xref: /openbsd-src/share/man/man9/smr_call.9 (revision 3a85e55fffc5492a9223f8200ce707f4f4839082)

.\"	$OpenBSD: smr_call.9,v 1.4 2022/06/22 14:10:49 visa Exp $
.\"
.\" Copyright (c) 2019 Visa Hankala
.\"
.\" Permission to use, copy, modify, and distribute this software for any
.\" purpose with or without fee is hereby granted, provided that the above
.\" copyright notice and this permission notice appear in all copies.
.\"
.\" THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
.\" WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
.\" MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
.\" ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
.\" WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
.\" ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
.\" OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
.\"
.Dd $Mdocdate: June 22 2022 $
.Dt SMR_CALL 9
.Os
.Sh NAME
.Nm smr_read_enter ,
.Nm smr_read_leave ,
.Nm smr_init ,
.Nm smr_call ,
.Nm smr_barrier ,
.Nm smr_flush ,
.Nm SMR_ASSERT_CRITICAL ,
.Nm SMR_ASSERT_NONCRITICAL
.Nd safe memory reclamation
.Sh SYNOPSIS
.In sys/smr.h
.Ft void
.Fn smr_read_enter
.Ft void
.Fn smr_read_leave
.Ft void
.Fn smr_init "struct smr_entry *smr"
.Ft void
.Fn smr_call "struct smr_entry *smr" "void (*fn)(void *)" "void *arg"
.Ft void
.Fn smr_barrier
.Ft void
.Fn smr_flush
.Ft void
.Fn SMR_ASSERT_CRITICAL
.Ft void
.Fn SMR_ASSERT_NONCRITICAL
.Sh DESCRIPTION
The safe memory reclamation API provides a mechanism for reclaiming
shared objects that readers can access without locking.
Objects that are reclaimed through SMR are called SMR-protected.
The mechanism guarantees that SMR-protected objects are not destroyed
while readers are using them.
However, it does not control how these objects are modified.
.Pp
Readers access SMR-protected objects inside an SMR read-side critical section
using
.Xr SMR_PTR_GET 9 .
.\" or with a higher level API built on top of it.
The section is entered with
.Fn smr_read_enter ,
and exited with
.Fn smr_read_leave .
These routines never block.
Sleeping is not allowed within SMR read-side critical section.
.Pp
.Fn smr_init
initializes the entry
.Fa smr
for use with
.Fn smr_call .
.Pp
.Fn smr_call
schedules a callback to be invoked after the entry
.Fa smr
cannot be referenced by a reader in SMR read-side critical section.
On invocation, the system calls function
.Fa fn
with argument
.Fa arg
in process context without any locks held.
The implementation may delay the call in order to reduce
overall system overhead by amortization.
.Pp
.Fn smr_barrier
sleeps until any SMR read-side critical sections that are active on other CPUs
at the time of invocation have ended.
Like with
.Fn smr_call ,
the processing of the request may be delayed.
.Pp
.Fn smr_flush
is like
.Fn smr_barrier
but the system is forced to process the request as soon as possible.
The use of this function is discouraged because of the heavy impact
on system performance.
.Pp
To avoid deadlocks, the caller of
.Fn smr_barrier
or
.Fn smr_flush
must not hold locks that can block the processing of SMR callbacks.
.Pp
The SMR implementation does not limit the number of deferred calls.
It is important to prevent arbitrary call rate of
.Fn smr_call .
Otherwise, it might be possible to exhaust system resources
if the system is not able to invoke callbacks quickly enough.
.Pp
.Fn SMR_ASSERT_CRITICAL
and
.Fn SMR_ASSERT_NONCRITICAL
can be used to assert that the current CPU is or is not
in SMR read-side critical section.
.Sh CONTEXT
.Fn smr_read_enter ,
.Fn smr_read_leave ,
.Fn smr_call
and
.Fn smr_init
can be called during autoconf, from process context, or from interrupt context.
.Pp
.Fn smr_barrier
and
.Fn smr_flush
can be called from process context.
.Sh SEE ALSO
.Xr mutex 9 ,
.Xr rwlock 9 ,
.Xr SMR_LIST_INIT 9 ,
.Xr SMR_PTR_GET 9
.Sh HISTORY
The SMR API first appeared in
.Ox 6.5 .