mirror of
git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
synced 2026-03-21 23:16:50 +08:00
Igor Ushakov reported that GC purged the receive queue of an alive socket due to a race with MSG_PEEK with a nice repro. This is the exact same issue previously fixed by commitcbcf01128d("af_unix: fix garbage collect vs MSG_PEEK"). After GC was replaced with the current algorithm, the cited commit removed the locking dance in unix_peek_fds() and reintroduced the same issue. The problem is that MSG_PEEK bumps a file refcount without interacting with GC. Consider an SCC containing sk-A and sk-B, where sk-A is close()d but can be recv()ed via sk-B. The bad thing happens if sk-A is recv()ed with MSG_PEEK from sk-B and sk-B is close()d while GC is checking unix_vertex_dead() for sk-A and sk-B. GC thread User thread --------- ----------- unix_vertex_dead(sk-A) -> true <------. \ `------ recv(sk-B, MSG_PEEK) invalidate !! -> sk-A's file refcount : 1 -> 2 close(sk-B) -> sk-B's file refcount : 2 -> 1 unix_vertex_dead(sk-B) -> true Initially, sk-A's file refcount is 1 by the inflight fd in sk-B recvq. GC thinks sk-A is dead because the file refcount is the same as the number of its inflight fds. However, sk-A's file refcount is bumped silently by MSG_PEEK, which invalidates the previous evaluation. At this moment, sk-B's file refcount is 2; one by the open fd, and one by the inflight fd in sk-A. The subsequent close() releases one refcount by the former. Finally, GC incorrectly concludes that both sk-A and sk-B are dead. One option is to restore the locking dance in unix_peek_fds(), but we can resolve this more elegantly thanks to the new algorithm. The point is that the issue does not occur without the subsequent close() and we actually do not need to synchronise MSG_PEEK with the dead SCC detection. When the issue occurs, close() and GC touch the same file refcount. If GC sees the refcount being decremented by close(), it can just give up garbage-collecting the SCC. Therefore, we only need to signal the race during MSG_PEEK with a proper memory barrier to make it visible to the GC. Let's use seqcount_t to notify GC when MSG_PEEK occurs and let it defer the SCC to the next run. This way no locking is needed on the MSG_PEEK side, and we can avoid imposing a penalty on every MSG_PEEK unnecessarily. Note that we can retry within unix_scc_dead() if MSG_PEEK is detected, but we do not do so to avoid hung task splat from abusive MSG_PEEK calls. Fixes:118f457da9("af_unix: Remove lock dance in unix_peek_fds().") Reported-by: Igor Ushakov <sysroot314@gmail.com> Signed-off-by: Kuniyuki Iwashima <kuniyu@google.com> Link: https://patch.msgid.link/20260311054043.1231316-1-kuniyu@google.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
660 lines
16 KiB
C
660 lines
16 KiB
C
// SPDX-License-Identifier: GPL-2.0-or-later
|
|
/*
|
|
* NET3: Garbage Collector For AF_UNIX sockets
|
|
*
|
|
* Garbage Collector:
|
|
* Copyright (C) Barak A. Pearlmutter.
|
|
*
|
|
* Chopped about by Alan Cox 22/3/96 to make it fit the AF_UNIX socket problem.
|
|
* If it doesn't work blame me, it worked when Barak sent it.
|
|
*
|
|
* Assumptions:
|
|
*
|
|
* - object w/ a bit
|
|
* - free list
|
|
*
|
|
* Current optimizations:
|
|
*
|
|
* - explicit stack instead of recursion
|
|
* - tail recurse on first born instead of immediate push/pop
|
|
* - we gather the stuff that should not be killed into tree
|
|
* and stack is just a path from root to the current pointer.
|
|
*
|
|
* Future optimizations:
|
|
*
|
|
* - don't just push entire root set; process in place
|
|
*
|
|
* Fixes:
|
|
* Alan Cox 07 Sept 1997 Vmalloc internal stack as needed.
|
|
* Cope with changing max_files.
|
|
* Al Viro 11 Oct 1998
|
|
* Graph may have cycles. That is, we can send the descriptor
|
|
* of foo to bar and vice versa. Current code chokes on that.
|
|
* Fix: move SCM_RIGHTS ones into the separate list and then
|
|
* skb_free() them all instead of doing explicit fput's.
