d/freebsd-skq
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
8925dffab1
freebsd-skq/sys/kern/subr_sleepqueue.c

1229 lines
32 KiB
C
Raw Normal View History

/* -> /*- for copyright notices, minor format tweaks as necessary
2005-01-06 23:35:40 +00:00
/*-
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
* Copyright (c) 2004 John Baldwin <jhb@FreeBSD.org>
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*/
/*
* Implementation of sleep queues used to hold queue of threads blocked on
* a wait channel. Sleep queues different from turnstiles in that wait
* channels are not owned by anyone, so there is no priority propagation.
* Sleep queues can also provide a timeout and can also be interrupted by
* signals. That said, there are several similarities between the turnstile
* and sleep queue implementations. (Note: turnstiles were implemented
* first.) For example, both use a hash table of the same size where each
* bucket is referred to as a "chain" that contains both a spin lock and
* a linked list of queues. An individual queue is located by using a hash
* to pick a chain, locking the chain, and then walking the chain searching
* for the queue. This means that a wait channel object does not need to
* embed it's queue head just as locks do not embed their turnstile queue
* head. Threads also carry around a sleep queue that they lend to the
* wait channel when blocking. Just as in turnstiles, the queue includes
* a free list of the sleep queues of other threads blocked on the same
* wait channel in the case of multiple waiters.
*
* Some additional functionality provided by sleep queues include the
* ability to set a timeout. The timeout is managed using a per-thread
* callout that resumes a thread if it is asleep. A thread may also
* catch signals while it is asleep (aka an interruptible sleep). The
* signal code uses sleepq_abort() to interrupt a sleeping thread. Finally,
* sleep queues also provide some extra assertions. One is not allowed to
* mix the sleep/wakeup and cv APIs for a given wait channel. Also, one
* must consistently use the same lock to synchronize with a wait channel,
* though this check is currently only a warning for sleep/wakeup due to
* pre-existing abuse of that API. The same lock must also be held when
* awakening threads, though that is currently only enforced for condition
* variables.
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
Add a new ddb command 'show sleepq'. It takes a wait channel as an argument and looks for a sleep queue associated with that wait channel. If it finds one it will display information such as the list of threads sleeping on that queue. If it can't find a sleep queue for that wait channel, then it will see if that address matches any of the active sleep queues. If so, it will display information about the sleepq at the specified address.
2006-01-27 22:24:07 +00:00
#include "opt_sleepqueue_profiling.h"
#include "opt_ddb.h"
- Include opt_sched.h for SCHED_STATS.
2007-06-12 23:27:31 +00:00
#include "opt_sched.h"
Add a new ddb command 'show sleepq'. It takes a wait channel as an argument and looks for a sleep queue associated with that wait channel. If it finds one it will display information such as the list of threads sleeping on that queue. If it can't find a sleep queue for that wait channel, then it will see if that address matches any of the active sleep queues. If so, it will display information about the sleepq at the specified address.
2006-01-27 22:24:07 +00:00
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/lock.h>
#include <sys/kernel.h>
#include <sys/ktr.h>
#include <sys/mutex.h>
#include <sys/proc.h>
- Add a facility similar to LOCK_PROFILING under SLEEPQUEUE_PROFILING. Keep a simple (wmesg, count) tuple in a hash to keep track of how many times we sleep at each wait message. We hash on message and not channel. No line number information is given as typically wait messages are not used in more than one place. Identical strings defined at different addresses will show up with seperate counters. - Use debug.sleepq.enable to enable, .reset to reset, and .stats dumps stats. - Do an unsynchronized check in sleepq_switch() prior to switching before calling sleepq_profile() which uses a global lock to synchronize the hash. Only sleeps which actually cause a context switch are counted.
2008-03-19 07:22:07 +00:00
#include <sys/sbuf.h>
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
#include <sys/sched.h>
Implement the DTrace sched provider. This implementation aims to be compatible with the sched provider implemented by Solaris and its open- source derivatives. Full documentation of the sched provider can be found on Oracle's DTrace wiki pages. Note that for compatibility with scripts originally written for Solaris, serveral probes are defined that will never fire. These probes are defined to fire when Solaris-specific features perform certain actions. As these features are not present in FreeBSD, the probes can never fire. Also, I have added a two probes that are not defined in Solaris, lend-pri and load-change. These probes have been added to make it possible to collect schedgraph data with DTrace. Finally, a few probes are defined in Solaris to take a cpuinfo_t * argument. As it was not immediately clear to me how to translate that to FreeBSD, currently those probes are passed NULL in place of a cpuinfo_t *. Sponsored by: Sandvine Incorporated MFC after: 2 weeks
2012-05-15 01:30:25 +00:00
#include <sys/sdt.h>
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
#include <sys/signalvar.h>
#include <sys/sleepqueue.h>
Add two new kernel options to allow rudimentary profiling of the internal hash tables used in the sleep queue and turnstile code. Each option adds a sysctl tree under debug containing the maximum depth of any bucket in the hash table as well as a separate node for each bucket (or chain) containing the current depth and maximum depth for that bucket.
2004-06-29 02:30:12 +00:00
#include <sys/sysctl.h>
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
- Convert turnstiles and sleepqueus to use UMA. This provides a modest speedup and will be more useful after each gains a spinlock in the impending thread_lock() commit. - Move initialization and asserts into init/fini routines. fini routines are only needed in the INVARIANTS case for now. Submitted by: Attilio Rao <attilio@FreeBSD.org> Tested by: kris, jeff
2007-05-18 06:32:24 +00:00
#include <vm/uma.h>
Add a new ddb command 'show sleepq'. It takes a wait channel as an argument and looks for a sleep queue associated with that wait channel. If it finds one it will display information such as the list of threads sleeping on that queue. If it can't find a sleep queue for that wait channel, then it will see if that address matches any of the active sleep queues. If so, it will display information about the sleepq at the specified address.
2006-01-27 22:24:07 +00:00
#ifdef DDB
#include <ddb/ddb.h>
#endif
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
/*
Make kern_nanosleep() and pause_sbt() to use per-CPU sleep queues. This removes significant sleep queue lock congestion on multithreaded microbenchmarks, making them scale to multiple CPUs almost linearly.
2013-03-12 06:58:49 +00:00
* Constants for the hash table of sleep queue chains.
* SC_TABLESIZE must be a power of two for SC_MASK to work properly.
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
*/
Make kern_nanosleep() and pause_sbt() to use per-CPU sleep queues. This removes significant sleep queue lock congestion on multithreaded microbenchmarks, making them scale to multiple CPUs almost linearly.
2013-03-12 06:58:49 +00:00
#define SC_TABLESIZE 256 /* Must be power of 2. */
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
#define SC_MASK (SC_TABLESIZE - 1)
#define SC_SHIFT 8
Make kern_nanosleep() and pause_sbt() to use per-CPU sleep queues. This removes significant sleep queue lock congestion on multithreaded microbenchmarks, making them scale to multiple CPUs almost linearly.
2013-03-12 06:58:49 +00:00
#define SC_HASH(wc) ((((uintptr_t)(wc) >> SC_SHIFT) ^ (uintptr_t)(wc)) & \
SC_MASK)
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
#define SC_LOOKUP(wc) &sleepq_chains[SC_HASH(wc)]
Add second sleep queue so that sx and lockmgr can have separate sleep queues for shared and exclusive acquisitions Submitted by: Attilio Rao Approved by: jhb
2006-12-16 06:54:09 +00:00
#define NR_SLEEPQS 2
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
/*
* There two different lists of sleep queues. Both lists are connected
* via the sq_hash entries. The first list is the sleep queue chain list
* that a sleep queue is on when it is attached to a wait channel. The
* second list is the free list hung off of a sleep queue that is attached
* to a wait channel.
*
* Each sleep queue also contains the wait channel it is attached to, the
* list of threads blocked on that wait channel, flags specific to the
* wait channel, and the lock used to synchronize with a wait channel.
* The flags are used to catch mismatches between the various consumers
* of the sleep queue API (e.g. sleep/wakeup and condition variables).
* The lock pointer is only used when invariants are enabled for various
* debugging checks.
*
* Locking key:
* c - sleep queue chain lock
*/
struct sleepqueue {
Add second sleep queue so that sx and lockmgr can have separate sleep queues for shared and exclusive acquisitions Submitted by: Attilio Rao Approved by: jhb
2006-12-16 06:54:09 +00:00
TAILQ_HEAD(, thread) sq_blocked[NR_SLEEPQS]; /* (c) Blocked threads. */
In current code, threads performing an interruptible sleep (on both sxlock, via the sx_{s, x}lock_sig() interface, or plain lockmgr), will leave the waiters flag on forcing the owner to do a wakeup even when if the waiter queue is empty. That operation may lead to a deadlock in the case of doing a fake wakeup on the "preferred" (based on the wakeup algorithm) queue while the other queue has real waiters on it, because nobody is going to wakeup the 2nd queue waiters and they will sleep indefinitively. A similar bug, is present, for lockmgr in the case the waiters are sleeping with LK_SLEEPFAIL on. In this case, even if the waiters queue is not empty, the waiters won't progress after being awake but they will just fail, still not taking care of the 2nd queue waiters (as instead the lock owned doing the wakeup would expect). In order to fix this bug in a cheap way (without adding too much locking and complicating too much the semantic) add a sleepqueue interface which does report the actual number of waiters on a specified queue of a waitchannel (sleepq_sleepcnt()) and use it in order to determine if the exclusive waiters (or shared waiters) are actually present on the lockmgr (or sx) before to give them precedence in the wakeup algorithm. This fix alone, however doesn't solve the LK_SLEEPFAIL bug. In order to cope with it, add the tracking of how many exclusive LK_SLEEPFAIL waiters a lockmgr has and if all the waiters on the exclusive waiters queue are LK_SLEEPFAIL just wake both queues. The sleepq_sleepcnt() introduction and ABI breakage require __FreeBSD_version bumping. Reported by: avg, kib, pho Reviewed by: kib Tested by: pho
2009-12-12 21:31:07 +00:00
u_int sq_blockedcnt[NR_SLEEPQS]; /* (c) N. of blocked threads. */
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
LIST_ENTRY(sleepqueue) sq_hash; /* (c) Chain and free list. */
LIST_HEAD(, sleepqueue) sq_free; /* (c) Free queues. */
void *sq_wchan; /* (c) Wait channel. */
Refine the turnstile and sleep queue interfaces just a bit: - Add a new _lock() call to each API that locks the associated chain lock for a lock_object pointer or wait channel. The _lookup() functions now require that the chain lock be locked via _lock() when they are called. - Change sleepq_add(), turnstile_wait() and turnstile_claim() to lookup the associated queue structure internally via _lookup() rather than accepting a pointer from the caller. For turnstiles, this means that the actual lookup of the turnstile in the hash table is only done when the thread actually blocks rather than being done on each loop iteration in _mtx_lock_sleep(). For sleep queues, this means that sleepq_lookup() is no longer used outside of the sleep queue code except to implement an assertion in cv_destroy(). - Change sleepq_broadcast() and sleepq_signal() to require that the chain lock is already required. For condition variables, this lets the cv_broadcast() and cv_signal() functions lock the sleep queue chain lock while testing the waiters count. This means that the waiters count internal to condition variables is no longer protected by the interlock mutex and cv_broadcast() and cv_signal() now no longer require that the interlock be held when they are called. This lets consumers of condition variables drop the lock before waking other threads which can result in fewer context switches. MFC after: 1 month
2004-10-12 18:36:20 +00:00
int sq_type; /* (c) Queue type. */
Introduce the new kernel thread called "deadlock resolver". While the name is pretentious, a good explanation of its targets is reported in this 17 months old presentation e-mail: http://lists.freebsd.org/pipermail/freebsd-arch/2008-August/008452.html In order to implement it, the sq_type in sleepqueues is mandatory and not only compiled along with INVARIANTS option. Additively, a new sleepqueue function, sleepq_type() is added, returning the type of the sleepqueue linked to a wchan. Three new sysctls are added in order to configure the thread: debug.deadlkres.slptime_threshold debug.deadlkres.blktime_threshold debug.deadlkres.sleepfreq rappresenting the thresholds for sleep and block time that will lead to a deadlock matching (when exceeded), while the sleepfreq rappresents the number of seconds between 2 consecutive thread runnings. In order to enable the deadlock resolver thread recompile your kernel with the option DEADLKRES. Reviewed by: jeff Tested by: pho, Giovanni Trematerra Sponsored by: Nokia Incorporated, Sandvine Incorporated MFC after: 2 weeks
2010-01-09 01:46:38 +00:00
#ifdef INVARIANTS
Change sleepq_add(9) argument from 'struct mtx *' to 'struct lock_object *', which allows to use it with different kinds of locks. For example it allows to implement Solaris conditions variables which will be used in ZFS port on top of sx(9) locks. Reviewed by: jhb
2006-11-16 01:02:00 +00:00
struct lock_object *sq_lock; /* (c) Associated lock. */
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
#endif
};
struct sleepqueue_chain {
LIST_HEAD(, sleepqueue) sc_queues; /* List of sleep queues. */
struct mtx sc_lock; /* Spin lock for this chain. */
Add two new kernel options to allow rudimentary profiling of the internal hash tables used in the sleep queue and turnstile code. Each option adds a sysctl tree under debug containing the maximum depth of any bucket in the hash table as well as a separate node for each bucket (or chain) containing the current depth and maximum depth for that bucket.
2004-06-29 02:30:12 +00:00
#ifdef SLEEPQUEUE_PROFILING
u_int sc_depth; /* Length of sc_queues. */
u_int sc_max_depth; /* Max length of sc_queues. */
#endif
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
};
Add two new kernel options to allow rudimentary profiling of the internal hash tables used in the sleep queue and turnstile code. Each option adds a sysctl tree under debug containing the maximum depth of any bucket in the hash table as well as a separate node for each bucket (or chain) containing the current depth and maximum depth for that bucket.
2004-06-29 02:30:12 +00:00
#ifdef SLEEPQUEUE_PROFILING
u_int sleepq_max_depth;
Mark all SYSCTL_NODEs static that have no corresponding SYSCTL_DECLs. The SYSCTL_NODE macro defines a list that stores all child-elements of that node. If there's no SYSCTL_DECL macro anywhere else, there's no reason why it shouldn't be static.
2011-11-07 15:43:11 +00:00
static SYSCTL_NODE(_debug, OID_AUTO, sleepq, CTLFLAG_RD, 0, "sleepq profiling");
static SYSCTL_NODE(_debug_sleepq, OID_AUTO, chains, CTLFLAG_RD, 0,
Add two new kernel options to allow rudimentary profiling of the internal hash tables used in the sleep queue and turnstile code. Each option adds a sysctl tree under debug containing the maximum depth of any bucket in the hash table as well as a separate node for each bucket (or chain) containing the current depth and maximum depth for that bucket.
2004-06-29 02:30:12 +00:00
"sleepq chain stats");
SYSCTL_UINT(_debug_sleepq, OID_AUTO, max_depth, CTLFLAG_RD, &sleepq_max_depth,
0, "maxmimum depth achieved of a single chain");
- Add a facility similar to LOCK_PROFILING under SLEEPQUEUE_PROFILING. Keep a simple (wmesg, count) tuple in a hash to keep track of how many times we sleep at each wait message. We hash on message and not channel. No line number information is given as typically wait messages are not used in more than one place. Identical strings defined at different addresses will show up with seperate counters. - Use debug.sleepq.enable to enable, .reset to reset, and .stats dumps stats. - Do an unsynchronized check in sleepq_switch() prior to switching before calling sleepq_profile() which uses a global lock to synchronize the hash. Only sleeps which actually cause a context switch are counted.
2008-03-19 07:22:07 +00:00
static void sleepq_profile(const char *wmesg);
static int prof_enabled;
Add two new kernel options to allow rudimentary profiling of the internal hash tables used in the sleep queue and turnstile code. Each option adds a sysctl tree under debug containing the maximum depth of any bucket in the hash table as well as a separate node for each bucket (or chain) containing the current depth and maximum depth for that bucket.
2004-06-29 02:30:12 +00:00
#endif
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
static struct sleepqueue_chain sleepq_chains[SC_TABLESIZE];
- Convert turnstiles and sleepqueus to use UMA. This provides a modest speedup and will be more useful after each gains a spinlock in the impending thread_lock() commit. - Move initialization and asserts into init/fini routines. fini routines are only needed in the INVARIANTS case for now. Submitted by: Attilio Rao <attilio@FreeBSD.org> Tested by: kris, jeff
2007-05-18 06:32:24 +00:00
static uma_zone_t sleepq_zone;
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
/*
* Prototypes for non-exported routines.
*/
- Pass the priority argument from *sleep() into sleepq and down into sched_sleep(). This removes extra thread_lock() acquisition and allows the scheduler to decide what to do with the static boost. - Change the priority arguments to cv_* to match sleepq/msleep/etc. where 0 means no priority change. Catch -1 in cv_broadcastpri() and convert it to 0 for now. - Set a flag when sleeping in a way that is compatible with swapping since direct priority comparisons are meaningless now. - Add a sysctl to ule, kern.sched.static_boost, that defaults to on which controls the boost behavior. Turning it off gives better performance in some workloads but needs more investigation. - While we're modifying sleepq, change signal and broadcast to both return with the lock held as the lock was held on enter. Reviewed by: jhb, peter
2008-03-12 06:31:06 +00:00
static int sleepq_catch_signals(void *wchan, int pri);
Add second sleep queue so that sx and lockmgr can have separate sleep queues for shared and exclusive acquisitions Submitted by: Attilio Rao Approved by: jhb
2006-12-16 06:54:09 +00:00
static int sleepq_check_signals(void);
Major callout subsystem cleanup and rewrite: - Close a migration race where callout_reset() failed to set the CALLOUT_ACTIVE flag. - Callout callback functions are now allowed to be protected by spinlocks. - Switching the callout CPU number cannot always be done on a per-callout basis. See the updated timeout(9) manual page for more information. - The timeout(9) manual page has been updated to reflect how all the functions inside the callout API are working. The manual page has been made function oriented to make it easier to deduce how each of the functions making up the callout API are working without having to first read the whole manual page. Group all functions into a handful of sections which should give a quick top-level overview when the different functions should be used. - The CALLOUT_SHAREDLOCK flag and its functionality has been removed to reduce the complexity in the callout code and to avoid problems about atomically stopping callouts via callout_stop(). If someone needs it, it can be re-added. From my quick grep there are no CALLOUT_SHAREDLOCK clients in the kernel. - A new callout API function named "callout_drain_async()" has been added. See the updated timeout(9) manual page for a complete description. - Update the callout clients in the "kern/" folder to use the callout API properly, like cv_timedwait(). Previously there was some custom sleepqueue code in the callout subsystem, which has been removed, because we now allow callouts to be protected by spinlocks. This allows us to tear down the callout like done with regular mutexes, and a "td_slpmutex" has been added to "struct thread" to atomically teardown the "td_slpcallout". Further the "TDF_TIMOFAIL" and "SWT_SLEEPQTIMO" states can now be completely removed. Currently they are marked as available and will be cleaned up in a follow up commit. - Bump the __FreeBSD_version to indicate kernel modules need recompilation. - There has been several reports that this patch "seems to squash a serious bug leading to a callout timeout and panic". Kernel build testing: all architectures were built MFC after: 2 weeks Differential Revision: https://reviews.freebsd.org/D1438 Sponsored by: Mellanox Technologies Reviewed by: jhb, adrian, sbruno and emaste
2015-01-15 15:32:30 +00:00
static int sleepq_check_timeout(struct thread *);
static void sleepq_stop_timeout(struct thread *);
- Convert turnstiles and sleepqueus to use UMA. This provides a modest speedup and will be more useful after each gains a spinlock in the impending thread_lock() commit. - Move initialization and asserts into init/fini routines. fini routines are only needed in the INVARIANTS case for now. Submitted by: Attilio Rao <attilio@FreeBSD.org> Tested by: kris, jeff
2007-05-18 06:32:24 +00:00
#ifdef INVARIANTS
static void sleepq_dtor(void *mem, int size, void *arg);
#endif
static int sleepq_init(void *mem, int size, int flags);
If a thread that is swapped out is made runnable, then the setrunnable() routine wakes up proc0 so that proc0 can swap the thread back in. Historically, this has been done by waking up proc0 directly from setrunnable() itself via a wakeup(). When waking up a sleeping thread that was swapped out (the usual case when waking proc0 since only sleeping threads are eligible to be swapped out), this resulted in a bit of recursion (e.g. wakeup() -> setrunnable() -> wakeup()). With sleep queues having separate locks in 6.x and later, this caused a spin lock LOR (sleepq lock -> sched_lock/thread lock -> sleepq lock). An attempt was made to fix this in 7.0 by making the proc0 wakeup use the ithread mechanism for doing the wakeup. However, this required grabbing proc0's thread lock to perform the wakeup. If proc0 was asleep elsewhere in the kernel (e.g. waiting for disk I/O), then this degenerated into the same LOR since the thread lock would be some other sleepq lock. Fix this by deferring the wakeup of the swapper until after the sleepq lock held by the upper layer has been locked. The setrunnable() routine now returns a boolean value to indicate whether or not proc0 needs to be woken up. The end result is that consumers of the sleepq API such as *sleep/wakeup, condition variables, sx locks, and lockmgr, have to wakeup proc0 if they get a non-zero return value from sleepq_abort(), sleepq_broadcast(), or sleepq_signal(). Discussed with: jeff Glanced at by: sam Tested by: Jurgen Weber jurgen - ish com au MFC after: 2 weeks
2008-08-05 20:02:31 +00:00
static int sleepq_resume_thread(struct sleepqueue *sq, struct thread *td,
- Convert turnstiles and sleepqueus to use UMA. This provides a modest speedup and will be more useful after each gains a spinlock in the impending thread_lock() commit. - Move initialization and asserts into init/fini routines. fini routines are only needed in the INVARIANTS case for now. Submitted by: Attilio Rao <attilio@FreeBSD.org> Tested by: kris, jeff
2007-05-18 06:32:24 +00:00
int pri);
- Pass the priority argument from *sleep() into sleepq and down into sched_sleep(). This removes extra thread_lock() acquisition and allows the scheduler to decide what to do with the static boost. - Change the priority arguments to cv_* to match sleepq/msleep/etc. where 0 means no priority change. Catch -1 in cv_broadcastpri() and convert it to 0 for now. - Set a flag when sleeping in a way that is compatible with swapping since direct priority comparisons are meaningless now. - Add a sysctl to ule, kern.sched.static_boost, that defaults to on which controls the boost behavior. Turning it off gives better performance in some workloads but needs more investigation. - While we're modifying sleepq, change signal and broadcast to both return with the lock held as the lock was held on enter. Reviewed by: jhb, peter
2008-03-12 06:31:06 +00:00
static void sleepq_switch(void *wchan, int pri);
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
static void sleepq_timeout(void *arg);
Implement the DTrace sched provider. This implementation aims to be compatible with the sched provider implemented by Solaris and its open- source derivatives. Full documentation of the sched provider can be found on Oracle's DTrace wiki pages. Note that for compatibility with scripts originally written for Solaris, serveral probes are defined that will never fire. These probes are defined to fire when Solaris-specific features perform certain actions. As these features are not present in FreeBSD, the probes can never fire. Also, I have added a two probes that are not defined in Solaris, lend-pri and load-change. These probes have been added to make it possible to collect schedgraph data with DTrace. Finally, a few probes are defined in Solaris to take a cpuinfo_t * argument. As it was not immediately clear to me how to translate that to FreeBSD, currently those probes are passed NULL in place of a cpuinfo_t *. Sponsored by: Sandvine Incorporated MFC after: 2 weeks
2012-05-15 01:30:25 +00:00
SDT_PROBE_DECLARE(sched, , , sleep);
SDT_PROBE_DECLARE(sched, , , wakeup);
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
/*
sysctl subsystem uses sxlocks so avoid to setup dynamic sysctl nodes before sleepinit() has been fully executed in the SLEEPQUEUE_PROFILING case. Sponsored by: EMC / Isilon storage division
2014-06-24 15:16:55 +00:00
* Initialize SLEEPQUEUE_PROFILING specific sysctl nodes.
* Note that it must happen after sleepinit() has been fully executed, so
* it must happen after SI_SUB_KMEM SYSINIT() subsystem setup.
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
*/
Add two new kernel options to allow rudimentary profiling of the internal hash tables used in the sleep queue and turnstile code. Each option adds a sysctl tree under debug containing the maximum depth of any bucket in the hash table as well as a separate node for each bucket (or chain) containing the current depth and maximum depth for that bucket.
2004-06-29 02:30:12 +00:00
#ifdef SLEEPQUEUE_PROFILING
sysctl subsystem uses sxlocks so avoid to setup dynamic sysctl nodes before sleepinit() has been fully executed in the SLEEPQUEUE_PROFILING case. Sponsored by: EMC / Isilon storage division
2014-06-24 15:16:55 +00:00
static void
init_sleepqueue_profiling(void)
{
Add two new kernel options to allow rudimentary profiling of the internal hash tables used in the sleep queue and turnstile code. Each option adds a sysctl tree under debug containing the maximum depth of any bucket in the hash table as well as a separate node for each bucket (or chain) containing the current depth and maximum depth for that bucket.
2004-06-29 02:30:12 +00:00
char chain_name[10];
sysctl subsystem uses sxlocks so avoid to setup dynamic sysctl nodes before sleepinit() has been fully executed in the SLEEPQUEUE_PROFILING case. Sponsored by: EMC / Isilon storage division
2014-06-24 15:16:55 +00:00
struct sysctl_oid *chain_oid;
u_int i;
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
for (i = 0; i < SC_TABLESIZE; i++) {
sysctl subsystem uses sxlocks so avoid to setup dynamic sysctl nodes before sleepinit() has been fully executed in the SLEEPQUEUE_PROFILING case. Sponsored by: EMC / Isilon storage division
2014-06-24 15:16:55 +00:00
snprintf(chain_name, sizeof(chain_name), "%u", i);
Add two new kernel options to allow rudimentary profiling of the internal hash tables used in the sleep queue and turnstile code. Each option adds a sysctl tree under debug containing the maximum depth of any bucket in the hash table as well as a separate node for each bucket (or chain) containing the current depth and maximum depth for that bucket.
2004-06-29 02:30:12 +00:00
chain_oid = SYSCTL_ADD_NODE(NULL,
SYSCTL_STATIC_CHILDREN(_debug_sleepq_chains), OID_AUTO,
chain_name, CTLFLAG_RD, NULL, "sleepq chain stats");
SYSCTL_ADD_UINT(NULL, SYSCTL_CHILDREN(chain_oid), OID_AUTO,
"depth", CTLFLAG_RD, &sleepq_chains[i].sc_depth, 0, NULL);
SYSCTL_ADD_UINT(NULL, SYSCTL_CHILDREN(chain_oid), OID_AUTO,
"max_depth", CTLFLAG_RD, &sleepq_chains[i].sc_max_depth, 0,
NULL);
sysctl subsystem uses sxlocks so avoid to setup dynamic sysctl nodes before sleepinit() has been fully executed in the SLEEPQUEUE_PROFILING case. Sponsored by: EMC / Isilon storage division
2014-06-24 15:16:55 +00:00
}
}
SYSINIT(sleepqueue_profiling, SI_SUB_LOCK, SI_ORDER_ANY,
init_sleepqueue_profiling, NULL);
Add two new kernel options to allow rudimentary profiling of the internal hash tables used in the sleep queue and turnstile code. Each option adds a sysctl tree under debug containing the maximum depth of any bucket in the hash table as well as a separate node for each bucket (or chain) containing the current depth and maximum depth for that bucket.
2004-06-29 02:30:12 +00:00
#endif
sysctl subsystem uses sxlocks so avoid to setup dynamic sysctl nodes before sleepinit() has been fully executed in the SLEEPQUEUE_PROFILING case. Sponsored by: EMC / Isilon storage division
2014-06-24 15:16:55 +00:00
/*
* Early initialization of sleep queues that is called from the sleepinit()
* SYSINIT.
*/
void
init_sleepqueues(void)
{
int i;
for (i = 0; i < SC_TABLESIZE; i++) {
LIST_INIT(&sleepq_chains[i].sc_queues);
mtx_init(&sleepq_chains[i].sc_lock, "sleepq chain", NULL,
MTX_SPIN | MTX_RECURSE);
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
}
- Convert turnstiles and sleepqueus to use UMA. This provides a modest speedup and will be more useful after each gains a spinlock in the impending thread_lock() commit. - Move initialization and asserts into init/fini routines. fini routines are only needed in the INVARIANTS case for now. Submitted by: Attilio Rao <attilio@FreeBSD.org> Tested by: kris, jeff
2007-05-18 06:32:24 +00:00
sleepq_zone = uma_zcreate("SLEEPQUEUE", sizeof(struct sleepqueue),
#ifdef INVARIANTS
NULL, sleepq_dtor, sleepq_init, NULL, UMA_ALIGN_CACHE, 0);
#else
NULL, NULL, sleepq_init, NULL, UMA_ALIGN_CACHE, 0);
#endif
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
thread0.td_sleepqueue = sleepq_alloc();
}
/*
- Convert turnstiles and sleepqueus to use UMA. This provides a modest speedup and will be more useful after each gains a spinlock in the impending thread_lock() commit. - Move initialization and asserts into init/fini routines. fini routines are only needed in the INVARIANTS case for now. Submitted by: Attilio Rao <attilio@FreeBSD.org> Tested by: kris, jeff
2007-05-18 06:32:24 +00:00
* Get a sleep queue for a new thread.
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
*/
struct sleepqueue *
sleepq_alloc(void)
{
- Convert turnstiles and sleepqueus to use UMA. This provides a modest speedup and will be more useful after each gains a spinlock in the impending thread_lock() commit. - Move initialization and asserts into init/fini routines. fini routines are only needed in the INVARIANTS case for now. Submitted by: Attilio Rao <attilio@FreeBSD.org> Tested by: kris, jeff
2007-05-18 06:32:24 +00:00
return (uma_zalloc(sleepq_zone, M_WAITOK));
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
}
/*
* Free a sleep queue when a thread is destroyed.
*/
void
sleepq_free(struct sleepqueue *sq)
{
- Convert turnstiles and sleepqueus to use UMA. This provides a modest speedup and will be more useful after each gains a spinlock in the impending thread_lock() commit. - Move initialization and asserts into init/fini routines. fini routines are only needed in the INVARIANTS case for now. Submitted by: Attilio Rao <attilio@FreeBSD.org> Tested by: kris, jeff
2007-05-18 06:32:24 +00:00
uma_zfree(sleepq_zone, sq);
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
}
Refine the turnstile and sleep queue interfaces just a bit: - Add a new _lock() call to each API that locks the associated chain lock for a lock_object pointer or wait channel. The _lookup() functions now require that the chain lock be locked via _lock() when they are called. - Change sleepq_add(), turnstile_wait() and turnstile_claim() to lookup the associated queue structure internally via _lookup() rather than accepting a pointer from the caller. For turnstiles, this means that the actual lookup of the turnstile in the hash table is only done when the thread actually blocks rather than being done on each loop iteration in _mtx_lock_sleep(). For sleep queues, this means that sleepq_lookup() is no longer used outside of the sleep queue code except to implement an assertion in cv_destroy(). - Change sleepq_broadcast() and sleepq_signal() to require that the chain lock is already required. For condition variables, this lets the cv_broadcast() and cv_signal() functions lock the sleep queue chain lock while testing the waiters count. This means that the waiters count internal to condition variables is no longer protected by the interlock mutex and cv_broadcast() and cv_signal() now no longer require that the interlock be held when they are called. This lets consumers of condition variables drop the lock before waking other threads which can result in fewer context switches. MFC after: 1 month
2004-10-12 18:36:20 +00:00
/*
* Lock the sleep queue chain associated with the specified wait channel.
