^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2) * Copyright (c) 2007, 2017 Oracle and/or its affiliates. All rights reserved.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4) * This software is available to you under a choice of one of two
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5) * licenses. You may choose to be licensed under the terms of the GNU
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 6) * General Public License (GPL) Version 2, available from the file
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 7) * COPYING in the main directory of this source tree, or the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 8) * OpenIB.org BSD license below:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 9) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 10) * Redistribution and use in source and binary forms, with or
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 11) * without modification, are permitted provided that the following
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 12) * conditions are met:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 13) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 14) * - Redistributions of source code must retain the above
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 15) * copyright notice, this list of conditions and the following
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 16) * disclaimer.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 17) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 18) * - Redistributions in binary form must reproduce the above
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 19) * copyright notice, this list of conditions and the following
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 20) * disclaimer in the documentation and/or other materials
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 21) * provided with the distribution.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 22) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 23) * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 24) * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 25) * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 26) * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 27) * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 28) * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 29) * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 30) * SOFTWARE.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 31) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 32) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 33) #include <linux/slab.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 34) #include <linux/types.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 35) #include <linux/rbtree.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 36) #include <linux/bitops.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 37) #include <linux/export.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 38)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 39) #include "rds.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 40)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 41) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 42) * This file implements the receive side of the unconventional congestion
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 43) * management in RDS.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 44) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 45) * Messages waiting in the receive queue on the receiving socket are accounted
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 46) * against the sockets SO_RCVBUF option value. Only the payload bytes in the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 47) * message are accounted for. If the number of bytes queued equals or exceeds
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 48) * rcvbuf then the socket is congested. All sends attempted to this socket's
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 49) * address should return block or return -EWOULDBLOCK.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 50) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 51) * Applications are expected to be reasonably tuned such that this situation
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 52) * very rarely occurs. An application encountering this "back-pressure" is
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 53) * considered a bug.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 54) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 55) * This is implemented by having each node maintain bitmaps which indicate
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 56) * which ports on bound addresses are congested. As the bitmap changes it is
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 57) * sent through all the connections which terminate in the local address of the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 58) * bitmap which changed.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 59) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 60) * The bitmaps are allocated as connections are brought up. This avoids
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 61) * allocation in the interrupt handling path which queues messages on sockets.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 62) * The dense bitmaps let transports send the entire bitmap on any bitmap change
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 63) * reasonably efficiently. This is much easier to implement than some
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 64) * finer-grained communication of per-port congestion. The sender does a very
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 65) * inexpensive bit test to test if the port it's about to send to is congested
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 66) * or not.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 67) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 68)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 69) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 70) * Interaction with poll is a tad tricky. We want all processes stuck in
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 71) * poll to wake up and check whether a congested destination became uncongested.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 72) * The really sad thing is we have no idea which destinations the application
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 73) * wants to send to - we don't even know which rds_connections are involved.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 74) * So until we implement a more flexible rds poll interface, we have to make
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 75) * do with this:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 76) * We maintain a global counter that is incremented each time a congestion map
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 77) * update is received. Each rds socket tracks this value, and if rds_poll
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 78) * finds that the saved generation number is smaller than the global generation
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 79) * number, it wakes up the process.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 80) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 81) static atomic_t rds_cong_generation = ATOMIC_INIT(0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 82)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 83) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 84) * Congestion monitoring
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 85) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 86) static LIST_HEAD(rds_cong_monitor);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 87) static DEFINE_RWLOCK(rds_cong_monitor_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 88)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 89) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 90) * Yes, a global lock. It's used so infrequently that it's worth keeping it
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 91) * global to simplify the locking. It's only used in the following
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 92) * circumstances:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 93) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 94) * - on connection buildup to associate a conn with its maps
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 95) * - on map changes to inform conns of a new map to send
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 96) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 97) * It's sadly ordered under the socket callback lock and the connection lock.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 98) * Receive paths can mark ports congested from interrupt context so the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 99) * lock masks interrupts.