|
|
* Another problem: since fput() may block somebody may
|
|
* create a new unix_socket when we are in the middle of sweep
|
|
* phase. Fix: revert the logic wrt MARKED. Mark everything
|
|
* upon the beginning and unmark non-junk ones.
|
|
*
|
|
* [12 Oct 1998] AAARGH! New code purges all SCM_RIGHTS
|
|
* sent to connect()'ed but still not accept()'ed sockets.
|
|
* Fixed. Old code had slightly different problem here:
|
|
* extra fput() in situation when we passed the descriptor via
|
|
* such socket and closed it (descriptor). That would happen on
|
|
* each unix_gc() until the accept(). Since the struct file in
|
|
* question would go to the free list and might be reused...
|
|
* That might be the reason of random oopses on filp_close()
|
|
* in unrelated processes.
|
|
*
|
|
* AV 28 Feb 1999
|
|
* Kill the explicit allocation of stack. Now we keep the tree
|
|
* with root in dummy + pointer (gc_current) to one of the nodes.
|
|
* Stack is represented as path from gc_current to dummy. Unmark
|
|
* now means "add to tree". Push == "make it a son of gc_current".
|
|
* Pop == "move gc_current to parent". We keep only pointers to
|
|
* parents (->gc_tree).
|
|
* AV 1 Mar 1999
|
|
* Damn. Added missing check for ->dead in listen queues scanning.
|
|
*
|
|
* Miklos Szeredi 25 Jun 2007
|
|
* Reimplement with a cycle collecting algorithm. This should
|
|
* solve several problems with the previous code, like being racy
|
|
* wrt receive and holding up unrelated socket operations.
|
|
*/
|
|
|
|
#include <linux/fs.h>
|
|
#include <linux/list.h>
|
|
#include <linux/skbuff.h>
|
|
#include <linux/socket.h>
|
|
#include <linux/workqueue.h>
|
|
#include <net/af_unix.h>
|
|
#include <net/scm.h>
|
|
#include <net/tcp_states.h>
|
|
|
|
#include "af_unix.h"
|
|
|
|
struct unix_vertex {
|
|
struct list_head edges;
|
|
struct list_head entry;
|
|
struct list_head scc_entry;
|
|
unsigned long out_degree;
|
|
unsigned long index;
|
|
unsigned long scc_index;
|
|
};
|
|
|
|
struct unix_edge {
|
|
struct unix_sock *predecessor;
|
|
struct unix_sock *successor;
|
|
struct list_head vertex_entry;
|
|
struct list_head stack_entry;
|
|
};
|
|
|
|
struct unix_sock *unix_get_socket(struct file *filp)
|
|
{
|
|
struct inode *inode = file_inode(filp);
|
|
|
|
/* Socket ? */
|
|
if (S_ISSOCK(inode->i_mode) && !(filp->f_mode & FMODE_PATH)) {
|
|
struct socket *sock = SOCKET_I(inode);
|
|
const struct proto_ops *ops;
|
|
struct sock *sk = sock->sk;
|
|
|
|
ops = READ_ONCE(sock->ops);
|
|
|
|
/* PF_UNIX ? */
|
|
if (sk && ops && ops->family == PF_UNIX)
|
|
return unix_sk(sk);
|
|
}
|
|
|
|
return NULL;
|
|
}
|
|
|
|
static struct unix_vertex *unix_edge_successor(struct unix_edge *edge)
|
|
{
|
|
/* If an embryo socket has a fd,
|
|
* the listener indirectly holds the fd's refcnt.
|
|
*/
|
|
if (edge->successor->listener)
|
|
return unix_sk(edge->successor->listener)->vertex;
|
|
|
|
return edge->successor->vertex;
|
|
}
|
|
|
|
enum {
|
|
UNIX_GRAPH_NOT_CYCLIC,
|
|
UNIX_GRAPH_MAYBE_CYCLIC,
|
|
UNIX_GRAPH_CYCLIC,
|
|
};
|
|
|
|
static unsigned char unix_graph_state;
|
|
|
|
static void unix_update_graph(struct unix_vertex *vertex)
|
|
{
|
|
/* If the receiver socket is not inflight, no cyclic
|
|
* reference could be formed.