*/
void
sleepq_lock(void *wchan)
{
struct sleepqueue_chain *sc;
sc = SC_LOOKUP(wchan);
mtx_lock_spin(&sc->sc_lock);
}
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
/*
* Look up the sleep queue associated with a given wait channel in the hash
Refine the turnstile and sleep queue interfaces just a bit: - Add a new _lock() call to each API that locks the associated chain lock for a lock_object pointer or wait channel. The _lookup() functions now require that the chain lock be locked via _lock() when they are called. - Change sleepq_add(), turnstile_wait() and turnstile_claim() to lookup the associated queue structure internally via _lookup() rather than accepting a pointer from the caller. For turnstiles, this means that the actual lookup of the turnstile in the hash table is only done when the thread actually blocks rather than being done on each loop iteration in _mtx_lock_sleep(). For sleep queues, this means that sleepq_lookup() is no longer used outside of the sleep queue code except to implement an assertion in cv_destroy(). - Change sleepq_broadcast() and sleepq_signal() to require that the chain lock is already required. For condition variables, this lets the cv_broadcast() and cv_signal() functions lock the sleep queue chain lock while testing the waiters count. This means that the waiters count internal to condition variables is no longer protected by the interlock mutex and cv_broadcast() and cv_signal() now no longer require that the interlock be held when they are called. This lets consumers of condition variables drop the lock before waking other threads which can result in fewer context switches. MFC after: 1 month
2004-10-12 18:36:20 +00:00
* table locking the associated sleep queue chain. If no queue is found in
* the table, NULL is returned.
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
*/
struct sleepqueue *
sleepq_lookup(void *wchan)
{
struct sleepqueue_chain *sc;
struct sleepqueue *sq;
KASSERT(wchan != NULL, ("%s: invalid NULL wait channel", __func__));
sc = SC_LOOKUP(wchan);
Refine the turnstile and sleep queue interfaces just a bit: - Add a new _lock() call to each API that locks the associated chain lock for a lock_object pointer or wait channel. The _lookup() functions now require that the chain lock be locked via _lock() when they are called. - Change sleepq_add(), turnstile_wait() and turnstile_claim() to lookup the associated queue structure internally via _lookup() rather than accepting a pointer from the caller. For turnstiles, this means that the actual lookup of the turnstile in the hash table is only done when the thread actually blocks rather than being done on each loop iteration in _mtx_lock_sleep(). For sleep queues, this means that sleepq_lookup() is no longer used outside of the sleep queue code except to implement an assertion in cv_destroy(). - Change sleepq_broadcast() and sleepq_signal() to require that the chain lock is already required. For condition variables, this lets the cv_broadcast() and cv_signal() functions lock the sleep queue chain lock while testing the waiters count. This means that the waiters count internal to condition variables is no longer protected by the interlock mutex and cv_broadcast() and cv_signal() now no longer require that the interlock be held when they are called. This lets consumers of condition variables drop the lock before waking other threads which can result in fewer context switches. MFC after: 1 month
2004-10-12 18:36:20 +00:00
mtx_assert(&sc->sc_lock, MA_OWNED);
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
LIST_FOREACH(sq, &sc->sc_queues, sq_hash)
if (sq->sq_wchan == wchan)
return (sq);
return (NULL);
}
/*
* Unlock the sleep queue chain associated with a given wait channel.
*/
void
sleepq_release(void *wchan)
{
struct sleepqueue_chain *sc;
sc = SC_LOOKUP(wchan);
mtx_unlock_spin(&sc->sc_lock);
}
/*
- Store threads on sleep queues in FIFO order rather than sorted by priority. The sleep queues don't get updated when the priority of threads changes, so sleepq_signal() might not always wakeup the highest priority thread. Updating the queues when thread priorities change cannot be easily done due to lock orders, so instead we do an O(n) walk of the queue for a sleepq_signal() operation instead of O(1). On the other hand, adding a thread to a sleep queue now goes from O(n) to O(1) so it ends up as an even tradeoff. The correctness here with regards to priorities is actually fairly important. msleep() gives interactive threads their priority "boost" after they are placed on the queue, but before this fix that "boost" wasn't used to determine the highest priority thread that sleepq_signal() awoke. - Fix up some comments. Inspired by: ups, bde
2004-11-05 20:19:58 +00:00
* Places the current thread on the sleep queue for the specified wait
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
* channel. If INVARIANTS is enabled, then it associates the passed in
* lock with the sleepq to make sure it is held when that sleep queue is
* woken up.
*/
void
Add second sleep queue so that sx and lockmgr can have separate sleep queues for shared and exclusive acquisitions Submitted by: Attilio Rao Approved by: jhb
2006-12-16 06:54:09 +00:00
sleepq_add(void *wchan, struct lock_object *lock, const char *wmesg, int flags,
int queue)
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
{
struct sleepqueue_chain *sc;
Refine the turnstile and sleep queue interfaces just a bit: - Add a new _lock() call to each API that locks the associated chain lock for a lock_object pointer or wait channel. The _lookup() functions now require that the chain lock be locked via _lock() when they are called. - Change sleepq_add(), turnstile_wait() and turnstile_claim() to lookup the associated queue structure internally via _lookup() rather than accepting a pointer from the caller. For turnstiles, this means that the actual lookup of the turnstile in the hash table is only done when the thread actually blocks rather than being done on each loop iteration in _mtx_lock_sleep(). For sleep queues, this means that sleepq_lookup() is no longer used outside of the sleep queue code except to implement an assertion in cv_destroy(). - Change sleepq_broadcast() and sleepq_signal() to require that the chain lock is already required. For condition variables, this lets the cv_broadcast() and cv_signal() functions lock the sleep queue chain lock while testing the waiters count. This means that the waiters count internal to condition variables is no longer protected by the interlock mutex and cv_broadcast() and cv_signal() now no longer require that the interlock be held when they are called. This lets consumers of condition variables drop the lock before waking other threads which can result in fewer context switches. MFC after: 1 month
2004-10-12 18:36:20 +00:00
struct sleepqueue *sq;
- Store threads on sleep queues in FIFO order rather than sorted by priority. The sleep queues don't get updated when the priority of threads changes, so sleepq_signal() might not always wakeup the highest priority thread. Updating the queues when thread priorities change cannot be easily done due to lock orders, so instead we do an O(n) walk of the queue for a sleepq_signal() operation instead of O(1). On the other hand, adding a thread to a sleep queue now goes from O(n) to O(1) so it ends up as an even tradeoff. The correctness here with regards to priorities is actually fairly important. msleep() gives interactive threads their priority "boost" after they are placed on the queue, but before this fix that "boost" wasn't used to determine the highest priority thread that sleepq_signal() awoke. - Fix up some comments. Inspired by: ups, bde
2004-11-05 20:19:58 +00:00
struct thread *td;
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
td = curthread;
sc = SC_LOOKUP(wchan);
mtx_assert(&sc->sc_lock, MA_OWNED);
MPASS(td->td_sleepqueue != NULL);
MPASS(wchan != NULL);
Add second sleep queue so that sx and lockmgr can have separate sleep queues for shared and exclusive acquisitions Submitted by: Attilio Rao Approved by: jhb
2006-12-16 06:54:09 +00:00
MPASS((queue >= 0) && (queue < NR_SLEEPQS));
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
- Add a new simple facility for marking the current thread as being in a state where sleeping on a sleep queue is not allowed. The facility doesn't support recursion but uses a simple private per-thread flag (TDP_NOSLEEPING). The sleepq_add() function will panic if the flag is set and INVARIANTS is enabled. - Use this new facility to replace the g_xup and g_xdown mutexes that were (ab)used to achieve similar behavior. - Disallow sleeping in interrupt threads when invoking interrupt handlers. MFC after: 1 week Reviewed by: phk
2005-09-15 19:05:37 +00:00
/* If this thread is not allowed to sleep, die a horrible death. */
Replace the TDP_NOSLEEPING flag with a counter so that the THREAD_NO_SLEEPING() and THREAD_SLEEPING_OK() macros can nest. Reviewed by: attilio
2013-03-01 22:03:31 +00:00
KASSERT(td->td_no_sleeping == 0,
("%s: td %p to sleep on wchan %p with sleeping prohibited",
Tweak the commit message in case of panic for sleeping from threads with TDP_NOSLEEPING on. The current message has no informations on the thread and wchan involed, which may be useful in case where dumps have mangled dwarf informations. Reported by: kib Reviewed by: bde, jhb, kib MFC after: 1 week
2012-09-12 22:05:54 +00:00
__func__, td, wchan));
- Add a new simple facility for marking the current thread as being in a state where sleeping on a sleep queue is not allowed. The facility doesn't support recursion but uses a simple private per-thread flag (TDP_NOSLEEPING). The sleepq_add() function will panic if the flag is set and INVARIANTS is enabled. - Use this new facility to replace the g_xup and g_xdown mutexes that were (ab)used to achieve similar behavior. - Disallow sleeping in interrupt threads when invoking interrupt handlers. MFC after: 1 week Reviewed by: phk
2005-09-15 19:05:37 +00:00
Refine the turnstile and sleep queue interfaces just a bit: - Add a new _lock() call to each API that locks the associated chain lock for a lock_object pointer or wait channel. The _lookup() functions now require that the chain lock be locked via _lock() when they are called. - Change sleepq_add(), turnstile_wait() and turnstile_claim() to lookup the associated queue structure internally via _lookup() rather than accepting a pointer from the caller. For turnstiles, this means that the actual lookup of the turnstile in the hash table is only done when the thread actually blocks rather than being done on each loop iteration in _mtx_lock_sleep(). For sleep queues, this means that sleepq_lookup() is no longer used outside of the sleep queue code except to implement an assertion in cv_destroy(). - Change sleepq_broadcast() and sleepq_signal() to require that the chain lock is already required. For condition variables, this lets the cv_broadcast() and cv_signal() functions lock the sleep queue chain lock while testing the waiters count. This means that the waiters count internal to condition variables is no longer protected by the interlock mutex and cv_broadcast() and cv_signal() now no longer require that the interlock be held when they are called. This lets consumers of condition variables drop the lock before waking other threads which can result in fewer context switches. MFC after: 1 month
2004-10-12 18:36:20 +00:00
/* Look up the sleep queue associated with the wait channel 'wchan'. */
sq = sleepq_lookup(wchan);
/*
* If the wait channel does not already have a sleep queue, use
* this thread's sleep queue. Otherwise, insert the current thread
* into the sleep queue already in use by this wait channel.
*/
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
if (sq == NULL) {
Add second sleep queue so that sx and lockmgr can have separate sleep queues for shared and exclusive acquisitions Submitted by: Attilio Rao Approved by: jhb
2006-12-16 06:54:09 +00:00
#ifdef INVARIANTS
Cleaner fix for handling declaration of loop variable under INVARIANTS - in trying to avoid nested brackets and #ifdef INVARIANTS around i at the top, I broke booting for INVARIANTS all together :-( - the cleanest fix is to simply assign to sq twice if INVARIANTS is enabled - tested both with and without INVARIANTS :-/
2006-12-17 00:14:20 +00:00
int i;
Don't intermix assignments and variable declarations in prev. commit
2006-12-16 21:17:27 +00:00
Cleaner fix for handling declaration of loop variable under INVARIANTS - in trying to avoid nested brackets and #ifdef INVARIANTS around i at the top, I broke booting for INVARIANTS all together :-( - the cleanest fix is to simply assign to sq twice if INVARIANTS is enabled - tested both with and without INVARIANTS :-/
2006-12-17 00:14:20 +00:00
sq = td->td_sleepqueue;
In current code, threads performing an interruptible sleep (on both sxlock, via the sx_{s, x}lock_sig() interface, or plain lockmgr), will leave the waiters flag on forcing the owner to do a wakeup even when if the waiter queue is empty. That operation may lead to a deadlock in the case of doing a fake wakeup on the "preferred" (based on the wakeup algorithm) queue while the other queue has real waiters on it, because nobody is going to wakeup the 2nd queue waiters and they will sleep indefinitively. A similar bug, is present, for lockmgr in the case the waiters are sleeping with LK_SLEEPFAIL on. In this case, even if the waiters queue is not empty, the waiters won't progress after being awake but they will just fail, still not taking care of the 2nd queue waiters (as instead the lock owned doing the wakeup would expect). In order to fix this bug in a cheap way (without adding too much locking and complicating too much the semantic) add a sleepqueue interface which does report the actual number of waiters on a specified queue of a waitchannel (sleepq_sleepcnt()) and use it in order to determine if the exclusive waiters (or shared waiters) are actually present on the lockmgr (or sx) before to give them precedence in the wakeup algorithm. This fix alone, however doesn't solve the LK_SLEEPFAIL bug. In order to cope with it, add the tracking of how many exclusive LK_SLEEPFAIL waiters a lockmgr has and if all the waiters on the exclusive waiters queue are LK_SLEEPFAIL just wake both queues. The sleepq_sleepcnt() introduction and ABI breakage require __FreeBSD_version bumping. Reported by: avg, kib, pho Reviewed by: kib Tested by: pho
2009-12-12 21:31:07 +00:00
for (i = 0; i < NR_SLEEPQS; i++) {
Cleaner fix for handling declaration of loop variable under INVARIANTS - in trying to avoid nested brackets and #ifdef INVARIANTS around i at the top, I broke booting for INVARIANTS all together :-( - the cleanest fix is to simply assign to sq twice if INVARIANTS is enabled - tested both with and without INVARIANTS :-/
2006-12-17 00:14:20 +00:00
KASSERT(TAILQ_EMPTY(&sq->sq_blocked[i]),
In current code, threads performing an interruptible sleep (on both sxlock, via the sx_{s, x}lock_sig() interface, or plain lockmgr), will leave the waiters flag on forcing the owner to do a wakeup even when if the waiter queue is empty. That operation may lead to a deadlock in the case of doing a fake wakeup on the "preferred" (based on the wakeup algorithm) queue while the other queue has real waiters on it, because nobody is going to wakeup the 2nd queue waiters and they will sleep indefinitively. A similar bug, is present, for lockmgr in the case the waiters are sleeping with LK_SLEEPFAIL on. In this case, even if the waiters queue is not empty, the waiters won't progress after being awake but they will just fail, still not taking care of the 2nd queue waiters (as instead the lock owned doing the wakeup would expect). In order to fix this bug in a cheap way (without adding too much locking and complicating too much the semantic) add a sleepqueue interface which does report the actual number of waiters on a specified queue of a waitchannel (sleepq_sleepcnt()) and use it in order to determine if the exclusive waiters (or shared waiters) are actually present on the lockmgr (or sx) before to give them precedence in the wakeup algorithm. This fix alone, however doesn't solve the LK_SLEEPFAIL bug. In order to cope with it, add the tracking of how many exclusive LK_SLEEPFAIL waiters a lockmgr has and if all the waiters on the exclusive waiters queue are LK_SLEEPFAIL just wake both queues. The sleepq_sleepcnt() introduction and ABI breakage require __FreeBSD_version bumping. Reported by: avg, kib, pho Reviewed by: kib Tested by: pho
2009-12-12 21:31:07 +00:00
("thread's sleep queue %d is not empty", i));
KASSERT(sq->sq_blockedcnt[i] == 0,
("thread's sleep queue %d count mismatches", i));
}
Add second sleep queue so that sx and lockmgr can have separate sleep queues for shared and exclusive acquisitions Submitted by: Attilio Rao Approved by: jhb
2006-12-16 06:54:09 +00:00
KASSERT(LIST_EMPTY(&sq->sq_free),
("thread's sleep queue has a non-empty free list"));
KASSERT(sq->sq_wchan == NULL, ("stale sq_wchan pointer"));
Cleaner fix for handling declaration of loop variable under INVARIANTS - in trying to avoid nested brackets and #ifdef INVARIANTS around i at the top, I broke booting for INVARIANTS all together :-( - the cleanest fix is to simply assign to sq twice if INVARIANTS is enabled - tested both with and without INVARIANTS :-/
2006-12-17 00:14:20 +00:00
sq->sq_lock = lock;
Add second sleep queue so that sx and lockmgr can have separate sleep queues for shared and exclusive acquisitions Submitted by: Attilio Rao Approved by: jhb
2006-12-16 06:54:09 +00:00
#endif
Add two new kernel options to allow rudimentary profiling of the internal hash tables used in the sleep queue and turnstile code. Each option adds a sysctl tree under debug containing the maximum depth of any bucket in the hash table as well as a separate node for each bucket (or chain) containing the current depth and maximum depth for that bucket.
2004-06-29 02:30:12 +00:00
#ifdef SLEEPQUEUE_PROFILING
sc->sc_depth++;
if (sc->sc_depth > sc->sc_max_depth) {
sc->sc_max_depth = sc->sc_depth;
if (sc->sc_max_depth > sleepq_max_depth)
sleepq_max_depth = sc->sc_max_depth;
}
#endif
Cleaner fix for handling declaration of loop variable under INVARIANTS - in trying to avoid nested brackets and #ifdef INVARIANTS around i at the top, I broke booting for INVARIANTS all together :-( - the cleanest fix is to simply assign to sq twice if INVARIANTS is enabled - tested both with and without INVARIANTS :-/
2006-12-17 00:14:20 +00:00
sq = td->td_sleepqueue;
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
LIST_INSERT_HEAD(&sc->sc_queues, sq, sq_hash);
sq->sq_wchan = wchan;
Introduce the new kernel thread called "deadlock resolver". While the name is pretentious, a good explanation of its targets is reported in this 17 months old presentation e-mail: http://lists.freebsd.org/pipermail/freebsd-arch/2008-August/008452.html In order to implement it, the sq_type in sleepqueues is mandatory and not only compiled along with INVARIANTS option. Additively, a new sleepqueue function, sleepq_type() is added, returning the type of the sleepqueue linked to a wchan. Three new sysctls are added in order to configure the thread: debug.deadlkres.slptime_threshold debug.deadlkres.blktime_threshold debug.deadlkres.sleepfreq rappresenting the thresholds for sleep and block time that will lead to a deadlock matching (when exceeded), while the sleepfreq rappresents the number of seconds between 2 consecutive thread runnings. In order to enable the deadlock resolver thread recompile your kernel with the option DEADLKRES. Reviewed by: jeff Tested by: pho, Giovanni Trematerra Sponsored by: Nokia Incorporated, Sandvine Incorporated MFC after: 2 weeks
2010-01-09 01:46:38 +00:00
sq->sq_type = flags & SLEEPQ_TYPE;
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
} else {
MPASS(wchan == sq->sq_wchan);
Always assert that the passed in lock is the same as the saved lock in the sleep queue now that the one abnormal case has been fixed.
2004-03-02 15:02:08 +00:00
MPASS(lock == sq->sq_lock);
Refine the turnstile and sleep queue interfaces just a bit: - Add a new _lock() call to each API that locks the associated chain lock for a lock_object pointer or wait channel. The _lookup() functions now require that the chain lock be locked via _lock() when they are called. - Change sleepq_add(), turnstile_wait() and turnstile_claim() to lookup the associated queue structure internally via _lookup() rather than accepting a pointer from the caller. For turnstiles, this means that the actual lookup of the turnstile in the hash table is only done when the thread actually blocks rather than being done on each loop iteration in _mtx_lock_sleep(). For sleep queues, this means that sleepq_lookup() is no longer used outside of the sleep queue code except to implement an assertion in cv_destroy(). - Change sleepq_broadcast() and sleepq_signal() to require that the chain lock is already required. For condition variables, this lets the cv_broadcast() and cv_signal() functions lock the sleep queue chain lock while testing the waiters count. This means that the waiters count internal to condition variables is no longer protected by the interlock mutex and cv_broadcast() and cv_signal() now no longer require that the interlock be held when they are called. This lets consumers of condition variables drop the lock before waking other threads which can result in fewer context switches. MFC after: 1 month
2004-10-12 18:36:20 +00:00
MPASS((flags & SLEEPQ_TYPE) == sq->sq_type);
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
LIST_INSERT_HEAD(&sq->sq_free, td->td_sleepqueue, sq_hash);
}
subr_sleepqueue.c presents a thread lock missing which leads to dangerous races for some struct thread members. More specifically, this bug seems responsible for some memory dumping problems people were experiencing. Fix this adding correct thread locking. Tested by: rwatson Submitted by: tegge Approved by: jeff Approved by: re
2007-09-13 09:12:36 +00:00
thread_lock(td);
Add second sleep queue so that sx and lockmgr can have separate sleep queues for shared and exclusive acquisitions Submitted by: Attilio Rao Approved by: jhb
2006-12-16 06:54:09 +00:00
TAILQ_INSERT_TAIL(&sq->sq_blocked[queue], td, td_slpq);
In current code, threads performing an interruptible sleep (on both sxlock, via the sx_{s, x}lock_sig() interface, or plain lockmgr), will leave the waiters flag on forcing the owner to do a wakeup even when if the waiter queue is empty. That operation may lead to a deadlock in the case of doing a fake wakeup on the "preferred" (based on the wakeup algorithm) queue while the other queue has real waiters on it, because nobody is going to wakeup the 2nd queue waiters and they will sleep indefinitively. A similar bug, is present, for lockmgr in the case the waiters are sleeping with LK_SLEEPFAIL on. In this case, even if the waiters queue is not empty, the waiters won't progress after being awake but they will just fail, still not taking care of the 2nd queue waiters (as instead the lock owned doing the wakeup would expect). In order to fix this bug in a cheap way (without adding too much locking and complicating too much the semantic) add a sleepqueue interface which does report the actual number of waiters on a specified queue of a waitchannel (sleepq_sleepcnt()) and use it in order to determine if the exclusive waiters (or shared waiters) are actually present on the lockmgr (or sx) before to give them precedence in the wakeup algorithm. This fix alone, however doesn't solve the LK_SLEEPFAIL bug. In order to cope with it, add the tracking of how many exclusive LK_SLEEPFAIL waiters a lockmgr has and if all the waiters on the exclusive waiters queue are LK_SLEEPFAIL just wake both queues. The sleepq_sleepcnt() introduction and ABI breakage require __FreeBSD_version bumping. Reported by: avg, kib, pho Reviewed by: kib Tested by: pho
2009-12-12 21:31:07 +00:00
sq->sq_blockedcnt[queue]++;
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
td->td_sleepqueue = NULL;
Add second sleep queue so that sx and lockmgr can have separate sleep queues for shared and exclusive acquisitions Submitted by: Attilio Rao Approved by: jhb
2006-12-16 06:54:09 +00:00
td->td_sqqueue = queue;
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
td->td_wchan = wchan;
td->td_wmesg = wmesg;
Fix a long standing race between sleep queue and thread suspension code. When a thread A is going to sleep, it calls sleepq_catch_signals() to detect any pending signals or thread suspension request, if nothing happens, it returns without holding process lock or scheduler lock, this opens a race window which allows thread B to come in and do process suspension work, however since A is still at running state, thread B can do nothing to A, thread A continues, and puts itself into actually sleeping state, but B has never seen it, and it sits there forever until B is woken up by other threads sometimes later(this can be very long delay or never happen). Fix this bug by forcing sleepq_catch_signals to return with scheduler lock held. Fix sleepq_abort() by passing it an interrupted code, previously, it worked as wakeup_one(), and the interruption can not be identified correctly by sleep queue code when the sleeping thread is resumed. Let thread_suspend_check() returns EINTR or ERESTART, so sleep queue no longer has to use SIGSTOP as a hack to build a return value. Reviewed by: jhb MFC after: 1 week
2006-02-15 23:52:01 +00:00
if (flags & SLEEPQ_INTERRUPTIBLE) {
Now that the return value semantics of cv's for multithreaded processes have been unified with that of msleep(9), further refine the sleepq interface and consolidate some duplicated code: - Move the pre-sleep checks for theaded processes into a thread_sleep_check() function in kern_thread.c. - Move all handling of TDF_SINTR to be internal to subr_sleepqueue.c. Specifically, if a thread is awakened by something other than a signal while checking for signals before going to sleep, clear TDF_SINTR in sleepq_catch_signals(). This removes a sched_lock lock/unlock combo in that edge case during an interruptible sleep. Also, fix sleepq_check_signals() to properly handle the condition if TDF_SINTR is clear rather than requiring the callers of the sleepq API to notice this edge case and call a non-_sig variant of sleepq_wait(). - Clarify the flags arguments to sleepq_add(), sleepq_signal() and sleepq_broadcast() by creating an explicit submask for sleepq types. Also, add an explicit SLEEPQ_MSLEEP type rather than a magic number of 0. Also, add a SLEEPQ_INTERRUPTIBLE flag for use with sleepq_add() and move the setting of TDF_SINTR to sleepq_add() if this flag is set rather than sleepq_catch_signals(). Note that it is the caller's responsibility to ensure that sleepq_catch_signals() is called if and only if this flag is passed to the preceeding sleepq_add(). Note that this also removes a sched_lock lock/unlock pair from sleepq_catch_signals(). It also ensures that for an interruptible sleep, TDF_SINTR is always set when TD_ON_SLEEPQ() is true.
2004-08-19 11:31:42 +00:00
td->td_flags |= TDF_SINTR;
Fix a long standing race between sleep queue and thread suspension code. When a thread A is going to sleep, it calls sleepq_catch_signals() to detect any pending signals or thread suspension request, if nothing happens, it returns without holding process lock or scheduler lock, this opens a race window which allows thread B to come in and do process suspension work, however since A is still at running state, thread B can do nothing to A, thread A continues, and puts itself into actually sleeping state, but B has never seen it, and it sits there forever until B is woken up by other threads sometimes later(this can be very long delay or never happen). Fix this bug by forcing sleepq_catch_signals to return with scheduler lock held. Fix sleepq_abort() by passing it an interrupted code, previously, it worked as wakeup_one(), and the interruption can not be identified correctly by sleep queue code when the sleeping thread is resumed. Let thread_suspend_check() returns EINTR or ERESTART, so sleep queue no longer has to use SIGSTOP as a hack to build a return value. Reviewed by: jhb MFC after: 1 week
2006-02-15 23:52:01 +00:00
td->td_flags &= ~TDF_SLEEPABORT;
}
subr_sleepqueue.c presents a thread lock missing which leads to dangerous races for some struct thread members. More specifically, this bug seems responsible for some memory dumping problems people were experiencing. Fix this adding correct thread locking. Tested by: rwatson Submitted by: tegge Approved by: jeff Approved by: re
2007-09-13 09:12:36 +00:00
thread_unlock(td);
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
}
/*
* Sets a timeout that will remove the current thread from the specified
* sleep queue after timo ticks if the thread has not already been awakened.
*/
void
MFcalloutng: Convert sleepqueue(9) bits to the new callout KPI. Take advantage of the possibility to run callback directly from hw interrupt context. Sponsored by: Google Summer of Code 2012, iXsystems inc. Tested by: flo, marius, ian, markj, Fabian Keil
2013-03-04 11:51:46 +00:00
sleepq_set_timeout_sbt(void *wchan, sbintime_t sbt, sbintime_t pr,
int flags)
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
{
struct thread *td;
td = curthread;
Major callout subsystem cleanup and rewrite: - Close a migration race where callout_reset() failed to set the CALLOUT_ACTIVE flag. - Callout callback functions are now allowed to be protected by spinlocks. - Switching the callout CPU number cannot always be done on a per-callout basis. See the updated timeout(9) manual page for more information. - The timeout(9) manual page has been updated to reflect how all the functions inside the callout API are working. The manual page has been made function oriented to make it easier to deduce how each of the functions making up the callout API are working without having to first read the whole manual page. Group all functions into a handful of sections which should give a quick top-level overview when the different functions should be used. - The CALLOUT_SHAREDLOCK flag and its functionality has been removed to reduce the complexity in the callout code and to avoid problems about atomically stopping callouts via callout_stop(). If someone needs it, it can be re-added. From my quick grep there are no CALLOUT_SHAREDLOCK clients in the kernel. - A new callout API function named "callout_drain_async()" has been added. See the updated timeout(9) manual page for a complete description. - Update the callout clients in the "kern/" folder to use the callout API properly, like cv_timedwait(). Previously there was some custom sleepqueue code in the callout subsystem, which has been removed, because we now allow callouts to be protected by spinlocks. This allows us to tear down the callout like done with regular mutexes, and a "td_slpmutex" has been added to "struct thread" to atomically teardown the "td_slpcallout". Further the "TDF_TIMOFAIL" and "SWT_SLEEPQTIMO" states can now be completely removed. Currently they are marked as available and will be cleaned up in a follow up commit. - Bump the __FreeBSD_version to indicate kernel modules need recompilation. - There has been several reports that this patch "seems to squash a serious bug leading to a callout timeout and panic". Kernel build testing: all architectures were built MFC after: 2 weeks Differential Revision: https://reviews.freebsd.org/D1438 Sponsored by: Mellanox Technologies Reviewed by: jhb, adrian, sbruno and emaste
2015-01-15 15:32:30 +00:00
mtx_lock_spin(&td->td_slpmutex);
MFcalloutng: Convert sleepqueue(9) bits to the new callout KPI. Take advantage of the possibility to run callback directly from hw interrupt context. Sponsored by: Google Summer of Code 2012, iXsystems inc. Tested by: flo, marius, ian, markj, Fabian Keil
2013-03-04 11:51:46 +00:00
callout_reset_sbt_on(&td->td_slpcallout, sbt, pr,
sleepq_timeout, td, PCPU_GET(cpuid), flags | C_DIRECT_EXEC);
Major callout subsystem cleanup and rewrite: - Close a migration race where callout_reset() failed to set the CALLOUT_ACTIVE flag. - Callout callback functions are now allowed to be protected by spinlocks. - Switching the callout CPU number cannot always be done on a per-callout basis. See the updated timeout(9) manual page for more information. - The timeout(9) manual page has been updated to reflect how all the functions inside the callout API are working. The manual page has been made function oriented to make it easier to deduce how each of the functions making up the callout API are working without having to first read the whole manual page. Group all functions into a handful of sections which should give a quick top-level overview when the different functions should be used. - The CALLOUT_SHAREDLOCK flag and its functionality has been removed to reduce the complexity in the callout code and to avoid problems about atomically stopping callouts via callout_stop(). If someone needs it, it can be re-added. From my quick grep there are no CALLOUT_SHAREDLOCK clients in the kernel. - A new callout API function named "callout_drain_async()" has been added. See the updated timeout(9) manual page for a complete description. - Update the callout clients in the "kern/" folder to use the callout API properly, like cv_timedwait(). Previously there was some custom sleepqueue code in the callout subsystem, which has been removed, because we now allow callouts to be protected by spinlocks. This allows us to tear down the callout like done with regular mutexes, and a "td_slpmutex" has been added to "struct thread" to atomically teardown the "td_slpcallout". Further the "TDF_TIMOFAIL" and "SWT_SLEEPQTIMO" states can now be completely removed. Currently they are marked as available and will be cleaned up in a follow up commit. - Bump the __FreeBSD_version to indicate kernel modules need recompilation. - There has been several reports that this patch "seems to squash a serious bug leading to a callout timeout and panic". Kernel build testing: all architectures were built MFC after: 2 weeks Differential Revision: https://reviews.freebsd.org/D1438 Sponsored by: Mellanox Technologies Reviewed by: jhb, adrian, sbruno and emaste
2015-01-15 15:32:30 +00:00
mtx_unlock_spin(&td->td_slpmutex);
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
}
In current code, threads performing an interruptible sleep (on both sxlock, via the sx_{s, x}lock_sig() interface, or plain lockmgr), will leave the waiters flag on forcing the owner to do a wakeup even when if the waiter queue is empty. That operation may lead to a deadlock in the case of doing a fake wakeup on the "preferred" (based on the wakeup algorithm) queue while the other queue has real waiters on it, because nobody is going to wakeup the 2nd queue waiters and they will sleep indefinitively. A similar bug, is present, for lockmgr in the case the waiters are sleeping with LK_SLEEPFAIL on. In this case, even if the waiters queue is not empty, the waiters won't progress after being awake but they will just fail, still not taking care of the 2nd queue waiters (as instead the lock owned doing the wakeup would expect). In order to fix this bug in a cheap way (without adding too much locking and complicating too much the semantic) add a sleepqueue interface which does report the actual number of waiters on a specified queue of a waitchannel (sleepq_sleepcnt()) and use it in order to determine if the exclusive waiters (or shared waiters) are actually present on the lockmgr (or sx) before to give them precedence in the wakeup algorithm. This fix alone, however doesn't solve the LK_SLEEPFAIL bug. In order to cope with it, add the tracking of how many exclusive LK_SLEEPFAIL waiters a lockmgr has and if all the waiters on the exclusive waiters queue are LK_SLEEPFAIL just wake both queues. The sleepq_sleepcnt() introduction and ABI breakage require __FreeBSD_version bumping. Reported by: avg, kib, pho Reviewed by: kib Tested by: pho
2009-12-12 21:31:07 +00:00
/*
* Return the number of actual sleepers for the specified queue.
