^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) * NET3: Garbage Collector For AF_UNIX sockets
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5) * Garbage Collector:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 6) * Copyright (C) Barak A. Pearlmutter.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 7) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 8) * Chopped about by Alan Cox 22/3/96 to make it fit the AF_UNIX socket problem.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 9) * If it doesn't work blame me, it worked when Barak sent it.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 10) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 11) * Assumptions:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 12) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 13) * - object w/ a bit
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 14) * - free list
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 15) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 16) * Current optimizations:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 17) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 18) * - explicit stack instead of recursion
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 19) * - tail recurse on first born instead of immediate push/pop
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 20) * - we gather the stuff that should not be killed into tree
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 21) * and stack is just a path from root to the current pointer.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 22) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 23) * Future optimizations:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 24) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 25) * - don't just push entire root set; process in place
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 26) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 27) * Fixes:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 28) * Alan Cox 07 Sept 1997 Vmalloc internal stack as needed.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 29) * Cope with changing max_files.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 30) * Al Viro 11 Oct 1998
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 31) * Graph may have cycles. That is, we can send the descriptor
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 32) * of foo to bar and vice versa. Current code chokes on that.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 33) * Fix: move SCM_RIGHTS ones into the separate list and then
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 34) * skb_free() them all instead of doing explicit fput's.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 35) * Another problem: since fput() may block somebody may
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 36) * create a new unix_socket when we are in the middle of sweep
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 37) * phase. Fix: revert the logic wrt MARKED. Mark everything
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 38) * upon the beginning and unmark non-junk ones.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 39) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 40) * [12 Oct 1998] AAARGH! New code purges all SCM_RIGHTS
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 41) * sent to connect()'ed but still not accept()'ed sockets.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 42) * Fixed. Old code had slightly different problem here:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 43) * extra fput() in situation when we passed the descriptor via
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 44) * such socket and closed it (descriptor). That would happen on
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 45) * each unix_gc() until the accept(). Since the struct file in
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 46) * question would go to the free list and might be reused...
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 47) * That might be the reason of random oopses on filp_close()
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 48) * in unrelated processes.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 49) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 50) * AV 28 Feb 1999
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 51) * Kill the explicit allocation of stack. Now we keep the tree
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 52) * with root in dummy + pointer (gc_current) to one of the nodes.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 53) * Stack is represented as path from gc_current to dummy. Unmark
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 54) * now means "add to tree". Push == "make it a son of gc_current".
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 55) * Pop == "move gc_current to parent". We keep only pointers to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 56) * parents (->gc_tree).
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 57) * AV 1 Mar 1999
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 58) * Damn. Added missing check for ->dead in listen queues scanning.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 59) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 60) * Miklos Szeredi 25 Jun 2007
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 61) * Reimplement with a cycle collecting algorithm. This should
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 62) * solve several problems with the previous code, like being racy
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 63) * wrt receive and holding up unrelated socket operations.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 64) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 65)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 66) #include <linux/kernel.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 67) #include <linux/string.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 68) #include <linux/socket.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 69) #include <linux/un.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 70) #include <linux/net.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 71) #include <linux/fs.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 72) #include <linux/skbuff.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 73) #include <linux/netdevice.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 74) #include <linux/file.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 75) #include <linux/proc_fs.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 76) #include <linux/mutex.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 77) #include <linux/wait.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 78)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 79) #include <net/sock.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 80) #include <net/af_unix.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 81) #include <net/scm.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 82) #include <net/tcp_states.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 83)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 84) #include "scm.