|
|
*/
|
|
if (!vertex)
|
|
return;
|
|
|
|
WRITE_ONCE(unix_graph_state, UNIX_GRAPH_MAYBE_CYCLIC);
|
|
}
|
|
|
|
static LIST_HEAD(unix_unvisited_vertices);
|
|
|
|
enum unix_vertex_index {
|
|
UNIX_VERTEX_INDEX_MARK1,
|
|
UNIX_VERTEX_INDEX_MARK2,
|
|
UNIX_VERTEX_INDEX_START,
|
|
};
|
|
|
|
static unsigned long unix_vertex_unvisited_index = UNIX_VERTEX_INDEX_MARK1;
|
|
static unsigned long unix_vertex_max_scc_index = UNIX_VERTEX_INDEX_START;
|
|
|
|
static void unix_add_edge(struct scm_fp_list *fpl, struct unix_edge *edge)
|
|
{
|
|
struct unix_vertex *vertex = edge->predecessor->vertex;
|
|
|
|
if (!vertex) {
|
|
vertex = list_first_entry(&fpl->vertices, typeof(*vertex), entry);
|
|
vertex->index = unix_vertex_unvisited_index;
|
|
vertex->scc_index = ++unix_vertex_max_scc_index;
|
|
vertex->out_degree = 0;
|
|
INIT_LIST_HEAD(&vertex->edges);
|
|
INIT_LIST_HEAD(&vertex->scc_entry);
|
|
|
|
list_move_tail(&vertex->entry, &unix_unvisited_vertices);
|
|
edge->predecessor->vertex = vertex;
|
|
}
|
|
|
|
vertex->out_degree++;
|
|
list_add_tail(&edge->vertex_entry, &vertex->edges);
|
|
|
|
unix_update_graph(unix_edge_successor(edge));
|
|
}
|
|
|
|
static void unix_del_edge(struct scm_fp_list *fpl, struct unix_edge *edge)
|
|
{
|
|
struct unix_vertex *vertex = edge->predecessor->vertex;
|
|
|
|
if (!fpl->dead)
|
|
unix_update_graph(unix_edge_successor(edge));
|
|
|
|
list_del(&edge->vertex_entry);
|
|
vertex->out_degree--;
|
|
|
|
if (!vertex->out_degree) {
|
|
edge->predecessor->vertex = NULL;
|
|
list_move_tail(&vertex->entry, &fpl->vertices);
|
|
}
|
|
}
|
|
|
|
static void unix_free_vertices(struct scm_fp_list *fpl)
|
|
{
|
|
struct unix_vertex *vertex, *next_vertex;
|
|
|
|
list_for_each_entry_safe(vertex, next_vertex, &fpl->vertices, entry) {
|
|
list_del(&vertex->entry);
|
|
kfree(vertex);
|
|
}
|
|
}
|
|
|
|
static __cacheline_aligned_in_smp DEFINE_SPINLOCK(unix_gc_lock);
|
|
|
|
void unix_add_edges(struct scm_fp_list *fpl, struct unix_sock *receiver)
|
|
{
|
|
int i = 0, j = 0;
|
|
|
|
spin_lock(&unix_gc_lock);
|
|
|
|
if (!fpl->count_unix)
|
|
goto out;
|
|
|
|
do {
|
|
struct unix_sock *inflight = unix_get_socket(fpl->fp[j++]);
|
|
struct unix_edge *edge;
|
|
|
|
if (!inflight)
|
|
continue;
|
|
|
|
edge = fpl->edges + i++;
|
|
edge->predecessor = inflight;
|
|
edge->successor = receiver;
|
|
|
|
unix_add_edge(fpl, edge);
|
|
} while (i < fpl->count_unix);
|
|
|
|
receiver->scm_stat.nr_unix_fds += fpl->count_unix;
|
|
out:
|
|
WRITE_ONCE(fpl->user->unix_inflight, fpl->user->unix_inflight + fpl->count);
|
|
|
|
spin_unlock(&unix_gc_lock);
|
|
|
|
fpl->inflight = true;
|
|
|
|
unix_free_vertices(fpl);
|
|
}
|
|
|
|
void unix_del_edges(struct scm_fp_list *fpl)
|
|
{
|
|
struct unix_sock *receiver;
|
|
int i = 0;
|
|
|
|
spin_lock(&unix_gc_lock);
|
|
|
|
if (!fpl->count_unix)
|
|
goto out;
|
|
|
|
do {
|
|
struct unix_edge *edge = fpl->edges + i++;
|
|
|
|
unix_del_edge(fpl, edge);
|
|
} while (i < fpl->count_unix);
|
|
|
|
if (!fpl->dead) {
|
|
receiver = fpl->edges[0].successor;
|
|
receiver->scm_stat.nr_unix_fds -= fpl->count_unix;
|
|
}
|
|
out:
|
|
WRITE_ONCE(fpl->user->unix_inflight, fpl->user->unix_inflight - fpl->count);
|
|
|
|
spin_unlock(&unix_gc_lock);
|
|
|
|
fpl->inflight = false;
|
|
}
|
|
|
|
void unix_update_edges(struct unix_sock *receiver)
|
|
{
|
|
/* nr_unix_fds is only updated under unix_state_lock().