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 100) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 101) static DEFINE_SPINLOCK(rds_cong_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 102) static struct rb_root rds_cong_tree = RB_ROOT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 103)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 104) static struct rds_cong_map *rds_cong_tree_walk(const struct in6_addr *addr,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 105) struct rds_cong_map *insert)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 106) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 107) struct rb_node **p = &rds_cong_tree.rb_node;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 108) struct rb_node *parent = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 109) struct rds_cong_map *map;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 110)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 111) while (*p) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 112) int diff;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 113)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 114) parent = *p;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 115) map = rb_entry(parent, struct rds_cong_map, m_rb_node);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 116)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 117) diff = rds_addr_cmp(addr, &map->m_addr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 118) if (diff < 0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 119) p = &(*p)->rb_left;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 120) else if (diff > 0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 121) p = &(*p)->rb_right;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 122) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 123) return map;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 124) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 125)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 126) if (insert) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 127) rb_link_node(&insert->m_rb_node, parent, p);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 128) rb_insert_color(&insert->m_rb_node, &rds_cong_tree);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 129) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 130) return NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 131) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 132)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 133) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 134) * There is only ever one bitmap for any address. Connections try and allocate
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 135) * these bitmaps in the process getting pointers to them. The bitmaps are only
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 136) * ever freed as the module is removed after all connections have been freed.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 137) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 138) static struct rds_cong_map *rds_cong_from_addr(const struct in6_addr *addr)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 139) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 140) struct rds_cong_map *map;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 141) struct rds_cong_map *ret = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 142) unsigned long zp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 143) unsigned long i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 144) unsigned long flags;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 145)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 146) map = kzalloc(sizeof(struct rds_cong_map), GFP_KERNEL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 147) if (!map)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 148) return NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 149)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 150) map->m_addr = *addr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 151) init_waitqueue_head(&map->m_waitq);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 152) INIT_LIST_HEAD(&map->m_conn_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 153)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 154) for (i = 0; i < RDS_CONG_MAP_PAGES; i++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 155) zp = get_zeroed_page(GFP_KERNEL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 156) if (zp == 0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 157) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 158) map->m_page_addrs[i] = zp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 159) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 160)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 161) spin_lock_irqsave(&rds_cong_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 162) ret = rds_cong_tree_walk(addr, map);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 163) spin_unlock_irqrestore(&rds_cong_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 164)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 165) if (!ret) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 166) ret = map;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 167) map = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 168) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 169)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 170) out:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 171) if (map) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 172) for (i = 0; i < RDS_CONG_MAP_PAGES && map->m_page_addrs[i]; i++)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 173) free_page(map->m_page_addrs[i]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 174) kfree(map);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 175) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 176)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 177) rdsdebug("map %p for addr %pI6c\n", ret, addr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 178)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 179) return ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 180) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 181)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 182) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 183) * Put the conn on its local map's list. This is called when the conn is
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 184) * really added to the hash. It's nested under the rds_conn_lock, sadly.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 185) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 186) void rds_cong_add_conn(struct rds_connection *conn)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 187) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 188) unsigned long flags;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 189)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 190) rdsdebug("conn %p now on map %p\n", conn, conn->c_lcong);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 191) spin_lock_irqsave(&rds_cong_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 192) list_add_tail(&conn->c_map_item, &conn->c_lcong->m_conn_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 193) spin_unlock_irqrestore(&rds_cong_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 194) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 195)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 196) void rds_cong_remove_conn(struct rds_connection *conn)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 197) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 198) unsigned long flags;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 199)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 200) rdsdebug("removing conn %p from map %p\n", conn, conn->c_lcong);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 201) spin_lock_irqsave(&rds_cong_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 202) list_del_init(&conn->c_map_item);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 203) spin_unlock_irqrestore(&rds_cong_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 204) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 205)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 206) int rds_cong_get_maps(struct rds_connection *conn)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 207) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 208) conn->c_lcong = rds_cong_from_addr(&conn->c_laddr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 209) conn->c_fcong = rds_cong_from_addr(&conn->c_faddr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 210)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 211) if (!(conn->c_lcong && conn->c_fcong))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 212) return -ENOMEM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 213)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 214) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 215) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 216)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 217) void rds_cong_queue_updates(struct rds_cong_map *map)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 218) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 219) struct rds_connection *conn;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 220) unsigned long flags;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 221)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 222) spin_lock_irqsave(&rds_cong_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 223)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 224) list_for_each_entry(conn, &map->m_conn_list, c_map_item) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 225) struct rds_conn_path *cp = &conn->c_path[0];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 226)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 227) rcu_read_lock();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 228) if (!test_and_set_bit(0, &conn->c_map_queued) &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 229) !rds_destroy_pending(cp->cp_conn)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 230) rds_stats_inc(s_cong_update_queued);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 231) /* We cannot inline the call to rds_send_xmit() here
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 232) * for two reasons (both pertaining to a TCP transport):
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 233) * 1. When we get here from the receive path, we
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 234) * are already holding the sock_lock (held by
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 235) * tcp_v4_rcv()). So inlining calls to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 236) * tcp_setsockopt and/or tcp_sendmsg will deadlock
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 237) * when it tries to get the sock_lock())
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 238) * 2. Interrupts are masked so that we can mark the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 239) * port congested from both send and recv paths.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 240) * (See comment around declaration of rdc_cong_lock).