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 85)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 86) /* Internal data structures and random procedures: */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 87)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 88) static LIST_HEAD(gc_candidates);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 89) static DECLARE_WAIT_QUEUE_HEAD(unix_gc_wait);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 90)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 91) static void scan_inflight(struct sock *x, void (*func)(struct unix_sock *),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 92) struct sk_buff_head *hitlist)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 93) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 94) struct sk_buff *skb;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 95) struct sk_buff *next;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 96)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 97) spin_lock(&x->sk_receive_queue.lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 98) skb_queue_walk_safe(&x->sk_receive_queue, skb, next) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 99) /* Do we have file descriptors ? */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 100) if (UNIXCB(skb).fp) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 101) bool hit = false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 102) /* Process the descriptors of this socket */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 103) int nfd = UNIXCB(skb).fp->count;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 104) struct file **fp = UNIXCB(skb).fp->fp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 105)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 106) while (nfd--) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 107) /* Get the socket the fd matches if it indeed does so */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 108) struct sock *sk = unix_get_socket(*fp++);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 109)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 110) if (sk) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 111) struct unix_sock *u = unix_sk(sk);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 112)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 113) /* Ignore non-candidates, they could
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 114) * have been added to the queues after
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 115) * starting the garbage collection
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 116) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 117) if (test_bit(UNIX_GC_CANDIDATE, &u->gc_flags)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 118) hit = true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 119)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 120) func(u);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 121) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 122) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 123) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 124) if (hit && hitlist != NULL) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 125) __skb_unlink(skb, &x->sk_receive_queue);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 126) __skb_queue_tail(hitlist, skb);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 127) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 128) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 129) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 130) spin_unlock(&x->sk_receive_queue.lock);
^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) static void scan_children(struct sock *x, void (*func)(struct unix_sock *),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 134) struct sk_buff_head *hitlist)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 135) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 136) if (x->sk_state != TCP_LISTEN) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 137) scan_inflight(x, func, hitlist);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 138) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 139) struct sk_buff *skb;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 140) struct sk_buff *next;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 141) struct unix_sock *u;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 142) LIST_HEAD(embryos);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 143)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 144) /* For a listening socket collect the queued embryos
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 145) * and perform a scan on them as well.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 146) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 147) spin_lock(&x->sk_receive_queue.lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 148) skb_queue_walk_safe(&x->sk_receive_queue, skb, next) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 149) u = unix_sk(skb->sk);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 150)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 151) /* An embryo cannot be in-flight, so it's safe
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 152) * to use the list link.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 153) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 154) BUG_ON(!list_empty(&u->link));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 155) list_add_tail(&u->link, &embryos);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 156) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 157) spin_unlock(&x->sk_receive_queue.lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 158)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 159) while (!list_empty(&embryos)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 160) u = list_entry(embryos.next, struct unix_sock, link);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 161) scan_inflight(&u->sk, func, hitlist);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 162) list_del_init(&u->link);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 163) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 164) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 165) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 166)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 167) static void dec_inflight(struct unix_sock *usk)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 168) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 169) atomic_long_dec(&usk->inflight);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 170) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 171)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 172) static void inc_inflight(struct unix_sock *usk)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 173) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 174) atomic_long_inc(&usk->inflight);
^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) static void inc_inflight_move_tail(struct unix_sock *u)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 178) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 179) atomic_long_inc(&u->inflight);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 180) /* If this still might be part of a cycle, move it to the end
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 181) * of the list, so that it's checked even if it was already
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 182) * passed over
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 183) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 184) if (test_bit(UNIX_GC_MAYBE_CYCLE, &u->gc_flags))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 185) list_move_tail(&u->link, &gc_candidates);
^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) static bool gc_in_progress;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 189) #define UNIX_INFLIGHT_TRIGGER_GC 16000
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 190)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 191) void wait_for_unix_gc(void)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 192) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 193) /* If number of inflight sockets is insane,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 194) * force a garbage collect right now.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 195) * Paired with the WRITE_ONCE() in unix_inflight(),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 196) * unix_notinflight() and gc_in_progress().