|
|
* If it's 0 here, the embryo socket is not part of the
|
|
* inflight graph, and GC will not see it, so no lock needed.
|
|
*/
|
|
if (!receiver->scm_stat.nr_unix_fds) {
|
|
receiver->listener = NULL;
|
|
} else {
|
|
spin_lock(&unix_gc_lock);
|
|
unix_update_graph(unix_sk(receiver->listener)->vertex);
|
|
receiver->listener = NULL;
|
|
spin_unlock(&unix_gc_lock);
|
|
}
|
|
}
|
|
|
|
int unix_prepare_fpl(struct scm_fp_list *fpl)
|
|
{
|
|
struct unix_vertex *vertex;
|
|
int i;
|
|
|
|
if (!fpl->count_unix)
|
|
return 0;
|
|
|
|
for (i = 0; i < fpl->count_unix; i++) {
|
|
vertex = kmalloc_obj(*vertex);
|
|
if (!vertex)
|
|
goto err;
|
|
|
|
list_add(&vertex->entry, &fpl->vertices);
|
|
}
|
|
|
|
fpl->edges = kvmalloc_objs(*fpl->edges, fpl->count_unix,
|
|
GFP_KERNEL_ACCOUNT);
|
|
if (!fpl->edges)
|
|
goto err;
|
|
|
|
unix_schedule_gc(fpl->user);
|
|
|
|
return 0;
|
|
|
|
err:
|
|
unix_free_vertices(fpl);
|
|
return -ENOMEM;
|
|
}
|
|
|
|
void unix_destroy_fpl(struct scm_fp_list *fpl)
|
|
{
|
|
if (fpl->inflight)
|
|
unix_del_edges(fpl);
|
|
|
|
kvfree(fpl->edges);
|
|
unix_free_vertices(fpl);
|
|
}
|
|
|
|
static bool gc_in_progress;
|
|
static seqcount_t unix_peek_seq = SEQCNT_ZERO(unix_peek_seq);
|
|
|
|
void unix_peek_fpl(struct scm_fp_list *fpl)
|
|
{
|
|
static DEFINE_SPINLOCK(unix_peek_lock);
|
|
|
|
if (!fpl || !fpl->count_unix)
|
|
return;
|
|
|
|
if (!READ_ONCE(gc_in_progress))
|
|
return;
|
|
|
|
/* Invalidate the final refcnt check in unix_vertex_dead(). */
|
|
spin_lock(&unix_peek_lock);
|
|
raw_write_seqcount_barrier(&unix_peek_seq);
|
|
spin_unlock(&unix_peek_lock);
|
|
}
|
|
|
|
static bool unix_vertex_dead(struct unix_vertex *vertex)
|
|
{
|
|
struct unix_edge *edge;
|
|
struct unix_sock *u;
|
|
long total_ref;
|
|
|
|
list_for_each_entry(edge, &vertex->edges, vertex_entry) {
|
|
struct unix_vertex *next_vertex = unix_edge_successor(edge);
|
|
|
|
/* The vertex's fd can be received by a non-inflight socket. */
|
|
if (!next_vertex)
|
|
return false;
|
|
|
|
/* The vertex's fd can be received by an inflight socket in
|
|
* another SCC.