*/
u_int
sleepq_sleepcnt(void *wchan, int queue)
{
struct sleepqueue *sq;
KASSERT(wchan != NULL, ("%s: invalid NULL wait channel", __func__));
MPASS((queue >= 0) && (queue < NR_SLEEPQS));
sq = sleepq_lookup(wchan);
if (sq == NULL)
return (0);
return (sq->sq_blockedcnt[queue]);
}
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
/*
* Marks the pending sleep of the current thread as interruptible and
* makes an initial check for pending signals before putting a thread
Commit 2/14 of sched_lock decomposition. - Adapt sleepqueues to the new thread_lock() mechanism. - Delay assigning the sleep queue spinlock as the thread lock until after we've checked for signals. It is illegal for a thread to return in mi_switch() with any lock assigned to td_lock other than the scheduler locks. - Change sleepq_catch_signals() to do the switch if necessary to simplify the callers. - Simplify timeout handling now that locking a sleeping thread has the side-effect of locking the sleepqueue. Some previous races are no longer possible. Tested by: kris, current@ Tested on: i386, amd64, ULE, 4BSD, libthr, libkse, PREEMPTION, etc. Discussed with: kris, attilio, kmacy, jhb, julian, bde (small parts each)
2007-06-04 23:50:56 +00:00
* to sleep. Enters and exits with the thread lock held. Thread lock
* may have transitioned from the sleepq lock to a run lock.
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
*/
Fix a long standing race between sleep queue and thread suspension code. When a thread A is going to sleep, it calls sleepq_catch_signals() to detect any pending signals or thread suspension request, if nothing happens, it returns without holding process lock or scheduler lock, this opens a race window which allows thread B to come in and do process suspension work, however since A is still at running state, thread B can do nothing to A, thread A continues, and puts itself into actually sleeping state, but B has never seen it, and it sits there forever until B is woken up by other threads sometimes later(this can be very long delay or never happen). Fix this bug by forcing sleepq_catch_signals to return with scheduler lock held. Fix sleepq_abort() by passing it an interrupted code, previously, it worked as wakeup_one(), and the interruption can not be identified correctly by sleep queue code when the sleeping thread is resumed. Let thread_suspend_check() returns EINTR or ERESTART, so sleep queue no longer has to use SIGSTOP as a hack to build a return value. Reviewed by: jhb MFC after: 1 week
2006-02-15 23:52:01 +00:00
static int
- Pass the priority argument from *sleep() into sleepq and down into sched_sleep(). This removes extra thread_lock() acquisition and allows the scheduler to decide what to do with the static boost. - Change the priority arguments to cv_* to match sleepq/msleep/etc. where 0 means no priority change. Catch -1 in cv_broadcastpri() and convert it to 0 for now. - Set a flag when sleeping in a way that is compatible with swapping since direct priority comparisons are meaningless now. - Add a sysctl to ule, kern.sched.static_boost, that defaults to on which controls the boost behavior. Turning it off gives better performance in some workloads but needs more investigation. - While we're modifying sleepq, change signal and broadcast to both return with the lock held as the lock was held on enter. Reviewed by: jhb, peter
2008-03-12 06:31:06 +00:00
sleepq_catch_signals(void *wchan, int pri)
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
{
struct sleepqueue_chain *sc;
struct sleepqueue *sq;
struct thread *td;
struct proc *p;
Fix a long standing race between sleep queue and thread suspension code. When a thread A is going to sleep, it calls sleepq_catch_signals() to detect any pending signals or thread suspension request, if nothing happens, it returns without holding process lock or scheduler lock, this opens a race window which allows thread B to come in and do process suspension work, however since A is still at running state, thread B can do nothing to A, thread A continues, and puts itself into actually sleeping state, but B has never seen it, and it sits there forever until B is woken up by other threads sometimes later(this can be very long delay or never happen). Fix this bug by forcing sleepq_catch_signals to return with scheduler lock held. Fix sleepq_abort() by passing it an interrupted code, previously, it worked as wakeup_one(), and the interruption can not be identified correctly by sleep queue code when the sleeping thread is resumed. Let thread_suspend_check() returns EINTR or ERESTART, so sleep queue no longer has to use SIGSTOP as a hack to build a return value. Reviewed by: jhb MFC after: 1 week
2006-02-15 23:52:01 +00:00
struct sigacts *ps;
Partially revert r195702. Deferring stops is now implemented via a set of calls to toggle TDF_SBDRY rather than passing PBDRY to individual sleep calls. - Remove the stop_allowed parameters from cursig() and issignal(). issignal() checks TDF_SBDRY directly. - Remove the PBDRY and SLEEPQ_STOP_ON_BDRY flags.
2013-03-18 17:23:58 +00:00
int sig, ret;
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
td = curthread;
Fix a long standing race between sleep queue and thread suspension code. When a thread A is going to sleep, it calls sleepq_catch_signals() to detect any pending signals or thread suspension request, if nothing happens, it returns without holding process lock or scheduler lock, this opens a race window which allows thread B to come in and do process suspension work, however since A is still at running state, thread B can do nothing to A, thread A continues, and puts itself into actually sleeping state, but B has never seen it, and it sits there forever until B is woken up by other threads sometimes later(this can be very long delay or never happen). Fix this bug by forcing sleepq_catch_signals to return with scheduler lock held. Fix sleepq_abort() by passing it an interrupted code, previously, it worked as wakeup_one(), and the interruption can not be identified correctly by sleep queue code when the sleeping thread is resumed. Let thread_suspend_check() returns EINTR or ERESTART, so sleep queue no longer has to use SIGSTOP as a hack to build a return value. Reviewed by: jhb MFC after: 1 week
2006-02-15 23:52:01 +00:00
p = curproc;
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
sc = SC_LOOKUP(wchan);
mtx_assert(&sc->sc_lock, MA_OWNED);
MPASS(wchan != NULL);
If thread set a TDP_WAKEUP for itself, clears the flag and returns EINTR immediately, this is used for implementing reliable pthread cancellation.
2010-08-20 04:28:30 +00:00
if ((td->td_pflags & TDP_WAKEUP) != 0) {
td->td_pflags &= ~TDP_WAKEUP;
ret = EINTR;
make sure thread lock is locked.
2010-08-20 23:51:34 +00:00
thread_lock(td);
If thread set a TDP_WAKEUP for itself, clears the flag and returns EINTR immediately, this is used for implementing reliable pthread cancellation.
2010-08-20 04:28:30 +00:00
goto out;
}
- At the top of sleepq_catch_signals() lock the thread and check TDF_NEEDSIGCHK before doing the very expensive cursig() and related locking. NEEDSIGCHK is updated whenever our signal mask change or when a signal is delivered and should be sufficient to avoid the more expensive tests. This eliminates another source of PROC_LOCK contention in multithreaded programs.
2008-03-19 07:35:14 +00:00
/*
* See if there are any pending signals for this thread. If not
* we can switch immediately. Otherwise do the signal processing
* directly.
*/
thread_lock(td);
- Add a new td flag TDF_NEEDSUSPCHK that is set whenever a thread needs to enter thread_suspend_check(). - Set TDF_ASTPENDING along with TDF_NEEDSUSPCHK so we can move the thread_suspend_check() to ast() rather than userret(). - Check TDF_NEEDSUSPCHK in the sleepq_catch_signals() optimization so that we don't miss a suspend request. If this is set use the expensive signal path. - Set NEEDSUSPCHK when creating a new thread in thr in case the creating thread is due to be suspended as well but has not yet. Reviewed by: davidxu (Authored original patch)
2008-03-21 08:23:25 +00:00
if ((td->td_flags & (TDF_NEEDSIGCHK | TDF_NEEDSUSPCHK)) == 0) {
- At the top of sleepq_catch_signals() lock the thread and check TDF_NEEDSIGCHK before doing the very expensive cursig() and related locking. NEEDSIGCHK is updated whenever our signal mask change or when a signal is delivered and should be sufficient to avoid the more expensive tests. This eliminates another source of PROC_LOCK contention in multithreaded programs.
2008-03-19 07:35:14 +00:00
sleepq_switch(wchan, pri);
return (0);
}
thread_unlock(td);
mtx_unlock_spin(&sc->sc_lock);
Fixed some common printf format errors. Don't assume that "struct foo *" is "void *" (it isn't) or that the default promotion of pid_t is int. Instead, assume that casting "struct foo *" to "void *" and printing the result with %p is useful, and that all pid_t's are representable as longs. Fixed some minor style bugs (mainly spelling errors in comments).
2004-05-14 20:51:42 +00:00
CTR3(KTR_PROC, "sleepq catching signals: thread %p (pid %ld, %s)",
A bunch more files that should probably print out a thread name instead of a process name.
2007-11-14 06:51:33 +00:00
(void *)td, (long)p->p_pid, td->td_name);
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
PROC_LOCK(p);
Fix a long standing race between sleep queue and thread suspension code. When a thread A is going to sleep, it calls sleepq_catch_signals() to detect any pending signals or thread suspension request, if nothing happens, it returns without holding process lock or scheduler lock, this opens a race window which allows thread B to come in and do process suspension work, however since A is still at running state, thread B can do nothing to A, thread A continues, and puts itself into actually sleeping state, but B has never seen it, and it sits there forever until B is woken up by other threads sometimes later(this can be very long delay or never happen). Fix this bug by forcing sleepq_catch_signals to return with scheduler lock held. Fix sleepq_abort() by passing it an interrupted code, previously, it worked as wakeup_one(), and the interruption can not be identified correctly by sleep queue code when the sleeping thread is resumed. Let thread_suspend_check() returns EINTR or ERESTART, so sleep queue no longer has to use SIGSTOP as a hack to build a return value. Reviewed by: jhb MFC after: 1 week
2006-02-15 23:52:01 +00:00
ps = p->p_sigacts;
mtx_lock(&ps->ps_mtx);
Partially revert r195702. Deferring stops is now implemented via a set of calls to toggle TDF_SBDRY rather than passing PBDRY to individual sleep calls. - Remove the stop_allowed parameters from cursig() and issignal(). issignal() checks TDF_SBDRY directly. - Remove the PBDRY and SLEEPQ_STOP_ON_BDRY flags.
2013-03-18 17:23:58 +00:00
sig = cursig(td);
Fix a long standing race between sleep queue and thread suspension code. When a thread A is going to sleep, it calls sleepq_catch_signals() to detect any pending signals or thread suspension request, if nothing happens, it returns without holding process lock or scheduler lock, this opens a race window which allows thread B to come in and do process suspension work, however since A is still at running state, thread B can do nothing to A, thread A continues, and puts itself into actually sleeping state, but B has never seen it, and it sits there forever until B is woken up by other threads sometimes later(this can be very long delay or never happen). Fix this bug by forcing sleepq_catch_signals to return with scheduler lock held. Fix sleepq_abort() by passing it an interrupted code, previously, it worked as wakeup_one(), and the interruption can not be identified correctly by sleep queue code when the sleeping thread is resumed. Let thread_suspend_check() returns EINTR or ERESTART, so sleep queue no longer has to use SIGSTOP as a hack to build a return value. Reviewed by: jhb MFC after: 1 week
2006-02-15 23:52:01 +00:00
if (sig == 0) {
mtx_unlock(&ps->ps_mtx);
ret = thread_suspend_check(1);
MPASS(ret == 0 || ret == EINTR || ret == ERESTART);
} else {
if (SIGISMEMBER(ps->ps_sigintr, sig))
ret = EINTR;
else
ret = ERESTART;
mtx_unlock(&ps->ps_mtx);
}
Revert rev 184216 and 184199, due to the way the thread_lock works, it may cause a lockup. Noticed by: peter, jhb
2008-11-05 03:01:23 +00:00
/*
* Lock the per-process spinlock prior to dropping the PROC_LOCK
* to avoid a signal delivery race. PROC_LOCK, PROC_SLOCK, and
Update comment for tdsignal() -> tdsendsignal() rename. Forgot to include this in 209592.
2010-06-30 18:00:45 +00:00
* thread_lock() are currently held in tdsendsignal().
Revert rev 184216 and 184199, due to the way the thread_lock works, it may cause a lockup. Noticed by: peter, jhb
2008-11-05 03:01:23 +00:00
*/
PROC_SLOCK(p);
Actually, for signal and thread suspension, extra process spin lock is unnecessary, the normal process lock and thread lock are enough. The spin lock is still needed for process and thread exiting to mimic single sched_lock.
2008-10-23 07:55:38 +00:00
mtx_lock_spin(&sc->sc_lock);
partly revert revision 184199, because TDF_NEEDSIGCHK is persitent when thread is in kernel mode, it can cause dead loop, now unlock process lock after acquired sleep queue lock and thread lock to avoid the problem. This means TDF_NEEDSIGCHK and TDF_NEEDSUSPCHK must be set with process lock and thread lock being hold at same time.
2008-10-24 01:03:31 +00:00
PROC_UNLOCK(p);
Revert rev 184216 and 184199, due to the way the thread_lock works, it may cause a lockup. Noticed by: peter, jhb
2008-11-05 03:01:23 +00:00
thread_lock(td);
PROC_SUNLOCK(p);
Revision 184199 had not been fully reverted, add missing piece. Reported by: phk
2008-12-01 01:54:55 +00:00
if (ret == 0) {
sleepq_switch(wchan, pri);
return (0);
}
If thread set a TDP_WAKEUP for itself, clears the flag and returns EINTR immediately, this is used for implementing reliable pthread cancellation.
2010-08-20 04:28:30 +00:00
out:
Move comments to more accurate place.
2006-02-23 03:42:17 +00:00
/*
* There were pending signals and this thread is still
* on the sleep queue, remove it from the sleep queue.
*/
Commit 2/14 of sched_lock decomposition. - Adapt sleepqueues to the new thread_lock() mechanism. - Delay assigning the sleep queue spinlock as the thread lock until after we've checked for signals. It is illegal for a thread to return in mi_switch() with any lock assigned to td_lock other than the scheduler locks. - Change sleepq_catch_signals() to do the switch if necessary to simplify the callers. - Simplify timeout handling now that locking a sleeping thread has the side-effect of locking the sleepqueue. Some previous races are no longer possible. Tested by: kris, current@ Tested on: i386, amd64, ULE, 4BSD, libthr, libkse, PREEMPTION, etc. Discussed with: kris, attilio, kmacy, jhb, julian, bde (small parts each)
2007-06-04 23:50:56 +00:00
if (TD_ON_SLEEPQ(td)) {
sq = sleepq_lookup(wchan);
If a thread that is swapped out is made runnable, then the setrunnable() routine wakes up proc0 so that proc0 can swap the thread back in. Historically, this has been done by waking up proc0 directly from setrunnable() itself via a wakeup(). When waking up a sleeping thread that was swapped out (the usual case when waking proc0 since only sleeping threads are eligible to be swapped out), this resulted in a bit of recursion (e.g. wakeup() -> setrunnable() -> wakeup()). With sleep queues having separate locks in 6.x and later, this caused a spin lock LOR (sleepq lock -> sched_lock/thread lock -> sleepq lock). An attempt was made to fix this in 7.0 by making the proc0 wakeup use the ithread mechanism for doing the wakeup. However, this required grabbing proc0's thread lock to perform the wakeup. If proc0 was asleep elsewhere in the kernel (e.g. waiting for disk I/O), then this degenerated into the same LOR since the thread lock would be some other sleepq lock. Fix this by deferring the wakeup of the swapper until after the sleepq lock held by the upper layer has been locked. The setrunnable() routine now returns a boolean value to indicate whether or not proc0 needs to be woken up. The end result is that consumers of the sleepq API such as *sleep/wakeup, condition variables, sx locks, and lockmgr, have to wakeup proc0 if they get a non-zero return value from sleepq_abort(), sleepq_broadcast(), or sleepq_signal(). Discussed with: jeff Glanced at by: sam Tested by: Jurgen Weber jurgen - ish com au MFC after: 2 weeks
2008-08-05 20:02:31 +00:00
if (sleepq_resume_thread(sq, td, 0)) {
#ifdef INVARIANTS
/*
* This thread hasn't gone to sleep yet, so it
* should not be swapped out.
*/
panic("not waking up swapper");
#endif
}
Commit 2/14 of sched_lock decomposition. - Adapt sleepqueues to the new thread_lock() mechanism. - Delay assigning the sleep queue spinlock as the thread lock until after we've checked for signals. It is illegal for a thread to return in mi_switch() with any lock assigned to td_lock other than the scheduler locks. - Change sleepq_catch_signals() to do the switch if necessary to simplify the callers. - Simplify timeout handling now that locking a sleeping thread has the side-effect of locking the sleepqueue. Some previous races are no longer possible. Tested by: kris, current@ Tested on: i386, amd64, ULE, 4BSD, libthr, libkse, PREEMPTION, etc. Discussed with: kris, attilio, kmacy, jhb, julian, bde (small parts each)
2007-06-04 23:50:56 +00:00
}
mtx_unlock_spin(&sc->sc_lock);
MPASS(td->td_lock != &sc->sc_lock);
Fix a long standing race between sleep queue and thread suspension code. When a thread A is going to sleep, it calls sleepq_catch_signals() to detect any pending signals or thread suspension request, if nothing happens, it returns without holding process lock or scheduler lock, this opens a race window which allows thread B to come in and do process suspension work, however since A is still at running state, thread B can do nothing to A, thread A continues, and puts itself into actually sleeping state, but B has never seen it, and it sits there forever until B is woken up by other threads sometimes later(this can be very long delay or never happen). Fix this bug by forcing sleepq_catch_signals to return with scheduler lock held. Fix sleepq_abort() by passing it an interrupted code, previously, it worked as wakeup_one(), and the interruption can not be identified correctly by sleep queue code when the sleeping thread is resumed. Let thread_suspend_check() returns EINTR or ERESTART, so sleep queue no longer has to use SIGSTOP as a hack to build a return value. Reviewed by: jhb MFC after: 1 week
2006-02-15 23:52:01 +00:00
return (ret);
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
}
/*
Commit 2/14 of sched_lock decomposition. - Adapt sleepqueues to the new thread_lock() mechanism. - Delay assigning the sleep queue spinlock as the thread lock until after we've checked for signals. It is illegal for a thread to return in mi_switch() with any lock assigned to td_lock other than the scheduler locks. - Change sleepq_catch_signals() to do the switch if necessary to simplify the callers. - Simplify timeout handling now that locking a sleeping thread has the side-effect of locking the sleepqueue. Some previous races are no longer possible. Tested by: kris, current@ Tested on: i386, amd64, ULE, 4BSD, libthr, libkse, PREEMPTION, etc. Discussed with: kris, attilio, kmacy, jhb, julian, bde (small parts each)
2007-06-04 23:50:56 +00:00
* Switches to another thread if we are still asleep on a sleep queue.
* Returns with thread lock.
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
*/
static void
- Pass the priority argument from *sleep() into sleepq and down into sched_sleep(). This removes extra thread_lock() acquisition and allows the scheduler to decide what to do with the static boost. - Change the priority arguments to cv_* to match sleepq/msleep/etc. where 0 means no priority change. Catch -1 in cv_broadcastpri() and convert it to 0 for now. - Set a flag when sleeping in a way that is compatible with swapping since direct priority comparisons are meaningless now. - Add a sysctl to ule, kern.sched.static_boost, that defaults to on which controls the boost behavior. Turning it off gives better performance in some workloads but needs more investigation. - While we're modifying sleepq, change signal and broadcast to both return with the lock held as the lock was held on enter. Reviewed by: jhb, peter
2008-03-12 06:31:06 +00:00
sleepq_switch(void *wchan, int pri)
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
{
struct sleepqueue_chain *sc;
Fix a race in the sleepqueue timeout code that resulted in sleeps not being properly cancelled by a timeout. In general there is a race between a the sleepq timeout handler firing while the thread is still in the process of going to sleep. In 6.x with sched_lock, the race was largely protected by sched_lock. The only place it was "exposed" and had to be handled was while checking for any pending signals in sleepq_catch_signals(). With the thread lock changes, the thread lock is dropped in between sleepq_add() and sleepq_*wait*() opening up a new window for this race. Thus, if the timeout fired while the sleeping thread was in between sleepq_add() and sleepq_*wait*(), the thread would be marked as timed out, but the thread would not be dequeued and sleepq_switch() would still block the thread until it was awakened via some other means. In the case of pause(9) where there is no other wakeup, the thread would never be awakened. Fix this by teaching sleepq_switch() to check if the thread has had its sleep canceled before blocking by checking the TDF_TIMEOUT flag and aborting the sleep and dequeueing the thread if it is set. MFC after: 3 days Reported by: dwhite, peter
2008-01-25 02:09:38 +00:00
struct sleepqueue *sq;
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
struct thread *td;
td = curthread;
sc = SC_LOOKUP(wchan);
mtx_assert(&sc->sc_lock, MA_OWNED);
Commit 2/14 of sched_lock decomposition. - Adapt sleepqueues to the new thread_lock() mechanism. - Delay assigning the sleep queue spinlock as the thread lock until after we've checked for signals. It is illegal for a thread to return in mi_switch() with any lock assigned to td_lock other than the scheduler locks. - Change sleepq_catch_signals() to do the switch if necessary to simplify the callers. - Simplify timeout handling now that locking a sleeping thread has the side-effect of locking the sleepqueue. Some previous races are no longer possible. Tested by: kris, current@ Tested on: i386, amd64, ULE, 4BSD, libthr, libkse, PREEMPTION, etc. Discussed with: kris, attilio, kmacy, jhb, julian, bde (small parts each)
2007-06-04 23:50:56 +00:00
THREAD_LOCK_ASSERT(td, MA_OWNED);
Fix a race in the sleepqueue timeout code that resulted in sleeps not being properly cancelled by a timeout. In general there is a race between a the sleepq timeout handler firing while the thread is still in the process of going to sleep. In 6.x with sched_lock, the race was largely protected by sched_lock. The only place it was "exposed" and had to be handled was while checking for any pending signals in sleepq_catch_signals(). With the thread lock changes, the thread lock is dropped in between sleepq_add() and sleepq_*wait*() opening up a new window for this race. Thus, if the timeout fired while the sleeping thread was in between sleepq_add() and sleepq_*wait*(), the thread would be marked as timed out, but the thread would not be dequeued and sleepq_switch() would still block the thread until it was awakened via some other means. In the case of pause(9) where there is no other wakeup, the thread would never be awakened. Fix this by teaching sleepq_switch() to check if the thread has had its sleep canceled before blocking by checking the TDF_TIMEOUT flag and aborting the sleep and dequeueing the thread if it is set. MFC after: 3 days Reported by: dwhite, peter
2008-01-25 02:09:38 +00:00
/*
* If we have a sleep queue, then we've already been woken up, so
* just return.
*/
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
if (td->td_sleepqueue != NULL) {
mtx_unlock_spin(&sc->sc_lock);
return;
}
Fix a race in the sleepqueue timeout code that resulted in sleeps not being properly cancelled by a timeout. In general there is a race between a the sleepq timeout handler firing while the thread is still in the process of going to sleep. In 6.x with sched_lock, the race was largely protected by sched_lock. The only place it was "exposed" and had to be handled was while checking for any pending signals in sleepq_catch_signals(). With the thread lock changes, the thread lock is dropped in between sleepq_add() and sleepq_*wait*() opening up a new window for this race. Thus, if the timeout fired while the sleeping thread was in between sleepq_add() and sleepq_*wait*(), the thread would be marked as timed out, but the thread would not be dequeued and sleepq_switch() would still block the thread until it was awakened via some other means. In the case of pause(9) where there is no other wakeup, the thread would never be awakened. Fix this by teaching sleepq_switch() to check if the thread has had its sleep canceled before blocking by checking the TDF_TIMEOUT flag and aborting the sleep and dequeueing the thread if it is set. MFC after: 3 days Reported by: dwhite, peter
2008-01-25 02:09:38 +00:00
/*
* If TDF_TIMEOUT is set, then our sleep has been timed out
* already but we are still on the sleep queue, so dequeue the
* thread and return.
*/
if (td->td_flags & TDF_TIMEOUT) {
MPASS(TD_ON_SLEEPQ(td));
sq = sleepq_lookup(wchan);
If a thread that is swapped out is made runnable, then the setrunnable() routine wakes up proc0 so that proc0 can swap the thread back in. Historically, this has been done by waking up proc0 directly from setrunnable() itself via a wakeup(). When waking up a sleeping thread that was swapped out (the usual case when waking proc0 since only sleeping threads are eligible to be swapped out), this resulted in a bit of recursion (e.g. wakeup() -> setrunnable() -> wakeup()). With sleep queues having separate locks in 6.x and later, this caused a spin lock LOR (sleepq lock -> sched_lock/thread lock -> sleepq lock). An attempt was made to fix this in 7.0 by making the proc0 wakeup use the ithread mechanism for doing the wakeup. However, this required grabbing proc0's thread lock to perform the wakeup. If proc0 was asleep elsewhere in the kernel (e.g. waiting for disk I/O), then this degenerated into the same LOR since the thread lock would be some other sleepq lock. Fix this by deferring the wakeup of the swapper until after the sleepq lock held by the upper layer has been locked. The setrunnable() routine now returns a boolean value to indicate whether or not proc0 needs to be woken up. The end result is that consumers of the sleepq API such as *sleep/wakeup, condition variables, sx locks, and lockmgr, have to wakeup proc0 if they get a non-zero return value from sleepq_abort(), sleepq_broadcast(), or sleepq_signal(). Discussed with: jeff Glanced at by: sam Tested by: Jurgen Weber jurgen - ish com au MFC after: 2 weeks
2008-08-05 20:02:31 +00:00
if (sleepq_resume_thread(sq, td, 0)) {
#ifdef INVARIANTS
/*
* This thread hasn't gone to sleep yet, so it
* should not be swapped out.
*/
panic("not waking up swapper");
#endif
}
Fix a race in the sleepqueue timeout code that resulted in sleeps not being properly cancelled by a timeout. In general there is a race between a the sleepq timeout handler firing while the thread is still in the process of going to sleep. In 6.x with sched_lock, the race was largely protected by sched_lock. The only place it was "exposed" and had to be handled was while checking for any pending signals in sleepq_catch_signals(). With the thread lock changes, the thread lock is dropped in between sleepq_add() and sleepq_*wait*() opening up a new window for this race. Thus, if the timeout fired while the sleeping thread was in between sleepq_add() and sleepq_*wait*(), the thread would be marked as timed out, but the thread would not be dequeued and sleepq_switch() would still block the thread until it was awakened via some other means. In the case of pause(9) where there is no other wakeup, the thread would never be awakened. Fix this by teaching sleepq_switch() to check if the thread has had its sleep canceled before blocking by checking the TDF_TIMEOUT flag and aborting the sleep and dequeueing the thread if it is set. MFC after: 3 days Reported by: dwhite, peter
2008-01-25 02:09:38 +00:00
mtx_unlock_spin(&sc->sc_lock);
return;
}
- Add a facility similar to LOCK_PROFILING under SLEEPQUEUE_PROFILING. Keep a simple (wmesg, count) tuple in a hash to keep track of how many times we sleep at each wait message. We hash on message and not channel. No line number information is given as typically wait messages are not used in more than one place. Identical strings defined at different addresses will show up with seperate counters. - Use debug.sleepq.enable to enable, .reset to reset, and .stats dumps stats. - Do an unsynchronized check in sleepq_switch() prior to switching before calling sleepq_profile() which uses a global lock to synchronize the hash. Only sleeps which actually cause a context switch are counted.