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 197) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 198) if (READ_ONCE(unix_tot_inflight) > UNIX_INFLIGHT_TRIGGER_GC &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 199) !READ_ONCE(gc_in_progress))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 200) unix_gc();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 201) wait_event(unix_gc_wait, gc_in_progress == false);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 202) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 203)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 204) /* The external entry point: unix_gc() */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 205) void unix_gc(void)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 206) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 207) struct unix_sock *u;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 208) struct unix_sock *next;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 209) struct sk_buff_head hitlist;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 210) struct list_head cursor;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 211) LIST_HEAD(not_cycle_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 212)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 213) spin_lock(&unix_gc_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 214)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 215) /* Avoid a recursive GC. */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 216) if (gc_in_progress)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 217) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 218)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 219) /* Paired with READ_ONCE() in wait_for_unix_gc(). */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 220) WRITE_ONCE(gc_in_progress, true);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 221)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 222) /* First, select candidates for garbage collection. Only
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 223) * in-flight sockets are considered, and from those only ones
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 224) * which don't have any external reference.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 225) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 226) * Holding unix_gc_lock will protect these candidates from
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 227) * being detached, and hence from gaining an external
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 228) * reference. Since there are no possible receivers, all
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 229) * buffers currently on the candidates' queues stay there
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 230) * during the garbage collection.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 231) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 232) * We also know that no new candidate can be added onto the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 233) * receive queues. Other, non candidate sockets _can_ be
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 234) * added to queue, so we must make sure only to touch
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 235) * candidates.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 236) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 237) list_for_each_entry_safe(u, next, &gc_inflight_list, link) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 238) long total_refs;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 239) long inflight_refs;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 240)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 241) total_refs = file_count(u->sk.sk_socket->file);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 242) inflight_refs = atomic_long_read(&u->inflight);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 243)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 244) BUG_ON(inflight_refs < 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 245) BUG_ON(total_refs < inflight_refs);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 246) if (total_refs == inflight_refs) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 247) list_move_tail(&u->link, &gc_candidates);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 248) __set_bit(UNIX_GC_CANDIDATE, &u->gc_flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 249) __set_bit(UNIX_GC_MAYBE_CYCLE, &u->gc_flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 250) }
^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) /* Now remove all internal in-flight reference to children of
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 254) * the candidates.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 255) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 256) list_for_each_entry(u, &gc_candidates, link)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 257) scan_children(&u->sk, dec_inflight, NULL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 258)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 259) /* Restore the references for children of all candidates,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 260) * which have remaining references. Do this recursively, so
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 261) * only those remain, which form cyclic references.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 262) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 263) * Use a "cursor" link, to make the list traversal safe, even
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 264) * though elements might be moved about.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 265) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 266) list_add(&cursor, &gc_candidates);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 267) while (cursor.next != &gc_candidates) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 268) u = list_entry(cursor.next, struct unix_sock, link);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 269)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 270) /* Move cursor to after the current position. */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 271) list_move(&cursor, &u->link);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 272)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 273) if (atomic_long_read(&u->inflight) > 0) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 274) list_move_tail(&u->link, ¬_cycle_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 275) __clear_bit(UNIX_GC_MAYBE_CYCLE, &u->gc_flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 276) scan_children(&u->sk, inc_inflight_move_tail, NULL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 277) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 278) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 279) list_del(&cursor);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 280)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 281) /* Now gc_candidates contains only garbage. Restore original
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 282) * inflight counters for these as well, and remove the skbuffs
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 283) * which are creating the cycle(s).
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 284) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 285) skb_queue_head_init(&hitlist);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 286) list_for_each_entry(u, &gc_candidates, link)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 287) scan_children(&u->sk, inc_inflight, &hitlist);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 288)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 289) /* not_cycle_list contains those sockets which do not make up a
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 290) * cycle. Restore these to the inflight list.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 291) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 292) while (!list_empty(¬_cycle_list)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 293) u = list_entry(not_cycle_list.next, struct unix_sock, link);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 294) __clear_bit(UNIX_GC_CANDIDATE, &u->gc_flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 295) list_move_tail(&u->link, &gc_inflight_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 296) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 297)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 298) spin_unlock(&unix_gc_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 299)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 300) /* Here we are. Hitlist is filled. Die. */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 301) __skb_queue_purge(&hitlist);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 302)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 303) spin_lock(&unix_gc_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 304)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 305) /* All candidates should have been detached by now. */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 306) BUG_ON(!list_empty(&gc_candidates));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 307)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 308) /* Paired with READ_ONCE() in wait_for_unix_gc(). */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 309) WRITE_ONCE(gc_in_progress, false);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 310)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 311) wake_up(&unix_gc_wait);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 312)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 313) out:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 314) spin_unlock(&unix_gc_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 315) }
Orange Pi5 kernel
Deprecated Linux kernel 5.10.110 for OrangePi 5/5B/5+ boards
3 Commits
0 Branches
0 Tags