|
|
*/
|
|
if (next_vertex->scc_index != vertex->scc_index)
|
|
return false;
|
|
}
|
|
|
|
/* No receiver exists out of the same SCC. */
|
|
|
|
edge = list_first_entry(&vertex->edges, typeof(*edge), vertex_entry);
|
|
u = edge->predecessor;
|
|
total_ref = file_count(u->sk.sk_socket->file);
|
|
|
|
/* If not close()d, total_ref > out_degree. */
|
|
if (total_ref != vertex->out_degree)
|
|
return false;
|
|
|
|
return true;
|
|
}
|
|
|
|
static LIST_HEAD(unix_visited_vertices);
|
|
static unsigned long unix_vertex_grouped_index = UNIX_VERTEX_INDEX_MARK2;
|
|
|
|
static bool unix_scc_dead(struct list_head *scc, bool fast)
|
|
{
|
|
struct unix_vertex *vertex;
|
|
bool scc_dead = true;
|
|
unsigned int seq;
|
|
|
|
seq = read_seqcount_begin(&unix_peek_seq);
|
|
|
|
list_for_each_entry_reverse(vertex, scc, scc_entry) {
|
|
/* Don't restart DFS from this vertex. */
|
|
list_move_tail(&vertex->entry, &unix_visited_vertices);
|
|
|
|
/* Mark vertex as off-stack for __unix_walk_scc(). */
|
|
if (!fast)
|
|
vertex->index = unix_vertex_grouped_index;
|
|
|
|
if (scc_dead)
|
|
scc_dead = unix_vertex_dead(vertex);
|
|
}
|
|
|
|
/* If MSG_PEEK intervened, defer this SCC to the next round. */
|
|
if (read_seqcount_retry(&unix_peek_seq, seq))
|
|
return false;
|
|
|
|
return scc_dead;
|
|
}
|
|
|
|
static void unix_collect_skb(struct list_head *scc, struct sk_buff_head *hitlist)
|
|
{
|
|
struct unix_vertex *vertex;
|
|
|
|
list_for_each_entry_reverse(vertex, scc, scc_entry) {
|
|
struct sk_buff_head *queue;
|
|
struct unix_edge *edge;
|
|
struct unix_sock *u;
|
|
|
|
edge = list_first_entry(&vertex->edges, typeof(*edge), vertex_entry);
|
|
u = edge->predecessor;
|
|
queue = &u->sk.sk_receive_queue;
|
|
|
|
spin_lock(&queue->lock);
|
|
|
|
if (u->sk.sk_state == TCP_LISTEN) {
|
|
struct sk_buff *skb;
|
|
|
|
skb_queue_walk(queue, skb) {
|
|
struct sk_buff_head *embryo_queue = &skb->sk->sk_receive_queue;
|
|
|
|
spin_lock(&embryo_queue->lock);
|
|
skb_queue_splice_init(embryo_queue, hitlist);
|
|
spin_unlock(&embryo_queue->lock);
|
|
}
|
|
} else {
|
|
skb_queue_splice_init(queue, hitlist);
|
|
}
|
|
|
|
spin_unlock(&queue->lock);
|
|
}
|
|
}
|
|
|
|
static bool unix_scc_cyclic(struct list_head *scc)
|
|
{
|
|
struct unix_vertex *vertex;
|
|
struct unix_edge *edge;
|
|
|
|
/* SCC containing multiple vertices ? */
|
|
if (!list_is_singular(scc))
|
|
return true;
|
|
|
|
vertex = list_first_entry(scc, typeof(*vertex), scc_entry);
|
|
|
|
/* Self-reference or a embryo-listener circle ? */
|
|
list_for_each_entry(edge, &vertex->edges, vertex_entry) {
|
|
if (unix_edge_successor(edge) == vertex)
|
|
return true;
|
|
}
|
|
|
|
return false;
|
|
}
|
|
|
|
static unsigned long __unix_walk_scc(struct unix_vertex *vertex,
|
|
unsigned long *last_index,
|
|
struct sk_buff_head *hitlist)
|
|
{
|
|
unsigned long cyclic_sccs = 0;
|
|
LIST_HEAD(vertex_stack);
|
|
struct unix_edge *edge;
|
|
LIST_HEAD(edge_stack);
|
|
|
|
next_vertex:
|
|
/* Push vertex to vertex_stack and mark it as on-stack
|
|
* (index >= UNIX_VERTEX_INDEX_START).