2008-03-19 07:22:07 +00:00
#ifdef SLEEPQUEUE_PROFILING
if (prof_enabled)
sleepq_profile(td->td_wmesg);
#endif
Commit 2/14 of sched_lock decomposition. - Adapt sleepqueues to the new thread_lock() mechanism. - Delay assigning the sleep queue spinlock as the thread lock until after we've checked for signals. It is illegal for a thread to return in mi_switch() with any lock assigned to td_lock other than the scheduler locks. - Change sleepq_catch_signals() to do the switch if necessary to simplify the callers. - Simplify timeout handling now that locking a sleeping thread has the side-effect of locking the sleepqueue. Some previous races are no longer possible. Tested by: kris, current@ Tested on: i386, amd64, ULE, 4BSD, libthr, libkse, PREEMPTION, etc. Discussed with: kris, attilio, kmacy, jhb, julian, bde (small parts each)
2007-06-04 23:50:56 +00:00
MPASS(td->td_sleepqueue == NULL);
- Pass the priority argument from *sleep() into sleepq and down into sched_sleep(). This removes extra thread_lock() acquisition and allows the scheduler to decide what to do with the static boost. - Change the priority arguments to cv_* to match sleepq/msleep/etc. where 0 means no priority change. Catch -1 in cv_broadcastpri() and convert it to 0 for now. - Set a flag when sleeping in a way that is compatible with swapping since direct priority comparisons are meaningless now. - Add a sysctl to ule, kern.sched.static_boost, that defaults to on which controls the boost behavior. Turning it off gives better performance in some workloads but needs more investigation. - While we're modifying sleepq, change signal and broadcast to both return with the lock held as the lock was held on enter. Reviewed by: jhb, peter
2008-03-12 06:31:06 +00:00
sched_sleep(td, pri);
thread_lock_set(td, &sc->sc_lock);
Implement the DTrace sched provider. This implementation aims to be compatible with the sched provider implemented by Solaris and its open- source derivatives. Full documentation of the sched provider can be found on Oracle's DTrace wiki pages. Note that for compatibility with scripts originally written for Solaris, serveral probes are defined that will never fire. These probes are defined to fire when Solaris-specific features perform certain actions. As these features are not present in FreeBSD, the probes can never fire. Also, I have added a two probes that are not defined in Solaris, lend-pri and load-change. These probes have been added to make it possible to collect schedgraph data with DTrace. Finally, a few probes are defined in Solaris to take a cpuinfo_t * argument. As it was not immediately clear to me how to translate that to FreeBSD, currently those probes are passed NULL in place of a cpuinfo_t *. Sponsored by: Sandvine Incorporated MFC after: 2 weeks
2012-05-15 01:30:25 +00:00
SDT_PROBE0(sched, , , sleep);
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
TD_SET_SLEEPING(td);
- Make SCHED_STATS more generic by adding a wrapper to create the variables and sysctl nodes. - In reset walk the children of kern_sched_stats and reset the counters via the oid_arg1 pointer. This allows us to add arbitrary counters to the tree and still reset them properly. - Define a set of switch types to be passed with flags to mi_switch(). These types are named SWT_*. These types correspond to SCHED_STATS counters and are automatically handled in this way. - Make the new SWT_ types more specific than the older switch stats. There are now stats for idle switches, remote idle wakeups, remote preemption ithreads idling, etc. - Add switch statistics for ULE's pickcpu algorithm. These stats include how much migration there is, how often affinity was successful, how often threads were migrated to the local cpu on wakeup, etc. Sponsored by: Nokia
2008-04-17 04:20:10 +00:00
mi_switch(SW_VOL | SWT_SLEEPQ, NULL);
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
KASSERT(TD_IS_RUNNING(td), ("running but not TDS_RUNNING"));
Fixed some common printf format errors. Don't assume that "struct foo *" is "void *" (it isn't) or that the default promotion of pid_t is int. Instead, assume that casting "struct foo *" to "void *" and printing the result with %p is useful, and that all pid_t's are representable as longs. Fixed some minor style bugs (mainly spelling errors in comments).
2004-05-14 20:51:42 +00:00
CTR3(KTR_PROC, "sleepq resume: thread %p (pid %ld, %s)",
generally we are interested in what thread did something as opposed to what process. Since threads by default have teh name of the process unless over-written with more useful information, just print the thread name instead.
2007-11-14 06:21:24 +00:00
(void *)td, (long)td->td_proc->p_pid, (void *)td->td_name);
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
}
/*
* Check to see if we timed out.
*/
static int
Major callout subsystem cleanup and rewrite: - Close a migration race where callout_reset() failed to set the CALLOUT_ACTIVE flag. - Callout callback functions are now allowed to be protected by spinlocks. - Switching the callout CPU number cannot always be done on a per-callout basis. See the updated timeout(9) manual page for more information. - The timeout(9) manual page has been updated to reflect how all the functions inside the callout API are working. The manual page has been made function oriented to make it easier to deduce how each of the functions making up the callout API are working without having to first read the whole manual page. Group all functions into a handful of sections which should give a quick top-level overview when the different functions should be used. - The CALLOUT_SHAREDLOCK flag and its functionality has been removed to reduce the complexity in the callout code and to avoid problems about atomically stopping callouts via callout_stop(). If someone needs it, it can be re-added. From my quick grep there are no CALLOUT_SHAREDLOCK clients in the kernel. - A new callout API function named "callout_drain_async()" has been added. See the updated timeout(9) manual page for a complete description. - Update the callout clients in the "kern/" folder to use the callout API properly, like cv_timedwait(). Previously there was some custom sleepqueue code in the callout subsystem, which has been removed, because we now allow callouts to be protected by spinlocks. This allows us to tear down the callout like done with regular mutexes, and a "td_slpmutex" has been added to "struct thread" to atomically teardown the "td_slpcallout". Further the "TDF_TIMOFAIL" and "SWT_SLEEPQTIMO" states can now be completely removed. Currently they are marked as available and will be cleaned up in a follow up commit. - Bump the __FreeBSD_version to indicate kernel modules need recompilation. - There has been several reports that this patch "seems to squash a serious bug leading to a callout timeout and panic". Kernel build testing: all architectures were built MFC after: 2 weeks Differential Revision: https://reviews.freebsd.org/D1438 Sponsored by: Mellanox Technologies Reviewed by: jhb, adrian, sbruno and emaste
2015-01-15 15:32:30 +00:00
sleepq_check_timeout(struct thread *td)
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
{
Commit 2/14 of sched_lock decomposition. - Adapt sleepqueues to the new thread_lock() mechanism. - Delay assigning the sleep queue spinlock as the thread lock until after we've checked for signals. It is illegal for a thread to return in mi_switch() with any lock assigned to td_lock other than the scheduler locks. - Change sleepq_catch_signals() to do the switch if necessary to simplify the callers. - Simplify timeout handling now that locking a sleeping thread has the side-effect of locking the sleepqueue. Some previous races are no longer possible. Tested by: kris, current@ Tested on: i386, amd64, ULE, 4BSD, libthr, libkse, PREEMPTION, etc. Discussed with: kris, attilio, kmacy, jhb, julian, bde (small parts each)
2007-06-04 23:50:56 +00:00
THREAD_LOCK_ASSERT(td, MA_OWNED);
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
/*
* If TDF_TIMEOUT is set, we timed out.
*/
if (td->td_flags & TDF_TIMEOUT) {
td->td_flags &= ~TDF_TIMEOUT;
return (EWOULDBLOCK);
}
return (0);
}
Major callout subsystem cleanup and rewrite: - Close a migration race where callout_reset() failed to set the CALLOUT_ACTIVE flag. - Callout callback functions are now allowed to be protected by spinlocks. - Switching the callout CPU number cannot always be done on a per-callout basis. See the updated timeout(9) manual page for more information. - The timeout(9) manual page has been updated to reflect how all the functions inside the callout API are working. The manual page has been made function oriented to make it easier to deduce how each of the functions making up the callout API are working without having to first read the whole manual page. Group all functions into a handful of sections which should give a quick top-level overview when the different functions should be used. - The CALLOUT_SHAREDLOCK flag and its functionality has been removed to reduce the complexity in the callout code and to avoid problems about atomically stopping callouts via callout_stop(). If someone needs it, it can be re-added. From my quick grep there are no CALLOUT_SHAREDLOCK clients in the kernel. - A new callout API function named "callout_drain_async()" has been added. See the updated timeout(9) manual page for a complete description. - Update the callout clients in the "kern/" folder to use the callout API properly, like cv_timedwait(). Previously there was some custom sleepqueue code in the callout subsystem, which has been removed, because we now allow callouts to be protected by spinlocks. This allows us to tear down the callout like done with regular mutexes, and a "td_slpmutex" has been added to "struct thread" to atomically teardown the "td_slpcallout". Further the "TDF_TIMOFAIL" and "SWT_SLEEPQTIMO" states can now be completely removed. Currently they are marked as available and will be cleaned up in a follow up commit. - Bump the __FreeBSD_version to indicate kernel modules need recompilation. - There has been several reports that this patch "seems to squash a serious bug leading to a callout timeout and panic". Kernel build testing: all architectures were built MFC after: 2 weeks Differential Revision: https://reviews.freebsd.org/D1438 Sponsored by: Mellanox Technologies Reviewed by: jhb, adrian, sbruno and emaste
2015-01-15 15:32:30 +00:00
/*
* Atomically stop the timeout by using a mutex.
*/
static void
sleepq_stop_timeout(struct thread *td)
{
mtx_lock_spin(&td->td_slpmutex);
callout_stop(&td->td_slpcallout);
mtx_unlock_spin(&td->td_slpmutex);
}
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
/*
* Check to see if we were awoken by a signal.
*/
static int
sleepq_check_signals(void)
{
struct thread *td;
td = curthread;
Commit 2/14 of sched_lock decomposition. - Adapt sleepqueues to the new thread_lock() mechanism. - Delay assigning the sleep queue spinlock as the thread lock until after we've checked for signals. It is illegal for a thread to return in mi_switch() with any lock assigned to td_lock other than the scheduler locks. - Change sleepq_catch_signals() to do the switch if necessary to simplify the callers. - Simplify timeout handling now that locking a sleeping thread has the side-effect of locking the sleepqueue. Some previous races are no longer possible. Tested by: kris, current@ Tested on: i386, amd64, ULE, 4BSD, libthr, libkse, PREEMPTION, etc. Discussed with: kris, attilio, kmacy, jhb, julian, bde (small parts each)
2007-06-04 23:50:56 +00:00
THREAD_LOCK_ASSERT(td, MA_OWNED);
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
/* We are no longer in an interruptible sleep. */
Fix a long standing race between sleep queue and thread suspension code. When a thread A is going to sleep, it calls sleepq_catch_signals() to detect any pending signals or thread suspension request, if nothing happens, it returns without holding process lock or scheduler lock, this opens a race window which allows thread B to come in and do process suspension work, however since A is still at running state, thread B can do nothing to A, thread A continues, and puts itself into actually sleeping state, but B has never seen it, and it sits there forever until B is woken up by other threads sometimes later(this can be very long delay or never happen). Fix this bug by forcing sleepq_catch_signals to return with scheduler lock held. Fix sleepq_abort() by passing it an interrupted code, previously, it worked as wakeup_one(), and the interruption can not be identified correctly by sleep queue code when the sleeping thread is resumed. Let thread_suspend_check() returns EINTR or ERESTART, so sleep queue no longer has to use SIGSTOP as a hack to build a return value. Reviewed by: jhb MFC after: 1 week
2006-02-15 23:52:01 +00:00
if (td->td_flags & TDF_SINTR)
Rework the handling of stop signals in the NFS client. The changes in 195702, 195703, and 195821 prevented a thread from suspending while holding locks inside of NFS by forcing the thread to fail sleeps with EINTR or ERESTART but defer the thread suspension to the user boundary. However, this had the effect that stopping a process during an NFS request could abort the request and trigger EINTR errors that were visible to userland processes (previously the thread would have suspended and completed the request once it was resumed). This change instead effectively masks stop signals while in the NFS client. It uses the existing TDF_SBDRY flag to effect this since SIGSTOP cannot be masked directly. Also, instead of setting PBDRY on individual sleeps, the NFS client now sets the TDF_SBDRY flag around each NFS request and stop signals are masked for all sleeps during that region (the previous change missed sleeps in lockmgr locks). The end result is that stop signals sent to threads performing an NFS request are completely ignored until after the NFS request has finished processing and the thread prepares to return to userland. This restores the behavior of stop signals being transparent to userland processes while still preventing threads from suspending while holding NFS locks. Reviewed by: kib MFC after: 1 month
2013-02-06 17:06:51 +00:00
td->td_flags &= ~TDF_SINTR;
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
Fix a long standing race between sleep queue and thread suspension code. When a thread A is going to sleep, it calls sleepq_catch_signals() to detect any pending signals or thread suspension request, if nothing happens, it returns without holding process lock or scheduler lock, this opens a race window which allows thread B to come in and do process suspension work, however since A is still at running state, thread B can do nothing to A, thread A continues, and puts itself into actually sleeping state, but B has never seen it, and it sits there forever until B is woken up by other threads sometimes later(this can be very long delay or never happen). Fix this bug by forcing sleepq_catch_signals to return with scheduler lock held. Fix sleepq_abort() by passing it an interrupted code, previously, it worked as wakeup_one(), and the interruption can not be identified correctly by sleep queue code when the sleeping thread is resumed. Let thread_suspend_check() returns EINTR or ERESTART, so sleep queue no longer has to use SIGSTOP as a hack to build a return value. Reviewed by: jhb MFC after: 1 week
2006-02-15 23:52:01 +00:00
if (td->td_flags & TDF_SLEEPABORT) {
td->td_flags &= ~TDF_SLEEPABORT;
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
return (td->td_intrval);
Fix a long standing race between sleep queue and thread suspension code. When a thread A is going to sleep, it calls sleepq_catch_signals() to detect any pending signals or thread suspension request, if nothing happens, it returns without holding process lock or scheduler lock, this opens a race window which allows thread B to come in and do process suspension work, however since A is still at running state, thread B can do nothing to A, thread A continues, and puts itself into actually sleeping state, but B has never seen it, and it sits there forever until B is woken up by other threads sometimes later(this can be very long delay or never happen). Fix this bug by forcing sleepq_catch_signals to return with scheduler lock held. Fix sleepq_abort() by passing it an interrupted code, previously, it worked as wakeup_one(), and the interruption can not be identified correctly by sleep queue code when the sleeping thread is resumed. Let thread_suspend_check() returns EINTR or ERESTART, so sleep queue no longer has to use SIGSTOP as a hack to build a return value. Reviewed by: jhb MFC after: 1 week
2006-02-15 23:52:01 +00:00
}
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
Fix a long standing race between sleep queue and thread suspension code. When a thread A is going to sleep, it calls sleepq_catch_signals() to detect any pending signals or thread suspension request, if nothing happens, it returns without holding process lock or scheduler lock, this opens a race window which allows thread B to come in and do process suspension work, however since A is still at running state, thread B can do nothing to A, thread A continues, and puts itself into actually sleeping state, but B has never seen it, and it sits there forever until B is woken up by other threads sometimes later(this can be very long delay or never happen). Fix this bug by forcing sleepq_catch_signals to return with scheduler lock held. Fix sleepq_abort() by passing it an interrupted code, previously, it worked as wakeup_one(), and the interruption can not be identified correctly by sleep queue code when the sleeping thread is resumed. Let thread_suspend_check() returns EINTR or ERESTART, so sleep queue no longer has to use SIGSTOP as a hack to build a return value. Reviewed by: jhb MFC after: 1 week
2006-02-15 23:52:01 +00:00
return (0);
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
}
/*
* Block the current thread until it is awakened from its sleep queue.
*/
void
- Pass the priority argument from *sleep() into sleepq and down into sched_sleep(). This removes extra thread_lock() acquisition and allows the scheduler to decide what to do with the static boost. - Change the priority arguments to cv_* to match sleepq/msleep/etc. where 0 means no priority change. Catch -1 in cv_broadcastpri() and convert it to 0 for now. - Set a flag when sleeping in a way that is compatible with swapping since direct priority comparisons are meaningless now. - Add a sysctl to ule, kern.sched.static_boost, that defaults to on which controls the boost behavior. Turning it off gives better performance in some workloads but needs more investigation. - While we're modifying sleepq, change signal and broadcast to both return with the lock held as the lock was held on enter. Reviewed by: jhb, peter
2008-03-12 06:31:06 +00:00
sleepq_wait(void *wchan, int pri)
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
{
Commit 2/14 of sched_lock decomposition. - Adapt sleepqueues to the new thread_lock() mechanism. - Delay assigning the sleep queue spinlock as the thread lock until after we've checked for signals. It is illegal for a thread to return in mi_switch() with any lock assigned to td_lock other than the scheduler locks. - Change sleepq_catch_signals() to do the switch if necessary to simplify the callers. - Simplify timeout handling now that locking a sleeping thread has the side-effect of locking the sleepqueue. Some previous races are no longer possible. Tested by: kris, current@ Tested on: i386, amd64, ULE, 4BSD, libthr, libkse, PREEMPTION, etc. Discussed with: kris, attilio, kmacy, jhb, julian, bde (small parts each)
2007-06-04 23:50:56 +00:00
struct thread *td;
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
Commit 2/14 of sched_lock decomposition. - Adapt sleepqueues to the new thread_lock() mechanism. - Delay assigning the sleep queue spinlock as the thread lock until after we've checked for signals. It is illegal for a thread to return in mi_switch() with any lock assigned to td_lock other than the scheduler locks. - Change sleepq_catch_signals() to do the switch if necessary to simplify the callers. - Simplify timeout handling now that locking a sleeping thread has the side-effect of locking the sleepqueue. Some previous races are no longer possible. Tested by: kris, current@ Tested on: i386, amd64, ULE, 4BSD, libthr, libkse, PREEMPTION, etc. Discussed with: kris, attilio, kmacy, jhb, julian, bde (small parts each)
2007-06-04 23:50:56 +00:00
td = curthread;
MPASS(!(td->td_flags & TDF_SINTR));
thread_lock(td);
- Pass the priority argument from *sleep() into sleepq and down into sched_sleep(). This removes extra thread_lock() acquisition and allows the scheduler to decide what to do with the static boost. - Change the priority arguments to cv_* to match sleepq/msleep/etc. where 0 means no priority change. Catch -1 in cv_broadcastpri() and convert it to 0 for now. - Set a flag when sleeping in a way that is compatible with swapping since direct priority comparisons are meaningless now. - Add a sysctl to ule, kern.sched.static_boost, that defaults to on which controls the boost behavior. Turning it off gives better performance in some workloads but needs more investigation. - While we're modifying sleepq, change signal and broadcast to both return with the lock held as the lock was held on enter. Reviewed by: jhb, peter
2008-03-12 06:31:06 +00:00
sleepq_switch(wchan, pri);
Commit 2/14 of sched_lock decomposition. - Adapt sleepqueues to the new thread_lock() mechanism. - Delay assigning the sleep queue spinlock as the thread lock until after we've checked for signals. It is illegal for a thread to return in mi_switch() with any lock assigned to td_lock other than the scheduler locks. - Change sleepq_catch_signals() to do the switch if necessary to simplify the callers. - Simplify timeout handling now that locking a sleeping thread has the side-effect of locking the sleepqueue. Some previous races are no longer possible. Tested by: kris, current@ Tested on: i386, amd64, ULE, 4BSD, libthr, libkse, PREEMPTION, etc. Discussed with: kris, attilio, kmacy, jhb, julian, bde (small parts each)
2007-06-04 23:50:56 +00:00
thread_unlock(td);
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
}
/*
* Block the current thread until it is awakened from its sleep queue
* or it is interrupted by a signal.
*/
int
- Pass the priority argument from *sleep() into sleepq and down into sched_sleep(). This removes extra thread_lock() acquisition and allows the scheduler to decide what to do with the static boost. - Change the priority arguments to cv_* to match sleepq/msleep/etc. where 0 means no priority change. Catch -1 in cv_broadcastpri() and convert it to 0 for now. - Set a flag when sleeping in a way that is compatible with swapping since direct priority comparisons are meaningless now. - Add a sysctl to ule, kern.sched.static_boost, that defaults to on which controls the boost behavior. Turning it off gives better performance in some workloads but needs more investigation. - While we're modifying sleepq, change signal and broadcast to both return with the lock held as the lock was held on enter. Reviewed by: jhb, peter
2008-03-12 06:31:06 +00:00
sleepq_wait_sig(void *wchan, int pri)
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
{
Fix a long standing race between sleep queue and thread suspension code. When a thread A is going to sleep, it calls sleepq_catch_signals() to detect any pending signals or thread suspension request, if nothing happens, it returns without holding process lock or scheduler lock, this opens a race window which allows thread B to come in and do process suspension work, however since A is still at running state, thread B can do nothing to A, thread A continues, and puts itself into actually sleeping state, but B has never seen it, and it sits there forever until B is woken up by other threads sometimes later(this can be very long delay or never happen). Fix this bug by forcing sleepq_catch_signals to return with scheduler lock held. Fix sleepq_abort() by passing it an interrupted code, previously, it worked as wakeup_one(), and the interruption can not be identified correctly by sleep queue code when the sleeping thread is resumed. Let thread_suspend_check() returns EINTR or ERESTART, so sleep queue no longer has to use SIGSTOP as a hack to build a return value. Reviewed by: jhb MFC after: 1 week
2006-02-15 23:52:01 +00:00
int rcatch;
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
int rval;
- Pass the priority argument from *sleep() into sleepq and down into sched_sleep(). This removes extra thread_lock() acquisition and allows the scheduler to decide what to do with the static boost. - Change the priority arguments to cv_* to match sleepq/msleep/etc. where 0 means no priority change. Catch -1 in cv_broadcastpri() and convert it to 0 for now. - Set a flag when sleeping in a way that is compatible with swapping since direct priority comparisons are meaningless now. - Add a sysctl to ule, kern.sched.static_boost, that defaults to on which controls the boost behavior. Turning it off gives better performance in some workloads but needs more investigation. - While we're modifying sleepq, change signal and broadcast to both return with the lock held as the lock was held on enter. Reviewed by: jhb, peter
2008-03-12 06:31:06 +00:00
rcatch = sleepq_catch_signals(wchan, pri);
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
rval = sleepq_check_signals();
Commit 2/14 of sched_lock decomposition. - Adapt sleepqueues to the new thread_lock() mechanism. - Delay assigning the sleep queue spinlock as the thread lock until after we've checked for signals. It is illegal for a thread to return in mi_switch() with any lock assigned to td_lock other than the scheduler locks. - Change sleepq_catch_signals() to do the switch if necessary to simplify the callers. - Simplify timeout handling now that locking a sleeping thread has the side-effect of locking the sleepqueue. Some previous races are no longer possible. Tested by: kris, current@ Tested on: i386, amd64, ULE, 4BSD, libthr, libkse, PREEMPTION, etc. Discussed with: kris, attilio, kmacy, jhb, julian, bde (small parts each)
2007-06-04 23:50:56 +00:00
thread_unlock(curthread);
Fix a long standing race between sleep queue and thread suspension code. When a thread A is going to sleep, it calls sleepq_catch_signals() to detect any pending signals or thread suspension request, if nothing happens, it returns without holding process lock or scheduler lock, this opens a race window which allows thread B to come in and do process suspension work, however since A is still at running state, thread B can do nothing to A, thread A continues, and puts itself into actually sleeping state, but B has never seen it, and it sits there forever until B is woken up by other threads sometimes later(this can be very long delay or never happen). Fix this bug by forcing sleepq_catch_signals to return with scheduler lock held. Fix sleepq_abort() by passing it an interrupted code, previously, it worked as wakeup_one(), and the interruption can not be identified correctly by sleep queue code when the sleeping thread is resumed. Let thread_suspend_check() returns EINTR or ERESTART, so sleep queue no longer has to use SIGSTOP as a hack to build a return value. Reviewed by: jhb MFC after: 1 week
2006-02-15 23:52:01 +00:00
if (rcatch)
return (rcatch);
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
return (rval);
}
/*
* Block the current thread until it is awakened from its sleep queue
* or it times out while waiting.
*/
int
- Pass the priority argument from *sleep() into sleepq and down into sched_sleep(). This removes extra thread_lock() acquisition and allows the scheduler to decide what to do with the static boost. - Change the priority arguments to cv_* to match sleepq/msleep/etc. where 0 means no priority change. Catch -1 in cv_broadcastpri() and convert it to 0 for now. - Set a flag when sleeping in a way that is compatible with swapping since direct priority comparisons are meaningless now. - Add a sysctl to ule, kern.sched.static_boost, that defaults to on which controls the boost behavior. Turning it off gives better performance in some workloads but needs more investigation. - While we're modifying sleepq, change signal and broadcast to both return with the lock held as the lock was held on enter. Reviewed by: jhb, peter
2008-03-12 06:31:06 +00:00
sleepq_timedwait(void *wchan, int pri)
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
{
Commit 2/14 of sched_lock decomposition. - Adapt sleepqueues to the new thread_lock() mechanism. - Delay assigning the sleep queue spinlock as the thread lock until after we've checked for signals. It is illegal for a thread to return in mi_switch() with any lock assigned to td_lock other than the scheduler locks. - Change sleepq_catch_signals() to do the switch if necessary to simplify the callers. - Simplify timeout handling now that locking a sleeping thread has the side-effect of locking the sleepqueue. Some previous races are no longer possible. Tested by: kris, current@ Tested on: i386, amd64, ULE, 4BSD, libthr, libkse, PREEMPTION, etc. Discussed with: kris, attilio, kmacy, jhb, julian, bde (small parts each)
2007-06-04 23:50:56 +00:00
struct thread *td;
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
int rval;
Commit 2/14 of sched_lock decomposition. - Adapt sleepqueues to the new thread_lock() mechanism. - Delay assigning the sleep queue spinlock as the thread lock until after we've checked for signals. It is illegal for a thread to return in mi_switch() with any lock assigned to td_lock other than the scheduler locks. - Change sleepq_catch_signals() to do the switch if necessary to simplify the callers. - Simplify timeout handling now that locking a sleeping thread has the side-effect of locking the sleepqueue. Some previous races are no longer possible. Tested by: kris, current@ Tested on: i386, amd64, ULE, 4BSD, libthr, libkse, PREEMPTION, etc. Discussed with: kris, attilio, kmacy, jhb, julian, bde (small parts each)
2007-06-04 23:50:56 +00:00
td = curthread;
MPASS(!(td->td_flags & TDF_SINTR));
thread_lock(td);
- Pass the priority argument from *sleep() into sleepq and down into sched_sleep(). This removes extra thread_lock() acquisition and allows the scheduler to decide what to do with the static boost. - Change the priority arguments to cv_* to match sleepq/msleep/etc. where 0 means no priority change. Catch -1 in cv_broadcastpri() and convert it to 0 for now. - Set a flag when sleeping in a way that is compatible with swapping since direct priority comparisons are meaningless now. - Add a sysctl to ule, kern.sched.static_boost, that defaults to on which controls the boost behavior. Turning it off gives better performance in some workloads but needs more investigation. - While we're modifying sleepq, change signal and broadcast to both return with the lock held as the lock was held on enter. Reviewed by: jhb, peter
2008-03-12 06:31:06 +00:00
sleepq_switch(wchan, pri);
Major callout subsystem cleanup and rewrite: - Close a migration race where callout_reset() failed to set the CALLOUT_ACTIVE flag. - Callout callback functions are now allowed to be protected by spinlocks. - Switching the callout CPU number cannot always be done on a per-callout basis. See the updated timeout(9) manual page for more information. - The timeout(9) manual page has been updated to reflect how all the functions inside the callout API are working. The manual page has been made function oriented to make it easier to deduce how each of the functions making up the callout API are working without having to first read the whole manual page. Group all functions into a handful of sections which should give a quick top-level overview when the different functions should be used. - The CALLOUT_SHAREDLOCK flag and its functionality has been removed to reduce the complexity in the callout code and to avoid problems about atomically stopping callouts via callout_stop(). If someone needs it, it can be re-added. From my quick grep there are no CALLOUT_SHAREDLOCK clients in the kernel. - A new callout API function named "callout_drain_async()" has been added. See the updated timeout(9) manual page for a complete description. - Update the callout clients in the "kern/" folder to use the callout API properly, like cv_timedwait(). Previously there was some custom sleepqueue code in the callout subsystem, which has been removed, because we now allow callouts to be protected by spinlocks. This allows us to tear down the callout like done with regular mutexes, and a "td_slpmutex" has been added to "struct thread" to atomically teardown the "td_slpcallout". Further the "TDF_TIMOFAIL" and "SWT_SLEEPQTIMO" states can now be completely removed. Currently they are marked as available and will be cleaned up in a follow up commit. - Bump the __FreeBSD_version to indicate kernel modules need recompilation. - There has been several reports that this patch "seems to squash a serious bug leading to a callout timeout and panic". Kernel build testing: all architectures were built MFC after: 2 weeks Differential Revision: https://reviews.freebsd.org/D1438 Sponsored by: Mellanox Technologies Reviewed by: jhb, adrian, sbruno and emaste
2015-01-15 15:32:30 +00:00
rval = sleepq_check_timeout(td);
Commit 2/14 of sched_lock decomposition. - Adapt sleepqueues to the new thread_lock() mechanism. - Delay assigning the sleep queue spinlock as the thread lock until after we've checked for signals. It is illegal for a thread to return in mi_switch() with any lock assigned to td_lock other than the scheduler locks. - Change sleepq_catch_signals() to do the switch if necessary to simplify the callers. - Simplify timeout handling now that locking a sleeping thread has the side-effect of locking the sleepqueue. Some previous races are no longer possible. Tested by: kris, current@ Tested on: i386, amd64, ULE, 4BSD, libthr, libkse, PREEMPTION, etc. Discussed with: kris, attilio, kmacy, jhb, julian, bde (small parts each)
2007-06-04 23:50:56 +00:00
thread_unlock(td);
Major callout subsystem cleanup and rewrite: - Close a migration race where callout_reset() failed to set the CALLOUT_ACTIVE flag. - Callout callback functions are now allowed to be protected by spinlocks. - Switching the callout CPU number cannot always be done on a per-callout basis. See the updated timeout(9) manual page for more information. - The timeout(9) manual page has been updated to reflect how all the functions inside the callout API are working. The manual page has been made function oriented to make it easier to deduce how each of the functions making up the callout API are working without having to first read the whole manual page. Group all functions into a handful of sections which should give a quick top-level overview when the different functions should be used. - The CALLOUT_SHAREDLOCK flag and its functionality has been removed to reduce the complexity in the callout code and to avoid problems about atomically stopping callouts via callout_stop(). If someone needs it, it can be re-added. From my quick grep there are no CALLOUT_SHAREDLOCK clients in the kernel. - A new callout API function named "callout_drain_async()" has been added. See the updated timeout(9) manual page for a complete description. - Update the callout clients in the "kern/" folder to use the callout API properly, like cv_timedwait(). Previously there was some custom sleepqueue code in the callout subsystem, which has been removed, because we now allow callouts to be protected by spinlocks. This allows us to tear down the callout like done with regular mutexes, and a "td_slpmutex" has been added to "struct thread" to atomically teardown the "td_slpcallout". Further the "TDF_TIMOFAIL" and "SWT_SLEEPQTIMO" states can now be completely removed. Currently they are marked as available and will be cleaned up in a follow up commit. - Bump the __FreeBSD_version to indicate kernel modules need recompilation. - There has been several reports that this patch "seems to squash a serious bug leading to a callout timeout and panic". Kernel build testing: all architectures were built MFC after: 2 weeks Differential Revision: https://reviews.freebsd.org/D1438 Sponsored by: Mellanox Technologies Reviewed by: jhb, adrian, sbruno and emaste
2015-01-15 15:32:30 +00:00
sleepq_stop_timeout(td);
Remove the signal_caught argument from sleepq_timedwait() as it was effectively always zero.
2004-06-28 18:57:06 +00:00
return (rval);
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
}
/*
* Block the current thread until it is awakened from its sleep queue,
* it is interrupted by a signal, or it times out waiting to be awakened.
