xref: /freebsd/share/man/man9/epoch.9 (revision a7dea1671b87c07d2d266f836bfa8b58efc7c134)
1.\"
2.\" Copyright (C) 2018 Matthew Macy <mmacy@FreeBSD.org>.
3.\"
4.\" Redistribution and use in source and binary forms, with or without
5.\" modification, are permitted provided that the following conditions
6.\" are met:
7.\" 1. Redistributions of source code must retain the above copyright
8.\"    notice(s), this list of conditions and the following disclaimer as
9.\"    the first lines of this file unmodified other than the possible
10.\"    addition of one or more copyright notices.
11.\" 2. Redistributions in binary form must reproduce the above copyright
12.\"    notice(s), this list of conditions and the following disclaimer in the
13.\"    documentation and/or other materials provided with the distribution.
14.\"
15.\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY
16.\" EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
17.\" WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
18.\" DISCLAIMED.  IN NO EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY
19.\" DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
20.\" (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
21.\" SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
22.\" CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
23.\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
24.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH
25.\" DAMAGE.
26.\"
27.\" $FreeBSD$
28.\"
29.Dd December 27, 2019
30.Dt EPOCH 9
31.Os
32.Sh NAME
33.Nm epoch ,
34.Nm epoch_context ,
35.Nm epoch_alloc ,
36.Nm epoch_free ,
37.Nm epoch_enter ,
38.Nm epoch_exit ,
39.Nm epoch_wait ,
40.Nm epoch_enter_preempt ,
41.Nm epoch_exit_preempt ,
42.Nm epoch_wait_preempt ,
43.Nm epoch_call ,
44.Nm epoch_drain_callbacks ,
45.Nm in_epoch ,
46.Nm in_epoch_verbose ,
47.Nd kernel epoch based reclamation
48.Sh SYNOPSIS
49.In sys/param.h
50.In sys/proc.h
51.In sys/epoch.h
52.\" Types
53.Bd -literal
54struct epoch;		/* Opaque */
55.Ed
56.Vt typedef "struct epoch *epoch_t" ;
57.Bd -literal
58struct epoch_context {
59	void	*data[2];
60};
61.Ed
62.Vt typedef "struct epoch_context *epoch_context_t" ;
63.Bd -literal
64struct epoch_tracker;	/* Opaque */
65.Ed
66.Vt typedef "struct epoch_tracker *epoch_tracker_t" ;
67.\" Declarations
68.Ft epoch_t
69.Fn epoch_alloc "const char *name" "int flags"
70.Ft void
71.Fn epoch_free "epoch_t epoch"
72.Ft void
73.Fn epoch_enter "epoch_t epoch"
74.Ft void
75.Fn epoch_exit "epoch_t epoch"
76.Ft void
77.Fn epoch_wait "epoch_t epoch"
78.Ft void
79.Fn epoch_enter_preempt "epoch_t epoch" "epoch_tracker_t et"
80.Ft void
81.Fn epoch_exit_preempt "epoch_t epoch" "epoch_tracker_t et"
82.Ft void
83.Fn epoch_wait_preempt "epoch_t epoch"
84.Ft void
85.Fn epoch_call "epoch_t epoch" "epoch_context_t ctx" "void (*callback)(epoch_context_t)"
86.Ft void
87.Fn epoch_drain_callbacks "epoch_t epoch"
88.Ft int
89.Fn in_epoch "epoch_t epoch"
90.Ft int
91.Fn in_epoch_verbose "epoch_t epoch" "int dump_onfail"
92.Sh DESCRIPTION
93Epochs are used to guarantee liveness and immutability of data by
94deferring reclamation and mutation until a grace period has elapsed.
95Epochs do not have any lock ordering issues.
96Entering and leaving an epoch section will never block.
97.Pp
98Epochs are allocated with
99.Fn epoch_alloc .
100The
101.Fa name
102argument is used for debugging convenience when the
103.Cd EPOCH_TRACE
104kernel option is configured.
105By default, epochs do not allow preemption during sections.
106By default mutexes cannot be held across
107.Fn epoch_wait_preempt .
108The
109.Fa flags
110specified are formed by
111.Em OR Ns 'ing
112the following values:
113.Bl -tag -offset indent -width Ds
114.It Dv EPOCH_LOCKED
115Permit holding mutexes across
116.Fn epoch_wait_preempt
117(requires
118.Dv EPOCH_PREEMPT ) .
119When doing this one must be cautious of creating a situation where a deadlock
120is possible.
121.It Dv EPOCH_PREEMPT
122The
123.Vt epoch
124will allow preemption during sections.
125Only non-sleepable locks may be acquired during a preemptible epoch.
126The functions
127.Fn epoch_enter_preempt ,
128.Fn epoch_exit_preempt ,
129and
130.Fn epoch_wait_preempt
131must be used in place of
132.Fn epoch_enter ,
133.Fn epoch_exit ,
134and
135.Fn epoch_wait ,
136respectively.
137.El
138.Pp
139.Vt epoch Ns s
140are freed with
141.Fn epoch_free .