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 241) * An attempt to get the sock_lock() here will
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 242) * therefore trigger warnings.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 243) * Defer the xmit to rds_send_worker() instead.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 244) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 245) queue_delayed_work(rds_wq, &cp->cp_send_w, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 246) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 247) rcu_read_unlock();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 248) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 249)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 250) spin_unlock_irqrestore(&rds_cong_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 251) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 252)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 253) void rds_cong_map_updated(struct rds_cong_map *map, uint64_t portmask)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 254) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 255) rdsdebug("waking map %p for %pI4\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 256) map, &map->m_addr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 257) rds_stats_inc(s_cong_update_received);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 258) atomic_inc(&rds_cong_generation);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 259) if (waitqueue_active(&map->m_waitq))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 260) wake_up(&map->m_waitq);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 261) if (waitqueue_active(&rds_poll_waitq))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 262) wake_up_all(&rds_poll_waitq);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 263)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 264) if (portmask && !list_empty(&rds_cong_monitor)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 265) unsigned long flags;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 266) struct rds_sock *rs;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 267)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 268) read_lock_irqsave(&rds_cong_monitor_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 269) list_for_each_entry(rs, &rds_cong_monitor, rs_cong_list) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 270) spin_lock(&rs->rs_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 271) rs->rs_cong_notify |= (rs->rs_cong_mask & portmask);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 272) rs->rs_cong_mask &= ~portmask;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 273) spin_unlock(&rs->rs_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 274) if (rs->rs_cong_notify)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 275) rds_wake_sk_sleep(rs);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 276) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 277) read_unlock_irqrestore(&rds_cong_monitor_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 278) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 279) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 280) EXPORT_SYMBOL_GPL(rds_cong_map_updated);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 281)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 282) int rds_cong_updated_since(unsigned long *recent)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 283) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 284) unsigned long gen = atomic_read(&rds_cong_generation);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 285)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 286) if (likely(*recent == gen))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 287) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 288) *recent = gen;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 289) return 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 290) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 291)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 292) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 293) * We're called under the locking that protects the sockets receive buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 294) * consumption. This makes it a lot easier for the caller to only call us
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 295) * when it knows that an existing set bit needs to be cleared, and vice versa.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 296) * We can't block and we need to deal with concurrent sockets working against
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 297) * the same per-address map.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 298) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 299) void rds_cong_set_bit(struct rds_cong_map *map, __be16 port)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 300) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 301) unsigned long i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 302) unsigned long off;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 303)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 304) rdsdebug("setting congestion for %pI4:%u in map %p\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 305) &map->m_addr, ntohs(port), map);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 306)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 307) i = be16_to_cpu(port) / RDS_CONG_MAP_PAGE_BITS;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 308) off = be16_to_cpu(port) % RDS_CONG_MAP_PAGE_BITS;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 309)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 310) set_bit_le(off, (void *)map->m_page_addrs[i]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 311) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 312)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 313) void rds_cong_clear_bit(struct rds_cong_map *map, __be16 port)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 314) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 315) unsigned long i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 316) unsigned long off;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 317)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 318) rdsdebug("clearing congestion for %pI4:%u in map %p\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 319) &map->m_addr, ntohs(port), map);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 320)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 321) i = be16_to_cpu(port) / RDS_CONG_MAP_PAGE_BITS;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 322) off = be16_to_cpu(port) % RDS_CONG_MAP_PAGE_BITS;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 323)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 324) clear_bit_le(off, (void *)map->m_page_addrs[i]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 325) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 326)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 327) static int rds_cong_test_bit(struct rds_cong_map *map, __be16 port)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 328) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 329) unsigned long i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 330) unsigned long off;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 331)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 332) i = be16_to_cpu(port) / RDS_CONG_MAP_PAGE_BITS;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 333) off = be16_to_cpu(port) % RDS_CONG_MAP_PAGE_BITS;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 