*/
int
- Pass the priority argument from *sleep() into sleepq and down into sched_sleep(). This removes extra thread_lock() acquisition and allows the scheduler to decide what to do with the static boost. - Change the priority arguments to cv_* to match sleepq/msleep/etc. where 0 means no priority change. Catch -1 in cv_broadcastpri() and convert it to 0 for now. - Set a flag when sleeping in a way that is compatible with swapping since direct priority comparisons are meaningless now. - Add a sysctl to ule, kern.sched.static_boost, that defaults to on which controls the boost behavior. Turning it off gives better performance in some workloads but needs more investigation. - While we're modifying sleepq, change signal and broadcast to both return with the lock held as the lock was held on enter. Reviewed by: jhb, peter
2008-03-12 06:31:06 +00:00
sleepq_timedwait_sig(void *wchan, int pri)
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
{
Major callout subsystem cleanup and rewrite: - Close a migration race where callout_reset() failed to set the CALLOUT_ACTIVE flag. - Callout callback functions are now allowed to be protected by spinlocks. - Switching the callout CPU number cannot always be done on a per-callout basis. See the updated timeout(9) manual page for more information. - The timeout(9) manual page has been updated to reflect how all the functions inside the callout API are working. The manual page has been made function oriented to make it easier to deduce how each of the functions making up the callout API are working without having to first read the whole manual page. Group all functions into a handful of sections which should give a quick top-level overview when the different functions should be used. - The CALLOUT_SHAREDLOCK flag and its functionality has been removed to reduce the complexity in the callout code and to avoid problems about atomically stopping callouts via callout_stop(). If someone needs it, it can be re-added. From my quick grep there are no CALLOUT_SHAREDLOCK clients in the kernel. - A new callout API function named "callout_drain_async()" has been added. See the updated timeout(9) manual page for a complete description. - Update the callout clients in the "kern/" folder to use the callout API properly, like cv_timedwait(). Previously there was some custom sleepqueue code in the callout subsystem, which has been removed, because we now allow callouts to be protected by spinlocks. This allows us to tear down the callout like done with regular mutexes, and a "td_slpmutex" has been added to "struct thread" to atomically teardown the "td_slpcallout". Further the "TDF_TIMOFAIL" and "SWT_SLEEPQTIMO" states can now be completely removed. Currently they are marked as available and will be cleaned up in a follow up commit. - Bump the __FreeBSD_version to indicate kernel modules need recompilation. - There has been several reports that this patch "seems to squash a serious bug leading to a callout timeout and panic". Kernel build testing: all architectures were built MFC after: 2 weeks Differential Revision: https://reviews.freebsd.org/D1438 Sponsored by: Mellanox Technologies Reviewed by: jhb, adrian, sbruno and emaste
2015-01-15 15:32:30 +00:00
struct thread *td;
Fix a long standing race between sleep queue and thread suspension code. When a thread A is going to sleep, it calls sleepq_catch_signals() to detect any pending signals or thread suspension request, if nothing happens, it returns without holding process lock or scheduler lock, this opens a race window which allows thread B to come in and do process suspension work, however since A is still at running state, thread B can do nothing to A, thread A continues, and puts itself into actually sleeping state, but B has never seen it, and it sits there forever until B is woken up by other threads sometimes later(this can be very long delay or never happen). Fix this bug by forcing sleepq_catch_signals to return with scheduler lock held. Fix sleepq_abort() by passing it an interrupted code, previously, it worked as wakeup_one(), and the interruption can not be identified correctly by sleep queue code when the sleeping thread is resumed. Let thread_suspend_check() returns EINTR or ERESTART, so sleep queue no longer has to use SIGSTOP as a hack to build a return value. Reviewed by: jhb MFC after: 1 week
2006-02-15 23:52:01 +00:00
int rcatch, rvalt, rvals;
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
Major callout subsystem cleanup and rewrite: - Close a migration race where callout_reset() failed to set the CALLOUT_ACTIVE flag. - Callout callback functions are now allowed to be protected by spinlocks. - Switching the callout CPU number cannot always be done on a per-callout basis. See the updated timeout(9) manual page for more information. - The timeout(9) manual page has been updated to reflect how all the functions inside the callout API are working. The manual page has been made function oriented to make it easier to deduce how each of the functions making up the callout API are working without having to first read the whole manual page. Group all functions into a handful of sections which should give a quick top-level overview when the different functions should be used. - The CALLOUT_SHAREDLOCK flag and its functionality has been removed to reduce the complexity in the callout code and to avoid problems about atomically stopping callouts via callout_stop(). If someone needs it, it can be re-added. From my quick grep there are no CALLOUT_SHAREDLOCK clients in the kernel. - A new callout API function named "callout_drain_async()" has been added. See the updated timeout(9) manual page for a complete description. - Update the callout clients in the "kern/" folder to use the callout API properly, like cv_timedwait(). Previously there was some custom sleepqueue code in the callout subsystem, which has been removed, because we now allow callouts to be protected by spinlocks. This allows us to tear down the callout like done with regular mutexes, and a "td_slpmutex" has been added to "struct thread" to atomically teardown the "td_slpcallout". Further the "TDF_TIMOFAIL" and "SWT_SLEEPQTIMO" states can now be completely removed. Currently they are marked as available and will be cleaned up in a follow up commit. - Bump the __FreeBSD_version to indicate kernel modules need recompilation. - There has been several reports that this patch "seems to squash a serious bug leading to a callout timeout and panic". Kernel build testing: all architectures were built MFC after: 2 weeks Differential Revision: https://reviews.freebsd.org/D1438 Sponsored by: Mellanox Technologies Reviewed by: jhb, adrian, sbruno and emaste
2015-01-15 15:32:30 +00:00
td = curthread;
- Pass the priority argument from *sleep() into sleepq and down into sched_sleep(). This removes extra thread_lock() acquisition and allows the scheduler to decide what to do with the static boost. - Change the priority arguments to cv_* to match sleepq/msleep/etc. where 0 means no priority change. Catch -1 in cv_broadcastpri() and convert it to 0 for now. - Set a flag when sleeping in a way that is compatible with swapping since direct priority comparisons are meaningless now. - Add a sysctl to ule, kern.sched.static_boost, that defaults to on which controls the boost behavior. Turning it off gives better performance in some workloads but needs more investigation. - While we're modifying sleepq, change signal and broadcast to both return with the lock held as the lock was held on enter. Reviewed by: jhb, peter
2008-03-12 06:31:06 +00:00
rcatch = sleepq_catch_signals(wchan, pri);
Major callout subsystem cleanup and rewrite: - Close a migration race where callout_reset() failed to set the CALLOUT_ACTIVE flag. - Callout callback functions are now allowed to be protected by spinlocks. - Switching the callout CPU number cannot always be done on a per-callout basis. See the updated timeout(9) manual page for more information. - The timeout(9) manual page has been updated to reflect how all the functions inside the callout API are working. The manual page has been made function oriented to make it easier to deduce how each of the functions making up the callout API are working without having to first read the whole manual page. Group all functions into a handful of sections which should give a quick top-level overview when the different functions should be used. - The CALLOUT_SHAREDLOCK flag and its functionality has been removed to reduce the complexity in the callout code and to avoid problems about atomically stopping callouts via callout_stop(). If someone needs it, it can be re-added. From my quick grep there are no CALLOUT_SHAREDLOCK clients in the kernel. - A new callout API function named "callout_drain_async()" has been added. See the updated timeout(9) manual page for a complete description. - Update the callout clients in the "kern/" folder to use the callout API properly, like cv_timedwait(). Previously there was some custom sleepqueue code in the callout subsystem, which has been removed, because we now allow callouts to be protected by spinlocks. This allows us to tear down the callout like done with regular mutexes, and a "td_slpmutex" has been added to "struct thread" to atomically teardown the "td_slpcallout". Further the "TDF_TIMOFAIL" and "SWT_SLEEPQTIMO" states can now be completely removed. Currently they are marked as available and will be cleaned up in a follow up commit. - Bump the __FreeBSD_version to indicate kernel modules need recompilation. - There has been several reports that this patch "seems to squash a serious bug leading to a callout timeout and panic". Kernel build testing: all architectures were built MFC after: 2 weeks Differential Revision: https://reviews.freebsd.org/D1438 Sponsored by: Mellanox Technologies Reviewed by: jhb, adrian, sbruno and emaste
2015-01-15 15:32:30 +00:00
rvalt = sleepq_check_timeout(td);
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
rvals = sleepq_check_signals();
Major callout subsystem cleanup and rewrite: - Close a migration race where callout_reset() failed to set the CALLOUT_ACTIVE flag. - Callout callback functions are now allowed to be protected by spinlocks. - Switching the callout CPU number cannot always be done on a per-callout basis. See the updated timeout(9) manual page for more information. - The timeout(9) manual page has been updated to reflect how all the functions inside the callout API are working. The manual page has been made function oriented to make it easier to deduce how each of the functions making up the callout API are working without having to first read the whole manual page. Group all functions into a handful of sections which should give a quick top-level overview when the different functions should be used. - The CALLOUT_SHAREDLOCK flag and its functionality has been removed to reduce the complexity in the callout code and to avoid problems about atomically stopping callouts via callout_stop(). If someone needs it, it can be re-added. From my quick grep there are no CALLOUT_SHAREDLOCK clients in the kernel. - A new callout API function named "callout_drain_async()" has been added. See the updated timeout(9) manual page for a complete description. - Update the callout clients in the "kern/" folder to use the callout API properly, like cv_timedwait(). Previously there was some custom sleepqueue code in the callout subsystem, which has been removed, because we now allow callouts to be protected by spinlocks. This allows us to tear down the callout like done with regular mutexes, and a "td_slpmutex" has been added to "struct thread" to atomically teardown the "td_slpcallout". Further the "TDF_TIMOFAIL" and "SWT_SLEEPQTIMO" states can now be completely removed. Currently they are marked as available and will be cleaned up in a follow up commit. - Bump the __FreeBSD_version to indicate kernel modules need recompilation. - There has been several reports that this patch "seems to squash a serious bug leading to a callout timeout and panic". Kernel build testing: all architectures were built MFC after: 2 weeks Differential Revision: https://reviews.freebsd.org/D1438 Sponsored by: Mellanox Technologies Reviewed by: jhb, adrian, sbruno and emaste
2015-01-15 15:32:30 +00:00
thread_unlock(td);
sleepq_stop_timeout(td);
Fix a long standing race between sleep queue and thread suspension code. When a thread A is going to sleep, it calls sleepq_catch_signals() to detect any pending signals or thread suspension request, if nothing happens, it returns without holding process lock or scheduler lock, this opens a race window which allows thread B to come in and do process suspension work, however since A is still at running state, thread B can do nothing to A, thread A continues, and puts itself into actually sleeping state, but B has never seen it, and it sits there forever until B is woken up by other threads sometimes later(this can be very long delay or never happen). Fix this bug by forcing sleepq_catch_signals to return with scheduler lock held. Fix sleepq_abort() by passing it an interrupted code, previously, it worked as wakeup_one(), and the interruption can not be identified correctly by sleep queue code when the sleeping thread is resumed. Let thread_suspend_check() returns EINTR or ERESTART, so sleep queue no longer has to use SIGSTOP as a hack to build a return value. Reviewed by: jhb MFC after: 1 week
2006-02-15 23:52:01 +00:00
if (rcatch)
return (rcatch);
if (rvals)
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
return (rvals);
Fix a long standing race between sleep queue and thread suspension code. When a thread A is going to sleep, it calls sleepq_catch_signals() to detect any pending signals or thread suspension request, if nothing happens, it returns without holding process lock or scheduler lock, this opens a race window which allows thread B to come in and do process suspension work, however since A is still at running state, thread B can do nothing to A, thread A continues, and puts itself into actually sleeping state, but B has never seen it, and it sits there forever until B is woken up by other threads sometimes later(this can be very long delay or never happen). Fix this bug by forcing sleepq_catch_signals to return with scheduler lock held. Fix sleepq_abort() by passing it an interrupted code, previously, it worked as wakeup_one(), and the interruption can not be identified correctly by sleep queue code when the sleeping thread is resumed. Let thread_suspend_check() returns EINTR or ERESTART, so sleep queue no longer has to use SIGSTOP as a hack to build a return value. Reviewed by: jhb MFC after: 1 week
2006-02-15 23:52:01 +00:00
return (rvalt);
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
}
Introduce the new kernel thread called "deadlock resolver". While the name is pretentious, a good explanation of its targets is reported in this 17 months old presentation e-mail: http://lists.freebsd.org/pipermail/freebsd-arch/2008-August/008452.html In order to implement it, the sq_type in sleepqueues is mandatory and not only compiled along with INVARIANTS option. Additively, a new sleepqueue function, sleepq_type() is added, returning the type of the sleepqueue linked to a wchan. Three new sysctls are added in order to configure the thread: debug.deadlkres.slptime_threshold debug.deadlkres.blktime_threshold debug.deadlkres.sleepfreq rappresenting the thresholds for sleep and block time that will lead to a deadlock matching (when exceeded), while the sleepfreq rappresents the number of seconds between 2 consecutive thread runnings. In order to enable the deadlock resolver thread recompile your kernel with the option DEADLKRES. Reviewed by: jeff Tested by: pho, Giovanni Trematerra Sponsored by: Nokia Incorporated, Sandvine Incorporated MFC after: 2 weeks
2010-01-09 01:46:38 +00:00
/*
* Returns the type of sleepqueue given a waitchannel.
*/
int
sleepq_type(void *wchan)
{
struct sleepqueue *sq;
int type;
MPASS(wchan != NULL);
sleepq_lock(wchan);
sq = sleepq_lookup(wchan);
if (sq == NULL) {
sleepq_release(wchan);
return (-1);
}
type = sq->sq_type;
sleepq_release(wchan);
return (type);
}
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
/*
Close a race between sleepq_broadcast() and sleepq_catch_signals(). Specifically, sleepq_broadcast() uses td_slpq for its private pending queue of threads that it is going to wake up after it takes them off the sleep queue. The problem is that if one of the threads is actually not asleep yet, then we can end up with td_slpq being corrupted and/or the thread being made runnable at the wrong time resulting in the td_sleepqueue == NULL assertion failures occasionally reported under heavy load. The fix is to stop being so fancy and ditch the whole pending queue bit. Instead, sleepq_remove_thread() and sleepq_resume_thread() were merged into one function that requires the caller to hold sched_lock. This fixes several places that unlocked sched_lock only to call a function that then locked sched_lock, so even though sched_lock is now held slightly longer, removing the extra lock acquires (1 pair instead of 3 in some cases) probably makes it an overall win if you don't include the fact that it closes a race. This is definitely a 5.4 candidate. PR: kern/79693 Submitted by: Steven Sears stevenjsears at yahoo dot com MFC after: 4 days
2005-04-14 06:30:32 +00:00
* Removes a thread from a sleep queue and makes it
* runnable.
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
*/
If a thread that is swapped out is made runnable, then the setrunnable() routine wakes up proc0 so that proc0 can swap the thread back in. Historically, this has been done by waking up proc0 directly from setrunnable() itself via a wakeup(). When waking up a sleeping thread that was swapped out (the usual case when waking proc0 since only sleeping threads are eligible to be swapped out), this resulted in a bit of recursion (e.g. wakeup() -> setrunnable() -> wakeup()). With sleep queues having separate locks in 6.x and later, this caused a spin lock LOR (sleepq lock -> sched_lock/thread lock -> sleepq lock). An attempt was made to fix this in 7.0 by making the proc0 wakeup use the ithread mechanism for doing the wakeup. However, this required grabbing proc0's thread lock to perform the wakeup. If proc0 was asleep elsewhere in the kernel (e.g. waiting for disk I/O), then this degenerated into the same LOR since the thread lock would be some other sleepq lock. Fix this by deferring the wakeup of the swapper until after the sleepq lock held by the upper layer has been locked. The setrunnable() routine now returns a boolean value to indicate whether or not proc0 needs to be woken up. The end result is that consumers of the sleepq API such as *sleep/wakeup, condition variables, sx locks, and lockmgr, have to wakeup proc0 if they get a non-zero return value from sleepq_abort(), sleepq_broadcast(), or sleepq_signal(). Discussed with: jeff Glanced at by: sam Tested by: Jurgen Weber jurgen - ish com au MFC after: 2 weeks
2008-08-05 20:02:31 +00:00
static int
Close a race between sleepq_broadcast() and sleepq_catch_signals(). Specifically, sleepq_broadcast() uses td_slpq for its private pending queue of threads that it is going to wake up after it takes them off the sleep queue. The problem is that if one of the threads is actually not asleep yet, then we can end up with td_slpq being corrupted and/or the thread being made runnable at the wrong time resulting in the td_sleepqueue == NULL assertion failures occasionally reported under heavy load. The fix is to stop being so fancy and ditch the whole pending queue bit. Instead, sleepq_remove_thread() and sleepq_resume_thread() were merged into one function that requires the caller to hold sched_lock. This fixes several places that unlocked sched_lock only to call a function that then locked sched_lock, so even though sched_lock is now held slightly longer, removing the extra lock acquires (1 pair instead of 3 in some cases) probably makes it an overall win if you don't include the fact that it closes a race. This is definitely a 5.4 candidate. PR: kern/79693 Submitted by: Steven Sears stevenjsears at yahoo dot com MFC after: 4 days
2005-04-14 06:30:32 +00:00
sleepq_resume_thread(struct sleepqueue *sq, struct thread *td, int pri)
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
{
struct sleepqueue_chain *sc;
MPASS(td != NULL);
MPASS(sq->sq_wchan != NULL);
MPASS(td->td_wchan == sq->sq_wchan);
Add second sleep queue so that sx and lockmgr can have separate sleep queues for shared and exclusive acquisitions Submitted by: Attilio Rao Approved by: jhb
2006-12-16 06:54:09 +00:00
MPASS(td->td_sqqueue < NR_SLEEPQS && td->td_sqqueue >= 0);
Commit 2/14 of sched_lock decomposition. - Adapt sleepqueues to the new thread_lock() mechanism. - Delay assigning the sleep queue spinlock as the thread lock until after we've checked for signals. It is illegal for a thread to return in mi_switch() with any lock assigned to td_lock other than the scheduler locks. - Change sleepq_catch_signals() to do the switch if necessary to simplify the callers. - Simplify timeout handling now that locking a sleeping thread has the side-effect of locking the sleepqueue. Some previous races are no longer possible. Tested by: kris, current@ Tested on: i386, amd64, ULE, 4BSD, libthr, libkse, PREEMPTION, etc. Discussed with: kris, attilio, kmacy, jhb, julian, bde (small parts each)
2007-06-04 23:50:56 +00:00
THREAD_LOCK_ASSERT(td, MA_OWNED);
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
sc = SC_LOOKUP(sq->sq_wchan);
mtx_assert(&sc->sc_lock, MA_OWNED);
Implement the DTrace sched provider. This implementation aims to be compatible with the sched provider implemented by Solaris and its open- source derivatives. Full documentation of the sched provider can be found on Oracle's DTrace wiki pages. Note that for compatibility with scripts originally written for Solaris, serveral probes are defined that will never fire. These probes are defined to fire when Solaris-specific features perform certain actions. As these features are not present in FreeBSD, the probes can never fire. Also, I have added a two probes that are not defined in Solaris, lend-pri and load-change. These probes have been added to make it possible to collect schedgraph data with DTrace. Finally, a few probes are defined in Solaris to take a cpuinfo_t * argument. As it was not immediately clear to me how to translate that to FreeBSD, currently those probes are passed NULL in place of a cpuinfo_t *. Sponsored by: Sandvine Incorporated MFC after: 2 weeks
2012-05-15 01:30:25 +00:00
SDT_PROBE2(sched, , , wakeup, td, td->td_proc);
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
/* Remove the thread from the queue. */
In current code, threads performing an interruptible sleep (on both sxlock, via the sx_{s, x}lock_sig() interface, or plain lockmgr), will leave the waiters flag on forcing the owner to do a wakeup even when if the waiter queue is empty. That operation may lead to a deadlock in the case of doing a fake wakeup on the "preferred" (based on the wakeup algorithm) queue while the other queue has real waiters on it, because nobody is going to wakeup the 2nd queue waiters and they will sleep indefinitively. A similar bug, is present, for lockmgr in the case the waiters are sleeping with LK_SLEEPFAIL on. In this case, even if the waiters queue is not empty, the waiters won't progress after being awake but they will just fail, still not taking care of the 2nd queue waiters (as instead the lock owned doing the wakeup would expect). In order to fix this bug in a cheap way (without adding too much locking and complicating too much the semantic) add a sleepqueue interface which does report the actual number of waiters on a specified queue of a waitchannel (sleepq_sleepcnt()) and use it in order to determine if the exclusive waiters (or shared waiters) are actually present on the lockmgr (or sx) before to give them precedence in the wakeup algorithm. This fix alone, however doesn't solve the LK_SLEEPFAIL bug. In order to cope with it, add the tracking of how many exclusive LK_SLEEPFAIL waiters a lockmgr has and if all the waiters on the exclusive waiters queue are LK_SLEEPFAIL just wake both queues. The sleepq_sleepcnt() introduction and ABI breakage require __FreeBSD_version bumping. Reported by: avg, kib, pho Reviewed by: kib Tested by: pho
2009-12-12 21:31:07 +00:00
sq->sq_blockedcnt[td->td_sqqueue]--;
Add second sleep queue so that sx and lockmgr can have separate sleep queues for shared and exclusive acquisitions Submitted by: Attilio Rao Approved by: jhb
2006-12-16 06:54:09 +00:00
TAILQ_REMOVE(&sq->sq_blocked[td->td_sqqueue], td, td_slpq);
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
/*
* Get a sleep queue for this thread. If this is the last waiter,
* use the queue itself and take it out of the chain, otherwise,
* remove a queue from the free list.
*/
if (LIST_EMPTY(&sq->sq_free)) {
td->td_sleepqueue = sq;
#ifdef INVARIANTS
sq->sq_wchan = NULL;
Add two new kernel options to allow rudimentary profiling of the internal hash tables used in the sleep queue and turnstile code. Each option adds a sysctl tree under debug containing the maximum depth of any bucket in the hash table as well as a separate node for each bucket (or chain) containing the current depth and maximum depth for that bucket.
2004-06-29 02:30:12 +00:00
#endif
#ifdef SLEEPQUEUE_PROFILING
sc->sc_depth--;
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
#endif
} else
td->td_sleepqueue = LIST_FIRST(&sq->sq_free);
LIST_REMOVE(td->td_sleepqueue, sq_hash);
Split sleepq_wakeup_thread() into two functions. sleepq_remove_thread() removes a specific thread from a sleep queue. sleepq_resume_thread() resumes scheduling of a thread that has been previously removed from a sleep queue. - sleepq_catch_signals() just removes a thread from the queue it was just added to when a pending signal is found. - sleepq_signal() and sleepq_broadcast() remove threads from a queue, drop the queue lock, and then resume all the previously removed threads. This doesn't completely fix the sched_lock <-> sleepq chain LOR, but it makes it a little better as we no longer call setrunnble() with a sleep queue lock held meaning if setrunnable() tries to wakeup the swapper we don't try to lock two sleep queue chains at the same time.
2004-05-13 20:00:43 +00:00
td->td_wmesg = NULL;
td->td_wchan = NULL;
Rework the handling of stop signals in the NFS client. The changes in 195702, 195703, and 195821 prevented a thread from suspending while holding locks inside of NFS by forcing the thread to fail sleeps with EINTR or ERESTART but defer the thread suspension to the user boundary. However, this had the effect that stopping a process during an NFS request could abort the request and trigger EINTR errors that were visible to userland processes (previously the thread would have suspended and completed the request once it was resumed). This change instead effectively masks stop signals while in the NFS client. It uses the existing TDF_SBDRY flag to effect this since SIGSTOP cannot be masked directly. Also, instead of setting PBDRY on individual sleeps, the NFS client now sets the TDF_SBDRY flag around each NFS request and stop signals are masked for all sleeps during that region (the previous change missed sleeps in lockmgr locks). The end result is that stop signals sent to threads performing an NFS request are completely ignored until after the NFS request has finished processing and the thread prepares to return to userland. This restores the behavior of stop signals being transparent to userland processes while still preventing threads from suspending while holding NFS locks. Reviewed by: kib MFC after: 1 month
2013-02-06 17:06:51 +00:00
td->td_flags &= ~TDF_SINTR;
Split sleepq_wakeup_thread() into two functions. sleepq_remove_thread() removes a specific thread from a sleep queue. sleepq_resume_thread() resumes scheduling of a thread that has been previously removed from a sleep queue. - sleepq_catch_signals() just removes a thread from the queue it was just added to when a pending signal is found. - sleepq_signal() and sleepq_broadcast() remove threads from a queue, drop the queue lock, and then resume all the previously removed threads. This doesn't completely fix the sched_lock <-> sleepq chain LOR, but it makes it a little better as we no longer call setrunnble() with a sleep queue lock held meaning if setrunnable() tries to wakeup the swapper we don't try to lock two sleep queue chains at the same time.
2004-05-13 20:00:43 +00:00
Fixed some common printf format errors. Don't assume that "struct foo *" is "void *" (it isn't) or that the default promotion of pid_t is int. Instead, assume that casting "struct foo *" to "void *" and printing the result with %p is useful, and that all pid_t's are representable as longs. Fixed some minor style bugs (mainly spelling errors in comments).
2004-05-14 20:51:42 +00:00
CTR3(KTR_PROC, "sleepq_wakeup: thread %p (pid %ld, %s)",
generally we are interested in what thread did something as opposed to what process. Since threads by default have teh name of the process unless over-written with more useful information, just print the thread name instead.
2007-11-14 06:21:24 +00:00
(void *)td, (long)td->td_proc->p_pid, td->td_name);
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
/* Adjust priority if requested. */
- Pass the priority argument from *sleep() into sleepq and down into sched_sleep(). This removes extra thread_lock() acquisition and allows the scheduler to decide what to do with the static boost. - Change the priority arguments to cv_* to match sleepq/msleep/etc. where 0 means no priority change. Catch -1 in cv_broadcastpri() and convert it to 0 for now. - Set a flag when sleeping in a way that is compatible with swapping since direct priority comparisons are meaningless now. - Add a sysctl to ule, kern.sched.static_boost, that defaults to on which controls the boost behavior. Turning it off gives better performance in some workloads but needs more investigation. - While we're modifying sleepq, change signal and broadcast to both return with the lock held as the lock was held on enter. Reviewed by: jhb, peter
2008-03-12 06:31:06 +00:00
MPASS(pri == 0 || (pri >= PRI_MIN && pri <= PRI_MAX));
Rework realtime priority support: - Move the realtime priority range up above kernel sleep priorities and just below interrupt thread priorities. - Contract the interrupt and kernel sleep priority ranges a bit so that the timesharing priority band can be increased. The new timeshare range is now slightly larger than the old realtime + timeshare ranges. - Change the ULE scheduler to no longer use realtime priorities for interactive threads. Instead, the larger timeshare range is now split into separate subranges for interactive and non-interactive ("batch") threads. The end result is that interactive threads and non-interactive threads still use the same priority ranges as before, but realtime threads now have a separate, dedicated priority range. - Do not modify the priority of non-timeshare threads in sched_sleep() or via cv_broadcastpri(). Realtime and idle priority threads will no longer have their priorities affected by sleeping in the kernel. Reviewed by: jeff
2011-01-14 17:06:54 +00:00
if (pri != 0 && td->td_priority > pri &&
PRI_BASE(td->td_pri_class) == PRI_TIMESHARE)
Directly modifying the priority of a thread that may be on the runqueue can break the sorting order of the ksegp run queue. Tested by: pho Reviewed by: jhb, julian Approved by: sam (mentor) MFC: ASAP
2004-10-12 16:31:23 +00:00
sched_prio(td, pri);
Don't bother calling setrunnable() and clearing the sleeping flag in sleepq_resume_thread() if the thread isn't asleep.
2008-11-04 19:13:53 +00:00
/*
* Note that thread td might not be sleeping if it is running
* sleepq_catch_signals() on another CPU or is blocked on its
* proc lock to check signals. There's no need to mark the
* thread runnable in that case.
*/
if (TD_IS_SLEEPING(td)) {
TD_CLR_SLEEPING(td);
return (setrunnable(td));
}
return (0);
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
}
- Convert turnstiles and sleepqueus to use UMA. This provides a modest speedup and will be more useful after each gains a spinlock in the impending thread_lock() commit. - Move initialization and asserts into init/fini routines. fini routines are only needed in the INVARIANTS case for now. Submitted by: Attilio Rao <attilio@FreeBSD.org> Tested by: kris, jeff
2007-05-18 06:32:24 +00:00
#ifdef INVARIANTS
/*
* UMA zone item deallocator.
*/
static void
sleepq_dtor(void *mem, int size, void *arg)
{
struct sleepqueue *sq;
int i;
sq = mem;
In current code, threads performing an interruptible sleep (on both sxlock, via the sx_{s, x}lock_sig() interface, or plain lockmgr), will leave the waiters flag on forcing the owner to do a wakeup even when if the waiter queue is empty. That operation may lead to a deadlock in the case of doing a fake wakeup on the "preferred" (based on the wakeup algorithm) queue while the other queue has real waiters on it, because nobody is going to wakeup the 2nd queue waiters and they will sleep indefinitively. A similar bug, is present, for lockmgr in the case the waiters are sleeping with LK_SLEEPFAIL on. In this case, even if the waiters queue is not empty, the waiters won't progress after being awake but they will just fail, still not taking care of the 2nd queue waiters (as instead the lock owned doing the wakeup would expect). In order to fix this bug in a cheap way (without adding too much locking and complicating too much the semantic) add a sleepqueue interface which does report the actual number of waiters on a specified queue of a waitchannel (sleepq_sleepcnt()) and use it in order to determine if the exclusive waiters (or shared waiters) are actually present on the lockmgr (or sx) before to give them precedence in the wakeup algorithm. This fix alone, however doesn't solve the LK_SLEEPFAIL bug. In order to cope with it, add the tracking of how many exclusive LK_SLEEPFAIL waiters a lockmgr has and if all the waiters on the exclusive waiters queue are LK_SLEEPFAIL just wake both queues. The sleepq_sleepcnt() introduction and ABI breakage require __FreeBSD_version bumping. Reported by: avg, kib, pho Reviewed by: kib Tested by: pho
2009-12-12 21:31:07 +00:00
for (i = 0; i < NR_SLEEPQS; i++) {
- Convert turnstiles and sleepqueus to use UMA. This provides a modest speedup and will be more useful after each gains a spinlock in the impending thread_lock() commit. - Move initialization and asserts into init/fini routines. fini routines are only needed in the INVARIANTS case for now. Submitted by: Attilio Rao <attilio@FreeBSD.org> Tested by: kris, jeff
2007-05-18 06:32:24 +00:00
MPASS(TAILQ_EMPTY(&sq->sq_blocked[i]));
In current code, threads performing an interruptible sleep (on both sxlock, via the sx_{s, x}lock_sig() interface, or plain lockmgr), will leave the waiters flag on forcing the owner to do a wakeup even when if the waiter queue is empty. That operation may lead to a deadlock in the case of doing a fake wakeup on the "preferred" (based on the wakeup algorithm) queue while the other queue has real waiters on it, because nobody is going to wakeup the 2nd queue waiters and they will sleep indefinitively. A similar bug, is present, for lockmgr in the case the waiters are sleeping with LK_SLEEPFAIL on. In this case, even if the waiters queue is not empty, the waiters won't progress after being awake but they will just fail, still not taking care of the 2nd queue waiters (as instead the lock owned doing the wakeup would expect). In order to fix this bug in a cheap way (without adding too much locking and complicating too much the semantic) add a sleepqueue interface which does report the actual number of waiters on a specified queue of a waitchannel (sleepq_sleepcnt()) and use it in order to determine if the exclusive waiters (or shared waiters) are actually present on the lockmgr (or sx) before to give them precedence in the wakeup algorithm. This fix alone, however doesn't solve the LK_SLEEPFAIL bug. In order to cope with it, add the tracking of how many exclusive LK_SLEEPFAIL waiters a lockmgr has and if all the waiters on the exclusive waiters queue are LK_SLEEPFAIL just wake both queues. The sleepq_sleepcnt() introduction and ABI breakage require __FreeBSD_version bumping. Reported by: avg, kib, pho Reviewed by: kib Tested by: pho
2009-12-12 21:31:07 +00:00
MPASS(sq->sq_blockedcnt[i] == 0);
}
- Convert turnstiles and sleepqueus to use UMA. This provides a modest speedup and will be more useful after each gains a spinlock in the impending thread_lock() commit. - Move initialization and asserts into init/fini routines. fini routines are only needed in the INVARIANTS case for now. Submitted by: Attilio Rao <attilio@FreeBSD.org> Tested by: kris, jeff
2007-05-18 06:32:24 +00:00
}
#endif
/*
* UMA zone item initializer.