|
|
* The vertex will be popped when finalising SCC later.
|
|
*/
|
|
list_add(&vertex->scc_entry, &vertex_stack);
|
|
|
|
vertex->index = *last_index;
|
|
vertex->scc_index = *last_index;
|
|
(*last_index)++;
|
|
|
|
/* Explore neighbour vertices (receivers of the current vertex's fd). */
|
|
list_for_each_entry(edge, &vertex->edges, vertex_entry) {
|
|
struct unix_vertex *next_vertex = unix_edge_successor(edge);
|
|
|
|
if (!next_vertex)
|
|
continue;
|
|
|
|
if (next_vertex->index == unix_vertex_unvisited_index) {
|
|
/* Iterative deepening depth first search
|
|
*
|
|
* 1. Push a forward edge to edge_stack and set
|
|
* the successor to vertex for the next iteration.
|
|
*/
|
|
list_add(&edge->stack_entry, &edge_stack);
|
|
|
|
vertex = next_vertex;
|
|
goto next_vertex;
|
|
|
|
/* 2. Pop the edge directed to the current vertex
|
|
* and restore the ancestor for backtracking.
|
|
*/
|
|
prev_vertex:
|
|
edge = list_first_entry(&edge_stack, typeof(*edge), stack_entry);
|
|
list_del_init(&edge->stack_entry);
|
|
|
|
next_vertex = vertex;
|
|
vertex = edge->predecessor->vertex;
|
|
|
|
/* If the successor has a smaller scc_index, two vertices
|
|
* are in the same SCC, so propagate the smaller scc_index
|
|
* to skip SCC finalisation.
|
|
*/
|
|
vertex->scc_index = min(vertex->scc_index, next_vertex->scc_index);
|
|
} else if (next_vertex->index != unix_vertex_grouped_index) {
|
|
/* Loop detected by a back/cross edge.
|
|
*
|
|
* The successor is on vertex_stack, so two vertices are in
|
|
* the same SCC. If the successor has a smaller *scc_index*,
|
|
* propagate it to skip SCC finalisation.
|
|
*/
|
|
vertex->scc_index = min(vertex->scc_index, next_vertex->scc_index);
|
|
} else {
|
|
/* The successor was already grouped as another SCC */
|
|
}
|
|
}
|
|
|
|
if (vertex->index == vertex->scc_index) {
|
|
struct list_head scc;
|
|
|
|
/* SCC finalised.
|
|
*
|
|
* If the scc_index was not updated, all the vertices above on
|
|
* vertex_stack are in the same SCC. Group them using scc_entry.
|
|
*/
|
|
__list_cut_position(&scc, &vertex_stack, &vertex->scc_entry);
|
|
|
|
if (unix_scc_dead(&scc, false)) {
|
|
unix_collect_skb(&scc, hitlist);
|
|
} else {
|
|
if (unix_vertex_max_scc_index < vertex->scc_index)
|
|
unix_vertex_max_scc_index = vertex->scc_index;
|
|
|
|
if (unix_scc_cyclic(&scc))
|
|
cyclic_sccs++;
|
|
}
|
|
|
|
list_del(&scc);
|
|
}
|
|
|
|
/* Need backtracking ? */
|
|
if (!list_empty(&edge_stack))
|
|
goto prev_vertex;
|
|
|
|
return cyclic_sccs;
|
|
}
|
|
|
|
static unsigned long unix_graph_cyclic_sccs;
|
|
|
|
static void unix_walk_scc(struct sk_buff_head *hitlist)
|
|
{
|
|
unsigned long last_index = UNIX_VERTEX_INDEX_START;
|
|
unsigned long cyclic_sccs = 0;
|
|
|
|
unix_vertex_max_scc_index = UNIX_VERTEX_INDEX_START;
|
|
|
|
/* Visit every vertex exactly once.
|
|
* __unix_walk_scc() moves visited vertices to unix_visited_vertices.