334)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 335) return test_bit_le(off, (void *)map->m_page_addrs[i]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 336) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 337)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 338) void rds_cong_add_socket(struct rds_sock *rs)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 339) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 340) unsigned long flags;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 341)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 342) write_lock_irqsave(&rds_cong_monitor_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 343) if (list_empty(&rs->rs_cong_list))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 344) list_add(&rs->rs_cong_list, &rds_cong_monitor);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 345) write_unlock_irqrestore(&rds_cong_monitor_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 346) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 347)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 348) void rds_cong_remove_socket(struct rds_sock *rs)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 349) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 350) unsigned long flags;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 351) struct rds_cong_map *map;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 352)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 353) write_lock_irqsave(&rds_cong_monitor_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 354) list_del_init(&rs->rs_cong_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 355) write_unlock_irqrestore(&rds_cong_monitor_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 356)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 357) /* update congestion map for now-closed port */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 358) spin_lock_irqsave(&rds_cong_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 359) map = rds_cong_tree_walk(&rs->rs_bound_addr, NULL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 360) spin_unlock_irqrestore(&rds_cong_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 361)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 362) if (map && rds_cong_test_bit(map, rs->rs_bound_port)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 363) rds_cong_clear_bit(map, rs->rs_bound_port);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 364) rds_cong_queue_updates(map);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 365) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 366) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 367)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 368) int rds_cong_wait(struct rds_cong_map *map, __be16 port, int nonblock,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 369) struct rds_sock *rs)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 370) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 371) if (!rds_cong_test_bit(map, port))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 372) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 373) if (nonblock) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 374) if (rs && rs->rs_cong_monitor) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 375) unsigned long flags;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 376)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 377) /* It would have been nice to have an atomic set_bit on
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 378) * a uint64_t. */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 379) spin_lock_irqsave(&rs->rs_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 380) rs->rs_cong_mask |= RDS_CONG_MONITOR_MASK(ntohs(port));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 381) spin_unlock_irqrestore(&rs->rs_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 382)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 383) /* Test again - a congestion update may have arrived in
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 384) * the meantime. */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 385) if (!rds_cong_test_bit(map, port))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 386) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 387) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 388) rds_stats_inc(s_cong_send_error);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 389) return -ENOBUFS;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 390) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 391)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 392) rds_stats_inc(s_cong_send_blocked);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 393) rdsdebug("waiting on map %p for port %u\n", map, be16_to_cpu(port));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 394)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 395) return wait_event_interruptible(map->m_waitq,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 396) !rds_cong_test_bit(map, port));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 397) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 398)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 399) void rds_cong_exit(void)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 400) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 401) struct rb_node *node;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 402) struct rds_cong_map *map;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 403) unsigned long i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 404)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 405) while ((node = rb_first(&rds_cong_tree))) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 406) map = rb_entry(node, struct rds_cong_map, m_rb_node);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 407) rdsdebug("freeing map %p\n", map);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 408) rb_erase(&map->m_rb_node, &rds_cong_tree);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 409) for (i = 0; i < RDS_CONG_MAP_PAGES && map->m_page_addrs[i]; i++)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 410) free_page(map->m_page_addrs[i]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 411) kfree(map);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 412) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 413) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 414)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 415) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 416) * Allocate a RDS message containing a congestion update.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 417) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 418) struct rds_message *rds_cong_update_alloc(struct rds_connection *conn)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 419) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 420) struct rds_cong_map *map = conn->c_lcong;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 421) struct rds_message *rm;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 422)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 423) rm = rds_message_map_pages(map->m_page_addrs, RDS_CONG_MAP_BYTES);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 424) if (!IS_ERR(rm))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 425) rm->m_inc.i_hdr.h_flags = RDS_FLAG_CONG_BITMAP;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 426)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 427) return rm;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 428) }
Orange Pi5 kernel
Deprecated Linux kernel 5.10.110 for OrangePi 5/5B/5+ boards
3 Commits
0 Branches
0 Tags