*/
static int
sleepq_init(void *mem, int size, int flags)
{
struct sleepqueue *sq;
int i;
bzero(mem, size);
sq = mem;
In current code, threads performing an interruptible sleep (on both sxlock, via the sx_{s, x}lock_sig() interface, or plain lockmgr), will leave the waiters flag on forcing the owner to do a wakeup even when if the waiter queue is empty. That operation may lead to a deadlock in the case of doing a fake wakeup on the "preferred" (based on the wakeup algorithm) queue while the other queue has real waiters on it, because nobody is going to wakeup the 2nd queue waiters and they will sleep indefinitively. A similar bug, is present, for lockmgr in the case the waiters are sleeping with LK_SLEEPFAIL on. In this case, even if the waiters queue is not empty, the waiters won't progress after being awake but they will just fail, still not taking care of the 2nd queue waiters (as instead the lock owned doing the wakeup would expect). In order to fix this bug in a cheap way (without adding too much locking and complicating too much the semantic) add a sleepqueue interface which does report the actual number of waiters on a specified queue of a waitchannel (sleepq_sleepcnt()) and use it in order to determine if the exclusive waiters (or shared waiters) are actually present on the lockmgr (or sx) before to give them precedence in the wakeup algorithm. This fix alone, however doesn't solve the LK_SLEEPFAIL bug. In order to cope with it, add the tracking of how many exclusive LK_SLEEPFAIL waiters a lockmgr has and if all the waiters on the exclusive waiters queue are LK_SLEEPFAIL just wake both queues. The sleepq_sleepcnt() introduction and ABI breakage require __FreeBSD_version bumping. Reported by: avg, kib, pho Reviewed by: kib Tested by: pho
2009-12-12 21:31:07 +00:00
for (i = 0; i < NR_SLEEPQS; i++) {
- Convert turnstiles and sleepqueus to use UMA. This provides a modest speedup and will be more useful after each gains a spinlock in the impending thread_lock() commit. - Move initialization and asserts into init/fini routines. fini routines are only needed in the INVARIANTS case for now. Submitted by: Attilio Rao <attilio@FreeBSD.org> Tested by: kris, jeff
2007-05-18 06:32:24 +00:00
TAILQ_INIT(&sq->sq_blocked[i]);
In current code, threads performing an interruptible sleep (on both sxlock, via the sx_{s, x}lock_sig() interface, or plain lockmgr), will leave the waiters flag on forcing the owner to do a wakeup even when if the waiter queue is empty. That operation may lead to a deadlock in the case of doing a fake wakeup on the "preferred" (based on the wakeup algorithm) queue while the other queue has real waiters on it, because nobody is going to wakeup the 2nd queue waiters and they will sleep indefinitively. A similar bug, is present, for lockmgr in the case the waiters are sleeping with LK_SLEEPFAIL on. In this case, even if the waiters queue is not empty, the waiters won't progress after being awake but they will just fail, still not taking care of the 2nd queue waiters (as instead the lock owned doing the wakeup would expect). In order to fix this bug in a cheap way (without adding too much locking and complicating too much the semantic) add a sleepqueue interface which does report the actual number of waiters on a specified queue of a waitchannel (sleepq_sleepcnt()) and use it in order to determine if the exclusive waiters (or shared waiters) are actually present on the lockmgr (or sx) before to give them precedence in the wakeup algorithm. This fix alone, however doesn't solve the LK_SLEEPFAIL bug. In order to cope with it, add the tracking of how many exclusive LK_SLEEPFAIL waiters a lockmgr has and if all the waiters on the exclusive waiters queue are LK_SLEEPFAIL just wake both queues. The sleepq_sleepcnt() introduction and ABI breakage require __FreeBSD_version bumping. Reported by: avg, kib, pho Reviewed by: kib Tested by: pho
2009-12-12 21:31:07 +00:00
sq->sq_blockedcnt[i] = 0;
}
- Convert turnstiles and sleepqueus to use UMA. This provides a modest speedup and will be more useful after each gains a spinlock in the impending thread_lock() commit. - Move initialization and asserts into init/fini routines. fini routines are only needed in the INVARIANTS case for now. Submitted by: Attilio Rao <attilio@FreeBSD.org> Tested by: kris, jeff
2007-05-18 06:32:24 +00:00
LIST_INIT(&sq->sq_free);
return (0);
}
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
/*
* Find the highest priority thread sleeping on a wait channel and resume it.
*/
If a thread that is swapped out is made runnable, then the setrunnable() routine wakes up proc0 so that proc0 can swap the thread back in. Historically, this has been done by waking up proc0 directly from setrunnable() itself via a wakeup(). When waking up a sleeping thread that was swapped out (the usual case when waking proc0 since only sleeping threads are eligible to be swapped out), this resulted in a bit of recursion (e.g. wakeup() -> setrunnable() -> wakeup()). With sleep queues having separate locks in 6.x and later, this caused a spin lock LOR (sleepq lock -> sched_lock/thread lock -> sleepq lock). An attempt was made to fix this in 7.0 by making the proc0 wakeup use the ithread mechanism for doing the wakeup. However, this required grabbing proc0's thread lock to perform the wakeup. If proc0 was asleep elsewhere in the kernel (e.g. waiting for disk I/O), then this degenerated into the same LOR since the thread lock would be some other sleepq lock. Fix this by deferring the wakeup of the swapper until after the sleepq lock held by the upper layer has been locked. The setrunnable() routine now returns a boolean value to indicate whether or not proc0 needs to be woken up. The end result is that consumers of the sleepq API such as *sleep/wakeup, condition variables, sx locks, and lockmgr, have to wakeup proc0 if they get a non-zero return value from sleepq_abort(), sleepq_broadcast(), or sleepq_signal(). Discussed with: jeff Glanced at by: sam Tested by: Jurgen Weber jurgen - ish com au MFC after: 2 weeks
2008-08-05 20:02:31 +00:00
int
Add second sleep queue so that sx and lockmgr can have separate sleep queues for shared and exclusive acquisitions Submitted by: Attilio Rao Approved by: jhb
2006-12-16 06:54:09 +00:00
sleepq_signal(void *wchan, int flags, int pri, int queue)
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
{
struct sleepqueue *sq;
- Store threads on sleep queues in FIFO order rather than sorted by priority. The sleep queues don't get updated when the priority of threads changes, so sleepq_signal() might not always wakeup the highest priority thread. Updating the queues when thread priorities change cannot be easily done due to lock orders, so instead we do an O(n) walk of the queue for a sleepq_signal() operation instead of O(1). On the other hand, adding a thread to a sleep queue now goes from O(n) to O(1) so it ends up as an even tradeoff. The correctness here with regards to priorities is actually fairly important. msleep() gives interactive threads their priority "boost" after they are placed on the queue, but before this fix that "boost" wasn't used to determine the highest priority thread that sleepq_signal() awoke. - Fix up some comments. Inspired by: ups, bde
2004-11-05 20:19:58 +00:00
struct thread *td, *besttd;
If a thread that is swapped out is made runnable, then the setrunnable() routine wakes up proc0 so that proc0 can swap the thread back in. Historically, this has been done by waking up proc0 directly from setrunnable() itself via a wakeup(). When waking up a sleeping thread that was swapped out (the usual case when waking proc0 since only sleeping threads are eligible to be swapped out), this resulted in a bit of recursion (e.g. wakeup() -> setrunnable() -> wakeup()). With sleep queues having separate locks in 6.x and later, this caused a spin lock LOR (sleepq lock -> sched_lock/thread lock -> sleepq lock). An attempt was made to fix this in 7.0 by making the proc0 wakeup use the ithread mechanism for doing the wakeup. However, this required grabbing proc0's thread lock to perform the wakeup. If proc0 was asleep elsewhere in the kernel (e.g. waiting for disk I/O), then this degenerated into the same LOR since the thread lock would be some other sleepq lock. Fix this by deferring the wakeup of the swapper until after the sleepq lock held by the upper layer has been locked. The setrunnable() routine now returns a boolean value to indicate whether or not proc0 needs to be woken up. The end result is that consumers of the sleepq API such as *sleep/wakeup, condition variables, sx locks, and lockmgr, have to wakeup proc0 if they get a non-zero return value from sleepq_abort(), sleepq_broadcast(), or sleepq_signal(). Discussed with: jeff Glanced at by: sam Tested by: Jurgen Weber jurgen - ish com au MFC after: 2 weeks
2008-08-05 20:02:31 +00:00
int wakeup_swapper;
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
CTR2(KTR_PROC, "sleepq_signal(%p, %d)", wchan, flags);
KASSERT(wchan != NULL, ("%s: invalid NULL wait channel", __func__));
Add second sleep queue so that sx and lockmgr can have separate sleep queues for shared and exclusive acquisitions Submitted by: Attilio Rao Approved by: jhb
2006-12-16 06:54:09 +00:00
MPASS((queue >= 0) && (queue < NR_SLEEPQS));
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
sq = sleepq_lookup(wchan);
Commit 2/14 of sched_lock decomposition. - Adapt sleepqueues to the new thread_lock() mechanism. - Delay assigning the sleep queue spinlock as the thread lock until after we've checked for signals. It is illegal for a thread to return in mi_switch() with any lock assigned to td_lock other than the scheduler locks. - Change sleepq_catch_signals() to do the switch if necessary to simplify the callers. - Simplify timeout handling now that locking a sleeping thread has the side-effect of locking the sleepqueue. Some previous races are no longer possible. Tested by: kris, current@ Tested on: i386, amd64, ULE, 4BSD, libthr, libkse, PREEMPTION, etc. Discussed with: kris, attilio, kmacy, jhb, julian, bde (small parts each)
2007-06-04 23:50:56 +00:00
if (sq == NULL)
If a thread that is swapped out is made runnable, then the setrunnable() routine wakes up proc0 so that proc0 can swap the thread back in. Historically, this has been done by waking up proc0 directly from setrunnable() itself via a wakeup(). When waking up a sleeping thread that was swapped out (the usual case when waking proc0 since only sleeping threads are eligible to be swapped out), this resulted in a bit of recursion (e.g. wakeup() -> setrunnable() -> wakeup()). With sleep queues having separate locks in 6.x and later, this caused a spin lock LOR (sleepq lock -> sched_lock/thread lock -> sleepq lock). An attempt was made to fix this in 7.0 by making the proc0 wakeup use the ithread mechanism for doing the wakeup. However, this required grabbing proc0's thread lock to perform the wakeup. If proc0 was asleep elsewhere in the kernel (e.g. waiting for disk I/O), then this degenerated into the same LOR since the thread lock would be some other sleepq lock. Fix this by deferring the wakeup of the swapper until after the sleepq lock held by the upper layer has been locked. The setrunnable() routine now returns a boolean value to indicate whether or not proc0 needs to be woken up. The end result is that consumers of the sleepq API such as *sleep/wakeup, condition variables, sx locks, and lockmgr, have to wakeup proc0 if they get a non-zero return value from sleepq_abort(), sleepq_broadcast(), or sleepq_signal(). Discussed with: jeff Glanced at by: sam Tested by: Jurgen Weber jurgen - ish com au MFC after: 2 weeks
2008-08-05 20:02:31 +00:00
return (0);
Now that the return value semantics of cv's for multithreaded processes have been unified with that of msleep(9), further refine the sleepq interface and consolidate some duplicated code: - Move the pre-sleep checks for theaded processes into a thread_sleep_check() function in kern_thread.c. - Move all handling of TDF_SINTR to be internal to subr_sleepqueue.c. Specifically, if a thread is awakened by something other than a signal while checking for signals before going to sleep, clear TDF_SINTR in sleepq_catch_signals(). This removes a sched_lock lock/unlock combo in that edge case during an interruptible sleep. Also, fix sleepq_check_signals() to properly handle the condition if TDF_SINTR is clear rather than requiring the callers of the sleepq API to notice this edge case and call a non-_sig variant of sleepq_wait(). - Clarify the flags arguments to sleepq_add(), sleepq_signal() and sleepq_broadcast() by creating an explicit submask for sleepq types. Also, add an explicit SLEEPQ_MSLEEP type rather than a magic number of 0. Also, add a SLEEPQ_INTERRUPTIBLE flag for use with sleepq_add() and move the setting of TDF_SINTR to sleepq_add() if this flag is set rather than sleepq_catch_signals(). Note that it is the caller's responsibility to ensure that sleepq_catch_signals() is called if and only if this flag is passed to the preceeding sleepq_add(). Note that this also removes a sched_lock lock/unlock pair from sleepq_catch_signals(). It also ensures that for an interruptible sleep, TDF_SINTR is always set when TD_ON_SLEEPQ() is true.
2004-08-19 11:31:42 +00:00
KASSERT(sq->sq_type == (flags & SLEEPQ_TYPE),
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
("%s: mismatch between sleep/wakeup and cv_*", __func__));
Split sleepq_wakeup_thread() into two functions. sleepq_remove_thread() removes a specific thread from a sleep queue. sleepq_resume_thread() resumes scheduling of a thread that has been previously removed from a sleep queue. - sleepq_catch_signals() just removes a thread from the queue it was just added to when a pending signal is found. - sleepq_signal() and sleepq_broadcast() remove threads from a queue, drop the queue lock, and then resume all the previously removed threads. This doesn't completely fix the sched_lock <-> sleepq chain LOR, but it makes it a little better as we no longer call setrunnble() with a sleep queue lock held meaning if setrunnable() tries to wakeup the swapper we don't try to lock two sleep queue chains at the same time.
2004-05-13 20:00:43 +00:00
- Store threads on sleep queues in FIFO order rather than sorted by priority. The sleep queues don't get updated when the priority of threads changes, so sleepq_signal() might not always wakeup the highest priority thread. Updating the queues when thread priorities change cannot be easily done due to lock orders, so instead we do an O(n) walk of the queue for a sleepq_signal() operation instead of O(1). On the other hand, adding a thread to a sleep queue now goes from O(n) to O(1) so it ends up as an even tradeoff. The correctness here with regards to priorities is actually fairly important. msleep() gives interactive threads their priority "boost" after they are placed on the queue, but before this fix that "boost" wasn't used to determine the highest priority thread that sleepq_signal() awoke. - Fix up some comments. Inspired by: ups, bde
2004-11-05 20:19:58 +00:00
/*
* Find the highest priority thread on the queue. If there is a
* tie, use the thread that first appears in the queue as it has
* been sleeping the longest since threads are always added to
* the tail of sleep queues.
*/
besttd = NULL;
Add second sleep queue so that sx and lockmgr can have separate sleep queues for shared and exclusive acquisitions Submitted by: Attilio Rao Approved by: jhb
2006-12-16 06:54:09 +00:00
TAILQ_FOREACH(td, &sq->sq_blocked[queue], td_slpq) {
- Store threads on sleep queues in FIFO order rather than sorted by priority. The sleep queues don't get updated when the priority of threads changes, so sleepq_signal() might not always wakeup the highest priority thread. Updating the queues when thread priorities change cannot be easily done due to lock orders, so instead we do an O(n) walk of the queue for a sleepq_signal() operation instead of O(1). On the other hand, adding a thread to a sleep queue now goes from O(n) to O(1) so it ends up as an even tradeoff. The correctness here with regards to priorities is actually fairly important. msleep() gives interactive threads their priority "boost" after they are placed on the queue, but before this fix that "boost" wasn't used to determine the highest priority thread that sleepq_signal() awoke. - Fix up some comments. Inspired by: ups, bde
2004-11-05 20:19:58 +00:00
if (besttd == NULL || td->td_priority < besttd->td_priority)
besttd = td;
}
MPASS(besttd != NULL);
Commit 2/14 of sched_lock decomposition. - Adapt sleepqueues to the new thread_lock() mechanism. - Delay assigning the sleep queue spinlock as the thread lock until after we've checked for signals. It is illegal for a thread to return in mi_switch() with any lock assigned to td_lock other than the scheduler locks. - Change sleepq_catch_signals() to do the switch if necessary to simplify the callers. - Simplify timeout handling now that locking a sleeping thread has the side-effect of locking the sleepqueue. Some previous races are no longer possible. Tested by: kris, current@ Tested on: i386, amd64, ULE, 4BSD, libthr, libkse, PREEMPTION, etc. Discussed with: kris, attilio, kmacy, jhb, julian, bde (small parts each)
2007-06-04 23:50:56 +00:00
thread_lock(besttd);
If a thread that is swapped out is made runnable, then the setrunnable() routine wakes up proc0 so that proc0 can swap the thread back in. Historically, this has been done by waking up proc0 directly from setrunnable() itself via a wakeup(). When waking up a sleeping thread that was swapped out (the usual case when waking proc0 since only sleeping threads are eligible to be swapped out), this resulted in a bit of recursion (e.g. wakeup() -> setrunnable() -> wakeup()). With sleep queues having separate locks in 6.x and later, this caused a spin lock LOR (sleepq lock -> sched_lock/thread lock -> sleepq lock). An attempt was made to fix this in 7.0 by making the proc0 wakeup use the ithread mechanism for doing the wakeup. However, this required grabbing proc0's thread lock to perform the wakeup. If proc0 was asleep elsewhere in the kernel (e.g. waiting for disk I/O), then this degenerated into the same LOR since the thread lock would be some other sleepq lock. Fix this by deferring the wakeup of the swapper until after the sleepq lock held by the upper layer has been locked. The setrunnable() routine now returns a boolean value to indicate whether or not proc0 needs to be woken up. The end result is that consumers of the sleepq API such as *sleep/wakeup, condition variables, sx locks, and lockmgr, have to wakeup proc0 if they get a non-zero return value from sleepq_abort(), sleepq_broadcast(), or sleepq_signal(). Discussed with: jeff Glanced at by: sam Tested by: Jurgen Weber jurgen - ish com au MFC after: 2 weeks
2008-08-05 20:02:31 +00:00
wakeup_swapper = sleepq_resume_thread(sq, besttd, pri);
Commit 2/14 of sched_lock decomposition. - Adapt sleepqueues to the new thread_lock() mechanism. - Delay assigning the sleep queue spinlock as the thread lock until after we've checked for signals. It is illegal for a thread to return in mi_switch() with any lock assigned to td_lock other than the scheduler locks. - Change sleepq_catch_signals() to do the switch if necessary to simplify the callers. - Simplify timeout handling now that locking a sleeping thread has the side-effect of locking the sleepqueue. Some previous races are no longer possible. Tested by: kris, current@ Tested on: i386, amd64, ULE, 4BSD, libthr, libkse, PREEMPTION, etc. Discussed with: kris, attilio, kmacy, jhb, julian, bde (small parts each)
2007-06-04 23:50:56 +00:00
thread_unlock(besttd);
If a thread that is swapped out is made runnable, then the setrunnable() routine wakes up proc0 so that proc0 can swap the thread back in. Historically, this has been done by waking up proc0 directly from setrunnable() itself via a wakeup(). When waking up a sleeping thread that was swapped out (the usual case when waking proc0 since only sleeping threads are eligible to be swapped out), this resulted in a bit of recursion (e.g. wakeup() -> setrunnable() -> wakeup()). With sleep queues having separate locks in 6.x and later, this caused a spin lock LOR (sleepq lock -> sched_lock/thread lock -> sleepq lock). An attempt was made to fix this in 7.0 by making the proc0 wakeup use the ithread mechanism for doing the wakeup. However, this required grabbing proc0's thread lock to perform the wakeup. If proc0 was asleep elsewhere in the kernel (e.g. waiting for disk I/O), then this degenerated into the same LOR since the thread lock would be some other sleepq lock. Fix this by deferring the wakeup of the swapper until after the sleepq lock held by the upper layer has been locked. The setrunnable() routine now returns a boolean value to indicate whether or not proc0 needs to be woken up. The end result is that consumers of the sleepq API such as *sleep/wakeup, condition variables, sx locks, and lockmgr, have to wakeup proc0 if they get a non-zero return value from sleepq_abort(), sleepq_broadcast(), or sleepq_signal(). Discussed with: jeff Glanced at by: sam Tested by: Jurgen Weber jurgen - ish com au MFC after: 2 weeks
2008-08-05 20:02:31 +00:00
return (wakeup_swapper);
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
}
/*
* Resume all threads sleeping on a specified wait channel.
*/
If a thread that is swapped out is made runnable, then the setrunnable() routine wakes up proc0 so that proc0 can swap the thread back in. Historically, this has been done by waking up proc0 directly from setrunnable() itself via a wakeup(). When waking up a sleeping thread that was swapped out (the usual case when waking proc0 since only sleeping threads are eligible to be swapped out), this resulted in a bit of recursion (e.g. wakeup() -> setrunnable() -> wakeup()). With sleep queues having separate locks in 6.x and later, this caused a spin lock LOR (sleepq lock -> sched_lock/thread lock -> sleepq lock). An attempt was made to fix this in 7.0 by making the proc0 wakeup use the ithread mechanism for doing the wakeup. However, this required grabbing proc0's thread lock to perform the wakeup. If proc0 was asleep elsewhere in the kernel (e.g. waiting for disk I/O), then this degenerated into the same LOR since the thread lock would be some other sleepq lock. Fix this by deferring the wakeup of the swapper until after the sleepq lock held by the upper layer has been locked. The setrunnable() routine now returns a boolean value to indicate whether or not proc0 needs to be woken up. The end result is that consumers of the sleepq API such as *sleep/wakeup, condition variables, sx locks, and lockmgr, have to wakeup proc0 if they get a non-zero return value from sleepq_abort(), sleepq_broadcast(), or sleepq_signal(). Discussed with: jeff Glanced at by: sam Tested by: Jurgen Weber jurgen - ish com au MFC after: 2 weeks
2008-08-05 20:02:31 +00:00
int
Add second sleep queue so that sx and lockmgr can have separate sleep queues for shared and exclusive acquisitions Submitted by: Attilio Rao Approved by: jhb
2006-12-16 06:54:09 +00:00
sleepq_broadcast(void *wchan, int flags, int pri, int queue)
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
{
struct sleepqueue *sq;
Close a race in sleepq_broadcast() where the sleepq could be reused after it had been assigned to the last sleeping thread. That thread might have started running on another CPU and have reused that sleep queue. Fix it by just walking the thread queue using TAILQ_FOREACH_SAFE() rather than a while loop. PR: amd64/124200 Discovered by: tegge Tested by: benjsc MFC after: 1 week
2008-09-08 19:44:57 +00:00
struct thread *td, *tdn;
If a thread that is swapped out is made runnable, then the setrunnable() routine wakes up proc0 so that proc0 can swap the thread back in. Historically, this has been done by waking up proc0 directly from setrunnable() itself via a wakeup(). When waking up a sleeping thread that was swapped out (the usual case when waking proc0 since only sleeping threads are eligible to be swapped out), this resulted in a bit of recursion (e.g. wakeup() -> setrunnable() -> wakeup()). With sleep queues having separate locks in 6.x and later, this caused a spin lock LOR (sleepq lock -> sched_lock/thread lock -> sleepq lock). An attempt was made to fix this in 7.0 by making the proc0 wakeup use the ithread mechanism for doing the wakeup. However, this required grabbing proc0's thread lock to perform the wakeup. If proc0 was asleep elsewhere in the kernel (e.g. waiting for disk I/O), then this degenerated into the same LOR since the thread lock would be some other sleepq lock. Fix this by deferring the wakeup of the swapper until after the sleepq lock held by the upper layer has been locked. The setrunnable() routine now returns a boolean value to indicate whether or not proc0 needs to be woken up. The end result is that consumers of the sleepq API such as *sleep/wakeup, condition variables, sx locks, and lockmgr, have to wakeup proc0 if they get a non-zero return value from sleepq_abort(), sleepq_broadcast(), or sleepq_signal(). Discussed with: jeff Glanced at by: sam Tested by: Jurgen Weber jurgen - ish com au MFC after: 2 weeks
2008-08-05 20:02:31 +00:00
int wakeup_swapper;
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
CTR2(KTR_PROC, "sleepq_broadcast(%p, %d)", wchan, flags);
KASSERT(wchan != NULL, ("%s: invalid NULL wait channel", __func__));
Add second sleep queue so that sx and lockmgr can have separate sleep queues for shared and exclusive acquisitions Submitted by: Attilio Rao Approved by: jhb
2006-12-16 06:54:09 +00:00
MPASS((queue >= 0) && (queue < NR_SLEEPQS));
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
sq = sleepq_lookup(wchan);
- Pass the priority argument from *sleep() into sleepq and down into sched_sleep(). This removes extra thread_lock() acquisition and allows the scheduler to decide what to do with the static boost. - Change the priority arguments to cv_* to match sleepq/msleep/etc. where 0 means no priority change. Catch -1 in cv_broadcastpri() and convert it to 0 for now. - Set a flag when sleeping in a way that is compatible with swapping since direct priority comparisons are meaningless now. - Add a sysctl to ule, kern.sched.static_boost, that defaults to on which controls the boost behavior. Turning it off gives better performance in some workloads but needs more investigation. - While we're modifying sleepq, change signal and broadcast to both return with the lock held as the lock was held on enter. Reviewed by: jhb, peter
2008-03-12 06:31:06 +00:00
if (sq == NULL)
If a thread that is swapped out is made runnable, then the setrunnable() routine wakes up proc0 so that proc0 can swap the thread back in. Historically, this has been done by waking up proc0 directly from setrunnable() itself via a wakeup(). When waking up a sleeping thread that was swapped out (the usual case when waking proc0 since only sleeping threads are eligible to be swapped out), this resulted in a bit of recursion (e.g. wakeup() -> setrunnable() -> wakeup()). With sleep queues having separate locks in 6.x and later, this caused a spin lock LOR (sleepq lock -> sched_lock/thread lock -> sleepq lock). An attempt was made to fix this in 7.0 by making the proc0 wakeup use the ithread mechanism for doing the wakeup. However, this required grabbing proc0's thread lock to perform the wakeup. If proc0 was asleep elsewhere in the kernel (e.g. waiting for disk I/O), then this degenerated into the same LOR since the thread lock would be some other sleepq lock. Fix this by deferring the wakeup of the swapper until after the sleepq lock held by the upper layer has been locked. The setrunnable() routine now returns a boolean value to indicate whether or not proc0 needs to be woken up. The end result is that consumers of the sleepq API such as *sleep/wakeup, condition variables, sx locks, and lockmgr, have to wakeup proc0 if they get a non-zero return value from sleepq_abort(), sleepq_broadcast(), or sleepq_signal(). Discussed with: jeff Glanced at by: sam Tested by: Jurgen Weber jurgen - ish com au MFC after: 2 weeks
2008-08-05 20:02:31 +00:00
return (0);
Now that the return value semantics of cv's for multithreaded processes have been unified with that of msleep(9), further refine the sleepq interface and consolidate some duplicated code: - Move the pre-sleep checks for theaded processes into a thread_sleep_check() function in kern_thread.c. - Move all handling of TDF_SINTR to be internal to subr_sleepqueue.c. Specifically, if a thread is awakened by something other than a signal while checking for signals before going to sleep, clear TDF_SINTR in sleepq_catch_signals(). This removes a sched_lock lock/unlock combo in that edge case during an interruptible sleep. Also, fix sleepq_check_signals() to properly handle the condition if TDF_SINTR is clear rather than requiring the callers of the sleepq API to notice this edge case and call a non-_sig variant of sleepq_wait(). - Clarify the flags arguments to sleepq_add(), sleepq_signal() and sleepq_broadcast() by creating an explicit submask for sleepq types. Also, add an explicit SLEEPQ_MSLEEP type rather than a magic number of 0. Also, add a SLEEPQ_INTERRUPTIBLE flag for use with sleepq_add() and move the setting of TDF_SINTR to sleepq_add() if this flag is set rather than sleepq_catch_signals(). Note that it is the caller's responsibility to ensure that sleepq_catch_signals() is called if and only if this flag is passed to the preceeding sleepq_add(). Note that this also removes a sched_lock lock/unlock pair from sleepq_catch_signals(). It also ensures that for an interruptible sleep, TDF_SINTR is always set when TD_ON_SLEEPQ() is true.
2004-08-19 11:31:42 +00:00
KASSERT(sq->sq_type == (flags & SLEEPQ_TYPE),
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
("%s: mismatch between sleep/wakeup and cv_*", __func__));
Split sleepq_wakeup_thread() into two functions. sleepq_remove_thread() removes a specific thread from a sleep queue. sleepq_resume_thread() resumes scheduling of a thread that has been previously removed from a sleep queue. - sleepq_catch_signals() just removes a thread from the queue it was just added to when a pending signal is found. - sleepq_signal() and sleepq_broadcast() remove threads from a queue, drop the queue lock, and then resume all the previously removed threads. This doesn't completely fix the sched_lock <-> sleepq chain LOR, but it makes it a little better as we no longer call setrunnble() with a sleep queue lock held meaning if setrunnable() tries to wakeup the swapper we don't try to lock two sleep queue chains at the same time.