|
|
*/
|
|
while (!list_empty(&unix_unvisited_vertices)) {
|
|
struct unix_vertex *vertex;
|
|
|
|
vertex = list_first_entry(&unix_unvisited_vertices, typeof(*vertex), entry);
|
|
cyclic_sccs += __unix_walk_scc(vertex, &last_index, hitlist);
|
|
}
|
|
|
|
list_replace_init(&unix_visited_vertices, &unix_unvisited_vertices);
|
|
swap(unix_vertex_unvisited_index, unix_vertex_grouped_index);
|
|
|
|
WRITE_ONCE(unix_graph_cyclic_sccs, cyclic_sccs);
|
|
WRITE_ONCE(unix_graph_state,
|
|
cyclic_sccs ? UNIX_GRAPH_CYCLIC : UNIX_GRAPH_NOT_CYCLIC);
|
|
}
|
|
|
|
static void unix_walk_scc_fast(struct sk_buff_head *hitlist)
|
|
{
|
|
unsigned long cyclic_sccs = unix_graph_cyclic_sccs;
|
|
|
|
while (!list_empty(&unix_unvisited_vertices)) {
|
|
struct unix_vertex *vertex;
|
|
struct list_head scc;
|
|
|
|
vertex = list_first_entry(&unix_unvisited_vertices, typeof(*vertex), entry);
|
|
list_add(&scc, &vertex->scc_entry);
|
|
|
|
if (unix_scc_dead(&scc, true)) {
|
|
cyclic_sccs--;
|
|
unix_collect_skb(&scc, hitlist);
|
|
}
|
|
|
|
list_del(&scc);
|
|
}
|
|
|
|
list_replace_init(&unix_visited_vertices, &unix_unvisited_vertices);
|
|
|
|
WRITE_ONCE(unix_graph_cyclic_sccs, cyclic_sccs);
|
|
WRITE_ONCE(unix_graph_state,
|
|
cyclic_sccs ? UNIX_GRAPH_CYCLIC : UNIX_GRAPH_NOT_CYCLIC);
|
|
}
|
|
|
|
static void unix_gc(struct work_struct *work)
|
|
{
|
|
struct sk_buff_head hitlist;
|
|
struct sk_buff *skb;
|
|
|
|
spin_lock(&unix_gc_lock);
|
|
|
|
if (unix_graph_state == UNIX_GRAPH_NOT_CYCLIC) {
|
|
spin_unlock(&unix_gc_lock);
|
|
goto skip_gc;
|
|
}
|
|
|
|
__skb_queue_head_init(&hitlist);
|
|
|
|
if (unix_graph_state == UNIX_GRAPH_CYCLIC)
|
|
unix_walk_scc_fast(&hitlist);
|
|
else
|
|
unix_walk_scc(&hitlist);
|
|
|
|
spin_unlock(&unix_gc_lock);
|
|
|
|
skb_queue_walk(&hitlist, skb) {
|
|
if (UNIXCB(skb).fp)
|
|
UNIXCB(skb).fp->dead = true;
|
|
}
|
|
|
|
__skb_queue_purge_reason(&hitlist, SKB_DROP_REASON_SOCKET_CLOSE);
|
|
skip_gc:
|
|
WRITE_ONCE(gc_in_progress, false);
|
|
}
|
|
|
|
static DECLARE_WORK(unix_gc_work, unix_gc);
|
|
|
|
#define UNIX_INFLIGHT_SANE_USER (SCM_MAX_FD * 8)
|
|
|
|
void unix_schedule_gc(struct user_struct *user)
|
|
{
|
|
if (READ_ONCE(unix_graph_state) == UNIX_GRAPH_NOT_CYCLIC)
|
|
return;
|
|
|
|
/* Penalise users who want to send AF_UNIX sockets
|
|
* but whose sockets have not been received yet.
|
|
*/
|
|
if (user &&
|
|
READ_ONCE(user->unix_inflight) < UNIX_INFLIGHT_SANE_USER)
|
|
return;
|
|
|
|
if (!READ_ONCE(gc_in_progress)) {
|
|
WRITE_ONCE(gc_in_progress, true);
|
|
queue_work(system_dfl_wq, &unix_gc_work);
|
|
}
|
|
|
|
if (user && READ_ONCE(unix_graph_cyclic_sccs))
|
|
flush_work(&unix_gc_work);
|
|
}
|