)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2471) return err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2472) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2473)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2474) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2475) * init_sess() - establishes all session connections and does handshake
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2476) * @sess: client session.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2477) * In case of error full close or reconnect procedure should be taken,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2478) * because reconnect or close async works can be started.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2479) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2480) static int init_sess(struct rtrs_clt_sess *sess)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2481) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2482) int err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2483)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2484) mutex_lock(&sess->init_mutex);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2485) err = init_conns(sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2486) if (err) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2487) rtrs_err(sess->clt, "init_conns(), err: %d\n", err);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2488) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2489) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2490) err = rtrs_send_sess_info(sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2491) if (err) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2492) rtrs_err(sess->clt, "rtrs_send_sess_info(), err: %d\n", err);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2493) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2494) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2495) rtrs_clt_sess_up(sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2496) out:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2497) mutex_unlock(&sess->init_mutex);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2498)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2499) return err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2500) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2501)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2502) static void rtrs_clt_reconnect_work(struct work_struct *work)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2503) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2504) struct rtrs_clt_sess *sess;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2505) struct rtrs_clt *clt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2506) unsigned int delay_ms;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2507) int err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2508)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2509) sess = container_of(to_delayed_work(work), struct rtrs_clt_sess,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2510) reconnect_dwork);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2511) clt = sess->clt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2512)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2513) if (READ_ONCE(sess->state) != RTRS_CLT_RECONNECTING)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2514) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2515)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2516) if (sess->reconnect_attempts >= clt->max_reconnect_attempts) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2517) /* Close a session completely if max attempts is reached */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2518) rtrs_clt_close_conns(sess, false);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2519) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2520) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2521) sess->reconnect_attempts++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2522)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2523) /* Stop everything */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2524) rtrs_clt_stop_and_destroy_conns(sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2525) msleep(RTRS_RECONNECT_BACKOFF);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2526) if (rtrs_clt_change_state(sess, RTRS_CLT_CONNECTING)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2527) err = init_sess(sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2528) if (err)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2529) goto reconnect_again;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2530) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2531)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2532) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2533)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2534) reconnect_again:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2535) if (rtrs_clt_change_state(sess, RTRS_CLT_RECONNECTING)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2536) sess->stats->reconnects.fail_cnt++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2537) delay_ms = clt->reconnect_delay_sec * 1000;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2538) queue_delayed_work(rtrs_wq, &sess->reconnect_dwork,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2539) msecs_to_jiffies(delay_ms +
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2540) prandom_u32() %
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2541) RTRS_RECONNECT_SEED));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2542) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2543) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2544)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2545) static void rtrs_clt_dev_release(struct device *dev)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2546) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2547) struct rtrs_clt *clt = container_of(dev, struct rtrs_clt, dev);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2548)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2549) mutex_destroy(&clt->paths_ev_mutex);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2550) mutex_destroy(&clt->paths_mutex);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2551) kfree(clt);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2552) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2553)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2554) static struct rtrs_clt *alloc_clt(const char *sessname, size_t paths_num,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2555) u16 port, size_t pdu_sz, void *priv,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2556) void (*link_ev)(void *priv,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2557) enum rtrs_clt_link_ev ev),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2558) unsigned int max_segments,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2559) size_t max_segment_size,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2560) unsigned int reconnect_delay_sec,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2561) unsigned int max_reconnect_attempts)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2562) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2563) struct rtrs_clt *clt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2564) int err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2565)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2566) if (!paths_num || paths_num > MAX_PATHS_NUM)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2567) return ERR_PTR(-EINVAL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2568)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2569) if (strlen(sessname) >= sizeof(clt->sessname))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2570) return ERR_PTR(-EINVAL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2571)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2572) clt = kzalloc(sizeof(*clt), GFP_KERNEL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2573) if (!clt)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2574) return ERR_PTR(-ENOMEM);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2575)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2576) clt->pcpu_path = alloc_percpu(typeof(*clt->pcpu_path));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2577) if (!clt->pcpu_path) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2578) kfree(clt);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2579) return ERR_PTR(-ENOMEM);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2580) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2581)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2582) clt->dev.class = rtrs_clt_dev_class;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2583) clt->dev.release = rtrs_clt_dev_release;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2584) uuid_gen(&clt->paths_uuid);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2585) INIT_LIST_HEAD_RCU(&clt->paths_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2586) clt->paths_num = paths_num;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2587) clt->paths_up = MAX_PATHS_NUM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2588) clt->port = port;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2589) clt->pdu_sz = pdu_sz;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2590) clt->max_segments = max_segments;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2591) clt->max_segment_size = max_segment_size;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2592) clt->reconnect_delay_sec = reconnect_delay_sec;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2593) clt->max_reconnect_attempts = max_reconnect_attempts;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2594) clt->priv = priv;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2595) clt->link_ev = link_ev;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2596) clt->mp_policy = MP_POLICY_MIN_INFLIGHT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2597) strlcpy(clt->sessname, sessname, sizeof(clt->sessname));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2598) init_waitqueue_head(&clt->permits_wait);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2599) mutex_init(&clt->paths_ev_mutex);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2600) mutex_init(&clt->paths_mutex);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2601) device_initialize(&clt->dev);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2602)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2603) err = dev_set_name(&clt->dev, "%s", sessname);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2604) if (err)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2605) goto err_put;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2606)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2607) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2608) * Suppress user space notification until
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2609) * sysfs files are created
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2610) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2611) dev_set_uevent_suppress(&clt->dev, true);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2612) err = device_add(&clt->dev);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2613) if (err)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2614) goto err_put;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2615)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2616) clt->kobj_paths = kobject_create_and_add("paths", &clt->dev.kobj);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2617) if (!clt->kobj_paths) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2618) err = -ENOMEM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2619) goto err_del;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2620) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2621) err = rtrs_clt_create_sysfs_root_files(clt);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2622) if (err) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2623) kobject_del(clt->kobj_paths);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2624) kobject_put(clt->kobj_paths);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2625) goto err_del;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2626) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2627) dev_set_uevent_suppress(&clt->dev, false);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2628) kobject_uevent(&clt->dev.kobj, KOBJ_ADD);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2629)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2630) return clt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2631) err_del:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2632) device_del(&clt->dev);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2633) err_put:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2634) free_percpu(clt->pcpu_path);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2635) put_device(&clt->dev);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2636) return ERR_PTR(err);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2637) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2638)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2639) static void free_clt(struct rtrs_clt *clt)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2640) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2641) free_percpu(clt->pcpu_path);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2642)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2643) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2644) * release callback will free clt and destroy mutexes in last put
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2645) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2646) device_unregister(&clt->dev);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2647) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2648)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2649) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2650) * rtrs_clt_open() - Open a session to an RTRS server
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2651) * @ops: holds the link event callback and the private pointer.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2652) * @sessname: name of the session
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2653) * @paths: Paths to be established defined by their src and dst addresses
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2654) * @paths_num: Number of elements in the @paths array
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2655) * @port: port to be used by the RTRS session
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2656) * @pdu_sz: Size of extra payload which can be accessed after permit allocation.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2657) * @reconnect_delay_sec: time between reconnect tries
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2658) * @max_segments: Max. number of segments per IO request
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2659) * @max_segment_size: Max. size of one segment
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2660) * @max_reconnect_attempts: Number of times to reconnect on error before giving
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2661) * up, 0 for * disabled, -1 for forever
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2662) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2663) * Starts session establishment with the rtrs_server. The function can block
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2664) * up to ~2000ms before it returns.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2665) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2666) * Return a valid pointer on success otherwise PTR_ERR.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2667) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2668) struct rtrs_clt *rtrs_clt_open(struct rtrs_clt_ops *ops,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2669) const char *sessname,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2670) const struct rtrs_addr *paths,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2671) size_t paths_num, u16 port,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2672) size_t pdu_sz, u8 reconnect_delay_sec,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2673) u16 max_segments,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2674) size_t max_segment_size,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2675) s16 max_reconnect_attempts)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2676) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2677) struct rtrs_clt_sess *sess, *tmp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2678) struct rtrs_clt *clt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2679) int err, i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2680)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2681) clt = alloc_clt(sessname, paths_num, port, pdu_sz, ops->priv,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2682) ops->link_ev,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2683) max_segments, max_segment_size, reconnect_delay_sec,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2684) max_reconnect_attempts);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2685) if (IS_ERR(clt)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2686) err = PTR_ERR(clt);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2687) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2688) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2689) for (i = 0; i < paths_num; i++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2690) struct rtrs_clt_sess *sess;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2691)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2692) sess = alloc_sess(clt, &paths[i], nr_cpu_ids,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2693) max_segments, max_segment_size);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2694) if (IS_ERR(sess)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2695) err = PTR_ERR(sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2696) goto close_all_sess;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2697) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2698) if (!i)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2699) sess->for_new_clt = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2700) list_add_tail_rcu(&sess->s.entry, &clt->paths_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2701)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2702) err = init_sess(sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2703) if (err) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2704) list_del_rcu(&sess->s.entry);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2705) rtrs_clt_close_conns(sess, true);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2706) free_percpu(sess->stats->pcpu_stats);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2707) kfree(sess->stats);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2708) free_sess(sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2709) goto close_all_sess;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2710) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2711)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2712) err = rtrs_clt_create_sess_files(sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2713) if (err) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2714) list_del_rcu(&sess->s.entry);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2715) rtrs_clt_close_conns(sess, true);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2716) free_percpu(sess->stats->pcpu_stats);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2717) kfree(sess->stats);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2718) free_sess(sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2719) goto close_all_sess;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2720) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2721) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2722) err = alloc_permits(clt);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2723) if (err)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2724) goto close_all_sess;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2725)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2726) return clt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2727)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2728) close_all_sess:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2729) list_for_each_entry_safe(sess, tmp, &clt->paths_list, s.entry) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2730) rtrs_clt_destroy_sess_files(sess, NULL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2731) rtrs_clt_close_conns(sess, true);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2732) kobject_put(&sess->kobj);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2733) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2734) rtrs_clt_destroy_sysfs_root_files(clt);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2735) rtrs_clt_destroy_sysfs_root_folders(clt);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2736) free_clt(clt);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2737)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2738) out:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2739) return ERR_PTR(err);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2740) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2741) EXPORT_SYMBOL(rtrs_clt_open);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2742)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2743) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2744) * rtrs_clt_close() - Close a session
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2745) * @clt: Session handle. Session is freed upon return.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2746) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2747) void rtrs_clt_close(struct rtrs_clt *clt)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2748) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2749) struct rtrs_clt_sess *sess, *tmp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2750)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2751) /* Firstly forbid sysfs access */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2752) rtrs_clt_destroy_sysfs_root_files(clt);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2753) rtrs_clt_destroy_sysfs_root_folders(clt);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2754)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2755) /* Now it is safe to iterate over all paths without locks */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2756) list_for_each_entry_safe(sess, tmp, &clt->paths_list, s.entry) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2757) rtrs_clt_close_conns(sess, true);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2758) rtrs_clt_destroy_sess_files(sess, NULL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2759) kobject_put(&sess->kobj);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2760) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2761) free_permits(clt);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2762) free_clt(clt);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2763) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2764) EXPORT_SYMBOL(rtrs_clt_close);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2765)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2766) int rtrs_clt_reconnect_from_sysfs(struct rtrs_clt_sess *sess)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2767) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2768) enum rtrs_clt_state old_state;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2769) int err = -EBUSY;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2770) bool changed;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2771)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2772) changed = rtrs_clt_change_state_get_old(sess, RTRS_CLT_RECONNECTING,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2773) &old_state);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2774) if (changed) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2775) sess->reconnect_attempts = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2776) queue_delayed_work(rtrs_wq, &sess->reconnect_dwork, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2777) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2778) if (changed || old_state == RTRS_CLT_RECONNECTING) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2779) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2780) * flush_delayed_work() queues pending work for immediate
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2781) * execution, so do the flush if we have queued something
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2782) * right now or work is pending.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2783) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2784) flush_delayed_work(&sess->reconnect_dwork);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2785) err = (READ_ONCE(sess->state) ==
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2786) RTRS_CLT_CONNECTED ? 0 : -ENOTCONN);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2787) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2788)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2789) return err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2790) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2791)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2792) int rtrs_clt_disconnect_from_sysfs(struct rtrs_clt_sess *sess)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2793) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2794) rtrs_clt_close_conns(sess, true);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2795)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2796) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2797) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2798)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2799) int rtrs_clt_remove_path_from_sysfs(struct rtrs_clt_sess *sess,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2800) const struct attribute *sysfs_self)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2801) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2802) enum rtrs_clt_state old_state;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2803) bool changed;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2804)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2805) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2806) * Continue stopping path till state was changed to DEAD or