2004-05-13 20:00:43 +00:00
Close a race between sleepq_broadcast() and sleepq_catch_signals(). Specifically, sleepq_broadcast() uses td_slpq for its private pending queue of threads that it is going to wake up after it takes them off the sleep queue. The problem is that if one of the threads is actually not asleep yet, then we can end up with td_slpq being corrupted and/or the thread being made runnable at the wrong time resulting in the td_sleepqueue == NULL assertion failures occasionally reported under heavy load. The fix is to stop being so fancy and ditch the whole pending queue bit. Instead, sleepq_remove_thread() and sleepq_resume_thread() were merged into one function that requires the caller to hold sched_lock. This fixes several places that unlocked sched_lock only to call a function that then locked sched_lock, so even though sched_lock is now held slightly longer, removing the extra lock acquires (1 pair instead of 3 in some cases) probably makes it an overall win if you don't include the fact that it closes a race. This is definitely a 5.4 candidate. PR: kern/79693 Submitted by: Steven Sears stevenjsears at yahoo dot com MFC after: 4 days
2005-04-14 06:30:32 +00:00
/* Resume all blocked threads on the sleep queue. */
If a thread that is swapped out is made runnable, then the setrunnable() routine wakes up proc0 so that proc0 can swap the thread back in. Historically, this has been done by waking up proc0 directly from setrunnable() itself via a wakeup(). When waking up a sleeping thread that was swapped out (the usual case when waking proc0 since only sleeping threads are eligible to be swapped out), this resulted in a bit of recursion (e.g. wakeup() -> setrunnable() -> wakeup()). With sleep queues having separate locks in 6.x and later, this caused a spin lock LOR (sleepq lock -> sched_lock/thread lock -> sleepq lock). An attempt was made to fix this in 7.0 by making the proc0 wakeup use the ithread mechanism for doing the wakeup. However, this required grabbing proc0's thread lock to perform the wakeup. If proc0 was asleep elsewhere in the kernel (e.g. waiting for disk I/O), then this degenerated into the same LOR since the thread lock would be some other sleepq lock. Fix this by deferring the wakeup of the swapper until after the sleepq lock held by the upper layer has been locked. The setrunnable() routine now returns a boolean value to indicate whether or not proc0 needs to be woken up. The end result is that consumers of the sleepq API such as *sleep/wakeup, condition variables, sx locks, and lockmgr, have to wakeup proc0 if they get a non-zero return value from sleepq_abort(), sleepq_broadcast(), or sleepq_signal(). Discussed with: jeff Glanced at by: sam Tested by: Jurgen Weber jurgen - ish com au MFC after: 2 weeks
2008-08-05 20:02:31 +00:00
wakeup_swapper = 0;
Close a race in sleepq_broadcast() where the sleepq could be reused after it had been assigned to the last sleeping thread. That thread might have started running on another CPU and have reused that sleep queue. Fix it by just walking the thread queue using TAILQ_FOREACH_SAFE() rather than a while loop. PR: amd64/124200 Discovered by: tegge Tested by: benjsc MFC after: 1 week
2008-09-08 19:44:57 +00:00
TAILQ_FOREACH_SAFE(td, &sq->sq_blocked[queue], td_slpq, tdn) {
Commit 2/14 of sched_lock decomposition. - Adapt sleepqueues to the new thread_lock() mechanism. - Delay assigning the sleep queue spinlock as the thread lock until after we've checked for signals. It is illegal for a thread to return in mi_switch() with any lock assigned to td_lock other than the scheduler locks. - Change sleepq_catch_signals() to do the switch if necessary to simplify the callers. - Simplify timeout handling now that locking a sleeping thread has the side-effect of locking the sleepqueue. Some previous races are no longer possible. Tested by: kris, current@ Tested on: i386, amd64, ULE, 4BSD, libthr, libkse, PREEMPTION, etc. Discussed with: kris, attilio, kmacy, jhb, julian, bde (small parts each)
2007-06-04 23:50:56 +00:00
thread_lock(td);
If a thread that is swapped out is made runnable, then the setrunnable() routine wakes up proc0 so that proc0 can swap the thread back in. Historically, this has been done by waking up proc0 directly from setrunnable() itself via a wakeup(). When waking up a sleeping thread that was swapped out (the usual case when waking proc0 since only sleeping threads are eligible to be swapped out), this resulted in a bit of recursion (e.g. wakeup() -> setrunnable() -> wakeup()). With sleep queues having separate locks in 6.x and later, this caused a spin lock LOR (sleepq lock -> sched_lock/thread lock -> sleepq lock). An attempt was made to fix this in 7.0 by making the proc0 wakeup use the ithread mechanism for doing the wakeup. However, this required grabbing proc0's thread lock to perform the wakeup. If proc0 was asleep elsewhere in the kernel (e.g. waiting for disk I/O), then this degenerated into the same LOR since the thread lock would be some other sleepq lock. Fix this by deferring the wakeup of the swapper until after the sleepq lock held by the upper layer has been locked. The setrunnable() routine now returns a boolean value to indicate whether or not proc0 needs to be woken up. The end result is that consumers of the sleepq API such as *sleep/wakeup, condition variables, sx locks, and lockmgr, have to wakeup proc0 if they get a non-zero return value from sleepq_abort(), sleepq_broadcast(), or sleepq_signal(). Discussed with: jeff Glanced at by: sam Tested by: Jurgen Weber jurgen - ish com au MFC after: 2 weeks
2008-08-05 20:02:31 +00:00
if (sleepq_resume_thread(sq, td, pri))
wakeup_swapper = 1;
Commit 2/14 of sched_lock decomposition. - Adapt sleepqueues to the new thread_lock() mechanism. - Delay assigning the sleep queue spinlock as the thread lock until after we've checked for signals. It is illegal for a thread to return in mi_switch() with any lock assigned to td_lock other than the scheduler locks. - Change sleepq_catch_signals() to do the switch if necessary to simplify the callers. - Simplify timeout handling now that locking a sleeping thread has the side-effect of locking the sleepqueue. Some previous races are no longer possible. Tested by: kris, current@ Tested on: i386, amd64, ULE, 4BSD, libthr, libkse, PREEMPTION, etc. Discussed with: kris, attilio, kmacy, jhb, julian, bde (small parts each)
2007-06-04 23:50:56 +00:00
thread_unlock(td);
}
If a thread that is swapped out is made runnable, then the setrunnable() routine wakes up proc0 so that proc0 can swap the thread back in. Historically, this has been done by waking up proc0 directly from setrunnable() itself via a wakeup(). When waking up a sleeping thread that was swapped out (the usual case when waking proc0 since only sleeping threads are eligible to be swapped out), this resulted in a bit of recursion (e.g. wakeup() -> setrunnable() -> wakeup()). With sleep queues having separate locks in 6.x and later, this caused a spin lock LOR (sleepq lock -> sched_lock/thread lock -> sleepq lock). An attempt was made to fix this in 7.0 by making the proc0 wakeup use the ithread mechanism for doing the wakeup. However, this required grabbing proc0's thread lock to perform the wakeup. If proc0 was asleep elsewhere in the kernel (e.g. waiting for disk I/O), then this degenerated into the same LOR since the thread lock would be some other sleepq lock. Fix this by deferring the wakeup of the swapper until after the sleepq lock held by the upper layer has been locked. The setrunnable() routine now returns a boolean value to indicate whether or not proc0 needs to be woken up. The end result is that consumers of the sleepq API such as *sleep/wakeup, condition variables, sx locks, and lockmgr, have to wakeup proc0 if they get a non-zero return value from sleepq_abort(), sleepq_broadcast(), or sleepq_signal(). Discussed with: jeff Glanced at by: sam Tested by: Jurgen Weber jurgen - ish com au MFC after: 2 weeks
2008-08-05 20:02:31 +00:00
return (wakeup_swapper);
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
}
/*
* Time sleeping threads out. When the timeout expires, the thread is
* removed from the sleep queue and made runnable if it is still asleep.
*/
static void
sleepq_timeout(void *arg)
{
Major callout subsystem cleanup and rewrite: - Close a migration race where callout_reset() failed to set the CALLOUT_ACTIVE flag. - Callout callback functions are now allowed to be protected by spinlocks. - Switching the callout CPU number cannot always be done on a per-callout basis. See the updated timeout(9) manual page for more information. - The timeout(9) manual page has been updated to reflect how all the functions inside the callout API are working. The manual page has been made function oriented to make it easier to deduce how each of the functions making up the callout API are working without having to first read the whole manual page. Group all functions into a handful of sections which should give a quick top-level overview when the different functions should be used. - The CALLOUT_SHAREDLOCK flag and its functionality has been removed to reduce the complexity in the callout code and to avoid problems about atomically stopping callouts via callout_stop(). If someone needs it, it can be re-added. From my quick grep there are no CALLOUT_SHAREDLOCK clients in the kernel. - A new callout API function named "callout_drain_async()" has been added. See the updated timeout(9) manual page for a complete description. - Update the callout clients in the "kern/" folder to use the callout API properly, like cv_timedwait(). Previously there was some custom sleepqueue code in the callout subsystem, which has been removed, because we now allow callouts to be protected by spinlocks. This allows us to tear down the callout like done with regular mutexes, and a "td_slpmutex" has been added to "struct thread" to atomically teardown the "td_slpcallout". Further the "TDF_TIMOFAIL" and "SWT_SLEEPQTIMO" states can now be completely removed. Currently they are marked as available and will be cleaned up in a follow up commit. - Bump the __FreeBSD_version to indicate kernel modules need recompilation. - There has been several reports that this patch "seems to squash a serious bug leading to a callout timeout and panic". Kernel build testing: all architectures were built MFC after: 2 weeks Differential Revision: https://reviews.freebsd.org/D1438 Sponsored by: Mellanox Technologies Reviewed by: jhb, adrian, sbruno and emaste
2015-01-15 15:32:30 +00:00
struct thread *td = arg;
int wakeup_swapper = 0;
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
Fixed some common printf format errors. Don't assume that "struct foo *" is "void *" (it isn't) or that the default promotion of pid_t is int. Instead, assume that casting "struct foo *" to "void *" and printing the result with %p is useful, and that all pid_t's are representable as longs. Fixed some minor style bugs (mainly spelling errors in comments).
2004-05-14 20:51:42 +00:00
CTR3(KTR_PROC, "sleepq_timeout: thread %p (pid %ld, %s)",
generally we are interested in what thread did something as opposed to what process. Since threads by default have teh name of the process unless over-written with more useful information, just print the thread name instead.
2007-11-14 06:21:24 +00:00
(void *)td, (long)td->td_proc->p_pid, (void *)td->td_name);
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
Major callout subsystem cleanup and rewrite: - Close a migration race where callout_reset() failed to set the CALLOUT_ACTIVE flag. - Callout callback functions are now allowed to be protected by spinlocks. - Switching the callout CPU number cannot always be done on a per-callout basis. See the updated timeout(9) manual page for more information. - The timeout(9) manual page has been updated to reflect how all the functions inside the callout API are working. The manual page has been made function oriented to make it easier to deduce how each of the functions making up the callout API are working without having to first read the whole manual page. Group all functions into a handful of sections which should give a quick top-level overview when the different functions should be used. - The CALLOUT_SHAREDLOCK flag and its functionality has been removed to reduce the complexity in the callout code and to avoid problems about atomically stopping callouts via callout_stop(). If someone needs it, it can be re-added. From my quick grep there are no CALLOUT_SHAREDLOCK clients in the kernel. - A new callout API function named "callout_drain_async()" has been added. See the updated timeout(9) manual page for a complete description. - Update the callout clients in the "kern/" folder to use the callout API properly, like cv_timedwait(). Previously there was some custom sleepqueue code in the callout subsystem, which has been removed, because we now allow callouts to be protected by spinlocks. This allows us to tear down the callout like done with regular mutexes, and a "td_slpmutex" has been added to "struct thread" to atomically teardown the "td_slpcallout". Further the "TDF_TIMOFAIL" and "SWT_SLEEPQTIMO" states can now be completely removed. Currently they are marked as available and will be cleaned up in a follow up commit. - Bump the __FreeBSD_version to indicate kernel modules need recompilation. - There has been several reports that this patch "seems to squash a serious bug leading to a callout timeout and panic". Kernel build testing: all architectures were built MFC after: 2 weeks Differential Revision: https://reviews.freebsd.org/D1438 Sponsored by: Mellanox Technologies Reviewed by: jhb, adrian, sbruno and emaste
2015-01-15 15:32:30 +00:00
/* Handle the three cases which can happen */
Fix a race in the sleepqueue timeout code that resulted in sleeps not being properly cancelled by a timeout. In general there is a race between a the sleepq timeout handler firing while the thread is still in the process of going to sleep. In 6.x with sched_lock, the race was largely protected by sched_lock. The only place it was "exposed" and had to be handled was while checking for any pending signals in sleepq_catch_signals(). With the thread lock changes, the thread lock is dropped in between sleepq_add() and sleepq_*wait*() opening up a new window for this race. Thus, if the timeout fired while the sleeping thread was in between sleepq_add() and sleepq_*wait*(), the thread would be marked as timed out, but the thread would not be dequeued and sleepq_switch() would still block the thread until it was awakened via some other means. In the case of pause(9) where there is no other wakeup, the thread would never be awakened. Fix this by teaching sleepq_switch() to check if the thread has had its sleep canceled before blocking by checking the TDF_TIMEOUT flag and aborting the sleep and dequeueing the thread if it is set. MFC after: 3 days Reported by: dwhite, peter
2008-01-25 02:09:38 +00:00
Major callout subsystem cleanup and rewrite: - Close a migration race where callout_reset() failed to set the CALLOUT_ACTIVE flag. - Callout callback functions are now allowed to be protected by spinlocks. - Switching the callout CPU number cannot always be done on a per-callout basis. See the updated timeout(9) manual page for more information. - The timeout(9) manual page has been updated to reflect how all the functions inside the callout API are working. The manual page has been made function oriented to make it easier to deduce how each of the functions making up the callout API are working without having to first read the whole manual page. Group all functions into a handful of sections which should give a quick top-level overview when the different functions should be used. - The CALLOUT_SHAREDLOCK flag and its functionality has been removed to reduce the complexity in the callout code and to avoid problems about atomically stopping callouts via callout_stop(). If someone needs it, it can be re-added. From my quick grep there are no CALLOUT_SHAREDLOCK clients in the kernel. - A new callout API function named "callout_drain_async()" has been added. See the updated timeout(9) manual page for a complete description. - Update the callout clients in the "kern/" folder to use the callout API properly, like cv_timedwait(). Previously there was some custom sleepqueue code in the callout subsystem, which has been removed, because we now allow callouts to be protected by spinlocks. This allows us to tear down the callout like done with regular mutexes, and a "td_slpmutex" has been added to "struct thread" to atomically teardown the "td_slpcallout". Further the "TDF_TIMOFAIL" and "SWT_SLEEPQTIMO" states can now be completely removed. Currently they are marked as available and will be cleaned up in a follow up commit. - Bump the __FreeBSD_version to indicate kernel modules need recompilation. - There has been several reports that this patch "seems to squash a serious bug leading to a callout timeout and panic". Kernel build testing: all architectures were built MFC after: 2 weeks Differential Revision: https://reviews.freebsd.org/D1438 Sponsored by: Mellanox Technologies Reviewed by: jhb, adrian, sbruno and emaste
2015-01-15 15:32:30 +00:00
thread_lock(td);
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
if (TD_ON_SLEEPQ(td)) {
Major callout subsystem cleanup and rewrite: - Close a migration race where callout_reset() failed to set the CALLOUT_ACTIVE flag. - Callout callback functions are now allowed to be protected by spinlocks. - Switching the callout CPU number cannot always be done on a per-callout basis. See the updated timeout(9) manual page for more information. - The timeout(9) manual page has been updated to reflect how all the functions inside the callout API are working. The manual page has been made function oriented to make it easier to deduce how each of the functions making up the callout API are working without having to first read the whole manual page. Group all functions into a handful of sections which should give a quick top-level overview when the different functions should be used. - The CALLOUT_SHAREDLOCK flag and its functionality has been removed to reduce the complexity in the callout code and to avoid problems about atomically stopping callouts via callout_stop(). If someone needs it, it can be re-added. From my quick grep there are no CALLOUT_SHAREDLOCK clients in the kernel. - A new callout API function named "callout_drain_async()" has been added. See the updated timeout(9) manual page for a complete description. - Update the callout clients in the "kern/" folder to use the callout API properly, like cv_timedwait(). Previously there was some custom sleepqueue code in the callout subsystem, which has been removed, because we now allow callouts to be protected by spinlocks. This allows us to tear down the callout like done with regular mutexes, and a "td_slpmutex" has been added to "struct thread" to atomically teardown the "td_slpcallout". Further the "TDF_TIMOFAIL" and "SWT_SLEEPQTIMO" states can now be completely removed. Currently they are marked as available and will be cleaned up in a follow up commit. - Bump the __FreeBSD_version to indicate kernel modules need recompilation. - There has been several reports that this patch "seems to squash a serious bug leading to a callout timeout and panic". Kernel build testing: all architectures were built MFC after: 2 weeks Differential Revision: https://reviews.freebsd.org/D1438 Sponsored by: Mellanox Technologies Reviewed by: jhb, adrian, sbruno and emaste
2015-01-15 15:32:30 +00:00
if (TD_IS_SLEEPING(td)) {
struct sleepqueue_chain *sc;
struct sleepqueue *sq;
void *wchan;
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
Major callout subsystem cleanup and rewrite: - Close a migration race where callout_reset() failed to set the CALLOUT_ACTIVE flag. - Callout callback functions are now allowed to be protected by spinlocks. - Switching the callout CPU number cannot always be done on a per-callout basis. See the updated timeout(9) manual page for more information. - The timeout(9) manual page has been updated to reflect how all the functions inside the callout API are working. The manual page has been made function oriented to make it easier to deduce how each of the functions making up the callout API are working without having to first read the whole manual page. Group all functions into a handful of sections which should give a quick top-level overview when the different functions should be used. - The CALLOUT_SHAREDLOCK flag and its functionality has been removed to reduce the complexity in the callout code and to avoid problems about atomically stopping callouts via callout_stop(). If someone needs it, it can be re-added. From my quick grep there are no CALLOUT_SHAREDLOCK clients in the kernel. - A new callout API function named "callout_drain_async()" has been added. See the updated timeout(9) manual page for a complete description. - Update the callout clients in the "kern/" folder to use the callout API properly, like cv_timedwait(). Previously there was some custom sleepqueue code in the callout subsystem, which has been removed, because we now allow callouts to be protected by spinlocks. This allows us to tear down the callout like done with regular mutexes, and a "td_slpmutex" has been added to "struct thread" to atomically teardown the "td_slpcallout". Further the "TDF_TIMOFAIL" and "SWT_SLEEPQTIMO" states can now be completely removed. Currently they are marked as available and will be cleaned up in a follow up commit. - Bump the __FreeBSD_version to indicate kernel modules need recompilation. - There has been several reports that this patch "seems to squash a serious bug leading to a callout timeout and panic". Kernel build testing: all architectures were built MFC after: 2 weeks Differential Revision: https://reviews.freebsd.org/D1438 Sponsored by: Mellanox Technologies Reviewed by: jhb, adrian, sbruno and emaste
2015-01-15 15:32:30 +00:00
/*
* Case I - thread is asleep and needs to be
* awoken:
*/
wchan = td->td_wchan;
sc = SC_LOOKUP(wchan);
THREAD_LOCKPTR_ASSERT(td, &sc->sc_lock);
sq = sleepq_lookup(wchan);
MPASS(sq != NULL);
td->td_flags |= TDF_TIMEOUT;
wakeup_swapper = sleepq_resume_thread(sq, td, 0);
} else {
/*
* Case II - cancel going to sleep by setting
* the timeout flag because the target thread
* is not asleep yet. It can be on another CPU
* in between sleepq_add() and one of the
* sleepq_*wait*() routines or it can be in
* sleepq_catch_signals().
*/
td->td_flags |= TDF_TIMEOUT;
}
} else {
/*
* Case III - thread is already woken up by a wakeup
* call and should not timeout. Nothing to do!
*/
}
Commit 2/14 of sched_lock decomposition. - Adapt sleepqueues to the new thread_lock() mechanism. - Delay assigning the sleep queue spinlock as the thread lock until after we've checked for signals. It is illegal for a thread to return in mi_switch() with any lock assigned to td_lock other than the scheduler locks. - Change sleepq_catch_signals() to do the switch if necessary to simplify the callers. - Simplify timeout handling now that locking a sleeping thread has the side-effect of locking the sleepqueue. Some previous races are no longer possible. Tested by: kris, current@ Tested on: i386, amd64, ULE, 4BSD, libthr, libkse, PREEMPTION, etc. Discussed with: kris, attilio, kmacy, jhb, julian, bde (small parts each)
2007-06-04 23:50:56 +00:00
thread_unlock(td);
If a thread that is swapped out is made runnable, then the setrunnable() routine wakes up proc0 so that proc0 can swap the thread back in. Historically, this has been done by waking up proc0 directly from setrunnable() itself via a wakeup(). When waking up a sleeping thread that was swapped out (the usual case when waking proc0 since only sleeping threads are eligible to be swapped out), this resulted in a bit of recursion (e.g. wakeup() -> setrunnable() -> wakeup()). With sleep queues having separate locks in 6.x and later, this caused a spin lock LOR (sleepq lock -> sched_lock/thread lock -> sleepq lock). An attempt was made to fix this in 7.0 by making the proc0 wakeup use the ithread mechanism for doing the wakeup. However, this required grabbing proc0's thread lock to perform the wakeup. If proc0 was asleep elsewhere in the kernel (e.g. waiting for disk I/O), then this degenerated into the same LOR since the thread lock would be some other sleepq lock. Fix this by deferring the wakeup of the swapper until after the sleepq lock held by the upper layer has been locked. The setrunnable() routine now returns a boolean value to indicate whether or not proc0 needs to be woken up. The end result is that consumers of the sleepq API such as *sleep/wakeup, condition variables, sx locks, and lockmgr, have to wakeup proc0 if they get a non-zero return value from sleepq_abort(), sleepq_broadcast(), or sleepq_signal(). Discussed with: jeff Glanced at by: sam Tested by: Jurgen Weber jurgen - ish com au MFC after: 2 weeks
2008-08-05 20:02:31 +00:00
if (wakeup_swapper)
kick_proc0();
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
}
/*
* Resumes a specific thread from the sleep queue associated with a specific
* wait channel if it is on that queue.
*/
void
sleepq_remove(struct thread *td, void *wchan)
{
struct sleepqueue *sq;
If a thread that is swapped out is made runnable, then the setrunnable() routine wakes up proc0 so that proc0 can swap the thread back in. Historically, this has been done by waking up proc0 directly from setrunnable() itself via a wakeup(). When waking up a sleeping thread that was swapped out (the usual case when waking proc0 since only sleeping threads are eligible to be swapped out), this resulted in a bit of recursion (e.g. wakeup() -> setrunnable() -> wakeup()). With sleep queues having separate locks in 6.x and later, this caused a spin lock LOR (sleepq lock -> sched_lock/thread lock -> sleepq lock). An attempt was made to fix this in 7.0 by making the proc0 wakeup use the ithread mechanism for doing the wakeup. However, this required grabbing proc0's thread lock to perform the wakeup. If proc0 was asleep elsewhere in the kernel (e.g. waiting for disk I/O), then this degenerated into the same LOR since the thread lock would be some other sleepq lock. Fix this by deferring the wakeup of the swapper until after the sleepq lock held by the upper layer has been locked. The setrunnable() routine now returns a boolean value to indicate whether or not proc0 needs to be woken up. The end result is that consumers of the sleepq API such as *sleep/wakeup, condition variables, sx locks, and lockmgr, have to wakeup proc0 if they get a non-zero return value from sleepq_abort(), sleepq_broadcast(), or sleepq_signal(). Discussed with: jeff Glanced at by: sam Tested by: Jurgen Weber jurgen - ish com au MFC after: 2 weeks
2008-08-05 20:02:31 +00:00
int wakeup_swapper;
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
/*
* Look up the sleep queue for this wait channel, then re-check
* that the thread is asleep on that channel, if it is not, then
* bail.
*/
MPASS(wchan != NULL);
Refine the turnstile and sleep queue interfaces just a bit: - Add a new _lock() call to each API that locks the associated chain lock for a lock_object pointer or wait channel. The _lookup() functions now require that the chain lock be locked via _lock() when they are called. - Change sleepq_add(), turnstile_wait() and turnstile_claim() to lookup the associated queue structure internally via _lookup() rather than accepting a pointer from the caller. For turnstiles, this means that the actual lookup of the turnstile in the hash table is only done when the thread actually blocks rather than being done on each loop iteration in _mtx_lock_sleep(). For sleep queues, this means that sleepq_lookup() is no longer used outside of the sleep queue code except to implement an assertion in cv_destroy(). - Change sleepq_broadcast() and sleepq_signal() to require that the chain lock is already required. For condition variables, this lets the cv_broadcast() and cv_signal() functions lock the sleep queue chain lock while testing the waiters count. This means that the waiters count internal to condition variables is no longer protected by the interlock mutex and cv_broadcast() and cv_signal() now no longer require that the interlock be held when they are called. This lets consumers of condition variables drop the lock before waking other threads which can result in fewer context switches. MFC after: 1 month
2004-10-12 18:36:20 +00:00
sleepq_lock(wchan);
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
sq = sleepq_lookup(wchan);
Commit 2/14 of sched_lock decomposition. - Adapt sleepqueues to the new thread_lock() mechanism. - Delay assigning the sleep queue spinlock as the thread lock until after we've checked for signals. It is illegal for a thread to return in mi_switch() with any lock assigned to td_lock other than the scheduler locks. - Change sleepq_catch_signals() to do the switch if necessary to simplify the callers. - Simplify timeout handling now that locking a sleeping thread has the side-effect of locking the sleepqueue. Some previous races are no longer possible. Tested by: kris, current@ Tested on: i386, amd64, ULE, 4BSD, libthr, libkse, PREEMPTION, etc. Discussed with: kris, attilio, kmacy, jhb, julian, bde (small parts each)
2007-06-04 23:50:56 +00:00
/*
* We can not lock the thread here as it may be sleeping on a
* different sleepq. However, holding the sleepq lock for this
* wchan can guarantee that we do not miss a wakeup for this
* channel. The asserts below will catch any false positives.
*/
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
if (!TD_ON_SLEEPQ(td) || td->td_wchan != wchan) {
sleepq_release(wchan);
return;
}
/* Thread is asleep on sleep queue sq, so wake it up. */
Commit 2/14 of sched_lock decomposition. - Adapt sleepqueues to the new thread_lock() mechanism. - Delay assigning the sleep queue spinlock as the thread lock until after we've checked for signals. It is illegal for a thread to return in mi_switch() with any lock assigned to td_lock other than the scheduler locks. - Change sleepq_catch_signals() to do the switch if necessary to simplify the callers. - Simplify timeout handling now that locking a sleeping thread has the side-effect of locking the sleepqueue. Some previous races are no longer possible. Tested by: kris, current@ Tested on: i386, amd64, ULE, 4BSD, libthr, libkse, PREEMPTION, etc. Discussed with: kris, attilio, kmacy, jhb, julian, bde (small parts each)
2007-06-04 23:50:56 +00:00
thread_lock(td);
MPASS(sq != NULL);
MPASS(td->td_wchan == wchan);
If a thread that is swapped out is made runnable, then the setrunnable() routine wakes up proc0 so that proc0 can swap the thread back in. Historically, this has been done by waking up proc0 directly from setrunnable() itself via a wakeup(). When waking up a sleeping thread that was swapped out (the usual case when waking proc0 since only sleeping threads are eligible to be swapped out), this resulted in a bit of recursion (e.g. wakeup() -> setrunnable() -> wakeup()). With sleep queues having separate locks in 6.x and later, this caused a spin lock LOR (sleepq lock -> sched_lock/thread lock -> sleepq lock). An attempt was made to fix this in 7.0 by making the proc0 wakeup use the ithread mechanism for doing the wakeup. However, this required grabbing proc0's thread lock to perform the wakeup. If proc0 was asleep elsewhere in the kernel (e.g. waiting for disk I/O), then this degenerated into the same LOR since the thread lock would be some other sleepq lock. Fix this by deferring the wakeup of the swapper until after the sleepq lock held by the upper layer has been locked. The setrunnable() routine now returns a boolean value to indicate whether or not proc0 needs to be woken up. The end result is that consumers of the sleepq API such as *sleep/wakeup, condition variables, sx locks, and lockmgr, have to wakeup proc0 if they get a non-zero return value from sleepq_abort(), sleepq_broadcast(), or sleepq_signal(). Discussed with: jeff Glanced at by: sam Tested by: Jurgen Weber jurgen - ish com au MFC after: 2 weeks
2008-08-05 20:02:31 +00:00
wakeup_swapper = sleepq_resume_thread(sq, td, 0);
Commit 2/14 of sched_lock decomposition. - Adapt sleepqueues to the new thread_lock() mechanism. - Delay assigning the sleep queue spinlock as the thread lock until after we've checked for signals. It is illegal for a thread to return in mi_switch() with any lock assigned to td_lock other than the scheduler locks. - Change sleepq_catch_signals() to do the switch if necessary to simplify the callers. - Simplify timeout handling now that locking a sleeping thread has the side-effect of locking the sleepqueue. Some previous races are no longer possible. Tested by: kris, current@ Tested on: i386, amd64, ULE, 4BSD, libthr, libkse, PREEMPTION, etc. Discussed with: kris, attilio, kmacy, jhb, julian, bde (small parts each)
2007-06-04 23:50:56 +00:00
thread_unlock(td);
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
sleepq_release(wchan);
If a thread that is swapped out is made runnable, then the setrunnable() routine wakes up proc0 so that proc0 can swap the thread back in. Historically, this has been done by waking up proc0 directly from setrunnable() itself via a wakeup(). When waking up a sleeping thread that was swapped out (the usual case when waking proc0 since only sleeping threads are eligible to be swapped out), this resulted in a bit of recursion (e.g. wakeup() -> setrunnable() -> wakeup()). With sleep queues having separate locks in 6.x and later, this caused a spin lock LOR (sleepq lock -> sched_lock/thread lock -> sleepq lock). An attempt was made to fix this in 7.0 by making the proc0 wakeup use the ithread mechanism for doing the wakeup. However, this required grabbing proc0's thread lock to perform the wakeup. If proc0 was asleep elsewhere in the kernel (e.g. waiting for disk I/O), then this degenerated into the same LOR since the thread lock would be some other sleepq lock. Fix this by deferring the wakeup of the swapper until after the sleepq lock held by the upper layer has been locked. The setrunnable() routine now returns a boolean value to indicate whether or not proc0 needs to be woken up. The end result is that consumers of the sleepq API such as *sleep/wakeup, condition variables, sx locks, and lockmgr, have to wakeup proc0 if they get a non-zero return value from sleepq_abort(), sleepq_broadcast(), or sleepq_signal(). Discussed with: jeff Glanced at by: sam Tested by: Jurgen Weber jurgen - ish com au MFC after: 2 weeks
2008-08-05 20:02:31 +00:00
if (wakeup_swapper)
kick_proc0();
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
}
/*
Fixed some common printf format errors. Don't assume that "struct foo *" is "void *" (it isn't) or that the default promotion of pid_t is int. Instead, assume that casting "struct foo *" to "void *" and printing the result with %p is useful, and that all pid_t's are representable as longs. Fixed some minor style bugs (mainly spelling errors in comments).
2004-05-14 20:51:42 +00:00
* Abort a thread as if an interrupt had occurred. Only abort
* interruptible waits (unfortunately it isn't safe to abort others).