142.Pp
143Threads indicate the start of an epoch critical section by calling
144.Fn epoch_enter
145(or
146.Fn epoch_enter_preempt
147for preemptible epochs).
148Threads call
149.Fn epoch_exit
150(or
151.Fn epoch_exit_preempt
152for preemptible epochs)
153to indicate the end of a critical section.
154.Vt struct epoch_tracker Ns s
155are stack objects whose pointers are passed to
156.Fn epoch_enter_preempt
157and
158.Fn epoch_exit_preempt
159(much like
160.Vt struct rm_priotracker ) .
161.Pp
162Threads can defer work until a grace period has expired since any thread has
163entered the epoch either synchronously or asynchronously.
164.Fn epoch_call
165defers work asynchronously by invoking the provided
166.Fa callback
167at a later time.
168.Fn epoch_wait
169(or
170.Fn epoch_wait_preempt )
171blocks the current thread until the grace period has expired and the work can be
172done safely.
173.Pp
174Default, non-preemptible epoch wait
175.Fn ( epoch_wait )
176is guaranteed to have much shorter completion times relative to
177preemptible epoch wait
178.Fn ( epoch_wait_preempt ) .
179(In the default type, none of the threads in an epoch section will be preempted
180before completing its section.)
181.Pp
182INVARIANTS can assert that a thread is in an epoch by using
183.Fn in_epoch .
184.Fn in_epoch "epoch"
185is equivalent to invoking
186.Fn in_epoch_verbose "epoch" "0" .
187If
188.Cd EPOCH_TRACE
189is enabled,
190.Fn in_epoch_verbose "epoch" "1"
191provides additional verbose debugging information.
192.Pp
193The epoch API currently does not support sleeping in epoch_preempt sections.
194A caller should never call
195.Fn epoch_wait
196in the middle of an epoch section for the same epoch as this will lead to a deadlock.
197.Pp
198The
199.Fn epoch_drain_callbacks
200function is used to drain all pending callbacks which have been invoked by prior
201.Fn epoch_call
202function calls on the same epoch.
203This function is useful when there are shared memory structure(s)
204referred to by the epoch callback(s) which are not refcounted and are
205rarely freed.
206The typical place for calling this function is right before freeing or
207invalidating the shared resource(s) used by the epoch callback(s).
208This function can sleep and is not optimized for performance.
209.Sh RETURN VALUES
210.Fn in_epoch curepoch
211will return 1 if curthread is in curepoch, 0 otherwise.
212.Sh CAVEATS
213One must be cautious when using
214.Fn epoch_wait_preempt .
215Threads are pinned during epoch sections, so if a thread in a section is then
216preempted by a higher priority compute bound thread on that CPU, it can be
217prevented from leaving the section indefinitely.
218.Pp
219Epochs are not a straight replacement for read locks.
220Callers must use safe list and tailq traversal routines in an epoch (see ck_queue).
221When modifying a list referenced from an epoch section safe removal
222routines must be used and the caller can no longer modify a list entry
223in place.
224An item to be modified must be handled with copy on write
225and frees must be deferred until after a grace period has elapsed.
226.Sh EXAMPLES
227Async free example:
228Thread 1:
229.Bd -literal
230int
231in_pcbladdr(struct inpcb *inp, struct in_addr *faddr, struct in_laddr *laddr,
232    struct ucred *cred)
233{
234    /* ... */
235    epoch_enter(net_epoch);
236    CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
237        sa = ifa->ifa_addr;
238	if (sa->sa_family != AF_INET)
239	    continue;
240	sin = (struct sockaddr_in *)sa;
241	if (prison_check_ip4(cred, &sin->sin_addr) == 0) {
242	     ia = (struct in_ifaddr *)ifa;
243	     break;
244	}
245    }
246    epoch_exit(net_epoch);
247    /* ... */
248}
249.Ed
250Thread 2:
251.Bd -literal
252void
253ifa_free(struct ifaddr *ifa)
254{
255
256    if (refcount_release(&ifa->ifa_refcnt))
257        epoch_call(net_epoch, &ifa->ifa_epoch_ctx, ifa_destroy);
258}
259
260void
261if_purgeaddrs(struct ifnet *ifp)
262{
263
264    /* .... *
265    IF_ADDR_WLOCK(ifp);
266    CK_STAILQ_REMOVE(&ifp->if_addrhead, ifa, ifaddr, ifa_link);
267    IF_ADDR_WUNLOCK(ifp);
268    ifa_free(ifa);
269}
270.Ed
271.Pp
272Thread 1 traverses the ifaddr list in an epoch.
273Thread 2 unlinks with the corresponding epoch safe macro, marks as logically free,
274and then defers deletion.
275More general mutation or a synchronous
276free would have to follow a call to
277.Fn epoch_wait .
278.Sh NOTES
279The
280.Nm
281kernel programming interface is under development and is subject to change.
282.El
283.Sh SEE ALSO
284.Xr locking 9 ,
285.Xr mtx_pool 9 ,
286.Xr mutex 9 ,
287.Xr rwlock 9 ,
288.Xr sema 9 ,
289.Xr sleep 9 ,
290.Xr sx 9 ,
291.Xr timeout 9
292