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2807) * state was observed as DEAD:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2808) * 1. State was changed to DEAD - we were fast and nobody
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2809) * invoked rtrs_clt_reconnect(), which can again start
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2810) * reconnecting.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2811) * 2. State was observed as DEAD - we have someone in parallel
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2812) * removing the path.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2813) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2814) do {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2815) rtrs_clt_close_conns(sess, true);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2816) changed = rtrs_clt_change_state_get_old(sess,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2817) RTRS_CLT_DEAD,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2818) &old_state);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2819) } while (!changed && old_state != RTRS_CLT_DEAD);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2820)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2821) if (likely(changed)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2822) rtrs_clt_remove_path_from_arr(sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2823) rtrs_clt_destroy_sess_files(sess, sysfs_self);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2824) kobject_put(&sess->kobj);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2825) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2826)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2827) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2828) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2829)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2830) void rtrs_clt_set_max_reconnect_attempts(struct rtrs_clt *clt, int value)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2831) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2832) clt->max_reconnect_attempts = (unsigned int)value;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2833) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2834)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2835) int rtrs_clt_get_max_reconnect_attempts(const struct rtrs_clt *clt)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2836) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2837) return (int)clt->max_reconnect_attempts;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2838) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2839)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2840) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2841) * rtrs_clt_request() - Request data transfer to/from server via RDMA.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2842) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2843) * @dir: READ/WRITE
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2844) * @ops: callback function to be called as confirmation, and the pointer.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2845) * @clt: Session
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2846) * @permit: Preallocated permit
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2847) * @vec: Message that is sent to server together with the request.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2848) * Sum of len of all @vec elements limited to <= IO_MSG_SIZE.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2849) * Since the msg is copied internally it can be allocated on stack.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2850) * @nr: Number of elements in @vec.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2851) * @data_len: length of data sent to/from server
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2852) * @sg: Pages to be sent/received to/from server.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2853) * @sg_cnt: Number of elements in the @sg
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2854) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2855) * Return:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2856) * 0: Success
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2857) * <0: Error
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2858) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2859) * On dir=READ rtrs client will request a data transfer from Server to client.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2860) * The data that the server will respond with will be stored in @sg when
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2861) * the user receives an %RTRS_CLT_RDMA_EV_RDMA_REQUEST_WRITE_COMPL event.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2862) * On dir=WRITE rtrs client will rdma write data in sg to server side.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2863) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2864) int rtrs_clt_request(int dir, struct rtrs_clt_req_ops *ops,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2865) struct rtrs_clt *clt, struct rtrs_permit *permit,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2866) const struct kvec *vec, size_t nr, size_t data_len,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2867) struct scatterlist *sg, unsigned int sg_cnt)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2868) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2869) struct rtrs_clt_io_req *req;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2870) struct rtrs_clt_sess *sess;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2871)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2872) enum dma_data_direction dma_dir;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2873) int err = -ECONNABORTED, i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2874) size_t usr_len, hdr_len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2875) struct path_it it;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2876)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2877) /* Get kvec length */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2878) for (i = 0, usr_len = 0; i < nr; i++)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2879) usr_len += vec[i].iov_len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2880)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2881) if (dir == READ) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2882) hdr_len = sizeof(struct rtrs_msg_rdma_read) +
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2883) sg_cnt * sizeof(struct rtrs_sg_desc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2884) dma_dir = DMA_FROM_DEVICE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2885) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2886) hdr_len = sizeof(struct rtrs_msg_rdma_write);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2887) dma_dir = DMA_TO_DEVICE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2888) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2889)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2890) rcu_read_lock();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2891) for (path_it_init(&it, clt);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2892) (sess = it.next_path(&it)) && it.i < it.clt->paths_num; it.i++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2893) if (unlikely(READ_ONCE(sess->state) != RTRS_CLT_CONNECTED))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2894) continue;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2895)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2896) if (unlikely(usr_len + hdr_len > sess->max_hdr_size)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2897) rtrs_wrn_rl(sess->clt,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2898) "%s request failed, user message size is %zu and header length %zu, but max size is %u\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2899) dir == READ ? "Read" : "Write",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2900) usr_len, hdr_len, sess->max_hdr_size);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2901) err = -EMSGSIZE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2902) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2903) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2904) req = rtrs_clt_get_req(sess, ops->conf_fn, permit, ops->priv,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2905) vec, usr_len, sg, sg_cnt, data_len,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2906) dma_dir);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2907) if (dir == READ)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2908) err = rtrs_clt_read_req(req);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2909) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2910) err = rtrs_clt_write_req(req);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2911) if (unlikely(err)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2912) req->in_use = false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2913) continue;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2914) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2915) /* Success path */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2916) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2917) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2918) path_it_deinit(&it);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2919) rcu_read_unlock();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2920)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2921) return err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2922) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2923) EXPORT_SYMBOL(rtrs_clt_request);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2924)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2925) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2926) * rtrs_clt_query() - queries RTRS session attributes