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
*/
If a thread that is swapped out is made runnable, then the setrunnable() routine wakes up proc0 so that proc0 can swap the thread back in. Historically, this has been done by waking up proc0 directly from setrunnable() itself via a wakeup(). When waking up a sleeping thread that was swapped out (the usual case when waking proc0 since only sleeping threads are eligible to be swapped out), this resulted in a bit of recursion (e.g. wakeup() -> setrunnable() -> wakeup()). With sleep queues having separate locks in 6.x and later, this caused a spin lock LOR (sleepq lock -> sched_lock/thread lock -> sleepq lock). An attempt was made to fix this in 7.0 by making the proc0 wakeup use the ithread mechanism for doing the wakeup. However, this required grabbing proc0's thread lock to perform the wakeup. If proc0 was asleep elsewhere in the kernel (e.g. waiting for disk I/O), then this degenerated into the same LOR since the thread lock would be some other sleepq lock. Fix this by deferring the wakeup of the swapper until after the sleepq lock held by the upper layer has been locked. The setrunnable() routine now returns a boolean value to indicate whether or not proc0 needs to be woken up. The end result is that consumers of the sleepq API such as *sleep/wakeup, condition variables, sx locks, and lockmgr, have to wakeup proc0 if they get a non-zero return value from sleepq_abort(), sleepq_broadcast(), or sleepq_signal(). Discussed with: jeff Glanced at by: sam Tested by: Jurgen Weber jurgen - ish com au MFC after: 2 weeks
2008-08-05 20:02:31 +00:00
int
Fix a long standing race between sleep queue and thread suspension code. When a thread A is going to sleep, it calls sleepq_catch_signals() to detect any pending signals or thread suspension request, if nothing happens, it returns without holding process lock or scheduler lock, this opens a race window which allows thread B to come in and do process suspension work, however since A is still at running state, thread B can do nothing to A, thread A continues, and puts itself into actually sleeping state, but B has never seen it, and it sits there forever until B is woken up by other threads sometimes later(this can be very long delay or never happen). Fix this bug by forcing sleepq_catch_signals to return with scheduler lock held. Fix sleepq_abort() by passing it an interrupted code, previously, it worked as wakeup_one(), and the interruption can not be identified correctly by sleep queue code when the sleeping thread is resumed. Let thread_suspend_check() returns EINTR or ERESTART, so sleep queue no longer has to use SIGSTOP as a hack to build a return value. Reviewed by: jhb MFC after: 1 week
2006-02-15 23:52:01 +00:00
sleepq_abort(struct thread *td, int intrval)
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
{
Commit 2/14 of sched_lock decomposition. - Adapt sleepqueues to the new thread_lock() mechanism. - Delay assigning the sleep queue spinlock as the thread lock until after we've checked for signals. It is illegal for a thread to return in mi_switch() with any lock assigned to td_lock other than the scheduler locks. - Change sleepq_catch_signals() to do the switch if necessary to simplify the callers. - Simplify timeout handling now that locking a sleeping thread has the side-effect of locking the sleepqueue. Some previous races are no longer possible. Tested by: kris, current@ Tested on: i386, amd64, ULE, 4BSD, libthr, libkse, PREEMPTION, etc. Discussed with: kris, attilio, kmacy, jhb, julian, bde (small parts each)
2007-06-04 23:50:56 +00:00
struct sleepqueue *sq;
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
void *wchan;
Commit 2/14 of sched_lock decomposition. - Adapt sleepqueues to the new thread_lock() mechanism. - Delay assigning the sleep queue spinlock as the thread lock until after we've checked for signals. It is illegal for a thread to return in mi_switch() with any lock assigned to td_lock other than the scheduler locks. - Change sleepq_catch_signals() to do the switch if necessary to simplify the callers. - Simplify timeout handling now that locking a sleeping thread has the side-effect of locking the sleepqueue. Some previous races are no longer possible. Tested by: kris, current@ Tested on: i386, amd64, ULE, 4BSD, libthr, libkse, PREEMPTION, etc. Discussed with: kris, attilio, kmacy, jhb, julian, bde (small parts each)
2007-06-04 23:50:56 +00:00
THREAD_LOCK_ASSERT(td, MA_OWNED);
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
MPASS(TD_ON_SLEEPQ(td));
MPASS(td->td_flags & TDF_SINTR);
Fix a long standing race between sleep queue and thread suspension code. When a thread A is going to sleep, it calls sleepq_catch_signals() to detect any pending signals or thread suspension request, if nothing happens, it returns without holding process lock or scheduler lock, this opens a race window which allows thread B to come in and do process suspension work, however since A is still at running state, thread B can do nothing to A, thread A continues, and puts itself into actually sleeping state, but B has never seen it, and it sits there forever until B is woken up by other threads sometimes later(this can be very long delay or never happen). Fix this bug by forcing sleepq_catch_signals to return with scheduler lock held. Fix sleepq_abort() by passing it an interrupted code, previously, it worked as wakeup_one(), and the interruption can not be identified correctly by sleep queue code when the sleeping thread is resumed. Let thread_suspend_check() returns EINTR or ERESTART, so sleep queue no longer has to use SIGSTOP as a hack to build a return value. Reviewed by: jhb MFC after: 1 week
2006-02-15 23:52:01 +00:00
MPASS(intrval == EINTR || intrval == ERESTART);
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
/*
* If the TDF_TIMEOUT flag is set, just leave. A
* timeout is scheduled anyhow.
*/
if (td->td_flags & TDF_TIMEOUT)
If a thread that is swapped out is made runnable, then the setrunnable() routine wakes up proc0 so that proc0 can swap the thread back in. Historically, this has been done by waking up proc0 directly from setrunnable() itself via a wakeup(). When waking up a sleeping thread that was swapped out (the usual case when waking proc0 since only sleeping threads are eligible to be swapped out), this resulted in a bit of recursion (e.g. wakeup() -> setrunnable() -> wakeup()). With sleep queues having separate locks in 6.x and later, this caused a spin lock LOR (sleepq lock -> sched_lock/thread lock -> sleepq lock). An attempt was made to fix this in 7.0 by making the proc0 wakeup use the ithread mechanism for doing the wakeup. However, this required grabbing proc0's thread lock to perform the wakeup. If proc0 was asleep elsewhere in the kernel (e.g. waiting for disk I/O), then this degenerated into the same LOR since the thread lock would be some other sleepq lock. Fix this by deferring the wakeup of the swapper until after the sleepq lock held by the upper layer has been locked. The setrunnable() routine now returns a boolean value to indicate whether or not proc0 needs to be woken up. The end result is that consumers of the sleepq API such as *sleep/wakeup, condition variables, sx locks, and lockmgr, have to wakeup proc0 if they get a non-zero return value from sleepq_abort(), sleepq_broadcast(), or sleepq_signal(). Discussed with: jeff Glanced at by: sam Tested by: Jurgen Weber jurgen - ish com au MFC after: 2 weeks
2008-08-05 20:02:31 +00:00
return (0);
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
Fixed some common printf format errors. Don't assume that "struct foo *" is "void *" (it isn't) or that the default promotion of pid_t is int. Instead, assume that casting "struct foo *" to "void *" and printing the result with %p is useful, and that all pid_t's are representable as longs. Fixed some minor style bugs (mainly spelling errors in comments).
2004-05-14 20:51:42 +00:00
CTR3(KTR_PROC, "sleepq_abort: thread %p (pid %ld, %s)",
generally we are interested in what thread did something as opposed to what process. Since threads by default have teh name of the process unless over-written with more useful information, just print the thread name instead.
2007-11-14 06:21:24 +00:00
(void *)td, (long)td->td_proc->p_pid, (void *)td->td_name);
Commit 2/14 of sched_lock decomposition. - Adapt sleepqueues to the new thread_lock() mechanism. - Delay assigning the sleep queue spinlock as the thread lock until after we've checked for signals. It is illegal for a thread to return in mi_switch() with any lock assigned to td_lock other than the scheduler locks. - Change sleepq_catch_signals() to do the switch if necessary to simplify the callers. - Simplify timeout handling now that locking a sleeping thread has the side-effect of locking the sleepqueue. Some previous races are no longer possible. Tested by: kris, current@ Tested on: i386, amd64, ULE, 4BSD, libthr, libkse, PREEMPTION, etc. Discussed with: kris, attilio, kmacy, jhb, julian, bde (small parts each)
2007-06-04 23:50:56 +00:00
td->td_intrval = intrval;
td->td_flags |= TDF_SLEEPABORT;
/*
* If the thread has not slept yet it will find the signal in
* sleepq_catch_signals() and call sleepq_resume_thread. Otherwise
* we have to do it here.
*/
if (!TD_IS_SLEEPING(td))
If a thread that is swapped out is made runnable, then the setrunnable() routine wakes up proc0 so that proc0 can swap the thread back in. Historically, this has been done by waking up proc0 directly from setrunnable() itself via a wakeup(). When waking up a sleeping thread that was swapped out (the usual case when waking proc0 since only sleeping threads are eligible to be swapped out), this resulted in a bit of recursion (e.g. wakeup() -> setrunnable() -> wakeup()). With sleep queues having separate locks in 6.x and later, this caused a spin lock LOR (sleepq lock -> sched_lock/thread lock -> sleepq lock). An attempt was made to fix this in 7.0 by making the proc0 wakeup use the ithread mechanism for doing the wakeup. However, this required grabbing proc0's thread lock to perform the wakeup. If proc0 was asleep elsewhere in the kernel (e.g. waiting for disk I/O), then this degenerated into the same LOR since the thread lock would be some other sleepq lock. Fix this by deferring the wakeup of the swapper until after the sleepq lock held by the upper layer has been locked. The setrunnable() routine now returns a boolean value to indicate whether or not proc0 needs to be woken up. The end result is that consumers of the sleepq API such as *sleep/wakeup, condition variables, sx locks, and lockmgr, have to wakeup proc0 if they get a non-zero return value from sleepq_abort(), sleepq_broadcast(), or sleepq_signal(). Discussed with: jeff Glanced at by: sam Tested by: Jurgen Weber jurgen - ish com au MFC after: 2 weeks
2008-08-05 20:02:31 +00:00
return (0);
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
wchan = td->td_wchan;
Commit 2/14 of sched_lock decomposition. - Adapt sleepqueues to the new thread_lock() mechanism. - Delay assigning the sleep queue spinlock as the thread lock until after we've checked for signals. It is illegal for a thread to return in mi_switch() with any lock assigned to td_lock other than the scheduler locks. - Change sleepq_catch_signals() to do the switch if necessary to simplify the callers. - Simplify timeout handling now that locking a sleeping thread has the side-effect of locking the sleepqueue. Some previous races are no longer possible. Tested by: kris, current@ Tested on: i386, amd64, ULE, 4BSD, libthr, libkse, PREEMPTION, etc. Discussed with: kris, attilio, kmacy, jhb, julian, bde (small parts each)
2007-06-04 23:50:56 +00:00
MPASS(wchan != NULL);
sq = sleepq_lookup(wchan);
MPASS(sq != NULL);
/* Thread is asleep on sleep queue sq, so wake it up. */
If a thread that is swapped out is made runnable, then the setrunnable() routine wakes up proc0 so that proc0 can swap the thread back in. Historically, this has been done by waking up proc0 directly from setrunnable() itself via a wakeup(). When waking up a sleeping thread that was swapped out (the usual case when waking proc0 since only sleeping threads are eligible to be swapped out), this resulted in a bit of recursion (e.g. wakeup() -> setrunnable() -> wakeup()). With sleep queues having separate locks in 6.x and later, this caused a spin lock LOR (sleepq lock -> sched_lock/thread lock -> sleepq lock). An attempt was made to fix this in 7.0 by making the proc0 wakeup use the ithread mechanism for doing the wakeup. However, this required grabbing proc0's thread lock to perform the wakeup. If proc0 was asleep elsewhere in the kernel (e.g. waiting for disk I/O), then this degenerated into the same LOR since the thread lock would be some other sleepq lock. Fix this by deferring the wakeup of the swapper until after the sleepq lock held by the upper layer has been locked. The setrunnable() routine now returns a boolean value to indicate whether or not proc0 needs to be woken up. The end result is that consumers of the sleepq API such as *sleep/wakeup, condition variables, sx locks, and lockmgr, have to wakeup proc0 if they get a non-zero return value from sleepq_abort(), sleepq_broadcast(), or sleepq_signal(). Discussed with: jeff Glanced at by: sam Tested by: Jurgen Weber jurgen - ish com au MFC after: 2 weeks
2008-08-05 20:02:31 +00:00
return (sleepq_resume_thread(sq, td, 0));
Add an implementation of a generic sleep queue abstraction that is used to queue threads sleeping on a wait channel similar to how turnstiles are used to queue threads waiting for a lock. This subsystem will be used as the backend for sleep/wakeup and condition variables initially. Eventually it will also be used to replace the ithread-specific iwait thread inhibitor. Sleep queues are also not locked by sched_lock, so this splits sched_lock up a bit further increasing concurrency within the scheduler. Sleep queues also natively support timeouts on sleeps and interruptible sleeps allowing for the reduction of a lot of duplicated code between the sleep/wakeup and condition variable implementations. For more details on the sleep queue implementation, check the comments in sys/sleepqueue.h and kern/subr_sleepqueue.c.
2004-02-27 18:33:09 +00:00
}
Add a new ddb command 'show sleepq'. It takes a wait channel as an argument and looks for a sleep queue associated with that wait channel. If it finds one it will display information such as the list of threads sleeping on that queue. If it can't find a sleep queue for that wait channel, then it will see if that address matches any of the active sleep queues. If so, it will display information about the sleepq at the specified address.
2006-01-27 22:24:07 +00:00
- Add a facility similar to LOCK_PROFILING under SLEEPQUEUE_PROFILING. Keep a simple (wmesg, count) tuple in a hash to keep track of how many times we sleep at each wait message. We hash on message and not channel. No line number information is given as typically wait messages are not used in more than one place. Identical strings defined at different addresses will show up with seperate counters. - Use debug.sleepq.enable to enable, .reset to reset, and .stats dumps stats. - Do an unsynchronized check in sleepq_switch() prior to switching before calling sleepq_profile() which uses a global lock to synchronize the hash. Only sleeps which actually cause a context switch are counted.
2008-03-19 07:22:07 +00:00
#ifdef SLEEPQUEUE_PROFILING
#define SLEEPQ_PROF_LOCATIONS 1024
Re-add r212370 now that the LOR in powerpc64 has been resolved: Add a drain function for struct sysctl_req, and use it for a variety of handlers, some of which had to do awkward things to get a large enough SBUF_FIXEDLEN buffer. Note that some sysctl handlers were explicitly outputting a trailing NUL byte. This behaviour was preserved, though it should not be necessary. Reviewed by: phk (original patch)
2010-09-16 16:13:12 +00:00
#define SLEEPQ_SBUFSIZE 512
- Add a facility similar to LOCK_PROFILING under SLEEPQUEUE_PROFILING. Keep a simple (wmesg, count) tuple in a hash to keep track of how many times we sleep at each wait message. We hash on message and not channel. No line number information is given as typically wait messages are not used in more than one place. Identical strings defined at different addresses will show up with seperate counters. - Use debug.sleepq.enable to enable, .reset to reset, and .stats dumps stats. - Do an unsynchronized check in sleepq_switch() prior to switching before calling sleepq_profile() which uses a global lock to synchronize the hash. Only sleeps which actually cause a context switch are counted.
2008-03-19 07:22:07 +00:00
struct sleepq_prof {
LIST_ENTRY(sleepq_prof) sp_link;
const char *sp_wmesg;
long sp_count;
};
LIST_HEAD(sqphead, sleepq_prof);
struct sqphead sleepq_prof_free;
struct sqphead sleepq_hash[SC_TABLESIZE];
static struct sleepq_prof sleepq_profent[SLEEPQ_PROF_LOCATIONS];
static struct mtx sleepq_prof_lock;
MTX_SYSINIT(sleepq_prof_lock, &sleepq_prof_lock, "sleepq_prof", MTX_SPIN);
static void
sleepq_profile(const char *wmesg)
{
struct sleepq_prof *sp;
mtx_lock_spin(&sleepq_prof_lock);
if (prof_enabled == 0)
goto unlock;
LIST_FOREACH(sp, &sleepq_hash[SC_HASH(wmesg)], sp_link)
if (sp->sp_wmesg == wmesg)
goto done;
sp = LIST_FIRST(&sleepq_prof_free);
if (sp == NULL)
goto unlock;
sp->sp_wmesg = wmesg;
LIST_REMOVE(sp, sp_link);
LIST_INSERT_HEAD(&sleepq_hash[SC_HASH(wmesg)], sp, sp_link);
done:
sp->sp_count++;
unlock:
mtx_unlock_spin(&sleepq_prof_lock);
return;
}
static void
sleepq_prof_reset(void)
{
struct sleepq_prof *sp;
int enabled;
int i;
mtx_lock_spin(&sleepq_prof_lock);
enabled = prof_enabled;
prof_enabled = 0;
for (i = 0; i < SC_TABLESIZE; i++)
LIST_INIT(&sleepq_hash[i]);
LIST_INIT(&sleepq_prof_free);
for (i = 0; i < SLEEPQ_PROF_LOCATIONS; i++) {
sp = &sleepq_profent[i];
sp->sp_wmesg = NULL;
sp->sp_count = 0;
LIST_INSERT_HEAD(&sleepq_prof_free, sp, sp_link);
}
prof_enabled = enabled;
mtx_unlock_spin(&sleepq_prof_lock);
}
static int
enable_sleepq_prof(SYSCTL_HANDLER_ARGS)
{
int error, v;
v = prof_enabled;
error = sysctl_handle_int(oidp, &v, v, req);
if (error)
return (error);
if (req->newptr == NULL)
return (error);
if (v == prof_enabled)
return (0);
if (v == 1)
sleepq_prof_reset();
mtx_lock_spin(&sleepq_prof_lock);
prof_enabled = !!v;
mtx_unlock_spin(&sleepq_prof_lock);
return (0);
}
static int
reset_sleepq_prof_stats(SYSCTL_HANDLER_ARGS)
{
int error, v;
v = 0;
error = sysctl_handle_int(oidp, &v, 0, req);
if (error)
return (error);
if (req->newptr == NULL)
return (error);
if (v == 0)
return (0);
sleepq_prof_reset();
return (0);
}
static int
dump_sleepq_prof_stats(SYSCTL_HANDLER_ARGS)
{
struct sleepq_prof *sp;
struct sbuf *sb;
int enabled;
int error;
int i;
Explicitly wire the user buffer rather than doing it implicitly in sbuf_new_for_sysctl(9). This allows using an sbuf with a SYSCTL_OUT drain for extremely large amounts of data where the caller knows that appropriate references are held, and sleeping is not an issue. Inspired by: rwatson
2011-01-27 00:34:12 +00:00
error = sysctl_wire_old_buffer(req, 0);
if (error != 0)
return (error);
Re-add r212370 now that the LOR in powerpc64 has been resolved: Add a drain function for struct sysctl_req, and use it for a variety of handlers, some of which had to do awkward things to get a large enough SBUF_FIXEDLEN buffer. Note that some sysctl handlers were explicitly outputting a trailing NUL byte. This behaviour was preserved, though it should not be necessary. Reviewed by: phk (original patch)
2010-09-16 16:13:12 +00:00
sb = sbuf_new_for_sysctl(NULL, NULL, SLEEPQ_SBUFSIZE, req);
- Add a facility similar to LOCK_PROFILING under SLEEPQUEUE_PROFILING. Keep a simple (wmesg, count) tuple in a hash to keep track of how many times we sleep at each wait message. We hash on message and not channel. No line number information is given as typically wait messages are not used in more than one place. Identical strings defined at different addresses will show up with seperate counters. - Use debug.sleepq.enable to enable, .reset to reset, and .stats dumps stats. - Do an unsynchronized check in sleepq_switch() prior to switching before calling sleepq_profile() which uses a global lock to synchronize the hash. Only sleeps which actually cause a context switch are counted.
2008-03-19 07:22:07 +00:00
sbuf_printf(sb, "\nwmesg\tcount\n");
enabled = prof_enabled;
mtx_lock_spin(&sleepq_prof_lock);
prof_enabled = 0;
mtx_unlock_spin(&sleepq_prof_lock);
for (i = 0; i < SC_TABLESIZE; i++) {
LIST_FOREACH(sp, &sleepq_hash[i], sp_link) {
sbuf_printf(sb, "%s\t%ld\n",
sp->sp_wmesg, sp->sp_count);
}
}
mtx_lock_spin(&sleepq_prof_lock);
prof_enabled = enabled;
mtx_unlock_spin(&sleepq_prof_lock);
Re-add r212370 now that the LOR in powerpc64 has been resolved: Add a drain function for struct sysctl_req, and use it for a variety of handlers, some of which had to do awkward things to get a large enough SBUF_FIXEDLEN buffer. Note that some sysctl handlers were explicitly outputting a trailing NUL byte. This behaviour was preserved, though it should not be necessary. Reviewed by: phk (original patch)
2010-09-16 16:13:12 +00:00
error = sbuf_finish(sb);
- Add a facility similar to LOCK_PROFILING under SLEEPQUEUE_PROFILING. Keep a simple (wmesg, count) tuple in a hash to keep track of how many times we sleep at each wait message. We hash on message and not channel. No line number information is given as typically wait messages are not used in more than one place. Identical strings defined at different addresses will show up with seperate counters. - Use debug.sleepq.enable to enable, .reset to reset, and .stats dumps stats. - Do an unsynchronized check in sleepq_switch() prior to switching before calling sleepq_profile() which uses a global lock to synchronize the hash. Only sleeps which actually cause a context switch are counted.
2008-03-19 07:22:07 +00:00
sbuf_delete(sb);
return (error);
}
SYSCTL_PROC(_debug_sleepq, OID_AUTO, stats, CTLTYPE_STRING | CTLFLAG_RD,
NULL, 0, dump_sleepq_prof_stats, "A", "Sleepqueue profiling statistics");
SYSCTL_PROC(_debug_sleepq, OID_AUTO, reset, CTLTYPE_INT | CTLFLAG_RW,
NULL, 0, reset_sleepq_prof_stats, "I",
"Reset sleepqueue profiling statistics");
SYSCTL_PROC(_debug_sleepq, OID_AUTO, enable, CTLTYPE_INT | CTLFLAG_RW,
NULL, 0, enable_sleepq_prof, "I", "Enable sleepqueue profiling");
#endif
Add a new ddb command 'show sleepq'. It takes a wait channel as an argument and looks for a sleep queue associated with that wait channel. If it finds one it will display information such as the list of threads sleeping on that queue. If it can't find a sleep queue for that wait channel, then it will see if that address matches any of the active sleep queues. If so, it will display information about the sleepq at the specified address.
2006-01-27 22:24:07 +00:00
#ifdef DDB
DB_SHOW_COMMAND(sleepq, db_show_sleepqueue)
{
struct sleepqueue_chain *sc;
struct sleepqueue *sq;
lock unused when INVARIANTS not defined, so don't declare it then
2006-01-28 00:49:31 +00:00
#ifdef INVARIANTS
Add a new ddb command 'show sleepq'. It takes a wait channel as an argument and looks for a sleep queue associated with that wait channel. If it finds one it will display information such as the list of threads sleeping on that queue. If it can't find a sleep queue for that wait channel, then it will see if that address matches any of the active sleep queues. If so, it will display information about the sleepq at the specified address.
2006-01-27 22:24:07 +00:00
struct lock_object *lock;
lock unused when INVARIANTS not defined, so don't declare it then
2006-01-28 00:49:31 +00:00
#endif
Add a new ddb command 'show sleepq'. It takes a wait channel as an argument and looks for a sleep queue associated with that wait channel. If it finds one it will display information such as the list of threads sleeping on that queue. If it can't find a sleep queue for that wait channel, then it will see if that address matches any of the active sleep queues. If so, it will display information about the sleepq at the specified address.
2006-01-27 22:24:07 +00:00
struct thread *td;
void *wchan;
int i;
if (!have_addr)
return;
/*
* First, see if there is an active sleep queue for the wait channel
* indicated by the address.
*/
wchan = (void *)addr;
sc = SC_LOOKUP(wchan);
LIST_FOREACH(sq, &sc->sc_queues, sq_hash)
if (sq->sq_wchan == wchan)
goto found;
/*
* Second, see if there is an active sleep queue at the address
* indicated.
*/
for (i = 0; i < SC_TABLESIZE; i++)
LIST_FOREACH(sq, &sleepq_chains[i].sc_queues, sq_hash) {
if (sq == (struct sleepqueue *)addr)
goto found;
}
db_printf("Unable to locate a sleep queue via %p\n", (void *)addr);
return;
found:
db_printf("Wait channel: %p\n", sq->sq_wchan);
db_printf("Queue type: %d\n", sq->sq_type);
Introduce the new kernel thread called "deadlock resolver". While the name is pretentious, a good explanation of its targets is reported in this 17 months old presentation e-mail: http://lists.freebsd.org/pipermail/freebsd-arch/2008-August/008452.html In order to implement it, the sq_type in sleepqueues is mandatory and not only compiled along with INVARIANTS option. Additively, a new sleepqueue function, sleepq_type() is added, returning the type of the sleepqueue linked to a wchan. Three new sysctls are added in order to configure the thread: debug.deadlkres.slptime_threshold debug.deadlkres.blktime_threshold debug.deadlkres.sleepfreq rappresenting the thresholds for sleep and block time that will lead to a deadlock matching (when exceeded), while the sleepfreq rappresents the number of seconds between 2 consecutive thread runnings. In order to enable the deadlock resolver thread recompile your kernel with the option DEADLKRES. Reviewed by: jeff Tested by: pho, Giovanni Trematerra Sponsored by: Nokia Incorporated, Sandvine Incorporated MFC after: 2 weeks
2010-01-09 01:46:38 +00:00
#ifdef INVARIANTS
Add a new ddb command 'show sleepq'. It takes a wait channel as an argument and looks for a sleep queue associated with that wait channel. If it finds one it will display information such as the list of threads sleeping on that queue. If it can't find a sleep queue for that wait channel, then it will see if that address matches any of the active sleep queues. If so, it will display information about the sleepq at the specified address.
2006-01-27 22:24:07 +00:00
if (sq->sq_lock) {
Change sleepq_add(9) argument from 'struct mtx *' to 'struct lock_object *', which allows to use it with different kinds of locks. For example it allows to implement Solaris conditions variables which will be used in ZFS port on top of sx(9) locks. Reviewed by: jhb
2006-11-16 01:02:00 +00:00
lock = sq->sq_lock;
Add a new ddb command 'show sleepq'. It takes a wait channel as an argument and looks for a sleep queue associated with that wait channel. If it finds one it will display information such as the list of threads sleeping on that queue. If it can't find a sleep queue for that wait channel, then it will see if that address matches any of the active sleep queues. If so, it will display information about the sleepq at the specified address.
2006-01-27 22:24:07 +00:00
db_printf("Associated Interlock: %p - (%s) %s\n", lock,
LOCK_CLASS(lock)->lc_name, lock->lo_name);
}
#endif
db_printf("Blocked threads:\n");
Add second sleep queue so that sx and lockmgr can have separate sleep queues for shared and exclusive acquisitions Submitted by: Attilio Rao Approved by: jhb
2006-12-16 06:54:09 +00:00
for (i = 0; i < NR_SLEEPQS; i++) {
db_printf("\nQueue[%d]:\n", i);
if (TAILQ_EMPTY(&sq->sq_blocked[i]))
db_printf("\tempty\n");
else
TAILQ_FOREACH(td, &sq->sq_blocked[0],
td_slpq) {
db_printf("\t%p (tid %d, pid %d, \"%s\")\n", td,
td->td_tid, td->td_proc->p_pid,
Really fix this.
2008-07-28 18:33:43 +00:00
td->td_name);
Add second sleep queue so that sx and lockmgr can have separate sleep queues for shared and exclusive acquisitions Submitted by: Attilio Rao Approved by: jhb
2006-12-16 06:54:09 +00:00
}
In current code, threads performing an interruptible sleep (on both sxlock, via the sx_{s, x}lock_sig() interface, or plain lockmgr), will leave the waiters flag on forcing the owner to do a wakeup even when if the waiter queue is empty. That operation may lead to a deadlock in the case of doing a fake wakeup on the "preferred" (based on the wakeup algorithm) queue while the other queue has real waiters on it, because nobody is going to wakeup the 2nd queue waiters and they will sleep indefinitively. A similar bug, is present, for lockmgr in the case the waiters are sleeping with LK_SLEEPFAIL on. In this case, even if the waiters queue is not empty, the waiters won't progress after being awake but they will just fail, still not taking care of the 2nd queue waiters (as instead the lock owned doing the wakeup would expect). In order to fix this bug in a cheap way (without adding too much locking and complicating too much the semantic) add a sleepqueue interface which does report the actual number of waiters on a specified queue of a waitchannel (sleepq_sleepcnt()) and use it in order to determine if the exclusive waiters (or shared waiters) are actually present on the lockmgr (or sx) before to give them precedence in the wakeup algorithm. This fix alone, however doesn't solve the LK_SLEEPFAIL bug. In order to cope with it, add the tracking of how many exclusive LK_SLEEPFAIL waiters a lockmgr has and if all the waiters on the exclusive waiters queue are LK_SLEEPFAIL just wake both queues. The sleepq_sleepcnt() introduction and ABI breakage require __FreeBSD_version bumping. Reported by: avg, kib, pho Reviewed by: kib Tested by: pho
2009-12-12 21:31:07 +00:00
db_printf("(expected: %u)\n", sq->sq_blockedcnt[i]);
Add second sleep queue so that sx and lockmgr can have separate sleep queues for shared and exclusive acquisitions Submitted by: Attilio Rao Approved by: jhb
2006-12-16 06:54:09 +00:00
}
Add a new ddb command 'show sleepq'. It takes a wait channel as an argument and looks for a sleep queue associated with that wait channel. If it finds one it will display information such as the list of threads sleeping on that queue. If it can't find a sleep queue for that wait channel, then it will see if that address matches any of the active sleep queues. If so, it will display information about the sleepq at the specified address.
2006-01-27 22:24:07 +00:00
}
Add a 'show sleepqueue' alias for 'show sleepq' in DDB.
2006-04-17 20:16:32 +00:00
/* Alias 'show sleepqueue' to 'show sleepq'. */
Make ddb command registration dynamic so modules can extend the command set (only so long as the module is present): o add db_command_register and db_command_unregister to add and remove commands, respectively o replace linker sets with SYSINIT's (and SYSUINIT's) that register commands o expose 3 list heads: db_cmd_table, db_show_table, and db_show_all_table for registering top-level commands, show operands, and show all operands, respectively While here also: o sort command lists o add DB_ALIAS, DB_SHOW_ALIAS, and DB_SHOW_ALL_ALIAS to add aliases for existing commands o add "show all trace" as an alias for "show alltrace" o add "show all locks" as an alias for "show alllocks" Submitted by: Guillaume Ballet <gballet@gmail.com> (original version) Reviewed by: jhb MFC after: 1 month
2008-09-15 22:45:14 +00:00
DB_SHOW_ALIAS(sleepqueue, db_show_sleepqueue);
Add a new ddb command 'show sleepq'. It takes a wait channel as an argument and looks for a sleep queue associated with that wait channel. If it finds one it will display information such as the list of threads sleeping on that queue. If it can't find a sleep queue for that wait channel, then it will see if that address matches any of the active sleep queues. If so, it will display information about the sleepq at the specified address.
2006-01-27 22:24:07 +00:00
#endif
Reference in New Issue Copy Permalink