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2927) *@clt: session pointer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2928) *@attr: query results for session attributes.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2929) * Returns:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2930) * 0 on success
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2931) * -ECOMM no connection to the server
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2932) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2933) int rtrs_clt_query(struct rtrs_clt *clt, struct rtrs_attrs *attr)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2934) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2935) if (!rtrs_clt_is_connected(clt))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2936) return -ECOMM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2937)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2938) attr->queue_depth = clt->queue_depth;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2939) attr->max_io_size = clt->max_io_size;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2940) attr->sess_kobj = &clt->dev.kobj;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2941) strlcpy(attr->sessname, clt->sessname, sizeof(attr->sessname));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2942)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2943) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2944) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2945) EXPORT_SYMBOL(rtrs_clt_query);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2946)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2947) int rtrs_clt_create_path_from_sysfs(struct rtrs_clt *clt,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2948) struct rtrs_addr *addr)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2949) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2950) struct rtrs_clt_sess *sess;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2951) int err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2952)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2953) sess = alloc_sess(clt, addr, nr_cpu_ids, clt->max_segments,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2954) clt->max_segment_size);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2955) if (IS_ERR(sess))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2956) return PTR_ERR(sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2957)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2958) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2959) * It is totally safe to add path in CONNECTING state: coming
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2960) * IO will never grab it. Also it is very important to add
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2961) * path before init, since init fires LINK_CONNECTED event.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2962) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2963) rtrs_clt_add_path_to_arr(sess, addr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2964)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2965) err = init_sess(sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2966) if (err)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2967) goto close_sess;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2968)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2969) err = rtrs_clt_create_sess_files(sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2970) if (err)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2971) goto close_sess;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2972)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2973) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2974)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2975) close_sess:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2976) rtrs_clt_remove_path_from_arr(sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2977) rtrs_clt_close_conns(sess, true);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2978) free_percpu(sess->stats->pcpu_stats);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2979) kfree(sess->stats);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2980) free_sess(sess);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2981)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2982) return err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2983) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2984)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2985) static int rtrs_clt_ib_dev_init(struct rtrs_ib_dev *dev)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2986) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2987) if (!(dev->ib_dev->attrs.device_cap_flags &
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2988) IB_DEVICE_MEM_MGT_EXTENSIONS)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2989) pr_err("Memory registrations not supported.\n");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2990) return -ENOTSUPP;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2991) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2992)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2993) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2994) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2995)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2996) static const struct rtrs_rdma_dev_pd_ops dev_pd_ops = {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2997) .init = rtrs_clt_ib_dev_init
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2998) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2999)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3000) static int __init rtrs_client_init(void)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3001) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3002) rtrs_rdma_dev_pd_init(0, &dev_pd);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3003)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3004) rtrs_clt_dev_class = class_create(THIS_MODULE, "rtrs-client");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3005) if (IS_ERR(rtrs_clt_dev_class)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3006) pr_err("Failed to create rtrs-client dev class\n");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3007) return PTR_ERR(rtrs_clt_dev_class);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3008) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3009) rtrs_wq = alloc_workqueue("rtrs_client_wq", 0, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3010) if (!rtrs_wq) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3011) class_destroy(rtrs_clt_dev_class);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3012) return -ENOMEM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3013) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3014)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3015) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3016) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3017)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3018) static void __exit rtrs_client_exit(void)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3019) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3020) destroy_workqueue(rtrs_wq);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3021) class_destroy(rtrs_clt_dev_class);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3022) rtrs_rdma_dev_pd_deinit(&dev_pd);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3023) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3024)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3025) module_init(rtrs_client_init);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3026) module_exit(rtrs_client_exit);
Orange Pi5 kernel
Deprecated Linux kernel 5.10.110 for OrangePi 5/5B/5+ boards
3 Commits
0 Branches
0 Tags